linux/kernel/rcutree_trace.c

516 lines
14 KiB
C
Raw Normal View History

"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
/*
* Read-Copy Update tracing for classic implementation
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
* Copyright IBM Corporation, 2008
*
* Papers: http://www.rdrop.com/users/paulmck/RCU
*
* For detailed explanation of Read-Copy Update mechanism see -
* Documentation/RCU
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
*
*/
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/spinlock.h>
#include <linux/smp.h>
#include <linux/rcupdate.h>
#include <linux/interrupt.h>
#include <linux/sched.h>
#include <asm/atomic.h>
#include <linux/bitops.h>
#include <linux/module.h>
#include <linux/completion.h>
#include <linux/moduleparam.h>
#include <linux/percpu.h>
#include <linux/notifier.h>
#include <linux/cpu.h>
#include <linux/mutex.h>
#include <linux/debugfs.h>
#include <linux/seq_file.h>
#define RCU_TREE_NONCORE
#include "rcutree.h"
#ifdef CONFIG_RCU_BOOST
DECLARE_PER_CPU(unsigned int, rcu_cpu_kthread_status);
DECLARE_PER_CPU(unsigned int, rcu_cpu_kthread_cpu);
DECLARE_PER_CPU(unsigned int, rcu_cpu_kthread_loops);
DECLARE_PER_CPU(char, rcu_cpu_has_work);
static char convert_kthread_status(unsigned int kthread_status)
{
if (kthread_status > RCU_KTHREAD_MAX)
return '?';
return "SRWOY"[kthread_status];
}
#endif /* #ifdef CONFIG_RCU_BOOST */
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
static void print_one_rcu_data(struct seq_file *m, struct rcu_data *rdp)
{
if (!rdp->beenonline)
return;
seq_printf(m, "%3d%cc=%lu g=%lu pq=%d pqc=%lu qp=%d",
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->cpu,
cpu_is_offline(rdp->cpu) ? '!' : ' ',
rdp->completed, rdp->gpnum,
rdp->passed_quiesc, rdp->passed_quiesc_completed,
rcu: Make hierarchical RCU less IPI-happy This patch fixes a hierarchical-RCU performance bug located by Anton Blanchard. The problem stems from a misguided attempt to provide a work-around for jiffies-counter failure. This work-around uses a per-CPU n_rcu_pending counter, which is incremented on each call to rcu_pending(), which in turn is called from each scheduling-clock interrupt. Each CPU then treats this counter as a surrogate for the jiffies counter, so that if the jiffies counter fails to advance, the per-CPU n_rcu_pending counter will cause RCU to invoke force_quiescent_state(), which in turn will (among other things) send resched IPIs to CPUs that have thus far failed to pass through an RCU quiescent state. Unfortunately, each CPU resets only its own counter after sending a batch of IPIs. This means that the other CPUs will also (needlessly) send -another- round of IPIs, for a full N-squared set of IPIs in the worst case every three scheduler-clock ticks until the grace period finally ends. It is not reasonable for a given CPU to reset each and every n_rcu_pending for all the other CPUs, so this patch instead simply disables the jiffies-counter "training wheels", thus eliminating the excessive IPIs. Note that the jiffies-counter IPIs do not have this problem due to the fact that the jiffies counter is global, so that the CPU sending the IPIs can easily reset things, thus preventing the other CPUs from sending redundant IPIs. Note also that the n_rcu_pending counter remains, as it will continue to be used for tracing. It may also see use to update the jiffies counter, should an appropriate kick-the-jiffies-counter API appear. Located-by: Anton Blanchard <anton@au1.ibm.com> Tested-by: Anton Blanchard <anton@au1.ibm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: anton@samba.org Cc: akpm@linux-foundation.org Cc: dipankar@in.ibm.com Cc: manfred@colorfullife.com Cc: cl@linux-foundation.org Cc: josht@linux.vnet.ibm.com Cc: schamp@sgi.com Cc: niv@us.ibm.com Cc: dvhltc@us.ibm.com Cc: ego@in.ibm.com Cc: laijs@cn.fujitsu.com Cc: rostedt@goodmis.org Cc: peterz@infradead.org Cc: penberg@cs.helsinki.fi Cc: andi@firstfloor.org Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> LKML-Reference: <12396834793575-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-04-14 04:31:16 +00:00
rdp->qs_pending);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
#ifdef CONFIG_NO_HZ
rcu: Decrease memory-barrier usage based on semi-formal proof (Note: this was reverted, and is now being re-applied in pieces, with this being the fifth and final piece. See below for the reason that it is now felt to be safe to re-apply this.) Commit d09b62d fixed grace-period synchronization, but left some smp_mb() invocations in rcu_process_callbacks() that are no longer needed, but sheer paranoia prevented them from being removed. This commit removes them and provides a proof of correctness in their absence. It also adds a memory barrier to rcu_report_qs_rsp() immediately before the update to rsp->completed in order to handle the theoretical possibility that the compiler or CPU might move massive quantities of code into a lock-based critical section. This also proves that the sheer paranoia was not entirely unjustified, at least from a theoretical point of view. In addition, the old dyntick-idle synchronization depended on the fact that grace periods were many milliseconds in duration, so that it could be assumed that no dyntick-idle CPU could reorder a memory reference across an entire grace period. Unfortunately for this design, the addition of expedited grace periods breaks this assumption, which has the unfortunate side-effect of requiring atomic operations in the functions that track dyntick-idle state for RCU. (There is some hope that the algorithms used in user-level RCU might be applied here, but some work is required to handle the NMIs that user-space applications can happily ignore. For the short term, better safe than sorry.) This proof assumes that neither compiler nor CPU will allow a lock acquisition and release to be reordered, as doing so can result in deadlock. The proof is as follows: 1. A given CPU declares a quiescent state under the protection of its leaf rcu_node's lock. 2. If there is more than one level of rcu_node hierarchy, the last CPU to declare a quiescent state will also acquire the ->lock of the next rcu_node up in the hierarchy, but only after releasing the lower level's lock. The acquisition of this lock clearly cannot occur prior to the acquisition of the leaf node's lock. 3. Step 2 repeats until we reach the root rcu_node structure. Please note again that only one lock is held at a time through this process. The acquisition of the root rcu_node's ->lock must occur after the release of that of the leaf rcu_node. 4. At this point, we set the ->completed field in the rcu_state structure in rcu_report_qs_rsp(). However, if the rcu_node hierarchy contains only one rcu_node, then in theory the code preceding the quiescent state could leak into the critical section. We therefore precede the update of ->completed with a memory barrier. All CPUs will therefore agree that any updates preceding any report of a quiescent state will have happened before the update of ->completed. 5. Regardless of whether a new grace period is needed, rcu_start_gp() will propagate the new value of ->completed to all of the leaf rcu_node structures, under the protection of each rcu_node's ->lock. If a new grace period is needed immediately, this propagation will occur in the same critical section that ->completed was set in, but courtesy of the memory barrier in #4 above, is still seen to follow any pre-quiescent-state activity. 6. When a given CPU invokes __rcu_process_gp_end(), it becomes aware of the end of the old grace period and therefore makes any RCU callbacks that were waiting on that grace period eligible for invocation. If this CPU is the same one that detected the end of the grace period, and if there is but a single rcu_node in the hierarchy, we will still be in the single critical section. In this case, the memory barrier in step #4 guarantees that all callbacks will be seen to execute after each CPU's quiescent state. On the other hand, if this is a different CPU, it will acquire the leaf rcu_node's ->lock, and will again be serialized after each CPU's quiescent state for the old grace period. On the strength of this proof, this commit therefore removes the memory barriers from rcu_process_callbacks() and adds one to rcu_report_qs_rsp(). The effect is to reduce the number of memory barriers by one and to reduce the frequency of execution from about once per scheduling tick per CPU to once per grace period. This was reverted do to hangs found during testing by Yinghai Lu and Ingo Molnar. Frederic Weisbecker supplied Yinghai with tracing that located the underlying problem, and Frederic also provided the fix. The underlying problem was that the HARDIRQ_ENTER() macro from lib/locking-selftest.c invoked irq_enter(), which in turn invokes rcu_irq_enter(), but HARDIRQ_EXIT() invoked __irq_exit(), which does not invoke rcu_irq_exit(). This situation resulted in calls to rcu_irq_enter() that were not balanced by the required calls to rcu_irq_exit(). Therefore, after these locking selftests completed, RCU's dyntick-idle nesting count was a large number (for example, 72), which caused RCU to to conclude that the affected CPU was not in dyntick-idle mode when in fact it was. RCU would therefore incorrectly wait for this dyntick-idle CPU, resulting in hangs. In contrast, with Frederic's patch, which replaces the irq_enter() in HARDIRQ_ENTER() with an __irq_enter(), these tests don't ever call either rcu_irq_enter() or rcu_irq_exit(), which works because the CPU running the test is already marked as not being in dyntick-idle mode. This means that the rcu_irq_enter() and rcu_irq_exit() calls and RCU then has no problem working out which CPUs are in dyntick-idle mode and which are not. The reason that the imbalance was not noticed before the barrier patch was applied is that the old implementation of rcu_enter_nohz() ignored the nesting depth. This could still result in delays, but much shorter ones. Whenever there was a delay, RCU would IPI the CPU with the unbalanced nesting level, which would eventually result in rcu_enter_nohz() being called, which in turn would force RCU to see that the CPU was in dyntick-idle mode. The reason that very few people noticed the problem is that the mismatched irq_enter() vs. __irq_exit() occured only when the kernel was built with CONFIG_DEBUG_LOCKING_API_SELFTESTS. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org>
2010-09-07 17:38:22 +00:00
seq_printf(m, " dt=%d/%d/%d df=%lu",
atomic_read(&rdp->dynticks->dynticks),
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->dynticks->dynticks_nesting,
rcu: Decrease memory-barrier usage based on semi-formal proof (Note: this was reverted, and is now being re-applied in pieces, with this being the fifth and final piece. See below for the reason that it is now felt to be safe to re-apply this.) Commit d09b62d fixed grace-period synchronization, but left some smp_mb() invocations in rcu_process_callbacks() that are no longer needed, but sheer paranoia prevented them from being removed. This commit removes them and provides a proof of correctness in their absence. It also adds a memory barrier to rcu_report_qs_rsp() immediately before the update to rsp->completed in order to handle the theoretical possibility that the compiler or CPU might move massive quantities of code into a lock-based critical section. This also proves that the sheer paranoia was not entirely unjustified, at least from a theoretical point of view. In addition, the old dyntick-idle synchronization depended on the fact that grace periods were many milliseconds in duration, so that it could be assumed that no dyntick-idle CPU could reorder a memory reference across an entire grace period. Unfortunately for this design, the addition of expedited grace periods breaks this assumption, which has the unfortunate side-effect of requiring atomic operations in the functions that track dyntick-idle state for RCU. (There is some hope that the algorithms used in user-level RCU might be applied here, but some work is required to handle the NMIs that user-space applications can happily ignore. For the short term, better safe than sorry.) This proof assumes that neither compiler nor CPU will allow a lock acquisition and release to be reordered, as doing so can result in deadlock. The proof is as follows: 1. A given CPU declares a quiescent state under the protection of its leaf rcu_node's lock. 2. If there is more than one level of rcu_node hierarchy, the last CPU to declare a quiescent state will also acquire the ->lock of the next rcu_node up in the hierarchy, but only after releasing the lower level's lock. The acquisition of this lock clearly cannot occur prior to the acquisition of the leaf node's lock. 3. Step 2 repeats until we reach the root rcu_node structure. Please note again that only one lock is held at a time through this process. The acquisition of the root rcu_node's ->lock must occur after the release of that of the leaf rcu_node. 4. At this point, we set the ->completed field in the rcu_state structure in rcu_report_qs_rsp(). However, if the rcu_node hierarchy contains only one rcu_node, then in theory the code preceding the quiescent state could leak into the critical section. We therefore precede the update of ->completed with a memory barrier. All CPUs will therefore agree that any updates preceding any report of a quiescent state will have happened before the update of ->completed. 5. Regardless of whether a new grace period is needed, rcu_start_gp() will propagate the new value of ->completed to all of the leaf rcu_node structures, under the protection of each rcu_node's ->lock. If a new grace period is needed immediately, this propagation will occur in the same critical section that ->completed was set in, but courtesy of the memory barrier in #4 above, is still seen to follow any pre-quiescent-state activity. 6. When a given CPU invokes __rcu_process_gp_end(), it becomes aware of the end of the old grace period and therefore makes any RCU callbacks that were waiting on that grace period eligible for invocation. If this CPU is the same one that detected the end of the grace period, and if there is but a single rcu_node in the hierarchy, we will still be in the single critical section. In this case, the memory barrier in step #4 guarantees that all callbacks will be seen to execute after each CPU's quiescent state. On the other hand, if this is a different CPU, it will acquire the leaf rcu_node's ->lock, and will again be serialized after each CPU's quiescent state for the old grace period. On the strength of this proof, this commit therefore removes the memory barriers from rcu_process_callbacks() and adds one to rcu_report_qs_rsp(). The effect is to reduce the number of memory barriers by one and to reduce the frequency of execution from about once per scheduling tick per CPU to once per grace period. This was reverted do to hangs found during testing by Yinghai Lu and Ingo Molnar. Frederic Weisbecker supplied Yinghai with tracing that located the underlying problem, and Frederic also provided the fix. The underlying problem was that the HARDIRQ_ENTER() macro from lib/locking-selftest.c invoked irq_enter(), which in turn invokes rcu_irq_enter(), but HARDIRQ_EXIT() invoked __irq_exit(), which does not invoke rcu_irq_exit(). This situation resulted in calls to rcu_irq_enter() that were not balanced by the required calls to rcu_irq_exit(). Therefore, after these locking selftests completed, RCU's dyntick-idle nesting count was a large number (for example, 72), which caused RCU to to conclude that the affected CPU was not in dyntick-idle mode when in fact it was. RCU would therefore incorrectly wait for this dyntick-idle CPU, resulting in hangs. In contrast, with Frederic's patch, which replaces the irq_enter() in HARDIRQ_ENTER() with an __irq_enter(), these tests don't ever call either rcu_irq_enter() or rcu_irq_exit(), which works because the CPU running the test is already marked as not being in dyntick-idle mode. This means that the rcu_irq_enter() and rcu_irq_exit() calls and RCU then has no problem working out which CPUs are in dyntick-idle mode and which are not. The reason that the imbalance was not noticed before the barrier patch was applied is that the old implementation of rcu_enter_nohz() ignored the nesting depth. This could still result in delays, but much shorter ones. Whenever there was a delay, RCU would IPI the CPU with the unbalanced nesting level, which would eventually result in rcu_enter_nohz() being called, which in turn would force RCU to see that the CPU was in dyntick-idle mode. The reason that very few people noticed the problem is that the mismatched irq_enter() vs. __irq_exit() occured only when the kernel was built with CONFIG_DEBUG_LOCKING_API_SELFTESTS. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org>
2010-09-07 17:38:22 +00:00
rdp->dynticks->dynticks_nmi_nesting,
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->dynticks_fqs);
#endif /* #ifdef CONFIG_NO_HZ */
seq_printf(m, " of=%lu ri=%lu", rdp->offline_fqs, rdp->resched_ipi);
seq_printf(m, " ql=%ld qs=%c%c%c%c",
rdp->qlen,
".N"[rdp->nxttail[RCU_NEXT_READY_TAIL] !=
rdp->nxttail[RCU_NEXT_TAIL]],
".R"[rdp->nxttail[RCU_WAIT_TAIL] !=
rdp->nxttail[RCU_NEXT_READY_TAIL]],
".W"[rdp->nxttail[RCU_DONE_TAIL] !=
rdp->nxttail[RCU_WAIT_TAIL]],
".D"[&rdp->nxtlist != rdp->nxttail[RCU_DONE_TAIL]]);
#ifdef CONFIG_RCU_BOOST
seq_printf(m, " kt=%d/%c/%d ktl=%x",
per_cpu(rcu_cpu_has_work, rdp->cpu),
convert_kthread_status(per_cpu(rcu_cpu_kthread_status,
rdp->cpu)),
per_cpu(rcu_cpu_kthread_cpu, rdp->cpu),
per_cpu(rcu_cpu_kthread_loops, rdp->cpu) & 0xffff);
#endif /* #ifdef CONFIG_RCU_BOOST */
seq_printf(m, " b=%ld", rdp->blimit);
seq_printf(m, " ci=%lu co=%lu ca=%lu\n",
rdp->n_cbs_invoked, rdp->n_cbs_orphaned, rdp->n_cbs_adopted);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
}
#define PRINT_RCU_DATA(name, func, m) \
do { \
int _p_r_d_i; \
\
for_each_possible_cpu(_p_r_d_i) \
func(m, &per_cpu(name, _p_r_d_i)); \
} while (0)
static int show_rcudata(struct seq_file *m, void *unused)
{
rcu: Merge preemptable-RCU functionality into hierarchical RCU Create a kernel/rcutree_plugin.h file that contains definitions for preemptable RCU (or, under the #else branch of the #ifdef, empty definitions for the classic non-preemptable semantics). These definitions fit into plugins defined in kernel/rcutree.c for this purpose. This variant of preemptable RCU uses a new algorithm whose read-side expense is roughly that of classic hierarchical RCU under CONFIG_PREEMPT. This new algorithm's update-side expense is similar to that of classic hierarchical RCU, and, in absence of read-side preemption or blocking, is exactly that of classic hierarchical RCU. Perhaps more important, this new algorithm has a much simpler implementation, saving well over 1,000 lines of code compared to mainline's implementation of preemptable RCU, which will hopefully be retired in favor of this new algorithm. The simplifications are obtained by maintaining per-task nesting state for running tasks, and using a simple lock-protected algorithm to handle accounting when tasks block within RCU read-side critical sections, making use of lessons learned while creating numerous user-level RCU implementations over the past 18 months. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: laijs@cn.fujitsu.com Cc: dipankar@in.ibm.com Cc: akpm@linux-foundation.org Cc: mathieu.desnoyers@polymtl.ca Cc: josht@linux.vnet.ibm.com Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org LKML-Reference: <12509746134003-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-22 20:56:52 +00:00
#ifdef CONFIG_TREE_PREEMPT_RCU
seq_puts(m, "rcu_preempt:\n");
PRINT_RCU_DATA(rcu_preempt_data, print_one_rcu_data, m);
#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
seq_puts(m, "rcu_sched:\n");
PRINT_RCU_DATA(rcu_sched_data, print_one_rcu_data, m);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
seq_puts(m, "rcu_bh:\n");
PRINT_RCU_DATA(rcu_bh_data, print_one_rcu_data, m);
return 0;
}
static int rcudata_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcudata, NULL);
}
static const struct file_operations rcudata_fops = {
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
.owner = THIS_MODULE,
.open = rcudata_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static void print_one_rcu_data_csv(struct seq_file *m, struct rcu_data *rdp)
{
if (!rdp->beenonline)
return;
seq_printf(m, "%d,%s,%lu,%lu,%d,%lu,%d",
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->cpu,
cpu_is_offline(rdp->cpu) ? "\"N\"" : "\"Y\"",
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->completed, rdp->gpnum,
rdp->passed_quiesc, rdp->passed_quiesc_completed,
rcu: Make hierarchical RCU less IPI-happy This patch fixes a hierarchical-RCU performance bug located by Anton Blanchard. The problem stems from a misguided attempt to provide a work-around for jiffies-counter failure. This work-around uses a per-CPU n_rcu_pending counter, which is incremented on each call to rcu_pending(), which in turn is called from each scheduling-clock interrupt. Each CPU then treats this counter as a surrogate for the jiffies counter, so that if the jiffies counter fails to advance, the per-CPU n_rcu_pending counter will cause RCU to invoke force_quiescent_state(), which in turn will (among other things) send resched IPIs to CPUs that have thus far failed to pass through an RCU quiescent state. Unfortunately, each CPU resets only its own counter after sending a batch of IPIs. This means that the other CPUs will also (needlessly) send -another- round of IPIs, for a full N-squared set of IPIs in the worst case every three scheduler-clock ticks until the grace period finally ends. It is not reasonable for a given CPU to reset each and every n_rcu_pending for all the other CPUs, so this patch instead simply disables the jiffies-counter "training wheels", thus eliminating the excessive IPIs. Note that the jiffies-counter IPIs do not have this problem due to the fact that the jiffies counter is global, so that the CPU sending the IPIs can easily reset things, thus preventing the other CPUs from sending redundant IPIs. Note also that the n_rcu_pending counter remains, as it will continue to be used for tracing. It may also see use to update the jiffies counter, should an appropriate kick-the-jiffies-counter API appear. Located-by: Anton Blanchard <anton@au1.ibm.com> Tested-by: Anton Blanchard <anton@au1.ibm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: anton@samba.org Cc: akpm@linux-foundation.org Cc: dipankar@in.ibm.com Cc: manfred@colorfullife.com Cc: cl@linux-foundation.org Cc: josht@linux.vnet.ibm.com Cc: schamp@sgi.com Cc: niv@us.ibm.com Cc: dvhltc@us.ibm.com Cc: ego@in.ibm.com Cc: laijs@cn.fujitsu.com Cc: rostedt@goodmis.org Cc: peterz@infradead.org Cc: penberg@cs.helsinki.fi Cc: andi@firstfloor.org Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> LKML-Reference: <12396834793575-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-04-14 04:31:16 +00:00
rdp->qs_pending);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
#ifdef CONFIG_NO_HZ
seq_printf(m, ",%d,%d,%d,%lu",
rcu: Decrease memory-barrier usage based on semi-formal proof (Note: this was reverted, and is now being re-applied in pieces, with this being the fifth and final piece. See below for the reason that it is now felt to be safe to re-apply this.) Commit d09b62d fixed grace-period synchronization, but left some smp_mb() invocations in rcu_process_callbacks() that are no longer needed, but sheer paranoia prevented them from being removed. This commit removes them and provides a proof of correctness in their absence. It also adds a memory barrier to rcu_report_qs_rsp() immediately before the update to rsp->completed in order to handle the theoretical possibility that the compiler or CPU might move massive quantities of code into a lock-based critical section. This also proves that the sheer paranoia was not entirely unjustified, at least from a theoretical point of view. In addition, the old dyntick-idle synchronization depended on the fact that grace periods were many milliseconds in duration, so that it could be assumed that no dyntick-idle CPU could reorder a memory reference across an entire grace period. Unfortunately for this design, the addition of expedited grace periods breaks this assumption, which has the unfortunate side-effect of requiring atomic operations in the functions that track dyntick-idle state for RCU. (There is some hope that the algorithms used in user-level RCU might be applied here, but some work is required to handle the NMIs that user-space applications can happily ignore. For the short term, better safe than sorry.) This proof assumes that neither compiler nor CPU will allow a lock acquisition and release to be reordered, as doing so can result in deadlock. The proof is as follows: 1. A given CPU declares a quiescent state under the protection of its leaf rcu_node's lock. 2. If there is more than one level of rcu_node hierarchy, the last CPU to declare a quiescent state will also acquire the ->lock of the next rcu_node up in the hierarchy, but only after releasing the lower level's lock. The acquisition of this lock clearly cannot occur prior to the acquisition of the leaf node's lock. 3. Step 2 repeats until we reach the root rcu_node structure. Please note again that only one lock is held at a time through this process. The acquisition of the root rcu_node's ->lock must occur after the release of that of the leaf rcu_node. 4. At this point, we set the ->completed field in the rcu_state structure in rcu_report_qs_rsp(). However, if the rcu_node hierarchy contains only one rcu_node, then in theory the code preceding the quiescent state could leak into the critical section. We therefore precede the update of ->completed with a memory barrier. All CPUs will therefore agree that any updates preceding any report of a quiescent state will have happened before the update of ->completed. 5. Regardless of whether a new grace period is needed, rcu_start_gp() will propagate the new value of ->completed to all of the leaf rcu_node structures, under the protection of each rcu_node's ->lock. If a new grace period is needed immediately, this propagation will occur in the same critical section that ->completed was set in, but courtesy of the memory barrier in #4 above, is still seen to follow any pre-quiescent-state activity. 6. When a given CPU invokes __rcu_process_gp_end(), it becomes aware of the end of the old grace period and therefore makes any RCU callbacks that were waiting on that grace period eligible for invocation. If this CPU is the same one that detected the end of the grace period, and if there is but a single rcu_node in the hierarchy, we will still be in the single critical section. In this case, the memory barrier in step #4 guarantees that all callbacks will be seen to execute after each CPU's quiescent state. On the other hand, if this is a different CPU, it will acquire the leaf rcu_node's ->lock, and will again be serialized after each CPU's quiescent state for the old grace period. On the strength of this proof, this commit therefore removes the memory barriers from rcu_process_callbacks() and adds one to rcu_report_qs_rsp(). The effect is to reduce the number of memory barriers by one and to reduce the frequency of execution from about once per scheduling tick per CPU to once per grace period. This was reverted do to hangs found during testing by Yinghai Lu and Ingo Molnar. Frederic Weisbecker supplied Yinghai with tracing that located the underlying problem, and Frederic also provided the fix. The underlying problem was that the HARDIRQ_ENTER() macro from lib/locking-selftest.c invoked irq_enter(), which in turn invokes rcu_irq_enter(), but HARDIRQ_EXIT() invoked __irq_exit(), which does not invoke rcu_irq_exit(). This situation resulted in calls to rcu_irq_enter() that were not balanced by the required calls to rcu_irq_exit(). Therefore, after these locking selftests completed, RCU's dyntick-idle nesting count was a large number (for example, 72), which caused RCU to to conclude that the affected CPU was not in dyntick-idle mode when in fact it was. RCU would therefore incorrectly wait for this dyntick-idle CPU, resulting in hangs. In contrast, with Frederic's patch, which replaces the irq_enter() in HARDIRQ_ENTER() with an __irq_enter(), these tests don't ever call either rcu_irq_enter() or rcu_irq_exit(), which works because the CPU running the test is already marked as not being in dyntick-idle mode. This means that the rcu_irq_enter() and rcu_irq_exit() calls and RCU then has no problem working out which CPUs are in dyntick-idle mode and which are not. The reason that the imbalance was not noticed before the barrier patch was applied is that the old implementation of rcu_enter_nohz() ignored the nesting depth. This could still result in delays, but much shorter ones. Whenever there was a delay, RCU would IPI the CPU with the unbalanced nesting level, which would eventually result in rcu_enter_nohz() being called, which in turn would force RCU to see that the CPU was in dyntick-idle mode. The reason that very few people noticed the problem is that the mismatched irq_enter() vs. __irq_exit() occured only when the kernel was built with CONFIG_DEBUG_LOCKING_API_SELFTESTS. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org>
2010-09-07 17:38:22 +00:00
atomic_read(&rdp->dynticks->dynticks),
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->dynticks->dynticks_nesting,
rcu: Decrease memory-barrier usage based on semi-formal proof (Note: this was reverted, and is now being re-applied in pieces, with this being the fifth and final piece. See below for the reason that it is now felt to be safe to re-apply this.) Commit d09b62d fixed grace-period synchronization, but left some smp_mb() invocations in rcu_process_callbacks() that are no longer needed, but sheer paranoia prevented them from being removed. This commit removes them and provides a proof of correctness in their absence. It also adds a memory barrier to rcu_report_qs_rsp() immediately before the update to rsp->completed in order to handle the theoretical possibility that the compiler or CPU might move massive quantities of code into a lock-based critical section. This also proves that the sheer paranoia was not entirely unjustified, at least from a theoretical point of view. In addition, the old dyntick-idle synchronization depended on the fact that grace periods were many milliseconds in duration, so that it could be assumed that no dyntick-idle CPU could reorder a memory reference across an entire grace period. Unfortunately for this design, the addition of expedited grace periods breaks this assumption, which has the unfortunate side-effect of requiring atomic operations in the functions that track dyntick-idle state for RCU. (There is some hope that the algorithms used in user-level RCU might be applied here, but some work is required to handle the NMIs that user-space applications can happily ignore. For the short term, better safe than sorry.) This proof assumes that neither compiler nor CPU will allow a lock acquisition and release to be reordered, as doing so can result in deadlock. The proof is as follows: 1. A given CPU declares a quiescent state under the protection of its leaf rcu_node's lock. 2. If there is more than one level of rcu_node hierarchy, the last CPU to declare a quiescent state will also acquire the ->lock of the next rcu_node up in the hierarchy, but only after releasing the lower level's lock. The acquisition of this lock clearly cannot occur prior to the acquisition of the leaf node's lock. 3. Step 2 repeats until we reach the root rcu_node structure. Please note again that only one lock is held at a time through this process. The acquisition of the root rcu_node's ->lock must occur after the release of that of the leaf rcu_node. 4. At this point, we set the ->completed field in the rcu_state structure in rcu_report_qs_rsp(). However, if the rcu_node hierarchy contains only one rcu_node, then in theory the code preceding the quiescent state could leak into the critical section. We therefore precede the update of ->completed with a memory barrier. All CPUs will therefore agree that any updates preceding any report of a quiescent state will have happened before the update of ->completed. 5. Regardless of whether a new grace period is needed, rcu_start_gp() will propagate the new value of ->completed to all of the leaf rcu_node structures, under the protection of each rcu_node's ->lock. If a new grace period is needed immediately, this propagation will occur in the same critical section that ->completed was set in, but courtesy of the memory barrier in #4 above, is still seen to follow any pre-quiescent-state activity. 6. When a given CPU invokes __rcu_process_gp_end(), it becomes aware of the end of the old grace period and therefore makes any RCU callbacks that were waiting on that grace period eligible for invocation. If this CPU is the same one that detected the end of the grace period, and if there is but a single rcu_node in the hierarchy, we will still be in the single critical section. In this case, the memory barrier in step #4 guarantees that all callbacks will be seen to execute after each CPU's quiescent state. On the other hand, if this is a different CPU, it will acquire the leaf rcu_node's ->lock, and will again be serialized after each CPU's quiescent state for the old grace period. On the strength of this proof, this commit therefore removes the memory barriers from rcu_process_callbacks() and adds one to rcu_report_qs_rsp(). The effect is to reduce the number of memory barriers by one and to reduce the frequency of execution from about once per scheduling tick per CPU to once per grace period. This was reverted do to hangs found during testing by Yinghai Lu and Ingo Molnar. Frederic Weisbecker supplied Yinghai with tracing that located the underlying problem, and Frederic also provided the fix. The underlying problem was that the HARDIRQ_ENTER() macro from lib/locking-selftest.c invoked irq_enter(), which in turn invokes rcu_irq_enter(), but HARDIRQ_EXIT() invoked __irq_exit(), which does not invoke rcu_irq_exit(). This situation resulted in calls to rcu_irq_enter() that were not balanced by the required calls to rcu_irq_exit(). Therefore, after these locking selftests completed, RCU's dyntick-idle nesting count was a large number (for example, 72), which caused RCU to to conclude that the affected CPU was not in dyntick-idle mode when in fact it was. RCU would therefore incorrectly wait for this dyntick-idle CPU, resulting in hangs. In contrast, with Frederic's patch, which replaces the irq_enter() in HARDIRQ_ENTER() with an __irq_enter(), these tests don't ever call either rcu_irq_enter() or rcu_irq_exit(), which works because the CPU running the test is already marked as not being in dyntick-idle mode. This means that the rcu_irq_enter() and rcu_irq_exit() calls and RCU then has no problem working out which CPUs are in dyntick-idle mode and which are not. The reason that the imbalance was not noticed before the barrier patch was applied is that the old implementation of rcu_enter_nohz() ignored the nesting depth. This could still result in delays, but much shorter ones. Whenever there was a delay, RCU would IPI the CPU with the unbalanced nesting level, which would eventually result in rcu_enter_nohz() being called, which in turn would force RCU to see that the CPU was in dyntick-idle mode. The reason that very few people noticed the problem is that the mismatched irq_enter() vs. __irq_exit() occured only when the kernel was built with CONFIG_DEBUG_LOCKING_API_SELFTESTS. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org>
2010-09-07 17:38:22 +00:00
rdp->dynticks->dynticks_nmi_nesting,
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rdp->dynticks_fqs);
#endif /* #ifdef CONFIG_NO_HZ */
seq_printf(m, ",%lu,%lu", rdp->offline_fqs, rdp->resched_ipi);
seq_printf(m, ",%ld,\"%c%c%c%c\"", rdp->qlen,
".N"[rdp->nxttail[RCU_NEXT_READY_TAIL] !=
rdp->nxttail[RCU_NEXT_TAIL]],
".R"[rdp->nxttail[RCU_WAIT_TAIL] !=
rdp->nxttail[RCU_NEXT_READY_TAIL]],
".W"[rdp->nxttail[RCU_DONE_TAIL] !=
rdp->nxttail[RCU_WAIT_TAIL]],
".D"[&rdp->nxtlist != rdp->nxttail[RCU_DONE_TAIL]]);
#ifdef CONFIG_RCU_BOOST
seq_printf(m, ",%d,\"%c\"",
per_cpu(rcu_cpu_has_work, rdp->cpu),
convert_kthread_status(per_cpu(rcu_cpu_kthread_status,
rdp->cpu)));
#endif /* #ifdef CONFIG_RCU_BOOST */
seq_printf(m, ",%ld", rdp->blimit);
seq_printf(m, ",%lu,%lu,%lu\n",
rdp->n_cbs_invoked, rdp->n_cbs_orphaned, rdp->n_cbs_adopted);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
}
static int show_rcudata_csv(struct seq_file *m, void *unused)
{
rcu: Make hierarchical RCU less IPI-happy This patch fixes a hierarchical-RCU performance bug located by Anton Blanchard. The problem stems from a misguided attempt to provide a work-around for jiffies-counter failure. This work-around uses a per-CPU n_rcu_pending counter, which is incremented on each call to rcu_pending(), which in turn is called from each scheduling-clock interrupt. Each CPU then treats this counter as a surrogate for the jiffies counter, so that if the jiffies counter fails to advance, the per-CPU n_rcu_pending counter will cause RCU to invoke force_quiescent_state(), which in turn will (among other things) send resched IPIs to CPUs that have thus far failed to pass through an RCU quiescent state. Unfortunately, each CPU resets only its own counter after sending a batch of IPIs. This means that the other CPUs will also (needlessly) send -another- round of IPIs, for a full N-squared set of IPIs in the worst case every three scheduler-clock ticks until the grace period finally ends. It is not reasonable for a given CPU to reset each and every n_rcu_pending for all the other CPUs, so this patch instead simply disables the jiffies-counter "training wheels", thus eliminating the excessive IPIs. Note that the jiffies-counter IPIs do not have this problem due to the fact that the jiffies counter is global, so that the CPU sending the IPIs can easily reset things, thus preventing the other CPUs from sending redundant IPIs. Note also that the n_rcu_pending counter remains, as it will continue to be used for tracing. It may also see use to update the jiffies counter, should an appropriate kick-the-jiffies-counter API appear. Located-by: Anton Blanchard <anton@au1.ibm.com> Tested-by: Anton Blanchard <anton@au1.ibm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: anton@samba.org Cc: akpm@linux-foundation.org Cc: dipankar@in.ibm.com Cc: manfred@colorfullife.com Cc: cl@linux-foundation.org Cc: josht@linux.vnet.ibm.com Cc: schamp@sgi.com Cc: niv@us.ibm.com Cc: dvhltc@us.ibm.com Cc: ego@in.ibm.com Cc: laijs@cn.fujitsu.com Cc: rostedt@goodmis.org Cc: peterz@infradead.org Cc: penberg@cs.helsinki.fi Cc: andi@firstfloor.org Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> LKML-Reference: <12396834793575-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-04-14 04:31:16 +00:00
seq_puts(m, "\"CPU\",\"Online?\",\"c\",\"g\",\"pq\",\"pqc\",\"pq\",");
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
#ifdef CONFIG_NO_HZ
rcu: Decrease memory-barrier usage based on semi-formal proof (Note: this was reverted, and is now being re-applied in pieces, with this being the fifth and final piece. See below for the reason that it is now felt to be safe to re-apply this.) Commit d09b62d fixed grace-period synchronization, but left some smp_mb() invocations in rcu_process_callbacks() that are no longer needed, but sheer paranoia prevented them from being removed. This commit removes them and provides a proof of correctness in their absence. It also adds a memory barrier to rcu_report_qs_rsp() immediately before the update to rsp->completed in order to handle the theoretical possibility that the compiler or CPU might move massive quantities of code into a lock-based critical section. This also proves that the sheer paranoia was not entirely unjustified, at least from a theoretical point of view. In addition, the old dyntick-idle synchronization depended on the fact that grace periods were many milliseconds in duration, so that it could be assumed that no dyntick-idle CPU could reorder a memory reference across an entire grace period. Unfortunately for this design, the addition of expedited grace periods breaks this assumption, which has the unfortunate side-effect of requiring atomic operations in the functions that track dyntick-idle state for RCU. (There is some hope that the algorithms used in user-level RCU might be applied here, but some work is required to handle the NMIs that user-space applications can happily ignore. For the short term, better safe than sorry.) This proof assumes that neither compiler nor CPU will allow a lock acquisition and release to be reordered, as doing so can result in deadlock. The proof is as follows: 1. A given CPU declares a quiescent state under the protection of its leaf rcu_node's lock. 2. If there is more than one level of rcu_node hierarchy, the last CPU to declare a quiescent state will also acquire the ->lock of the next rcu_node up in the hierarchy, but only after releasing the lower level's lock. The acquisition of this lock clearly cannot occur prior to the acquisition of the leaf node's lock. 3. Step 2 repeats until we reach the root rcu_node structure. Please note again that only one lock is held at a time through this process. The acquisition of the root rcu_node's ->lock must occur after the release of that of the leaf rcu_node. 4. At this point, we set the ->completed field in the rcu_state structure in rcu_report_qs_rsp(). However, if the rcu_node hierarchy contains only one rcu_node, then in theory the code preceding the quiescent state could leak into the critical section. We therefore precede the update of ->completed with a memory barrier. All CPUs will therefore agree that any updates preceding any report of a quiescent state will have happened before the update of ->completed. 5. Regardless of whether a new grace period is needed, rcu_start_gp() will propagate the new value of ->completed to all of the leaf rcu_node structures, under the protection of each rcu_node's ->lock. If a new grace period is needed immediately, this propagation will occur in the same critical section that ->completed was set in, but courtesy of the memory barrier in #4 above, is still seen to follow any pre-quiescent-state activity. 6. When a given CPU invokes __rcu_process_gp_end(), it becomes aware of the end of the old grace period and therefore makes any RCU callbacks that were waiting on that grace period eligible for invocation. If this CPU is the same one that detected the end of the grace period, and if there is but a single rcu_node in the hierarchy, we will still be in the single critical section. In this case, the memory barrier in step #4 guarantees that all callbacks will be seen to execute after each CPU's quiescent state. On the other hand, if this is a different CPU, it will acquire the leaf rcu_node's ->lock, and will again be serialized after each CPU's quiescent state for the old grace period. On the strength of this proof, this commit therefore removes the memory barriers from rcu_process_callbacks() and adds one to rcu_report_qs_rsp(). The effect is to reduce the number of memory barriers by one and to reduce the frequency of execution from about once per scheduling tick per CPU to once per grace period. This was reverted do to hangs found during testing by Yinghai Lu and Ingo Molnar. Frederic Weisbecker supplied Yinghai with tracing that located the underlying problem, and Frederic also provided the fix. The underlying problem was that the HARDIRQ_ENTER() macro from lib/locking-selftest.c invoked irq_enter(), which in turn invokes rcu_irq_enter(), but HARDIRQ_EXIT() invoked __irq_exit(), which does not invoke rcu_irq_exit(). This situation resulted in calls to rcu_irq_enter() that were not balanced by the required calls to rcu_irq_exit(). Therefore, after these locking selftests completed, RCU's dyntick-idle nesting count was a large number (for example, 72), which caused RCU to to conclude that the affected CPU was not in dyntick-idle mode when in fact it was. RCU would therefore incorrectly wait for this dyntick-idle CPU, resulting in hangs. In contrast, with Frederic's patch, which replaces the irq_enter() in HARDIRQ_ENTER() with an __irq_enter(), these tests don't ever call either rcu_irq_enter() or rcu_irq_exit(), which works because the CPU running the test is already marked as not being in dyntick-idle mode. This means that the rcu_irq_enter() and rcu_irq_exit() calls and RCU then has no problem working out which CPUs are in dyntick-idle mode and which are not. The reason that the imbalance was not noticed before the barrier patch was applied is that the old implementation of rcu_enter_nohz() ignored the nesting depth. This could still result in delays, but much shorter ones. Whenever there was a delay, RCU would IPI the CPU with the unbalanced nesting level, which would eventually result in rcu_enter_nohz() being called, which in turn would force RCU to see that the CPU was in dyntick-idle mode. The reason that very few people noticed the problem is that the mismatched irq_enter() vs. __irq_exit() occured only when the kernel was built with CONFIG_DEBUG_LOCKING_API_SELFTESTS. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Josh Triplett <josh@joshtriplett.org>
2010-09-07 17:38:22 +00:00
seq_puts(m, "\"dt\",\"dt nesting\",\"dt NMI nesting\",\"df\",");
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
#endif /* #ifdef CONFIG_NO_HZ */
seq_puts(m, "\"of\",\"ri\",\"ql\",\"qs\"");
#ifdef CONFIG_RCU_BOOST
seq_puts(m, "\"kt\",\"ktl\"");
#endif /* #ifdef CONFIG_RCU_BOOST */
seq_puts(m, ",\"b\",\"ci\",\"co\",\"ca\"\n");
rcu: Merge preemptable-RCU functionality into hierarchical RCU Create a kernel/rcutree_plugin.h file that contains definitions for preemptable RCU (or, under the #else branch of the #ifdef, empty definitions for the classic non-preemptable semantics). These definitions fit into plugins defined in kernel/rcutree.c for this purpose. This variant of preemptable RCU uses a new algorithm whose read-side expense is roughly that of classic hierarchical RCU under CONFIG_PREEMPT. This new algorithm's update-side expense is similar to that of classic hierarchical RCU, and, in absence of read-side preemption or blocking, is exactly that of classic hierarchical RCU. Perhaps more important, this new algorithm has a much simpler implementation, saving well over 1,000 lines of code compared to mainline's implementation of preemptable RCU, which will hopefully be retired in favor of this new algorithm. The simplifications are obtained by maintaining per-task nesting state for running tasks, and using a simple lock-protected algorithm to handle accounting when tasks block within RCU read-side critical sections, making use of lessons learned while creating numerous user-level RCU implementations over the past 18 months. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: laijs@cn.fujitsu.com Cc: dipankar@in.ibm.com Cc: akpm@linux-foundation.org Cc: mathieu.desnoyers@polymtl.ca Cc: josht@linux.vnet.ibm.com Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org LKML-Reference: <12509746134003-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-22 20:56:52 +00:00
#ifdef CONFIG_TREE_PREEMPT_RCU
seq_puts(m, "\"rcu_preempt:\"\n");
PRINT_RCU_DATA(rcu_preempt_data, print_one_rcu_data_csv, m);
#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
seq_puts(m, "\"rcu_sched:\"\n");
PRINT_RCU_DATA(rcu_sched_data, print_one_rcu_data_csv, m);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
seq_puts(m, "\"rcu_bh:\"\n");
PRINT_RCU_DATA(rcu_bh_data, print_one_rcu_data_csv, m);
return 0;
}
static int rcudata_csv_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcudata_csv, NULL);
}
static const struct file_operations rcudata_csv_fops = {
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
.owner = THIS_MODULE,
.open = rcudata_csv_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
#ifdef CONFIG_RCU_BOOST
static void print_one_rcu_node_boost(struct seq_file *m, struct rcu_node *rnp)
{
seq_printf(m, "%d:%d tasks=%c%c%c%c kt=%c ntb=%lu neb=%lu nnb=%lu "
"j=%04x bt=%04x\n",
rnp->grplo, rnp->grphi,
"T."[list_empty(&rnp->blkd_tasks)],
"N."[!rnp->gp_tasks],
"E."[!rnp->exp_tasks],
"B."[!rnp->boost_tasks],
convert_kthread_status(rnp->boost_kthread_status),
rnp->n_tasks_boosted, rnp->n_exp_boosts,
rnp->n_normal_boosts,
(int)(jiffies & 0xffff),
(int)(rnp->boost_time & 0xffff));
seq_printf(m, "%s: nt=%lu egt=%lu bt=%lu nb=%lu ny=%lu nos=%lu\n",
" balk",
rnp->n_balk_blkd_tasks,
rnp->n_balk_exp_gp_tasks,
rnp->n_balk_boost_tasks,
rnp->n_balk_notblocked,
rnp->n_balk_notyet,
rnp->n_balk_nos);
}
static int show_rcu_node_boost(struct seq_file *m, void *unused)
{
struct rcu_node *rnp;
rcu_for_each_leaf_node(&rcu_preempt_state, rnp)
print_one_rcu_node_boost(m, rnp);
return 0;
}
static int rcu_node_boost_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcu_node_boost, NULL);
}
static const struct file_operations rcu_node_boost_fops = {
.owner = THIS_MODULE,
.open = rcu_node_boost_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
/*
* Create the rcuboost debugfs entry. Standard error return.
*/
static int rcu_boost_trace_create_file(struct dentry *rcudir)
{
return !debugfs_create_file("rcuboost", 0444, rcudir, NULL,
&rcu_node_boost_fops);
}
#else /* #ifdef CONFIG_RCU_BOOST */
static int rcu_boost_trace_create_file(struct dentry *rcudir)
{
return 0; /* There cannot be an error if we didn't create it! */
}
#endif /* #else #ifdef CONFIG_RCU_BOOST */
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
static void print_one_rcu_state(struct seq_file *m, struct rcu_state *rsp)
{
unsigned long gpnum;
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
int level = 0;
struct rcu_node *rnp;
gpnum = rsp->gpnum;
seq_printf(m, "c=%lu g=%lu s=%d jfq=%ld j=%x "
"nfqs=%lu/nfqsng=%lu(%lu) fqlh=%lu\n",
rsp->completed, gpnum, rsp->signaled,
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
(long)(rsp->jiffies_force_qs - jiffies),
(int)(jiffies & 0xffff),
rsp->n_force_qs, rsp->n_force_qs_ngp,
rsp->n_force_qs - rsp->n_force_qs_ngp,
rsp->n_force_qs_lh);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
for (rnp = &rsp->node[0]; rnp - &rsp->node[0] < NUM_RCU_NODES; rnp++) {
if (rnp->level != level) {
seq_puts(m, "\n");
level = rnp->level;
}
seq_printf(m, "%lx/%lx %c%c>%c %d:%d ^%d ",
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rnp->qsmask, rnp->qsmaskinit,
".G"[rnp->gp_tasks != NULL],
".E"[rnp->exp_tasks != NULL],
".T"[!list_empty(&rnp->blkd_tasks)],
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rnp->grplo, rnp->grphi, rnp->grpnum);
}
seq_puts(m, "\n");
}
static int show_rcuhier(struct seq_file *m, void *unused)
{
rcu: Merge preemptable-RCU functionality into hierarchical RCU Create a kernel/rcutree_plugin.h file that contains definitions for preemptable RCU (or, under the #else branch of the #ifdef, empty definitions for the classic non-preemptable semantics). These definitions fit into plugins defined in kernel/rcutree.c for this purpose. This variant of preemptable RCU uses a new algorithm whose read-side expense is roughly that of classic hierarchical RCU under CONFIG_PREEMPT. This new algorithm's update-side expense is similar to that of classic hierarchical RCU, and, in absence of read-side preemption or blocking, is exactly that of classic hierarchical RCU. Perhaps more important, this new algorithm has a much simpler implementation, saving well over 1,000 lines of code compared to mainline's implementation of preemptable RCU, which will hopefully be retired in favor of this new algorithm. The simplifications are obtained by maintaining per-task nesting state for running tasks, and using a simple lock-protected algorithm to handle accounting when tasks block within RCU read-side critical sections, making use of lessons learned while creating numerous user-level RCU implementations over the past 18 months. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: laijs@cn.fujitsu.com Cc: dipankar@in.ibm.com Cc: akpm@linux-foundation.org Cc: mathieu.desnoyers@polymtl.ca Cc: josht@linux.vnet.ibm.com Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org LKML-Reference: <12509746134003-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-22 20:56:52 +00:00
#ifdef CONFIG_TREE_PREEMPT_RCU
seq_puts(m, "rcu_preempt:\n");
print_one_rcu_state(m, &rcu_preempt_state);
#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
seq_puts(m, "rcu_sched:\n");
print_one_rcu_state(m, &rcu_sched_state);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
seq_puts(m, "rcu_bh:\n");
print_one_rcu_state(m, &rcu_bh_state);
return 0;
}
static int rcuhier_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcuhier, NULL);
}
static const struct file_operations rcuhier_fops = {
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
.owner = THIS_MODULE,
.open = rcuhier_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static void show_one_rcugp(struct seq_file *m, struct rcu_state *rsp)
{
unsigned long flags;
unsigned long completed;
unsigned long gpnum;
unsigned long gpage;
unsigned long gpmax;
struct rcu_node *rnp = &rsp->node[0];
raw_spin_lock_irqsave(&rnp->lock, flags);
completed = rsp->completed;
gpnum = rsp->gpnum;
if (rsp->completed == rsp->gpnum)
gpage = 0;
else
gpage = jiffies - rsp->gp_start;
gpmax = rsp->gp_max;
raw_spin_unlock_irqrestore(&rnp->lock, flags);
seq_printf(m, "%s: completed=%ld gpnum=%lu age=%ld max=%ld\n",
rsp->name, completed, gpnum, gpage, gpmax);
}
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
static int show_rcugp(struct seq_file *m, void *unused)
{
rcu: Merge preemptable-RCU functionality into hierarchical RCU Create a kernel/rcutree_plugin.h file that contains definitions for preemptable RCU (or, under the #else branch of the #ifdef, empty definitions for the classic non-preemptable semantics). These definitions fit into plugins defined in kernel/rcutree.c for this purpose. This variant of preemptable RCU uses a new algorithm whose read-side expense is roughly that of classic hierarchical RCU under CONFIG_PREEMPT. This new algorithm's update-side expense is similar to that of classic hierarchical RCU, and, in absence of read-side preemption or blocking, is exactly that of classic hierarchical RCU. Perhaps more important, this new algorithm has a much simpler implementation, saving well over 1,000 lines of code compared to mainline's implementation of preemptable RCU, which will hopefully be retired in favor of this new algorithm. The simplifications are obtained by maintaining per-task nesting state for running tasks, and using a simple lock-protected algorithm to handle accounting when tasks block within RCU read-side critical sections, making use of lessons learned while creating numerous user-level RCU implementations over the past 18 months. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: laijs@cn.fujitsu.com Cc: dipankar@in.ibm.com Cc: akpm@linux-foundation.org Cc: mathieu.desnoyers@polymtl.ca Cc: josht@linux.vnet.ibm.com Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org LKML-Reference: <12509746134003-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-22 20:56:52 +00:00
#ifdef CONFIG_TREE_PREEMPT_RCU
show_one_rcugp(m, &rcu_preempt_state);
rcu: Merge preemptable-RCU functionality into hierarchical RCU Create a kernel/rcutree_plugin.h file that contains definitions for preemptable RCU (or, under the #else branch of the #ifdef, empty definitions for the classic non-preemptable semantics). These definitions fit into plugins defined in kernel/rcutree.c for this purpose. This variant of preemptable RCU uses a new algorithm whose read-side expense is roughly that of classic hierarchical RCU under CONFIG_PREEMPT. This new algorithm's update-side expense is similar to that of classic hierarchical RCU, and, in absence of read-side preemption or blocking, is exactly that of classic hierarchical RCU. Perhaps more important, this new algorithm has a much simpler implementation, saving well over 1,000 lines of code compared to mainline's implementation of preemptable RCU, which will hopefully be retired in favor of this new algorithm. The simplifications are obtained by maintaining per-task nesting state for running tasks, and using a simple lock-protected algorithm to handle accounting when tasks block within RCU read-side critical sections, making use of lessons learned while creating numerous user-level RCU implementations over the past 18 months. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: laijs@cn.fujitsu.com Cc: dipankar@in.ibm.com Cc: akpm@linux-foundation.org Cc: mathieu.desnoyers@polymtl.ca Cc: josht@linux.vnet.ibm.com Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org LKML-Reference: <12509746134003-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-22 20:56:52 +00:00
#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
show_one_rcugp(m, &rcu_sched_state);
show_one_rcugp(m, &rcu_bh_state);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
return 0;
}
static int rcugp_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcugp, NULL);
}
static const struct file_operations rcugp_fops = {
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
.owner = THIS_MODULE,
.open = rcugp_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static void print_one_rcu_pending(struct seq_file *m, struct rcu_data *rdp)
{
seq_printf(m, "%3d%cnp=%ld "
"qsp=%ld rpq=%ld cbr=%ld cng=%ld "
"gpc=%ld gps=%ld nf=%ld nn=%ld\n",
rdp->cpu,
cpu_is_offline(rdp->cpu) ? '!' : ' ',
rdp->n_rcu_pending,
rdp->n_rp_qs_pending,
rdp->n_rp_report_qs,
rdp->n_rp_cb_ready,
rdp->n_rp_cpu_needs_gp,
rdp->n_rp_gp_completed,
rdp->n_rp_gp_started,
rdp->n_rp_need_fqs,
rdp->n_rp_need_nothing);
}
static void print_rcu_pendings(struct seq_file *m, struct rcu_state *rsp)
{
int cpu;
struct rcu_data *rdp;
for_each_possible_cpu(cpu) {
rdp = per_cpu_ptr(rsp->rda, cpu);
if (rdp->beenonline)
print_one_rcu_pending(m, rdp);
}
}
static int show_rcu_pending(struct seq_file *m, void *unused)
{
rcu: Merge preemptable-RCU functionality into hierarchical RCU Create a kernel/rcutree_plugin.h file that contains definitions for preemptable RCU (or, under the #else branch of the #ifdef, empty definitions for the classic non-preemptable semantics). These definitions fit into plugins defined in kernel/rcutree.c for this purpose. This variant of preemptable RCU uses a new algorithm whose read-side expense is roughly that of classic hierarchical RCU under CONFIG_PREEMPT. This new algorithm's update-side expense is similar to that of classic hierarchical RCU, and, in absence of read-side preemption or blocking, is exactly that of classic hierarchical RCU. Perhaps more important, this new algorithm has a much simpler implementation, saving well over 1,000 lines of code compared to mainline's implementation of preemptable RCU, which will hopefully be retired in favor of this new algorithm. The simplifications are obtained by maintaining per-task nesting state for running tasks, and using a simple lock-protected algorithm to handle accounting when tasks block within RCU read-side critical sections, making use of lessons learned while creating numerous user-level RCU implementations over the past 18 months. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: laijs@cn.fujitsu.com Cc: dipankar@in.ibm.com Cc: akpm@linux-foundation.org Cc: mathieu.desnoyers@polymtl.ca Cc: josht@linux.vnet.ibm.com Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org LKML-Reference: <12509746134003-git-send-email-> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-22 20:56:52 +00:00
#ifdef CONFIG_TREE_PREEMPT_RCU
seq_puts(m, "rcu_preempt:\n");
print_rcu_pendings(m, &rcu_preempt_state);
#endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
seq_puts(m, "rcu_sched:\n");
print_rcu_pendings(m, &rcu_sched_state);
seq_puts(m, "rcu_bh:\n");
print_rcu_pendings(m, &rcu_bh_state);
return 0;
}
static int rcu_pending_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcu_pending, NULL);
}
static const struct file_operations rcu_pending_fops = {
.owner = THIS_MODULE,
.open = rcu_pending_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static int show_rcutorture(struct seq_file *m, void *unused)
{
seq_printf(m, "rcutorture test sequence: %lu %s\n",
rcutorture_testseq >> 1,
(rcutorture_testseq & 0x1) ? "(test in progress)" : "");
seq_printf(m, "rcutorture update version number: %lu\n",
rcutorture_vernum);
return 0;
}
static int rcutorture_open(struct inode *inode, struct file *file)
{
return single_open(file, show_rcutorture, NULL);
}
static const struct file_operations rcutorture_fops = {
.owner = THIS_MODULE,
.open = rcutorture_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static struct dentry *rcudir;
static int __init rcutree_trace_init(void)
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
{
struct dentry *retval;
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
rcudir = debugfs_create_dir("rcu", NULL);
if (!rcudir)
goto free_out;
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
retval = debugfs_create_file("rcudata", 0444, rcudir,
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
NULL, &rcudata_fops);
if (!retval)
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
goto free_out;
retval = debugfs_create_file("rcudata.csv", 0444, rcudir,
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
NULL, &rcudata_csv_fops);
if (!retval)
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
goto free_out;
if (rcu_boost_trace_create_file(rcudir))
goto free_out;
retval = debugfs_create_file("rcugp", 0444, rcudir, NULL, &rcugp_fops);
if (!retval)
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
goto free_out;
retval = debugfs_create_file("rcuhier", 0444, rcudir,
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
NULL, &rcuhier_fops);
if (!retval)
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
goto free_out;
retval = debugfs_create_file("rcu_pending", 0444, rcudir,
NULL, &rcu_pending_fops);
if (!retval)
goto free_out;
retval = debugfs_create_file("rcutorture", 0444, rcudir,
NULL, &rcutorture_fops);
if (!retval)
goto free_out;
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
return 0;
free_out:
debugfs_remove_recursive(rcudir);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
return 1;
}
static void __exit rcutree_trace_cleanup(void)
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
{
debugfs_remove_recursive(rcudir);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
}
module_init(rcutree_trace_init);
module_exit(rcutree_trace_cleanup);
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 20:55:32 +00:00
MODULE_AUTHOR("Paul E. McKenney");
MODULE_DESCRIPTION("Read-Copy Update tracing for hierarchical implementation");
MODULE_LICENSE("GPL");