If a Virtual I/O server is rebooted, the client fibre channel sees a
transport event on its CRQ, which causes it to attempt to reconnect
to the CRQ. For a period of time during the VIOS reboot, the client's
attempts to register the CRQ will return H_CLOSED, indicating the server
side is not currently registered. The ibmvfc driver was not handling
this well and was taking the virtual adapter offline. Fix this by
re-enabling our interrupt and waiting for the event on our CRQ
indicating the server is back, at which point we can reconnect.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This removes the driver's get_host_def_dev_loss_tmo
callback and just has the driver set the dev loss
using the fc class fc_host_dev_loss_tmo macro like is
done for other fc params.
This patch also removes the module dev loss param.
To override the value the fc host sysfs value being
added in the fc class patch can be used instead of the driver
module param.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
By default, ibmvfc does not log any async events in order
to avoid flooding the log with them. Improve on this by
logging by default events that are not likely to flood the
log, such as link up/down. Having these events in the log
will improve the ability to debug issues with ibmvfc.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This fixes a bug where the driver was resetting the
rport dev_loss_tmo when devices were added by adding
support for the get_host_def_dev_loss_tmo callout.
Patch has only been compile tested.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The ibmvfc driver was incorrectly obtaining a scsi_target pointer
from an fc_rport. The way it is coded ensures that ibmvfc's
terminate_rport_io handler does absolutely nothing. Fix this up
to iterate through affected devices differently, sending cancel
and abort task set as appropriate. Without this patch,
fast_io_fail_tmo is broken for ibmvfc.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Commit 43c8da907c introduced a race
condition which can occur when adding/deleting rports. There are
two possible threads now that can be deleting rports in the ibmvfc
driver, which can result in list_del being called twice, resulting
in an oops. This patch adds a new state to the ibmvfc_target struct
to indicate the target has been removed from the list and is in
the process of being deleted.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Add the __init and __exit macros to the module_init / module_exit
functions from drivers/scsi/ibmvscsi/ibmvstgt.c
Signed-off-by: Peter Huewe <peterhuewe@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
If we encounter an error when sending a management datagram (i.e. non
SCSI command, such as virtual adapter initialization command), we
end up incrementing the request_limit, even though we don't decrement
it for these commands. Fix this up by doing this increment in
the error path for SRP commands only.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Fixes a deadlock that can occur if we hit a command timeout
during the virtual adapter initialization. The event done
functions are written with the assumption that no locks are held,
however, when purging requests this is not true. Fix up the
purge function to drop the lock so that the done function
is not called with the lock held, which can cause a deadlock.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This fixes a softlockup seen on resume. During resume, the CRQ
must be reenabled. However, the H_ENABLE_CRQ hcall used to do
this may return H_BUSY or H_LONG_BUSY. When this happens, the
caller is expected to retry later. This patch changes a simple
loop, which was causing the softlockup, to a loop at task level
which sleeps between retries rather than simply spinning.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Adds support for fc_block_scsi_eh to block the EH handlers if
the target device is in the blocked state to ensure we don't
take devices offline.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This fixes a softlockup seen on resume. During resume, the CRQ
must be reenabled. However, the H_ENABLE_CRQ hcall used to do
this may return H_BUSY or H_LONG_BUSY. When this happens, the
caller is expected to retry later. Normally the H_ENABLE_CRQ
succeeds relatively soon. However, we have seen cases where
this can take long enough to see softlockup warnings.
This patch changes a simple loop, which was causing the
softlockup, to a loop at task level which sleeps between
retries rather than simply spinning.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
A driver needs to be ready to take an interrupt as soon as it registers
an interrupt handler. I noticed the following oops when testing kdump:
ipr: IBM Power RAID SCSI Device Driver version: 2.5.0 (February 11, 2010)
ibmvscsi 30000002: SRP_VERSION: 16.a
ibmvscsi 30000002: SRP_VERSION: 16.a
Unable to handle kernel paging request for data at address 0x00000000
...
pc: c000000004085e34: .tasklet_action+0xf4/0x1dc
...
c000000004086fe4 .__do_softirq+0x16c/0x2c0
c00000000403138c .call_do_softirq+0x14/0x24
c00000000400ee14 .do_softirq+0xa0/0x104
c00000000408690c .irq_exit+0x70/0xd0
c00000000400f190 .do_IRQ+0x214/0x2a8
c000000004004804 hardware_interrupt_entry+0x1c/0x98
--- Exception: 501 (Hardware Interrupt) at c00000000400c544 .raw_local_irq_restore+0x48/0x54
c00000000465d2a8 ._raw_spin_unlock_irqrestore+0x74/0xa0
c0000000040e7f00 .__setup_irq+0x2ec/0x3f0
c0000000040e8198 .request_threaded_irq+0x194/0x22c
c00000000446d854 .rpavscsi_init_crq_queue+0x284/0x3f0
c00000000446c764 .ibmvscsi_probe+0x688/0x710
c00000000402903c .vio_bus_probe+0x37c/0x3e4
c000000004403f10 .driver_probe_device+0xec/0x1b8
c000000004404088 .__driver_attach+0xac/0xf4
c000000004403184 .bus_for_each_dev+0x98/0x104
c000000004403c98 .driver_attach+0x40/0x60
c0000000044026f0 .bus_add_driver+0x154/0x324
c0000000044045d0 .driver_register+0xe8/0x1ac
c00000000402b2a8 .vio_register_driver+0x54/0x74
c000000004933ea4 .ibmvscsi_module_init+0x80/0xc0
c000000004009834 .do_one_initcall+0x98/0x1d8
c0000000049005b4 .kernel_init+0x27c/0x33c
c000000004031550 .kernel_thread+0x54/0x70
srp_task needs to be setup before request_irq. The patch below fixes the oops.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Merging in current state of Linus' tree to deal with merge conflicts and
build failures in vio.c after merge.
Conflicts:
drivers/i2c/busses/i2c-cpm.c
drivers/i2c/busses/i2c-mpc.c
drivers/net/gianfar.c
Also fixed up one line in arch/powerpc/kernel/vio.c to use the
correct node pointer.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
This allows bin_attr->read,write,mmap callbacks to check file specific data
(such as inode owner) as part of any privilege validation.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The following structure elements duplicate the information in
'struct device.of_node' and so are being eliminated. This patch
makes all readers of these elements use device.of_node instead.
(struct of_device *)->node
(struct dev_archdata *)->prom_node (sparc)
(struct dev_archdata *)->of_node (powerpc & microblaze)
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
If a command times out resulting in EH getting invoked, we wait for the
aborted commands to come back after sending the abort. Shorten
the amount of time we wait for these responses, to ensure we don't
get stuck in EH for several minutes.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Commands which are completed by the VIOS are placed on a CRQ
in kernel memory for the ibmvfc driver to process. Each CRQ
entry is 16 bytes. The ibmvfc driver reads the first 8 bytes
to check if the entry is valid, then reads the next 8 bytes to get
the handle, which is a pointer the completed command. This fixes
an issue seen on Power 7 where the processor reordered the
loads from memory, resulting in processing command completion
with a stale handle. This could result in command timeouts,
and also early completion of commands.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
ibmvscsi uses dma_unmap_single() for buffers mapped via
dma_map_sg(). It works however it's the API violation. The DMA debug
facility complains about it:
http://marc.info/?l=linux-scsi&m=127018555013151&w=2
Reported-by: Sachin Sant <sachinp@in.ibm.com>
Tested-by: Sachin Sant <sachinp@in.ibm.com>
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Adds support for resuming from suspend for IBM VFC devices. We may have
lost an interrupt over the suspend, so we just kick the interrupt handler
to process anything that is outstanding. We expect to find a transport event
indicating we need to reestablish our CRQ.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Adds support for resuming from suspend for IBM VSCSI devices. We may have
lost an interrupt over the suspend, so we just kick the interrupt handler
to process anything that is outstanding. We expect to find a transport event
indicating we need to reestablish our CRQ.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
* 'for-2.6.34' of git://git.kernel.dk/linux-2.6-block: (38 commits)
block: don't access jiffies when initialising io_context
cfq: remove 8 bytes of padding from cfq_rb_root on 64 bit builds
block: fix for "Consolidate phys_segment and hw_segment limits"
cfq-iosched: quantum check tweak
blktrace: perform cleanup after setup error
blkdev: fix merge_bvec_fn return value checks
cfq-iosched: requests "in flight" vs "in driver" clarification
cciss: Fix problem with scatter gather elements in the scsi half of the driver
cciss: eliminate unnecessary pointer use in cciss scsi code
cciss: do not use void pointer for scsi hba data
cciss: factor out scatter gather chain block mapping code
cciss: fix scatter gather chain block dma direction kludge
cciss: simplify scatter gather code
cciss: factor out scatter gather chain block allocation and freeing
cciss: detect bad alignment of scsi commands at build time
cciss: clarify command list padding calculation
cfq-iosched: rethink seeky detection for SSDs
cfq-iosched: rework seeky detection
block: remove padding from io_context on 64bit builds
block: Consolidate phys_segment and hw_segment limits
...
Except for SCSI no device drivers distinguish between physical and
hardware segment limits. Consolidate the two into a single segment
limit.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Bart Van Assche <bart.vanassche@gmail.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This patch modifies scsi_host_template->change_queue_depth so that
it takes an argument indicating why it is being called. This will be
used so that if a LLD needs to do some extra processing when
handling queue fulls or later ramp ups, it can do so.
This is a simple port of the drivers setting a change_queue_depth
callback. In the patch I just have these LLDs adjust the queue depth
if the user was requesting it.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
[Vasu.Dev: v2
Also converted pmcraid_change_queue_depth and then verified
all modules compile using "make allmodconfig" for any new build
warnings on X86_64.
Updated original description after combing two original
patches from Mike to make this patch git bisectable.]
Signed-off-by: Vasu Dev <vasu.dev@intel.com>
[jejb: fixed up 53c700]
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
When issuing a Cancel to the virtual fibre channel adapter,
the interface specifies a flags field for the client to indicate
what kind of error recovery is being performed. Fix up these
flags for terminate_rport_io to indicate an abort task set
rather than a target reset.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Remove a parameter to ibmvfc_init_host which is always set to
zero by all callers.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Need to grab the host lock around the call to ibmvfc_link_down.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
When processing the response to either a LUN reset,
target reset, or an abort task set, the ibmvfc driver needs to
treat as success receiving a response with a non-zero
status in the response IU along with a general transport
error with the FCP response code being zero. The VIOS
currently guarantees this cannot happen, but a future version
of VIOS may allow this to be returned, so ensure we handle
this response combination correctly for TMFs, as we already
do for SCSI commands.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
fix the following 'make includecheck' warning:
drivers/scsi/ibmvscsi/ibmvscsi.c: asm/firmware.h is included more than once.
Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
LKML-Reference: <1247067016.4382.78.camel@ht.satnam>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
The allocated struct is manually zeroed after allocation, so avoid using
the (broken) kzalloc mempool (which does not re-zero previously used items
when they are returned to the pool).
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Fixes a regression seen in the ibmvscsi driver when using the VSCSI
server in SLES 9 and SLES 10. The VSCSI server in these releases
has a bug in it in which it does not send responses to unknown MADs.
Check the OS Type field in the adapter info response and do not send
these unsupported commands when talking to an older server.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Fixes a problem seen where sending a PRLI to a target
resulted in it sending a LOGO. This caused the ibmvfc
driver to go back through discovery again, which caused
another PRLI attempt, which caused another LOGO. Fix this
behavior by ignoring LOGO if we haven't even logged into
the target yet.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Since async events could indicate changes to link status, or
events which could affect decisions made during discovery, we should
process async events prior to command completion responses.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
In the near future, the driver core is going to not allow direct access
to the driver_data pointer in struct device. Instead, the functions
dev_get_drvdata() and dev_set_drvdata() should be used. These functions
have been around since the beginning, so are backwards compatible with
all older kernel versions.
Cc: linux-scsi@vger.kernel.org
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Brian King <brking@linux.vnet.ibm.com>
Cc: Robert Jennings <rcj@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add support to ibmvscsi for the capabilities MAD. This command gets sent
to the Virtual I/O server prior to login in order to communicate client
capabilities. Additionally it returns information regarding capabilities
that the server supports. The two main capabilities communicated in this
MAD are related to partition migration and client reserve. Client reserve
allows for SCSI-2 reservations to be sent to virtual disks which are backed
by physical LUNs and will result in the reservation being sent to the
physical LUN.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
A new mode of error reporting, fast fail, has been added to the VIOS
which allows failover to happen more quickly.
If this new fast fail mode is enabled on the VIOS and the vSCSI client
supports the mode, the VIOS will not return MEDIUM error on path failures,
but rather return VIOSRP_ADAPTER_FAIL in the crq response, which
ibmvscsi will translate to DID_ERROR.
This new mode can be enabled for single path configurations as well,
so it is the new default error reporting mode. A module parameter is
provided to disable this new behavior on the off chance it causes a
problem on some old VIOS version.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
The ibmvscsi driver currently sends the SRP Login before sending the Adapter
Info MAD, which can result in commands getting sent to the virtual adapter
before we are ready for them. This results in a slight window where the target
devices may not behave as expected. Change the order and close the window.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Previously we had one timeout that was used for all types of operations.
This adds specific timeout values for different operations (init, login,
adapter info MAD, abort task, and LUN reset).
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Adds support for 16 byte CDBs to the ibmvscsi driver.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
There are several scenarios where the ibmvfc driver needs to
try to log back into a target on the fabric. Today when these events
occur, we simply go through re-discovery for all attached targets,
assuming that either the query of the name server or an ADISC will
indicate we might need to log back into the target, which doesn't
work for all scenarios. Fix this by taking note of the affected target(s)
in these conditions and ensuring we try to PLOGI back into the target.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
For certain scenarios during device rediscovery, we detect we need
to log back into a target. Currently we do just that - PLOGI/PRLI
back into the target. Change the code to delete and add the target
from the FC transport layer as well, to ensure we handle any cases
where the target may have changed.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
