Fixed recovery of rport after race with dev_loss_tmo
Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Round 2 of Miscellaneous fixes:
- Ensure we don't prematurely re-enable IRQs in lpfc_sli_abort_fcp_cmpl().
- Prevent freeing of iocb after IOCB_TIMEDOUT error.
- Added code to cleanup REG_LOGIN mailbox command when a LOGO is received.
- Fix offline window where more work can sneak in after clearing work_ha
- Use target reset instead of LU reset in bus_device_reset_handler
- Fixed system hangs due to leaked host_lock.
- Fixed NULL pointer dereference during I/O with LIP.
- Fixed false iocb timeout.
- Fixed name server query response handling.
- Change rport dev_loss_tmo value when user change lpfc HBA's dev_loss_tmo.
- Fixed a memory leak in lpfc_sli_wake_mbox_wait.
- Fixed check for dropped frames.
- Removed hba queue depth calculation based on device PCI IDs
- Change min cr_count value specified in comment to agree with setting
Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Reorganize lpfc_nlp_list() and callers to prepare for nodelist simplification.
Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Modify ELS abort handling to prevent double completion
Rework portions of ELS abort handling to prevent double completion
- Rework ELS iotags and correct abort routine
- Move the (badly wrong) ELS completion logic from the initial ELS
abort request function to the ELS completion function.
- Fixup the iocb completion handling to account for the ELS abort
completions.
Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Misc bug fixes and code cleanup:
- Fix system hang while running on systems with IOMMU
- Fix use after free issues with rports
- Don't free mailbox structure if it's still on the mboxq list
- Decrement txq_cnt rather than txcmplq_cnt when parsing the txq list
- Use msleep for long delays to prevent soft lockup bug check
- Don't remove node during dev_loss_tmo if discovery is active
- Fix memory leaks in get/reset statistics and link attention paths
- Fixed lpfc_ns_rsp to handle entire GID_FT response.
- mbox interface should use MAILBOX_CMD_SIZE rather than sizeof(MAILBOX_t)
- Fixed bug check in add_timer.
- Fixup messages 0116, 0117, and 0128 to report ELS I/O tag.
- Remove unused parameter to lpfc_cleanup.
- Change mailbox timeout handling.
- Remove unused buflist. Code cleanup.
Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Use the sysfs configurable timeout when issuing a START_UNIT
command from the scsi error handler. This is needed for devices which
take longer than thirty seconds to respond to the start
unit. The problem was observed when sending a start unit
to a disk array device in an ipr RAID adapter, which results
in the adapter firmware sending potentially multiple commands
to physical devices as a result of this command, which ended
up timing out sometimes. This patch does not change the default
value used for this command.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Update the driver version reported by MEGAIOC_QDRVRVER to
match LSI_COMMON_MOD_VERSION.
Signed-off-by: David Milburn <dmilburn@redhat.com>
Acked-by: "Patro, Sumant" <Sumant.Patro@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Inspired somewhat by Vignesh Babu <vignesh.babu@wipro.com> patch to
dpt_i2o.c to replace kmalloc/memset sequences with kzalloc, doing the
same for the aacraid driver.
Signed-off-by: Mark Salyzyn <aacraid@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
It still used the long-deprecated "pci_module_init()" interface, rather
than the proper "pci_register_driver()" one.
[ I don't have the hardware, and I doubt many do, but the fix is
trivial and obvious, and can't be worse than not compiling ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/pci-2.6: (59 commits)
PCI: Free resource files in error path of pci_create_sysfs_dev_files()
pci-quirks: disable MSI on RS400-200 and RS480
PCI hotplug: Use menuconfig objects
PCI: ZT5550 CPCI Hotplug driver fix
PCI: rpaphp: Remove semaphores
PCI: rpaphp: Ensure more pcibios_add/pcibios_remove symmetry
PCI: rpaphp: Use pcibios_remove_pci_devices() symmetrically
PCI: rpaphp: Document is_php_dn()
PCI: rpaphp: Document find_php_slot()
PCI: rpaphp: Rename rpaphp_register_pci_slot() to rpaphp_enable_slot()
PCI: rpaphp: refactor tail call to rpaphp_register_slot()
PCI: rpaphp: remove rpaphp_set_attention_status()
PCI: rpaphp: remove print_slot_pci_funcs()
PCI: rpaphp: Remove setup_pci_slot()
PCI: rpaphp: remove a call that does nothing but a pointer lookup
PCI: rpaphp: Remove another wrappered function
PCI: rpaphp: Remve another call that is a wrapper
PCI: rpaphp: remove a function that does nothing but wrap debug printks
PCI: rpaphp: Remove un-needed goto
PCI: rpaphp: Fix a memleak; slot->location string was never freed
...
SCSI should be working on a TT (but someone should really try!) but causes
trouble on a Falcon (as in: it ate a filesystem of mine) at least when
used concurrently with IDE. I have the notion it's because locking of the
ST-DMA interrupt by IDE is broken in 2.6 (the IDE driver always complains
about trying to release an already-released ST-DMA). Needs more work, but
that's on the IDE or m68k interrupt side rather than SCSI.
Signed-off-by: Michael Schmitz <schmitz@debian.org>
Signed-off-by: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Convert code that allocs a struct pci_dev to use alloc_pci_dev().
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
I noticed that many source files include <linux/pci.h> while they do
not appear to need it. Here is an attempt to clean it all up.
In order to find all possibly affected files, I searched for all
files including <linux/pci.h> but without any other occurence of "pci"
or "PCI". I removed the include statement from all of these, then I
compiled an allmodconfig kernel on both i386 and x86_64 and fixed the
false positives manually.
My tests covered 66% of the affected files, so there could be false
positives remaining. Untested files are:
arch/alpha/kernel/err_common.c
arch/alpha/kernel/err_ev6.c
arch/alpha/kernel/err_ev7.c
arch/ia64/sn/kernel/huberror.c
arch/ia64/sn/kernel/xpnet.c
arch/m68knommu/kernel/dma.c
arch/mips/lib/iomap.c
arch/powerpc/platforms/pseries/ras.c
arch/ppc/8260_io/enet.c
arch/ppc/8260_io/fcc_enet.c
arch/ppc/8xx_io/enet.c
arch/ppc/syslib/ppc4xx_sgdma.c
arch/sh64/mach-cayman/iomap.c
arch/xtensa/kernel/xtensa_ksyms.c
arch/xtensa/platform-iss/setup.c
drivers/i2c/busses/i2c-at91.c
drivers/i2c/busses/i2c-mpc.c
drivers/media/video/saa711x.c
drivers/misc/hdpuftrs/hdpu_cpustate.c
drivers/misc/hdpuftrs/hdpu_nexus.c
drivers/net/au1000_eth.c
drivers/net/fec_8xx/fec_main.c
drivers/net/fec_8xx/fec_mii.c
drivers/net/fs_enet/fs_enet-main.c
drivers/net/fs_enet/mac-fcc.c
drivers/net/fs_enet/mac-fec.c
drivers/net/fs_enet/mac-scc.c
drivers/net/fs_enet/mii-bitbang.c
drivers/net/fs_enet/mii-fec.c
drivers/net/ibm_emac/ibm_emac_core.c
drivers/net/lasi_82596.c
drivers/parisc/hppb.c
drivers/sbus/sbus.c
drivers/video/g364fb.c
drivers/video/platinumfb.c
drivers/video/stifb.c
drivers/video/valkyriefb.c
include/asm-arm/arch-ixp4xx/dma.h
sound/oss/au1550_ac97.c
I would welcome test reports for these files. I am fine with removing
the untested files from the patch if the general opinion is that these
changes aren't safe. The tested part would still be nice to have.
Note that this patch depends on another header fixup patch I submitted
to LKML yesterday:
[PATCH] scatterlist.h needs types.h
http://lkml.org/lkml/2007/3/01/141
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* 'for-linus' of git://git.kernel.dk/data/git/linux-2.6-block:
[PATCH] elevator: elv_list_lock does not need irq disabling
[BLOCK] Don't pin lots of memory in mempools
cfq-iosched: speedup cic rb lookup
ll_rw_blk: add io_context private pointer
cfq-iosched: get rid of cfqq hash
cfq-iosched: tighten queue request overlap condition
cfq-iosched: improve sync vs async workloads
cfq-iosched: never allow an async queue idling
cfq-iosched: get rid of ->dispatch_slice
cfq-iosched: don't pass unused preemption variable around
cfq-iosched: get rid of ->cur_rr and ->cfq_list
cfq-iosched: slice offset should take ioprio into account
[PATCH] cfq-iosched: style cleanups and comments
cfq-iosched: sort IDLE queues into the rbtree
cfq-iosched: sort RT queues into the rbtree
[PATCH] cfq-iosched: speed up rbtree handling
cfq-iosched: rework the whole round-robin list concept
cfq-iosched: minor updates
cfq-iosched: development update
cfq-iosched: improve preemption for cooperating tasks
Some sparc64 boxes don't have a valid NVRAM (from which the driver
takes its WWPN) try to extract this from open firmware instead and if
that fails, fall back to a default, which would be invalid if more
than one machine on the same SAN does this, since two machines with
the same WWPN would be illegal, so warn when taking this potentially
invalid default.
Tested on SunBlade-1000:
Acked-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Currently we scale the mempool sizes depending on memory installed
in the machine, except for the bio pool itself which sits at a fixed
256 entry pre-allocation.
There's really no point in "optimizing" this OOM path, we just need
enough preallocated to make progress. A single unit is enough, lets
scale it down to 2 just to be on the safe side.
This patch saves ~150kb of pinned kernel memory on a 32-bit box.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
drivers/scsi/ipr.c: In function '__ipr_eh_dev_reset':
drivers/scsi/ipr.c:3865: warning: passing argument 4 of 'ata_do_eh' from incompatible pointer type
Cc: Tejun Heo <htejun@gmail.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
irqreturn.h for irqreturn_t and dma_addr_t being u128 warnings ;-)
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
SCSI was using the incorrect lock to protect walking the list of all
devices in the class. This patch fixes this.
Cc: James Bottomley <James.Bottomley@SteelEye.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6: (67 commits)
[SCSI] SUNESP: Complete driver rewrite to version 2.0
[SPARC64]: Convert PCI over to generic struct iommu/strbuf.
[SPARC]: device_node name constification fallout
[SPARC64]: Convert SBUS over to generic iommu/strbuf structs.
[SPARC64]: Add generic iommu and strbuf structs to iommu.h
[SPARC64]: Consolidate {sbus,pci}_iommu_arena.
[SPARC]: Make device_node name and type const
[SPARC64]: constify some paramaters of OF routines
[TIGON3]: of_get_property() returns const.
[SPARC64]: Fix PCI rework to adhere to of_get_property() const return.
[SPARC64]: Document and fix calculation of pages_avail.
[SPARC64]: Make sure pbm->prom_node is setup easly enough in psycho.c
[SPARC64]: Use bootmem_bootmap_pages() in choose_bootmap_pfn().
[SPARC64]: Add proper header file extern for cmdline_memory_size.
[SPARC64]: Kill sparc_ultra_dump_{i,d}tlb()
[SPARC64]: Use DECLARE_BITMAP and BITS_TO_LONGS in mm/init.c
[SPARC64]: Give move verbose show_mem() output just like i386.
[SPARC64]: Mark show_mem() printk's with KERN_INFO.
[SPARC64]: Kill kvaddr_to_phys() and friends.
[SPARC64]: Privatize sun4u_get_pte() and fix name.
...
Major features:
1) Tagged queuing support.
2) Will properly negotiate for synchronous transfers even on
devices that reject the wide negotiation message, such as
CDROMs
3) Significantly lower kernel stack usage in interrupt
handler path by elimination of function vector arrays,
replaced by a top-level switch statement state machine.
4) Uses generic scsi infrastructure as much as possible to
avoid code duplication.
5) Automatic request of sense data in response to CHECK_CONDITION
6) Portable to other platforms using ESP such as DEC and Sun3
systems.
Signed-off-by: David S. Miller <davem@davemloft.net>
The only unfortunate bit here is that the name field of struct map_info
is not const, so for now we put a cast on the assignment of it.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Switch cb_lock to mutex and allow netlink kernel users to override it
with a subsystem specific mutex for consistent locking in dump callbacks.
All netlink_dump_start users have been audited not to rely on any
side-effects of the previously used spinlock.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the common "(struct nlmsghdr *)skb->data" sequence, so that we reduce the
number of direct accesses to skb->data and for consistency with all the other
cast skb member helpers.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The following patch adds support for sysfs/uevent modalias
attribute for scsi devices (like disks, tapes, cdroms etc),
based on whatever current sd.c, sr.c, st.c and osst.c drivers
supports.
The modalias format is like this:
scsi:type-0x04
(for TYPE_WORM, handled by sr.c now).
Several comments.
o This hexadecimal type value is because all TYPE_XXX constants
in include/scsi/scsi.h are given in hex, but __stringify() will
not convert them to decimal (so it will NOT be scsi:type-4).
Since it does not really matter in which format it is, while
both modalias in module and modalias attribute match each other,
I descided to go for that 0x%02x format (and added a comment in
include/scsi/scsi.h to keep them that way), instead of changing
them all to decimal.
o There was no .uevent routine for SCSI bus. It might be a good
idea to add some more ueven environment variables in there.
o osst.c driver handles tapes too, like st.c, but only SOME tapes.
With this setup, hotplug scripts (or whatever is used by the
user) will try to load both st and osst modules for all SCSI
tapes found, because both modules have scsi:type-0x01 alias).
It is not harmful, but one extra module is no good either.
It is possible to solve this, by exporting more info in
modalias attribute, including vendor and device identification
strings, so that modalias becomes something like
scsi:type-0x12:vendor-Adaptec LTD:device-OnStream Tape Drive
and having that, match for all 3 attributes, not only device
type. But oh well, vendor and device strings may be large,
and they do contain spaces and whatnot.
So I left them for now, awaiting for comments first.
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This patch (as857) modifies the SG_GET_RESERVED_SIZE and
SG_SET_RESERVED_SIZE ioctls in the sg driver, capping the values at
the device's request_queue's max_sectors value. This will permit
cdrecord to obtain a legal value for the maximum transfer length,
fixing Bugzilla #7026.
The patch also caps the initial reserved_size value. There's no
reason to have a reserved buffer larger than max_sectors, since it
would be impossible to use the extra space.
The corresponding ioctls in the block layer are modified similarly,
and the initial value for the reserved_size is set as large as
possible. This will effectively make it default to max_sectors.
Note that the actual value is meaningless anyway, since block devices
don't have a reserved buffer.
Finally, the BLKSECTGET ioctl is added to sg, so that there will be a
uniform way for users to determine the actual max_sectors value for
any raw SCSI transport.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Acked-by: Douglas Gilbert <dougg@torque.net>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
I got so sick of seing the check_region warnings from BusLogic.c I actually
fixed it properly. Never use check region, reserve it before the probe
with request region instead and check the error result; free region if
setup fails. Should be functionally identical to the original except for
fixing the potential race.
Signed-off-by: Zachary Amsden <zach@vmware.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
- ibmstgt hits the max rdma transfer size (128k).
- libsrp returns a rdma transfer error properly.
Signed-off-by: Bastian Blank <bastian@waldi.eu.org>
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Thanks for the help from Steve Fox and Duane Cox investigating this
issue, I'd like to report that we found the problem. The issue is with
the patch Steve Fox isolated below, by not accommodating older adapters
properly and issuing a command they do not support when retrieving
storage parameters about the arrays. This simple patch resolves the
problem (and more accurately mimics the logic of the original code
before the patch).
Signed-off-by: Mark Salyzyn <aacraid@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Currently, the scsi error handler will issue a START_UNIT
command if the drive indicates it needs its motor started
and the allow_restart flag is set in the scsi_device. If,
after the scsi error handler invokes a host adapter reset
due to error recovery, a device is in a unit attention
state AND also needs a START_UNIT, that device will be placed
offline. The disk array devices on an ipr RAID adapter
will do exactly this when in a dual initiator configuration.
This patch adds a single retry to the EH initiated
START_UNIT.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Patch modified and
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
We only map DMA when cmd->request_bufflen is non-zero for non-sg
buffers, we thus should make the same check when unmapping.
Based upon a report from Pasi Pirhonen.
Signed-off-by: David S. Miller <davem@davemloft.net>
3w-xxxx emulates a REQUEST_SENSE response by simply returning nothing.
Unfortunately, it's assuming that the REQUEST_SENSE command is
implemented with use_sg == 0, which is no longer the case. The oops
occurs because it's clearing the scatterlist in request_buffer instead
of the memory region.
This is fixed by using tw_transfer_internal() to transfer correctly to
the scatterlist.
Acked-by: adam radford <aradford@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This fixes a regression caused by commit:
2dc611de5a
The sense buffer code in scsi_send_eh_cmnd was changed to use
alloc_page() and a scatter list, but the sense data copy was not
updated to match so what we actually get in the sense buffer is total
grabage starting with the kernel address of the struct page we got.
Basically the stack frame of scsi_send_eh_cmd() is what ends up
in the sense buffer.
Depending upon how pointers look on a given platform, you can
end up getting sr_ioctl.c errors when you mount a cdrom. If
the CDROM gives a check condition for GPCMD_GET_CONFIGURATION issued
by drivers/cdrom/cdrom.c:cdrom_mmc_profile(), sr_ioctl will
spit out this error message in sr_do_ioctl() with the way pointers
are on sparc64:
default:
printk(KERN_ERR "%s: CDROM (ioctl) error, command: ", cd->cdi.name);
__scsi_print_command(cgc->cmd);
scsi_print_sense_hdr("sr", &sshdr);
err = -EIO;
This is the error Tom Callaway reported in:
http://marc.info/?l=linux-sparc&m=117407453208101&w=2
Anyways, fix this by using page_address(sgl.page) which is OK
because we know this is low-mem due to GFP_ATOMIC.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Christoph Hellwig <hch@lst.de>
Attached is the patch I feel will address this issue. As an added
'perk' I have also added the code to detect if the controller was
previously initialized for interrupted operations by ANY operating
system should the reset_devices kernel parameter not be set and we are
dealing with a naïve kexec without the addition of this kernel
parameter. The reset handler is also improved. Related to reset
operations, but not pertinent specifically to this issue, I have also
altered the handling somewhat so that we reset the adapter if we feel
it is taking too long (three minutes) to start up.
We have not unit tested the reset_devices flag propagation to this
driver code, nor have we unit tested the check for the interrupted
operations under the conditions of a naively issued kexec. We are
submitting this modified driver to our Q/A department for integration
testing in our current programs. I would appreciate an ACK to this
patch should it resolve the issue described in this thread...
Signed-off-by: Mark Salyzyn <aacraid@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>