commit 2a1b7e575b upstream.
I may have an explanation for the LSI 1068 HBA hangs provoked by ATA
pass-through commands, in particular by smartctl.
First, my version of the symptoms. On an LSI SAS1068E B3 HBA running
01.29.00.00 firmware, with SATA disks, and with smartd running, I'm seeing
occasional task, bus, and host resets, some of which lead to hard faults of
the HBA requiring a reboot. Abusively looping the smartctl command,
# while true; do smartctl -a /dev/sdb > /dev/null; done
dramatically increases the frequency of these failures to nearly one per
minute. A high IO load through the HBA while looping smartctl seems to
improve the chance of a full scsi host reset or a non-recoverable hang.
I reduced what smartctl was doing down to a simple test case which
causes the hang with a single IO when pointed at the sd interface. See
the code at the bottom of this e-mail. It uses an SG_IO ioctl to issue
a single pass-through ATA identify device command. If the buffer
userspace gives for the read data has certain alignments, the task is
issued to the HBA but the HBA fails to respond. If run against the sg
interface, neither the test code nor smartctl causes a hang.
sd and sg handle the SG_IO ioctl slightly differently. Unless you
specifically set a flag to do direct IO, sg passes a buffer of its own,
which is page-aligned, to the block layer and later copies the result
into the userspace buffer regardless of its alignment. sd, on the other
hand, always does direct IO unless the userspace buffer fails an
alignment test at block/blk-map.c line 57, in which case a page-aligned
buffer is created and used for the transfer.
The alignment test currently checks for word-alignment, the default
setup by scsi_lib.c; therefore, userspace buffers of almost any
alignment are given directly to the HBA as DMA targets. The LSI 1068
hardware doesn't seem to like at least a couple of the alignments which
cross a page boundary (see the test code below). Curiously, many
page-boundary-crossing alignments do work just fine.
So, either the hardware has an bug handling certain alignments or the
hardware has a stricter alignment requirement than the driver is
advertising. If stricter alignment is required, then in no case should
misaligned buffers from userspace be allowed through without being
bounced or at least causing an error to be returned.
It seems the mptsas driver could use blk_queue_dma_alignment() to advertise
a stricter alignment requirement. If it does, sd does the right thing and
bounces misaligned buffers (see block/blk-map.c line 57). The following
patch to 2.6.34-rc5 makes my symptoms go away. I'm sure this is the wrong
place for this code, but it gets my idea across.
Acked-by: Kashyap Desai <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 9b53b39243 upstream.
Errata:
Certain conditions on the scsi bus may casue the 53C1030 to incorrectly signal
a SCSI_DATA_UNDERRUN to the host.
Workaround 1:
For an Errata on LSI53C1030 When the length of request data
and transfer data are different with result of command (READ or VERIFY),
DID_SOFT_ERROR is set.
Workaround 2:
For potential trouble on LSI53C1030. It is checked whether the length of
request data is equal to the length of transfer and residual.
MEDIUM_ERROR is set by incorrect data.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Cc: maximilian attems <max@stro.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 9858ae3801 upstream.
retval should be SUCCESS/FAILED which is defined at scsi.h
retval = 0 is directing wrong return value. It must be retval = SUCCESS.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit f1053a7ca9 upstream.
Since commit 9d2e9d66a3
mptsas driver fails to allocate memory for the MPT chain buffers
for second LSI adapter on PPC440SPe Katmai platform:
...
ioc1: LSISAS1068E B3: Capabilities={Initiator}
mptbase: ioc1: ERROR - Unable to allocate Reply, Request, Chain Buffers!
mptbase: ioc1: ERROR - didn't initialize properly! (-3)
mptsas: probe of 0002:31:00.0 failed with error -3
This commit increased MPT_FC_CAN_QUEUE value but initChainBuffers()
doesn't differentiate between SAS and FC causing increased allocation
for SAS case, too. Later pci_alloc_consistent() fails to allocate
increased chain buffer pool size for SAS case.
Provide a fix by looking at the bus type and using appropriate
MPT_SAS_CAN_QUEUE value while calculation of the number of chain
buffers.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Acked-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
After m68k's task_thread_info() doesn't refer to current,
it's possible to remove sched.h from interrupt.h and not break m68k!
Many thanks to Heiko Carstens for allowing this.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial: (34 commits)
trivial: fix typo in aic7xxx comment
trivial: fix comment typo in drivers/ata/pata_hpt37x.c
trivial: typo in kernel-parameters.txt
trivial: fix typo in tracing documentation
trivial: add __init/__exit macros in drivers/gpio/bt8xxgpio.c
trivial: add __init macro/ fix of __exit macro location in ipmi_poweroff.c
trivial: remove unnecessary semicolons
trivial: Fix duplicated word "options" in comment
trivial: kbuild: remove extraneous blank line after declaration of usage()
trivial: improve help text for mm debug config options
trivial: doc: hpfall: accept disk device to unload as argument
trivial: doc: hpfall: reduce risk that hpfall can do harm
trivial: SubmittingPatches: Fix reference to renumbered step
trivial: fix typos "man[ae]g?ment" -> "management"
trivial: media/video/cx88: add __init/__exit macros to cx88 drivers
trivial: fix typo in CONFIG_DEBUG_FS in gcov doc
trivial: fix missing printk space in amd_k7_smp_check
trivial: fix typo s/ketymap/keymap/ in comment
trivial: fix typo "to to" in multiple files
trivial: fix typos in comments s/DGBU/DBGU/
...
* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (209 commits)
[SCSI] fix oops during scsi scanning
[SCSI] libsrp: fix memory leak in srp_ring_free()
[SCSI] libiscsi, bnx2i: make bound ep check common
[SCSI] libiscsi: add completion function for drivers that do not need pdu processing
[SCSI] scsi_dh_rdac: changes for rdac debug logging
[SCSI] scsi_dh_rdac: changes to collect the rdac debug information during the initialization
[SCSI] scsi_dh_rdac: move the init code from rdac_activate to rdac_bus_attach
[SCSI] sg: fix oops in the error path in sg_build_indirect()
[SCSI] mptsas : Bump version to 3.04.12
[SCSI] mptsas : FW event thread and scsi mid layer deadlock in SYNCHRONIZE CACHE command
[SCSI] mptsas : Send DID_NO_CONNECT for pending IOs of removed device
[SCSI] mptsas : PAE Kernel more than 4 GB kernel panic
[SCSI] mptsas : NULL pointer on big endian systems causing Expander not to tear off
[SCSI] mptsas : Sanity check for phyinfo is added
[SCSI] scsi_dh_rdac: Add support for Sun StorageTek ST2500, ST2510 and ST2530
[SCSI] pmcraid: PMC-Sierra MaxRAID driver to support 6Gb/s SAS RAID controller
[SCSI] qla2xxx: Update version number to 8.03.01-k6.
[SCSI] qla2xxx: Properly delete rports attached to a vport.
[SCSI] qla2xxx: Correct various NPIV issues.
[SCSI] qla2xxx: Correct qla2x00_eh_wait_on_command() to wait correctly.
...
Normally In HBA reset path MPT driver will flush existing work in current work
queue (mpt/0) . This is just a dummy activity for MPT driver point of
view, since HBA reset will turn off Work queue events.
It means we will simply returns from work queue without doing anything.
But for the case where Work is already done (half the way), we have to have
that work to be done.
Considering above condition we stuck forever since Deadlock in scsi midlayer
and MPT driver. sd_sync_cache() will wait forever since HBA is not in
Running state, and it will never come into Running state since
sd_sync_cache() is called from HBA reset context.
Now new code will not wait for half cooked work to be finished
before returning from HBA reset.
Once we are out of HBA reset, EH thread will change host state to running from
recovery and work waiting for running state of HBA will be finished.
New code is turning ON firmware event from another special work called
Rescan toplogy.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Driver is modified to return DID_NO_CONNECT for all pending I/O
requests for bus type SAS, if it founds the target is removed at
the firmware level.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This patch is solving problem for PAE kernel DMA operation.
On PAE system dma_addr and unsigned long will have different
values.
Now dma_addr is not type casted using unsigned long.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
On Big endian system kernel will crash due to address translation
is not handle properly.
Signed-off-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Do not set max_id value received from FW. Once SAS transport layer is
introduced max_id value is missleading to SCSI mid layer. Use max_id to
infinite value.
logic of can queue of scsi host is changed.
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This patch is the result of an automatic spatch transformation to convert
all ndo_start_xmit() return values of 0 to NETDEV_TX_OK.
Some occurences are missed by the automatic conversion, those will be
handled in a seperate patch.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
