scsi_device_quiesce() and scsi_device_resume() are called during
system-wide suspend and resume. scsi_device_quiesce() only succeeds for
SCSI devices that are in one of the RUNNING, OFFLINE or TRANSPORT_OFFLINE
states (see also scsi_set_device_state()). This patch avoids that the
following warning is triggered when resuming a system for which quiescing a
SCSI device failed:
WARNING: CPU: 2 PID: 11303 at drivers/scsi/scsi_lib.c:2600 scsi_device_resume+0x4f/0x58
CPU: 2 PID: 11303 Comm: kworker/u8:70 Not tainted 5.0.0-rc1+ #50
Hardware name: LENOVO 80E3/Lancer 5B2, BIOS A2CN45WW(V2.13) 08/04/2016
Workqueue: events_unbound async_run_entry_fn
Call Trace:
scsi_dev_type_resume+0x2e/0x60
async_run_entry_fn+0x32/0xd8
process_one_work+0x1f4/0x420
worker_thread+0x28/0x3c0
kthread+0x118/0x130
ret_from_fork+0x22/0x40
Cc: Przemek Socha <soprwa@gmail.com>
Reported-by: Przemek Socha <soprwa@gmail.com>
Fixes: 3a0a529971 ("block, scsi: Make SCSI quiesce and resume work reliably") # v4.15
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask() incorrectly made a second call with the 32
bit DMA mask value when the call with the 64 bit DMA mask value succeeded.
Fixes: 453cd3700c ("scsi: hptiop: use dma_set_mask")
Cc: <stable@vger.kernel.org>
Suggested-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA
mask value succeeded.
[mkp: fixed commit message]
Fixes: e4db40e7a1 ("scsi: hisi_sas: use dma_set_mask_and_coherent")
Cc: <stable@vger.kernel.org>
Suggested-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA mask
value succeeded.
Fixes: c22b332d81 ("scsi: csiostor: switch to generic DMA API")
Cc: <stable@vger.kernel.org>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA mask
value succeeded.
[mkp: fixed commit message]
Fixes: a69b080025 ("scsi: bfa: use dma_set_mask_and_coherent")
Cc: <stable@vger.kernel.org>
Suggested-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA mask
value succeeded.
[mkp: fixed subject]
Fixes: 3a21986f1a ("scsi: aic94xx: fully convert to the generic DMA API")
Cc: <stable@vger.kernel.org>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA mask
value succeeded.
Fixes: b1fa122930 ("scsi: 3w-sas: fully convert to the generic DMA API")
Cc: <stable@vger.kernel.org>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA mask
value succeeded.
Fixes: b000bced57 ("scsi: 3w-9xxx: fully convert to the generic DMA API")
Cc: <stable@vger.kernel.org>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The change to use dma_set_mask_and_coherent() incorrectly made a second
call with the 32 bit DMA mask value when the call with the 64 bit DMA mask
value succeeded. This resulted in NVMe/FC connections failing due to
corrupted data buffers, and various other SCSI/FCP I/O errors.
Fixes: f30e1bfd61 ("scsi: lpfc: use dma_set_mask_and_coherent")
Cc: <stable@vger.kernel.org>
Suggested-by: Don Dutile <ddutile@redhat.com>
Signed-off-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Up to 4.12, __scsi_error_from_host_byte() would reset the host byte to
DID_OK for various cases including DID_NEXUS_FAILURE. Commit
2a842acab1 ("block: introduce new block status code type") replaced this
function with scsi_result_to_blk_status() and removed the host-byte
resetting code for the DID_NEXUS_FAILURE case. As the line
set_host_byte(cmd, DID_OK) was preserved for the other cases, I suppose
this was an editing mistake.
The fact that the host byte remains set after 4.13 is causing problems with
the sg_persist tool, which now returns success rather then exit status 24
when a RESERVATION CONFLICT error is encountered.
Fixes: 2a842acab1 "block: introduce new block status code type"
Signed-off-by: Martin Wilck <mwilck@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The function sd_zbc_do_report_zones() issues a REPORT ZONES command with a
buffer size calculated based on the number of zones requested by the
caller. This value should however not exceed the capabilities of the
hardware maximum command size, that is, should not exceed the
max_hw_sectors limit of the device. This problem leads to failures of
report zones commands when re-validating disks with some SAS HBAs.
Fix this by limiting a report zone command buffer size to the minimum of
the device max_hw_sectors and calculated value based on the requested
number of zones. This does not change the semantic of the report_zones file
operation as report zones can always return less zone reports than
requested. Short reports are handled using a loop execution of the
report_zones file operation in the function blk_report_zones().
[Damien]
Before patch 'e76239a3748c ("block: add a report_zones method")', report
zones buffer allocation was limited to max_sectors when allocated in
blk_report_zones(). This however does not consider the actual format of the
device reply which is interface dependent. Limiting the allocation based
on the size of the expected reply format rather than the size of the array
of generic sturct blkzone passed by blk_report_zones() makes more sense.
Fixes: e76239a374 ("block: add a report_zones method")
Cc: stable@vger.kernel.org
Signed-off-by: Masato Suzuki <masato.suzuki@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In qla2x00_async_tm_cmd, we reference off sp after it has been freed. This
caused a panic on a system running a slub debug kernel. Since fcport is
passed in anyways, just use that instead.
Signed-off-by: Bill Kuzeja <william.kuzeja@stratus.com>
Acked-by: Giridhar Malavali <gmalavali@marvell.com>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The problem is that the default for MQ is not to gather entropy, whereas
the default for the legacy queue was always to gather it. The original
attempt to fix entropy gathering for rotational disks under MQ added an
else branch in sd_read_block_characteristics(). Unfortunately, the entire
check isn't reached if the device has no characteristics VPD page. Since
this page was only introduced in SBC-3 and its optional anyway, most less
expensive rotational disks don't have one, meaning they all stopped
gathering entropy when we made MQ the default. In a wholly unrelated
change, openssl and openssh won't function until the random number
generator is initialised, meaning lots of people have been seeing large
delays before they could log into systems with default MQ kernels due to
this lack of entropy, because it now can take tens of minutes to initialise
the kernel random number generator.
The fix is to set the non-rotational and add-randomness flags
unconditionally early on in the disk initialization path, so they can be
reset only if the device actually reports being non-rotational via the VPD
page.
Reported-by: Mikael Pettersson <mikpelinux@gmail.com>
Fixes: 83e32a5910 ("scsi: sd: Contribute to randomness when running rotational device")
Cc: stable@vger.kernel.org
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Xuewei Zhang <xueweiz@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Presently when an error is encountered during probe of the cxlflash
adapter, a deadlock is seen with cpu thread stuck inside
cxlflash_remove(). Below is the trace of the deadlock as logged by
khungtaskd:
cxlflash 0006:00:00.0: cxlflash_probe: init_afu failed rc=-16
INFO: task kworker/80:1:890 blocked for more than 120 seconds.
Not tainted 5.0.0-rc4-capi2-kexec+ #2
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
kworker/80:1 D 0 890 2 0x00000808
Workqueue: events work_for_cpu_fn
Call Trace:
0x4d72136320 (unreliable)
__switch_to+0x2cc/0x460
__schedule+0x2bc/0xac0
schedule+0x40/0xb0
cxlflash_remove+0xec/0x640 [cxlflash]
cxlflash_probe+0x370/0x8f0 [cxlflash]
local_pci_probe+0x6c/0x140
work_for_cpu_fn+0x38/0x60
process_one_work+0x260/0x530
worker_thread+0x280/0x5d0
kthread+0x1a8/0x1b0
ret_from_kernel_thread+0x5c/0x80
INFO: task systemd-udevd:5160 blocked for more than 120 seconds.
The deadlock occurs as cxlflash_remove() is called from cxlflash_probe()
without setting 'cxlflash_cfg->state' to STATE_PROBED and the probe thread
starts to wait on 'cxlflash_cfg->reset_waitq'. Since the device was never
successfully probed the 'cxlflash_cfg->state' never changes from
STATE_PROBING hence the deadlock occurs.
We fix this deadlock by setting the variable 'cxlflash_cfg->state' to
STATE_PROBED in case an error occurs during cxlflash_probe() and just
before calling cxlflash_remove().
Cc: stable@vger.kernel.org
Fixes: c21e0bbfc485("cxlflash: Base support for IBM CXL Flash Adapter")
Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This reverts commit bbc0f8bd88.
It added a warning whose intent was to check whether the rport was still
linked into the peer list. It doesn't work as intended and gives false
positive warnings for two reasons:
1) If the rport is never linked into the peer list it will not be
considered empty since the list_head is never initialized.
2) If the rport is deleted from the peer list using list_del_rcu(), then
the list_head is in an undefined state and it is not considered empty.
Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Commit bf50545696 ("block: Introduce blk_revalidate_disk_zones()")
inadvertently broke the message output of sd_zbc_print_zones() because the
zone information initialization of the scsi disk structure was moved to the
second scan run while sd_zbc_print_zones() is called on the first
scan. This leads to the following incorrect message to be printed for any
ZBC or ZAC zoned disks.
"...[sdX] 4294967295 zones of 0 logical blocks + 1 runt zone"
Fix this by initializing sdkp zone size and number of zones early on the
first scan. This does not impact the execution of
blk_revalidate_zones(). This functions is still called only once the block
device capacity is set on the second revalidate run on boot, or if the disk
zone configuration changed (i.e. the disk changed).
Fixes: bf50545696 ("block: Introduce blk_revalidate_disk_zones()")
Cc: stable@vger.kernel.org
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
pi_prot_format conversion to write-only caused userspace breakage. Make the
ConfigFS path readable again and hardcode the "0\n" content, matching
previous output.
Fixes: 6baca7601b ("scsi: target: drop unused pi_prot_format attribute storage")
Link: https://bugzilla.redhat.com/show_bug.cgi?id=1667505
Reported-by: Lee Duncan <lduncan@suse.com>
Reported-by: Laura Abbott <labbott@redhat.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The aic94xx driver is currently failing to load with errors like
sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:03.0/0000:02:00.3/0000:07:02.0/revision'
Because the PCI code had recently added a file named 'revision' to every
PCI device. Fix this by renaming the aic94xx revision file to
aic_revision. This is safe to do for us because as far as I can tell,
there's nothing in userspace relying on the current aic94xx revision file
so it can be renamed without breaking anything.
Fixes: 702ed3be1b (PCI: Create revision file in sysfs)
Cc: stable@vger.kernel.org
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The "hostdata->dev" pointer is NULL here. We set "hostdata->dev = dev;"
later in the function and we also use "hostdata->dev" when we call
dma_free_attrs() in NCR_700_release().
This bug predates git version control.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
There are two issues here. First if cmgr->hba is not set early enough then
it leads to a NULL dereference. Second if we don't completely initialize
cmgr->io_bdt_pool[] then we end up dereferencing uninitialized pointers.
Fixes: 853e2bd210 ("[SCSI] bnx2fc: Broadcom FCoE offload driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The WRITE SAME(10) and (16) implementations didn't take account of the
buffer wrap required when the virtual_gb parameter is greater than 0.
Fix that and rename the fake_store() function to lba2fake_store() to lessen
confusion with the global fake_storep pointer. Bump version date.
Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Reported-by: Bart Van Assche <bvanassche@acm.org>
Tested by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The issue to be fixed in this commit is when libfc found it received a
invalid FLOGI response from FC switch, it would return without freeing the
fc frame, which is just the skb data. This would cause memory leak if FC
switch keeps sending invalid FLOGI responses.
This fix is just to make it execute `fc_frame_free(fp)` before returning
from function `fc_lport_flogi_resp`.
Signed-off-by: Ming Lu <ming.lu@citrix.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Since v2.6.35 commit 683229845f ("[SCSI] zfcp: Report scatter-gather
limits to SCSI and block layer"), zfcp set dma_parms.max_segment_size ==
PAGE_SIZE (but without using the setter dma_set_max_seg_size()) and
scsi_host_template.dma_boundary == PAGE_SIZE - 1.
v5.0-rc1 commit 50c2e9107f ("scsi: introduce a max_segment_size
host_template parameters") introduced a new field
scsi_host_template.max_segment_size. If an LLDD such as zfcp does not set
it, scsi_host_alloc() uses BLK_MAX_SEGMENT_SIZE = 65536 for
Scsi_Host.max_segment_size. __scsi_init_queue() announced the minimum of
Scsi_Host.max_segment_size and dma_parms.max_segment_size to the block
layer. For zfcp: min(65536, 4096) == 4096 which was still good.
v5.0 commit a8cf59a669 ("scsi: communicate max segment size to the DMA
mapping code") announces Scsi_Host.max_segment_size to the block layer and
overwrites dma_parms.max_segment_size with Scsi_Host.max_segment_size. For
zfcp dma_parms.max_segment_size == Scsi_Host.max_segment_size == 65536
which is also reflected in block queue limits.
$ cd /sys/bus/ccw/drivers/zfcp
$ cd 0.0.3c40/host5/rport-5:0-4/target5:0:4/5:0:4:10/block/sdi/queue
$ cat max_segment_size
65536
Zfcp I/O still works because dma_boundary implicitly still keeps the
effective max segment size <= PAGE_SIZE. However, dma_boundary does not
seem visible to user space, but max_segment_size is visible and shows a
misleading wrong value. Fix it and inherit the stable tag of a8cf59a669.
Devices on our bus ccw support DMA but no DMA mapping. Of multiple device
types on the ccw bus, only zfcp needs dma_parms for SCSI limits. So, leave
dma_parms setup in zfcp and do not move it to the bus.
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Fixes: 50c2e9107f ("scsi: introduce a max_segment_size host_template parameters")
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Fixes: a94a2572b9 ("scsi: tcmu: avoid cmd/qfull timers updated whenever a new cmd comes")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Mike Christie <mchristi@redhat.com>
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Mike Christie <mchristi@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Assign fc_vport to ln->fc_vport before calling csio_fcoe_alloc_vnp() to
avoid a NULL pointer dereference in csio_vport_set_state().
ln->fc_vport is dereferenced in csio_vport_set_state().
Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
We cannot wait on a completion object in the lpfc_nvme_targetport structure
in the _destroy_targetport() code path because the NVMe/fc transport will
free that structure immediately after the .targetport_delete() callback.
This results in a use-after-free, and a hang if slub_debug=FZPU is enabled.
Fix this by putting the completion on the stack.
Signed-off-by: Ewan D. Milne <emilne@redhat.com>
Acked-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
We cannot wait on a completion object in the lpfc_nvme_lport structure in
the _destroy_localport() code path because the NVMe/fc transport will free
that structure immediately after the .localport_delete() callback. This
results in a use-after-free, and a hang if slub_debug=FZPU is enabled.
Fix this by putting the completion on the stack.
Signed-off-by: Ewan D. Milne <emilne@redhat.com>
Acked-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When a host driver sets a maximum segment size we should not only propagate
that setting to the block layer, which can merge segments, but also to the
DMA mapping layer which can merge segments as well.
Fixes: 50c2e9107f ("scsi: introduce a max_segment_size host_template parameters")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In case of ->set_param() and ->bind_conn() cxgb4i driver does not wait for
cmd completion, this can create race conditions, to avoid this add
wait_for_completion().
Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
After Commit 54aed4dd35 ("MIPS: IP27: use dma_direct_ops") qla1280 driver
failed on SGI IP27 machines with
qla1280: QLA1040 found on PCI bus 0, dev 0
qla1280 0000:00:00.0: enabling device (0006 -> 0007)
qla1280: Failed to get request memory
qla1280: probe of 0000:00:00.0 failed with error -12
Reason is that SGI IP27 always generates 64bit DMA addresses and has no
fallback mode for 32bit DMA addresses implemented. QLA1280 supports 64bit
addressing for all DMA accesses so setting coherent mask to 64bit fixes the
issue.
Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Albeit we no longer rely on those hard-coded descriptor sizes, we still use
them as our defaults, so better get it right. While adding its sysfs
entries, we forgot to update the geometry descriptor size. It is 0x48
according to UFS2.1, and wasn't changed in UFS3.0.
[mkp: typo]
Fixes: c720c09122 (scsi: ufs: sysfs: geometry descriptor)
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
commit 272652fcbf ("scsi: megaraid_sas: add retry logic in megasas_readl")
missed changing readl to megasas_readl in megasas_clear_intr_fusion(). For
Aero controllers, reads of outbound_intr_status register needs to be
retried.
Reported-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When the driver finds invalid destination MAC for the first un-reachable
target, and before completes the PATH_REQ operation, set new ep_state to
OFFLDCONN_NONE so that as part of driver ep_poll mechanism, the upper
open-iscsi layer is notified to complete the login process on the first
un-reachable target and thus proceed login to other reachable targets.
Signed-off-by: Manish Rangankar <mrangankar@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
hba->is_sys_suspended is set after successful system suspend but
not clear after successful system resume.
According to current behavior, hba->is_sys_suspended will not be set if
host is runtime-suspended but not system-suspended. Thus we shall aligh the
same policy: clear this flag even if host remains runtime-suspended after
ufshcd_system_resume is successfully returned.
Simply fix this flag to correct host status logs.
Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When SCSI-MQ is enabled, in some case system would present
nr_possible_cpus() which is greater than requested vectors by the
driver. This results into driver being able to get larger number of MSI-X
vectors than actual online CPUs. Driver then uses
pci_alloc_irq_vectors_affinity() to assign 1:1 mapping and affinity for
each MSI-x vector to CPUs. When the command is submitted using MSI-x
vector, assigned to offline CPU, it results in an ABTS and system
hang. This hang is result of a driver not being able to process interrupt
on a vector assigned to an Off-line CPUs
This patch fixes this issue by setting irq_offset value for the
blk_mq_pci_map_queues() to use only those CPUs which has CPU mask affinity
assigned and are online. By using the irq_offset value, driver will allow
online cpumask to decide which vectors are used in blk_mq_pci_map_queues().
Fixes: 5601236b6f ("scsi: qla2xxx: Add Block Multi Queue functionality.")
Cc: <stable@vger.kernel.org> #4.19
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Himanshu Madhani <hmadhani@marvell.com>
Tested-by: Himanshu Madhani <hmadhani@marvell.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently we set the protection parameters after calling scsi_add_host()
for v3 hw.
They should be set beforehand, so make this change.
Appearantly this fixes our DIX issue (not mainline yet) also, but more
testing required.
Fixes: d6a9000b81 ("scsi: hisi_sas: Add support for DIF feature for v2 hw")
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently there is one cmd timeout timer and one qfull timer for each udev,
and whenever any new command is coming in we will update the cmd timer or
qfull timer. For some corner cases the timers are always working only for
the ringbuffer's and full queue's newest cmd. That's to say the timer won't
be fired even if one cmd has been stuck for a very long time and the
deadline is reached.
This fix will keep the cmd/qfull timers to be pended for the oldest cmd in
ringbuffer and full queue, and will update them with the next cmd's
deadline only when the old cmd's deadline is reached or removed from the
ringbuffer and full queue.
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Acked-by: Mike Christie <mchristi@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
scsi_mq_setup_tags(), which is called by scsi_add_host(), calculates the
command size to allocate based on the prot_capabilities. In the isci
driver, scsi_host_set_prot() is called after scsi_add_host() so the command
size gets calculated to be smaller than it needs to be. Eventually,
scsi_mq_init_request() locates the 'prot_sdb' after the command assuming it
was sized correctly and a buffer overrun may occur.
However, seeing blk_mq_alloc_rqs() rounds up to the nearest cache line
size, the mistake can go unnoticed.
The bug was noticed after the struct request size was reduced by commit
9d037ad707 ("block: remove req->timeout_list")
Which likely reduced the allocated space for the request by an entire cache
line, enough that the overflow could be hit and it caused a panic, on boot,
at:
RIP: 0010:t10_pi_complete+0x77/0x1c0
Call Trace:
<IRQ>
sd_done+0xf5/0x340
scsi_finish_command+0xc3/0x120
blk_done_softirq+0x83/0xb0
__do_softirq+0xa1/0x2e6
irq_exit+0xbc/0xd0
call_function_single_interrupt+0xf/0x20
</IRQ>
sd_done() would call scsi_prot_sg_count() which reads the number of
entities in 'prot_sdb', but seeing 'prot_sdb' is located after the end of
the allocated space it reads a garbage number and erroneously calls
t10_pi_complete().
To prevent this, the calls to scsi_host_set_prot() are moved into
isci_host_alloc() before the call to scsi_add_host(). Out of caution, also
move the similar call to scsi_host_set_guard().
Fixes: 3d2d752549 ("[SCSI] isci: T10 DIF support")
Link: http://lkml.kernel.org/r/da851333-eadd-163a-8c78-e1f4ec5ec857@deltatee.com
Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
Cc: Intel SCU Linux support <intel-linux-scu@intel.com>
Cc: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Jeff Moyer <jmoyer@redhat.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases where
we are expecting to fall through.
Notice that, in this particular case, I replaced "Drop thru" and "Fall
Thru" with "fall through" annotations, which is what GCC is expecting to
find.
Also, in some cases a dash is added as a token in order to separate the
"fall through" annotation from the rest of the comment on the same line,
which is what GCC is expecting to find.
Addresses-Coverity-ID: 114979 ("Missing break in switch")
Addresses-Coverity-ID: 114980 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Fix boolean expression by using logical AND operator '&&' instead of
bitwise operator '&'.
This issue was detected with the help of Coccinelle.
Fixes: 1e46731efd ("scsi: smartpqi: check for null device pointers")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The commit 356fd2663c ("scsi: Set request queue runtime PM status back to
active on resume") fixed up the inconsistent RPM status between request
queue and device. However changing request queue RPM status shall be done
only on successful resume, otherwise status may be still inconsistent as
below,
Request queue: RPM_ACTIVE
Device: RPM_SUSPENDED
This ends up soft lockup because requests can be submitted to underlying
devices but those devices and their required resource are not resumed.
For example,
After above inconsistent status happens, IO request can be submitted to UFS
device driver but required resource (like clock) is not resumed yet thus
lead to warning as below call stack,
WARN_ON(hba->clk_gating.state != CLKS_ON);
ufshcd_queuecommand
scsi_dispatch_cmd
scsi_request_fn
__blk_run_queue
cfq_insert_request
__elv_add_request
blk_flush_plug_list
blk_finish_plug
jbd2_journal_commit_transaction
kjournald2
We may see all behind IO requests hang because of no response from storage
host or device and then soft lockup happens in system. In the end, system
may crash in many ways.
Fixes: 356fd2663c (scsi: Set request queue runtime PM status back to active on resume)
Cc: stable@vger.kernel.org
Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Delete tab aligning a statement with the right hand side of a preceding
assignment rather than the left hand side.
Found with the help of Coccinelle.
[mkp: added space]
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Jack Wang <jinpu.wang@cloud.ionos.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The return code should be check while qla4xxx_copy_from_fwddb_param fails.
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Manish Rangankar <mrangankar@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
This was apparently forgotten in
894169db1 ("scsi: megaraid_sas: Use 63-bit DMA addressing").
Signed-off-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Changing of caching mode via /sys/devices/.../scsi_disk/.../cache_type may
fail if device responds to MODE SENSE command with DPOFUA flag set, and
then checks this flag to be not set on MODE SELECT command.
In this scenario, when trying to change cache_type, write always fails:
# echo "none" >cache_type
bash: echo: write error: Invalid argument
And following appears in dmesg:
[13007.865745] sd 1:0:1:0: [sda] Sense Key : Illegal Request [current]
[13007.865753] sd 1:0:1:0: [sda] Add. Sense: Invalid field in parameter list
From SBC-4 r15, 6.5.1 "Mode pages overview", description of DEVICE-SPECIFIC
PARAMETER field in the mode parameter header:
...
The write protect (WP) bit for mode data sent with a MODE SELECT
command shall be ignored by the device server.
...
The DPOFUA bit is reserved for mode data sent with a MODE SELECT
command.
...
The remaining bits in the DEVICE-SPECIFIC PARAMETER byte are also reserved
and shall be set to zero.
[mkp: shuffled commentary to commit description]
Cc: stable@vger.kernel.org
Signed-off-by: Ivan Mironov <mironov.ivan@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
