Merge patch series "lpfc: Update lpfc to revision 14.2.0.13"

Justin Tee <justintee8345@gmail.com> says:

Update lpfc to revision 14.2.0.13

This patch set contains discovery bug fixes, firmware logging
improvements, clean up of CQ handling, and statistics collection
enhancements.

The patches were cut against Martin's 6.5/scsi-queue tree.

Link: https://lore.kernel.org/r/20230523183206.7728-1-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Copyright updates for 14.2.0.13 patches

Update copyrights to 2023 for files modified in the 14.2.0.13 patch set.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-10-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Update lpfc version to 14.2.0.13

Update lpfc version to 14.2.0.13

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-9-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Enhance congestion statistics collection

Various improvements are made for collecting congestion statistics:

 - Pre-existing logic is replaced with use of an hrtimer for increased
   reporting accuracy.

 - Congestion timestamp information is reorganized into a single struct.

 - Common statistic collection logic is refactored into a helper routine.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-8-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Clean up SLI-4 CQE status handling

There is mishandling of SLI-4 CQE status values larger than what is allowed
by the LPFC_IOCB_STATUS_MASK of 4 bits.  The LPFC_IOCB_STATUS_MASK is a
leftover SLI-3 construct and serves no purpose in SLI-4 path.

Remove the LPFC_IOCB_STATUS_MASK and clean up general CQE status handling
in SLI-4 completion paths.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-7-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Change firmware upgrade logging to KERN_NOTICE instead of TRACE_EVENT

A firmware upgrade does not necessitate dumping of phba->dbg_log[] to kmsg
via LOG_TRACE_EVENT.  A simple KERN_NOTICE log message should suffice to
notify the user of successful or unsuccessful firmware upgrade.  As such,
firmware upgrade log messages are updated to use KERN_NOTICE instead of
LOG_TRACE_EVENT.  Additionally, in order to notify the user of reset type
for instantiating newly downloaded firmware, lpfc_log_msg's default
KERN_LEVEL is updated to 5 or KERN_NOTICE.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-6-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Revise NPIV ELS unsol rcv cmpl logic to drop ndlp based on nlp_state

When NPIV ports are zoned to devices that support both initiator and target
mode, a remote device's initiated PRLI results in unintended final kref
clean up of the device's ndlp structure.  This disrupts NPIV ports'
discovery for target devices that support both initiator and target mode.

Modify the NPIV lpfc_drop_node clause such that we allow the ndlp to live
so long as it was in NLP_STE_PLOGI_ISSUE, NLP_STE_REG_LOGIN_ISSUE, or
NLP_STE_PRLI_ISSUE nlp_state.  This allows lpfc's issued PRLI completion
routine to determine if the final kref clean up should execute rather than
a remote device's issued PRLI.

Fixes: db651ec22524 ("scsi: lpfc: Correct used_rpi count when devloss tmo fires with no recovery")
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-5-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Account for fabric domain ctlr device loss recovery

Pre-existing device loss recovery logic via the NLP_IN_RECOV_POST_DEV_LOSS
flag only handled Fabric Port Login, Fabric Controller, Management, and
Name Server addresses.

Fabric domain controllers fall under the same category for usage of the
NLP_IN_RECOV_POST_DEV_LOSS flag.  Add a default case statement to mark an
ndlp for device loss recovery.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-4-justintee8345@gmail.com
Acked-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Clear NLP_IN_DEV_LOSS flag if already in rediscovery

In dev_loss_tmo callback routine, we early return if the ndlp is in a state
of rediscovery.  This occurs when a target proactively PLOGIs or PRLIs
after an RSCN before the dev_loss_tmo callback routine is scheduled to run.
Move clear of the NLP_IN_DEV_LOSS flag before the ndlp state check in such
cases.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-3-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Fix use-after-free rport memory access in lpfc_register_remote_port()

Due to a target port D_ID swap, it is possible for the
lpfc_register_remote_port() routine to touch post mortem fc_rport memory
when trying to access fc_rport->dd_data.

The D_ID swap causes a simultaneous call to lpfc_unregister_remote_port(),
where fc_remote_port_delete() reclaims fc_rport memory.

Remove the fc_rport->dd_data->pnode NULL assignment because the following
line reassigns ndlp->rport with an fc_rport object from
fc_remote_port_add() anyways.  The pnode nullification is superfluous.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230523183206.7728-2-justintee8345@gmail.com
Acked-by: Martin Wilck <mwilck@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230530164131.987213-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230530162321.984035-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Acked-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sym53c8xx: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230530160323.412484-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230530160248.411637-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libfcoe: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230530155818.368562-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230530155745.343032-1-azeemshaikh38@gmail.com
Reviewed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: dt-bindings: ufs: qcom: Add compatible for sa8775p

Add the compatible string for the UFS on sa8775p platforms.

Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Link: https://lore.kernel.org/r/20230411130446.401440-2-brgl@bgdev.pl
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: core: Do not open code SZ_x

Do not open code SZ_x.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/20230531070009.4593-1-avri.altman@wdc.com
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Keoseong Park <keosung.park@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "ufs: Do not requeue while ungating the clock"

Bart Van Assche <bvanassche@acm.org> says:

In the traces we recorded while testing zoned storage we noticed that UFS
commands are requeued while the clock is being ungated. Command requeueing
makes it harder than necessary to preserve the command order. Hence this
patch series that modifies the SCSI core and also the UFS driver such that
clock ungating does not trigger command requeueing.

Link: https://lore.kernel.org/r/20230529202640.11883-1-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Ungate the clock synchronously

Ungating the clock asynchronously causes ufshcd_queuecommand() to return
SCSI_MLQUEUE_HOST_BUSY and hence causes commands to be requeued.  This is
suboptimal. Allow ufshcd_queuecommand() to sleep such that clock ungating
does not trigger command requeuing. Remove the ufshcd_scsi_block_requests()
and ufshcd_scsi_unblock_requests() calls because these are no longer
needed. The flush_work(&hba->clk_gating.ungate_work) call is sufficient to
make the SCSI core wait for clock ungating to complete.

Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230529202640.11883-6-bvanassche@acm.org
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Bao D. Nguyen <quic_nguyenb@quicinc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Declare ufshcd_{hold,release}() once

ufshcd_hold() and ufshcd_release are declared twice: once in
drivers/ufs/core/ufshcd-priv.h and a second time in include/ufs/ufshcd.h.
Remove the declarations from ufshcd-priv.h.

Fixes: dd11376b9f1b ("scsi: ufs: Split the drivers/scsi/ufs directory")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230529202640.11883-5-bvanassche@acm.org
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Keoseong Park <keosung.park@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Conditionally enable the BLK_MQ_F_BLOCKING flag

Prepare for adding code in ufshcd_queuecommand() that may sleep.

Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230529202640.11883-4-bvanassche@acm.org
Reviewed-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Support setting BLK_MQ_F_BLOCKING

Prepare for adding code in ufshcd_queuecommand() that may sleep. This patch
is similar to a patch posted last year by Mike Christie. See also
https://lore.kernel.org/all/20220308003957.123312-2-michael.christie@oracle.com/

Cc: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230529202640.11883-3-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Rework scsi_host_block()

Make scsi_host_block() easier to read by converting it to the widely used
early-return style. See also commit f983622ae605 ("scsi: core: Avoid
calling synchronize_rcu() for each device in scsi_host_block()").

Reviewed-by: Mike Christie <michael.christie@oracle.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Cc: Ye Bin <yebin10@huawei.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230529202640.11883-2-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: virtio_scsi: Remove a useless function call

'inq_result' is known to be NULL. There is no point calling kfree().

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/08740635cdb0f8293e57c557b22e048daae50961.1685345683.git.christophe.jaillet@wanadoo.fr
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata-scsi: Fix ata_msense_control kdoc comment

Add missing description of the spg argument of ata_msense_control().

Fixes: df60f9c64576 ("scsi: ata: libata: Add ATA feature control sub-page translation")
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://lore.kernel.org/r/20230523074701.293502-1-dlemoal@kernel.org
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix end of loop test

This loop will exit successfully when "found" is false or in the failure
case it times out with "wait_iter" set to -1.  The test for timeouts is
impossible as is.

Fixes: b843adde8d49 ("scsi: qla2xxx: Fix mem access after free")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Link: https://lore.kernel.org/r/cea5a62f-b873-4347-8f8e-c67527ced8d2@kili.mountain
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Only kick the requeue list if necessary

Instead of running the request queue of each device associated with a host
every 3 ms (BLK_MQ_RESOURCE_DELAY) while host error handling is in
progress, run the request queue after error handling has finished.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230518193159.1166304-4-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Trace SCSI sense data

If a command fails, SCSI sense data is essential to determine why it
failed. Hence make the sense key, ASC and ASCQ codes available in the
ftrace output.

Cc: Niklas Cassel <niklas.cassel@wdc.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230518193159.1166304-3-bvanassche@acm.org
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Use min() instead of open-coding it

Use min() instead of open-coding it in scsi_normalize_sense().

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Cc: Douglas Gilbert <dgilbert@interlog.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20230518193159.1166304-2-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: sym53c8xx_2: Shorten chapter heading

Make the chapter heading concise yet still descriptive.
This makes the subsystem table of contents more readable (IMO).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-12-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: ncr53c8xx: Shorten chapter heading

Make the chapter heading concise yet still descriptive.
This makes the subsystem table of contents more readable (IMO).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-11-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: megaraid: Clarify chapter heading

Include "Megaraid" in the chapter heading so that it is clear
what subject the document is about. This improves viewing in the TOC.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-10-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: Kashyap Desai <kashyap.desai@broadcom.com>
Cc: Sumit Saxena <sumit.saxena@broadcom.com>
Cc: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Cc: megaraidlinux.pdl@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: g_NCR5380: Shorten chapter heading

Make the chapter heading be concise yet still descriptive.
This makes the subsystem table of contents more readable (IMO).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-9-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: Finn Thain <fthain@linux-m68k.org>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Acked-by: Finn Thain <fthain@linux-m68k.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: scsi-generic: Multiple cleanups

Make the heading be concise yet still descriptive.
This makes the subsystem table of contents more readable (IMO).

Spell "CDROM" as "CD-ROM".
Capitalize "Linux".

Use https instead of http for URLs.

Drop the Linux Documentation Project URL for the SCSI generic HOWTO
since it hasn't been updated since 2002. Use Doug Gilbert's URL
for it instead.

Drop some outdated documentation & references.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-8-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: Doug Gilbert <dgilbert@interlog.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: scsi_fc_transport: Fix typo in heading

Fix the typo "Tansport" to be "Transport".
Update email address for James Smart.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-7-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: dc395x: Shorten the chapter heading

Make the heading be concise yet still descriptive.
This makes the subsystem table of contents more readable (IMO).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-6-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: Oliver Neukum <oliver@neukum.org>
Cc: Ali Akcaagac <aliakc@web.de>
Cc: Jamie Lenehan <lenehan@twibble.org>
Cc: dc395x@twibble.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: scsi-changer: Shorten the chapter heading

Make the heading be concise yet still descriptive.
This makes the subsystem table of contents more readable (IMO).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-5-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Cc: Gerd Knorr <kraxel@bytesex.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: arcmsr: Use a chapter heading for clarity

Add a chapter heading so that the document sections are not all
at the same level, mucking up the SCSI subsystem contents.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-4-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: introduction: Multiple cleanups

Modify URLs to use https instead of http.
Remove ancient URLs that don't work.
Change "scsi" in text to "SCSI".
Change "cdrom" in text to "CD-ROM".
Drop the reference to "autoclean" for modules since I can't
  find it in any current documentation.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-3-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: docs: Organize the SCSI documentation

Break the SCSI documentation up into categories:

  Introduction, APIs, driver parameters, and host adapter drivers instead
  of alphabetical by document file name (i.e., no organization).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230518212749.18266-2-rdunlap@infradead.org
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: linux-doc@vger.kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Replace one-element array with flexible-array member

One-element arrays are deprecated, and we are replacing them with flexible
array members instead. So, replace one-element arrays with flexible-array
members in a couple of structures, and refactor the rest of the code,
accordingly.

This helps with the ongoing efforts to tighten the FORTIFY_SOURCE routines
on memcpy() and help us make progress towards globally enabling
-fstrict-flex-arrays=3 [1].

This results in no differences in binary output.

Link: https://github.com/KSPP/linux/issues/79
Link: https://github.com/KSPP/linux/issues/295
Link: https://gcc.gnu.org/pipermail/gcc-patches/2022-October/602902.html
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Link: https://lore.kernel.org/r/6c6dcab88524c14c47fd06b9332bd96162656db5.1684358315.git.gustavoars@kernel.org
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: Fix the type used for pointers to bitmap

Bitmaps are "unsigned long[]", so better use "unsigned long *" instead of a
plain "void *" when dealing with pointers to bitmaps.

This is more informative.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/8bdf9148ce1a5d01aac11c46c8617b477813457e.1683473011.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: 3w-xxxx: Add error handling for initialization failure in tw_probe()

Smatch complains that:

tw_probe() warn: missing error code 'retval'

This patch adds error checking to tw_probe() to handle initialization
failure. If tw_reset_sequence() function returns a non-zero value, the
function will return -EINVAL to indicate initialization failure.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Yuchen Yang <u202114568@hust.edu.cn>
Link: https://lore.kernel.org/r/20230505141259.7730-1-u202114568@hust.edu.cn
Reviewed-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: core: Return earlier if ufshcd_hba_init_crypto_capabilities() fails

The 'err' variable is used only as the result of
ufshcd_hba_init_crypto_capabilities(), so return 'err' immediately when
failed. If it is not an error, explicitly return 0.

Signed-off-by: Keoseong Park <keosung.park@samsung.com>
Link: https://lore.kernel.org/r/20230503104630epcms2p8b82734102ffb920531e9264604086372@epcms2p8
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "Add Command Duration Limits support"

Niklas Cassel <nks@flawful.org> says:

This series adds support for Command Duration Limits.
The series is based on linux tag: v6.4-rc1
The series can also be found in git: https://github.com/floatious/linux/commits/cdl-v7

=================
CDL in ATA / SCSI
=================
Command Duration Limits is defined in:
T13 ATA Command Set - 5 (ACS-5) and
T10 SCSI Primary Commands - 6 (SPC-6) respectively
(a simpler version of CDL is defined in T10 SPC-5).

CDL defines Duration Limits Descriptors (DLD).
7 DLDs for read commands and 7 DLDs for write commands.
Simply put, a DLD contains a limit and a policy.

A command can specify that a certain limit should be applied by setting
the DLD index field (3 bits, so 0-7) in the command itself.

The DLD index points to one of the 7 DLDs.
DLD index 0 means no descriptor, so no limit.
DLD index 1-7 means DLD 1-7.

A DLD can have a few different policies, but the two major ones are:
-Policy 0xF (abort), command will be completed with command aborted error
(ATA) or status CHECK CONDITION (SCSI), with sense data indicating that
the command timed out.
-Policy 0xD (complete-unavailable), command will be completed without
error (ATA) or status GOOD (SCSI), with sense data indicating that the
command timed out. Note that the command will not have transferred any
data to/from the device when the command timed out, even though the
command returned success.

Regardless of the CDL policy, in case of a CDL timeout, the I/O will
result in a -ETIME error to user-space.

The DLDs are defined in the CDL log page(s) and are readable and writable.
Reading and writing the CDL DLDs are outside the scope of the kernel.
If a user wants to read or write the descriptors, they can do so using a
user-space application that sends passthrough commands, such as cdl-tools:
https://github.com/westerndigitalcorporation/cdl-tools

================================
The introduction of ioprio hints
================================
What the kernel does provide, is a method to let I/O use one of the CDL DLDs
defined in the device. Note that the kernel will simply forward the DLD index
to the device, so the kernel currently does not know, nor does it need to know,
how the DLDs are defined inside the device.

The way that the CDL DLD index is supplied to the kernel is by introducing a
new 10 bit "ioprio hint" field within the existing 16 bit ioprio definition.

Currently, only 6 out of the 16 ioprio bits are in use, the remaining 10 bits
are unused, and are currently explicitly disallowed to be set by the kernel.

For now, we only add ioprio hints representing CDL DLD index 1-7. Additional
ioprio hints for other QoS features could be defined in the future.

A theoretical future work could be to make an I/O scheduler aware of these
hints. E.g. for CDL, an I/O scheduler could make use of the duration limit
in each descriptor, and take that information into account while scheduling
commands. Right now, the ioprio hints will be ignored by the I/O schedulers.

==============================
How to use CDL from user-space
==============================
Since CDL is mutually exclusive with NCQ priority
(see ncq_prio_enable and sas_ncq_prio_enable in
Documentation/ABI/testing/sysfs-block-device),
CDL has to be explicitly enabled using:
echo 1 > /sys/block/$bdev/device/cdl_enable

Since the ioprio hints are supplied through the existing I/O priority API,
it should be simple for an application to make use of the ioprio hints.

It simply has to reuse one of the new macros defined in
include/uapi/linux/ioprio.h: IOPRIO_PRIO_HINT() or IOPRIO_PRIO_VALUE_HINT(),
and supply one of the new hints defined in include/uapi/linux/ioprio.h:
IOPRIO_HINT_DEV_DURATION_LIMIT_[1-7], which indicates that the I/O should
use the corresponding CDL DLD index 1-7.

By reusing the I/O priority API, the user can both define a DLD to use per
AIO (io_uring sqe->ioprio or libaio iocb->aio_reqprio) or per-thread
(ioprio_set()).

=======
Testing
=======
With the following fio patches:
https://github.com/floatious/fio/commits/cdl

fio adds support for ioprio hints, such that CDL can be tested using e.g.:
fio --ioengine=io_uring --cmdprio_percentage=10 --cmdprio_hint=DLD_index

A simple way to test is to use a DLD with a very short duration limit,
and send large reads. Regardless of the CDL policy, in case of a CDL
timeout, the I/O will result in a -ETIME error to user-space.

We also provide a CDL test suite located in the cdl-tools repo, see:
https://github.com/westerndigitalcorporation/cdl-tools#testing-a-system-command-duration-limits-support

We have tested this patch series using:
-real hardware
-the following QEMU implementation:
https://github.com/floatious/qemu/tree/cdl
(NOTE: the QEMU implementation requires you to define the CDL policy at compile
time, so you currently need to recompile QEMU when switching between policies.)

===================
Further information
===================
For further information about CDL, see Damien's slides:

Presented at SDC 2021:
https://www.snia.org/sites/default/files/SDC/2021/pdfs/SNIA-SDC21-LeMoal-Be-On-Time-command-duration-limits-Feature-Support-in%20Linux.pdf

Presented at Lund Linux Con 2022:
https://drive.google.com/file/d/1I6ChFc0h4JY9qZdO1bY5oCAdYCSZVqWw/view?usp=sharing

================
Changes since V6
================
-Rebased series on v6.4-rc1.
-Picked up Reviewed-by tags from Hannes (Thank you Hannes!)
-Picked up Reviewed-by tag from Christoph (Thank you Christoph!)
-Changed KernelVersion from 6.4 to 6.5 for new sysfs attributes.

For older change logs, see previous patch series versions:
https://lore.kernel.org/linux-scsi/20230406113252.41211-1-nks@flawful.org/
https://lore.kernel.org/linux-scsi/20230404182428.715140-1-nks@flawful.org/
https://lore.kernel.org/linux-scsi/20230309215516.3800571-1-niklas.cassel@wdc.com/
https://lore.kernel.org/linux-scsi/20230124190308.127318-1-niklas.cassel@wdc.com/
https://lore.kernel.org/linux-scsi/20230112140412.667308-1-niklas.cassel@wdc.com/
https://lore.kernel.org/linux-scsi/20221208105947.2399894-1-niklas.cassel@wdc.com/

Link: https://lore.kernel.org/r/20230511011356.227789-1-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata: Handle completion of CDL commands using policy 0xD

A CDL timeout for policy 0xF is defined as a NCQ error, just with a CDL
specific sk/asc/ascq in the sense data. Therefore, the existing code in
libata does not need to be modified to handle a policy 0xF CDL timeout.

For Command Duration Limits policy 0xD:

  The device shall complete the command without error with the additional
  sense code set to DATA CURRENTLY UNAVAILABLE.

Since a CDL timeout for policy 0xD is not an error, we cannot use the NCQ
Command Error log (10h).

Instead, we need to read the Sense Data for Successful NCQ Commands log
(0Fh).

In the success case, just like in the error case, we cannot simply read a
log page from the interrupt handler itself, since reading a log page
involves sending a READ LOG DMA EXT or READ LOG EXT command.

Therefore, we add a new EH action ATA_EH_GET_SUCCESS_SENSE.  When a command
completes without error, and when the ATA_SENSE bit is set, this new action
is set as pending, and EH is scheduled.

This way, similar to the NCQ error case, the log page will be read from EH
context.

An alternative would have been to add a new kthread or workqueue to handle
this. However, extending EH can be done with minimal changes and avoids the
need to synchronize a new kthread/workqueue with EH.

Co-developed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-20-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata: Set read/write commands CDL index

For devices supporting the command duration limits feature, translate the
dld field of read and write operation to set the command duration limit
index field of the command task file when the duration limit feature is
enabled.

The function ata_set_tf_cdl() is introduced to do this. For unqueued (non
NCQ) read and write operations, this function sets the command duration
limit index set as the lower 3 bits of the feature field.  For queued NCQ
read/write commands, the index is set as the lower 3 bits of the auxiliary
field.

The flag ATA_QCFLAG_HAS_CDL is introduced to indicate that a command
taskfile has a non zero cdl field.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Igor Pylypiv <ipylypiv@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-19-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata: Add ATA feature control sub-page translation

Add support for the ATA feature control sub-page of the control mode page
to enable/disable the command duration limits feature using the cdl_ctrl
field of the ATA feature control sub-page.

Both mode sense and mode select translation are supported. For mode sense,
the ata device flag ATA_DFLAG_CDL_ENABLED is used to cache the status of
the command duration limits feature. Enabling this feature is done using a
SET FEATURES command with a cdl action set to 1 when the page cdl_ctrl
field value is 0x2 (T2A and T2B pages supported). If this field is 0, CDL
is disabled using the SET FEATURES command with a cdl action set to 0.

Since a device CDL and NCQ priority features should not be used
simultaneously, ata_mselect_control_ata_feature() returns an error when
attempting to enable CDL with the device priority feature enabled.
Conversely, the function ata_ncq_prio_enable_store() used to enable the use
of the device NCQ priority feature through sysfs is modified to return an
error if the device CDL feature is enabled.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-18-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata-scsi: Add support for CDL pages mode sense

Modify ata_scsiop_mode_sense() and ata_msense_control() to support mode
sense access to the T2A and T2B sub-pages of the control mode page.
ata_msense_control() is modified to support sub-pages. The T2A sub-page is
generated using the read descriptors of the command duration limits log
page 18h. The T2B sub-page is generated using the write descriptors of the
same log page. With the addition of these sub-pages, getting all sub-pages
of the control mode page is also supported by increasing the value of
ATA_SCSI_RBUF_SIZE from 576B up to 2048B to ensure that all sub-pages fit
in the fill buffer.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-17-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata-scsi: Handle CDL bits in ata_scsiop_maint_in()

For a scsi MAINTENANCE_IN/MI_REPORT_SUPPORTED_OPERATION_CODES operation,
add the translation of the rwcdlp and cdlp bits for the READ 16 and WRITE
16 commands. If the ATA device does not support command duration limits,
these bits are always 0. If the ATA device supports command duration
limits, the rwcdlp bit is set to 1 for READ 16 and WRITE 16 and the cdlp
bits are set to 0x1 for READ 16 and 0x2 for WRITE 16. These correspond to
the T2A mode page containing the read descriptors and to the T2B mode page
containing the write descriptors, as defined in SAT-5.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-16-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata: Detect support for command duration limits

Use the supported capabilities identify device data log page to detect if a
device supports the command duration limits feature. For devices supporting
this feature, set the device flag ATA_DFLAG_CDL. To support SCSI-ATA
translation, retrieve the command duration limits log page 18h and cache
this page content using the cdl array added to the ata_device data
structure.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-15-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata: Change ata_eh_request_sense() to not set CHECK_CONDITION

Currently, ata_eh_request_sense() unconditionally sets the scsicmd->result
to SAM_STAT_CHECK_CONDITION.

For Command Duration Limits policy 0xD:

  The device shall complete the command without error (SAM_STAT_GOOD) with
  the additional sense code set to DATA CURRENTLY UNAVAILABLE.

It is perfectly fine to have sense data for a command that returned
completion without error.

In order to support for CDL policy 0xD, we have to remove this assumption
that having sense data means that the command failed
(SAM_STAT_CHECK_CONDITION).

Change ata_eh_request_sense() to not set SAM_STAT_CHECK_CONDITION, and
instead move the setting of SAM_STAT_CHECK_CONDITION to the single caller
that wants SAM_STAT_CHECK_CONDITION set, that way ata_eh_request_sense()
can be reused in a follow-up patch that adds support for CDL policy 0xD.

The only caller of ata_eh_request_sense() is protected by: if (!(qc->flags
& ATA_QCFLAG_SENSE_VALID)), so we can remove this duplicated check from
ata_eh_request_sense() itself.

Additionally, ata_eh_request_sense() is only called from
ata_eh_analyze_tf(), which is only called when iteratating the QCs using
ata_qc_for_each_raw(), which does not include the internal tag, so cmd can
never be NULL (all non-internal commands have qc->scsicmd set), so remove
the !cmd check as well.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-14-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: libata-scsi: Remove unnecessary !cmd checks

There is no need to check if !cmd as this can only happen for ATA internal
commands which uses the ATA internal tag (32).

Most users of ata_scsi_set_sense() are from _xlat functions that translate
a scsicmd to an ATA command. These obviously have a qc->scsicmd.

ata_scsi_qc_complete() can also call ata_scsi_set_sense() via
ata_gen_passthru_sense() / ata_gen_ata_sense(), called via
ata_scsi_qc_complete(). This callback is only called for translated
commands, so it also has a qc->scsicmd.

ata_eh_analyze_ncq_error(): the NCQ error log can only contain a 0-31
value, so it will never be able to get the ATA internal tag (32).

ata_eh_request_sense(): only called by ata_eh_analyze_tf(), which is only
called when iteratating the QCs using ata_qc_for_each_raw(), which does not
include the internal tag.

Since there is no existing call site where cmd can be NULL, remove the !cmd
check from ata_scsi_set_sense() and ata_scsi_set_sense_information().

Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-13-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Handle read/write CDL timeout failures

Commands using a duration limit descriptor that has limit policies set to a
value other than 0x0 may be failed by the device if one of the limits are
exceeded. For such commands, since the failure is the result of the user
duration limit configuration and workload, the commands should not be
retried and terminated immediately. Furthermore, to allow the user to
differentiate these "soft" failures from hard errors due to hardware
problem, a different error code than EIO should be returned.

There are 2 cases to consider:

(1) The failure is due to a limit policy failing the command with a check
condition sense key, that is, any limit policy other than 0xD.  For this
case, scsi_check_sense() is modified to detect failures with the ABORTED
COMMAND sense key and the COMMAND TIMEOUT BEFORE PROCESSING or COMMAND
TIMEOUT DURING PROCESSING or COMMAND TIMEOUT DURING PROCESSING DUE TO ERROR
RECOVERY additional sense code. For these failures, a SUCCESS disposition
is returned so that scsi_finish_command() is called to terminate the
command.

(2) The failure is due to a limit policy set to 0xD, which result in the
command being terminated with a GOOD status, COMPLETED sense key, and DATA
CURRENTLY UNAVAILABLE additional sense code. To handle this case, the
scsi_check_sense() is modified to return a SUCCESS disposition so that
scsi_finish_command() is called to terminate the command.  In addition,
scsi_decide_disposition() has to be modified to see if a command being
terminated with GOOD status has sense data.  This is as defined in SCSI
Primary Commands - 6 (SPC-6), so all according to spec, even if GOOD status
commands were not checked before.

If scsi_check_sense() detects sense data representing a duration limit,
scsi_check_sense() will set the newly introduced SCSI ML byte
SCSIML_STAT_DL_TIMEOUT. This SCSI ML byte is checked in scsi_noretry_cmd(),
so that a command that failed because of a CDL timeout cannot be
retried. The SCSI ML byte is also checked in scsi_result_to_blk_status() to
complete the command request with the BLK_STS_DURATION_LIMIT status, which
result in the user seeing ETIME errors for the failed commands.

Co-developed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-12-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Set read/write command CDL index

Introduce the command duration limits helper function sd_cdl_dld() to set
the DLD bits of READ/WRITE 16 and READ/WRITE 32 commands to indicate to the
device the command duration limit descriptor to apply to the commands.

When command duration limits are enabled, sd_cdl_dld() obtains the index of
the descriptor to apply to the command using the hints field of the request
IO priority value (hints IOPRIO_HINT_DEV_DURATION_LIMIT_1 to
IOPRIO_HINT_DEV_DURATION_LIMIT_7).

If command duration limits is disabled (which is the default), the limit
index "0" is always used to indicate "no limit" for a command.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-11-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Allow enabling and disabling command duration limits

Add the sysfs scsi_device attribute cdl_enable to allow a user to enable or
disable a device command duration limits feature. CDL is disabled by
default. This feature must be explicitly enabled by a user by setting the
cdl_enable attribute to 1.

The new function scsi_cdl_enable() does not do anything beside setting the
cdl_enable field of struct scsi_device in the case of a (real) SCSI device
(e.g. a SAS HDD). For ATA devices, the command duration limits feature
needs to be enabled/disabled using the ATA feature sub-page of the control
mode page. To do so, the scsi_cdl_enable() function checks if this mode
page is supported using scsi_mode_sense(). If it is, scsi_mode_select() is
used to enable and disable CDL.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-10-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Detect support for command duration limits

Introduce the function scsi_cdl_check() to detect if a device supports
command duration limits (CDL). Support for the READ 16, WRITE 16, READ 32
and WRITE 32 commands are checked using the function scsi_report_opcode()
to probe the rwcdlp and cdlp bits as they indicate the mode page defining
the command duration limits descriptors that apply to the command being
tested.

If any of these commands support CDL, the field cdl_supported of struct
scsi_device is set to 1 to indicate that the device supports CDL.

Support for CDL for a device is advertizes through sysfs using the new
cdl_supported device attribute. This attribute value is 1 for a device
supporting CDL and 0 otherwise.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-9-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Support Service Action in scsi_report_opcode()

The REPORT_SUPPORTED_OPERATION_CODES command allows checking for support of
commands that have the same opcode but different service actions, such as
READ 32 and WRITE 32. However, the current implementation of
scsi_report_opcode() only allows checking an operation code without a
service action differentiation.

Add the "sa" argument to scsi_report_opcode() to allow passing a service
action. If a non-zero service action is specified, the reporting options
field value is set to 3 to have the service action field taken into account
by the device. If no service action field is specified (zero), the
reporting options field is set to 1 as before.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-8-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Support retrieving sub-pages of mode pages

Allow scsi_mode_sense() to retrieve sub-pages of mode pages by adding the
subpage argument. Change all the current caller sites to specify the
subpage 0.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-7-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Rename and move get_scsi_ml_byte()

SCSI has two different getters:

 - get_XXX_byte() (in scsi_cmnd.h) which takes a struct scsi_cmnd *, and

 - XXX_byte() (in scsi.h) which takes a scmd->result.

The proper name for get_scsi_ml_byte() should thus be without the get_
prefix, as it takes a scmd->result. Rename the function to rectify this.
(This change was suggested by Mike Christie.)

Additionally, move get_scsi_ml_byte() to scsi_priv.h since both scsi_lib.c
and scsi_error.c will need to use this helper in a follow-up patch.

Cc: Mike Christie <michael.christie@oracle.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-6-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Allow libata to complete successful commands via EH

In SCSI, we get the sense data as part of the completion, for ATA however,
we need to fetch the sense data as an extra step. For an aborted ATA
command the sense data is fetched via libata's ->eh_strategy_handler().

For Command Duration Limits policy 0xD:

  The device shall complete the command without error with the additional
  sense code set to DATA CURRENTLY UNAVAILABLE.

In order to handle this policy in libata, we intend to send a successful
command via SCSI EH, and let libata's ->eh_strategy_handler() fetch the
sense data for the good command. This is similar to how we handle an
aborted ATA command, just that we need to read the Successful NCQ Commands
log instead of the NCQ Command Error log.

When we get a SATA completion with successful commands, ATA_SENSE will be
set, indicating that some commands in the completion have sense data.

The sense_valid bitmask in the Sense Data for Successful NCQ Commands log
will inform exactly which commands that had sense data, which might be a
subset of all the commands that was completed in the same completion. (Yet
all will have ATA_SENSE set, since the status is per completion.)

The successful commands that have e.g. a "DATA CURRENTLY UNAVAILABLE" sense
data will have a SCSI ML byte set, so scsi_eh_flush_done_q() will not set
the scmd->result to DID_TIME_OUT for these commands. However, the
successful commands that did not have sense data, must not get their result
marked as DID_TIME_OUT by SCSI EH.

Add a new flag SCMD_FORCE_EH_SUCCESS, which tells SCSI EH to not mark a
command as DID_TIME_OUT, even if it has scmd->result == SAM_STAT_GOOD.

This will be used by libata in a subsequent commit.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-5-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: Introduce BLK_STS_DURATION_LIMIT

Introduce the new block I/O status BLK_STS_DURATION_LIMIT for LLDDs to
report command that failed due to a command duration limit being
exceeded. This new status is mapped to the ETIME error code to allow users
to differentiate "soft" duration limit failures from other more serious
hardware related errors.

If we compare BLK_STS_DURATION_LIMIT with BLK_STS_TIMEOUT:
-BLK_STS_DURATION_LIMIT means that the drive gave a reply indicating that
the command duration limit was exceeded before the command could be
completed. This I/O status is mapped to ETIME for user space.

-BLK_STS_TIMEOUT means that the drive never gave a reply at all.
This I/O status is mapped to ETIMEDOUT for user space.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Co-developed-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-4-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: Introduce ioprio hints

I/O priorities currently only use 6-bits of the 16-bits ioprio value: the
3-upper bits are used to define up to 8 priority classes (4 of which are
valid) and the 3 lower bits of the value are used to define a priority
level for the real-time and best-effort class.

The remaining 10-bits between the I/O priority class and level are unused,
and in fact, cannot be used by the user as doing so would either result in
the value being completely ignored, or in an error returned by
ioprio_check_cap().

Use these 10-bits of an ioprio value to allow a user to specify I/O
hints. An I/O hint is defined as a 10-bitsvalue, allowing up to 1023
different hints to be specified, with the value 0 being reserved as the "no
hint" case. An I/O hint can apply to any I/O that specifies a valid
priority class other than NONE, regardless of the I/O priority level
specified.

To do so, the macros IOPRIO_PRIO_HINT() and IOPRIO_PRIO_VALUE_HINT() are
introduced in include/uapi/linux/ioprio.h to respectively allow a user to
get and set a hint in an ioprio value.

To support the ATA and SCSI command duration limits feature, 7 hints are
defined: IOPRIO_HINT_DEV_DURATION_LIMIT_1 to
IOPRIO_HINT_DEV_DURATION_LIMIT_7, allowing a user to specify which command
duration limit descriptor should be applied to the commands serving an
I/O. Specifying these hints has for now no effect whatsoever if the target
block devices do not support the command duration limits feature. However,
in the future, block I/O schedulers can be modified to optimize I/O issuing
order based on these hints, even for devices that do not support the
command duration limits feature.

Given that the 7 duration limits hints defined have no effect on any block
layer component, the actual definition of the duration limits implied by
these hints remains at the device level.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-3-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: ioprio: Clean up interface definition

The I/O priority user interface defines the 16-bits ioprio values as the
combination of the upper 3-bits for an I/O priority class and the lower
13-bits as priority data. However, the kernel only uses the lower 3-bits of
the priority data to define priority levels for the RT and BE priority
classes. The data part of an ioprio value is completely ignored for the
IDLE and NONE classes. This is enforced by checks done in
ioprio_check_cap(), which is called for all paths that allow defining an
I/O priority for I/Os: the per-context ioprio_set() system call, aio
interface and io_uring interface.

Clarify this fact in the uapi ioprio.h header file and introduce the
IOPRIO_PRIO_LEVEL_MASK and IOPRIO_PRIO_LEVEL() macros for users to define
and get priority levels in an ioprio value. The coarser macro
IOPRIO_PRIO_DATA() is retained for backward compatibility with old
applications already using it. There is no functional change introduced
with this.

In-kernel users of the IOPRIO_PRIO_DATA() macro which are explicitly
handling I/O priority data as a priority level are modified to use the new
IOPRIO_PRIO_LEVEL() macro without any functional change. Since f2fs is the
only user of this macro not explicitly using that value as a priority
level, it is left unchanged.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230511011356.227789-2-nks@flawful.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "Use block pr_ops in LIO"

Mike Christie <michael.christie@oracle.com> says:

The patches in this thread allow us to use the block pr_ops with LIO's
target_core_iblock module to support cluster applications in VMs. They
were built over Linus's tree. They also apply over linux-next and
Martin's tree and Jens's trees.

Currently, to use windows clustering or linux clustering (pacemaker +
cluster labs scsi fence agents) in VMs with LIO and vhost-scsi, you
have to use tcmu or pscsi or use a cluster aware FS/framework for the
LIO pr file. Setting up a cluster FS/framework is pain and waste when
your real backend device is already a distributed device, and pscsi
and tcmu are nice for specific use cases, but iblock gives you the
best performance and allows you to use stacked devices like
dm-multipath. So these patches allow iblock to work like pscsi/tcmu
where they can pass a PR command to the backend module. And then
iblock will use the pr_ops to pass the PR command to the real devices
similar to what we do for unmap today.

The patches are separated in the following groups:

Patch 1 - 2:

 - Add block layer callouts for reading reservations and rename reservation
   error code.

Patch 3 - 5:

 - SCSI support for new callouts.

Patch 6:

 - DM support for new callouts.

Patch 7 - 13:

 - NVMe support for new callouts.

Patch 14 - 18:

 - LIO support for new callouts.

This patchset has been tested with the libiscsi PGR ops and with
window's failover cluster verification test. Note that for scsi
backend devices we need this patchset:

https://lore.kernel.org/linux-scsi/20230123221046.125483-1-michael.christie@oracle.com/T/#m4834a643ffb5bac2529d65d40906d3cfbdd9b1b7

to handle UAs. To reduce the size of this patchset that's being done
separately to make reviewing easier. And to make merging easier this
patchset and the one above do not have any conflicts so can be merged
in different trees.

Link: https://lore.kernel.org/r/20230407200551.12660-1-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: dc395x: Documentation: Reword original driver attribution

The Linux kernel isn't in 2.6.x anymore, but rather the major version
has advanced much (currently 6.x). Reword the attribution.

Also, replace 404'ed 2.4 driver link with web.archive.org snapshot [1].

Link: https://web.archive.org/web/20140129181343/http://www.garloff.de/kurt/linux/dc395/
Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Link: https://lore.kernel.org/r/20230510093933.19985-4-bagasdotme@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: dc395x: Documentation: Replace non-functional twibble.org list

Sync mailing list address in the documentation to follow MAINTAINERS.

Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Link: https://lore.kernel.org/r/20230510093933.19985-3-bagasdotme@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: MAINTAINERS: Drop DC395x list and site

Emails to DC395x list bounce (550 error) and visiting the site returns 404
page.

Drop both twibble.org links. The driver should now be covered by linux-scsi
list.

Signed-off-by: Bagas Sanjaya <bagasdotme@gmail.com>
Link: https://lore.kernel.org/r/20230510093933.19985-2-bagasdotme@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: MAINTAINERS: Add a libsas entry

John has been reviewing libsas patches for years. And I have been
contributing to libsas for years and I am interested in reviewing and
testing libsas patches too. So add a libsas entry and add John and me as
reviewer.

Cc: John Garry <john.g.garry@oracle.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Link: https://lore.kernel.org/r/20230516110131.388634-1-yanaijie@huawei.com
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Acked-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230516025404.2843867-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla4xxx: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230516025355.2835898-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230516025322.2804923-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bfa: Replace all non-returning strlcpy() with strscpy()

strlcpy() reads the entire source buffer first.  This read may exceed the
destination size limit.  This is both inefficient and can lead to linear
read overflows if a source string is not NUL-terminated [1].  In an effort
to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.

[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89

Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
Link: https://lore.kernel.org/r/20230516013345.723623-1-azeemshaikh38@gmail.com
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "scsi: hisi_sas: Some misc changes"

Xiang Chen <chenxiang66@hisilicon.com> says:

This series contains some fixes including:

 - Configure initial value of some registers according to HBA model

 - Change DMA setup lock timeout from 100ms to 2.5s

 - Fix warnings detected by sparse

Link: https://lore.kernel.org/r/1684118481-95908-1-git-send-email-chenxiang66@hisilicon.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: Fix warnings detected by sparse

This patch fixes the following warning:

drivers/scsi/hisi_sas/hisi_sas_v3_hw.c:2168:43: sparse: sparse: restricted __le32 degrades to integer

Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/oe-kbuild-all/202304161254.NztCVZIO-lkp@intel.com/
Signed-off-by: Xingui Yang <yangxingui@huawei.com>
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Link: https://lore.kernel.org/r/1684118481-95908-4-git-send-email-chenxiang66@hisilicon.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: Change DMA setup lock timeout to 2.5s

DMA setup lock timeout protection is added when DMA setup frames are
received. It's a function outside the protocol and used to prevent SATA
disk I/Os from being delivered for a long time. The default value is 100ms,
it's too strict and easily triggered timeout when the disk is overloaded or
faulty. Based on the average I/O latency of 300 disks, we adjust the value
to 2.5s.

Signed-off-by: Xingui Yang <yangxingui@huawei.com>
Signed-off-by: Yihang Li <liyihang9@huawei.com>
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Link: https://lore.kernel.org/r/1684118481-95908-3-git-send-email-chenxiang66@hisilicon.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: Configure initial value of some registers according to HBA model

For SAS HBAs of 920 and previous version, we use init_reg_v3_hw() to set
some registers which are related to HW boards. For SAS HBAs of 920B and
later version, those HW registers are set through firmware. And different
HBA models are distinguished through pci_dev->revision.

Signed-off-by: Yihang Li <liyihang9@huawei.com>
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Link: https://lore.kernel.org/r/1684118481-95908-2-git-send-email-chenxiang66@hisilicon.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Convert union megasas_sgl to flex-arrays

In the ongoing effort to replace all fake flexible arrays with true
flexible arrays, replace the sge32, sge64, and sge_skinny members of union
megasas_sgl with true flexible arrays. No binary differences are seen after
this change; sizes were already being manually calculated using the member
struct sizes directly.

Cc: Kashyap Desai <kashyap.desai@broadcom.com>
Cc: Sumit Saxena <sumit.saxena@broadcom.com>
Cc: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Cc: "James E.J. Bottomley" <jejb@linux.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: Gustavo A. R. Silva <gustavoars@kernel.org>
Cc: megaraidlinux.pdl@broadcom.com
Cc: linux-scsi@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20230511220957.never.919-kees@kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: hwmon: Constify pointers to hwmon_channel_info

Statically allocated array of pointers to hwmon_channel_info can be made
const for safety.

Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20230511175204.281038-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "smartpqi updates"

Don Brace <don.brace@microchip.com> says:

These patches are based on Martin Petersen's 6.4/scsi-queue tree
  https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git
  6.4/scsi-queue

This set of changes consists of:

 * Map entire BAR 0.  The driver was mapping up to and including the
   controller registers, but not all of BAR 0.

 * Add PCI IDs to support new controllers.

 * Clean up some code by removing unnecessary NULL checks.  This cleanup is
   a result of a Coverity report.

 * Correct a rare memory leak whenever pqi_sas_port_add_rhpy() returns an
   error. This was Suggested by: Yang Yingliang <yangyingliang@huawei.com>

 * Remove atomic operations on variable raid_bypass_cnt. Accuracy is not
   required for driver operation. Change type from atomic_t to unsigned
   int.

 * Correct a rare drive hot-plug removal issue where we get a NULL
   io_request. We added a check for this condition.

 * Turn on NCQ priority for AIO requests to disks comprising RAID devices.

 * Correct byte aligned writew() operations on some ARM servers. Changed
   the writew() to two writeb() operations.

 * Change how the driver checks for a sanitize operation in progress.  We
   were using TEST UNIT READY. We removed the TEST UNIT READY code and are
   now using the controller's firmware information in order to avoid issues
   caused by drives failing to complete TEST UNIT READY.

 * Some customers have been requesting that we add the NUMA node to
   /sys/block/sd<scsi device>/device like the nvme driver does.

 * Update the copyright information to match the current year.

 * Bump the driver version to 2.1.22-040.

Link: https://lore.kernel.org/r/20230428153712.297638-1-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "scsi: pm80xx: Enhanced debug logs for HW events"

Pranav Prasad <pranavpp@google.com> says:

This patch series enhances debug logs for pm80xx HW events, and provides a
minor fix in the case of a hard reset. The log enhancement involves changing
the log severity level to enable logging for HW events which consequently
help debug disk discovery issues.

 1. Changed log severity level from MSG to EVENT for HW events.  Enhanced
    the HW event logs by adding the phyid.

 2. Enabled INIT logging.

 3. Log portid along with the PHY_UP event.

 4. Print phyid and portid sent as part of device registration request.

 5. Log port state during HW events.

 6. Update phy_state and phy_attached to correct values after a hard reset.

Link: https://lore.kernel.org/r/20230418190101.696345-1-pranavpp@google.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "scsi: libsas: remove empty branches and code simplification"

Jason Yan <yanaijie@huawei.com> says:

Three patches to remove two empty branches and a little code simplification.

Link: https://lore.kernel.org/r/20230421093744.1583609-1-yanaijie@huawei.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "qla2xxx driver update"

Nilesh Javali <njavali@marvell.com> says:

Please apply the qla2xxx driver enhancement and bug fixes to the scsi tree
at your earliest convenience.

Link: https://lore.kernel.org/r/20230428075339.32551-1-njavali@marvell.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "lpfc: Update lpfc to revision 14.2.0.12"

Justin Tee <justintee8345@gmail.com> says:

Update lpfc to revision 14.2.0.12

This patch set contains fixes flagged by code analyzer tools, introduces a
new CQE status to handle DMA errors, and replaces the usage of blk
interrupts with threaded interrupts.

The patches were cut against Martin's 6.4/scsi-queue tree.

Link: https://lore.kernel.org/r/20230417191558.83100-1-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufs-mediatek: Delete some dead code

There is already a test for "if (val == state)" earlier so it's not
possible here.  Delete the dead code.

Fixes: 9006e3986f66 ("scsi: ufs-mediatek: Do not gate clocks if auto-hibern8 is not entered yet")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Link: https://lore.kernel.org/r/68fce64f-4970-45f1-807e-6c0eecdfcdc2@kili.mountain
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Fix NULL dereference in error handling

Smatch reported:

drivers/scsi/qedf/qedf_main.c:3056 qedf_alloc_global_queues()
warn: missing unwind goto?

At this point in the function, nothing has been allocated so we can return
directly. In particular the "qedf->global_queues" have not been allocated
so calling qedf_free_global_queues() will lead to a NULL dereference when
we check if (!gl[i]) and "gl" is NULL.

Fixes: 61d8658b4a43 ("scsi: qedf: Add QLogic FastLinQ offload FCoE driver framework.")
Signed-off-by: Jinhong Zhu <jinhongzhu@hust.edu.cn>
Link: https://lore.kernel.org/r/20230502140022.2852-1-jinhongzhu@hust.edu.cn
Reviewed-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: core: Change the module parameter macro of use_mcq_mode

mcq_mode_ops uses only param_{set,get}_bool(). Therefore, convert
module_param_cb() to module_param() and remove the mcq_mode_ops.

Signed-off-by: Keoseong Park <keosung.park@samsung.com>
Link: https://lore.kernel.org/r/20230427094420epcms2p1043333a3e0c0cf58e66164e0b83b3b02@epcms2p1
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: Use -ENOMEM instead of -1 in mpi3mr_expander_add()

smatch warnings:

drivers/scsi/mpi3mr/mpi3mr_transport.c:1449 mpi3mr_expander_add() warn:
returning -1 instead of -ENOMEM is sloppy

No functional change.

Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <error27@gmail.com>
Link: https://lore.kernel.org/r/202303202027.ZeDQE5Ug-lkp@intel.com/
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Link: https://lore.kernel.org/r/20230419064256.2532069-1-harshit.m.mogalapalli@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Update version to 2.1.22-040

Reviewed-by: Gerry Morong <gerry.morong@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-13-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Update copyright to 2023

Update copyright to current year.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-12-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add sysfs entry for NUMA node in /sys/block/sdX/device

Although NUMA node is a PCIe device level attribute, it was requested the
NUMA node be added for each exposed device similar to NVMe disks.

Example for NVMe:

  /sys/block/nvme1c1n1/device/numa_node

Example for smartpqi:

  /sys/block/sdh/device/numa_node

  cat /sys/block/sdh/device/numa_node
  0

Reviewed-by: David Strahan <david.strahan@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-11-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Stop sending driver-initiated TURs

Stop sending driver-initiated TURs to physical devices during driver
load/rescan.

Note: This does not affect SML initiated TURs.

Some Linux kernels can cause lengthy delays in OS boot if the kernel
detects that a drive is being sanitized/erased. We were using TURs to
detect if a sanitize/erase was in progress.

Some devices do not return the TUR in a timely manner, causing driver
load/rescan stalls.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-10-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Fix byte aligned writew for ARM servers

Correct OOPs on ARM servers during driver init.

The driver attempts to update FW with max_feature_supported value using a
writew() kernel call using a byte aligned address. This fails on some ARM
systems.

Change the writew() to two writeb() calls to update this value.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-9-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add support for RAID NCQ priority

Enable NCQ priority feature for the RAID path when AIO path is disabled.

Move function pqi_is_io_high_priority() up to avoid adding a prototype.
Remove unused argument ctrl_info.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Gilbert Wu <Gilbert.Wu@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-8-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Validate block layer host tag

Prevent OS crashes when a drive is hot removed during I/O stress test.

The I/O request pointer can be invalid if block layer provides incorrect
multi-queue host tag. This can lead to invalid I/O request pointer
dereference.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-7-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Remove contention for raid_bypass_cnt

Reduce CPU contention when incrementing variable raid_bypass_cnt.

Remove the atomic operations for this variable by changing the atomic to an
unsigned int and replace atomic operations with standard operations. The
value is only checked that it is increasing and accuracy is not required.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-6-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Fix rare SAS transport memory leak

Free rphy when pqi_sas_port_add_rphy() returns an error.

If pqi_sas_port_add_rphy() returns an error, the 'rphy' allocated in
sas_end_device_alloc() needs to be freed.

It should be noted that no issues were ever reported.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Suggested-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-5-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Remove NULL pointer check

Remove an unnecessary check for a NULL pointer.  This unnecessary check was
flagged by Coverity.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-4-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add new controller PCI IDs

All PCI ID entries in Hex.

Add PCI IDs for ZTE controllers:
                                            VID  / DID  / SVID / SDID
                                            ----   ----   ----   ----
    ZTE SmartROC3200 RS344-16i 4G           9005 / 028f / 1cf2 / 0804
    ZTE SmartROC3200 RS345-16i 8G           9005 / 028f / 1cf2 / 0805
    ZTE SmartIOC2200 RS346-16i              9005 / 028f / 1cf2 / 0806
    ZTE SmartROC3200 RM344-16i 4G           9005 / 028f / 1cf2 / 54da
    ZTE SmartROC3200 RM345-16i 8G           9005 / 028f / 1cf2 / 54db
    ZTE SmartIOC2200 RM346-16i              9005 / 028f / 1cf2 / 54dc

Add PCI IDs for ByteDance controllers:
                                            VID  / DID  / SVID / SDID
                                            ----   ----   ----   ----
    ByteHBA JGH43014-8                      9005 / 028f / 1e93 / 1005

Add PCI IDs for IBM controllers:
                                            VID  / DID  / SVID / SDID
                                            ----   ----   ----   ----
    IBM 4-Port 24G SAS                      9005 / 028f / 1014 / 0718

Add PCI IDs for Cloudnine controllers:
                                            VID  / DID  / SVID / SDID
                                            ----   ----   ----   ----
    SmartHBA P6600-8i                       9005 / 028f / 1f51 / 1001
    SmartRAID P7604-8i                      9005 / 028f / 1f51 / 1002
    SmartHBA P6600-8e                       9005 / 028f / 1f51 / 1003
    SmartRAID P7604-8e                      9005 / 028f / 1f51 / 1004
    SmartHBA P6600-16i                      9005 / 028f / 1f51 / 1005
    SmartRAID P7608-16i                     9005 / 028f / 1f51 / 1006
    SmartHBA P6600-8i8e                     9005 / 028f / 1f51 / 1007
    SmartRAID P7608-8i8e                    9005 / 028f / 1f51 / 1008
    SmartHBA P6600-16e                      9005 / 028f / 1f51 / 1009
    SmartRAID P7608-16e                     9005 / 028f / 1f51 / 100a

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: David Strahan <David.Strahan@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20230428153712.297638-3-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>