scsi: sd: Retry START STOP UNIT commands

During system resume, sd_start_stop_device() submits a START STOP UNIT
command to the SCSI device that is being resumed. That command is not
retried in case of a unit attention and hence may fail. An example:

[16575.983359] sd 0:0:0:3: [sdd] Starting disk
[16575.983693] sd 0:0:0:3: [sdd] Start/Stop Unit failed: Result: hostbyte=0x00 driverbyte=DRIVER_OK
[16575.983712] sd 0:0:0:3: [sdd] Sense Key : 0x6
[16575.983730] sd 0:0:0:3: [sdd] ASC=0x29 ASCQ=0x0
[16575.983738] sd 0:0:0:3: PM: dpm_run_callback(): scsi_bus_resume+0x0/0xa0 returns -5
[16575.983783] sd 0:0:0:3: PM: failed to resume async: error -5

Make the SCSI core retry the START STOP UNIT command if the device reports
that it has been powered on or that it has been reset.

Cc: Damien Le Moal <dlemoal@kernel.org>
Cc: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240904210304.2947789-1-bvanassche@acm.org
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: A performance fix

Commit 0c52310f2600 ("hrtimer: Ignore slack time for RT tasks in
schedule_hrtimeout_range()") effectivelly shortens a sleep in a polling
function in the driver.  That is causing a performance regression as the
new value of just 2us is too low, in certain tests the perf drop is ~30%.
Fix this by adjusting the sleep to 20us (close to the previous value).

Reported-by: Jan Jurca <jjurca@redhat.com>
Signed-off-by: Tomas Henzl <thenzl@redhat.com>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Link: https://lore.kernel.org/r/20240903144729.37218-1-thenzl@redhat.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: qcom: Update MODE_MAX cfg_bw value

Commit 8db8f6ce556a ("scsi: ufs: qcom: Add missing interconnect bandwidth
values for Gear 5") updated the ufs_qcom_bw_table for Gear 5. However, it
missed updating the cfg_bw value for the max mode.

Hence update the cfg_bw value for the max mode for UFS 4.x devices.

Fixes: 8db8f6ce556a ("scsi: ufs: qcom: Add missing interconnect bandwidth values for Gear 5")
Cc: stable@vger.kernel.org
Signed-off-by: Manish Pandey <quic_mapa@quicinc.com>
Link: https://lore.kernel.org/r/20240903063709.4335-1-quic_mapa@quicinc.com
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Remove trailing space after \n newline

There is a extraneous space after a newline in an ioc_info message.
Remove it and join to split literal strings into one.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902172708.369741-1-colin.i.king@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Remove trailing space after \n newline

There is a extraneous space after a newline in two lpfc_printf_log()
messages. Remove the space.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902150042.311157-1-colin.i.king@gmail.com
Reviewed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Remove trailing space after \n newline

There is a extraneous space after a newline in a QEDF_INFO message.
Remove it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902145138.310883-1-colin.i.king@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: Remove trailing space after \n newline

There is a extraneous space after a newline in a dev_info message.
Remove it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902144153.309920-1-colin.i.king@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Remove trailing space after \n newline

There is a extraneous space after a newline in a dev_err message.
Remove it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902142252.309232-1-colin.i.king@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: pm8001: Remove trailing space after \n newline

There is a extraneous space after a newline in a pm8001_dbg message.
Remove it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902141537.308914-1-colin.i.king@gmail.com
Acked-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: zalon: Remove trailing space after \n newline

There is a extraneous space after a newline in a dev_printk message,
remove it. Also fix non-tabbed indentation of the statement.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20240902141202.308632-1-colin.i.king@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: fusion: mptctl: Use min() macro

Using the real macro is usually more intuitive and readable when the
original file is guaranteed to contain the minmax.h header file and compile
correctly.

Signed-off-by: Yan Zhen <yanzhen@vivo.com>
Link: https://lore.kernel.org/r/20240902013303.909316-1-yanzhen@vivo.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libcxgbi: Remove an unused field in struct cxgbi_device

Usage of .dev_ddp_cleanup() in libcxgbi was removed by commit 5999299f1ce9
("cxgb3i,cxgb4i,libcxgbi: remove iSCSI DDP support") on 2016-07.

.csk_rx_pdu_ready() and debugfs_root have apparently never been used since
introduction by commit 9ba682f01e2f ("[SCSI] libcxgbi: common library for
cxgb3i and cxgb4i")

Remove the now unused function pointer from struct cxgbi_device.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/58f77f690d85e2c653447e3e3fc4f8d3c3ce8563.1725223504.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ibmvfc: Add max_sectors module parameter

There are some scenarios that can occur, such as performing an upgrade of
the virtual I/O server, where the supported max transfer of the backing
device for an ibmvfc HBA can change. If the max transfer of the backing
device decreases, this can cause issues with previously discovered
LUNs. This patch accomplishes two things.  First, it changes the default
ibmvfc max transfer value to 1MB.  This is generally supported by all
backing devices, which should mitigate this issue out of the box. Secondly,
it adds a module parameter, enabling a user to increase the max transfer
value to values that are larger than 1MB, as long as they have configured
these larger values on the virtual I/O server as well.

[mkp: fix checkpatch warnings]

Signed-off-by: Brian King <brking@linux.ibm.com>
Link: https://lore.kernel.org/r/20240903134708.139645-2-brking@linux.ibm.com
Reviewed-by: Martin Wilck <mwilck@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Remove duplicate included header file linux/bio-integrity.h

The header file linux/bio-integrity.h is included twice.  Remove the last
one. The compilation test has passed.

Signed-off-by: Hongbo Li <lihongbo22@huawei.com>
Link: https://lore.kernel.org/r/20240830075858.3541907-1-lihongbo22@huawei.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: st: Fix input/output error on empty drive reset

A previous change was introduced to prevent data loss during a power-on
reset when a tape is present inside the drive. This commit set the
"pos_unknown" flag to true to avoid operations that could compromise data
by performing actions from an untracked position. The relevant change is
commit 9604eea5bd3a ("scsi: st: Add third party poweron reset handling")

As a consequence of this change, a new issue has surfaced: the driver now
returns an "Input/output error" even for empty drives when the drive, host,
or bus is reset. This issue stems from the "flush_buffer" function, which
first checks whether the "pos_unknown" flag is set. If the flag is set, the
user will encounter an "Input/output error" until the tape position is
known again. This behavior differs from the previous implementation, where
empty drives were not affected at system start up time, allowing tape
software to send commands to the driver to retrieve the drive's status and
other information.

The current behavior prioritizes the "pos_unknown" flag over the
"ST_NO_TAPE" status, leading to issues for software that detects drives
during system startup. This software will receive an "Input/output error"
until a tape is loaded and its position is known.

To resolve this, the "ST_NO_TAPE" status should take priority when the
drive is empty, allowing communication with the drive following a power-on
reset. At the same time, the change should continue to protect data by
maintaining the "pos_unknown" flag when the drive contains a tape and its
position is unknown.

Signed-off-by: Rafael Rocha <rrochavi@fnal.gov>
Link: https://lore.kernel.org/r/20240905173921.10944-1-rrochavi@fnal.gov
Fixes: 9604eea5bd3a ("scsi: st: Add third party poweron reset handling")
Acked-by: Kai Mäkisara <kai.makisara@kolumbus.fi>
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 6.12/scsi-queue tree
  https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git
  6.12/scsi-queue

There are two functional changes:
    smartpqi-add-fw-log-to-kdump
    smartpqi-add-counter-for-parity-write-stream-requests

There are three minor bug fixes:
    smartpqi-fix-stream-detection
    smartpqi-fix-rare-system-hang-during-LUN-reset
    smartpqi-fix-volume-size-updates

The other two patches add PCI-IDs for new controllers and change the
driver version.

This set of changes consists of:
* smartpqi-add-fw-log-to-kdump

  During a kdump, the driver tells the controller to copy its logging
  information to some pre-allocated buffers that can be analyzed
  later.

  This is a "feature" driven capability and is backward compatible
  with existing controller FW.

  This patch renames some prefixes for OFA (Online-Firmware Activation
  ofa_*) buffers to host_memory_*. So, not a lot of actual functional
  changes to smartpqi_init.c, mainly determining the memory size
  allocation.

  We added a function to notify the controller to copy debug data into
  host memory before continuing kdump.

  Most of the functional changes are in smartpqi_sis.c where the
  actual handshaking is done.

* smartpqi-fix-stream-detection

  Correct some false write-stream detections. The data structure used
  to check for write-streams was not initialized to all 0's causing
  some false write stream detections. The driver sends down streamed
  requests to the raid engine instead of using AIO bypass for some
  extra performance.  (Potential full-stripe write verses Read Modify
  Write).

  False detections have not caused any data corruption.  Found by
  internal testing. No known externally reported bugs.

* smartpqi-add-counter-for-parity-write-stream-requests

  Adding some counters for raid_bypass and write streams. These two
  counters are related because write stream detection is only checked
  if an I/O request is eligible for bypass (AIO).

  The bypass counter (raid_bypass_cnt) was moved into a common
  structure (pqi_raid_io_stats) and changed to type __percpu. The
  write stream counter is (write_stream_cnt) has been added to this
  same structure.

  These counters are __percpu counters for performance. We added a
  sysfs entry to show the write stream count. The raid bypass counter
  sysfs entry already exists.

  Useful for checking streaming writes. The change in the sysfs entry
  write_stream_cnt can be checked during AIO eligible write
  operations.

* smartpqi-add-new-controller-PCI-IDs

  Adding support for new controller HW.  No functional changes.

* smartpqi-fix-rare-system-hang-during-LUN-reset

  We found a rare race condition that can occur during a LUN reset. We
  were not emptying our internal queue completely.

  There have been some rare conditions where our internal request
  queue has requests for multiple LUNs and a reset comes in for one of
  the LUNs. The driver waits for this internal queue to empty. We were
  only clearing out the requests for the LUN being reset so the
  request queue was never empty causing a hang.

  The Fix:

     For all requests in our internal request queue:

        Complete requests with DID_RESET for queued requests for the
        device undergoing a reset.

        Complete requests with DID_REQUEUE for all other queued requests.

  Found by internal testing. No known externally reported bugs.

* smartpqi-fix-volume-size-updates

  The current code only checks for a size change if there is also a
  queue depth change.  We are separating the check for queue depth and
  the size changes.

  Found by internal testing. No known bugs were filed.

* smartpqi-update-version-to-2.1.30-031
  No functional changes.

Link: https://lore.kernel.org/r/20240827185501.692804-1-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: update driver version to 2.1.30-031

Update driver version to 2.1.30-031.

Reviewed-by: David Strahan <david.strahan@microchip.com>
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: Gerry Morong <gerry.morong@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240827185501.692804-8-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: fix volume size updates

Correct logical volume size changes by moving the check for a volume rescan
outside of the check for a queue depth change.

Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240827185501.692804-7-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: fix rare system hang during LUN reset

Correct a rare case where in a LUN reset occurs on a device and I/O
requests for other devices persist in the driver's internal request queue.

Part of a LUN reset involves waiting for our internal request queue to
empty before proceeding. The internal request queue contains requests not
yet sent down to the controller.

We were clearing the requests queued for the LUN undergoing a reset, but
not all of the queued requests. Causing a hang.

For all requests in our internal request queue:

   Complete requests with DID_RESET for queued requests for the device
   undergoing a reset.

   Complete requests with DID_REQUEUE for all other queued requests.

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: Murthy Bhat <Murthy.Bhat@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240827185501.692804-6-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 new cisco pci ids:
                                             VID  / DID  / SVID / SDID
                                             ----   ----   ----   ----
                                             9005   028f   1137   02fe
                                             9005   028f   1137   02ff
                                             9005   028f   1137   0300

Add new h3c pci ids:
                                             VID  / DID  / SVID / SDID
                                             ----   ----   ----   ----
                                             9005   028f   193d   0462
                                             9005   028f   193d   8462

Add new ieit pci ids:
                                             VID  / DID  / SVID / SDID
                                             ----   ----   ----   ----
                                             9005   028f   1ff9   00a3

Reviewed-by: Scott Benesh <scott.benesh@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/20240827185501.692804-5-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: add counter for parity write stream requests

Add sysfs entry to check for write stream requests.

Move existing raid_bypass_cnt into a structure named pqi_raid_io_stats and
add member write_stream_cnt. These two counters are related because write
stream detection is only checked if an I/O request is eligible for bypass
(AIO).

Example usage:

lsscsi
[15:1:0:0]   disk    Adaptec  LOGICAL VOLUME   0129  /dev/sdae

cat /sys/block/sdae/device/ssd_smart_path_enabled
1
^
|
+---- NOTE: here bypass has been enabled on device sdae

To read the counter for parity write stream requests:

cat /sys/block/sdae/device/write_stream_cnt
0x60cd507

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Co-developed-by: Kevin Barnett <kevin.barnett@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/20240827185501.692804-4-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: correct stream detection

Correct stream detection by initializing the structure
pqi_scsi_dev_raid_map_data to 0s.

When the OS issues SCSI READ commands, the driver erroneously considers
them as SCSI WRITES. If they are identified as sequential IOs, the driver
then submits those requests via the RAID path instead of the AIO path.

The 'is_write' flag might be set for SCSI READ commands also.  The driver
may interpret SCSI READ commands as SCSI WRITE commands, resulting in IOs
being submitted through the RAID path.

Note: This does not cause data corruption.

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: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240827185501.692804-3-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add fw log to kdump

Add controller logs to kdump.

Driver allocates DMA memory and communicates this address to FW.  In the
event of system crash, host driver notifies the firmware about the crash
and firmware posts all the necessary logs in the pre-allocated host buffer
for firmware debugging.

Once firmware notifies the completion of the log uploading to the host
memory and host continues with the OS crash dump saving.

This is a "feature" driven capability and is backward compatible with
existing controller FW.

Rename some prefixes for OFA (Online-Firmware Activation ofa_*) buffers to
host_memory_*. So, not a lot of actual functional changes to
smartpqi_init.c, mainly determining the memory size allocation.

Added a function to notify the controller to copy debug data into host
memory before continuing kdump.

Most of the functional changes are in smartpqi_sis.c where the actual
handshaking is done.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@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/20240827185501.692804-2-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bnx2fc: Remove some unused fields in struct bnx2fc_rport

Some fields are unused in struct bnx2fc_rport.  Remove them in order to
save 96 bytes on a x86_64.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/42e20b159f3bbb12da7796463a521ca051bd5274.1724399924.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove the unused 'del_list_entry' field in struct fc_port

The 'del_list_entry' field in "struct fc_port" is unused.

The field was introduced in commit 2d70c103fd2a ("[SCSI] qla2xxx: Add LLD
target-mode infrastructure for >= 24xx series") in 2012-05 and the last
user was removed in commit 726b85487067 ("qla2xxx: Add framework for async
fabric discovery") in 2017-02.

Remove this unused field.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/69155321ab26c1f4d473d5bb6cd44b59b9b6a020.1724094686.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: core: Remove ufshcd_urgent_bkops()

ufshcd_urgent_bkops() is a wrapper function. It only calls
ufshcd_bkops_ctrl(). Remove it to simplify the ufs core driver. Replace any
references to ufshcd_urgent_bkops() with ufshcd_bkops_ctrl().

In addition, remove the second parameter in the ufshcd_bkops_ctrl() because
the information can be retrieved from the first parameter.

Signed-off-by: Bao D. Nguyen <quic_nguyenb@quicinc.com>
Link: https://lore.kernel.org/r/0c7f2c8d68408e39c28e3e81addce09cc0ee3969.1724800328.git.quic_nguyenb@quicinc.com
Acked-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Remove obsoleted declaration for scsi_driverbyte_string()

scsi_driverbyte_string() has been unused since commit 54c29086195f ("scsi:
core: Drop the now obsolete driver_byte definitions"). Remove it.

Signed-off-by: Gaosheng Cui <cuigaosheng1@huawei.com>
Link: https://lore.kernel.org/r/20240826032005.4007834-1-cuigaosheng1@huawei.com
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bnx2i: Remove unused declarations

Commit cf4e6363859d ("[SCSI] bnx2i: Add bnx2i iSCSI driver.") declared but
never implemented these.

Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/r/20240824084724.3647307-1-yuehaibing@huawei.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "Simplify multiple create*_workqueue() invocations"

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

Hi Martin,

Multiple SCSI drivers use snprintf() to format a workqueue name before
invoking one of the create*_workqueue() macros. This patch series
simplifies such code by passing the format string and arguments to
alloc_workqueue(). Additionally, the structure members that are only
used as a temporary buffer for formatting workqueue names are
removed. Please consider this patch series for the next merge window.

Thanks,

Bart.

Link: https://lore.kernel.org/r/20240822195944.654691-1-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Simplify an alloc_workqueue() invocation

Let alloc_workqueue() format the workqueue name. Remove the
work_q_name[] member from struct Scsi_Host because it is no longer
used by any SCSI driver nor by the SCSI core.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-19-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Simplify alloc*_workqueue() invocation

Let alloc*_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Peter Wang <peter.wang@mediatek.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-18-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: stex: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-17-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_transport_fc: Simplify alloc_workqueue() invocations

Let alloc_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-16-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: snic: Simplify alloc_workqueue() invocations

Let alloc_workqueue() format the workqueue name instead of calling
snprintf() explicitly. Not setting shost->work_q_name is safe because
there is no code that reads the value set by the removed code.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-15-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Simplify an alloc_workqueue() invocation

Let alloc_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-14-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Simplify alloc_workqueue() invocations

Let alloc_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-13-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: myrs: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-12-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: myrb: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-11-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-10-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-9-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ibmvscsi_tgt: Simplify an alloc_workqueue() invocation

Let alloc_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-8-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: fcoe: Simplify alloc_ordered_workqueue() invocations

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-7-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: esas2r: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-6-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bfa: Simplify an alloc_ordered_workqueue() invocation

Let alloc_ordered_workqueue() format the workqueue name instead of
calling snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-5-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: be2iscsi: Simplify an alloc_workqueue() invocation

Let alloc_workqueue() format the workqueue name instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-4-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mptfusion: Simplify the alloc*_workqueue() invocations

Let alloc*_workqueue() format the workqueue names instead of calling
snprintf() explicitly.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-3-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: Expand all create*_workqueue() invocations

The workqueue maintainer wants to remove the create*_workqueue() macros
because these macros always set the WQ_MEM_RECLAIM flag and because these
only support literal workqueue names. Hence this patch that replaces the
create*_workqueue() invocations with the definition of this macro. The
WQ_MEM_RECLAIM flag has been retained because I think that flag is necessary
for workqueues created by storage drivers. This patch has been generated by
running spatch and git clang-format. spatch has been invoked as follows:

spatch --in-place --sp-file expand-create-workqueue.spatch $(git grep -lEw 'create_(freezable_|singlethread_|)workqueue' */scsi */ufs)

The contents of the expand-create-workqueue.spatch file is as follows:

@@
expression name;
@@
-create_workqueue(name)
+alloc_workqueue("%s", WQ_MEM_RECLAIM, 1, name)
@@
expression name;
@@
-create_freezable_workqueue(name)
+alloc_workqueue("%s", WQ_FREEZABLE | WQ_UNBOUND | WQ_MEM_RECLAIM, 1, name)
@@
expression name;
@@
-create_singlethread_workqueue(name)
+alloc_ordered_workqueue("%s", WQ_MEM_RECLAIM, name)

Reviewed-by: Peter Wang <peter.wang@mediatek.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240822195944.654691-2-bvanassche@acm.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target: Remove unused declarations

Commit 13247018d68f ("scsi: target: iscsi: Fix hang in the iSCSI login
code") removed iscsi_handle_login_thread_timeout() but left declaration.

Commit 3e1c81a95f0d ("iscsi-target: Refactor RX PDU logic + export request
PDU handling") left iscsi_target_get_initial_payload() declaration.

Commit d703ce2f7f4d ("iscsi/iser-target: Convert to command priv_size
usage") remove iscsit_alloc_cmd() but left declaration.

And finally, a few other declarations were never implenmented since
introduction in commit e48354ce078c ("iscsi-target: Add iSCSI fabric
support for target v4.1").

Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/r/20240810093437.2586476-1-yuehaibing@huawei.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: elx: libefc: Fix potential use after free in efc_nport_vport_del()

The kref_put() function will call nport->release if the refcount drops to
zero.  The nport->release release function is _efc_nport_free() which frees
"nport".  But then we dereference "nport" on the next line which is a use
after free.  Re-order these lines to avoid the use after free.

Fixes: fcd427303eb9 ("scsi: elx: libefc: SLI and FC PORT state machine interfaces")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Link: https://lore.kernel.org/r/b666ab26-6581-4213-9a3d-32a9147f0399@stanley.mountain
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Move UFS trace events to private header

UFS trace events are called exclusively from the UFS core drivers.  Make
those events private to the core driver.

The MAINTAINERS file does not need updating as the maintainership remains
the same and the relevant directory is already covered.

Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/20240821055411.3128159-1-avri.altman@wdc.com
Acked-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshcd-pltfrm: Signedness bug in ufshcd_parse_clock_info()

The "sz" variable needs to be a signed type for the error handling to work
as intended.  Fortunately, there is some sanity checking on "sz" on the
next line, so negative values would be caught and it doesn't really affect
runtime.

Fixes: eab0dce11dd9 ("scsi: ufs: ufshcd-pltfrm: Use of_property_count_u32_elems() to get property length")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Link: https://lore.kernel.org/r/404a4727-89c6-410b-9ece-301fa399d4db@stanley.mountain
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Add HCI capabilities sysfs group

The standard register map of UFSHCI is comprised of several groups.  The
first group (starting from offset 0x00), is the host capabilities group.
It contains some interesting information that otherwise is not available,
e.g. the UFS version of the platform etc.

Reviewed-by: Keoseong Park <keosung.park@samsung.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/20240811143757.2538212-3-avri.altman@wdc.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Prepare to add HCI capabilities sysfs

Prepare so we'll be able to read various other HCI registers.  While at it,
fix the HCPID & HCMID register names to stand for what they really are.
Also replace the pm_runtime_{get/put}_sync() calls in auto_hibern8_show to
ufshcd_rpm_{get/put}_sync() as any host controller register reads should.

Reviewed-by: Keoseong Park <keosung.park@samsung.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/20240811143757.2538212-2-avri.altman@wdc.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "NCR5380: Bug fixes and other improvements"

Finn Thain <fthain@linux-m68k.org> says:

This series begins with some work on the mac_scsi driver to improve
compatibility with SCSI2SD v5 devices. Better error handling is needed
there because the PDMA hardware does not tolerate the write latency
spikes which SD cards can produce.

A bug is fixed in the 5380 core driver so that scatter/gather can be
enabled in mac_scsi.

Several patches at the end of this series improve robustness and
correctness in the core driver.

This series has been tested on a variety of mac_scsi hosts. A variety
of SCSI targets was also tested, including Quantum HDD, Fujitsu HDD,
Iomega FDD, Ricoh CD-RW, Matsushita CD-ROM, SCSI2SD and BlueSCSI.

Link: https://lore.kernel.org/r/cover.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Clean up indentation

Tidy up a few indentation annoyances. No functional change.

Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/8541ea096fde9f8716b79e4f0707aed916a8c58d.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Remove obsolete comment

This comment should have been removed in commit e7734ef14ead ("scsi:
NCR5380: Remove context check") when the irqs_disabled() conditional was
removed.

Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/c54aff198b5a60be8ecfd50df0a9a77850730501.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Remove redundant result calculation from NCR5380_transfer_pio()

NCR5380_transfer_pio() returns an ambiguous value which is ignored by
callers. Make it void and remove the redundant calculation. Adopt
kernel-doc format for the updated description.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/c07a52d0d7610b4b9969d6dd4fc9a62458fe15de.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Drop redundant member from struct NCR5380_cmd

The 'message' member is stored but never loaded so just remove it.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/4dc903a95a814d0c9b09656f3651a1bd798fcbbb.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Handle BSY signal loss during information transfer phases

Improve robustness by checking for a lost BSY signal during the information
transfer loop. The status register is being polled anyway, so a BSY check
costs nothing. BSY signal loss could be caused by a target error or a
kicked plug etc. A bus reset is another possibility but that is already
handled and hostdata->connected would be NULL.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/d253dddaf4d9bc17b8ee02ea2b731d92f25b16f1.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Initialize buffer for MSG IN and STATUS transfers

Following an incomplete transfer in MSG IN phase, the driver would not
notice the problem and would make use of invalid data. Initialize 'tmp'
appropriately and bail out if no message was received. For STATUS phase,
preserve the existing status code unless a new value was transferred.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/52e02a8812ae1a2d810d7f9f7fd800c3ccc320c4.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mac_scsi: Enable scatter/gather by default

Now that FLAG_DMA_FIXUP has itself been fixed up, it can be used to enable
scatter/gather. Increase the default value for sg_tablesize to SG_ALL for
those systems which are compatible with FLAG_DMA_FIXUP.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/f155ba5ce93055cbc6ac6d4026673f40f826edb8.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Check for phase match during PDMA fixup

It's not an error for a target to change the bus phase during a transfer.
Unfortunately, the FLAG_DMA_FIXUP workaround does not allow for that -- a
phase change produces a DRQ timeout error and the device borken flag will
be set.

Check the phase match bit during FLAG_DMA_FIXUP processing. Don't forget to
decrement the command residual. While we are here, change shost_printk()
into scmd_printk() for better consistency with other DMA error messages.

Tested-by: Stan Johnson <userm57@yahoo.com>
Fixes: 55181be8ced1 ("ncr5380: Replace redundant flags with FLAG_NO_DMA_FIXUP")
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/99dc7d1f4c825621b5b120963a69f6cd3e9ca659.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mac_scsi: Disallow bus errors during PDMA send

SD cards can produce write latency spikes on the order of a hundred
milliseconds. If the target firmware does not hide that latency during DATA
IN and OUT phases it can cause the PDMA circuitry to raise a processor bus
fault which in turn leads to an unreliable byte count and a DMA overrun.

The Last Byte Sent flag is used to detect the overrun but this mechanism is
unreliable on some systems. Instead, set a DID_ERROR result whenever there
is a bus fault during a PDMA send, unless the cause was a phase mismatch.

Cc: stable@vger.kernel.org # 5.15+
Reported-and-tested-by: Stan Johnson <userm57@yahoo.com>
Fixes: 7c1f3e3447a1 ("scsi: mac_scsi: Treat Last Byte Sent time-out as failure")
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/cc38df687ace2c4ffc375a683b2502fc476b600d.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mac_scsi: Refactor polling loop

Before the error handling can be revised, some preparation is needed.
Refactor the polling loop with a new function, macscsi_wait_for_drq().
This function will gain more call sites in the next patch.

Cc: stable@vger.kernel.org # 5.15+
Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/6a5ffabb4290c0d138c6d285fda8fa3902e926f0.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mac_scsi: Revise printk(KERN_DEBUG ...) messages

After a bus fault, capture and log the chip registers immediately, if the
NDEBUG_PSEUDO_DMA macro is defined. Remove some printk(KERN_DEBUG ...)
messages that aren't needed any more.  Don't skip the debug message when
bytes == 0. Show all of the byte counters in the debug messages.

Cc: stable@vger.kernel.org # 5.15+
Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@linux-m68k.org>
Link: https://lore.kernel.org/r/7573c79f4e488fc00af2b8a191e257ca945e0409.1723001788.git.fthain@linux-m68k.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: Driver version update to 8.10.0.5.50

Update driver version to 8.10.0.5.50.

Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Link: https://lore.kernel.org/r/20240808125418.8832-4-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: Update consumer index of reply queues after every 100 replies

Instead of updating the ConsumerIndex of the Admin and Operational
ReplyQueues after processing all replies in the queue, the index will now
be periodically updated after processing every 100 replies.

Co-developed-by: Sathya Prakash <sathya.prakash@broadcom.com>
Signed-off-by: Sathya Prakash <sathya.prakash@broadcom.com>
Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Link: https://lore.kernel.org/r/20240808125418.8832-3-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpi3mr: Return complete ioc_status for ioctl commands

The driver masked the loginfo available bit in the iocstatus before passing
it to the applications, causing a mismatch in error messages between Linux
and other operating systems.

Modify driver to return unmasked (complete) iocstatus, including the
loginfo available bit, for the MPI commands sent through the ioctl
interface.

Co-developed-by: Sathya Prakash <sathya.prakash@broadcom.com>
Signed-off-by: Sathya Prakash <sathya.prakash@broadcom.com>
Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Link: https://lore.kernel.org/r/20240808125418.8832-2-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: snic: Avoid creating two slab caches with the same name

In the spirit of [1], fix the copy-paste typo and use unique names for both
caches.

[1]: https://lore.kernel.org/all/20240807090746.2146479-1-pedro.falcato@gmail.com/

Signed-off-by: Pedro Falcato <pedro.falcato@gmail.com>
Link: https://lore.kernel.org/r/20240807095709.2200728-1-pedro.falcato@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshcd-pltfrm: Use of_property_count_u32_elems() to get property length

Replace of_get_property() with the type specific
of_property_count_u32_elems() to get the property length.

This is part of a larger effort to remove callers of of_get_property() and
similar functions. of_get_property() leaks the DT property data pointer
which is a problem for dynamically allocated nodes which may be freed.

Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://lore.kernel.org/r/20240808170704.1438658-1-robh@kernel.org
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshcd-pltfrm: Use of_property_present()

Use of_property_present() to test for property presence rather than
of_find_property(). This is part of a larger effort to remove callers of
of_find_property() and similar functions. of_find_property() leaks the DT
struct property and data pointers which is a problem for dynamically
allocated nodes which may be freed.

Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://lore.kernel.org/r/20240808170644.1436991-1-robh@kernel.org
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: Don't check REQ_ATOMIC for reads

We check in submit_bio_noacct() if flag REQ_ATOMIC is set for both read and
write operations, and then validate the atomic operation if set. Flag
REQ_ATOMIC can only be set for writes, so don't bother checking for reads.

Fixes: 9da3d1e912f3 ("block: Add core atomic write support")
Signed-off-by: John Garry <john.g.garry@oracle.com>
Link: https://lore.kernel.org/r/20240805113315.1048591-3-john.g.garry@oracle.com
Reviewed-by: Kanchan Joshi <joshi.k@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Don't check if a write for REQ_ATOMIC

Flag REQ_ATOMIC can only be set for writes, so don't check if the operation
is also a write in sd_setup_read_write_cmnd().

Fixes: bf4ae8f2e640 ("scsi: sd: Atomic write support")
Signed-off-by: John Garry <john.g.garry@oracle.com>
Link: https://lore.kernel.org/r/20240805113315.1048591-2-john.g.garry@oracle.com
Reviewed-by: Kanchan Joshi <joshi.k@samsung.com>
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 6.11/scsi-queue tree

  https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git 6.11/scsi-queue

The functional changes of note to smartpqi are for: multipath failover
and improving the accuracy of our RAID bypass counter.

For multipath we are:

    Reverting commit 94a68c814328 ("scsi: smartpqi: Quickly propagate
    path failures to SCSI midlayer") because under certain rare
    conditions involving encryption-enabled devices, a false path
    failure is reported to the SML causing multipath to failover to
    the other path.

    Improving errors returned from the driver back to the SML by
    checking for error codes returned from the firmware and returning
    the correct ASC/ASCQ codes to the SML.

The other two patches add PCI-IDs for new controllers and change the
driver version.

Link: https://lore.kernel.org/r/20240711194704.982400-1-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Update driver version to 2.1.28-025

Update driver version to 2.1.28-025.

Reviewed-by: Mike Tran <mike.tran@microchip.com>
Reviewed-by: Gerry Morong <gerry.morong@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240711194704.982400-6-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Improve handling of multipath failover

Improve multipath failovers by mapping firmware errors into I/O errors.

In some rare instances, firmware does not return the proper error code for
I/O errors caused by a multipath path failure.

Map I/O errors returned by firmware into errors that help the multipath
layer to detect the failure of a path.

Reviewed-by: Gerry Morong <gerry.morong@microchip.com>
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/20240711194704.982400-5-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: revert propagate-the-multipath-failure-to-SML-quickly

Correct a rare multipath failure issue by reverting commit 94a68c814328
("scsi: smartpqi: Quickly propagate path failures to SCSI midlayer") [1].

Reason for revert: The patch propagated the path failure to SML quickly
when one of the path fails during IO and AIO path gets disabled for a
multipath device.

But it created a new issue: when creating a volume on an encryption-enabled
controller, the firmware reports the AIO path is disabled, which cause the
driver to report a path failure to SML for a multipath device.

There will be a new fix to handle "Illegal request" and "Invalid field in
parameter list" on RAID path when the AIO path is disabled on a multipath
device.

[1] https://lore.kernel.org/all/164375209313.440833.9992416628621839233.stgit@brunhilda.pdev.net/

Fixes: 94a68c814328 ("scsi: smartpqi: Quickly propagate path failures to SCSI midlayer")
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: Gilbert Wu <Gilbert.Wu@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240711194704.982400-4-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Improve accuracy/performance of raid-bypass-counter

The original implementation of this counter used an atomic variable.
However, this implementation negatively impacted performance in some
configurations.

Switch to using per_cpu variables.

Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Co-developed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Signed-off-by: Mahesh Rajashekhara <mahesh.rajashekhara@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/20240711194704.982400-3-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 new inagile PCI IDs:
                                             VID  / DID  / SVID / SDID
                                             ----   ----   ----   ----
            SMART-HBA 8242-24i               9005 / 028f / 1ff9 / 0045
            RAID 8236-16i                    9005 / 028f / 1ff9 / 0046
            RAID 8240-24i                    9005 / 028f / 1ff9 / 0047
            SMART-HBA 8238-16i               9005 / 028f / 1ff9 / 0048
            PM8222-SHBA                      9005 / 028f / 1ff9 / 004a
            RAID PM8204-2GB                  9005 / 028f / 1ff9 / 004b
            RAID PM8204-4GB                  9005 / 028f / 1ff9 / 004c
            PM8222-HBA                       9005 / 028f / 1ff9 / 004f
            MT0804M6R                        9005 / 028f / 1ff9 / 0051
            MT0801M6E                        9005 / 028f / 1ff9 / 0052
            MT0808M6R                        9005 / 028f / 1ff9 / 0053
            MT0800M6H                        9005 / 028f / 1ff9 / 0054
            RS0800M5H24i                     9005 / 028f / 1ff9 / 006b
            RS0800M5E8i                      9005 / 028f / 1ff9 / 006c
            RS0800M5H8i                      9005 / 028f / 1ff9 / 006d
            RS0804M5R16i                     9005 / 028f / 1ff9 / 006f
            RS0800M5E24i                     9005 / 028f / 1ff9 / 0070
            RS0800M5H16i                     9005 / 028f / 1ff9 / 0071
            RS0800M5E16i                     9005 / 028f / 1ff9 / 0072
            RT0800M7E                        9005 / 028f / 1ff9 / 0086
            RT0800M7H                        9005 / 028f / 1ff9 / 0087
            RT0804M7R                        9005 / 028f / 1ff9 / 0088
            RT0808M7R                        9005 / 028f / 1ff9 / 0089
            RT1608M6R16i                     9005 / 028f / 1ff9 / 00a1

Add new h3c pci_id:
                                             VID  / DID  / SVID / SDID
                                             ----   ----   ----   ----
            UN RAID P4408-Mr-2               9005 / 028f / 193d / 1110

Add new powerleader pci ids:
                                             VID  / DID  / SVID / SDID
                                             ----   ----   ----   ----
            PL SmartROC PM8204               9005 / 028f / 1f3a / 0104

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: David Strahan <David.Strahan@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Link: https://lore.kernel.org/r/20240711194704.982400-2-don.brace@microchip.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "Update lpfc to revision 14.4.0.4"

Justin Tee <justintee8345@gmail.com> says:

Update lpfc to revision 14.4.0.4

This patch set contains diagnostic logging improvements, a minor clean
up when submitting abort requests, a bug fix related to reset and
errata paths, and modifications to FLOGI and PRLO ELS command
handling.

The patches were cut against Martin's 6.11/scsi-queue tree.

Link: https://lore.kernel.org/r/20240726231512.92867-1-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Copyright updates for 14.4.0.4 patches

Update copyrights to 2024 for files modified in the 14.4.0.4 patch set.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-9-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Update lpfc version to 14.4.0.4

Update lpfc version to 14.4.0.4

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-8-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Update PRLO handling in direct attached topology

A kref imbalance occurs when handling an unsolicited PRLO in direct
attached topology.

Rework PRLO rcv handling when in MAPPED state.  Save the state that we were
handling a PRLO by setting nlp_last_elscmd to ELS_CMD_PRLO.  Then in the
lpfc_cmpl_els_logo_acc() completion routine, manually restart discovery.
By issuing the PLOGI, which nlp_gets, before nlp_put at the end of the
lpfc_cmpl_els_logo_acc() routine, we are saving us from a final nlp_put.
And, we are still allowing the unreg_rpi to happen.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-7-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Fix unsolicited FLOGI kref imbalance when in direct attached topology

In direct attached topology, certain target vendors that are quick to issue
FLOGI followed by a cable pull for more than dev_loss_tmo may result in a
kref imbalance for the remote port ndlp object.

Add an nlp_get when the defer_flogi_acc flag is set.  This is expected to
balance the nlp_put in the defer_flogi_acc clause in the
lpfc_issue_els_flogi() routine.  Because we need to retain the ndlp ptr,
reorganize all of the defer_flogi_acc information into one
lpfc_defer_flogi_acc struct.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-6-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Fix unintentional double clearing of vmid_flag

The vport->vmid_flag is unintentionally cleared twice after an issue_lip
via the lpfc_reinit_vmid routine().

The first call to lpfc_reinit_vmid() is in lpfc_cmpl_els_flogi().  Then
lpfc_cmpl_els_flogi_fabric() calls lpfc_register_new_vport(), which calls
lpfc_cmpl_reg_new_vport() when the mbox command completes and calls
lpfc_reinit_vmid() a second time.

Fix by moving the vmid_flag clear outside of the lpfc_reinit_vmid() routine
so that vmid_flag is only cleared once upon FLOGI completion.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-5-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Validate hdwq pointers before dereferencing in reset/errata paths

When the HBA is undergoing a reset or is handling an errata event, NULL ptr
dereference crashes may occur in routines such as
lpfc_sli_flush_io_rings(), lpfc_dev_loss_tmo_callbk(), or
lpfc_abort_handler().

Add NULL ptr checks before dereferencing hdwq pointers that may have been
freed due to operations colliding with a reset or errata event handler.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-4-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Remove redundant vport assignment when building an abort request

The lpfc_sli_issue_abort_iotag() routine has a redundant assignment of
abtsiocbp->vport = vport;

The duplicate lines are from a previous refactoring, and this patch removes
the accidental redundancy.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-3-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Change diagnostic log flag during receipt of unknown ELS cmds

During diagnostics, it has been determined that the 0115 log message for
receipt of unknown ELS cmds does not benefit from trace buffer dumps.  The
trace buffer dump floods the console with unnecessary information, and the
singular LOG_ELS flag has proven more beneficial in debugging efforts when
dealing with unknown ELS cmds.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20240726231512.92867-2-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: core: Support Updating UIC Command Timeout

The default UIC command timeout still remains 500ms.  Allow platform
drivers to override the UIC command timeout if desired.

In a real product, the 500ms timeout value is probably good enough.
However, during the product development where there are a lot of logging
and debug messages being printed to the UART console, interrupt starvations
happen occasionally because the UART may print long debug messages from
different modules in the system.  While printing, the UART may have
interrupts disabled for more than 500ms, causing UIC command timeout.  The
UIC command timeout would trigger more printing from the UFS driver, and
eventually a watchdog timeout may occur unnecessarily.

Add support for overriding the UIC command timeout value with the newly
created uic_cmd_timeout kernel module parameter.  Default value is
500ms. Supported values range from 500ms to 2 seconds.

Signed-off-by: Bao D. Nguyen <quic_nguyenb@quicinc.com>
Link: https://lore.kernel.org/r/e4e1c87f3f867f270a3d4b5d57a00139ff0e9741.1721792309.git.quic_nguyenb@quicinc.com
Suggested-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Peter Wang <peter.wang@mediatek.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "scsi: aacraid: struct sgmap: Replace 1-element arrays with flexible arrays"

Kees Cook <kees@kernel.org> says:

This replaces some of the last remaining uses in the kernel of
1-element "fake" flexible arrays with modern C99 flexible arrays. Some
refactoring is done to ease this, and binary differences are
identified. For the on stack size changes in patch 2, the "yes, that
is the source of the binary differences" debugging patch can be found
here[1].

[1] https://git.kernel.org/pub/scm/linux/kernel/git/kees/linux.git/commit/?h=dev/v6.10-rc2/1-element&id=45e6226bcbc5e982541754eca7ac29f403e82f5e

Link: https://lore.kernel.org/r/20240711212732.work.162-kees@kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aacraid: struct {user,}sgmap{,64,raw}: Replace 1-element arrays with flexible arrays

Replace the deprecated[1] use of 1-element arrays in struct sgmap, struct
sgmap64, struct sgmapraw, struct user_sgmap, and struct user_sgmap64 with
modern flexible arrays. Additionally remove struct user_sgmapraw as it is
unused.

The resulting binary output differences from this change are limited only
to stack space consumption of the smaller "srbu" variable in
aac_issue_safw_bmic_identify() and aac_get_safw_ciss_luns(), as well as the
smaller associated pair of memcpy()s in
aac_send_safw_bmic_cmd(). Artificially growing the size of srbu back to its
prior size removes all binary differences[2].

As an aside, after studying the aacraid driver code I wonder how
aac_send_wellness_command() ever works. It is reporting a size 4 bytes too
small for what it has constructed in memory in the DMA region: sgentry64 is
size 12, whereas sgentry is size 8. Perhaps the hardware doesn't
care. (Regardless, it is unchanged by this patch.)

Link: https://github.com/KSPP/linux/issues/79
Link: https://git.kernel.org/pub/scm/linux/kernel/git/kees/linux.git/commit/?h=dev/v6.10-rc2/1-element&id=45e6226bcbc5e982541754eca7ac29f403e82f5e
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711215739.208776-2-kees@kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aacraid: Rearrange order of struct aac_srb_unit

struct aac_srb_unit contains struct aac_srb, which contains struct sgmap,
which ends in a (currently) "fake" (1-element) flexible array.  Converting
this to a flexible array is needed so that runtime bounds checking won't
think the array is fixed size (i.e. under CONFIG_FORTIFY_SOURCE=y and/or
CONFIG_UBSAN_BOUNDS=y), as other parts of aacraid use struct sgmap as a
flexible array.

It is not legal to have a flexible array in the middle of a structure, so
it either needs to be split up or rearranged so that it is at the end of
the structure. Luckily, struct aac_srb_unit, which is exclusively
consumed/updated by aac_send_safw_bmic_cmd(), does not depend on member
ordering.

The values set in the on-stack struct aac_srb_unit instance "srbu" by the
only two callers, aac_issue_safw_bmic_identify() and
aac_get_safw_ciss_luns(), do not contain anything in srbu.srb.sgmap.sg, and
they both implicitly initialize srbu.srb.sgmap.count to 0 during
memset(). For example:

        memset(&srbu, 0, sizeof(struct aac_srb_unit));

        srbcmd = &srbu.srb;
        srbcmd->flags   = cpu_to_le32(SRB_DataIn);
        srbcmd->cdb[0]  = CISS_REPORT_PHYSICAL_LUNS;
        srbcmd->cdb[1]  = 2; /* extended reporting */
        srbcmd->cdb[8]  = (u8)(datasize >> 8);
        srbcmd->cdb[9]  = (u8)(datasize);

        rcode = aac_send_safw_bmic_cmd(dev, &srbu, phys_luns, datasize);

During aac_send_safw_bmic_cmd(), a separate srb is mapped into DMA, and has
srbu.srb copied into it:

        srb = fib_data(fibptr);
        memcpy(srb, &srbu->srb, sizeof(struct aac_srb));

Only then is srb.sgmap.count written and srb->sg populated:

        srb->count              = cpu_to_le32(xfer_len);

        sg64 = (struct sgmap64 *)&srb->sg;
        sg64->count             = cpu_to_le32(1);
        sg64->sg[0].addr[1]     = cpu_to_le32(upper_32_bits(addr));
        sg64->sg[0].addr[0]     = cpu_to_le32(lower_32_bits(addr));
        sg64->sg[0].count       = cpu_to_le32(xfer_len);

But this is happening in the DMA memory, not in srbu.srb. An attempt to
copy the changes back to srbu does happen:

        /*
         * Copy the updated data for other dumping or other usage if
         * needed
         */
        memcpy(&srbu->srb, srb, sizeof(struct aac_srb));

But this was never correct: the sg64 (3 u32s) overlap of srb.sg (2 u32s)
always meant that srbu.srb would have held truncated information and any
attempt to walk srbu.srb.sg.sg based on the value of srbu.srb.sg.count
would result in attempting to parse past the end of srbu.srb.sg.sg[0] into
srbu.srb_reply.

After getting a reply from hardware, the reply is copied into
srbu.srb_reply:

        srb_reply = (struct aac_srb_reply *)fib_data(fibptr);
        memcpy(&srbu->srb_reply, srb_reply, sizeof(struct aac_srb_reply));

This has always been fixed-size, so there's no issue here. It is worth
noting that the two callers _never check_ srbu contents -- neither
srbu.srb nor srbu.srb_reply is examined. (They depend on the mapped
xfer_buf instead.)

Therefore, the ordering of members in struct aac_srb_unit does not matter,
and the flexible array member can moved to the end.

(Additionally, the two memcpy()s that update srbu could be entirely
removed as they are never consumed, but I left that as-is.)

Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711215739.208776-1-kees@kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge patch series "scsi: message: fusion: Replace 1-element arrays with flexible arrays"

Kees Cook <kees@kernel.org> says:

Replace all remaining uses of deprecated 1-element "fake" flexible
arrays with modern C99 flexible arrays. Add __counted_by annotations
at the same time.

Link: https://lore.kernel.org/r/20240711172432.work.523-kees@kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: struct _CONFIG_PAGE_IOC_4: Replace 1-element array with flexible array

Replace the deprecated[1] use of a 1-element array in struct
_CONFIG_PAGE_IOC_4 with a modern flexible array.

Additionally add __counted_by annotation since SEP is only ever accessed
after updating ACtiveSEP:

lsi/mpi_cnfg.h:    IOC_4_SEP                   SEP[] __counted_by(ActiveSEP);  /* 08h */
mptsas.c:        ii = IOCPage4Ptr->ActiveSEP++;
mptsas.c:        IOCPage4Ptr->SEP[ii].SEPTargetID = id;
mptsas.c:        IOCPage4Ptr->SEP[ii].SEPBus = channel;

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711172821.123936-6-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: struct _CONFIG_PAGE_IOC_3: Replace 1-element array with flexible array

Replace the deprecated[1] use of a 1-element array in struct
_CONFIG_PAGE_IOC_3 with a modern flexible array.

Additionally add __counted_by annotation since PhysDisk is only ever
accessed via a loops bounded by NumPhysDisks:

lsi/mpi_cnfg.h:    IOC_3_PHYS_DISK             PhysDisk[] __counted_by(NumPhysDisks); /* 08h */
mptscsih.c: for (i = 0; i < ioc->raid_data.pIocPg3->NumPhysDisks; i++) {
mptscsih.c: if ((id == ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskID) &&
mptscsih.c:     (channel == ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskBus)) {
mptscsih.c: for (i = 0; i < ioc->raid_data.pIocPg3->NumPhysDisks; i++) {
mptscsih.c:     ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskNum);
mptscsih.c:     ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskNum,
mptscsih.c: for (i = 0; i < ioc->raid_data.pIocPg3->NumPhysDisks; i++) {
mptscsih.c: if ((id == ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskID) &&
mptscsih.c:     (channel == ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskBus)) {
mptscsih.c: rc = ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskNum;
mptscsih.c: for (i = 0; i < ioc->raid_data.pIocPg3->NumPhysDisks; i++) {
mptscsih.c:     ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskNum);
mptscsih.c:     ioc->raid_data.pIocPg3->PhysDisk[i].PhysDiskNum,

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711172821.123936-5-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: struct _CONFIG_PAGE_IOC_2: Replace 1-element array with flexible array

Replace the deprecated[1] use of a 1-element array in struct
_CONFIG_PAGE_IOC_2 with a modern flexible array.

Additionally add __counted_by annotation since RaidVolume is only ever
accessed from loops controlled by NumActiveVolumes:

lsi/mpi_cnfg.h:    CONFIG_PAGE_IOC_2_RAID_VOL  RaidVolume[] __counted_by(NumActiveVolumes); /* 0Ch */
mptbase.c:      for (i = 0; i < pIoc2->NumActiveVolumes ; i++)
mptbase.c:                  pIoc2->RaidVolume[i].VolumeBus,
mptbase.c:                  pIoc2->RaidVolume[i].VolumeID);
mptsas.c:               for (i = 0; i < ioc->raid_data.pIocPg2->NumActiveVolumes; i++) {
mptsas.c:                                       RaidVolume[i].VolumeID) {
mptsas.c:                                       RaidVolume[i].VolumeBus;
mptsas.c:       for (i = 0; i < ioc->raid_data.pIocPg2->NumActiveVolumes; i++) {
mptsas.c:                   ioc->raid_data.pIocPg2->RaidVolume[i].VolumeID, 0);
mptsas.c:                   ioc->raid_data.pIocPg2->RaidVolume[i].VolumeID);
mptsas.c:                   ioc->raid_data.pIocPg2->RaidVolume[i].VolumeID, 0);
mptsas.c:               for (i = 0; i < ioc->raid_data.pIocPg2->NumActiveVolumes; i++) {
mptsas.c:                       if (ioc->raid_data.pIocPg2->RaidVolume[i].VolumeID ==
mptsas.c:       for (i = 0; i < ioc->raid_data.pIocPg2->NumActiveVolumes; i++)
mptsas.c:               if (ioc->raid_data.pIocPg2->RaidVolume[i].VolumeID == id)
mptspi.c:       for (i=0; i < ioc->raid_data.pIocPg2->NumActiveVolumes; i++) {
mptspi.c:               if (ioc->raid_data.pIocPg2->RaidVolume[i].VolumeID == id) {

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711172821.123936-4-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: struct _CONFIG_PAGE_RAID_PHYS_DISK_1: Replace 1-element array with flexible array

Replace the deprecated[1] use of a 1-element array in struct
_CONFIG_PAGE_RAID_PHYS_DISK_1 with a modern flexible array.

Additionally add __counted_by annotation since Path is only ever accessed
via a loops bounded by NumPhysDiskPaths:

lsi/mpi_cnfg.h:    RAID_PHYS_DISK1_PATH            Path[] __counted_by(NumPhysDiskPaths);/* 0Ch */
mptbase.c:      phys_disk->NumPhysDiskPaths = buffer->NumPhysDiskPaths;
mptbase.c:      for (i = 0; i < phys_disk->NumPhysDiskPaths; i++) {
mptbase.c:              phys_disk->Path[i].PhysDiskID = buffer->Path[i].PhysDiskID;
mptbase.c:              phys_disk->Path[i].PhysDiskBus = buffer->Path[i].PhysDiskBus;

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711172821.123936-3-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: struct _CONFIG_PAGE_SAS_IO_UNIT_0: Replace 1-element array with flexible array

Replace the deprecated[1] use of a 1-element array in struct
_CONFIG_PAGE_SAS_IO_UNIT_0 with a modern flexible array.

Additionally add __counted_by annotation since PhyData is only ever
accessed via a loops bounded by NumPhys:

lsi/mpi_cnfg.h:    MPI_SAS_IO_UNIT0_PHY_DATA       PhyData[] __counted_by(NumPhys);    /* 10h */
mptsas.c:       port_info->num_phys = buffer->NumPhys;
mptsas.c:       for (i = 0; i < port_info->num_phys; i++) {
mptsas.c:               mptsas_print_phy_data(ioc, &buffer->PhyData[i]);
mptsas.c:               port_info->phy_info[i].phy_id = i;
mptsas.c:               port_info->phy_info[i].port_id =
mptsas.c:                   buffer->PhyData[i].Port;
mptsas.c:               port_info->phy_info[i].negotiated_link_rate =
mptsas.c:                   buffer->PhyData[i].NegotiatedLinkRate;

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711172821.123936-2-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: struct _RAID_VOL0_SETTINGS: Replace 1-element array with flexible array

Replace the deprecated[1] use of a 1-element array in struct
_RAID_VOL0_SETTINGS with a modern flexible array.

Additionally add __counted_by annotation since PhysDisk is only ever
accessed via a loops bounded by NumPhysDisks:

lsi/mpi_cnfg.h:    RAID_VOL0_PHYS_DISK     PhysDisk[] __counted_by(NumPhysDisks); /* 28h */
mptbase.c: for (i = 0; i < buffer->NumPhysDisks; i++) {
mptbase.c:     buffer->PhysDisk[i].PhysDiskNum, &phys_disk) != 0)
mptsas.c: for (i = 0; i < buffer->NumPhysDisks; i++) {
mptsas.c:     buffer->PhysDisk[i].PhysDiskNum, &phys_disk) != 0)
mptsas.c: for (i = 0; i < buffer->NumPhysDisks; i++) {
mptsas.c:     buffer->PhysDisk[i].PhysDiskNum, &phys_disk) != 0)

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711172821.123936-1-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ipr: Replace 1-element arrays with flexible arrays

Replace the deprecated[1] use of a 1-element arrays in struct
ipr_hostrcb_fabric_desc and struct ipr_hostrcb64_fabric_desc with modern
flexible arrays.

No binary differences are present after this conversion.

Link: https://github.com/KSPP/linux/issues/79
Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://lore.kernel.org/r/20240711180702.work.536-kees@kernel.org
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>