Firmware events are queued up using the fw_event_work's struct work, not
its delayed_work member.  The initial driver for SAS2 controllers had
handled firmware reset using the rescan barrier and was later redesigned
through "mpt2sas: [Resend] Host Reset code cleanup".  The delayed_work
variables are now unused and may provoke CONFIG_DEBUG_OBJECTS_TIMERS
"assert_init not available" false warnings in
_scsih_fw_event_cleanup_queue.

Cleanup fw_event_work's unused entries, update its kerneldoc, and
update _scsih_fw_event_cleanup_queue accordingly.

Fixes: 146b16c8 (mpt3sas: Refcount fw_events and fix unsafe list usage)
Signed-off-by: NJoe Lawrence <joe.lawrence@stratus.com>
Acked-by: NChaitra P B <chaitra.basappa@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add custom version of function to allocate device,
alloc_dev_quirk_v2_hw().  For sata devices the device id bit0 should be
0.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add v2 hw custom function slot_index_alloc_quirk_v2_hw().  SAS devices
should have IPTT bit0 equal to 1.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add methods to use HW specific versions of functions to allocate slot
and device.  HW specific methods are permitted to workaround device id
vs IPTT collision issue in v2 hw.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NSumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This patch will do code optmization for task management functions.
Below are key changes:

1. Remove reset_device hook as it was not being used and driver was
setting this to NULL.

2. Create wrapper functions for task abort and target reset and inside
   these functions adapter specific calls be made. e.g. fusion adapters
   support task abort and target reset so task abort and target reset
   should be issued to fusion adapters only and for MFI adapters, print
   a message saying feature not supported.
Signed-off-by: NSumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

In OCR path, before calling chip reset calls function
megasas_wait_for_outstanding_fusion to check reason for OCR. In case of
firmware FAULT initiated OCR and DCMD timeout initiated timeout, driver
will clear any outstanding reply (yet to be processed by driver) in
reply queues before going for chip reset. This code is added to handle a
scenario when IO timeout initiated adapter reset and management
application initiated adapter reset (by sending command to FAULT
firmware) happens simultaneously since adapter reset function is
safe-guarded by reset_mutex so only thread will be doing controller
reset. Consider IO timeout thread gets mutex and proceeds with adapter
reset process after disabling interrupts and by the time management
application has fired command to firmware to do adapter reset and the
same command is completed by firmware but since interrupts are disabled,
driver will not get completion and the same command will be in
outstanding/pending commands list of driver and refires same command
from IO timeout thread after chip reset which will again FAULT firmware
and eventually causes kill adapter.
Signed-off-by: NSumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This patch will reduce memory footprints of megaraid_sas driver when
booted in kdump mode.  Driver will not allocate memory for optional and
perfromance oriented features.  Below are key changes done in
megaraid_sas driver to do this:

1. Limit Controller's queue depth to 100 in kdump mode.

2. Do not allocate memory for system info buffer and PD info buffer.

3. Disable performance oriented features e.g. Disable RDPQ mode, disable
   dual queue depth, restrict to single MSI-x vector.
Signed-off-by: NSumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This patch avoids that building with W=1 causes gcc to report the
following type of warning:

    no previous prototype for ... [-Wmissing-prototypes]
Signed-off-by: NBart Van Assche <bart.vanassche@sandisk.com>
Reviewed-by: NHannes Reinicke <hare@suse.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ewan Milne <emilne@redhat.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

'struct timeval' will have its tv_sec value overflow on 32-bit systems
in year 2038 and beyond. This patch replaces the use of struct timeval
for computing mpi_request.TimeStamp, and instead uses ktime_t which
provides 64-bit seconds value. The timestamp computed remains
unaffected (milliseconds since Unix epoch).
Signed-off-by: NTina Ruchandani <ruchandani.tina@gmail.com>
Reviewed-by: NArnd Bergmann <arnd@arndb.de>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Acked-by: NSathya Prakash <sathya.prakash@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

It is unclear what the original intent of the masking was, but it is
clearly incorrect to truncate a physical address before calling
ioremap().  On systems where there are valid physical address bits above
bit-31 (arm64 for example) the result is an eventual OOPs when
initializing the driver.

Remove the bogus code to fix it.
Signed-off-by: NDavid Daney <david.daney@cavium.com>
Acked-by: NJack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

struct register_host_info stores a 64-bit UTC system time timestamp.
This patch removes the use of 'struct timeval' to obtain that timestamp
as its tv_sec value will overflow on 32-bit systems in year 2038 beyond.
The patch uses ktime_get_real_seconds() which returns a 64-bit seconds value.
Signed-off-by: NTina Ruchandani <ruchandani.tina@gmail.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Acked-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NZhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This patch fixes the ITCT table setup as it should be configured
differently for SAS and SATA devices.  For SATA disks there is no need
to set qw2 (already zeroed).  Also, link parameters for Bus inactive
limit, max connection time limit, and reject to open limit timers
parameters are changed to match global config register,
MAX_CON_TIME_LIMIT_TIME, as recommended by hw team.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NZhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This patch adds support for directly attaching SATA disks to phy
4-8. The problem was that only registers concerned with phy 0-3 were
being considered in sata_int_v2_hw().  The issue was not detected
previously as the development board only exposed phy 0-3; the new board
provides access to 8 phys.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NZhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Intermittently it is found that when multiple SATA disks are directly
connected to the host that some disks are not detected.  The problem is
that all set bitfields in ENT_INT_SRC1 are cleared for all phys in
sata_int_v2_hw() - it should clear the set bit for the phy being
serviced.

Also unnecessary double-write to ENT_INT_SRC1 and ENT_INT_SRC_MSK1 is
removed (remaining writes are done at end label).
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NZhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Contrary to the field name, the MCR (max connection rate) in the ITCT
should hold the device linkrate (linkrate of the connected phy), and not
the max linkrate.

This fixes an issue seen where some SATA drives connected through an
expander which would not attach.
Signed-off-by: NJohn Garry <john.garry@huawei.com>
Reviewed-by: NZhangfei Gao <zhangfei.gao@linaro.org>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Now it's ready to move the mempool based SG chained allocator code from
SCSI driver to lib/sg_pool.c, which will be compiled only based on a Kconfig
symbol CONFIG_SG_POOL.

SCSI selects CONFIG_SG_POOL.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NMing Lin <ming.l@ssi.samsung.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Rename SCSI_MAX_SG_SEGMENTS to SG_CHUNK_SIZE, which means the amount
we fit into a single scatterlist chunk.

Rename SCSI_MAX_SG_CHAIN_SEGMENTS to SG_MAX_SEGMENTS.

Will move these 2 generic definitions to scatterlist.h later.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Acked-by: Bart Van Assche <bart.vanassche@sandisk.com> (for ib_srp changes)
Signed-off-by: NMing Lin <ming.l@ssi.samsung.com>
Acked-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Rename SCSI specific struct and functions to more genenic names.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NMing Lin <ming.l@ssi.samsung.com>
Reviewed-by: NSagi Grimberg <sgi@grimberg.me>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Parameter "bool mq" is block driver specific.
Change it to "first_chunk" to make it more generic.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NMing Lin <ming.l@ssi.samsung.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Replace parameter "struct scsi_data_buffer" with "struct sg_table" in
SG alloc/free functions to make them generic.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NMing Lin <ming.l@ssi.samsung.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Claim Marvell 9485 controllers regardless of subdevice ID.

Tested on ASUS P9A-I/C2550/SAS/4L which uses vendor-specific 1043:8635.

[mkp: Tweaked commit message]
Signed-off-by: NLeonid Moiseichuk <leonid.moiseichuk@gmail.com>
Reviewed-by: NAndy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NChad Dupuis <chad.dupuis@qlogic.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

When we are in connection recovery and the internal command timer on a
request pops, either the scsi_cmnd->device or scsi_cmnd->device->host
back pointers may be NULL as the device that the command that the
request was submitted on may have been subsequently reaped due to the
connection recovery. This can cause one or both of the pointers above to
be NULL and cause a system crash if we try to return the command to the
midlayer.

Instead, double check the pointers before the return to the midlayer so
as to prevent the crash and let the upper layers finish the session
recovery and rediscover the device.
Signed-off-by: NChad Dupuis <chad.dupuis@qlogic.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NChad Dupuis <chad.dupuis@qlogic.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NChad Dupuis <chad.dupuis@qlogic.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Per customer request, add the following driver tunables:

o devloss_tmo
o max_luns
o queue_depth
o tm_timeout

tm_timeout is set per scsi_host in /sys/class/scsi_host/hostX/tm_timeout.
Signed-off-by: NJoe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: NChad Dupuis <chad.dupuis@qlogic.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Now that we've done a more comprehensive fix with the intermediate
target state we can remove the previous hack introduced with commit
90a88d6e ("scsi: fix soft lockup in scsi_remove_target() on module
removal").
Signed-off-by: NJohannes Thumshirn <jthumshirn@suse.de>
Cc: stable@vger.kernel.org
Reviewed-by: NEwan D. Milne <emilne@redhat.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add intermediate STARGET_REMOVE state to scsi_target_state to avoid
running into the BUG_ON() in scsi_target_reap(). The STARGET_REMOVE
state is only valid in the path from scsi_remove_target() to
scsi_target_destroy() indicating this target is going to be removed.

This re-fixes the problem introduced in commits bc3f02a7 ("[SCSI]
scsi_remove_target: fix softlockup regression on hot remove") and
40998193 ("scsi: restart list search after unlock in
scsi_remove_target") in a more comprehensive way.

[mkp: Included James' fix for scsi_target_destroy()]
Signed-off-by: NJohannes Thumshirn <jthumshirn@suse.de>
Fixes: 40998193
Cc: stable@vger.kernel.org
Reported-by: NSergey Senozhatsky <sergey.senozhatsky@gmail.com>
Tested-by: NSergey Senozhatsky <sergey.senozhatsky@gmail.com>
Reviewed-by: NEwan D. Milne <emilne@redhat.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Reviewed-by: NJames Bottomley <jejb@linux.vnet.ibm.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

scsi host is added after negotiating the max number of IOs with
Firmware.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

While HBA reset is in progress, if IO/TM completion is received for the
same IO then IO/TM completion path releases the driver private resources
associated with IO. This fix prevents releasing the resources in
IO and TM completion path if HBA reset is in progress.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

- On posting an IO to the firmware, adapter generates an interrupt.
  Due to hardware issues, sometimes the adapter fails to generate
  the interrupt. This behavior skips updating transmit queue-
  counters, which in turn causes the queue full condition. The fix
  addresses the queue full condition.

- The fix also reserves a slot in transmit queue for hba reset.
  when queue full is observed during IO, there will always be room
  to post hba reset command.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Fix deletes the snic targets synchronously prior to deletion of host.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Fix handles control path queue issues such as queue full and sudden
removal of hardware.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

- LUN goes offline if there are at least two scsi command timeouts
  Completing the IO with scsi_done() fixes the issue.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Adding additional stats, and fixed logging messages.
- Added qdepth change stats
- Added separate isr stats for each type of interrupt
- Fixed race in updating active IOs
- Suppressed Link event message for DAS backend.
Signed-off-by: NNarsimhulu Musini <nmusini@cisco.com>
Signed-off-by: NSesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NDick Kennedy <dick.kennedy@avagotech.com>
Signed-off-by: NJames Smart <james.smart@avagotech.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NDick Kennedy <dick.kennedy@avagotech.com>
Signed-off-by: NJames Smart <james.smart@avagotech.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NDick Kennedy <dick.kennedy@avagotech.com>
Signed-off-by: NJames Smart <james.smart@avagotech.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>