scsi: mpt3sas: fix oops in error handlers after shutdown/unload

This patch adds checks for 'ioc->remove_host' in the SCSI error handlers, so
not to access pointers/resources potentially freed in the PCI shutdown/module
unload path.  The error handlers may be invoked after shutdown/unload,
depending on other components.

This problem was observed with kexec on a system with a mpt3sas based adapter
and an infiniband adapter which takes long enough to shutdown:

The mpt3sas driver finished shutting down / disabled interrupt handling, thus
some commands have not finished and timed out.

Since the system was still running (waiting for the infiniband adapter to
shutdown), the scsi error handler for task abort of mpt3sas was invoked, and
hit an oops -- either in scsih_abort() because 'ioc->scsi_lookup' was NULL
without commit dbec4c9040ed ("scsi: mpt3sas: lockless command submission"), or
later up in scsih_host_reset() (with or without that commit), because it
eventually called mpt3sas_base_get_iocstate().

After the above commit, the oops in scsih_abort() does not occur anymore
(_scsih_scsi_lookup_find_by_scmd() is no longer called), but that commit is
too big and out of the scope of linux-stable, where this patch might help, so
still go for the changes.

Also, this might help to prevent similar errors in the future, in case code
changes and possibly tries to access freed stuff.

Note the fix in scsih_host_reset() is still important anyway.

Signed-off-by: Mauricio Faria de Oliveira <mauricfo@linux.vnet.ibm.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: storvsc: Spread interrupts when picking a channel for I/O requests

Update the algorithm in storvsc_do_io to look for a channel
starting with the current CPU + 1 and wrap around (within the
current NUMA node). This spreads VMbus interrupts more evenly
across CPUs. Previous code always started with first CPU in
the current NUMA node, skewing the interrupt load to that CPU.

Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Do not use 32-bit atomic request descriptor for Ventura controllers

Problem Statement: Sending I/O through 32 bit descriptors to Ventura series of
controller results in IO timeout on certain conditions.

This error only occurs on systems with high I/O activity on Ventura series
controllers.

Changes in this patch will prevent driver from using 32 bit descriptor and use
64 bit Descriptors.

Cc: <stable@vger.kernel.org>
Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla4xxx: skip error recovery in case of register disconnect.

A system crashes when continuously removing/re-adding the storage
controller.

Signed-off-by: Manish Rangankar <manish.rangankar@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aacraid: fix shutdown crash when init fails

When aacraid init fails with "AAC0: adapter self-test failed.", shutdown
leads to UBSAN warning and then oops:

[154316.118423] ================================================================================
[154316.118508] UBSAN: Undefined behaviour in drivers/scsi/scsi_lib.c:2328:27
[154316.118566] member access within null pointer of type 'struct Scsi_Host'
[154316.118631] CPU: 2 PID: 14530 Comm: reboot Tainted: G        W        4.15.0-dirty #89
[154316.118701] Hardware name: Hewlett Packard HP NetServer/HP System Board, BIOS 4.06.46 PW 06/25/2003
[154316.118774] Call Trace:
[154316.118848]  dump_stack+0x48/0x65
[154316.118916]  ubsan_epilogue+0xe/0x40
[154316.118976]  __ubsan_handle_type_mismatch+0xfb/0x180
[154316.119043]  scsi_block_requests+0x20/0x30
[154316.119135]  aac_shutdown+0x18/0x40 [aacraid]
[154316.119196]  pci_device_shutdown+0x33/0x50
[154316.119269]  device_shutdown+0x18a/0x390
[...]
[154316.123435] BUG: unable to handle kernel NULL pointer dereference at 000000f4
[154316.123515] IP: scsi_block_requests+0xa/0x30

This is because aac_shutdown() does

        struct Scsi_Host *shost = pci_get_drvdata(dev);
        scsi_block_requests(shost);

and that assumes shost has been assigned with pci_set_drvdata().

However, pci_set_drvdata(pdev, shost) is done in aac_probe_one() far
after bailing out with error from calling the init function
((*aac_drivers[index].init)(aac)), and when the init function fails, no
error is returned from aac_probe_one() so PCI layer assumes there is
driver attached, and tries to shut it down later.

Fix it by returning error from aac_probe_one() when card-specific init
function fails.

This fixes reboot on my HP NetRAID-4M with dead battery.

Signed-off-by: Meelis Roos <mroos@linux.ee>
Reviewed-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Cleanup local str variable

Signed-off-by: Nilesh Javali <nilesh.javali@cavium.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Acked-by: Chris Leech <cleech@redhat.com>
Acked-by: Lee Duncan <lduncan@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Fix truncation of CHAP name and secret

The data in NVRAM is not guaranteed to be NUL terminated.  Since
snprintf expects byte-stream to accommodate null byte, the CHAP secret
is truncated.  Use sprintf instead of snprintf to fix the truncation of
CHAP name and secret.

Signed-off-by: Andrew Vasquez <andrew.vasquez@cavium.com>
Signed-off-by: Nilesh Javali <nilesh.javali@cavium.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Acked-by: Chris Leech <cleech@redhat.com>
Acked-by: Lee Duncan <lduncan@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix incorrect handle for abort IOCB

This patch fixes incorrect handle used for abort IOCB.

Fixes: b027a5ace443 ("scsi: qla2xxx: Fix queue ID for async abort with Multiqueue")
Signed-off-by: Darren Trapp <darren.trapp@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix double free bug after firmware timeout

This patch is based on Max's original patch.

When the qla2xxx firmware is unavailable, eventually
qla2x00_sp_timeout() is reached, which calls the timeout function and
frees the srb_t instance.

The timeout function always resolves to qla2x00_async_iocb_timeout(),
which invokes another callback function called "done".  All of these
qla2x00_*_sp_done() callbacks also free the srb_t instance; after
returning to qla2x00_sp_timeout(), it is freed again.

The fix is to remove the "sp->free(sp)" call from qla2x00_sp_timeout()
and add it to those code paths in qla2x00_async_iocb_timeout() which
do not already free the object.

This is how it looks like with KASAN:

BUG: KASAN: use-after-free in qla2x00_sp_timeout+0x228/0x250
Read of size 8 at addr ffff88278147a590 by task swapper/2/0

Allocated by task 1502:
save_stack+0x33/0xa0
kasan_kmalloc+0xa0/0xd0
kmem_cache_alloc+0xb8/0x1c0
mempool_alloc+0xd6/0x260
qla24xx_async_gnl+0x3c5/0x1100

Freed by task 0:
save_stack+0x33/0xa0
kasan_slab_free+0x72/0xc0
kmem_cache_free+0x75/0x200
qla24xx_async_gnl_sp_done+0x556/0x9e0
qla2x00_async_iocb_timeout+0x1c7/0x420
qla2x00_sp_timeout+0x16d/0x250
call_timer_fn+0x36/0x200

The buggy address belongs to the object at ffff88278147a440
which belongs to the cache qla2xxx_srbs of size 344
The buggy address is located 336 bytes inside of
344-byte region [ffff88278147a440, ffff88278147a598)

Reported-by: Max Kellermann <mk@cm4all.com>
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Cc: Max Kellermann <mk@cm4all.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: storvsc: Increase cmd_per_lun for higher speed devices

Increase cmd_per_lun to allow more I/Os in progress per device,
particularly for NVMe's.  The Hyper-V host side can handle the higher
count with no issues.

Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix a locking imbalance in qlt_24xx_handle_els()

Ensure that upon return the tgt->ha->tgt.sess_lock spin lock is unlocked
no matter which code path is taken through this function.  This was
detected by sparse.

Fixes: 82abdcaf3ede ("scsi: qla2xxx: Allow target mode to accept PRLI in dual mode")
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Cc: Quinn Tran <quinn.tran@cavium.com>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_dh: Document alua_rtpg_queue() arguments

Since commit 3a025e1d1c2e ("Add optional check for bad kernel-doc
comments") building with W=1 causes warnings to appear for issues in
kernel-doc headers. This patch avoids that the following warnings are
reported when building with W=1:

drivers/scsi/device_handler/scsi_dh_alua.c:867: warning: No description found for parameter 'pg'
drivers/scsi/device_handler/scsi_dh_alua.c:867: warning: No description found for parameter 'sdev'
drivers/scsi/device_handler/scsi_dh_alua.c:867: warning: No description found for parameter 'qdata'
drivers/scsi/device_handler/scsi_dh_alua.c:867: warning: No description found for parameter 'force'

Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: Remove Makefile entry for oktagon files

Remove line using non-existent files which were removed in
commit 642978beb483 ("[SCSI] remove m68k NCR53C9x based drivers")

[mkp: tweaked patch description]

Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aic7xxx: remove aiclib.c

aiclib.c is unused (and contains no code) since commit 1ff927306e08
("[SCSI] aic7xxx: remove aiclib.c")

13 years later, finish the cleaning by removing it from tree.

[mkp: tweaked patch description]

Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Avoid triggering undefined behavior in qla2x00_mbx_completion()

A left shift must shift less than the bit width of the left argument.
Avoid triggering undefined behavior if ha->mbx_count == 32.

This patch avoids that UBSAN reports the following complaint:

UBSAN: Undefined behaviour in drivers/scsi/qla2xxx/qla_isr.c:275:14
shift exponent 32 is too large for 32-bit type 'int'
Call Trace:
 dump_stack+0x4e/0x6c
 ubsan_epilogue+0xd/0x3b
 __ubsan_handle_shift_out_of_bounds+0x112/0x14c
 qla2x00_mbx_completion+0x1c5/0x25d [qla2xxx]
 qla2300_intr_handler+0x1ea/0x3bb [qla2xxx]
 qla2x00_mailbox_command+0x77b/0x139a [qla2xxx]
 qla2x00_mbx_reg_test+0x83/0x114 [qla2xxx]
 qla2x00_chip_diag+0x354/0x45f [qla2xxx]
 qla2x00_initialize_adapter+0x2c2/0xa4e [qla2xxx]
 qla2x00_probe_one+0x1681/0x392e [qla2xxx]
 pci_device_probe+0x10b/0x1f1
 driver_probe_device+0x21f/0x3a4
 __driver_attach+0xa9/0xe1
 bus_for_each_dev+0x6e/0xb5
 driver_attach+0x22/0x3c
 bus_add_driver+0x1d1/0x2ae
 driver_register+0x78/0x130
 __pci_register_driver+0x75/0xa8
 qla2x00_module_init+0x21b/0x267 [qla2xxx]
 do_one_initcall+0x5a/0x1e2
 do_init_module+0x9d/0x285
 load_module+0x20db/0x38e3
 SYSC_finit_module+0xa8/0xbc
 SyS_finit_module+0x9/0xb
 do_syscall_64+0x77/0x271
 entry_SYSCALL64_slow_path+0x25/0x25

Reported-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mptfusion: Add bounds check in mptctl_hp_targetinfo()

My static checker complains about an out of bounds read:

    drivers/message/fusion/mptctl.c:2786 mptctl_hp_targetinfo()
    error: buffer overflow 'hd->sel_timeout' 255 <= u32max.

It's true that we probably should have a bounds check here.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sym53c8xx_2: iterator underflow in sym_getsync()

We wanted to exit the loop with "div" set to zero, but instead, if we
don't hit the break then "div" is -1 when we finish the loop.  It leads
to an array underflow a few lines later.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bnx2fc: Fix check in SCSI completion handler for timed out request

When a request times out we set the io_req flag BNX2FC_FLAG_IO_COMPL so
that if a subsequent completion comes in on that task ID we will ignore
it.  The issue is that in the check for this flag there is a missing
return so we will continue to process a request which may have already
been returned to the ownership of the SCSI layer.  This can cause
unpredictable results.

Solution is to add in the missing return.

[mkp: typo plus title shortening]

Signed-off-by: Chad Dupuis <chad.dupuis@cavium.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Tested-by: Laurence Oberman <loberman@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: csiostor: remove redundant assignment to pointer 'ln'

The pointer ln is assigned a value that is never read, it is re-assigned
a new value in the list_for_each loop hence the initialization is
redundant and can be removed.

Cleans up clang warning:
drivers/scsi/csiostor/csio_lnode.c:117:21: warning: Value stored to 'ln'
during its initialization is never read

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Enable quirk to ignore sending WRITE_SAME command

WRITE_SAME command is not supported by UFS. Enable a quirk for the upper
level drivers to not send WRITE SAME command.

[mkp: botched patch, applied by hand]

Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org>
Signed-off-by: Subhash Jadavani <subhashj@codeaurora.org>
Signed-off-by: Asutosh Das <asutoshd@codeaurora.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ibmvfc: fix misdefined reserved field in ibmvfc_fcp_rsp_info

The fcp_rsp_info structure as defined in the FC spec has an initial 3
bytes reserved field. The ibmvfc driver mistakenly defined this field as
4 bytes resulting in the rsp_code field being defined in what should be
the start of the second reserved field and thus always being reported as
zero by the driver.

Ideally, we should wire ibmvfc up with libfc for the sake of code
deduplication, and ease of maintaining standardized structures in a
single place. However, for now simply fixup the definition in ibmvfc for
backporting to distros on older kernels. Wiring up with libfc will be
done in a followup patch.

Cc: <stable@vger.kernel.org>
Reported-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix memory corruption during hba reset test

This patch fixes memory corrpution while performing HBA Reset test.

Following stack trace is seen:

[  466.397219] BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
[  466.433669] IP: [<ffffffffc06f5dd0>] qlt_free_session_done+0x260/0x5f0 [qla2xxx]
[  466.467731] PGD 0
[  466.476718] Oops: 0000 [#1] SMP

Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: fix an out of bound write

cpu_msix_table is allocated to store online cpus, but pci_irq_get_affinity
may return cpu_possible_mask which is then used to access cpu_msix_table.
That causes bad user experience.  Fix limits access to only online cpus,
I've also added an additional test to protect from an unlikely change in
cpu_online_mask.

[mkp: checkpatch]

Fixes: 1d55abc0e98a ("scsi: mpt3sas: switch to pci_alloc_irq_vectors")
Signed-off-by: Tomas Henzl <thenzl@redhat.com>
Acked-by: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix logo flag for qlt_free_session_done()

Commit 3515832cc614 ("scsi: qla2xxx: Reset the logo flag, after target
re-login.")fixed the target re-login after session relogin is complete,
but missed out the qlt_free_session_done() path.

This patch clears send_els_logo flag in qlt_free_session_done()
callback.

[mkp: checkpatch]

Fixes: 3515832cc614 ("scsi: qla2xxx: Reset the logo flag, after target re-login.")
Signed-off-by: Himanshu Madhani <hmadhani@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: arcmsr: avoid do_gettimeofday

The arcmsr uses its own implementation of time_to_tm(), along with
do_gettimeofday() to read the current time. While the algorithm used
here is fine in principle, it suffers from two problems:

- it assigns the seconds portion of the timeval to a 32-bit unsigned
  integer that overflows in 2106 even on 64-bit architectures.

- do_gettimeofday() returns a time_t that overflows in 2038 on all
  32-bit systems.

This changes the time retrieval function to ktime_get_real_seconds(),
which returns a proper 64-bit value, and replaces the open-coded
time_to_tm() algorithm with a call to the safe time64_to_tm().

I checked way all numbers are indexed and found that months are given in
range 0..11 while the days are in range 1..31, same as 'struct tm', but
the year value that the firmware expects starts in 2000 while 'struct
tm' is based on year 1900, so it needs a small adjustment.

[mkp: checkpatch tweaks]

Fixes: b416c099472a ("scsi: arcmsr: Add a function to set date and time to firmware")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Ching Huang <ching2048@areca.com.tw>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Add VENDOR_SPECIFIC sense code definitions

Some older devices will return vendor specific sense codes, so we should
be adding a definition for it.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Drop cqe response during connection recovery

We get stuck in the loop when firmware sends a cqe response during
connection recovery.

Signed-off-by: Manish Rangankar <manish.rangankar@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: fas216: fix sense buffer initialization

While testing with the ARM specific memset() macro removed, I ran into a
compiler warning that shows an old bug:

drivers/scsi/arm/fas216.c: In function 'fas216_rq_sns_done':
drivers/scsi/arm/fas216.c:2014:40: error: argument to 'sizeof' in 'memset' call is the same expression as the destination; did you mean to provide an explicit length? [-Werror=sizeof-pointer-memaccess]

It turns out that the definition of the scsi_cmd structure changed back
in linux-2.6.25, so now we clear only four bytes (sizeof(pointer))
instead of 96 (SCSI_SENSE_BUFFERSIZE). I did not check whether we
actually need to initialize the buffer here, but it's clear that if we
do it, we should use the correct size.

Fixes: de25deb18016 ("[SCSI] use dynamically allocated sense buffer")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ibmvfc: Remove unneeded semicolons

Trivial fix removes unneeded semicolons after switch blocks.

This issue was detected by using the Coccinelle software.

Signed-off-by: Christopher Díaz Riveros <chrisadr@gentoo.org>
Acked-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: fix a bug in hisi_sas_dev_gone()

When device gone, NULL pointer can be accessed in free_device callback
if during SAS controller reset as we clear structure sas_dev prior.

Actually we can only set dev_type as SAS_PHY_UNUSED and not clear
structure sas_dev as all the members of structure sas_dev will be
re-initialized after device found.

Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: directly attached disk LED feature for v2 hw

This patch implements LED feature of directly attached disk for v2 hw.
As libsas has provided an interface lldd_write_gpio() for this feature,
we just need realise the interface following SPGIO API.

We use an CPLD to finish the hardware part of this feature, and the base
address of CPLD should be configured through ACPI or DT tables.

Signed-off-by: Xiaofei Tan <tanxiaofei@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: devicetree: bindings: add LED feature for v2 hw

Add directly attached disk LED feature for v2 hw.

Signed-off-by: Xiaofei Tan <tanxiaofei@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: NVMe passthrough command support

NVMe passthrough via MFI interface. Current MegaRAID product supports
different types of encapsulation via the MFI framework.

NVMe native command should be framed by application and it should be
embedded in MFI as payload. The driver will provide interface to send
the MFI frame along with the payload (in this case, payload is NVMe
native command) to the firmware. Driver already has an existing, similar
interface for SATA and SMP passthrough.

1. Driver will pass MFI command to the firmware if the latter supports
   NVMe encapsulated processing (not all SAS3.5 firmware supports this
   feature).

2. Driver exposes sysfs entry support_nvme_encapsulation. This is
   required for backward compatibility for applications using earlier
   driver versions that did not process IOCTL frames and could result in
   host hang.

   This is already fixed as part of commit 82add4e1b354 ("scsi:
   megaraid_sas: Incorrect processing of IOCTL frames for SMP/STP
   commands")

[mkp: clarified commit message]

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid: use ktime_get_real for firmware time

do_gettimeofday() overflows in 2038 on 32-bit architectures and is
deprecated, so convert this driver to call ktime_get_real()
directly. This also simplifies the calculation.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: fnic: use 64-bit timestamps

struct timespec is deprecated since it overflows in 2038 on 32-bit
architectures, so we should use timespec64 consistently.

I'm slightly adapting the format strings here, to make sure we print the
nanoseconds with the correct number of leading zeroes.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Satish Kharat <satishkh@cisco.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Fix error return code in __qedf_probe()

Fix to return error code -ENOMEM from the error handling case instead of
0, as done elsewhere in this function.

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Acked-by: Chad Dupuis <chad.dupuis@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: devinfo: fix format of the device list

Replace "" with NULL for product revision level, and merge TEXEL
duplicate entries.

Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: James E.J. Bottomley <jejb@linux.vnet.ibm.com>
Cc: SCSI ML <linux-scsi@vger.kernel.org>
Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Update driver version to 10.00.00.05-k

Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Add XCB counters to debugfs

Signed-off-by: Anil Gurumurthy <anil.gurumurthy@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix queue ID for async abort with Multiqueue

[mkp: sparse warning]

Signed-off-by: Darren Trapp <darren.trapp@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix warning for code intentation in __qla24xx_handle_gpdb_event()

This patch fixes following smatch warning:

drivers/scsi/qla2xxx/qla_init.c:1054 __qla24xx_handle_gpdb_event() warn: inconsistent indenting

Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix warning during port_name debug print

This patch fixes following smatch warning:

drivers/scsi/qla2xxx/qla_iocb.c:2622 qla2x00_els_dcmd2_sp_done() error: '%pC' expects argument of type 'struct clk*', argument 8 has type 'uchar[]'

Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix warning in qla2x00_async_iocb_timeout()

This patch fixes following Smatch warning:

drivers/scsi/qla2xxx/qla_init.c:130 qla2x00_async_iocb_timeout() error: we previously assumed 'fcport' could be null (see line 107)

Fixes: 5c25d451163c ("scsi: qla2xxx: Fix NULL pointer access for fcport structure")
Reported by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix recursion while sending terminate exchange

During error test case where switch port status is toggled from enable to
disable, following stack trace is seen which indicates recursion trying to
send terminate exchange.  This regression was introduced by commit
82de802ad46e ("scsi: qla2xxx: Preparation for Target MQ.")

BUG: stack guard page was hit at ffffb96488383ff8 (stack is ffffb96488384000..ffffb96488387fff)
BUG: stack guard page was hit at ffffb964886c3ff8 (stack is ffffb964886c4000..ffffb964886c7fff)
kernel stack overflow (double-fault): 0000 [#1] SMP
qlt_term_ctio_exchange+0x9c/0xb0 [qla2xxx]
qlt_term_ctio_exchange+0x9c/0xb0 [qla2xxx]
qlt_term_ctio_exchange+0x9c/0xb0 [qla2xxx]
qlt_term_ctio_exchange+0x9c/0xb0 [qla2xxx]
qlt_term_ctio_exchange+0x9c/0xb0 [qla2xxx]

Fixes: 82de802ad46e ("scsi: qla2xxx: Preparation for Target MQ.")
Cc: <stable@vger.kernel.org> #4.10
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix NULL pointer crash due to probe failure

This patch fixes regression added by commit d74595278f4ab
("scsi: qla2xxx: Add multiple queue pair functionality.").

When driver is not able to get reqeusted IRQs from the system, driver will
attempt tp clean up memory before failing hardware probe. During this cleanup,
driver assigns NULL value to the pointer which has not been allocated by
driver yet. This results in a NULL pointer access.

Log file will show following message and stack trace

qla2xxx [0000:a3:00.1]-00c7:21: MSI-X: Failed to enable support, giving up -- 32/-1.
qla2xxx [0000:a3:00.1]-0037:21: Falling back-to MSI mode --1.
qla2xxx [0000:a3:00.1]-003a:21: Failed to reserve interrupt 821 already in use.
BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<ffffffffc010c4b6>] qla2x00_probe_one+0x18b6/0x2730 [qla2xxx]
PGD 0
Oops: 0002 [#1] SMP

Fixes: d74595278f4ab ("scsi: qla2xxx: Add multiple queue pair functionality.").
Cc: <stable@vger.kernel.org> # 4.10
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mptsas: remove duplicated assignment to pointer head

The pointer head is re-assigned the same value twice, so remove the
second redundant assignment.

Cleans up clang warning:
drivers/message/fusion/mptsas.c:1161:20: warning: Value stored to 'head'
during its initialization is never read

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: storvsc: missing error code in storvsc_probe()

We should set the error code if fc_remote_port_add() fails.

Cc: <stable@vger.kernel.org> #v4.12+
Fixes: daf0cd445a21 ("scsi: storvsc: Add support for FC rport.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Cathy Avery <cavery@redhat.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mptfusion: Use snprintf() instead of open coded divisions

Numbers up to 100 snprintf() prints without using a division.  Besides
that the code looks more readable.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: make function _get_st_from_smid static

The function _get_st_from_smid is local to the source and does not need
to be in global scope, so make it static.

Cleans up sparse warning:
symbol '_get_st_from_smid' was not declared. Should it be static?

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: make local symbol host_attrs static

Fixes the following sparse warning:

drivers/scsi/hisi_sas/hisi_sas_main.c:1691:25: warning:
 symbol 'host_attrs' was not declared. Should it be static?

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Acked-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_debug: delay stress fix

Introduce a state enum into sdebug_defer objects to indicate which, if
any, defer method has been used with the associated command.  Also add 2
bools to indicate which of the defer methods has been initialized. Those
objects are re-used but the initialization only needs to be done
once. This simplifies command cancellation handling.

Now the delay associated with a deferred response of a command cannot be
changed (once started) by changing the delay (and ndelay) parameters in
sysfs. Command aborts and driver shutdown are still honoured immediately
when received.

[mkp: applied by hand]

Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: fnic: use kzalloc in fnic_fcoe_process_vlan_resp

This saves a little .text and gets rid of the unmotivated line break and
the sizeof(...) style inconsistency.

Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: driver version upgrade

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: re-work DCMD refire code

No functional changes.

This patch is a re-work of DCMD refire code to better manage all the
different cases to decide whether to REFIRE or SKIP or COMPLETE certain
DCMD.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Expose fw_cmds_outstanding through sysfs

Expose FW outstanding commands (fw_outstanding) through sysfs interface.
This helps in debugging certain performance issues in the field.

Signed-off-by: Sasikumar Chandrasekaran <sasikumar.pc@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Selectively apply stream detection based on IO type

Performance improvement:

Current driver calls stream detection unconditionally for all IOs.
Stream Detection logic is not required for most of the fast path IO. To
improve performance, avoid stream detection logic and do it only if
required.

Below are the cases where stream detection is required in driver:

1. All non-FastPath IOs (IOs going to FW)
2. Fast Path reads sent to ReadAhead capable VDs.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Update LD map after populating drv_map driver map copy

Issue – There may be some IO accessing incorrect raid map, but driver
has checks in IO path to handle those cases. It is always better to move
to new raid map only once raid map is populated and validated.  No
functional defect. Fix is provided as part of review.  Fix – Update
instance->map_id after driver has populated new driver raid map from
firmware raid map.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Use megasas_wait_for_adapter_operational to detect controller state in IOCTL path

In IOCTL path, re-use megasas_wait_for_adapter_operational API to detect
controller state. This will make driver to use this API uniformly in all
cases where we need to wait for adapter to become operational.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Avoid firing DCMDs while OCR is in progress

Driver needs to avoid PCI writes while OCR is in progress.  Use
reset_mutex to synchronize between firing DCMDs MR_DCMD_PD_GET_INFO and
MR_DCMD_DRV_GET_TARGET_PROP while OCR is triggered.  Without this fix,
if Device/VD add/creation is in progress and at the same time MR
Firmware is going through OCR, user may see OCR never completed and it
may need system reboot.  This scenario is rare to occur. Fix is provided
as part of review.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: unload flag should be set after scsi_remove_host is called

Issue - Driver returns DID_NO_CONNECT when unload is in progress,
indicated using instance->unload flag. In case of dynamic unload of
driver, this flag is set before calling scsi_remove_host(). While doing
manual driver unload, user will see lots of prints for Sync Cache
command with DID_NO_CONNECT status.

Fix - Set the instance->unload flag after scsi_remove_host(). Allow
device removal process to be completed and do not block any command
before that.  SCSI commands (like SYNC_CACHE) are received (as part of
scsi_remove_host) by driver during unload will be submitted further down
to the drives.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Error handling for invalid ldcount provided by firmware in RAID map

Currently driver does not validate ldcount provided by firmware.  If the
value is invalid, fail RAID map validation accordingly.  This issue is
rare to hit in field and is fixed as part of code review.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Reset ldio_outstanding in megasas_resume

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Return the DCMD status from megasas_get_seq_num

In megasas_get_seq_num, the status of the DCMD fired to FW is not
returned, it always returns success.  We could end up registering AEN
request with incorrect sequence number if the DCMD failed.  Return the
DCMD status back to caller.  This was discovered during code review and
very rare to see issue in field to see AEN request failed bt FW.

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: memset IOC INIT frame using correct size

Commit b9637d14dc00 ("scsi: megaraid_sas: Resize MFA frame used for IOC
INIT to 4k") increased the size of IOC INIT frame to 4k.  Need to use
updated size when memsetting init_frame.

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: zero out IOC INIT and stream detection memory

Memory allocated for IOC_INIT command and stream detection array are not
zero'd before using. Use kzalloc instead of kmalloc to zero out the
memory allocated.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Change third __scsi_queue_insert() argument from int to bool

This patch does not change any functionality but makes the SCSI core
source code slightly easier to read.

Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aacraid: Delay for rescan worker needs to be 10 seconds

The delay for the rescan worker needs to 10 seconds, missed the HZ in
there.

Fixes: a1367e4adee207fe (scsi: aacraid: Reschedule host scan in case of failure)
Signed-off-by: Raghava Aditya Renukunta <RaghavaAditya.Renukunta@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aacraid: Get correct lun count

The correct lun count needs to be divided by 24, missed it in the
previous patch set.

Fixes: 4b00022753550055 (scsi: aacraid: Create helper functions to get lun info)
Signed-off-by: Raghava Aditya Renukunta <RaghavaAditya.Renukunta@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_dh_alua: skip RTPG for devices only supporting active/optimized

For hardware only supporting active/optimized there's no point in ever
re-issuing RTPG as the only new state we can possibly read is
active/optimized.  This avoid spurious errors during path failover on
such arrays.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bfa: use ARRAY_SIZE for array sizing calculation on array __pciids

Use the ARRAY_SIZE macro on array __pciids to determine size of the
array.  Improvement suggested by coccinelle.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: remove redundant assignment of d

The initialization of d is redundant as this value is never read and it
is overwritten inside the subsequent for-loop.  Remove this redundant
assignment.

Cleans up clang warning:
drivers/scsi/qla2xxx/qla_gs.c:3985:29: warning: Value stored to 'd'
during its initialization is never read

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Use zeroing allocator instead of allocator/memset

Use dma_zalloc_coherent instead of dma_alloc_coherent followed by memset
0.

Generated-by: scripts/coccinelle/api/alloc/kzalloc-simple.cocci
Suggested-by: Luis R. Rodriguez <mcgrof@kernel.org>
Signed-off-by: Himanshu Jha <himanshujha199640@gmail.com>
Acked-by: Manish Rangankar <Manish.Rangankar@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bnx2fc: Use zeroing allocator rather than allocator/memset

Use dma_zalloc_coherent instead of dma_alloc_coherent followed by memset
0.

Generated-by: scripts/coccinelle/api/alloc/kzalloc-simple.cocci
Suggested-by: Luis R. Rodriguez <mcgrof@kernel.org>
Signed-off-by: Himanshu Jha <himanshujha199640@gmail.com>
Acked-by: Chad Dupuis <chad.dupuis@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hisi_sas: Change frame type for SET MAX commands

According to ATA protocol, SET MAX commands belong to different frame
types. So judge features field of SET MAX commands to decide which
frame type they belongs to.

Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ata: enhance the definition of SET MAX feature field value

There are two other values for SET MAX feature field according to ata
protocol. So definite them.

Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: allow static build ("built-in")

If CONFIG_SCSI_SMARTPQI=y then don't build this driver as a module.

Signed-off-by: Steffen Weber <steffen.weber@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_debug: add resp_write_scat function

Add resp_write_scat() function to support decoding WRITE SCATTERED
(16 and 32). Also weave resp_write_scat() into the cdb decoding
logic.

Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_debug: ARRAY_SIZE and FF_MEDIA_IO

Reviewer suggested using the ARRAY_SIZE macro. That reduced one of the subtle
inter-dependencies in the parser's tables.

It is important that commands which simulate media access, indicate this in the
flags for that command. The flag to do that was FF_DIRECT_IO. On reflection
FF_MEDIA_IO seems a more accurate description.

Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_debug: do_device_access add sg offset argument

WRITE SCATTERED needs to take several "bites" out of the data-out buffer.
Expand the do_device_access() function to take a sg_skip argument.

Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_debug: fix group_number mask

Various cdb masks incorrectly assumed the GROUP NUMBER field
was 5 bits long. It is actually 6 bits long. Correct.
Also fix mask failure (in same byte) to allow DLD0 in READ(16)
and WRITE(16).

Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_debug: tab, kstrto changes

Some of my development tools tend to add spaces (my preference) rather
than tabs (kernel convention). Running unexpand to clean these spaces
up found more of them than checkpatch.pl did. Then checkpatch.pl
complained about other style violations in those newly tabbed lines.

Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: lockless command submission

Performance improvement using block layer tag.

Curent driver gets scsiio tracker and free smid from link list and array
based tracking managed by driver.  Accessing list in main io path is
performance pentaly because of protection using spinlock
"scsi_lookup_lock".

In this patch:

1. Driver removes all link list access from main io path and
   use scmd->request->tag to get free smid.

2. Instead of holding 'struct scsiio_tracker' in its own pool
   driver can embed it into the scsi command.

Driver provides cmd_size in scsi_host_template, so that struct
scsiio_tracker is preallocated by scsi mid layer for each scsi command.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: simplify _wait_for_commands_to_complete()

Use 'host_busy' instead of counting outstanding commands by hand.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: simplify mpt3sas_scsi_issue_tm()

Move the check for outstanding commands out of the function allowing us
to simplify the overall code.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: simplify task management functions

No functional change. Code optimization.

One can simply check 'target_busy' or 'device_busy' when figuring out if
there are outstanding commands; no need to painstakingly count them by
hand.

[mkp: tweaked patch description]

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: always use first reserved smid for ioctl passthrough

ioctl passthrough commands require a SCSIIO smid, but cannot easily
integrate with the block layer. But the driver already has reserved some
SCSIIO smids and we're only ever allowing one ioctl command at a time we
can use the first reserved smid for ioctl commands.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: check command status before attempting abort

When attempting a command abort we should check the command status prior
to sending the abort; the command might've been completed already.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Introduce mpt3sas_get_st_from_smid()

Abstract accesses to the scsi_lookup array by introducing
mpt3sas_get_st_from_smid().

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: open-code _scsih_scsi_lookup_get()

Just a wrapper around the scsi lookup array and only used in one place,
so open-code it.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: separate out _base_recovery_check()

No functional change.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: use list_splice_init()

Use 'list_splice_init()' instead of hand-crafted function.  No
functional change.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: set default value for cb_idx

No functional change

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxlflash: Staging to support future accelerators

As staging to support future accelerator transports, add a shim layer
such that the underlying services the cxlflash driver requires can be
conditional upon the accelerator infrastructure.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxlflash: Adapter context init can return error

Adapter context creation can return either NULL or an error pointer.
Updating the check condition to reflect this.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxlflash: Remove embedded CXL work structures

The CXL-specific work structure used to request the number of interrupts
currently resides as a nested member of both the context information and
hardware queue structures. It is used to cache values (specifically the
number of interrupts) required by the CXL layer when starting a context.

To facilitate staging that will ultimately allow the cxlflash core to
become agnostic of the underlying accelerator transport, remove these
embedded work structures.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxlflash: Explicitly cache number of interrupts per context

The number of interrupts a user requests during a context attach is
presently stored within the CXL work ioctl structure that is nested
alongside the per context metadata. Keeping this data in a structure
that is tied to a particular hardware implementation (CXL) will only
complicate matters when supporting newer accelerator transports.

Instead of relying upon the number of interrupts being cached within
a CXL-specific structure, explicitly cache the value within the context
information structure.

Signed-off-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxlflash: Update cxl-specific arguments to generic cookie

Convert cxl-specific pointers to generic cookies to facilitate future
enhancements.

Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxlflash: Reset command ioasc

In the event of a command failure, cxlflash returns the failure to the upper
layers to process. After processing the error, when the command is queued
again, the private command structure will not be zeroed and the ioasc could be
stale. Per the SISLite specification, the AFU only sets the ioasc in the
presence of a failure. Thus, even though the original command succeeds the
second time, the command is considered a failure due to stale ioasc. This
cycle repeats indefinitely and can cause a hang or IO failure.

To fix the issue, clear the ioasc before queuing any command.

[mkp: added Cc: stable per request]

Fixes: 479ad8e9d48c ("scsi: cxlflash: Remove zeroing of private command data")
Cc: <stable@vger.kernel.org>
Signed-off-by: Uma Krishnan <ukrishn@linux.vnet.ibm.com>
Acked-by: Matthew R. Ochs <mrochs@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: notify event PORTE_BROADCAST_RCVD in sas_enable_revalidation()

There are two places queuing the disco event DISCE_REVALIDATE_DOMAIN.
One is in sas_porte_broadcast_rcvd() and uses sas_chain_event() to queue
the event. The other is in sas_enable_revalidation() and uses
sas_queue_event() to queue the event. We have diffrent work queues for
event and discovery now, so the DISCE_REVALIDATE_DOMAIN event may be
processed in both event queue and discovery queue.

Now since we do synchronous event handling, we cannot do it in discovery
queue, so have to trigger a fake broadcast event to re-trigger the
revalidation from event queue.

Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: John Garry <john.garry@huawei.com>
CC: Johannes Thumshirn <jthumshirn@suse.de>
CC: Ewan Milne <emilne@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Tomas Henzl <thenzl@redhat.com>
CC: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: direct call probe and destruct

In commit 87c8331fcf72 ("[SCSI] libsas: prevent domain rediscovery
competing with ata error handling") introduced disco mutex to prevent
rediscovery competing with ata error handling and put the whole
revalidation in the mutex. But the rphy add/remove needs to wait for the
error handling which also grabs the disco mutex. This may leads to dead
lock.So the probe and destruct event were introduce to do the rphy
add/remove asynchronously and out of the lock.

The asynchronously processed workers makes the whole discovery process
not atomic, the other events may interrupt the process. For example,
if a loss of signal event inserted before the probe event, the
sas_deform_port() is called and the port will be deleted.

And sas_port_delete() may run before the destruct event, but the
port-x:x is the top parent of end device or expander. This leads to
a kernel WARNING such as:

[   82.042979] sysfs group 'power' not found for kobject 'phy-1:0:22'
[   82.042983] ------------[ cut here ]------------
[   82.042986] WARNING: CPU: 54 PID: 1714 at fs/sysfs/group.c:237
sysfs_remove_group+0x94/0xa0
[   82.043059] Call trace:
[   82.043082] [<ffff0000082e7624>] sysfs_remove_group+0x94/0xa0
[   82.043085] [<ffff00000864e320>] dpm_sysfs_remove+0x60/0x70
[   82.043086] [<ffff00000863ee10>] device_del+0x138/0x308
[   82.043089] [<ffff00000869a2d0>] sas_phy_delete+0x38/0x60
[   82.043091] [<ffff00000869a86c>] do_sas_phy_delete+0x6c/0x80
[   82.043093] [<ffff00000863dc20>] device_for_each_child+0x58/0xa0
[   82.043095] [<ffff000008696f80>] sas_remove_children+0x40/0x50
[   82.043100] [<ffff00000869d1bc>] sas_destruct_devices+0x64/0xa0
[   82.043102] [<ffff0000080e93bc>] process_one_work+0x1fc/0x4b0
[   82.043104] [<ffff0000080e96c0>] worker_thread+0x50/0x490
[   82.043105] [<ffff0000080f0364>] kthread+0xfc/0x128
[   82.043107] [<ffff0000080836c0>] ret_from_fork+0x10/0x50

Make probe and destruct a direct call in the disco and revalidate function,
but put them outside the lock. The whole discovery or revalidate won't
be interrupted by other events. And the DISCE_PROBE and DISCE_DESTRUCT
event are deleted as a result of the direct call.

Introduce a new list to destruct the sas_port and put the port delete after
the destruct. This makes sure the right order of destroying the sysfs
kobject and fix the warning above.

In sas_ex_revalidate_domain() have a loop to find all broadcasted
device, and sometimes we have a chance to find the same expander twice.
Because the sas_port will be deleted at the end of the whole revalidate
process, sas_port with the same name cannot be added before this.
Otherwise the sysfs will complain of creating duplicate filename. Since
the LLDD will send broadcast for every device change, we can only
process one expander's revalidation.

[mkp: kbuild test robot warning]

Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: John Garry <john.garry@huawei.com>
CC: Johannes Thumshirn <jthumshirn@suse.de>
CC: Ewan Milne <emilne@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Tomas Henzl <thenzl@redhat.com>
CC: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>