muen/linux.git
3 years agoscsi: mvsas: Remove set but not used variable 'id'
YueHaibing [Sun, 21 Oct 2018 09:50:54 +0000 (09:50 +0000)]
scsi: mvsas: Remove set but not used variable 'id'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/scsi/mvsas/mv_sas.c: In function 'mvs_work_queue':
drivers/scsi/mvsas/mv_sas.c:1909:31: warning:
 variable 'id' set but not used [-Wunused-but-set-variable]

It never used since introduction in commit
20b09c2992fe ("[SCSI] mvsas: add support for 94xx; layout change; bug fixes")

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Remove two arguments from qlafx00_error_entry()
Bart Van Assche [Thu, 18 Oct 2018 22:45:46 +0000 (15:45 -0700)]
scsi: qla2xxx: Remove two arguments from qlafx00_error_entry()

Move a debug statement from qlafx00_error_entry() into its caller. Remove
one unused argument from that function. This patch does not change the
behavior of the qla2xxx driver.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Make sure that qlafx00_ioctl_iosb_entry() initializes 'res'
Bart Van Assche [Thu, 18 Oct 2018 22:45:45 +0000 (15:45 -0700)]
scsi: qla2xxx: Make sure that qlafx00_ioctl_iosb_entry() initializes 'res'

Only one of the two code paths in qlafx00_ioctl_iosb_entry() initializes
the variable 'res'. Make sure that 'res' is initialized before
sp->done(sp, res) is called.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Remove a set-but-not-used variable
Bart Van Assche [Thu, 18 Oct 2018 22:45:44 +0000 (15:45 -0700)]
scsi: qla2xxx: Remove a set-but-not-used variable

This patch does not change any functionality.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Make qla2x00_sysfs_write_nvram() easier to analyze
Bart Van Assche [Thu, 18 Oct 2018 22:45:43 +0000 (15:45 -0700)]
scsi: qla2xxx: Make qla2x00_sysfs_write_nvram() easier to analyze

Modify the unlock statement such that it becomes easier for static
analyzers to analyze it. This patch does not change any functionality.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Declare local functions 'static'
Bart Van Assche [Thu, 18 Oct 2018 22:45:42 +0000 (15:45 -0700)]
scsi: qla2xxx: Declare local functions 'static'

This patch avoids that the compiler complains about missing declarations
when building with W=1.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Improve several kernel-doc headers
Bart Van Assche [Thu, 18 Oct 2018 22:45:41 +0000 (15:45 -0700)]
scsi: qla2xxx: Improve several kernel-doc headers

This patch avoids that complaints about kernel-doc headers are reported
when building with W=1.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Modify fall-through annotations
Bart Van Assche [Thu, 18 Oct 2018 22:45:40 +0000 (15:45 -0700)]
scsi: qla2xxx: Modify fall-through annotations

This patch avoids that the compiler complains about missing fall-through
annotations when building with W=1.

Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: 3w-sas: 3w-9xxx: Use unsigned char for cdb
Nathan Chancellor [Thu, 18 Oct 2018 21:55:41 +0000 (14:55 -0700)]
scsi: 3w-sas: 3w-9xxx: Use unsigned char for cdb

Clang warns a few times:

drivers/scsi/3w-sas.c:386:11: warning: implicit conversion from 'int' to
'char' changes value from 128 to -128 [-Wconstant-conversion]
        cdb[4] = TW_ALLOCATION_LENGTH; /* allocation length */
               ~ ^~~~~~~~~~~~~~~~~~~~

Update cdb's type to unsigned char, which matches the type of the cdb
member in struct TW_Command_Apache.

Link: https://github.com/ClangBuiltLinux/linux/issues/158
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Adam Radford <aradford@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: mvsas: Use dma_pool_zalloc
Sabyasachi Gupta [Thu, 18 Oct 2018 16:06:26 +0000 (21:36 +0530)]
scsi: mvsas: Use dma_pool_zalloc

Replace dma_pool_alloc + memset with dma_pool_zalloc.

Signed-off-by: Sabyasachi Gupta <sabyasachi.linux@gmail.com>
Reviewed-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: Don't request modules that aren't even built
Roland Dreier [Sat, 20 Oct 2018 16:45:16 +0000 (09:45 -0700)]
scsi: target: Don't request modules that aren't even built

If, for example, I don't enable CONFIG_TCM_PSCSI, then every time I load
the target subsystem, I get an annoying

    Unable to load target_core_pscsi

kernel log message.  Instead let's only request_module() on things if
that code is enabled.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: Set response length for REPORT TARGET PORT GROUPS
Roland Dreier [Sat, 20 Oct 2018 16:45:04 +0000 (09:45 -0700)]
scsi: target: Set response length for REPORT TARGET PORT GROUPS

One more place where we can return the length we actually fill in.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: myrs: Fix the processor absent message in processor_show()
Dan Carpenter [Fri, 19 Oct 2018 09:18:19 +0000 (12:18 +0300)]
scsi: myrs: Fix the processor absent message in processor_show()

If both processors are absent then it's supposed to print that, but
instead we print that just the second processor is absent.

Fixes: 77266186397c ("scsi: myrs: Add Mylex RAID controller (SCSI interface)")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: myrs: Fix a logical vs bitwise bug
Dan Carpenter [Fri, 19 Oct 2018 09:16:28 +0000 (12:16 +0300)]
scsi: myrs: Fix a logical vs bitwise bug

The || was supposed to be |.  The original code just sets ->result to 1.

Fixes: 77266186397c ("scsi: myrs: Add Mylex RAID controller (SCSI interface)")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: hisi_sas: Fix NULL pointer dereference
Gustavo A. R. Silva [Thu, 18 Oct 2018 16:59:39 +0000 (18:59 +0200)]
scsi: hisi_sas: Fix NULL pointer dereference

There is a NULL pointer dereference in case *slot* happens to be NULL at
lines 1053 and 1878:

struct hisi_sas_cq *cq =
&hisi_hba->cq[slot->dlvry_queue];

Notice that *slot* is being NULL checked at lines 1057 and 1881:
if (slot), which implies it may be NULL.

Fix this by placing the declaration and definition of variable cq, which
contains the pointer dereference slot->dlvry_queue, after slot has been
properly NULL checked.

Addresses-Coverity-ID: 1474515 ("Dereference before null check")
Addresses-Coverity-ID: 1474520 ("Dereference before null check")
Fixes: 584f53fe5f52 ("scsi: hisi_sas: Fix the race between IO completion and timeout for SMP/internal IO")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: myrs: fix build failure on 32 bit
James Bottomley [Thu, 18 Oct 2018 23:50:56 +0000 (16:50 -0700)]
scsi: myrs: fix build failure on 32 bit

For 32 bit versions we have to be careful about divisions of 64 bit
quantities so use do_div() instead of a direct division.  This fixes a
warning about _uldivmod being undefined in certain configurations

Fixes: 77266186397c ("scsi: myrs: Add Mylex RAID controller")
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Tested-by: Randy Dunlap <rdunlap@infradead.org> # build-tested
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: fnic: replace gross legacy tag hack with blk-mq hack
Jens Axboe [Tue, 16 Oct 2018 14:37:23 +0000 (08:37 -0600)]
scsi: fnic: replace gross legacy tag hack with blk-mq hack

Would be nice to fix up the SCSI midlayer instead, but this will do for
now.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Satish Kharat <satishkh@cisco.com>
Cc: linux-scsi@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: mesh: switch to generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 08:17:35 +0000 (10:17 +0200)]
scsi: mesh: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: ips: switch to generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 08:15:35 +0000 (10:15 +0200)]
scsi: ips: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: smartpqi: fully convert to the generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 07:47:59 +0000 (09:47 +0200)]
scsi: smartpqi: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Don Brace <don.brace@microchip.com>
Acked-by: Don Brace <don.brace@microchip.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: vmw_pscsi: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 18:22:40 +0000 (20:22 +0200)]
scsi: vmw_pscsi: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: snic: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 18:16:41 +0000 (20:16 +0200)]
scsi: snic: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla4xxx: fully convert to the generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 07:56:58 +0000 (09:56 +0200)]
scsi: qla4xxx: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: fully convert to the generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 07:42:07 +0000 (09:42 +0200)]
scsi: qla2xxx: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla1280: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 18:09:04 +0000 (20:09 +0200)]
scsi: qla1280: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qedi: fully convert to the generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 18:06:13 +0000 (20:06 +0200)]
scsi: qedi: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Manish Rangankar <Manish.Rangankar@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qedf: fully convert to the generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 18:04:15 +0000 (20:04 +0200)]
scsi: qedf: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Chad Dupuis <chad.dupuis@cavium.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: pm8001: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 17:59:50 +0000 (19:59 +0200)]
scsi: pm8001: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: nsp32: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 17:55:59 +0000 (19:55 +0200)]
scsi: nsp32: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: mvsas: fully convert to the generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 07:54:21 +0000 (09:54 +0200)]
scsi: mvsas: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: mvumi: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 17:53:14 +0000 (19:53 +0200)]
scsi: mvumi: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Also reuse an existing helper (after fixing the error return) to set the
DMA mask instead of having three copies of the code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: mpt3sas: switch to generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 07:35:25 +0000 (09:35 +0200)]
scsi: mpt3sas: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Also simplify setting the DMA mask a bit.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: megaraid_sas: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 17:31:25 +0000 (19:31 +0200)]
scsi: megaraid_sas: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: megaraid_mbox: switch to generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 07:34:24 +0000 (09:34 +0200)]
scsi: megaraid_mbox: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: hpsa: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:56:15 +0000 (18:56 +0200)]
scsi: hpsa: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Don Brace <don.brace@microchip.com>
Acked-by: Don Brace <don.brace@microchip.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: fnic: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:40:56 +0000 (18:40 +0200)]
scsi: fnic: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: csiostor: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:34:51 +0000 (18:34 +0200)]
scsi: csiostor: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: be2iscsi: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:22:24 +0000 (18:22 +0200)]
scsi: be2iscsi: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: atp870u: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:14:14 +0000 (18:14 +0200)]
scsi: atp870u: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: a100u2w: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:09:42 +0000 (18:09 +0200)]
scsi: a100u2w: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: BusLogic: switch to generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:03:24 +0000 (18:03 +0200)]
scsi: BusLogic: switch to generic DMA API

Switch from the legacy PCI DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: 3w-sas: fully convert to the generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 16:06:50 +0000 (18:06 +0200)]
scsi: 3w-sas: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Adam Radford <aradford@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: 3w-xxx: fully convert to the generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 15:58:50 +0000 (17:58 +0200)]
scsi: 3w-xxx: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Adam Radford <aradford@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: 3w-9xxx: fully convert to the generic DMA API
Christoph Hellwig [Wed, 10 Oct 2018 15:53:41 +0000 (17:53 +0200)]
scsi: 3w-9xxx: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Adam Radford <aradford@gmail.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: aic94xx: fully convert to the generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 08:02:55 +0000 (10:02 +0200)]
scsi: aic94xx: fully convert to the generic DMA API

The driver is currently using an odd mix of legacy PCI DMA API and
generic DMA API calls, switch it over to the generic API entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: esp_scsi: Optimize PIO loops
Finn Thain [Tue, 16 Oct 2018 05:31:25 +0000 (16:31 +1100)]
scsi: esp_scsi: Optimize PIO loops

Avoid function calls in the inner PIO loops. On a Centris 660av this
improves throughput for sequential read transfers by about 40% and
sequential write by about 10%.

Unfortunately it is not possible to have methods like .esp_write8 placed
inline so this is always going to be slow, even with LTO.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: esp_scsi: De-duplicate PIO routines
Finn Thain [Tue, 16 Oct 2018 05:31:25 +0000 (16:31 +1100)]
scsi: esp_scsi: De-duplicate PIO routines

As a temporary measure, the code to implement PIO transfers was
duplicated in zorro_esp and mac_esp. Now that it has stabilized move the
common code into the core driver but don't build it unless needed.

This replaces the inline assembler with more portable writesb() calls.
Optimizing the m68k writesb() implementation is a separate patch.

[mkp: applied by hand]

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Tested-by: Stan Johnson <userm57@yahoo.com>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: esp_scsi: Eliminate ESP_FLAG_DOING_SLOWCMD
Finn Thain [Tue, 16 Oct 2018 05:31:25 +0000 (16:31 +1100)]
scsi: esp_scsi: Eliminate ESP_FLAG_DOING_SLOWCMD

The concept of a 'slow command' as it appears in esp_scsi is confusing
because it could refer to an ESP command or a SCSI command. It turns out
that it refers to a particular ESP select command which the driver also
tracks as 'ESP_SELECT_MSGOUT'. For readability, it is better to use the
terminology from the datasheets.

The global ESP_FLAG_DOING_SLOWCMD flag is redundant anyway, as it can be
inferred from esp->select_state. Remove the ESP_FLAG_DOING_SLOWCMD cruft
and just use a boolean local variable.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: esp_scsi: Grant disconnect privilege for untagged commands
Finn Thain [Tue, 16 Oct 2018 05:31:25 +0000 (16:31 +1100)]
scsi: esp_scsi: Grant disconnect privilege for untagged commands

A SCSI device is not granted disconnect privilege by an esp_scsi host
unless that device has its simple_tags flag set. However, a device may
support disconnect/reselect and not support command queueing. Allow such
devices to disconnect and thereby improve bus utilization.

Drop the redundant 'lp' check. The mid-layer invokes .slave_alloc and
.slave_destroy in such a way that we may rely on scmd->device->hostdata
for as long as scmd belongs to the low-level driver.

Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: esp_scsi: Track residual for PIO transfers
Finn Thain [Tue, 16 Oct 2018 05:31:25 +0000 (16:31 +1100)]
scsi: esp_scsi: Track residual for PIO transfers

If a target disconnects during a PIO data transfer the command may fail
when the target reconnects:

scsi host1: DMA length is zero!
scsi host1: cur adr[04380000] len[00000000]

The scsi bus is then reset. This happens because the residual reached
zero before the transfer was completed.

The usual residual calculation relies on the Transfer Count registers.
That works for DMA transfers but not for PIO transfers. Fix the problem
by storing the PIO transfer residual and using that to correctly
calculate bytes_sent.

Fixes: 6fe07aaffbf0 ("[SCSI] m68k: new mac_esp scsi driver")
Tested-by: Stan Johnson <userm57@yahoo.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: zorro_esp: Limit DMA transfers to 65535 bytes
Finn Thain [Tue, 16 Oct 2018 05:31:25 +0000 (16:31 +1100)]
scsi: zorro_esp: Limit DMA transfers to 65535 bytes

The core driver, esp_scsi, does not use the ESP_CONFIG2_FENAB bit, so the
chip's Transfer Counter register is only 16 bits wide (not 24).  A larger
transfer cannot work and will theoretically result in a failed command
and a "DMA length is zero" error.

Fixes: 3109e5ae0311 ("scsi: zorro_esp: New driver for Amiga Zorro NCR53C9x boards")
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Reviewed-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: dc395x: use generic DMA API
Christoph Hellwig [Thu, 11 Oct 2018 14:17:16 +0000 (16:17 +0200)]
scsi: dc395x: use generic DMA API

Convert the driver from the legacy pci_* DMA API to the generic DMA API.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: dc395x: fix DMA API usage in sg_update_list
Christoph Hellwig [Thu, 11 Oct 2018 14:17:15 +0000 (16:17 +0200)]
scsi: dc395x: fix DMA API usage in sg_update_list

We need to transfer device ownership to the CPU before we can manipulate
the mapped data.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: dc395x: fix dma API usage in srb_done
Christoph Hellwig [Thu, 11 Oct 2018 14:17:14 +0000 (16:17 +0200)]
scsi: dc395x: fix dma API usage in srb_done

We can't just transfer ownership to the CPU and then unmap, as this will
break with swiotlb.

Instead unmap the command and sense buffer a little earlier in the I/O
completion handler and get rid of the pci_dma_sync_sg_for_cpu call
entirely.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: dc395x: simplify list handling
Christoph Hellwig [Thu, 11 Oct 2018 14:17:13 +0000 (16:17 +0200)]
scsi: dc395x: simplify list handling

Remove the list wrappers, including the pointless list iteration before
deletion.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: myrs: Add Mylex RAID controller (SCSI interface)
Hannes Reinecke [Wed, 17 Oct 2018 15:25:12 +0000 (17:25 +0200)]
scsi: myrs: Add Mylex RAID controller (SCSI interface)

This patch adds support for the Mylex DAC960 RAID controller,
supporting the newer, SCSI-based interface.  The driver is a
re-implementation of the original DAC960 driver.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: myrb: Add Mylex RAID controller (block interface)
Hannes Reinecke [Wed, 17 Oct 2018 15:25:11 +0000 (17:25 +0200)]
scsi: myrb: Add Mylex RAID controller (block interface)

This patch adds support for the Mylex DAC960 RAID controller,
supporting the older, block-based interface only.  The driver is a
re-implementation of the original DAC960 driver.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: advansys: remove unused variable 'srb_tag' in adv_isr_callback
YueHaibing [Wed, 17 Oct 2018 12:34:32 +0000 (20:34 +0800)]
scsi: advansys: remove unused variable 'srb_tag' in adv_isr_callback

drivers/scsi/advansys.c: In function 'adv_isr_callback':
drivers/scsi/advansys.c:5952:6: warning:
 variable 'srb_tag' set but not used [-Wunused-but-set-variable]

It never used since introduction in
commit 9c17c62aedb0 ("advansys: use shared host tag map for command lookup")

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: core: Remove scsi_block_when_processing_errors: message
Laurence Oberman [Tue, 16 Oct 2018 20:39:16 +0000 (16:39 -0400)]
scsi: core: Remove scsi_block_when_processing_errors: message

This message floods the log when enabling mask 0x7 for
/proc/sys/dev/scsi/logging_level:

 xxxxxxxx kernel: scsi_block_when_processing_errors: rtn: 1

It's not needed and makes tracing just scsi_eh* messages way too
verbose so get rid of it.

[mkp: mangled patch, applied by hand]

Signed-off-by: Laurence Oberman <loberman@redhat.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Chad Dupuis <chad.dupuis@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: ibmvscsi_tgt: Remove target_wait_for_sess_cmd()
Bryant G. Ly [Tue, 16 Oct 2018 17:34:26 +0000 (17:34 +0000)]
scsi: ibmvscsi_tgt: Remove target_wait_for_sess_cmd()

There is currently a bug with the driver where there is never a call to
target_sess_cmd_list_set_waiting(), it only called
target_wait_for_sess_cmd(), which basically means that the sess_wait_list
would always be empty.

Thus, list_empty(&sess->sess_wait_list) = true,
(eg: no se_cmd I/O is quiesced, because no se_cmd in sess_wait_list),
since commit 712db3eb2c35 ("scsi: ibmvscsis: Properly deregister
target sessions") in 4.9.y code.

ibmvscsi_tgt does not remove the I_T Nexus when a VM is active so we can
fix this issue by removing the call to target_wait_for_sess_cmd()
altogether.

Signed-off-by: Bryant G. Ly <bly@catalogicsoftware.com>
Reviewed-by: Mike Christie <mchristi@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: ufs: Schedule clk gating work on correct queue
Evan Green [Fri, 5 Oct 2018 17:27:32 +0000 (10:27 -0700)]
scsi: ufs: Schedule clk gating work on correct queue

With commit 10e5e37581fc ("scsi: ufs: Add clock ungating to a separate
workqueue"), clock gating work was moved to a separate work queue with
WQ_MEM_RECLAIM set, since clock gating could occur from a memory reclaim
context. Unfortunately, clk_gating.gate_work was left queued via
schedule_delayed_work, which is a system workqueue that does not have
WQ_MEM_RECLAIM set.  Because ufshcd_ungate_work attempts to cancel
gate_work, the following warning appears:

[   14.174170] workqueue: WQ_MEM_RECLAIM ufs_clk_gating_0:ufshcd_ungate_work is flushing !WQ_MEM_RECLAIM events:ufshcd_gate_work
[   14.174179] WARNING: CPU: 4 PID: 173 at kernel/workqueue.c:2440 check_flush_dependency+0x110/0x118
[   14.205725] CPU: 4 PID: 173 Comm: kworker/u16:3 Not tainted 4.14.68 #1
[   14.212437] Hardware name: Google Cheza (rev1) (DT)
[   14.217459] Workqueue: ufs_clk_gating_0 ufshcd_ungate_work
[   14.223107] task: ffffffc0f6a40080 task.stack: ffffff800a490000
[   14.229195] PC is at check_flush_dependency+0x110/0x118
[   14.234569] LR is at check_flush_dependency+0x110/0x118
[   14.239944] pc : [<ffffff80080cad14>] lr : [<ffffff80080cad14>] pstate: 60c001c9
[   14.333050] Call trace:
[   14.427767] [<ffffff80080cad14>] check_flush_dependency+0x110/0x118
[   14.434219] [<ffffff80080cafec>] start_flush_work+0xac/0x1fc
[   14.440046] [<ffffff80080caeec>] flush_work+0x40/0x94
[   14.445246] [<ffffff80080cb288>] __cancel_work_timer+0x11c/0x1b8
[   14.451433] [<ffffff80080cb4b8>] cancel_delayed_work_sync+0x20/0x30
[   14.457886] [<ffffff80085b9294>] ufshcd_ungate_work+0x24/0xd0
[   14.463800] [<ffffff80080cfb04>] process_one_work+0x32c/0x690
[   14.469713] [<ffffff80080d0154>] worker_thread+0x218/0x338
[   14.475361] [<ffffff80080d527c>] kthread+0x120/0x130
[   14.480470] [<ffffff8008084814>] ret_from_fork+0x10/0x18

The simple solution is to put the gate_work on the same WQ_MEM_RECLAIM
work queue as the ungate_work.

Fixes: 10e5e37581fc ("scsi: ufs: Add clock ungating to a separate workqueue")
Signed-off-by: Evan Green <evgreen@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: lpfc: Remove set but not used variables 'tgtp'
YueHaibing [Thu, 4 Oct 2018 10:51:02 +0000 (10:51 +0000)]
scsi: lpfc: Remove set but not used variables 'tgtp'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/scsi/lpfc/lpfc_debugfs.c: In function 'lpfc_debugfs_nodelist_data':
drivers/scsi/lpfc/lpfc_debugfs.c:553:29: warning:
 variable 'tgtp' set but not used [-Wunused-but-set-variable]

It never used since 2b65e18202fd ("scsi: lpfc: NVME Target: Add debugfs support")

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: aic94xx: mark expected switch fall-throughs
Gustavo A. R. Silva [Thu, 4 Oct 2018 07:10:09 +0000 (09:10 +0200)]
scsi: aic94xx: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114988 ("Missing break in switch")
Addresses-Coverity-ID: 114989 ("Missing break in switch")
Addresses-Coverity-ID: 114990 ("Missing break in switch")
Addresses-Coverity-ID: 114991 ("Missing break in switch")
Addresses-Coverity-ID: 114992 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: iscsi_tcp: Explicitly cast param in iscsi_sw_tcp_host_get_param
Nathan Chancellor [Thu, 4 Oct 2018 01:06:15 +0000 (18:06 -0700)]
scsi: iscsi_tcp: Explicitly cast param in iscsi_sw_tcp_host_get_param

Clang warns when one enumerated type is implicitly converted to another.

drivers/scsi/iscsi_tcp.c:803:15: warning: implicit conversion from
enumeration type 'enum iscsi_host_param' to different enumeration type
'enum iscsi_param' [-Wenum-conversion]
                                                 &addr, param, buf);
                                                        ^~~~~
1 warning generated.

iscsi_conn_get_addr_param handles ISCSI_HOST_PARAM_IPADDRESS just fine
so add an explicit cast to iscsi_param to make it clear to Clang that
this is expected behavior.

Link: https://github.com/ClangBuiltLinux/linux/issues/153
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: lpfc: Remove set but not used variable 'psli'
YueHaibing [Sat, 29 Sep 2018 13:53:07 +0000 (13:53 +0000)]
scsi: lpfc: Remove set but not used variable 'psli'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/scsi/lpfc/lpfc_hbadisc.c: In function 'lpfc_free_tx':
drivers/scsi/lpfc/lpfc_hbadisc.c:5431:19: warning:
 variable 'psli' set but not used [-Wunused-but-set-variable]

Since commit 895427bd012c ("scsi: lpfc: NVME Initiator: Base modifications")
'psli' is not used any more.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: bfa: Avoid implicit enum conversion in bfad_im_post_vendor_event
Nathan Chancellor [Thu, 27 Sep 2018 23:56:52 +0000 (16:56 -0700)]
scsi: bfa: Avoid implicit enum conversion in bfad_im_post_vendor_event

Clang warns when one enumerated type is implicitly converted to another.

drivers/scsi/bfa/bfa_fcs_lport.c:379:26: warning: implicit conversion
from enumeration type 'enum bfa_lport_aen_event' to different
enumeration type 'enum bfa_ioc_aen_event' [-Wenum-conversion]
                                  BFA_AEN_CAT_LPORT, event);
                                                     ^~~~~

The root cause of these warnings is the bfad_im_post_vendor_event
function, which expects a value from enum bfa_ioc_aen_event but there
are multiple instances of values from enums bfa_port_aen_event,
bfa_audit_aen_event, and bfa_lport_aen_event being used in this
function.

Given that this doesn't appear to be a problem since cat helps with
differentiating the events, just change evt's type to int so that no
conversion needs to happen and Clang won't warn. Update aen_type's type
in bfa_aen_entry_s as members that hold enumerated types should be int.

Link: https://github.com/ClangBuiltLinux/linux/issues/147
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: isci: Change sci_controller_start_task's return type to sci_status
Nathan Chancellor [Thu, 27 Sep 2018 00:12:00 +0000 (17:12 -0700)]
scsi: isci: Change sci_controller_start_task's return type to sci_status

Clang warns when an enumerated type is implicitly converted to another.

drivers/scsi/isci/request.c:3476:13: warning: implicit conversion from
enumeration type 'enum sci_task_status' to different enumeration type
'enum sci_status' [-Wenum-conversion]
                        status = sci_controller_start_task(ihost,
                               ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/scsi/isci/host.c:2744:10: warning: implicit conversion from
enumeration type 'enum sci_status' to different enumeration type 'enum
sci_task_status' [-Wenum-conversion]
                return SCI_SUCCESS;
                ~~~~~~ ^~~~~~~~~~~
drivers/scsi/isci/host.c:2753:9: warning: implicit conversion from
enumeration type 'enum sci_status' to different enumeration type 'enum
sci_task_status' [-Wenum-conversion]
        return status;
        ~~~~~~ ^~~~~~

Avoid all of these implicit conversion by just making
sci_controller_start_task use sci_status. This silences
Clang and has no functional change since sci_task_status
has all of its values mapped to something in sci_status.

Link: https://github.com/ClangBuiltLinux/linux/issues/153
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: isci: Use proper enumerated type in atapi_d2h_reg_frame_handler
Nathan Chancellor [Thu, 27 Sep 2018 00:11:50 +0000 (17:11 -0700)]
scsi: isci: Use proper enumerated type in atapi_d2h_reg_frame_handler

Clang warns when one enumerated type is implicitly converted to another.

drivers/scsi/isci/request.c:1629:13: warning: implicit conversion from
enumeration type 'enum sci_io_status' to different enumeration type
'enum sci_status' [-Wenum-conversion]
                        status = SCI_IO_FAILURE_RESPONSE_VALID;
                               ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/scsi/isci/request.c:1631:12: warning: implicit conversion from
enumeration type 'enum sci_io_status' to different enumeration type
'enum sci_status' [-Wenum-conversion]
                status = SCI_IO_FAILURE_RESPONSE_VALID;
                       ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

status is of type sci_status but SCI_IO_FAILURE_RESPONSE_VALID is of
type sci_io_status. Use SCI_FAILURE_IO_RESPONSE_VALID, which is from
sci_status and has SCI_IO_FAILURE_RESPONSE_VALID's exact value since
that is what SCI_IO_FAILURE_RESPONSE_VALID is mapped to in the isci.h
file.

Link: https://github.com/ClangBuiltLinux/linux/issues/153
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: lpfc: Remove set but not used variables 'fc_hdr' and 'hw_page_size'
YueHaibing [Fri, 14 Sep 2018 10:41:44 +0000 (10:41 +0000)]
scsi: lpfc: Remove set but not used variables 'fc_hdr' and 'hw_page_size'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/scsi/lpfc/lpfc_sli.c: In function 'lpfc_sli4_sp_handle_rcqe':
drivers/scsi/lpfc/lpfc_sli.c:13430:26: warning:
 variable 'fc_hdr' set but not used [-Wunused-but-set-variable]

drivers/scsi/lpfc/lpfc_sli.c: In function 'lpfc_cq_create':
drivers/scsi/lpfc/lpfc_sli.c:14852:11: warning:
 variable 'hw_page_size' set but not used [-Wunused-but-set-variable]

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: cxgb4i: add DCB support for iSCSI connections
Varun Prakash [Thu, 13 Sep 2018 15:56:00 +0000 (21:26 +0530)]
scsi: cxgb4i: add DCB support for iSCSI connections

Add IEEE and CEE DCBX support for iSCSI connections.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: megaraid_mbox: remove set but not used variables
YueHaibing [Wed, 26 Sep 2018 11:09:40 +0000 (19:09 +0800)]
scsi: megaraid_mbox: remove set but not used variables

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/scsi/megaraid/megaraid_mbox.c: In function 'megaraid_reset_handler':
drivers/scsi/megaraid/megaraid_mbox.c:2580:7: warning:
 variable 'recovering' set but not used [-Wunused-but-set-variable]

drivers/scsi/megaraid/megaraid_mbox.c: In function 'mbox_post_sync_cmd':
drivers/scsi/megaraid/megaraid_mbox.c:2728:12: warning:
 variable 'mbox64' set but not used [-Wunused-but-set-variable]

drivers/scsi/megaraid/megaraid_mbox.c: In function 'megaraid_mbox_support_random_del':
drivers/scsi/megaraid/megaraid_mbox.c:3138:11: warning:
 variable 'mbox' set but not used [-Wunused-but-set-variable]

drivers/scsi/megaraid/megaraid_mbox.c: In function 'megaraid_mbox_flush_cache':
drivers/scsi/megaraid/megaraid_mbox.c:3266:10: warning:
 variable 'mbox' set but not used [-Wunused-but-set-variable]

drivers/scsi/megaraid/megaraid_mbox.c: In function 'megaraid_mbox_fire_sync_cmd':
drivers/scsi/megaraid/megaraid_mbox.c:3302:12: warning:
 variable 'mbox64' set but not used [-Wunused-but-set-variable]

drivers/scsi/megaraid/megaraid_mbox.c: In function 'gather_hbainfo':
drivers/scsi/megaraid/megaraid_mbox.c:3797:10: warning:
 variable 'dmajor' set but not used [-Wunused-but-set-variable]

[mkp: applied by hand due to conflict with hch's DMA cleanup]

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Simplify conditional check
Nathan Chancellor [Fri, 21 Sep 2018 00:27:23 +0000 (17:27 -0700)]
scsi: qla2xxx: Simplify conditional check

Clang generates a warning when it sees a logical not followed by a
conditional operator like ==, >, or < because it thinks that the logical
not should be applied to the whole statement:

drivers/scsi/qla2xxx/qla_nx.c:3702:7: warning: logical not is only
applied to the left hand side of this comparison
[-Wlogical-not-parentheses]
                if (!qla2x00_eh_wait_for_pending_commands(vha, 0, 0,
                    ^
drivers/scsi/qla2xxx/qla_nx.c:3702:7: note: add parentheses after the
'!' to evaluate the comparison first
                if (!qla2x00_eh_wait_for_pending_commands(vha, 0, 0,
                    ^
                     (
drivers/scsi/qla2xxx/qla_nx.c:3702:7: note: add parentheses around left
hand side expression to silence this warning
                if (!qla2x00_eh_wait_for_pending_commands(vha, 0, 0,
                    ^
                    (
1 warning generated.

It assumes the author might have made a mistake in their logic:

if (!a == b) -> if (!(a == b))

Sometimes that is the case; other times, it's just a super convoluted
way of saying 'if (a)' when b = 0:

if (!1 == 0) -> if (0 == 0) -> if (true)

Alternatively:

if (!1 == 0) -> if (!!1) -> if (1)

Simplify this comparison so that Clang doesn't complain.

Link: https://github.com/ClangBuiltLinux/linux/issues/80
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: bfa: Remove unused functions
Nathan Chancellor [Thu, 20 Sep 2018 23:02:12 +0000 (16:02 -0700)]
scsi: bfa: Remove unused functions

Clang warns when a variable is assigned to itself.

drivers/scsi/bfa/bfa_fcbuild.c:199:6: warning: explicitly assigning
value of variable of type 'int' to itself [-Wself-assign]
        len = len;
        ~~~ ^ ~~~
drivers/scsi/bfa/bfa_fcbuild.c:838:6: warning: explicitly assigning
value of variable of type 'int' to itself [-Wself-assign]
        len = len;
        ~~~ ^ ~~~
drivers/scsi/bfa/bfa_fcbuild.c:917:6: warning: explicitly assigning
value of variable of type 'int' to itself [-Wself-assign]
        len = len;
        ~~~ ^ ~~~
drivers/scsi/bfa/bfa_fcbuild.c:981:6: warning: explicitly assigning
value of variable of type 'int' to itself [-Wself-assign]
        len = len;
        ~~~ ^ ~~~
drivers/scsi/bfa/bfa_fcbuild.c:1008:6: warning: explicitly assigning
value of variable of type 'int' to itself [-Wself-assign]
        len = len;
        ~~~ ^ ~~~
5 warnings generated.

This construct is usually used to avoid unused variable warnings, which
I assume is the case here. -Wunused-parameter is hidden behind -Wextra
with GCC 4.6, which is the minimum version to compile the kernel as of
commit cafa0010cd51 ("Raise the minimum required gcc version to 4.6").

However, upon further inspection, these functions aren't actually used
anywhere; they're just defined. Rather than just removing the self
assignments, remove all of this dead code.

Link: https://github.com/ClangBuiltLinux/linux/issues/148
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: qla2xxx: Remove unnecessary self assignment
Nathan Chancellor [Thu, 20 Sep 2018 22:17:06 +0000 (15:17 -0700)]
scsi: qla2xxx: Remove unnecessary self assignment

Clang warns when a variable is assigned to itself.

drivers/scsi/qla2xxx/qla_mbx.c:1514:4: warning: explicitly assigning
value of variable of type 'uint64_t' (aka 'unsigned long long') to
itself [-Wself-assign]
        l = l;
        ~ ^ ~
1 warning generated.

This construct is usually used to avoid unused variable warnings, which
I assume is the case here. -Wunused-parameter is hidden behind -Wextra
with GCC 4.6, which is the minimum version to compile the kernel as of
commit cafa0010cd51 ("Raise the minimum required gcc version to 4.6").
Just remove this line to silence Clang.

Link: https://github.com/ClangBuiltLinux/linux/issues/83
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: arcmsr: Remove set but not used variables 'id, lun'
YueHaibing [Fri, 12 Oct 2018 10:36:23 +0000 (10:36 +0000)]
scsi: arcmsr: Remove set but not used variables 'id, lun'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/scsi/arcmsr/arcmsr_hba.c: In function 'arcmsr_drain_donequeue':
drivers/scsi/arcmsr/arcmsr_hba.c:1320:10: warning:
 variable 'lun' set but not used [-Wunused-but-set-variable]

drivers/scsi/arcmsr/arcmsr_hba.c:1320:6: warning:
 variable 'id' set but not used [-Wunused-but-set-variable]

Never used since introduction in commit ae52e7f09ff5 ("arcmsr: Support 1024 scatter-gather list entries and improve AP while FW trapped and behaviors of EHs").

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: megaraid_sas: fix a missing-check bug
Wenwen Wang [Sat, 6 Oct 2018 18:34:21 +0000 (13:34 -0500)]
scsi: megaraid_sas: fix a missing-check bug

In megasas_mgmt_compat_ioctl_fw(), to handle the structure
compat_megasas_iocpacket 'cioc', a user-space structure megasas_iocpacket
'ioc' is allocated before megasas_mgmt_ioctl_fw() is invoked to handle
the packet. Since the two data structures have different fields, the data
is copied from 'cioc' to 'ioc' field by field. In the copy process,
'sense_ptr' is prepared if the field 'sense_len' is not null, because it
will be used in megasas_mgmt_ioctl_fw(). To prepare 'sense_ptr', the
user-space data 'ioc->sense_off' and 'cioc->sense_off' are copied and
saved to kernel-space variables 'local_sense_off' and 'user_sense_off'
respectively. Given that 'ioc->sense_off' is also copied from
'cioc->sense_off', 'local_sense_off' and 'user_sense_off' should have the
same value. However, 'cioc' is in the user space and a malicious user can
race to change the value of 'cioc->sense_off' after it is copied to
'ioc->sense_off' but before it is copied to 'user_sense_off'. By doing
so, the attacker can inject different values into 'local_sense_off' and
'user_sense_off'. This can cause undefined behavior in the following
execution, because the two variables are supposed to be same.

This patch enforces a check on the two kernel variables 'local_sense_off'
and 'user_sense_off' to make sure they are the same after the copy. In
case they are not, an error code EINVAL will be returned.

Signed-off-by: Wenwen Wang <wang6495@umn.edu>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: be2iscsi: fix spelling mistake "Retreiving" -> "Retrieving"
Colin Ian King [Tue, 16 Oct 2018 17:46:32 +0000 (18:46 +0100)]
scsi: be2iscsi: fix spelling mistake "Retreiving" -> "Retrieving"

Trivial fix to spelling mistake in beiscsi_log message.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: lpfc: fix spelling mistake "Resrouce" -> "Resource"
Colin Ian King [Tue, 16 Oct 2018 17:28:53 +0000 (18:28 +0100)]
scsi: lpfc: fix spelling mistake "Resrouce" -> "Resource"

Trivial fix to spelling mistake in lpfc_printf_log message text.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: hisi_sas: Fix spin lock management in slot_index_alloc_quirk_v2_hw()
John Garry [Tue, 16 Oct 2018 15:00:36 +0000 (23:00 +0800)]
scsi: hisi_sas: Fix spin lock management in slot_index_alloc_quirk_v2_hw()

Currently a spin_unlock_irqrestore() call is missing on the error path,
so add it.

Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: sg: remove bad blk_end_request_all() call
Jens Axboe [Tue, 16 Oct 2018 14:38:47 +0000 (08:38 -0600)]
scsi: sg: remove bad blk_end_request_all() call

We just need to free the request here. Additionally, this is currently
wrong for a queue that's using MQ currently, it'll crash.

Cc: Doug Gilbert <dgilbert@interlog.com>
Cc: linux-scsi@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Acked-by: Douglas Gilbert <dgilbert@interlog.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: osd: initiator should use mq variant of request ending
Jens Axboe [Tue, 16 Oct 2018 14:33:16 +0000 (08:33 -0600)]
scsi: osd: initiator should use mq variant of request ending

This is currently wrong since it isn't dependent on if we're using mq or
not. At least now it'll be correct when we force mq.

Cc: linux-scsi@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: ips: fix missing break in switch
Gustavo A. R. Silva [Tue, 16 Oct 2018 09:12:23 +0000 (11:12 +0200)]
scsi: ips: fix missing break in switch

Add missing break statement in order to prevent the code from falling
through to case TEST_UNIT_READY.

Addresses-Coverity-ID: 1357338 ("Missing break in switch")
Suggested-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target/core: Always call transport_complete_callback() upon failure
Bart Van Assche [Mon, 15 Oct 2018 15:51:39 +0000 (08:51 -0700)]
scsi: target/core: Always call transport_complete_callback() upon failure

COMPARE AND WRITE command execution starts with a call of
sbc_compare_and_write(). That function locks the caw_sem member in the
backend device data structure and submits a read request to the backend
driver. Upon successful completion of the read compare_and_write_callback()
gets called. That last function compares the data that has been read. If it
matches transport_complete_callback is set to compare_and_write_post and a
write request is submitted. compare_and_write_post() submits a write request
to the backend driver.

XDWRITEREAD command execution starts with sbc_execute_rw() submitting a
read to the backend device. Upon successful completion of the read the
xdreadwrite_callback() gets called. That function xors the data that has
been read with the data in the data-out buffer and stores the result in
the data-in buffer.

Call transport_complete_callback() not only if COMPARE AND WRITE fails but
also if XDWRITEREAD fails. This makes the code more systematic. Make sure
that the callback functions handle (cmd, false, NULL) argument triples fine.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target/core: Use sg_alloc_table() instead of open-coding it
Bart Van Assche [Mon, 15 Oct 2018 15:51:38 +0000 (08:51 -0700)]
scsi: target/core: Use sg_alloc_table() instead of open-coding it

The purpose of sg_alloc_table() is to allocate and initialize an
sg-list. Use that function instead of open-coding it. This patch will
make it easier to share code for caching sg-list allocations between the
SCSI and NVMe target cores.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target/core: Use the SECTOR_SHIFT constant
Bart Van Assche [Mon, 15 Oct 2018 15:51:37 +0000 (08:51 -0700)]
scsi: target/core: Use the SECTOR_SHIFT constant

Instead of duplicating the SECTOR_SHIFT definition from <linux/blkdev.h>,
use it. This patch does not change any functionality.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target/core: Remove the SCF_COMPARE_AND_WRITE_POST flag
Bart Van Assche [Mon, 15 Oct 2018 15:51:36 +0000 (08:51 -0700)]
scsi: target/core: Remove the SCF_COMPARE_AND_WRITE_POST flag

Commit 057085e522f8 ("target: Fix race for SCF_COMPARE_AND_WRITE_POST
checking") removed the code that checks the SCF_COMPARE_AND_WRITE_POST
flag.  Hence also remove the flag itself.

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target/core: Remove an unused data member from struct xcopy_pt_cmd
Bart Van Assche [Mon, 15 Oct 2018 15:51:35 +0000 (08:51 -0700)]
scsi: target/core: Remove an unused data member from struct xcopy_pt_cmd

A value is assigned to the xcopy_op member of struct xcopy_pt_cmd but
that value is never used. Hence remove the xcopy_op member.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target/core: Fix spelling in two source code comments
Bart Van Assche [Mon, 15 Oct 2018 15:51:34 +0000 (08:51 -0700)]
scsi: target/core: Fix spelling in two source code comments

Change one occurrence of "aleady" into "already" and one occurrence of
"is" into "if".

Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: stash sess_err_stats on Data-Out timeout
David Disseldorp [Fri, 12 Oct 2018 10:01:20 +0000 (12:01 +0200)]
scsi: target: stash sess_err_stats on Data-Out timeout

sess_err_stats are currently filled on NOP ping timeout, but not Data-Out
timeout. Stash details of Data-Out timeouts using a
ISCSI_SESS_ERR_CXN_TIMEOUT value for last_sess_failure_type.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: split out helper for cxn timeout error stashing
David Disseldorp [Sat, 13 Oct 2018 23:13:54 +0000 (01:13 +0200)]
scsi: target: split out helper for cxn timeout error stashing

Replace existing nested code blocks with helper function calls.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: log NOP ping timeouts as errors
David Disseldorp [Fri, 12 Oct 2018 10:01:18 +0000 (12:01 +0200)]
scsi: target: log NOP ping timeouts as errors

Events resulting in connection outages like this should be logged as
errors. Include the I_T Nexus in the message to aid path identification.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: log Data-Out timeouts as errors
David Disseldorp [Fri, 12 Oct 2018 10:01:17 +0000 (12:01 +0200)]
scsi: target: log Data-Out timeouts as errors

Data-Out timeouts resulting in connection outages should be logged as
errors. Include the I_T Nexus in the message to aid path identification.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: target: use ISCSI_IQN_LEN in iscsi_target_stat
David Disseldorp [Fri, 12 Oct 2018 10:01:16 +0000 (12:01 +0200)]
scsi: target: use ISCSI_IQN_LEN in iscsi_target_stat

Move the ISCSI_IQN_LEN definition up, so that it can be used in more
places instead of a hardcoded value.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: hisi_sas: Update v3 hw AIP_LIMIT and CFG_AGING_TIME register values
Xiang Chen [Mon, 24 Sep 2018 15:06:34 +0000 (23:06 +0800)]
scsi: hisi_sas: Update v3 hw AIP_LIMIT and CFG_AGING_TIME register values

Update registers as follows:
- Default value of AIP timer is 1ms, and it is easy for some expanders to
  cause IO error. Change the value to max value 65ms to avoid IO error for
  those expanders.

- A CQ completion will be reported by HW when 4 CQs have occurred or the
  aging timer expires, whichever happens first. Sor serial IO scenario, it
  will still wait 8us for every IO before it is reported. So in the
  situation, the performance is poor. So to improve it, change the limit
  time to the least value.
  For other scenario, it does little affect to the performance.

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>
3 years agoscsi: hisi_sas: Use block layer tag instead for IPTT
Xiang Chen [Mon, 24 Sep 2018 15:06:33 +0000 (23:06 +0800)]
scsi: hisi_sas: Use block layer tag instead for IPTT

Currently we use the IPTT defined in LLDD to identify IOs. Actually for
IOs which are from the block layer, they have tags to identify them. So
for those IOs, use tag of the block layer directly, and for IOs which is
not from the block layer (such as internal IOs from libsas/LLDD), reserve
96 IPTTs for them.

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>
3 years agoscsi: hisi_sas: unmask interrupts ent72 and ent74
Xiang Chen [Mon, 24 Sep 2018 15:06:32 +0000 (23:06 +0800)]
scsi: hisi_sas: unmask interrupts ent72 and ent74

The interrupts of ent72 and ent74 are not processed by PCIe AER handling,
so we need to unmask the interrupts and process them first in the driver.

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>
3 years agoscsi: hisi_sas: Free slot later in slot_complete_vx_hw()
Xiang Chen [Mon, 24 Sep 2018 15:06:31 +0000 (23:06 +0800)]
scsi: hisi_sas: Free slot later in slot_complete_vx_hw()

If an SSP/SMP IO times out, it may be actually in reality be
simultaneously processing completion of the slot in
slot_complete_vx_hw().

Then if the slot is freed in slot_complete_vx_hw() (this IPTT is freed
and it may be re-used by other slot), and we may abort the wrong slot in
hisi_sas_abort_task().

So to solve the issue, free the slot after the check of
SAS_TASK_STATE_ABORTED in slot_complete_vx_hw().

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>
3 years agoscsi: hisi_sas: Fix the race between IO completion and timeout for SMP/internal IO
Xiang Chen [Mon, 24 Sep 2018 15:06:30 +0000 (23:06 +0800)]
scsi: hisi_sas: Fix the race between IO completion and timeout for SMP/internal IO

If SMP/internal IO times out, we will possibly free the task immediately.

However if the IO actually completes at the same time, the IO completion
may refer to task which has been freed.

So to solve the issue, flush the tasklet to finish IO completion before
free'ing slot/task.

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>
3 years agoscsi: hisi_sas: Move evaluation of hisi_hba in hisi_sas_task_prep()
Luo Jiaxing [Mon, 24 Sep 2018 15:06:29 +0000 (23:06 +0800)]
scsi: hisi_sas: Move evaluation of hisi_hba in hisi_sas_task_prep()

In evaluating hisi_hba, the sas_port may be NULL, so for safety relocate
the the check to value possible NULL deference.

Signed-off-by: Luo Jiaxing <luojiaxing@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
3 years agoscsi: hisi_sas: Feed back linkrate(max/min) when re-attached
Luo Jiaxing [Mon, 24 Sep 2018 15:06:28 +0000 (23:06 +0800)]
scsi: hisi_sas: Feed back linkrate(max/min) when re-attached

At directly attached situation, if the user modifies the sysfs interface
of maximum_linkrate and minimum_linkrate to renegotiate the linkrate
between SAS controller and target, the value of both files mentioned
above should have change to user setting after renegotiate is over, but
it remains unchanged.

To fix this bug, maximum_linkrate and minimum_linkrate will be directly
fed back to relevant sas_phy structure.

Signed-off-by: Luo Jiaxing <luojiaxing@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>