platform/kernel/linux-starfive.git
5 years agoscsi: smartpqi: add sysfs attributes
Dave Carroll [Fri, 7 Dec 2018 22:28:47 +0000 (16:28 -0600)]
scsi: smartpqi: add sysfs attributes

- add sysfs device attributes, unique_id, lunid and path_info.

Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: smartpqi: refactor sending controller raid requests
Dave Carroll [Fri, 7 Dec 2018 22:28:41 +0000 (16:28 -0600)]
scsi: smartpqi: refactor sending controller raid requests

Clean up the common code which creates a raid path request for the
controller LUNID and sends it synchronously, into a common routine;

Reviewed-by: Murthy Bhat <murthy.bhat@microsemi.com>
Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: smartpqi: turn off lun data caching for ptraid
Dave Carroll [Fri, 7 Dec 2018 22:28:35 +0000 (16:28 -0600)]
scsi: smartpqi: turn off lun data caching for ptraid

- allow update the luns for PTRAID devices.

Reviewed-by: Ajish Koshy <ajish.koshy@microsemi.com>
Reviewed-by: Murthy Bhat <murthy.bhat@microsemi.com>
Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Reviewed-by: Justin Lindley <justin.lindley@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: smartpqi: correct host serial num for ssa
Mahesh Rajashekhara [Fri, 7 Dec 2018 22:28:29 +0000 (16:28 -0600)]
scsi: smartpqi: correct host serial num for ssa

Reviewed-by: Scott Benesh <scott.benesh@microsemi.com>
Reviewed-by: Ajish Koshy <ajish.koshy@microsemi.com>
Reviewed-by: Murthy Bhat <murthy.bhat@microsemi.com>
Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Reviewed-by: Dave Carroll <david.carroll@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: smartpqi: add no_write_same for logical volumes
Dave Carroll [Fri, 7 Dec 2018 22:28:23 +0000 (16:28 -0600)]
scsi: smartpqi: add no_write_same for logical volumes

During slave_alloc, for logical volumes include no_write_same into the
scsi_device structure. This will insure that WRITE_SAME will not be used
for LD's.

Reviewed-by: Ajish Koshy <ajish.koshy@microsemi.com>
Reviewed-by: Murthy Bhat <murthy.bhat@microsemi.com>
Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Reviewed-by: Justin Lindley <justin.lindley@microsemi.com>
Reviewed-by: Scott Benesh <scott.benesh@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: smartpqi: Add retries for device reset
Mahesh Rajashekhara [Fri, 7 Dec 2018 22:28:16 +0000 (16:28 -0600)]
scsi: smartpqi: Add retries for device reset

Reviewed-by: Ajish Koshy <ajish.koshy@microsemi.com>
Reviewed-by: Murthy Bhat <murthy.bhat@microsemi.com>
Reviewed-by: Justin Lindley <justin.lindley@microsemi.com>
Reviewed-by: Scott Benesh <scott.benesh@microsemi.com>
Reviewed-by: Dave Carroll <david.carroll@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: smartpqi: add support for PQI Config Table handshake
Kevin Barnett [Fri, 7 Dec 2018 22:28:10 +0000 (16:28 -0600)]
scsi: smartpqi: add support for PQI Config Table handshake

Add support for new IUs and parsing of the Firmware Features section of the
PQI Config Table to implement the "handshake" between the driver and
firmware to communicate firmware features supported and enabled by the
driver.

Reviewed-by: Ajish Koshy <ajish.koshy@microsemi.com>
Reviewed-by: Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com>
Reviewed-by: Murthy Bhat <murthy.bhat@microsemi.com>
Reviewed-by: Justin Lindley <justin.lindley@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Signed-off-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Update lpfc version to 12.0.0.10
James Smart [Thu, 13 Dec 2018 23:17:58 +0000 (15:17 -0800)]
scsi: lpfc: Update lpfc version to 12.0.0.10

Update lpfc version to 12.0.0.10

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Adding ability to reset chip via pci bus reset
James Smart [Thu, 13 Dec 2018 23:17:57 +0000 (15:17 -0800)]
scsi: lpfc: Adding ability to reset chip via pci bus reset

This patch adds a "pci_bus_reset" option to the board_mode sysfs attribute.
This option uses the pci_reset_bus() api to reset the PCIe link the adapter
is on, which will reset the chip/adapter.  Prior to issuing this option,
all functions on the same chip must be placed in the offline state by the
admin. After the reset, all of the instances may be brought online again.

The primary purpose of this functionality is to support cases where
firmware update required a chip reset but the admin did not want to reboot
the machine in order to instantiate the firmware update.

Sanity checks take place prior to the reset to ensure the adapter is the
sole entity on the PCIe bus and that all functions are in the offline
state.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Add log messages to aid in debugging fc4type discovery issues
James Smart [Thu, 13 Dec 2018 23:17:56 +0000 (15:17 -0800)]
scsi: lpfc: Add log messages to aid in debugging fc4type discovery issues

Current messages report generic actions (like send GID_FT), but misses
reporting for what protocol type the action is taken.

Revise the messages to reflect the FC4 protocol type being worked on.

[mkp: typo]

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix discovery failure when PLOGI is defered
James Smart [Thu, 13 Dec 2018 23:17:55 +0000 (15:17 -0800)]
scsi: lpfc: Fix discovery failure when PLOGI is defered

When a target's link dropped, an RSCN was received to communicate the
change. The driver detected the loss of the target and issued and UNREG_RPI
mailbox command.  While that was being processed, another RSCN was received
to communicate the port coming back.  The driver deferred the PLOGI to the
port until the mailbox command finishes. When the mailbox command completed
it saw the pending port and called the routines to issue the
PLOGI. However, it forgot to clear the UNREG_INP state flag, so the PLOGI
xmt routine nooped the PLOGI request assuming it needed to wait for the
mailbox command.  At this point, login would never be re-attempted.

Clear UNREG_INP before issuing the deferred PLOGI.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: update fault value on successful trunk events.
James Smart [Thu, 13 Dec 2018 23:17:54 +0000 (15:17 -0800)]
scsi: lpfc: update fault value on successful trunk events.

Currently, when a trunk link goes down due to some fault, the driver
snapshots the fault code.  If the link then comes back up, meaning there is
no fault, the driver is not clearing the fault code so the sysfs link_state
entry reports old/stale data.

Revise the logic so that on successful link up the fault code is cleared.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Correct MDS loopback diagnostics support
James Smart [Thu, 13 Dec 2018 23:17:53 +0000 (15:17 -0800)]
scsi: lpfc: Correct MDS loopback diagnostics support

The existing MDS loopback diagnostics support processing received frames in
the slowpath work thread. It caps the number of frames it will process at
64, before waiting for another event to indicate additional frame
reception. The net-net is this results in very slow frame processing during
loopback tests and sometimes orphans an io, causing the loopback test to
report failure by the switch.

Move MDS loopback frame processing out of the slow path worker thread and
into the normal RQ processing routines.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix link state reporting for trunking when adapter is offline
James Smart [Thu, 13 Dec 2018 23:17:52 +0000 (15:17 -0800)]
scsi: lpfc: Fix link state reporting for trunking when adapter is offline

If the adapter is taken offline, the trunk link port attributes continue to
report trunk links as up even though all links are down as the adapter is
offline.

Clear the trunk links state as part of taking the adapter offline.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: bfa: clean up a couple of indentation issues
Colin Ian King [Sat, 15 Dec 2018 14:28:09 +0000 (14:28 +0000)]
scsi: bfa: clean up a couple of indentation issues

There is a break statement with an extra space that needs removed and a
call to bfa_trc that is indented one level too much. Fix these.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: 3w-xxxx: fix indentation issue, add missing tab
Colin Ian King [Sat, 15 Dec 2018 14:17:08 +0000 (14:17 +0000)]
scsi: 3w-xxxx: fix indentation issue, add missing tab

There is a tab missing on a return statement, add the missing tab.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: csiostor: fix incorrect dma device in case of vport
Varun Prakash [Wed, 19 Dec 2018 10:12:50 +0000 (15:42 +0530)]
scsi: csiostor: fix incorrect dma device in case of vport

In case of ->vport_create() call scsi_add_host_with_dma() instead of
scsi_add_host() to pass correct dma device.

Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: myrs: remove the dma_boundary_limit
Christoph Hellwig [Wed, 19 Dec 2018 07:34:55 +0000 (08:34 +0100)]
scsi: myrs: remove the dma_boundary_limit

The old DAC960 driver was fine with merging over segment boundaries, so
this new driver should be too.

[mkp: typos]

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: myrb: remove the dma_boundary limit
Christoph Hellwig [Wed, 19 Dec 2018 07:34:54 +0000 (08:34 +0100)]
scsi: myrb: remove the dma_boundary limit

The old DAC960 driver was fine with merging over segment boundaries, so
this new driver should be too.

[mkp: typos]

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: driver version update
Shivasharan S [Tue, 18 Dec 2018 13:59:55 +0000 (05:59 -0800)]
scsi: megaraid_sas: driver version update

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: Use 63-bit DMA addressing
Shivasharan S [Tue, 18 Dec 2018 13:59:54 +0000 (05:59 -0800)]
scsi: megaraid_sas: Use 63-bit DMA addressing

Although MegaRAID controllers support 64-bit DMA addressing, as per
hardware design, DMA address with all 64-bits set
(0xFFFFFFFF-FFFFFFFF) results in a firmware fault.

Driver will set 63-bit DMA mask to ensure the above address will not be
used.

Cc: stable@vger.kernel.org
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: add retry logic in megasas_readl
Shivasharan S [Mon, 17 Dec 2018 08:47:40 +0000 (00:47 -0800)]
scsi: megaraid_sas: add retry logic in megasas_readl

Due to hardware errata in Aero controllers, reads to certain fusion
registers could intermittently return zero.  This behavior is
transient in nature and subsequent reads will return valid value.

For Aero controllers, any calls to readl to read from certain
registers will be retried for maximum three times, if read returns
zero.

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: changes to function prototypes
Shivasharan S [Mon, 17 Dec 2018 08:47:39 +0000 (00:47 -0800)]
scsi: megaraid_sas: changes to function prototypes

Instead of the register address, pass the instance pointer to clear_intr
and read_fw_status_reg functions.  This is done in preparation for adding
adapter type based checks in these functions in later patches of this
series.

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: Introduce new Aero adapter type
Shivasharan S [Mon, 17 Dec 2018 08:47:38 +0000 (00:47 -0800)]
scsi: megaraid_sas: Introduce new Aero adapter type

Identify all Aero controller PCI IDs with new adapter type.

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: Fix Ventura series based checks
Shivasharan S [Mon, 17 Dec 2018 08:47:37 +0000 (00:47 -0800)]
scsi: megaraid_sas: Fix Ventura series based checks

In preparation for the new Aero series adapter type, all the places where
we check adapter type for Ventura series needs to include any later adapter
types.

Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mpt3sas: fix memory ordering on 64bit writes
Stephan Günther [Sun, 16 Dec 2018 12:08:21 +0000 (13:08 +0100)]
scsi: mpt3sas: fix memory ordering on 64bit writes

With commit 09c2f95ad404 ("scsi: mpt3sas: Swap I/O memory read value back
to cpu endianness"), 64bit writes in _base_writeq() were rewritten to use
__raw_writeq() instad of writeq().

This introduced a bug apparent on powerpc64 systems such as the Raptor
Talos II that causes the HBA to drop from the PCIe bus under heavy load and
being reinitialized after a couple of seconds.

It can easily be triggered on affacted systems by using something like

  fio --name=random-write --iodepth=4 --rw=randwrite --bs=4k --direct=0 \
    --size=128M --numjobs=64 --end_fsync=1
  fio --name=random-write --iodepth=4 --rw=randwrite --bs=64k --direct=0 \
    --size=128M --numjobs=64 --end_fsync=1

a couple of times. In my case I tested it on both a ZFS raidz2 and a btrfs
raid6 using LSI 9300-8i and 9400-8i controllers.

The fix consists in resembling the write ordering of writeq() by adding a
mandatory write memory barrier before device access and a compiler barrier
afterwards. The additional MMIO barrier is superfluous.

Signed-off-by: Stephan Günther <moepi@moepi.net>
Reported-by: Matt Corallo <linux@bluematt.me>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: deadlock by configfs_depend_item
Anatoliy Glagolev [Thu, 6 Dec 2018 23:48:45 +0000 (16:48 -0700)]
scsi: qla2xxx: deadlock by configfs_depend_item

The intent of invoking configfs_depend_item in commit 7474f52a82d51
("tcm_qla2xxx: Perform configfs depend/undepend for base_tpg")
was to prevent a physical Fibre Channel port removal when
virtual (NPIV) ports announced through that physical port are active.
The change does not work as expected: it makes enabled physical port
dependent on target configfs subsystem (the port's parent), something
the configfs guarantees anyway.

Besides, scheduling work in a worker thread and waiting for the work's
completion is not really a valid workaround for the requirement not to call
configfs_depend_item from a configfs callback: the call occasionally
deadlocks.

Thus, removing configfs_depend_item calls does not break anything and fixes
the deadlock problem.

Signed-off-by: Anatoliy Glagolev <glagolig@gmail.com>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: block: remove the cluster flag
Christoph Hellwig [Thu, 13 Dec 2018 15:17:10 +0000 (16:17 +0100)]
scsi: block: remove the cluster flag

Now that the the SCSI layer replaced the use of the cluster flag with
segment size limits and the DMA boundary we can remove the cluster flag
from the block layer.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: hisi_sas: Add support for DIF feature for v2 hw
Xiang Chen [Mon, 17 Dec 2018 14:40:07 +0000 (22:40 +0800)]
scsi: hisi_sas: Add support for DIF feature for v2 hw

For v3 hw, we support DIF operation for SAS, but not SATA.

In addition, DIF CRC16 is supported.

This patchset adds the SW support for the described features. The main
components are as follows:

- Get protection mask from module param
- Fill PI fields
- Fill related to DIF in DQ and protection iu memories

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>
5 years agoscsi: lpfc: do not set queue->page_count to 0 if pc_sli4_params.wqpcnt is invalid
Ewan D. Milne [Thu, 13 Dec 2018 20:25:16 +0000 (15:25 -0500)]
scsi: lpfc: do not set queue->page_count to 0 if pc_sli4_params.wqpcnt is invalid

Certain older adapters such as the OneConnect OCe10100 may not have a valid
wqpcnt value.  In this case, do not set queue->page_count to 0 in
lpfc_sli4_queue_alloc() as this will prevent the driver from initializing.

Fixes: 895427bd01 ("scsi: lpfc: NVME Initiator: Base modifications")
Cc: stable@vger.kernel.org # 4.11+
Signed-off-by: Ewan D. Milne <emilne@redhat.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>
5 years agoscsi: remove the use_clustering flag
Christoph Hellwig [Thu, 13 Dec 2018 15:17:09 +0000 (16:17 +0100)]
scsi: remove the use_clustering flag

The same effects can be achieved by setting the dma_boundary to
PAGE_SIZE - 1 and the max_segment_size to PAGE_SIZE, so shift those
settings into the drivers.  Note that in many cases the setting might
be bogus, but this keeps the status quo.

[mkp: fix myrs and myrb]

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mac53c94: remove DISABLE_CLUSTERING
Christoph Hellwig [Thu, 13 Dec 2018 15:17:08 +0000 (16:17 +0100)]
scsi: mac53c94: remove DISABLE_CLUSTERING

mac53c94 has no limitations on crossing pages for segments.  Just make
the 65535 byte segment size limit explicit, even if it matches the
current block layer limit.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Paul Mackerras <paulus@ozlabs.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mesh: remove DISABLE_CLUSTERING
Christoph Hellwig [Thu, 13 Dec 2018 15:17:07 +0000 (16:17 +0100)]
scsi: mesh: remove DISABLE_CLUSTERING

mesh has no limitations on crossing pages for segments.  Just make
the 65535 byte segment size limit explicit, even if it matches the
current block layer limit.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Paul Mackerras <paulus@ozlabs.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: xen-scsifront: remove DISABLE_CLUSTERING
Christoph Hellwig [Thu, 13 Dec 2018 15:17:06 +0000 (16:17 +0100)]
scsi: xen-scsifront: remove DISABLE_CLUSTERING

There is no such limitation in the protocol or implementation, so
remove it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: powertec: remove the explicit use_clustering setting
Christoph Hellwig [Thu, 13 Dec 2018 15:17:05 +0000 (16:17 +0100)]
scsi: powertec: remove the explicit use_clustering setting

This driver already sets the dma_boundary to PAGE_SIZE - 1, which
has the same result.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: cumana_2: remove the explicit use_clustering setting
Christoph Hellwig [Thu, 13 Dec 2018 15:17:04 +0000 (16:17 +0100)]
scsi: cumana_2: remove the explicit use_clustering setting

This driver already sets the dma_boundary to PAGE_SIZE - 1, which
has the same result.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: introduce a max_segment_size host_template parameters
Christoph Hellwig [Thu, 13 Dec 2018 15:17:03 +0000 (16:17 +0100)]
scsi: introduce a max_segment_size host_template parameters

This allows the host driver to indicate the maximum supported
segment size in a nice an easy way, so that the driver doesn't
have to worry about DMA-layer imposed limitations.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: flip the default on use_clustering
Christoph Hellwig [Thu, 13 Dec 2018 15:17:02 +0000 (16:17 +0100)]
scsi: flip the default on use_clustering

Most SCSI drivers want to enable "clustering", that is merging of
segments so that they might span more than a single page.  Remove the
ENABLE_CLUSTERING define, and require drivers to explicitly set
DISABLE_CLUSTERING to disable this feature.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: make sure all drivers set the use_clustering flag
Christoph Hellwig [Thu, 13 Dec 2018 15:17:01 +0000 (16:17 +0100)]
scsi: make sure all drivers set the use_clustering flag

A few drivers were not setting the use_clustering flag at all and thus
default to disable.  Fix them up to explicitly set this field in
preparation for additional cleanups.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target/core: Use kmem_cache_free() instead of kfree()
Wei Yongjun [Mon, 17 Dec 2018 12:19:53 +0000 (12:19 +0000)]
scsi: target/core: Use kmem_cache_free() instead of kfree()

memory allocated by kmem_cache_alloc() should be freed using
kmem_cache_free(), not kfree().

Fixes: ad669505c4e9 ("scsi: target/core: Make sure that target_wait_for_sess_cmds() waits long enough")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: virtio_scsi: Remove per-target data because it is no longer used
Bart Van Assche [Fri, 14 Dec 2018 17:35:11 +0000 (09:35 -0800)]
scsi: virtio_scsi: Remove per-target data because it is no longer used

Commit b5b6e8c8d3b4 ("scsi: virtio_scsi: fix IO hang caused by automatic
irq vector affinity") removed all virtio_scsi hostdata users. Since the
SCSI host data is no longer used, also remove the host data itself.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: ufs: Fix platform_no_drv_owner.cocci warnings
YueHaibing [Tue, 13 Nov 2018 01:26:17 +0000 (01:26 +0000)]
scsi: ufs: Fix platform_no_drv_owner.cocci warnings

Remove .owner field if calls are used which set it automatically
Generated by: scripts/coccinelle/api/platform_no_drv_owner.cocci

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mpt3sas: Update driver version to 27.101.00.00
Suganath Prabu [Fri, 7 Dec 2018 07:28:35 +0000 (12:58 +0530)]
scsi: mpt3sas: Update driver version to 27.101.00.00

Update driver version from 27.100.00.00 to 27.101.00.00.

Signed-off-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mpt3sas: Replace readl with ioc->base_readl
Suganath Prabu [Fri, 7 Dec 2018 07:28:34 +0000 (12:58 +0530)]
scsi: mpt3sas: Replace readl with ioc->base_readl

Use ioc->base_readl to restrict the readl retries to only Aero controllers.

Signed-off-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mpt3sas: Add separate function for aero doorbell reads
Suganath Prabu [Fri, 7 Dec 2018 07:28:33 +0000 (12:58 +0530)]
scsi: mpt3sas: Add separate function for aero doorbell reads

Sometimes Aero controllers appears to be returning bad data (0) for
doorbell register read and if retries are performed immediately after the
bad read, they return good data.

Workaround is added to retry read from doorbell registers for maximum three
times if driver get the zero.  Added functions base_readl_aero for Aero IOC
and base_readl for gen35 and other controllers.

Signed-off-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: mpt3sas: Introduce flag for aero based controllers
Suganath Prabu [Fri, 7 Dec 2018 07:28:32 +0000 (12:58 +0530)]
scsi: mpt3sas: Introduce flag for aero based controllers

Adding flag "is_aero_ioc" to differentiate aero based controllers from
other gen35 controllers.

Signed-off-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: hisi_sas: Make sg_tablesize consistent value
Xiang Chen [Thu, 6 Dec 2018 13:34:42 +0000 (21:34 +0800)]
scsi: hisi_sas: Make sg_tablesize consistent value

Sht->sg_tablesize is set in the driver, and it will be assigned to
shost->sg_tablesize in SCSI mid-layer. So it is not necessary to assign
shost->sg_table one more time in the driver.

In addition to the change, change each scsi_host_template.sg_tablesize
to HISI_SAS_SGE_PAGE_CNT instead of SG_ALL.

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>
5 years agoscsi: hisi_sas: Relocate some code to reduce complexity
Xiang Chen [Thu, 6 Dec 2018 13:34:41 +0000 (21:34 +0800)]
scsi: hisi_sas: Relocate some code to reduce complexity

Relocate the codes related to dma_map/unmap in hisi_sas_task_prep() to
reduce complexity, with a view to add DIF/DIX support.

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>
5 years agoscsi: hisi_sas: Fix warnings detected by sparse
John Garry [Thu, 6 Dec 2018 13:34:40 +0000 (21:34 +0800)]
scsi: hisi_sas: Fix warnings detected by sparse

This patchset fixes some warnings detected by the sparse tool, like these:
drivers/scsi/hisi_sas/hisi_sas_main.c:1469:52: warning: incorrect type in assignment (different base types)
drivers/scsi/hisi_sas/hisi_sas_main.c:1469:52:    expected unsigned short [unsigned] [assigned] [usertype] tag_of_task_to_be_managed
drivers/scsi/hisi_sas/hisi_sas_main.c:1469:52:    got restricted __le16 [usertype] <noident>
drivers/scsi/hisi_sas/hisi_sas_main.c:1723:52: warning: incorrect type in assignment (different base types)
drivers/scsi/hisi_sas/hisi_sas_main.c:1723:52:    expected unsigned short [unsigned] [assigned] [usertype] tag_of_task_to_be_managed
drivers/scsi/hisi_sas/hisi_sas_main.c:1723:52:    got restricted __le16 [usertype] <noident>

Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: Update driver version to 10.00.00.12-k
Himanshu Madhani [Mon, 10 Dec 2018 20:36:25 +0000 (12:36 -0800)]
scsi: qla2xxx: Update driver version to 10.00.00.12-k

Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: Fix for FC-NVMe discovery for NPIV port
Giridhar Malavali [Mon, 10 Dec 2018 20:36:24 +0000 (12:36 -0800)]
scsi: qla2xxx: Fix for FC-NVMe discovery for NPIV port

This patch fixes NVMe discovery by setting SKIP_PRLI flag, so that PRLI is
driven by driver and is retried when the NPIV port is detected to have NVMe
capability.

Signed-off-by: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: Fix NPIV handling for FC-NVMe
Himanshu Madhani [Mon, 10 Dec 2018 20:36:23 +0000 (12:36 -0800)]
scsi: qla2xxx: Fix NPIV handling for FC-NVMe

This patch fixes issues with NPIV port with FC-NVMe. Clean up code for
remoteport delete and also call nvme_delete when deleting VPs.

Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: Enable FC-NVME on NPIV ports
Anil Gurumurthy [Mon, 10 Dec 2018 20:36:22 +0000 (12:36 -0800)]
scsi: qla2xxx: Enable FC-NVME on NPIV ports

Signed-off-by: Anil Gurumurthy <agurumurthy@marvell.com>
Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: aacraid: change event_wait to a completion
Arnd Bergmann [Mon, 10 Dec 2018 21:32:41 +0000 (22:32 +0100)]
scsi: aacraid: change event_wait to a completion

The event_wait semaphore has completion semantics, so we can
change it over to the completion interface for clarity without
changing the behavior.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: aacraid: change wait_sem to a completion
Arnd Bergmann [Mon, 10 Dec 2018 21:32:40 +0000 (22:32 +0100)]
scsi: aacraid: change wait_sem to a completion

The wait_sem member is used like a completion, so we should
use the respective API. The behavior is unchanged.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Dave Carroll <david.carroll@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Enable Management features for IF_TYPE=6
James Smart [Tue, 11 Dec 2018 03:37:01 +0000 (19:37 -0800)]
scsi: lpfc: Enable Management features for IF_TYPE=6

Addition of support for if_type=6 missed several checks for interface type,
resulting in the failure of several key management features such as
firmware dump and loopback testing.

Correct the checks on the if_type so that both SLI4 IF_TYPE's 2 and 6 are
supported.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoRevert "scsi: lpfc: ls_rjt erroneus FLOGIs"
Martin K. Petersen [Thu, 13 Dec 2018 01:26:56 +0000 (20:26 -0500)]
Revert "scsi: lpfc: ls_rjt erroneus FLOGIs"

This reverts commit 287aba2592870fa5b76134b28173b77f1f7a4492.

We killed the bad firmware and this mod is no longer necessary.

Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: update driver version to 12.0.0.9
James Smart [Fri, 30 Nov 2018 00:09:41 +0000 (16:09 -0800)]
scsi: lpfc: update driver version to 12.0.0.9

Update the driver version to 12.0.0.9

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix dif and first burst use in write commands
James Smart [Fri, 30 Nov 2018 00:09:40 +0000 (16:09 -0800)]
scsi: lpfc: Fix dif and first burst use in write commands

When dif and first burst is used in a write command wqe, the driver was not
properly setting fields in the io command request. This resulted in no dif
bytes being sent and invalid xfer_rdy's, resulting in the io being aborted
by the hardware.

Correct the wqe initializaton when both dif and first burst are used.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix driver release of fw-logging buffers
James Smart [Fri, 30 Nov 2018 00:09:39 +0000 (16:09 -0800)]
scsi: lpfc: Fix driver release of fw-logging buffers

On driver termination, after the driver stops fw logging by writing a
register on the chip, the driver immediately unmaps and frees the logging
buffer, without confirming in any way that the chip has received the write
and terminated the logging. As termination on the chip is not immediate,
the chip may issue a dma request to the now unmapped dma buffer, resulting
in a iommu fault.

Change the driver to receive a confirmation that logging ahs been
terminated. As the driver always issues an SLI reset with the device as
part of shutdown, and as part of that is receiving confirmation that the
reset is complete - the driver was modified to perform the write to disable
fw logging prior to the SLI reset and only free the fw log buffer after the
SLI reset is complete. That guarantees use of the fw log buffer is fully
terminated when it is unmapped.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Correct topology type reporting on G7 adapters
James Smart [Fri, 30 Nov 2018 00:09:38 +0000 (16:09 -0800)]
scsi: lpfc: Correct topology type reporting on G7 adapters

Driver missed classifying the chip type for G7 when reporting supported
topologies. This resulted in loop being shown as supported on FC links that
are not supported per the standard.

Add the chip classifications to the topology checks in the driver.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Correct code setting non existent bits in sli4 ABORT WQE
James Smart [Fri, 30 Nov 2018 00:09:37 +0000 (16:09 -0800)]
scsi: lpfc: Correct code setting non existent bits in sli4 ABORT WQE

Driver is setting bits in word 10 of the SLI4 ABORT WQE (the wqid).  The
field was a carry over from a prior SLI revision. The field does not exist
in SLI4, and the action may result in an overlap with future definition of
the WQE.

Remove the setting of WQID in the ABORT WQE.

Also cleaned up WQE field settings - initialize to zero, don't bother to
set fields to zero.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Defer LS_ACC to FLOGI on point to point logins
James Smart [Fri, 30 Nov 2018 00:09:36 +0000 (16:09 -0800)]
scsi: lpfc: Defer LS_ACC to FLOGI on point to point logins

The current discovery state machine the driver treated FLOGI oddly.  When
point to point, an FLOGI is to be exchanged by the two ports, with the port
with the most significant WWN then proceeding with PLOGI.  The
implementation in the driver was keyed to closely with "what have I sent",
not with what has happened between the two endpoints. Thus, it blatantly
would ACC an FLOGI, but reject PLOGI's until it had its FLOGI ACC'd. The
problem is - the sending of FLOGI may be delayed for some reason, or the
response to FLOGI held off by the other side. In the failing situation the
other side sent an FLOGI, which was ACC'd, then sent PLOGIs which were then
rjt'd until the retry count for the PLOGIs were exceeded and the port gave
up. The FLOGI may have been very late in transmit, or the response held off
until the PLOGIs failed. Given the other port had the higher WWN, no PLOGIs
would occur and communication stopped.

Correct the situation by changing the FLOGI handling. Defer any response to
an FLOGI until the driver has sent its FLOGI as well. Then, upon either
completion of the sent FLOGI, or upon sending an ACC to a received FLOGI
(which may be received before or just after FLOGI was sent). the driver
will act on who has the higher WWN. if the other port does, the driver will
noop any handling of an FLOGI response (if outstanding) and wait for PLOGI.
If the local port does, the driver will transition to sending PLOGI and
will noop any action on responding to an FLOGI (if not yet received).

Fortunately, to implement this, it only took another state flag and
deferring any FLOGI response if the FLOGI has yet to be transmit. All
subsequent actions were already in place.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: ls_rjt erroneus FLOGIs
James Smart [Fri, 30 Nov 2018 00:09:35 +0000 (16:09 -0800)]
scsi: lpfc: ls_rjt erroneus FLOGIs

In some link initialization sequences, the fw generates an erroneous FLOGI
payload to the driver without an intervening link bounce.  The driver, when
it sees a 2nd FLOGI without an intervening link bounce, automatically
performs a link bounce. In this, the link bounce causes the situate to
repeat and in a nasty loop of link bounces.

Resolve the issue by validating the FLOGI payload. The erroneous FLOGI will
contain VVL signatures that are not normal. When the driver sees these, it
will simply reject the flogi rather than bouncing the link.  The reject is
consumed within the firmware.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: rport port swap discovery issue.
James Smart [Fri, 30 Nov 2018 00:09:34 +0000 (16:09 -0800)]
scsi: lpfc: rport port swap discovery issue.

Two initiator ports were cable swapped and after swap both went down.  The
driver internally swaps the nlp nodes based on matching node wwn's but not
the same nport id as before. After detecting a change in the nodes RPI, the
driver sends an UNREG_RPI command and clears the NLP_RPI_REGISTERED flag,
then swaps the node information with the other node. But the other node's
NLP_RPI_REGISTERED flag is also cleared, but it is done so without an
UNREG_RPI being sent, which causes the later REG_RPI for that other node to
fail as the hardware believes its still registered.

Additionally, if the node swap occurred while the two nodes had PLOGI's in
flight, the fc4_types weren't properly getting swapped such that when the
PLOGIs commpleted and PRLI's were then sent, the PRLI's acted on bad
protocol types so the PRLI was for the wrong protocol. NVME devices saw
SCSI FCP PRLIs and vice versa.

Clean up the node swap so that the NLP_RPI_REGISTERED flag is handled
properly.

Fix the handling of the fc4_types when the nodes are swapped as well

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Cap NPIV vports to 256
James Smart [Fri, 30 Nov 2018 00:09:33 +0000 (16:09 -0800)]
scsi: lpfc: Cap NPIV vports to 256

Depending on the chipset, the number of NPIV vports may vary and be in
excess of what most switches support (256). To avoid confusion with the
users, limit the reported NPIV vports to 256.

Additionally correct the 16G adapter which is reporting a bogus NPIV vport
number if the link is down.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix kernel Oops due to null pring pointers
James Smart [Fri, 30 Nov 2018 00:09:32 +0000 (16:09 -0800)]
scsi: lpfc: Fix kernel Oops due to null pring pointers

Driver is hitting null pring pointers in lpfc_do_work().

Pointer assignment occurs based on SLI-revision. If recovering after an
error, its possible the sli revision for the port was cleared, making the
lpfc_phba_elsring() not return a ring pointer, thus the null pointer.

Add SLI revision checking to lpfc_phba_elsring() and status checking to all
callers.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix a duplicate 0711 log message number.
James Smart [Fri, 30 Nov 2018 00:09:31 +0000 (16:09 -0800)]
scsi: lpfc: Fix a duplicate 0711 log message number.

Renumber one of the 0711 log messages so there isn't a duplication.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix discovery failures during port failovers with lots of vports
James Smart [Fri, 30 Nov 2018 00:09:30 +0000 (16:09 -0800)]
scsi: lpfc: Fix discovery failures during port failovers with lots of vports

The driver is getting hit with 100s of RSCNs during remote port address
changes. Each of those RSCN's ends up generating UNREG_RPI and REG_PRI
mailbox commands.  The discovery engine within the driver doesn't wait for
the mailbox command completions. Instead it sets state flags and moves
forward. At some point, there's a massive backlog of mailbox commands which
take time for the adapter to process. Additionally, it appears there were
duplicate events from the switch so the driver generated duplicate mailbox
commands for the same remote port.  During this window, failures on PLOGI
and PRLI ELS's are see as the adapter is rejecting them as they are for
remote ports that still have pending mailbox commands.

Streamline the discovery engine so that PLOGI log checks for outstanding
UNREG_RPIs and defer the processing until the commands complete. This
better synchronizes the ELS transmission vs the RPI registrations.

Filter out multiple UNREG_RPIs being queued up for the same remote port.

Beef up log messages in this area.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: refactor mailbox structure context fields
James Smart [Fri, 30 Nov 2018 00:09:29 +0000 (16:09 -0800)]
scsi: lpfc: refactor mailbox structure context fields

The driver data structure for managing a mailbox command contained two
context fields. Unfortunately, the context were considered "generic" to be
used at the whim of the command code.  Of course, one section of code used
fields this way, while another did it that way, and eventually there were
mixups.

Refactored the structure so that the generic contexts become a node context
and a buffer context and all code standardizes on their use.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: update manufacturer attribute to reflect Broadcom
James Smart [Fri, 30 Nov 2018 00:09:28 +0000 (16:09 -0800)]
scsi: lpfc: update manufacturer attribute to reflect Broadcom

Update manufacturer attribute to reflect Broadcom Inc, not Emulex

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: lpfc: Fix panic when FW-log buffsize is not initialized
James Smart [Fri, 30 Nov 2018 00:09:27 +0000 (16:09 -0800)]
scsi: lpfc: Fix panic when FW-log buffsize is not initialized

While trying to get adapter fw-log for a function whose buffsize was set to
0, kernel panic occurred.

When buffsize is 0, the kernel buffer for the log won't be allocated.  When
fw log usage was enabled, it failed to check the buffer size, and log usage
was started. Eventually the driver referenced the unallocated log buffer.

Added checks of the buffer size before allowing fw logging to be enabled
and added check for valid buffer if enabling fw log.

Performed a couple other minor cleanups while fixing this:
 - clarified log messages
 - re-evaluated log message severity
 - treat any error as an error, not only a couple codes

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: aha1542: Fix zeroday __udivdi3 warning
James Bottomley [Wed, 28 Nov 2018 05:41:24 +0000 (21:41 -0800)]
scsi: aha1542: Fix zeroday __udivdi3 warning

dma_addr_t can be u64 on pae systems but isa_virt_to_bus only ever
returns unsigned long (because an ISA physical address can only be 24
bits). Cast to unsigned long to avoid division.

Fixes: 1794ef2b150d ("scsi: aha1542: convert to DMA mapping API")
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: zfcp: improve kdoc for return of zfcp_status_read_refill()
Steffen Maier [Thu, 6 Dec 2018 16:31:21 +0000 (17:31 +0100)]
scsi: zfcp: improve kdoc for return of zfcp_status_read_refill()

Complements

v2.6.35 commit 64deb6efdc55 ("[SCSI] zfcp: Use status_read_buf_num
provided by FCP channel") which replaced the hardcoded 16 with a
variable value

Also complements already existing fixups for above commit

v2.6.35 commit 8d88cf3f3b9a ("[SCSI] zfcp: Update status read mempool")
v3.10   commit 9edf7d75ee5f ("[SCSI] zfcp: status read buffers on first adapter open with link down")

Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: zfcp: fix posting too many status read buffers leading to adapter shutdown
Steffen Maier [Thu, 6 Dec 2018 16:31:20 +0000 (17:31 +0100)]
scsi: zfcp: fix posting too many status read buffers leading to adapter shutdown

Suppose adapter (open) recovery is between opened QDIO queues and before
(the end of) initial posting of status read buffers (SRBs). This time
window can be seconds long due to FSF_PROT_HOST_CONNECTION_INITIALIZING
causing by design looping with exponential increase sleeps in the function
performing exchange config data during recovery
[zfcp_erp_adapter_strat_fsf_xconf()]. Recovery triggered by local link up.

Suppose an event occurs for which the FCP channel would send an unsolicited
notification to zfcp by means of a previously posted SRB.  We saw it with
local cable pull (link down) in multi-initiator zoning with multiple
NPIV-enabled subchannels of the same shared FCP channel.

As soon as zfcp_erp_adapter_strategy_open_fsf() starts posting the initial
status read buffers from within the adapter's ERP thread, the channel does
send an unsolicited notification.

Since v2.6.27 commit d26ab06ede83 ("[SCSI] zfcp: receiving an unsolicted
status can lead to I/O stall"), zfcp_fsf_status_read_handler() schedules
adapter->stat_work to re-fill the just consumed SRB from a work item.

Now the ERP thread and the work item post SRBs in parallel.  Both contexts
call the helper function zfcp_status_read_refill().  The tracking of
missing (to be posted / re-filled) SRBs is not thread-safe due to separate
atomic_read() and atomic_dec(), in order to depend on posting
success. Hence, both contexts can see
atomic_read(&adapter->stat_miss) == 1. One of the two contexts posts
one too many SRB. Zfcp gets QDIO_ERROR_SLSB_STATE on the output queue
(trace tag "qdireq1") leading to zfcp_erp_adapter_shutdown() in
zfcp_qdio_handler_error().

An obvious and seemingly clean fix would be to schedule stat_work from the
ERP thread and wait for it to finish. This would serialize all SRB
re-fills. However, we already have another work item wait on the ERP
thread: adapter->scan_work runs zfcp_fc_scan_ports() which calls
zfcp_fc_eval_gpn_ft(). The latter calls zfcp_erp_wait() to wait for all the
open port recoveries during zfcp auto port scan, but in fact it waits for
any pending recovery including an adapter recovery. This approach leads to
a deadlock.  [see also v3.19 commit 18f87a67e6d6 ("zfcp: auto port scan
resiliency"); v2.6.37 commit d3e1088d6873
("[SCSI] zfcp: No ERP escalation on gpn_ft eval");
v2.6.28 commit fca55b6fb587
("[SCSI] zfcp: fix deadlock between wq triggered port scan and ERP")
fixing v2.6.27 commit c57a39a45a76
("[SCSI] zfcp: wait until adapter is finished with ERP during auto-port");
v2.6.27 commit cc8c282963bd
("[SCSI] zfcp: Automatically attach remote ports")]

Instead make the accounting of missing SRBs atomic for parallel execution
in both the ERP thread and adapter->stat_work.

Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Fixes: d26ab06ede83 ("[SCSI] zfcp: receiving an unsolicted status can lead to I/O stall")
Cc: <stable@vger.kernel.org> #2.6.27+
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: sun_esp: Use of_node_name_eq for node name comparisons
Rob Herring [Wed, 5 Dec 2018 19:50:39 +0000 (13:50 -0600)]
scsi: sun_esp: Use of_node_name_eq for node name comparisons

Convert string compares of DT node names to use of_node_name_eq helper
instead. This removes direct access to the node name pointer.

Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target: perform t10_wwn ID initialisation in target_alloc_device()
David Disseldorp [Wed, 5 Dec 2018 12:18:38 +0000 (13:18 +0100)]
scsi: target: perform t10_wwn ID initialisation in target_alloc_device()

Initialise the t10_wwn vendor, model and revision defaults when a device is
allocated instead of when it's enabled. This ensures that custom vendor or
model strings set prior to enablement are not later overwritten with
default values.

The TRANSPORT_FLAG_PASSTHROUGH conditional can be dropped for the following
reasons:

 - target_core_pscsi overwrites the defaults in the
   pscsi_configure_device() callback.
   + the contents is then only used for ConfigFS via
     $pscsi_dev/statistics/scsi_lu/vend, etc.

 - target_core_user doesn't touch the defaults, nor are they used for
   anything outside of ConfigFS.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target: remove hardcoded T10 Vendor ID in INQUIRY response
David Disseldorp [Wed, 5 Dec 2018 12:18:37 +0000 (13:18 +0100)]
scsi: target: remove hardcoded T10 Vendor ID in INQUIRY response

Use the value stored in t10_wwn.vendor, which defaults to "LIO-ORG", but
can be reconfigured via the vendor_id ConfigFS attribute.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Bryant G. Ly <bly@catalogicsoftware.com>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target: add device vendor_id configfs attribute
David Disseldorp [Wed, 5 Dec 2018 12:18:36 +0000 (13:18 +0100)]
scsi: target: add device vendor_id configfs attribute

The vendor_id attribute will allow for the modification of the T10 Vendor
Identification string returned in inquiry responses. Its value can be
viewed and modified via the ConfigFS path at:

target/core/$backstore/$name/wwn/vendor_id

"LIO-ORG" remains the default value, which is set when the backstore device
is enabled.

[mkp: corrected VPD page number]

Signed-off-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target: consistently null-terminate t10_wwn strings
David Disseldorp [Wed, 5 Dec 2018 12:18:35 +0000 (13:18 +0100)]
scsi: target: consistently null-terminate t10_wwn strings

In preparation for supporting user provided vendor strings, add an extra
byte to the vendor, model and revision arrays in struct t10_wwn. This
ensures that the full INQUIRY data can be carried in the arrays along with
a null-terminator.

Change a number of array readers and writers so that they account for
explicit null-termination:

- The pscsi_set_inquiry_info() and emulate_model_alias_store() codepaths
  don't currently explicitly null-terminate; fix this.

- Existing t10_wwn field dumps use for-loops which step over
  null-terminators for right-padding.
  + Use printf with width specifiers instead.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target: use consistent left-aligned ASCII INQUIRY data
David Disseldorp [Wed, 5 Dec 2018 12:18:34 +0000 (13:18 +0100)]
scsi: target: use consistent left-aligned ASCII INQUIRY data

spc5r17.pdf specifies:

  4.3.1 ASCII data field requirements
  ASCII data fields shall contain only ASCII printable characters (i.e.,
  code values 20h to 7Eh) and may be terminated with one or more ASCII null
  (00h) characters.  ASCII data fields described as being left-aligned
  shall have any unused bytes at the end of the field (i.e., highest
  offset) and the unused bytes shall be filled with ASCII space characters
  (20h).

LIO currently space-pads the T10 VENDOR IDENTIFICATION and PRODUCT
IDENTIFICATION fields in the standard INQUIRY data. However, the PRODUCT
REVISION LEVEL field in the standard INQUIRY data as well as the T10 VENDOR
IDENTIFICATION field in the INQUIRY Device Identification VPD Page are
zero-terminated/zero-padded.

Fix this inconsistency by using space-padding for all of the above fields.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bryant G. Ly <bly@catalogicsoftware.com>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: NULL check before some freeing functions is not needed
Thomas Meyer [Sun, 2 Dec 2018 20:52:11 +0000 (21:52 +0100)]
scsi: qla2xxx: NULL check before some freeing functions is not needed

NULL check before some freeing functions is not needed.

Signed-off-by: Thomas Meyer <thomas@m3y3r.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qedf: NULL check before some freeing functions is not needed
Thomas Meyer [Sun, 2 Dec 2018 20:52:11 +0000 (21:52 +0100)]
scsi: qedf: NULL check before some freeing functions is not needed

NULL check before some freeing functions is not needed.

Signed-off-by: Thomas Meyer <thomas@m3y3r.de>
Acked-by: Manish Rangankar <Manish.Rangankar@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: megaraid_sas: NULL check before some freeing functions is not needed
Thomas Meyer [Sun, 2 Dec 2018 20:52:11 +0000 (21:52 +0100)]
scsi: megaraid_sas: NULL check before some freeing functions is not needed

NULL check before some freeing functions is not needed.

Signed-off-by: Thomas Meyer <thomas@m3y3r.de>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: qla2xxx: Split the __qla2x00_abort_all_cmds() function
Bart Van Assche [Thu, 29 Nov 2018 18:25:11 +0000 (10:25 -0800)]
scsi: qla2xxx: Split the __qla2x00_abort_all_cmds() function

Nesting in __qla2x00_abort_all_cmds() is way too deep. Reduce the nesting
level by introducing a helper function. This patch does not change any
functionality.

Reviewed-by: Laurence Oberman <loberman@redhat.com>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: csiostor: remove flush_scheduled_work()
Varun Prakash [Thu, 29 Nov 2018 13:48:31 +0000 (19:18 +0530)]
scsi: csiostor: remove flush_scheduled_work()

flush_scheduled_work() is not required as csio_hw_exit_workers() calls
cancel_work_sync() for hw->evtq_work.

Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: zfcp: make DIX experimental, disabled, and independent of DIF
Fedor Loshakov [Thu, 29 Nov 2018 12:09:56 +0000 (13:09 +0100)]
scsi: zfcp: make DIX experimental, disabled, and independent of DIF

Introduce separate zfcp module parameters to individually select support
for: DIF which should work (zfcp.dif, which used to be DIF+DIX, disabled)
or DIX+DIF which can cause trouble (zfcp.dix, new, disabled).

If DIX is enabled, we warn on zfcp driver initialization.  As before, this
also reduces the maximum I/O request size to half, to support the worst
case of merged single sector requests with one protection data scatter
gather element per sector. This can impact the maximum throughput.

In DIF-only mode (zfcp.dif=1 zfcp.dix=0), we can use the full maximum I/O
request size as there is no protection data for zfcp.

Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Fedor Loshakov <loshakov@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: BusLogic: mark expected switch fall-through
Gustavo A. R. Silva [Wed, 28 Nov 2018 04:21:23 +0000 (22:21 -0600)]
scsi: BusLogic: mark expected switch fall-through

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

Addresses-Coverity-ID: 1056537 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Khalid Aziz <khalid@gonehiking.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: ufs: Remove redundant sense size definition
Avri Altman [Thu, 22 Nov 2018 18:04:56 +0000 (20:04 +0200)]
scsi: ufs: Remove redundant sense size definition

By spec, the ufs sense data is 18 bytes long.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: snic: Use vzalloc
Sabyasachi Gupta [Thu, 15 Nov 2018 16:34:04 +0000 (22:04 +0530)]
scsi: snic: Use vzalloc

Replaced vmalloc + memset with vzalloc

Signed-off-by: Sabyasachi Gupta <sabyasachi.linux@gmail.com>
Acked-by: Sesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: fnic: Use vzalloc
Sabyasachi Gupta [Wed, 14 Nov 2018 16:54:21 +0000 (22:24 +0530)]
scsi: fnic: Use vzalloc

Replaced vmalloc + memset with vzalloc

Signed-off-by: Sabyasachi Gupta <sabyasachi.linux@gmail.com>
Acked-by: Sesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
5 years agoscsi: target/core: Reduce the amount of code executed with a spinlock held
Bart Van Assche [Tue, 27 Nov 2018 23:52:04 +0000 (15:52 -0800)]
scsi: target/core: Reduce the amount of code executed with a spinlock held

Due to the "make ABORT and LUN RESET handling synchronous" patch, cmd->work
is only modified from the regular command execution path and no longer
asynchronously by the code that executes task management functions. Since
the regular command execution code is sequential per command, no locking is
required to manipulate cmd->work. Hence stop protecting cmd->work
manipulations with locking.

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Make ABORT and LUN RESET handling synchronous
Bart Van Assche [Tue, 27 Nov 2018 23:52:03 +0000 (15:52 -0800)]
scsi: target/core: Make ABORT and LUN RESET handling synchronous

Instead of invoking target driver callback functions from the context that
handles an abort or LUN RESET task management function, only set the abort
flag from that context and perform the actual abort handling from the
context of the regular command processing flow. This approach has the
advantage that the task management code becomes much easier to read and to
verify since the number of potential race conditions against the command
processing flow is strongly reduced.

This patch has been tested by running the following two shell commands
concurrently for about ten minutes for both the iSCSI and the SRP target
drivers ($dev is an initiator device node connected with storage provided
by the target driver under test):

 * fio with data verification enabled on a filesystem mounted on top of
   $dev.

 * while true; do sg_reset -d $dev; echo -n .; sleep .1; done

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Fix TAS handling for aborted commands
Bart Van Assche [Tue, 27 Nov 2018 23:52:02 +0000 (15:52 -0800)]
scsi: target/core: Fix TAS handling for aborted commands

The TASK ABORTED STATUS (TAS) bit is defined as follows in SAM:
"TASK_ABORTED: this status shall be returned if a command is aborted by a
command or task management function on another I_T nexus and the control
mode page TAS bit is set to one". TAS handling is spread over the target
core and the iSCSI target driver. If a LUN RESET is received, the target
core will send the TASK_ABORTED response for all commands for which such a
response has to be sent. If an ABORT TASK is received, only the iSCSI
target driver will send the TASK_ABORTED response for the commands for
which that response has to be sent.  That is a bug since all target drivers
have to honor the TAS bit. Fix this by moving the code that handles TAS
from the iSCSI target driver into the target core. Additionally, if a
command has been aborted, instead of sending the TASK_ABORTED status from
the context that processes the SCSI command send it from the context of the
ABORT TMF.  The core_tmr_abort_task() change in this patch causes the
CMD_T_TAS flag to be set if a TASK_ABORTED status has to be sent back to
the initiator that submitted the command. If that flag has been set
transport_cmd_finish_abort() will send the TASK_ABORTED response.

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Simplify the code for aborting SCSI commands
Bart Van Assche [Tue, 27 Nov 2018 23:52:01 +0000 (15:52 -0800)]
scsi: target/core: Simplify the code for aborting SCSI commands

Instead of allowing the code that aborts a SCSI command to finish before
all iSCSI data frames have been received, make that code wait until all
iSCSI data frames have been received. Introduce a new member variable in
the target driver template to communicate that information from the iSCSI
target driver to the target core. This change allows to leave out the check
whether or not it is already safe to send the TASK_ABORTED reply from
transport_send_task_abort().

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Make it possible to wait from more than one context for command...
Bart Van Assche [Tue, 27 Nov 2018 23:52:00 +0000 (15:52 -0800)]
scsi: target/core: Make it possible to wait from more than one context for command completion

This patch does not change any functionality but makes the patch that makes
TMF handling synchronous easier to read.

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Use system workqueues for TMF
Bart Van Assche [Tue, 27 Nov 2018 23:51:59 +0000 (15:51 -0800)]
scsi: target/core: Use system workqueues for TMF

A quote from SAM-5: "The order in which task management requests are
processed is not specified by the SCSI architecture model.  The SCSI
architecture model does not require in-order delivery of such task
management requests or processing by the task manager in the order
received. To guarantee the processing order of task management requests
referencing sent to a specific logical unit, an application client should
not have more than one such task management request pending to that logical
unit." This means that it is safe to use the system workqueues instead of
tmr_wq for processing TMFs. An intended side effect of this patch is that
it enables concurrent processing of TMFs.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Make sure that target_wait_for_sess_cmds() waits long enough
Bart Van Assche [Tue, 27 Nov 2018 23:51:58 +0000 (15:51 -0800)]
scsi: target/core: Make sure that target_wait_for_sess_cmds() waits long enough

A session must only be released after all code that accesses the session
structure has finished. Make sure that this is the case by introducing a
new command counter per session that is only decremented after the
.release_cmd() callback has finished. This patch fixes the following crash:

BUG: KASAN: use-after-free in do_raw_spin_lock+0x1c/0x130
Read of size 4 at addr ffff8801534b16e4 by task rmdir/14805
CPU: 16 PID: 14805 Comm: rmdir Not tainted 4.18.0-rc2-dbg+ #5
Call Trace:
dump_stack+0xa4/0xf5
print_address_description+0x6f/0x270
kasan_report+0x241/0x360
__asan_load4+0x78/0x80
do_raw_spin_lock+0x1c/0x130
_raw_spin_lock_irqsave+0x52/0x60
srpt_set_ch_state+0x27/0x70 [ib_srpt]
srpt_disconnect_ch+0x1b/0xc0 [ib_srpt]
srpt_close_session+0xa8/0x260 [ib_srpt]
target_shutdown_sessions+0x170/0x180 [target_core_mod]
core_tpg_del_initiator_node_acl+0xf3/0x200 [target_core_mod]
target_fabric_nacl_base_release+0x25/0x30 [target_core_mod]
config_item_release+0x9c/0x110 [configfs]
config_item_put+0x26/0x30 [configfs]
configfs_rmdir+0x3b8/0x510 [configfs]
vfs_rmdir+0xb3/0x1e0
do_rmdir+0x262/0x2c0
do_syscall_64+0x77/0x230
entry_SYSCALL_64_after_hwframe+0x49/0xbe

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Simplify transport_clear_lun_ref()
Bart Van Assche [Tue, 27 Nov 2018 23:51:57 +0000 (15:51 -0800)]
scsi: target/core: Simplify transport_clear_lun_ref()

Since transport_clear_lun_ref() already waits until the percpu-refcount
.release() method is called, it is not necessary to wait first until
percpu_ref_kill_and_confirm() has finished transitioning the refcount into
atomic mode. Remove the code that waits for percpu_ref_kill_and_confirm()
to complete and also the completion object that is used by that code.  This
patch does not change the behavior of the SCSI target code.

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.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>
5 years agoscsi: target/core: Use kvcalloc() instead of open-coding it
Bart Van Assche [Tue, 27 Nov 2018 23:51:56 +0000 (15:51 -0800)]
scsi: target/core: Use kvcalloc() instead of open-coding it

This patch does not change any functionality. Note: the code that frees
sess_cmd_map already uses kvfree() so that code does not need to be
modified.

Reviewed-by: David Disseldorp <ddiss@suse.de>
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>