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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

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>

scsi: target: Fix target_wait_for_sess_cmds breakage with active signals

With the addition of commit 00d909a10710 ("scsi: target: Make the session
shutdown code also wait for commands that are being aborted") in v4.19-rc, it
incorrectly assumes no signals will be pending for task_struct executing the
normal session shutdown and I/O quiesce code-path.

For example, iscsi-target and iser-target issue SIGINT to all kthreads as part
of session shutdown.  This has been the behaviour since day one.

As-is when signals are pending with se_cmds active in se_sess->sess_cmd_list,
wait_event_interruptible_lock_irq_timeout() returns a negative number and
immediately kills the machine because of the do while (ret <= 0) loop that was
added in commit 00d909a107 to spin while backend I/O is taking any amount of
extended time (say 30 seconds) to complete.

Here's what it looks like in action with debug plus delayed backend I/O
completion:

[ 4951.909951] se_sess: 000000003e7e08fa before target_wait_for_sess_cmds
[ 4951.914600] target_wait_for_sess_cmds: signal_pending: 1
[ 4951.918015] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 0
[ 4951.921639] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 1
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 2
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 3
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 4
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 5
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 6
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 7
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 8
[ 4951.921944] wait_event_interruptible_lock_irq_timeout ret: -512 signal_pending: 1 loop count: 9

... followed by the usual RCU CPU stalls and deadlock.

There was never a case pre commit 00d909a107 where
wait_for_complete(&se_cmd->cmd_wait_comp) was able to be interrupted, so to
address this for v4.19+ moving forward go ahead and use
wait_event_lock_irq_timeout() instead so new code works with all fabric
drivers.

Also for commit 00d909a107, fix a minor regression in
target_release_cmd_kref() to only wake_up the new se_sess->cmd_list_wq only
when shutdown has actually been triggered via se_sess->sess_tearing_down.

Fixes: 00d909a10710 ("scsi: target: Make the session shutdown code also wait for commands that are being aborted")
Cc: <stable@vger.kernel.org> # v4.19+
Cc: Bart Van Assche <bvanassche@acm.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Sagi Grimberg <sagi@grimberg.me>
Cc: Bryant G. Ly <bryantly@linux.vnet.ibm.com>
Tested-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Reviewed-by: Bryant G. Ly <bly@catalogicsoftware.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sched/wait: Add wait_event_lock_irq_timeout for TASK_UNINTERRUPTIBLE usage

Short of reverting commit 00d909a10710 ("scsi: target: Make the session
shutdown code also wait for commands that are being aborted") for v4.19,
target-core needs a wait_event_t macro can be executed using
TASK_UNINTERRUPTIBLE to function correctly with existing fabric drivers that
expect to run with signals pending during session shutdown and active se_cmd
I/O quiesce.

The most notable is iscsi-target/iser-target, while ibmvscsi_tgt invokes
session shutdown logic from userspace via configfs attribute that could also
potentially have signals pending.

So go ahead and introduce wait_event_lock_irq_timeout() to achieve this, and
update + rename __wait_event_lock_irq_timeout() to make it accept 'state' as a
parameter.

Fixes: 00d909a10710 ("scsi: target: Make the session shutdown code also wait for commands that are being aborted")
Cc: <stable@vger.kernel.org> # v4.19+
Cc: Bart Van Assche <bvanassche@acm.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Sagi Grimberg <sagi@grimberg.me>
Cc: Bryant G. Ly <bryantly@linux.vnet.ibm.com>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Reviewed-by: Bryant G. Ly <bly@catalogicsoftware.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libfc: retry PRLI if we cannot analyse the payload

When we fail to analyse the payload of a PRLI response we should reset
the state machine to retry the PRLI; eventually we will be getting a
proper frame.  Not doing so will result in a stuck state machine and the
port never to be presented to the systsm.

Suggested-by: Chad Dupuis <chad.dupuis@cavium.com>
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Arun Easi <arun.easi@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libfc: check fc_frame_payload_get() return value for null

We should not assume the payload of a PRLI or PLOGI respons is always
present.

Signed-off-by: Thomas Abraham <tabraham@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Arun Easi <arun.easi@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Allow state transitions from OFFLINE to BLOCKED

When an RSCN gets delayed (or not being sent at all), the transport class
will detect an error, EH kicks in, and eventually will be setting the
device to offline.  If we receive an RSCN after that, the device will
stay in 'offline'.  This patch allows for an 'offline' to 'blocked'
transition, thereby allowing the device to become active again.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: esp_scsi: remove union in esp_cmd_priv

The dma_addr_t member is unused ever since we switched the SCSI
layer to send down single-segement command using a scatterlist
as well many years ago.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: esp_scsi: move dma mapping into the core code

Except for the mac_esp driver, which uses PIO or pseudo DMA, all drivers
share the same dma mapping calls.  Move the dma mapping into the core
code using the scsi_dma_map / scsi_dma_unmap helpers, with a special
identify mapping variant triggered off a new ESP_FLAG_NO_DMA_MAP flag
for mac_esp.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: esp_scsi: remove the dev argument to scsi_esp_register

We can simplify use esp->dev now.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: esp_scsi: use strong typing for the dev field

esp->dev is a void pointer that points either to a struct device, or a
struct platform_device.  As we can easily get from the device to the
platform_device if needed change it to always point to a struct device
and properly type the pointer to avoid errors.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sun_esp: don't use GFP_ATOMIC for command block allocation

esp_sbus_map_command_block is called straight from the probe routine
without any locks held, so we can safely use GFP_KERNEL here.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: am53c974: use the generic DMA API

Remove usage of the legacy PCI DMA API.  To make this easier we also
store a struct device instead of pci_dev in the dev field of struct
esp.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Tested-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: make UFS Tx lane1 clock optional for QCOM platforms

Per Qcom's UFS host controller HW design, the UFS Tx lane1 clock could be
muxed with Tx lane0 clock, hence keep Tx lane1 clock optional by ignoring
it if it is not provided in device tree. This change also performs some
cleanup to lanes per direction checks when enable/disable lane clocks
just for symmetry.

Signed-off-by: Venkat Gopalakrishnan <venkatg@codeaurora.org>
Signed-off-by: Subhash Jadavani <subhashj@codeaurora.org>
Signed-off-by: Can Guo <cang@codeaurora.org>
Reviewed-by: Vivek Gautam <vivek.gautam@codeaurora.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: fix integer type usage in uapi header

We get a warning from 'make headers_check' about a newly introduced usage
of integer types in the scsi/scsi_bsg_ufs.h uapi header:

usr/include/scsi/scsi_bsg_ufs.h:18: found __[us]{8,16,32,64} type without #include <linux/types.h>

Aside from the missing linux/types.h inclusion, I also noticed that it
uses the wrong types: 'u32' is not available at all in user space, and
'uint32_t' depends on the inclusion of a standard header that we should
not include from kernel headers.

Change the all to __u32 and similar types here.

I also note the usage of '__be32' and '__be16' that seems unfortunate for
a user space API. I wonder if it would be better to define the interface
in terms of a CPU-endian structure and convert it in kernel space.

Fixes: e77044c5a842 ("scsi: ufs-bsg: Add support for uic commands in ufs_bsg_request()")
Fixes: df032bf27a41 ("scsi: ufs: Add a bsg endpoint that supports UPIUs")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: remove a bogus pci_dma_sync_single_for_device call

dma_alloc_coherent allocates memory that can be used by the cpu and the
device at the same time, calls to pci_dma_sync_* are not required, and in
fact actively harmful on some architectures like arm.

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>

scsi: megaraid_mbox: remove bogus use of pci_dma_sync_sg_* APIs

The dma_map_sg / dma_unmap_sg APIs called from scsi_dma_map /
scsi_dma_unmap already transfer memory ownership to the device or cpu
respectively.  Adding additional calls to pci_dma_sync_sg_* will in fact
lead to data corruption if we end up using swiotlb for some reason.

Also remove the now pointless megaraid_mbox_sync_scb function.

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>

scsi: arcmsr: clean up clang warning on extraneous parentheses

There are extraneous parantheses that are causing clang to produce a
warning so remove these.

Clean up 3 clang warnings:
equality comparison with extraneous parentheses [-Wparentheses-equality]

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Ching Huang <ching2048@areca.com.tw>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs-bsg: Add support for uic commands in ufs_bsg_request()

Make ufshcd_send_uic_cmd() public for that.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <Bart.VanAssche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs-bsg: Add support for raw upiu in ufs_bsg_request()

Do that for the currently supported UPIUs: query, nop out, and task
management.

We do not support UPIU of type scsi command yet, while we are using the
job's request and reply pointers to hold the payload.  We will look into
it in later patches. We might need to elaborate the raw upiu api for
that.

We also still not supporting uic commands: For first phase, we plan to
use the existing api, and send only uic commands that are already
supported. Anyway, all that will come in the next patch.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <Bart.VanAssche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Add API to execute raw upiu commands

The UFS host software uses a combination of a host register set and
Transfer Request Descriptors in system memory to communicate with host
controller hardware. In its mmio space, a separate places are assigned
to UTP Transfer Request Descriptor ("utrd") list, and to UTP Task
Management Request Descriptor ("utmrd") list.

The provided API supports utrd-typed requests: nop out and device
management commands. It also supports utmrd-type requests:
task management requests. Other UPIU types are not supported for now.

We utilize the already existing code for tag and task work queues.
That is, all utrd-typed UPIUs are "disguised" as device management
commands. Similarly, the utmrd-typed UPUIs uses the task management
infrastructure.

It is up to the caller to fill the upiu request properly, as it will be
copied without any further input validations.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <Bart.VanAssche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Use data structure size in pointer arithmetic

Use the structure size in pointer arithmetic instead of an opaque 32
bytes for the over-allocation of descriptors.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Add a bsg endpoint that supports UPIUs

For now, just provide an API to allocate and remove ufs-bsg node. We
will use this framework to manage ufs devices by sending UPIU
transactions.

For the time being, implements an empty bsg_request() - will add some
more functionality in coming patches.

Nonetheless, we reveal here the protocol we are planning to use: UFS
Transport Protocol Transactions. UFS transactions consist of packets
called UFS Protocol Information Units (UPIU).

There are UPIU’s defined for UFS SCSI commands, responses, data in and
data out, task management, utility functions, vendor functions,
transaction synchronization and control, and more.

By using UPIUs, we get access to the most fine-grained internals of this
protocol, and able to communicate with the device in ways, that are
sometimes beyond the capacity of the ufs driver.

Moreover and as a result, our core structure - ufs_bsg_node has a pretty
lean structure: using upiu transactions that contains the outmost
detailed info, so we don't really need complex constructs to support it.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <Bart.VanAssche@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: uapi: ufs: Make utp_upiu_req visible to user space

in preparation to send UPIU requests via bsg.

Signed-off-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Bart Van Assche <Bart.VanAssche@wdc.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: add a low-level __ufshcd_issue_tm_cmd helper

Add a helper that takes a utp_task_req_desc and issues it, which will
be useful for UFS bsg support.  Rewrite ufshcd_issue_tm_cmd0x to use
this new helper.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: cleanup struct utp_task_req_desc

Remove the pointless task_req_upiu and task_rsp_upiu indirections,
which are __le32 arrays always cast to given structures and just add
the members directly.  Also clean up variables names in use in the
callers a bit to make the code more readable.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: snic: Replace spin_is_locked() with lockdep

lockdep_assert_held() is better suited to checking locking requirements, since
it won't get confused when someone else holds the lock. This is also a step
towards possibly removing spin_is_locked().

Signed-off-by: Lance Roy <ldr709@gmail.com>
Cc: Karan Tilak Kumar <kartilak@cisco.com>
Cc: Sesidhar Baddela <sebaddel@cisco.com>
Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: <linux-scsi@vger.kernel.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Remove unnecessary parentheses and simplify null checks

Clang warns when multiple pairs of parentheses are used for a single
conditional statement.

drivers/scsi/mpt3sas/mpt3sas_base.c:535:11: warning: equality comparison
with extraneous parentheses [-Wparentheses-equality]
        if ((ioc == NULL))
             ~~~~^~~~~~~
drivers/scsi/mpt3sas/mpt3sas_base.c:535:11: note: remove extraneous
parentheses around the comparison to silence this warning
        if ((ioc == NULL))
            ~    ^      ~
drivers/scsi/mpt3sas/mpt3sas_base.c:535:11: note: use '=' to turn this
equality comparison into an assignment
        if ((ioc == NULL))
                 ^~
                 =
drivers/scsi/mpt3sas/mpt3sas_base.c:539:12: warning: equality comparison
with extraneous parentheses [-Wparentheses-equality]
        if ((pdev == NULL))
             ~~~~~^~~~~~~
drivers/scsi/mpt3sas/mpt3sas_base.c:539:12: note: remove extraneous
parentheses around the comparison to silence this warning
        if ((pdev == NULL))
            ~     ^      ~
drivers/scsi/mpt3sas/mpt3sas_base.c:539:12: note: use '=' to turn this
equality comparison into an assignment
        if ((pdev == NULL))
                  ^~
                  =
2 warnings generated.

Remove them and while we're at it, simplify the NULL checks as '!var' is
used more than 'var == NULL'.

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Use dma_pool_zalloc

Replaced dma_pool_alloc + memset with dma_pool_zalloc.

Signed-off-by: Brajeswar Ghosh <brajeswar.linux@gmail.com>
Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Remove unused macro MPT3SAS_FMT

All the uses have been removed, delete the macro.

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Convert logging uses with MPT3SAS_FMT without logging levels

Convert these uses to ioc_<level> where appropriate.

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Remove KERN_WARNING from panic uses

Remove the logging level as panic calls stop the machine and should always be
emitted regardless of requested logging level.

These existing panic uses are perhaps inappropriate.

Miscellanea:

o Coalesce formats and convert MPT3SAS_FMT to "%s: " to improve clarity

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Convert logging uses with MPT3SAS_FMT and reply_q_name to %s:

Convert the existing 2 uses to make the format and arguments matching more
obvious.

Miscellanea:

o Move the word "enabled" into the format to trivially reduce object size
o Remove unnecessary parentheses

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Convert mlsleading uses of pr_<level> with MPT3SAS_FMT

These have misordered uses of __func__ and ioc->name that could mismatch
MPT3SAS_FMT and "%s: ".

Convert them to ioc_<level>.

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Convert uses of pr_<level> with MPT3SAS_FMT to ioc_<level>

Use a more common logging style.

Done using the perl script below and some typing

$ git grep --name-only -w MPT3SAS_FMT -- "*.c" | \
  xargs perl -i -e 'local $/; while (<>) { s/\bpr_(info|err|notice|warn)\s*\(\s*MPT3SAS_FMT\s*("[^"]+"(?:\s*\\?\s*"[^"]+"\s*){0,5}\s*),\s*ioc->name\s*/ioc_\1(ioc, \2/g; print;}'

Miscellanea for these conversions:

o Coalesce formats
o Realign arguments
o Remove unnecessary parentheses
o Use casts to u64 instead of unsigned long long where appropriate
o Convert broken pr_info uses to pr_cont
o Fix broken format string concatenation with line continuations and
  excess whitespace

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mpt3sas: Add ioc_<level> logging macros

These macros can help identify specific logging uses and eventually perhaps
reduce object sizes.

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: MAINTAINERS: Fix typo in cxlflash stanza

The uapi header file listed in the cxlflash stanza has a typo.

Removed the trailing 's' from the filename.

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

scsi: qla2xxx: Return switch command on a timeout

This patch fixes commit bcc71cc3cde1 ("scsi: qla2xxx: Fix for double
free of SRB structure") which placed code in wrong routines.

Also updated the use of WARN_ON() to WARN_ON_ONCE() to prevent
flooding log messages.

Fixes: bcc71cc3cde1 ("scsi: qla2xxx: Fix for double free of SRB structure")
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Check for bus reset

The SR_RST bit isn't latched. Hence, detecting a bus reset isn't reliable.
When it is detected, the right thing to do is to drop all connected and
disconnected commands. The code for that is already present so refactor it and
call it when SR_RST is set.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Handle BUS FREE during reselection

The X3T9.2 specification (draft) says, under "6.1.4.2 RESELECTION time-out
procedure", that a target may assert RST or go to BUS FREE phase if the
initiator does not respond within 200 us. Something like this has been
observed with AztecMonster II target. When it happens, all we can do is wait
for the target to try again.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Don't call dsprintk() following reselection interrupt

The X3T9.2 specification (draft) says, under "6.1.4.1 RESELECTION",

    ... The reselected initiator shall then assert the BSY signal
    within a selection abort time of its most recent detection of being
    reselected; this is required for correct operation of the time-out
    procedure.

The selection abort time is only 200 us which may be insufficient time for a
printk() call. Move the diagnostics to the error paths.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Don't clear busy flag when abort fails

When NCR5380_abort() returns FAILED, the driver forgets that the target is
still busy. Hence, further commands may be sent to the target, which may fail
during selection and produce the error message, "reselection after won
arbitration?". Prevent this by leaving the busy flag set when NCR5380_abort()
fails.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Check for invalid reselection target

The X3T9.2 specification (draft) says, under "6.1.4.1 RESELECTION", that "the
initiator shall not respond to a RESELECTION phase if other than two SCSI ID
bits are on the DATA BUS." This issue (too many bits set) has been observed in
the wild, so add a check.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Use DRIVER_SENSE to indicate valid sense data

When sense data is valid, call set_driver_byte(cmd, DRIVER_SENSE).  Otherwise
some callers of scsi_execute() will ignore sense data.  Don't set DID_ERROR or
DID_RESET just because sense data is missing.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Withhold disconnect privilege for REQUEST SENSE

This is mostly needed because an AztecMonster II target has been observed
disconnecting REQUEST SENSE commands and then failing to reselect properly.

Suggested-by: Michael Schmitz <schmitzmic@gmail.com>
Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Have NCR5380_select() return a bool

The return value is taken to mean "retry" or "don't retry". Change it to bool
to improve readability. Fix related comments. No functional change.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Reduce goto statements in NCR5380_select()

Replace a 'goto' statement with a simple 'return' where possible.  This
improves readability. No functional change.

Tested-by: Michael Schmitz <schmitzmic@gmail.com>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: NCR5380: Clear all unissued commands on host reset

When doing a host reset we should be clearing all outstanding commands, not
just the command triggering the reset.

[mkp: adjusted Hannes' SoB address]

Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Ondrey Zary <linux@rainbow-software.org>
Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: mptfusion: Remove unnecessary parentheses

Clang warns when multiple pairs of parentheses are used for a single
conditional statement.

drivers/message/fusion/mptbase.c:338:11: warning: equality comparison
with extraneous parentheses [-Wparentheses-equality]
        if ((ioc == NULL))
             ~~~~^~~~~~~
drivers/message/fusion/mptbase.c:338:11: note: remove extraneous
parentheses around the comparison to silence this warning
        if ((ioc == NULL))
            ~    ^      ~
drivers/message/fusion/mptbase.c:338:11: note: use '=' to turn this
equality comparison into an assignment
        if ((ioc == NULL))
                 ^~
                 =
drivers/message/fusion/mptbase.c:342:12: warning: equality comparison
with extraneous parentheses [-Wparentheses-equality]
        if ((pdev == NULL))
             ~~~~~^~~~~~~
drivers/message/fusion/mptbase.c:342:12: note: remove extraneous
parentheses around the comparison to silence this warning
        if ((pdev == NULL))
            ~     ^      ~
drivers/message/fusion/mptbase.c:342:12: note: use '=' to turn this
equality comparison into an assignment
        if ((pdev == NULL))
                  ^~
                  =
2 warnings generated.

Remove them and while we're at it, simplify the NULL checks as '!var' is
used more than 'var == NULL'.

Reported-by: Nick Desaulniers <ndesaulniers@google.com>
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>

scsi: qla4xxx: Remove redundant check on drvr_wait

The check for a non-zero drvr_wait is redundant as the same check is performed
earlier in the outer while loop, the inner check will always be true if we
reached this point inside the while loop.  Remove the redundant if check.

Detected by cppcheck:
(warning) Identical inner 'if' condition is always true.

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

scsi: qla2xxx: Move log messages before issuing command to firmware

There is a probability that the SRB structure might have been released by the
time the debug log message dereferences it.  This patch moved the log messages
before the command is issued to the firmware to prevent unknown behavior and
kernel crash

Fixes: 726b85487067 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org>
Signed-off-by: Giridhar Malavali <giridhar.malavali@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix for double free of SRB structure

This patch fixes issue during switch command query where driver was freeing
SRB resources multiple times

Following stack trace will be seen
[  853.436234] BUG: unable to handle kernel NULL pointer dereference at
0000000000000001
[  853.436348] IP: [<ffffffff811df514>] kmem_cache_alloc+0x74/0x1e0
[  853.436476] PGD 0
[  853.436601] Oops: 0000 [#1] SMP

[  853.454700]  [<ffffffff81099f6a>] ? mod_timer+0x14a/0x220
[  853.455543]  [<ffffffff81185465>] mempool_alloc_slab+0x15/0x20
[  853.456395]  [<ffffffff811855a9>] mempool_alloc+0x69/0x170
[  853.457257]  [<ffffffff81098af2>] ? internal_add_timer+0x32/0x70
[  853.458136]  [<ffffffffc0092d2b>] qla2xxx_queuecommand+0x29b/0x3f0 [qla2xxx]
[  853.459024]  [<ffffffff8146535a>] scsi_dispatch_cmd+0xaa/0x230
[  853.459923]  [<ffffffff8146e11f>] scsi_request_fn+0x4df/0x680
[  853.460829]  [<ffffffff81029557>] ? __switch_to+0xd7/0x510
[  853.461747]  [<ffffffff812f7113>] __blk_run_queue+0x33/0x40
[  853.462670]  [<ffffffff812f7735>] blk_delay_work+0x25/0x40
[  853.463603]  [<ffffffff810a882a>] process_one_work+0x17a/0x440
[  853.464546]  [<ffffffff810a94f6>] worker_thread+0x126/0x3c0
[  853.465501]  [<ffffffff810a93d0>] ? manage_workers.isra.24+0x2a0/0x2a0
[  853.466447]  [<ffffffff810b099f>] kthread+0xcf/0xe0
[  853.467379]  [<ffffffff810b08d0>] ? insert_kthread_work+0x40/0x40
[  853.470172] Code: db e2 7e 49 8b 50 08 4d 8b 20 49 8b 40 10 4d 85 e4 0f 84 20
01 00 00 48 85 c0 0f 84 17 01 00 00 49 63 46 20 48 8d 4a 01 4d 8b 06 <49> 8b 1c
04 4c 89 e0 65 49 0f c7 08 0f 94 c0 84 c0 74 ba 49 63
[  853.472072] RIP  [<ffffffff811df514>] kmem_cache_alloc+0x74/0x1e0
[  853.472971]  RSP <ffff88103726fc50>

Fixes: 726b85487067 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org>
Signed-off-by: Giridhar Malavali <giridhar.malavali@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix recursive mailbox timeout

This patch prevents user space mailbox request from doing chip reset if the
mailbox timed out. The chip reset is only reserved for the DPC thread to
ensure all mailbox requests are flushed properly. The DPC thread is
responsible for the flushing all MBs and chip reset.

Fixes: b2000805a975 ("scsi: qla2xxx: Flush mailbox commands on chip reset")
Cc: <stable@ger.kernel.org>
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix driver hang when FC-NVMe LUNs are configured

This patch fixes multiple call for qla_nvme_unregister_remote_port() as part
of qlt_schedule_session_for_deletion(), Do not call it again during
qla_nvme_delete()

Fixes: e473b3074104 ("scsi: qla2xxx: Add FC-NVMe abort processing")
Cc: <stable@vger.kernel.org>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix re-using LoopID when handle is in use

This patch fixes issue where driver clears NPort ID map instead of marking
handle in use. Once driver clears NPort ID from the database, it can reuse
the same NPort ID resulting in a PLOGI failure.

[mkp: fixed Himanshu's SoB]

Fixes: a084fd68e1d2 ("scsi: qla2xxx: Fix re-login for Nport Handle in use")
Cc: <stable@vger.kernel.org>
Signed-of-by: Quinn Tran <quinn.tran@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <hmadhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix duplicate switch database entries

The response data buffer used in switch scan is reused 4 times.  (For example,
for commands GPN_FT, GNN_FT for FCP and FC-NVME) Before driver reuses this
buffer, clear it to prevent duplicate entries in our database.

Fixes: a4239945b8ad1 ("scsi: qla2xxx: Add switch command to simplify fabric discovery"
Cc: <stable@vger.kernel.org>
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix NVMe Target discovery

This patch fixes issue when remoteport registers itself as both FCP and
FC-NVMe with the switch, driver will pick FC-NVMe personality as default when
scanning for targets.

Driver was using comaprative operator instead of bitwise operator to check for
fc4_type for both FCP and FC-NVME.

Fixes: 2b5b96473efc ("scsi: qla2xxx: Fix FC-NVMe LUN discovery")
Cc: <stable@vger.kernel.org>
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix NVMe session hang on unload

Send aborts only when chip is active.

Fixes: 623ee824e579 ("scsi: qla2xxx: Fix FC-NVMe IO abort during driver reset")
Cc: <stable@vger.kernel.org> # 4.14
Signed-off-by: Quinn Tran <quinn.tran@cavium.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: don't allow negative thresholds

We shouldn't allow negative thresholds.  I don't know what it would do
but it can't be good.

Fixes: 8b4673ba3a1b ("scsi: qla2xxx: Add support for ZIO6 interrupt threshold")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix comment in MODULE_PARM_DESC in qla2xxx

Default value of ql2xasynctmfenable for qla2xxx driver was set to 1 in
commit 043dc1d7e850 ("scsi: qla2xxx: Enable Async TMF processing") but
comment in MODULE_PARAM_DESC was not modified.

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target: iscsi: cxgbit: fix csk leak

csk leak can happen if a new TCP connection gets established after
cxgbit_accept_np() returns, to fix this leak free remaining csk in
cxgbit_free_np().

Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: fix a race condition when smp task timeout

When the lldd is processing the complete sas task in interrupt and set the
task stat as SAS_TASK_STATE_DONE, the smp timeout timer is able to be
triggered at the same time. And smp_task_timedout() will complete the task
wheter the SAS_TASK_STATE_DONE is set or not. Then the sas task may freed
before lldd end the interrupt process. Thus a use-after-free will happen.

Fix this by calling the complete() only when SAS_TASK_STATE_DONE is not
set. And remove the check of the return value of the del_timer(). Once the
LLDD sets DONE, it must call task->done(), which will call
smp_task_done()->complete() and the task will be completed and freed
correctly.

Reported-by: chenxiang <chenxiang66@hisilicon.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: John Garry <john.garry@huawei.com>
CC: Johannes Thumshirn <jthumshirn@suse.de>
CC: Ewan Milne <emilne@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Tomas Henzl <thenzl@redhat.com>
CC: Dan Williams <dan.j.williams@intel.com>
CC: Hannes Reinecke <hare@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: John Garry <john.garry@huawei.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: check the ata device status by ata_dev_enabled()

When ata device IDENTIFY failed, the ata device status is ATA_DEV_UNKNOWN. The
libata reported like:

[113518.620433] ata5.00: qc timeout (cmd 0xec)
[113518.653646] ata5.00: failed to IDENTIFY (I/O error, err_mask=0x4)

But libsas verifies the device status by ata_dev_disabled(), which skipped
ATA_DEV_UNKNOWN. This will make libsas think the ata device probing succeed
the device cannot be actually brought up. And even the new bcast of this
device will be considered as flutter and will not probe this device again.

Change ata_dev_disabled() to !ata_dev_enabled() so that libsas can deal with
this if the ata device probe failed. New bcasts can let us try to probe the
device again and bring it up if it is fine to IDENTIFY.

Tested-by: Zhou Yupeng <zhouyupeng1@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: John Garry <john.garry@huawei.com>
CC: Johannes Thumshirn <jthumshirn@suse.de>
CC: Ewan Milne <emilne@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Tomas Henzl <thenzl@redhat.com>
CC: Dan Williams <dan.j.williams@intel.com>
CC: Hannes Reinecke <hare@suse.com>
Reviewed-by: John Garry <john.garry@huawei.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: always unregister the old device if going to discover new

If we went into sas_rediscover_dev() the attached_sas_addr was already insured
not to be zero. So it's unnecessary to check if the attached_sas_addr is zero.

And although if the sas address is not changed, we always have to unregister
the old device when we are going to register a new one. We cannot just leave
the device there and bring up the new.

Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: chenxiang <chenxiang66@hisilicon.com>
CC: John Garry <john.garry@huawei.com>
CC: Johannes Thumshirn <jthumshirn@suse.de>
CC: Ewan Milne <emilne@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Tomas Henzl <thenzl@redhat.com>
CC: Dan Williams <dan.j.williams@intel.com>
CC: Hannes Reinecke <hare@suse.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: make the lldd_port_deformed method optional

Now LLDDs have to implement lldd_port_deformed method otherwise NULL
dereference will happen. Make it optional and remove the dummy implementation
in hisi_sas.

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

scsi: libsas: delete dead code in scsi_transport_sas.c

This code is dead and no clue implies that it will be back again.

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

scsi: megaraid: fix spelling mistake "maibox" -> "mailbox"

Trivial fix to spelling mistake in warning message and comments

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

scsi: FlashPoint: Remove unnecessary parentheses

Clang warns when multiple pairs of parentheses are used for a single
conditional statement.

In file included from drivers/scsi/BusLogic.c:57:
drivers/scsi/FlashPoint.c:2947:34: warning: equality comparison with extraneous parentheses [-Wparentheses-equality]
                                if ((currSCCB->Sccb_scsistat == SELECT_SN_ST)) {
                                     ~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~
drivers/scsi/FlashPoint.c:2947:34: note: remove extraneous parentheses around the comparison to silence this warning
                                if ((currSCCB->Sccb_scsistat == SELECT_SN_ST)) {
                                    ~                        ^              ~
drivers/scsi/FlashPoint.c:2947:34: note: use '=' to turn this equality comparison into an assignment
                                if ((currSCCB->Sccb_scsistat == SELECT_SN_ST)) {
                                                             ^~
                                                             =
drivers/scsi/FlashPoint.c:2956:39: warning: equality comparison with extraneous parentheses [-Wparentheses-equality]
                                else if ((currSCCB->Sccb_scsistat ==
                                          ~~~~~~~~~~~~~~~~~~~~~~~~^~
drivers/scsi/FlashPoint.c:2956:39: note: remove extraneous parentheses around the comparison to silence this warning
                                else if ((currSCCB->Sccb_scsistat ==
                                         ~                        ^
drivers/scsi/FlashPoint.c:2956:39: note: use '=' to turn this equality comparison into an assignment
                                else if ((currSCCB->Sccb_scsistat ==
                                                                  ^~
                                                                  =
2 warnings generated.

Link: https://github.com/ClangBuiltLinux/linux/issues/156
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Khalid Aziz <khalid@gonehiking.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: hpsa: Use vmemdup_user to replace the open code

vmemdup_user is better than duplicating its implementation, So just replace
the open code.

The issue is detected with the help of Coccinelle.

Tested-by: Don Brace <don.brace@microsemi.com>
Acked-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: advansys: Remove unnecessary parentheses

Clang warns when multiple pairs of parentheses are used for a single
conditional statement.

drivers/scsi/advansys.c:6451:20: warning: equality comparison with
extraneous parentheses [-Wparentheses-equality]
                                if ((sdtr_data == 0xFF)) {
                                     ~~~~~~~~~~^~~~~~~
drivers/scsi/advansys.c:6451:20: note: remove extraneous parentheses
around the comparison to silence this warning
                                if ((sdtr_data == 0xFF)) {
                                    ~          ^      ~
drivers/scsi/advansys.c:6451:20: note: use '=' to turn this equality
comparison into an assignment
                                if ((sdtr_data == 0xFF)) {
                                               ^~
                                               =
1 warning generated.

Link: https://github.com/ClangBuiltLinux/linux/issues/155
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target: iblock: split T10 PI SGL across command bios

When T10 PI is enabled on a backing device for the iblock backstore, the PI
SGL for the entire command is attached to the first bio only.  This works fine
if the command is covered by a single bio, but can result in ref tag errors in
the client for the other bios in a multi-bio command, e.g.

[   47.631236] sda: ref tag error at location 2048 (rcvd 0)
[   47.637658] sda: ref tag error at location 4096 (rcvd 0)
[   47.644228] sda: ref tag error at location 6144 (rcvd 0)

The command will be split into multiple bios if the number of data SG elements
exceeds BIO_MAX_PAGES (see iblock_get_bio()).

The bios may later be split again in the block layer on the host after
iblock_submit_bios(), depending on the queue limits of the backing device.
The block and SCSI layers will pass through the whole PI SGL down to the LLDD
however that first bio is split up, but the LLDD may only use the portion that
corresponds to the data length (depends on the LLDD, tested with scsi_debug).

Split the PI SGL across the bios in the command, so each bio's
bio_integrity_payload contains the protection information for the data in the
bio.  Use an sg_mapping_iter to keep track of where we are in PI SGL, so we
know where to start with the next bio.

Signed-off-by: Greg Edwards <gedwards@ddn.com>
Reviewed-by: Mike Christie <mchristi@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove set but not used variable 'ptr_dma'

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

drivers/scsi/qla2xxx/qla_iocb.c: In function 'qla24xx_els_dcmd2_iocb':
drivers/scsi/qla2xxx/qla_iocb.c:2644:13: warning:
 variable 'ptr_dma' set but not used [-Wunused-but-set-variable]

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

scsi: lpfc: Remove set but not used variable 'sgl_size'

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

drivers/scsi/lpfc/lpfc_nvme.c: In function 'lpfc_new_nvme_buf':
drivers/scsi/lpfc/lpfc_nvme.c:2238:24: warning:
 variable 'sgl_size' set but not used [-Wunused-but-set-variable]
  int bcnt, num_posted, sgl_size;
                        ^
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bnx2fc: Remove set but not used variable 'oxid'

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

drivers/scsi/bnx2fc/bnx2fc_fcoe.c: In function 'bnx2fc_rcv':
drivers/scsi/bnx2fc/bnx2fc_fcoe.c:435:17: warning:
 variable 'oxid' 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>

scsi: core: remove unnecessary unlikely()

BUG_ON() already contains an unlikely(), there is no need for another one.

Signed-off-by: Igor Stoppa <igor.stoppa@huawei.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com>
Cc: linux-scsi@vger.kernel.org
CC: linux-kernel@vger.kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aic7xxx: remove set but not used variable 'shared_scb_data'

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

drivers/scsi/aic7xxx/aic79xx_pci.c: In function 'ahd_pci_config':
drivers/scsi/aic7xxx/aic79xx_pci.c:291:19: warning:
 variable 'shared_scb_data' 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>

scsi: iscsi: target: fix spelling mistake "entires" -> "entries"

Trivial fix to spelling mistake in function name and comment

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

scsi: pm80xx: Remove set but not used variable 'page_code'

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

drivers/scsi/pm8001/pm80xx_hwi.c: In function 'pm8001_set_phy_profile':
drivers/scsi/pm8001/pm80xx_hwi.c:4679:6: warning:
 variable 'page_code' set but not used [-Wunused-but-set-variable]

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: pm80xx: Remove set but not used variable 'device_id'

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

drivers/scsi/pm8001/pm8001_sas.c: In function 'pm8001_I_T_nexus_event_handler':
drivers/scsi/pm8001/pm8001_sas.c:1052:6: warning:
 variable 'device_id' set but not used [-Wunused-but-set-variable]

drivers/scsi/pm8001/pm8001_sas.c: In function 'pm8001_abort_task':
drivers/scsi/pm8001/pm8001_sas.c:1191:6: warning:
 variable 'device_id' set but not used [-Wunused-but-set-variable]

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufshcd: Fix NULL pointer dereference for in ufshcd_init

Error paths in ufshcd_init() ufshcd_hba_exit() killed clk_scaling workqueue
when the workqueue is actually created quite late in ufshcd_init().  So, we
end up getting NULL pointer dereference in such error paths.  Fix this by
moving clk_scaling initialization and kill codes to two separate methods, and
call them at required places.

Fixes: 401f1e4490ee ("scsi: ufs: don't suspend clock scaling during clock
gating")

Signed-off-by: Vivek Gautam <vivek.gautam@codeaurora.org>
Cc: Bjorn Andersson <bjorn.andersson@linaro.org>
Cc: Subhash Jadavani <subhashj@codeaurora.org>
Cc: Matthias Kaehlcke <mka@chromium.org>
Cc: Evan Green <evgreen@chromium.org>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Evan Green <evgreen@chromium.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: message: fusion: fix a few trivial spelling mistakes

Trival fix to spelling mistakes:

 PrimativeSeqErrCount -> PrimitiveSeqErrCount
 Primative -> Primitive
 primative -> primitive
 mptsas_broadcast_primative_work -> mptsas_broadcast_primitive_work
 Broadcase -> Broadcast

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

scsi: pm80xx: Update driver version to 0.1.39

Updated the driver version from 0.1.38 to 0.1.39.

Signed-off-by: Deepak Ukey <deepak.ukey@microchip.com>
Signed-off-by: Viswas G <Viswas.G@microchip.com>
Acked-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: pm80xx: Fixed system hang issue during kexec boot

When the firmware is not responding, execution of kexec boot causes a system
hang. When firmware assertion happened, driver get notified with interrupt
vector updated in MPI configuration table. Then, the driver will read
scratchpad register and set controller_fatal_error flag to true.

Signed-off-by: Deepak Ukey <deepak.ukey@microchip.com>
Signed-off-by: Viswas G <Viswas.G@microchip.com>
Acked-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: pm80xx: Corrected dma_unmap_sg() parameter

For the function dma_unmap_sg(), the <nents> parameter should be number of
elements in the scatter list prior to the mapping, not after the mapping.

Signed-off-by: Deepak Ukey <deepak.ukey@microchip.com>
Signed-off-by: Viswas G <Viswas.G@microchip.com>
Acked-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: pm80xx: Fix for phy enable/disable functionality

Added proper mask for phy id in mpi_phy_stop_resp().

Signed-off-by: Deepak Ukey <deepak.ukey@microchip.com>
Signed-off-by: Viswas G <Viswas.G@microchip.com>
Acked-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: fix typo "CT-PASSTRHU" -> "CT-PASSTHRU"

Trivial fix to typo in debug message text.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Himanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>