scsi: ufs: ufshpb: Add "cold" regions timer

In order not to hang on to "cold" regions, we inactivate a region that has
had no READ access for a predefined amount of time - READ_TO_MS. For that
purpose monitor the active regions list, polling it on every
POLLING_INTERVAL_MS. On timeout expiry add the region to the
"to-be-inactivated" list unless it is clean and did not exhaust its
READ_TO_EXPIRIES - another parameter.

None of this applies to pinned regions.

Link: https://lore.kernel.org/r/20210712095039.8093-9-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Add HPB dev reset response

The spec does not define what the host's recommended response is when the
device sends HPB dev reset response (oper 0x2).

Update all active HPB regions.

Link: https://lore.kernel.org/r/20210712095039.8093-8-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Region inactivation in host mode

In host mode, the host is expected to send HPB WRITE BUFFER with buffer-id
= 0x1 when it inactivates a region.

Use the map-requests pool as there is no point in assigning a designated
cache for umap-requests.

[mkp: REQ_OP_DRV_*]

Link: https://lore.kernel.org/r/20210712095039.8093-7-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Make eviction depend on region's reads

In host mode, eviction is considered an extreme measure. Verify that the
entering region has enough reads, and the exiting region has fewer reads.

Link: https://lore.kernel.org/r/20210712095039.8093-6-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Add reads counter

In host control mode, reads are the major source of activation trials.
Keep track of those reads counters, for both active as well inactive
regions.

We reset the read counter upon write - we are only interested in "clean"
reads.

Keep those counters normalized, as we are using those reads as a
comparative score, to make various decisions.  If during consecutive
normalizations an active region has exhaust its reads - inactivate it.

While at it, protect the {active,inactive}_count stats by adding them into
the applicable handler.

Link: https://lore.kernel.org/r/20210712095039.8093-5-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Transform set_dirty to iterate_rgn

Given a transfer length, set_dirty meticulously iterates over all the
entries, across subregions and regions if needed. Currently its only use is
to mark dirty blocks, but HCM may benefit from it as well to manage its
read counters.

Link: https://lore.kernel.org/r/20210712095039.8093-4-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Add host control mode support to rsp_upiu

In device control mode, the device may recommend the host to either
activate or inactivate a region, and the host should follow. Meaning those
are not actually recommendations, but more of instructions.

Conversely, in host control mode, the recommendation protocol is slightly
changed:

 a) The device may only recommend the host to update a subregion of an
    already-active region. And,

 b) The device may *not* recommend to inactivate a region.

Furthermore, in host control mode, the host may choose not to follow any of
the device's recommendations. However, in case of a recommendation to
update an active and clean subregion, it is better to follow those
recommendation because otherwise the host has no other way to know that
some internal relocation took place.

Link: https://lore.kernel.org/r/20210712095039.8093-3-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Cache HPB Control mode on init

We will use control_mode later when we need to differentiate between device
and host control modes.

Link: https://lore.kernel.org/r/20210712095039.8093-2-avri.altman@wdc.com
Reviewed-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Add HPB 2.0 support

Version 2.0 of HBP supports reads of varying sizes from 4KB to 1MB.

A read operation <= 32KB is supported as single HPB read. A read between
36KB and 1MB is supported by a combination of write buffer command and HPB
read command to deliver more PPN. The write buffer commands may not be
issued immediately due to busy tags. To use HPB read more aggressively, the
driver can requeue the write buffer command. The requeue threshold is
implemented as timeout and can be modified with requeue_timeout_ms entry in
sysfs.

[mkp: REQ_OP_DRV_* and blk_rq_is_passthrough()]

Link: https://lore.kernel.org/r/20210712090025epcms2p3b3d94f6f1b2cfa394e3d9ba130ca0fa7@epcms2p3
Tested-by: Can Guo <cang@codeaurora.org>
Tested-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Can Guo <cang@codeaurora.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Prepare HPB read for cached sub-region

If the logical address of a read I/O belongs to an active sub-region, the
HPB driver modifies the read I/O command to an HPB read. The driver
modifies the UFS UPIU instead of modifying the existing SCSI command.

In HPB version 1.0, the maximum read I/O size that can be converted to HPB
read is 4KB.

The dirty map of the active sub-region prevents an incorrect HPB read that
has stale physical page number which is updated by previous write I/O.

[mkp: REQ_OP_DRV_* and blk_rq_is_passthrough()]

Link: https://lore.kernel.org/r/20210712085936epcms2p4b0ec5c8cecdeea6cc043d684363842b6@epcms2p4
Tested-by: Bean Huo <beanhuo@micron.com>
Tested-by: Can Guo <cang@codeaurora.org>
Tested-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Can Guo <cang@codeaurora.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Acked-by: Avri Altman <Avri.Altman@wdc.com>
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: L2P map management for HPB read

Implement L2P map management in HPB.

The HPB divides logical addresses into several regions. A region consists
of several sub-regions. The sub-region is a basic unit where L2P mapping is
managed. The driver loads L2P mapping data of each sub-region. The loaded
sub-region is called active-state. The HPB driver unloads L2P mapping data
as region unit. The unloaded region is called inactive-state.

Sub-region/region candidates to be loaded and unloaded are delivered from
the UFS device. The UFS device delivers the recommended active sub-region
and inactivate region to the driver using sense data. The HPB module
performs L2P mapping management on the host through the delivered
information.

A pinned region is a preset region on the UFS device that is always
in activate-state.

The data structures for map data requests and L2P mappings use the mempool
API, minimizing allocation overhead while avoiding static allocation.

The mininum size of the memory pool used in the HPB is implemented
as a module parameter so that it can be configurable by the user.

To guarantee a minimum memory pool size of 4MB: ufshpb_host_map_kbytes=4096.

The map_work manages active/inactive via 2 "to-do" lists:

 - hpb->lh_inact_rgn: regions to be inactivated
 - hpb->lh_act_srgn: subregions to be activated

These lists are maintained on I/O completion.

[mkp: switch to REQ_OP_DRV_*]

Link: https://lore.kernel.org/r/20210712085859epcms2p36e420f19564f6cd0c4a45d54949619eb@epcms2p3
Tested-by: Bean Huo <beanhuo@micron.com>
Tested-by: Can Guo <cang@codeaurora.org>
Tested-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Can Guo <cang@codeaurora.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Acked-by: Avri Altman <Avri.Altman@wdc.com>
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: ufshpb: Introduce Host Performance Buffer feature

Implement Host Performance Buffer (HPB) initialization and add function
calls to UFS core driver.

NAND flash-based storage devices, including UFS, have mechanisms to
translate logical addresses of I/O requests to the corresponding physical
addresses of the flash storage.  In UFS, logical-to-physical-address (L2P)
map data, which is required to identify the physical address for the
requested I/Os, can only be partially stored in SRAM from NAND flash. Due
to this partial loading, accessing the flash address area, where the L2P
information for that address is not loaded in the SRAM, can result in
serious performance degradation.

The basic concept of HPB is to cache L2P mapping entries in host system
memory so that both physical block address (PBA) and logical block address
(LBA) can be delivered in HPB read command. The HPB read command allows to
read data faster than a regular read command in UFS since it provides the
physical address (HPB Entry) of the desired logical block in addition to
its logical address. The UFS device can access the physical block in NAND
directly without searching and uploading L2P mapping table. This improves
read performance because the NAND read operation for uploading L2P mapping
table is removed.

In HPB initialization, the host checks if the UFS device supports HPB
feature and retrieves related device capabilities. Then, HPB parameters are
configured in the device.

Total start-up time of popular applications was measured and the difference
observed between HPB being enabled and disabled. Popular applications are
12 game apps and 24 non-game apps. Each test cycle consists of running 36
applications in sequence. We repeated the cycle for observing performance
improvement by L2P mapping cache hit in HPB.

The following is the test environment:

 - kernel version: 4.4.0
 - RAM: 8GB
 - UFS 2.1 (64GB)

Results:

   +-------+----------+----------+-------+
   | cycle | baseline | with HPB | diff  |
   +-------+----------+----------+-------+
   | 1     | 272.4    | 264.9    | -7.5  |
   | 2     | 250.4    | 248.2    | -2.2  |
   | 3     | 226.2    | 215.6    | -10.6 |
   | 4     | 230.6    | 214.8    | -15.8 |
   | 5     | 232.0    | 218.1    | -13.9 |
   | 6     | 231.9    | 212.6    | -19.3 |
   +-------+----------+----------+-------+

We also measured HPB performance using iozone:

   $ iozone -r 4k -+n -i2 -ecI -t 16 -l 16 -u 16 -s $IO_RANGE/16 -F \
   mnt/tmp_1 mnt/tmp_2 mnt/tmp_3 mnt/tmp_4 mnt/tmp_5 mnt/tmp_6 mnt/tmp_7 \
   mnt/tmp_8 mnt/tmp_9 mnt/tmp_10 mnt/tmp_11 mnt/tmp_12 mnt/tmp_13 \
   mnt/tmp_14 mnt/tmp_15 mnt/tmp_16

Results:

   +----------+--------+---------+
   | IO range | HPB on | HPB off |
   +----------+--------+---------+
   |   1 GB   | 294.8  | 300.87  |
   |   4 GB   | 293.51 | 179.35  |
   |   8 GB   | 294.85 | 162.52  |
   |  16 GB   | 293.45 | 156.26  |
   |  32 GB   | 277.4  | 153.25  |
   +----------+--------+---------+

Link: https://lore.kernel.org/r/20210712085830epcms2p8c1288b7f7a81b044158a18232617b572@epcms2p8
Reported-by: kernel test robot <lkp@intel.com>
Tested-by: Bean Huo <beanhuo@micron.com>
Tested-by: Can Guo <cang@codeaurora.org>
Tested-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Can Guo <cang@codeaurora.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Acked-by: Avri Altman <Avri.Altman@wdc.com>
Signed-off-by: Daejun Park <daejun7.park@samsung.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla4xxx: Convert uses of __constant_cpu_to_<foo> to cpu_to_<foo>

The macros cpu_to_le16() and cpu_to_le32() have special cases for
constants.  Their __constant_<foo> versions are not required.

On little endian systems, both cpu_to_le16() and __constant_cpu_to_le16()
expand to the same expression. Same is the case with cpu_to_le32().

On big endian systems, cpu_to_le16() expands to __swab16() which has a
__builtin_constant_p check. Similarly, cpu_to_le32() expands to __swab32().

Consequently these macros can be safely used with constants, and hence all
those uses are converted. This was discovered as a part of a checkpatch
evaluation, looking at all reports of WARNING:CONSTANT_CONVERSION error
type.

Link: https://lore.kernel.org/r/20210716112852.24598-1-dwaipayanray1@gmail.com
Signed-off-by: Dwaipayan Ray <dwaipayanray1@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: BusLogic: Use %X for u32 sized integer rather than %lX

An earlier fix changed the print format specifier for adapter->bios_addr to
use %lX. However, the integer is a u32 so the fix was wrong. Fix this by
using the correct %X format specifier.

Link: https://lore.kernel.org/r/20210730095031.26981-1-colin.king@canonical.com
Fixes: 43622697117c ("scsi: BusLogic: use %lX for unsigned long rather than %X")
Acked-by: Khalid Aziz <khalid@gonehiking.org>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Addresses-Coverity: ("Invalid type in argument")

scsi: BusLogic: Avoid unbounded vsprintf() use

Existing blogic_msg() invocations do not appear to overrun its internal
buffer of a fixed length of 100, which would cause stack corruption, but
it's easy to miss with possible further updates and a fix is cheap in
performance terms, so limit the output produced into the buffer by using
vscnprintf() rather than vsprintf().

Link: https://lore.kernel.org/r/alpine.DEB.2.21.2104201939390.44318@angie.orcam.me.uk
Acked-by: Khalid Aziz <khalid@gonehiking.org>
Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: BusLogic: Fix missing pr_cont() use

Update BusLogic driver's messaging system to use pr_cont() for continuation
lines, bringing messy output:

pci 0000:00:13.0: PCI->APIC IRQ transform: INT A -> IRQ 17
scsi: ***** BusLogic SCSI Driver Version 2.1.17 of 12 September 2013 *****
scsi: Copyright 1995-1998 by Leonard N. Zubkoff <lnz@dandelion.com>
scsi0: Configuring BusLogic Model BT-958 PCI Wide Ultra SCSI Host Adapter
scsi0:   Firmware Version: 5.07B, I/O Address: 0x7000, IRQ Channel: 17/Level
scsi0:   PCI Bus: 0, Device: 19, Address:
0xE0012000,
Host Adapter SCSI ID: 7
scsi0:   Parity Checking: Enabled, Extended Translation: Enabled
scsi0:   Synchronous Negotiation: Ultra, Wide Negotiation: Enabled
scsi0:   Disconnect/Reconnect: Enabled, Tagged Queuing: Enabled
scsi0:   Scatter/Gather Limit: 128 of 8192 segments, Mailboxes: 211
scsi0:   Driver Queue Depth: 211, Host Adapter Queue Depth: 192
scsi0:   Tagged Queue Depth:
Automatic
, Untagged Queue Depth: 3
scsi0:   SCSI Bus Termination: Both Enabled
, SCAM: Disabled

scsi0: *** BusLogic BT-958 Initialized Successfully ***
scsi host0: BusLogic BT-958

back to order:

pci 0000:00:13.0: PCI->APIC IRQ transform: INT A -> IRQ 17
scsi: ***** BusLogic SCSI Driver Version 2.1.17 of 12 September 2013 *****
scsi: Copyright 1995-1998 by Leonard N. Zubkoff <lnz@dandelion.com>
scsi0: Configuring BusLogic Model BT-958 PCI Wide Ultra SCSI Host Adapter
scsi0:   Firmware Version: 5.07B, I/O Address: 0x7000, IRQ Channel: 17/Level
scsi0:   PCI Bus: 0, Device: 19, Address: 0xE0012000, Host Adapter SCSI ID: 7
scsi0:   Parity Checking: Enabled, Extended Translation: Enabled
scsi0:   Synchronous Negotiation: Ultra, Wide Negotiation: Enabled
scsi0:   Disconnect/Reconnect: Enabled, Tagged Queuing: Enabled
scsi0:   Scatter/Gather Limit: 128 of 8192 segments, Mailboxes: 211
scsi0:   Driver Queue Depth: 211, Host Adapter Queue Depth: 192
scsi0:   Tagged Queue Depth: Automatic, Untagged Queue Depth: 3
scsi0:   SCSI Bus Termination: Both Enabled, SCAM: Disabled
scsi0: *** BusLogic BT-958 Initialized Successfully ***
scsi host0: BusLogic BT-958

Also diagnostic output such as with the BusLogic=TraceConfiguration
parameter is affected and becomes vertical and therefore hard to read.
This has now been corrected, e.g.:

pci 0000:00:13.0: PCI->APIC IRQ transform: INT A -> IRQ 17
blogic_cmd(86) Status = 30:  4 ==>  4: FF 05 93 00
blogic_cmd(95) Status = 28: (Modify I/O Address)
blogic_cmd(91) Status = 30:  1 ==>  1: 01
blogic_cmd(04) Status = 30:  4 ==>  4: 41 41 35 30
blogic_cmd(8D) Status = 30: 14 ==> 14: 45 DC 00 20 00 00 00 00 00 40 30 37 42 1D
scsi: ***** BusLogic SCSI Driver Version 2.1.17 of 12 September 2013 *****
scsi: Copyright 1995-1998 by Leonard N. Zubkoff <lnz@dandelion.com>
blogic_cmd(04) Status = 30:  4 ==>  4: 41 41 35 30
blogic_cmd(0B) Status = 30:  3 ==>  3: 00 08 07
blogic_cmd(0D) Status = 30: 34 ==> 34: 03 01 07 04 00 00 00 00 00 00 00 00 00 00 00 00 FF 42 44 46 FF 00 00 00 00 00 00 00 00 00 FF 00 FF 00
blogic_cmd(8D) Status = 30: 14 ==> 14: 45 DC 00 20 00 00 00 00 00 40 30 37 42 1D
blogic_cmd(84) Status = 30:  1 ==>  1: 37
blogic_cmd(8B) Status = 30:  5 ==>  5: 39 35 38 20 20
blogic_cmd(85) Status = 30:  1 ==>  1: 42
blogic_cmd(86) Status = 30:  4 ==>  4: FF 05 93 00
blogic_cmd(91) Status = 30: 64 ==> 64: 41 46 3E 20 39 35 38 20 20 00 C4 00 04 01 07 2F 07 04 35 FF FF FF FF FF FF FF FF FF FF 01 00 FE FF 08 FF FF 00 00 00 00 00 00 00 01 00 01 00 00 FF FF 00 00 00 00 00 00 00 00 00 00 00 00 00 FC
scsi0: Configuring BusLogic Model BT-958 PCI Wide Ultra SCSI Host Adapter

etc.

Link: https://lore.kernel.org/r/alpine.DEB.2.21.2104201940430.44318@angie.orcam.me.uk
Fixes: 4bcc595ccd80 ("printk: reinstate KERN_CONT for printing continuation lines")
Cc: stable@vger.kernel.org # v4.9+
Acked-by: Khalid Aziz <khalid@gonehiking.org>
Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg-lib: Fix commands without data transfer in bsg_transport_sg_io_fn()

Set ret to 0 after the initial permission checks to avoid leaking -EPERM
for commands without data transfer.

Link: https://lore.kernel.org/r/20210731074027.1185545-3-hch@lst.de
Fixes: 75ca56409e5b ("scsi: bsg: Move the whole request execution into the SCSI/transport handlers")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg: Fix commands without data transfer in scsi_bsg_sg_io_fn()

Set ret to 0 after the initial permission checks to avoid leaking -EPERM
for commands without data transfer.

Link: https://lore.kernel.org/r/20210731074027.1185545-2-hch@lst.de
Fixes: 75ca56409e5b ("scsi: bsg: Move the whole request execution into the SCSI/transport handlers")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Allow async suspend/resume callbacks

Allow UFS suspend/resume callbacks to run in parallel with other
suspend/resume callbacks. This can recoup dozens of milliseconds on the
resume path if UFS hardware needs to be powered back on.

Suspending and resuming asynchronously is safe to do so long as the driver
callbacks only depend on resources made available by either a) parent
devices or b) devices explicitly marked as suppliers with device_link_add.

Link: https://lore.kernel.org/r/20210728012743.1063928-1-paillon@google.com
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Bart Van Assche <bvanassche@acm.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Stanley Chu <stanley.chu@mediatek.com>
Cc: Can Guo <cang@codeaurora.org>
Cc: Asutosh Das <asutoshd@codeaurora.org>
Cc: Avri Altman <avri.altman@wdc.com>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Vincent Palomares <paillon@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Fix possible ABBA deadlock in nvmet_xri_aborted()

The lpfc_sli4_nvmet_xri_aborted() routine takes out the abts_buf_list_lock
and traverses the buffer contexts to match the xri. Upon match, it then
takes the context lock before potentially removing the context from the
associated buffer list. This violates the lock hierarchy used elsewhere in
the driver of locking context, then the abts_buf_list_lock - thus a
possible deadlock.

Resolve by: after matching, release the abts_buf_list_lock, then take the
context lock, and if to be deleted from the list, retake the
abts_buf_list_lock, maintaining lock hierarchy. This matches same list lock
hierarchy as elsewhere in the driver

Link: https://lore.kernel.org/r/20210730163309.25809-1-jsmart2021@gmail.com
Reported-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix spelling mistakes "allloc" -> "alloc"

There are two spelling mistakes with the same triple l in alloc, one in a
comment, the other in a ql_dbg() debug message. Fix them.

Link: https://lore.kernel.org/r/20210729082413.4761-1-colin.king@canonical.com
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg: Move the whole request execution into the SCSI/transport handlers

Remove the amount of indirect calls by making the handler responsible for
the entire execution of the request.

Link: https://lore.kernel.org/r/20210729064845.1044147-5-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: Remove the remaining SG_IO-related fields from struct request_queue

Move the sg_timeout and sg_reserved_size fields into the bsg_device and
scsi_device structures as they have nothing to do with generic block I/O.
Note that these values are now separate for bsg vs. SCSI device node
access, but that just matches how /dev/sg vs the other nodes has always
behaved.

Link: https://lore.kernel.org/r/20210729064845.1044147-4-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: Remove BLK_SCSI_MAX_CMDS

This was used for the table based SCSI passthough permission checking that
is gone now.

Link: https://lore.kernel.org/r/20210729064845.1044147-3-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg: Simplify device registration

Use the per-device cdev_device_interface to store the bsg data in the char
device inode, and thus remove the need to embedd the bsg_class_device
structure in the request_queue.

Link: https://lore.kernel.org/r/20210729064845.1044147-2-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sr: cdrom: Move cdrom_read_cdda_bpc() into the sr driver

cdrom_read_cdda_bpc() relies on sending SCSI command to the low level
driver using a REQ_OP_SCSI_IN request.  This isn't generic block layer
functionality, so move the actual low-level code into the sr driver and
call it through a new read_cdda_bpc method in the cdrom_device_ops
structure.

With this the CDROM code does not have to pull in scsi_normalize_sense()
and depend on CONFIG_SCSI_COMMON.

Link: https://lore.kernel.org/r/20210730072752.GB23847%40lst.de
Tested-by: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Unexport sg_scsi_ioctl()

Just call scsi_ioctl() in sg as that has the same effect.

Link: https://lore.kernel.org/r/20210724072033.1284840-25-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Factor SG_IO handling into a helper

Split the SG_IO handler from the main scsi_ioctl() routine.

Link: https://lore.kernel.org/r/20210724072033.1284840-24-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Factor SCSI_IOCTL_GET_IDLUN handling into a helper

Split the SCSI_IOCTL_GET_IDLUN handler from the main scsi_ioctl() routine.

Link: https://lore.kernel.org/r/20210724072033.1284840-23-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Consolidate the START STOP UNIT handling

Factor out a helper for the various flavors of START STOP UNIT command
ioctls.

Link: https://lore.kernel.org/r/20210724072033.1284840-22-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Remove a very misleading comment

Remove the comment above ioctl_internal_command() which doesn't document
this function at all.

Link: https://lore.kernel.org/r/20210724072033.1284840-21-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Rename CONFIG_BLK_SCSI_REQUEST to CONFIG_SCSI_COMMON

CONFIG_BLK_SCSI_REQUEST is rather misnamed as it enables building a small
amount of code shared by the SCSI initiator, target, and consumers of the
scsi_request passthrough API.  Rename it and also allow building it as a
module.

[mkp: add module license]

Link: https://lore.kernel.org/r/20210724072033.1284840-20-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Move the "block layer" SCSI ioctl handling to drivers/scsi

Merge the ioctl handling in block/scsi_ioctl.c into its only caller in
drivers/scsi/scsi_ioctl.c.

Link: https://lore.kernel.org/r/20210724072033.1284840-19-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Simplify SCSI passthrough permission checking

Remove the separate command filter structure and just use a switch
statement (which also cought two duplicate commands), return a bool and
give the function a sensible name.

Link: https://lore.kernel.org/r/20210724072033.1284840-18-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Move scsi_command_size_tbl to scsi_common.c

Move the SCSI command size table to common SCSI code.

Link: https://lore.kernel.org/r/20210724072033.1284840-17-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Remove scsi_req_init()

Merge scsi_req_init() into its only caller.

Link: https://lore.kernel.org/r/20210724072033.1284840-16-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg: Move bsg_scsi_ops to drivers/scsi/

Move the SCSI-specific bsg code in the SCSI midlayer instead of in the
common bsg code.  This just keeps the common bsg code block/ and also
allows building it as a module.

Link: https://lore.kernel.org/r/20210724072033.1284840-15-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg: Decouple from scsi_cmd_ioctl()

Decouple bsg from scsi_cmd_ioctl().  This requires a small amount of code
duplication, but will allow moving all SCSI ioctl handling into SCSI
midlayer.

Link: https://lore.kernel.org/r/20210724072033.1284840-14-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: block: Add a queue_max_bytes() helper

Return the max_sectors value in bytes.  Lifted from scsi_ioctl.c.

Link: https://lore.kernel.org/r/20210724072033.1284840-13-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Call scsi_cmd_ioctl() from scsi_ioctl()

Ensure SCSI ULD only has to call a single ioctl helper.  This also adds a
bunch of missing ioctls to the ch driver, and removes the need for a
duplicate implementation of SCSI_IOCTL_SEND_COMMAND command.

Link: https://lore.kernel.org/r/20210724072033.1284840-12-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Remove scsi_verify_blk_ioctl()

Manually verify that the device is not a partition and the caller has admin
privіleges at the beginning of the sr ioctl method and open code the
trivial check for sd as well.

Link: https://lore.kernel.org/r/20210724072033.1284840-11-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_ioctl: Remove scsi_cmd_blk_ioctl()

Open code scsi_cmd_blk_ioctl() in its two callers.

Link: https://lore.kernel.org/r/20210724072033.1284840-10-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cdrom: Remove the call to scsi_cmd_blk_ioctl() from cdrom_ioctl()

Only the sr driver can handle SCSI passthrough requests, so move the call
to scsi_cmd_blk_ioctl() there.

Link: https://lore.kernel.org/r/20210724072033.1284840-9-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: st: Simplify ioctl handling

Merge st_ioctl_common() into st_ioctl() and streamline the invocation of
the common ioctl helpers.

Link: https://lore.kernel.org/r/20210724072033.1284840-8-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Remove scsi_compat_ioctl()

Just handle the compat case in scsi_ioctl() using in_compat_syscall().

Link: https://lore.kernel.org/r/20210724072033.1284840-7-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sg: Consolidate compat ioctl handling

Merge the native and compat ioctl handlers into a single one using
in_compat_syscall().

Link: https://lore.kernel.org/r/20210724072033.1284840-6-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ch: Consolidate compat ioctl handling

Merge the native and compat ioctl handlers into a single one using
in_compat_syscall().

Link: https://lore.kernel.org/r/20210724072033.1284840-5-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Consolidate compat ioctl handling

Merge the native and compat ioctl handlers into a single one using
in_compat_syscall(), and also simplify the calling conventions by merging
sd_ioctl_common() into sd_ioctl().

Link: https://lore.kernel.org/r/20210724072033.1284840-4-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sr: Consolidate compat ioctl handling

Merge the native and compat ioctl handlers into a single one using
in_compat_syscall().

Link: https://lore.kernel.org/r/20210724072033.1284840-3-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: bsg: Remove support for SCSI_IOCTL_SEND_COMMAND

SCSI_IOCTL_SEND_COMMAND has been deprecated longer than bsg exists and has
been warning for just as long.  More importantly it harcodes SCSI CDBs and
thus will do the wrong thing on non-SCSI bsg nodes.

Link: https://lore.kernel.org/r/20210724072033.1284840-2-hch@lst.de
Fixes: aa387cc89567 ("block: add bsg helper library")
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Update version to 10.02.00.107-k

Link: https://lore.kernel.org/r/20210624052606.21613-12-njavali@marvell.com
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Increment command and completion counts

Increment the command and the completion counts.

Link: https://lore.kernel.org/r/20210624052606.21613-11-njavali@marvell.com
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add encryption to I/O path

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

After the completion of PLOGI, both sides have authenticated and PRLI
completed, encrypted I/Os are allowed to proceed.

 - Use new firmware API to encrypt traffic on the wire

 - Add driver parameter to enable|disable EDIF feature

   # modprobe qla2xxx ql2xsecenable=1

Link: https://lore.kernel.org/r/20210624052606.21613-10-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add doorbell notification for app

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

During runtime, driver and authentication application need to stay in sync
in terms of: Session being down|up, arrival of new authentication
message (AUTH ELS) and SADB update completion.

These events are queued up as doorbell to the authentication
application. Application would read this doorbell on regular basis to stay
up to date. Each SCSI host would have a separate doorbell queue.

The doorbell interface can daisy chain a list of events for each read. Each
event contains an event code + hint to help application steer the next
course of action.

Link: https://lore.kernel.org/r/20210624052606.21613-9-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add detection of secure device

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

There is no FC switch scan service that can indicate whether a device is
secure or non-secure.

In order to detect whether the remote port supports encrypted operation,
driver must first do a PLOGI with the remote device. On completion of the
PLOGI, driver will query firmware to see if the device supports secure
login. To do that, driver + firmware must advertise the security bit via
PLOGI's service parameter. The remote device shall respond using the same
service parameter whether it supports it or not.

Link: https://lore.kernel.org/r/20210624052606.21613-8-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add authentication pass + fail bsgs

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

On completion of the authentication process, the authentication application
will notify driver on whether it is successful or not.

In case of success, application will use the QL_VND_SC_AUTH_OK BSG call to
tell driver to proceed to the PRLI phase.

In case of failure, application will use the QL_VND_SC_AUTH_FAIL bsg call
to tell driver to tear down the connection and retry. In the case where an
existing session is active, the re-key process can fail. The session tear
down ensures data is not further compromised.

Link: https://lore.kernel.org/r/20210624052606.21613-7-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add key update

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

As part of the authentication process, the authentication application will
generate a SADB entry (Security Association/SA, key, SPI value, etc). This
SADB is then passed to driver to be programmed into hardware. There will be
a pair of SADB's (Tx and Rx) for each connection.

After some period, the application can choose to change the key. At that
time, a new set of SADB pair is given to driver. The old set of SADB will
be deleted.

Add a new bsg call (QL_VND_SC_SA_UPDATE) to allow application to allow
adding or deleting SADB entries.  Driver will not keep the key in
memory. It will pass it to HW.

It is assumed that application will assign a unique SPI value to this SADB
(SA + key). Driver + hardware will assign a handle to track this unique
SPI/SADB.

Link: https://lore.kernel.org/r/20210624052606.21613-6-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add extraction of auth_els from the wire

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

Once authentication messages sent from a remote device have arrived, each
message is extracted and placed in a buffer for application to retrieve.
The FC frame header will be stripped, leaving behind the AUTH ELS payload.
It is up to the application to strip the AUTH ELS header to get to the
actual authentication message.

Link: https://lore.kernel.org/r/20210624052606.21613-5-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add send, receive, and accept for auth_els

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

Add the ability for authentication application to send and retrieve
messages as part of the authentication process via existing
FC_BSG_HST_ELS_NOLOGIN BSG interface.

To send a message, application is expected to format the data in the AUTH
ELS format. Refer to FC-SP2 for details.

If a message was received, application is required to reply with either a
LS_ACC or LS_RJT complete the exchange using the same interface. Otherwise,
remote device will treat it as a timeout.

Link: https://lore.kernel.org/r/20210624052606.21613-4-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add getfcinfo and statistic bsgs

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

Add two new BSG calls:

 - QL_VND_SC_GET_FCINFO: Application can from time to time request a list
   of all FC ports or a single device that supports secure connection.  If
   driver sees a new or old device has logged into the switch, this call is
   used to check for the WWPN.

 - QL_VND_SC_GET_STATS: Application request for various statistics for each
   FC port.

Link: https://lore.kernel.org/r/20210624052606.21613-3-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: edif: Add start + stop bsgs

Some FC adapters from Marvell offer the ability to encrypt data in flight
(EDIF). This feature requires an application to act as an authenticator.

Add two new BSG calls:

 - QL_VND_SC_APP_START: Application will announce its presence to driver
   with this call. Driver will restart all connections to see if remote
   device supports security or not.

 - QL_VND_SC_APP_STOP: Application announces it is in the process of
   exiting. Driver will restart all connections to revert back to
   non-secure. Provided the remote device is willing to allow a non-secure
   connection.

Link: https://lore.kernel.org/r/20210624052606.21613-2-njavali@marvell.com
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Co-developed-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Signed-off-by: Larry Wisneski <Larry.Wisneski@marvell.com>
Co-developed-by: Duane Grigsby <duane.grigsby@marvell.com>
Signed-off-by: Duane Grigsby <duane.grigsby@marvell.com>
Co-developed-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Rick Hicksted Jr <rhicksted@marvell.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove unused variable 'status'

Fix the clang build warning:

  drivers/scsi/qla2xxx/qla_nx.c:2209:6: error: variable 'status' set but not used [-Werror,-Wunused-but-set-variable]
        int status = 0;

Link: https://lore.kernel.org/r/20210726201924.3202278-4-morbo@google.com
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Bill Wendling <morbo@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: Drop BLK_DEV_BSGLIB selection

SCSI_SAS_ATTRS already selects BLK_DEV_BSGLIB in drivers/scsi/Kconfig.
Remove selection in libsas/Kconfig.

Link: https://lore.kernel.org/r/20210723084624.2596297-1-guoqing.jiang@linux.dev
Signed-off-by: Guoqing Jiang <jiangguoqing@kylinos.cn>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target: Remove redundant assignment to variable ret

The variable ret is being initialized with a value that is never read, the
assignment is redundant and can be removed.

Link: https://lore.kernel.org/r/20210721101519.42299-1-colin.king@canonical.com
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Addresses-Coverity: ("Unused value")

scsi: lpfc: Remove redundant assignment to pointer pcmd

The pointer pcmd is being initialized with a value that is never read, the
assignment is redundant and can be removed.

Link: https://lore.kernel.org/r/20210721095350.41564-1-colin.king@canonical.com
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Addresses-Coverity: ("Unused value")

scsi: lpfc: Copyright updates for 14.0.0.0 patches

Update copyrights to 2021 for files modified in the 14.0.0.0 patch set.

Link: https://lore.kernel.org/r/20210722221721.74388-7-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Update lpfc version to 14.0.0.0

Update lpfc version to 14.0.0.0.

Link: https://lore.kernel.org/r/20210722221721.74388-6-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Add 256 Gb link speed support

Update routines to support 256 Gb link speed for LPe37000/LPe38000
adapters. 256 Gb speeds can be seen on trunk links.

Link: https://lore.kernel.org/r/20210722221721.74388-5-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Revise Topology and RAS support checks for new adapters

Support for Topology and RAS logging capabilities were qualified by PCIe
device ID checks necessitating additional driver changes for new device
IDs.

Reduce reliance on specific PCIe device IDs by substituting checks for SLI
family information. This automatically picks up support on the newest
hardware.

Link: https://lore.kernel.org/r/20210722221721.74388-4-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Fix cq_id truncation in rq create

On the newer hardware, CQ_ID values can be larger than seen on previous
generations. This exposed an issue in the driver where its definition of
cq_id in the RQ Create mailbox cmd was too small, thus the cq_id was
truncated, causing the command to fail.

Revise the RQ_CREATE CQ_ID field to its proper size (16 bits).

Link: https://lore.kernel.org/r/20210722221721.74388-3-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Add PCI ID support for LPe37000/LPe38000 series adapters

Update supported pci_device_id table to include the values for the G7+ ASIC
Device ID utilized by LPe37xxx and LPe38xxx series of adapters.  The
default reporting string will be "LPe38000".

Link: https://lore.kernel.org/r/20210722221721.74388-2-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Add BLIST_IGN_MEDIA_CHANGE for Ultra HS-SD/MMC USB card readers

Ultra HS-SD/MMC card reader devices establish a MEDIUM MAY HAVE CHANGED
unit attention not only when the medium changes but also when resuming from
suspend.

Setting the BLIST_IGN_MEDIA_CHANGE flag permits using runtime PM for these
readers.

[mkp: renamed flag]

Link: https://lore.kernel.org/r/20210704075403.147114-4-martin.kepplinger@puri.sm
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: REQUEST SENSE for BLIST_IGN_MEDIA_CHANGE devices in runtime_resume()

For SD card reader devices that have the BLIST_IGN_MEDIA_CHANGE flag
set, a MEDIUM MAY HAVE CHANGED unit attention is established after
resuming from runtime suspend. Send a REQUEST SENSE to consume the UA.

The "downside" is that for these devices we now rely on users to not
change the medium (SD card) *during* a runtime suspend/resume cycle,
i.e. when not unmounting.

To enable runtime PM for an SD cardreader (device number 0:0:0:0), do:

echo 0 > /sys/module/block/parameters/events_dfl_poll_msecs
echo 1000 > /sys/bus/scsi/devices/0:0:0:0/power/autosuspend_delay_ms
echo auto > /sys/bus/scsi/devices/0:0:0:0/power/control

[mkp: use scsi_device flag instead of poking at BLIST]

Link: https://lore.kernel.org/r/20210704075403.147114-3-martin.kepplinger@puri.sm
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: Add new flag BLIST_IGN_MEDIA_CHANGE

Add a new flag for devices that erroneously establish MEDIUM MAY HAVE
CHANGED unit attentions. Drivers can set this flag to make the SCSI
layer ignore media change events during resume.

[mkp: add "ignore" and add corresponding flag to struct scsi_device]

Link: https://lore.kernel.org/r/20210704075403.147114-2-martin.kepplinger@puri.sm
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: Allow libsas to include SCSI header files directly

libsas needs to include some header files in the scsi directory. However
these are currently hardcoded with the path "../" in the C files. Do this
in the Makefile to avoid hardcoding the path.

Link: https://lore.kernel.org/r/20210716074551.771312-1-yanaijie@huawei.com
Cc: John Garry <john.garry@huawei.com>
Reviewed-by: John Garry <john.garry@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: core: Use scsi_get_lba() to get LBA

Use the scsi_get_lba() helper instead of a function internal to the
SCSI disk driver. Remove #include "sd.h".

Link: https://lore.kernel.org/r/20210609033929.3815-16-martin.petersen@oracle.com
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-16-martin.petersen@oracle.com>

scsi: core: Make scsi_get_lba() return the LBA

scsi_get_lba() confusingly returned the block layer sector number expressed
in units of 512 bytes. Now that we have a more aptly named
scsi_get_sector() function, make scsi_get_lba() return the actual LBA.

Link: https://lore.kernel.org/r/20210609033929.3815-13-martin.petersen@oracle.com
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-13-martin.petersen@oracle.com>

scsi: iser: Use scsi_get_sector() instead of scsi_get_lba()

Use scsi_get_sector() instead of scsi_get_lba() since the name of the
latter is confusing. This patch does not change any functionality.

Link: https://lore.kernel.org/r/20210513223757.3938-3-bvanassche@acm.org
Link: https://lore.kernel.org/r/20210609033929.3815-12-martin.petersen@oracle.com
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-12-martin.petersen@oracle.com>

scsi: core: Introduce scsi_get_sector()

Since scsi_get_lba() returns a sector_t value instead of the LBA, the name
of that function is confusing. Introduce an identical function
scsi_get_sector().

Link: https://lore.kernel.org/r/20210513223757.3938-2-bvanassche@acm.org
Link: https://lore.kernel.org/r/20210609033929.3815-11-martin.petersen@oracle.com
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-11-martin.petersen@oracle.com>

scsi: scsi_debug: Improve RDPROTECT/WRPROTECT handling

It is useful for testing purposes to be able to inject errors by writing
bad protection information to media with checking disabled and then
attempting to read it back. Extend scsi_debug's PI verification logic to
give the driver feature parity with commercially available drives. Almost
all devices with PI capability support RDPROTECT and WRPROTECT values of 0,
1, and 3.

Link: https://lore.kernel.org/r/20210609033929.3815-10-martin.petersen@oracle.com
Reviewed-by: Douglas Gilbert <dgilbert@interlog.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-10-martin.petersen@oracle.com>

scsi: scsi_debug: Remove dump_sector()

The function used to dump sectors containing protection information errors
was useful during initial development over a decade ago.  However,
dump_sector() substantially slows down the system during testing due to
writing an entire sector's worth of data to syslog on every error.

We now log plenty of information about the nature of detected protection
information errors throughout the stack. Dumping the entire contents of an
offending sector is no longer needed.

Link: https://lore.kernel.org/r/20210609033929.3815-9-martin.petersen@oracle.com
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Douglas Gilbert <dgilbert@interlog.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-9-martin.petersen@oracle.com>

scsi: zfcp: Use the proper SCSI midlayer interfaces for PI

Use scsi_prot_ref_tag() and scsi_prot_interval() instead scsi_get_lba() and
sector_size.

Link: https://lore.kernel.org/r/20210609033929.3815-7-martin.petersen@oracle.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-7-martin.petersen@oracle.com>

scsi: qla2xxx: Use the proper SCSI midlayer interfaces for PI

Use the SCSI midlayer interfaces to query protection interval, reference
tag, and per-command DIX flags.

Link: https://lore.kernel.org/r/20210609033929.3815-4-martin.petersen@oracle.com
Reviewed-by: Arun Easi <aeasi@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-4-martin.petersen@oracle.com>

scsi: core: Add scsi_prot_ref_tag() helper

We are about to remove the request pointer from struct scsi_cmnd and that
will complicate getting to the ref_tag via t10_pi_ref_tag() in the various
drivers. Introduce a helper function to retrieve the reference tag so
drivers will not have to worry about the details.

Link: https://lore.kernel.org/r/20210609033929.3815-2-martin.petersen@oracle.com
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-2-martin.petersen@oracle.com>

scsi: smartpqi: Update version to 2.1.10-020

Link: https://lore.kernel.org/r/20210714182847.50360-10-don.brace@microchip.com
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Gerry Morong <gerry.morong@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Fix ISR accessing uninitialized data

Correct driver's ISR accessing a data structure member that has not been
fully initialized during driver initialization.

The pqi queue groups can have uninitialized members when an interrupt
fires. This has not resulted in any driver crashes. This was found during
our own internal testing. No bugs were ever filed.

Link: https://lore.kernel.org/r/20210714182847.50360-9-don.brace@microchip.com
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add PCI IDs for new ZTE controllers

Add support for ZTE RM241-18i 2G device ID:

    VID_9005, DID_028F, SVID_1CF2 and SDID_5445

Add support for ZTE RM242-18i 4G device ID:

    VID_9005, DID_028F, SVID_1CF2 and SDID_5446

Add support for ZTE RM243-18i device ID:

    VID_9005, DID_028F, SVID_1CF2 and SDID_5447

Add support for ZTE SDPSA/B-18i 4G device ID:

    VID_9005, DID_028F, SVID_1CF2 and SDID_0B27

Add support for ZTE SDPSA/B_I-18i device ID:

    VID_9005, DID_028F, SVID_1CF2 and SDID_0B29

Add support for ZTE SDPSA/B_L-18i 2G device ID:

    VID_9005, DID_028F, SVID_1CF2 and SDID_0B45

Link: https://lore.kernel.org/r/20210714182847.50360-8-don.brace@microchip.com
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Balsundar P <balsundar.p@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add PCI ID for new ntcom controller

Add support for Norsi ntcom Raid-24i controller:

    VID_0x9005, DID_0x028f, SVID_0x1dfc, SDID_0x3161

Link: https://lore.kernel.org/r/20210714182847.50360-7-don.brace@microchip.com
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Mike McGowen <mike.mcgowen@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add SCSI cmd info for resets

Report on SCSI command that has triggered the reset.  Also add check for
NULL SCSI commands resulting from issuing sg_reset when there is no
outstanding commands.

   Example:

   sg_reset -d /dev/sgXY
   smartpqi 0000:39:00.0: resetting scsi 4:0:1:0 due to cmd 0x12

Link: https://lore.kernel.org/r/20210714182847.50360-6-don.brace@microchip.com
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Change Kconfig menu entry to Microchip

Change Microsemi to Microchip.

Link: https://lore.kernel.org/r/20210714182847.50360-5-don.brace@microchip.com
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Gerry Morong <gerry.morong@microchip.com>
Reviewed-by: Justin Lindley <Justin.lindley@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Change driver module macros to Microchip

Change driver module macros to reflect copyright changes: Microsemi to
Microchip.

Link: https://lore.kernel.org/r/20210714182847.50360-4-don.brace@microchip.com
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Gerry Morong <gerry.morong@microchip.com>
Reviewed-by: Justin Lindley <Justin.lindley@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Update copyright notices

Updated copyright notices and company name strings to reflect Microchip
ownership.

Link: https://lore.kernel.org/r/20210714182847.50360-3-don.brace@microchip.com
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: smartpqi: Add PCI IDs for H3C P4408 controllers

Add support for H3C P4408-Ma-8i-2GB device ID:

    VID_9005, DID_028F, SVID_193D and SDID_1108
    VID_9005, DID_028F, SVID_193D and SDID_1109

Link: https://lore.kernel.org/r/20210714182847.50360-2-don.brace@microchip.com
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Reviewed-by: Mike McGowen <mike.mcgowen@microchip.com>
Reviewed-by: Murthy Bhat <Murthy.Bhat@microchip.com>
Reviewed-by: Scott Benesh <scott.benesh@microchip.com>
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Signed-off-by: Mahesh Rajashekhara <mahesh.rajashekhara@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Copyright updates for 12.8.0.11 patches

Update copyrights for files modified by the 12.8.0.11 patch set.

Link: https://lore.kernel.org/r/20210707184351.67872-21-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Update lpfc version to 12.8.0.11

Update lpfc version to 12.8.0.11.

Link: https://lore.kernel.org/r/20210707184351.67872-20-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Skip issuing ADISC when node is in NPR state

When a node moves to NPR state due to a device recovery event, the
nlp_fc4_types in the node are cleared. An ADISC received for a node in the
NPR state triggers an ADISC. Without fc4 types being known, the calls to
register with the transport are no-op'd, thus no additional references are
placed on the node by transport re-registrations. A subsequent RSCN could
trigger another unregister request, which will decrement the reference
counts, leading to the ref count hitting zero and the node being freed
while futher discovery on the node is being attempted by the RSCN event
handling.

Fix by skipping the trigger of an ADISC when in NPR state. The normal ADISC
process will kick off in the regular discovery path after receiving a
response from name server.

Link: https://lore.kernel.org/r/20210707184351.67872-19-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Skip reg_vpi when link is down for SLI3 in ADISC cmpl path

During RSCN storms, some instances of LIP on SLI-3 adapters lead to a
situation where FLOGIs keep failing with firmware indicating an illegal
command error code.  This situation was preceded by an ADISC completion
that was processed while the link was down. This path on SLI-3 performs a
CLEAR_LA and attempts to activate a VPI with REG_VPI.  Later, as the FLOGI
completes, it's no longer in sync with the VPI state.  In SLI-3 it is
illegal to have an active VPI during FLOGI.

Resolve by circumventing the SLI-3 path that performs the CLEAR_LA and
REG_VPI. The path will be taken after the FLOGI after the next Link Up.

Link: https://lore.kernel.org/r/20210707184351.67872-18-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Call discovery state machine when handling PLOGI/ADISC completions

In the PLOGI and ADISC completion handling, the device removal event could
be skipped during some link errors. This could leave a stale node in UNUSED
state.  Driver unload would hang for a long time waiting for this node to
be freed.

Resolve by taking the following steps:

 - Always post ADISC completion events to discovery state machine upon
   ADISC completion.

 - In case of a completion error for PLOGI/ADISC, ensure that init refcount
   is dropped if not registered with transport.

Link: https://lore.kernel.org/r/20210707184351.67872-17-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Delay unregistering from transport until GIDFT or ADISC completes

On an RSCN event, the nodes specified in RSCN payload and in MAPPED state
are moved to NPR state in order to revalidate the login. This triggers an
immediate unregister from SCSI/NVMe backend. The assumption is that the
node may be missing. The re-registration with the backend happens after
either relogin (PLOGI/PRLI; if ADISC is disabled or login truly lost) or
when ADISC completes successfully (rediscover with ADISC enabled).

However, the NVMe-FC standard provides for an RSCN to be triggered when
the remote port supports a discovery controller and there was a change
of discovery log content. As the remote port typically also supports
storage subsystems, this unregister causes all storage controller
connections to fail and require reconnect.

Correct by reworking the code to ensure that the unregistration only occurs
when a login state is truly terminated, thereby leaving the NVMe storage
controllers in place.

The changes made are:

 - Retain node state in ADISC_ISSUE when scheduling ADISC ELS retry.

 - Do not clear wwpn/wwnn values upon ADISC failure.

 - Move MAPPED nodes to NPR during RSCN processing, but do not unregister
   with transport.  On GIDFT completion, identify missing nodes (not marked
   NLP_NPR_2B_DISC) and unregister them.

 - Perform unregistration for nodes that will go through ADISC processing
   if ADISC completion fails.

 - Successful ADISC completion will move node back to MAPPED state.

Link: https://lore.kernel.org/r/20210707184351.67872-16-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Enable adisc discovery after RSCN by default

Assign a default value of 1 to driver module parameter lpfc_use_adisc.

Link: https://lore.kernel.org/r/20210707184351.67872-15-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>