scsi: qla2xxx: Make qla2x00_mem_free() easier to verify

Instead of clearing all freed pointers at the end of qla2x00_mem_free(),
clear freed pointers immediately after having freed the memory these
pointers point at.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Increase the size of the mailbox arrays from 4 to 8

This patch avoids that Coverity complains that qla2x00_async_event() writes
outside the bounds of the mb[] arrays (MBA_IDC_AEN case).

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Log the status code if a firmware command fails

It is important to know why a firmware command failed. Hence log 'rval'
together with the values of the mailbox registers if a firwmare command
fails.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Avoid that Coverity complains about dereferencing a NULL rport pointer

Since Coverity cannot know that rport != NULL in qla2xxx_queuecommand() and
since there is code in that function that dereferences the rport pointer,
modify qla2xxx_queuecommand() such that it fails SCSI commands if rport ==
NULL.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove the fcport test from qla_nvme_abort_work()

Testing whether a pointer is not NULL after it has been dereferenced is not
useful. Hence remove the if (fcport) test. This was detected by Coverity.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Uninline qla2x00_init_timer()

Since qla2x00_init_timer() is not used for I/O commands there is no need to
inline this function. Hence uninline this function.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Move qla2x00_is_reserved_id() from qla_inline.h into qla_init.c

The previous patch moved all qla2x00_is_reserved_id() callers into
qla_init.c. Hence also move the qla2x00_is_reserved_id() definition into
qla_init.c.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Move qla2x00_clear_loop_id() from qla_inline.h into qla_init.c

Since qla2x00_clear_loop_id() is not in the hot path, uninline it.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove a set-but-not-used variable

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Declare qla2x00_find_new_loop_id() static

Since all qla2x00_find_new_loop_id() calls occur in the same source file as
the definition of this function, move that function to just before its
first caller and declare it static. Convert the header above this function
into kernel-doc format.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Move qla2x00_set_reserved_loop_ids() definition

Since qla2x00_set_reserved_loop_ids() only has a single caller, move it
into the source file from where it is called.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix a format specifier

Since mcmd->sess->port_name is eight bytes long, use %8phC to format that
port name instead of %phC.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Fixes: 726b85487067 ("qla2xxx: Add framework for async fabric discovery") # v4.11.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Update two source code comments

Change one occurrence of "*(" into "()" and change one occurrence of
"lcoate" into "locate". Fix the reference to qla_tgt_handle_cmd_for_atio():
there has never been a function with that name.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Fixes: 75f8c1f693ee ("[SCSI] tcm_qla2xxx: Add >= 24xx series fabric module for target-core") # v3.5.
Fixes: 2d70c103fd2a ("[SCSI] qla2xxx: Add LLD target-mode infrastructure for >= 24xx series") # v3.5.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Revert "Rely on the driver core for asynchronous probing"

Hibernation hangs as follows due to commit 21e6ba3f0e02 when using SATA:

Call Trace:
 __schedule+0x464/0xe70
 schedule+0x4e/0xd0
 blk_queue_enter+0x5fe/0x7e0
 generic_make_request+0x313/0x950
 submit_bio+0x9b/0x250
 submit_bio_wait+0xc9/0x110
 hib_submit_io+0x17d/0x1c0
 write_page+0x61/0xa0
 swap_write_page+0x4b/0x1f0
 swsusp_write+0x2f9/0x3d0
 hibernate.cold.10+0x108/0x231
 state_store+0xf7/0x100
 kobj_attr_store+0x37/0x50
 sysfs_kf_write+0x87/0xa0
 kernfs_fop_write+0x186/0x240
 __vfs_write+0x4d/0x90
 vfs_write+0xfa/0x260
 ksys_write+0xb9/0x1a0
 __x64_sys_write+0x43/0x50
 do_syscall_64+0x71/0x210
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Hence revert commit 21e6ba3f0e02.

Cc: Pavel Machek <pavel@ucw.cz>
Reported-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: sd: Revert "Inline sd_probe_part2()"

Reverts commit d16ece577bf2 to make a clean revert of its predecessor
possible.

Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Fix RX_TERMINATION_FORCE_ENABLE define value

Fix RX_TERMINATION_FORCE_ENABLE define value from 0x0089 to 0x00A9
according to MIPI Alliance MPHY specification.

Fixes: e785060ea3a1 ("ufs: definitions for phy interface")
Signed-off-by: Pedro Sousa <sousa@synopsys.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: remove set but not used variables

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

drivers/scsi/qedf/qedf_els.c: In function 'qedf_process_els_compl':
drivers/scsi/qedf/qedf_els.c:149:20: warning: variable 'sc_cmd' set but not used [-Wunused-but-set-variable]
drivers/scsi/qedf/qedf_els.c:148:28: warning: variable 'task_ctx' set but not used [-Wunused-but-set-variable]
drivers/scsi/qedf/qedf_els.c: In function 'qedf_send_srr':
drivers/scsi/qedf/qedf_els.c:612:6: warning: variable 'sid' set but not used [-Wunused-but-set-variable]

They are never used since introduction.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Update the driver version to 8.37.25.20

Update the driver version to 8.37.25.20.

Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Add return value to log message if scsi_add_host fails

Print return value of scsi_add_host on failure.

Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Print fcport information on wait for upload timeout

Log fcport for which upload failed.

Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Check the return value of start_xmit

Log the reason for start xmit failure.

Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Log message if scsi_add_host fails

Print message on scsi_add_host failure.

Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Check for fcoe_libfc_config failure

Print the fcoe_libfc_config failure and return proper failure.

Signed-off-by: Saurav Kashyap <saurav.kashyap@cavium.com>
Signed-off-by: Chad Dupuis <chad.dupuis@cavium.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Add comment to display logging levels

Comment will help in decoding the logging level.

Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Add port_id for fcport into initiate_cleanup debug message

Port ID will help in debugging.

Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Add LBA to underrun debug messages

Print LBA information for underrun cases.

Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Print scsi_cmd backpointer in good completion path if the command is still being used

Printing scsi command pointer will help in crash dump analysis.

Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Add driver state to 'driver_stats' debugfs node

Add debugfs node for driver stats.

[mkp: typo]

Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: Change MSI-X load error message

Change the message to display load failure.

Signed-off-by: Chad Dupuis <cdupuis@marvell.com>
Signed-off-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Adjust termination and offload ramrod timers

Whenever offload ramrod is issued, firmware wants driver to wait for max 5
secs, otherwise driver can initiate further corrective action. Similarly,
when termination ramrod is issued, irrespective of abortive or non-abortive
termination, driver should wait for 60 sec * max TCP-RT timeout.

[mkp: typos]

Signed-off-by: Manish Rangankar <mrangankar@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedi: Abort ep termination if offload not scheduled

Sometimes during connection recovery when there is a failure to resolve
ARP, and offload connection was not issued, driver tries to flush pending
offload connection work which was not queued up.

kernel: WARNING: CPU: 19 PID: 10110 at kernel/workqueue.c:3030 __flush_work.isra.34+0x19c/0x1b0
kernel: CPU: 19 PID: 10110 Comm: iscsid Tainted: G W 5.1.0-rc4 #11
kernel: Hardware name: Dell Inc. PowerEdge R730/0599V5, BIOS 2.9.1 12/04/2018
kernel: RIP: 0010:__flush_work.isra.34+0x19c/0x1b0
kernel: Code: 8b fb 66 0f 1f 44 00 00 31 c0 eb ab 48 89 ef c6 07 00 0f 1f 40 00 fb 66 0f 1f 44 00 00 31 c0 eb 96 e8 08 16 fe ff 0f 0b eb 8d <0f> 0b 31 c0 eb 87 0f 1f 40 00 66 2e 0f 1
f 84 00 00 00 00 00 0f 1f
kernel: RSP: 0018:ffffa6b4054dba68 EFLAGS: 00010246
kernel: RAX: 0000000000000000 RBX: ffff91df21c36fc0 RCX: 0000000000000000
kernel: RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff91df21c36fc0
kernel: RBP: ffff91df21c36ef0 R08: 0000000000000000 R09: 0000000000000000
kernel: R10: 0000000000000038 R11: ffffa6b4054dbd60 R12: ffffffffc05e72c0
kernel: R13: ffff91db10280820 R14: 0000000000000048 R15: 0000000000000000
kernel: FS:  00007f5d83cc1740(0000) GS:ffff91df2f840000(0000) knlGS:0000000000000000
kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
kernel: CR2: 0000000001cc5000 CR3: 0000000465450002 CR4: 00000000001606e0
kernel: Call Trace:
kernel: ? try_to_del_timer_sync+0x4d/0x80
kernel: qedi_ep_disconnect+0x3b/0x410 [qedi]
kernel: ? 0xffffffffc083c000
kernel: ? klist_iter_exit+0x14/0x20
kernel: ? class_find_device+0x93/0xf0
kernel: iscsi_if_ep_disconnect.isra.18+0x58/0x70 [scsi_transport_iscsi]
kernel: iscsi_if_recv_msg+0x10e2/0x1510 [scsi_transport_iscsi]
kernel: ? copyout+0x22/0x30
kernel: ? _copy_to_iter+0xa0/0x430
kernel: ? _cond_resched+0x15/0x30
kernel: ? __kmalloc_node_track_caller+0x1f9/0x270
kernel: iscsi_if_rx+0xa5/0x1e0 [scsi_transport_iscsi]
kernel: netlink_unicast+0x17f/0x230
kernel: netlink_sendmsg+0x2d2/0x3d0
kernel: sock_sendmsg+0x36/0x50
kernel: ___sys_sendmsg+0x280/0x2a0
kernel: ? timerqueue_add+0x54/0x80
kernel: ? enqueue_hrtimer+0x38/0x90
kernel: ? hrtimer_start_range_ns+0x19f/0x2c0
kernel: __sys_sendmsg+0x58/0xa0
kernel: do_syscall_64+0x5b/0x180
kernel: entry_SYSCALL_64_after_hwframe+0x44/0xa9

Signed-off-by: Manish Rangankar <mrangankar@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: csiostor: create per port irq affinity mask set

csiostor driver allocates per port num_online_cpus() irq vectors, so create
per-port irq affinity mask set to spread irq vectors evenly.

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

scsi: qla2xxx: Silence Successful ELS IOCB message

ELS IOCB done message should be moved to verbose logging to prevent
confusion about the error case v/s successful submission case.

[mkp: typos]

Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix device staying in blocked state

This patch fixes issue reported by some of the customers, who discovered
that after cable pull scenario the devices disappear and path seems to
remain in blocked state. Once the device reappears, driver does not seem to
update path to online. This issue appears because of the defer flag
creating race condition where the same session reappears.  This patch fixes
this issue by indicating SCSI-ML of device lost when
qlt_free_session_done() is called from qlt_unreg_sess().

Fixes: 41dc529a4602a ("qla2xxx: Improve RSCN handling in driver")
Signed-off-by: Quinn Tran <qtran@marvell.com>
Cc: stable@vger.kernel.org #4.19
Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: aic7xxx: improve the Kconfig entry

There is no old vs new driver anymore.

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

scsi: aic7xxx: fix spelling mistake "recevied" -> "received"

There is a spelling mistake in a kernel message, fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: lpfc: Make lpfc_sli4_oas_verify static

Fix sparse warning:

drivers/scsi/lpfc/lpfc_init.c:13091:1: warning:
 symbol 'lpfc_sli4_oas_verify' was not declared. Should it be static?

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: fix spelling mistake "oustanding" -> "outstanding"

There are a couple of spelling mistakes in some kernel info and notice
messages. Fix these.

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

scsi: qedi: fix spelling mistake "oflload" -> "offload"

There are a couple of spelling mistakes in some kernel alert messages.  Fix
these.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Manish Rangankar <mrangankar@marvell.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qedf: remove memset/memcpy to nfunc and use func instead

Currently the qedf_dbg_* family of functions can overrun the end of the
source string if it is less than the destination buffer length because of
the use of a fixed sized memcpy. Remove the memset/memcpy calls to nfunc
and just use func instead as it is always a null terminated string.

Addresses-Coverity: ("Out-of-bounds access")
Fixes: 61d8658b4a43 ("scsi: qedf: Add QLogic FastLinQ offload FCoE driver framework.")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Saurav Kashyap <skashyap@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: core: map PQ=1, PDT=other values to SCSI_SCAN_TARGET_PRESENT

commit 84961f28e9d1 ("[SCSI] Don't add scsi_device for devices that return
PQ=1, PDT=0x1f") returns SCSI_SCAN_TARGET_PRESENT if inquiry returns PQ=1,
and PDT = 0x1f. However, from the scsi spec, it seemed setting PQ=1, and
PDT to the type it is capable to support, can also mean the device is not
connected. E.g. we see an IBM/2145 returns PQ=1 and PDT=0 for a non-mapped
lun (details attached at the end).

This patch changes the check condition a bit, so the check don't require
PTD to be 0x1f when PQ=1.

$ echo 0 0 1 > /sys/class/scsi_host/host1/scan
[ 2483.722186] scsi 1:0:0:1: scsi scan: INQUIRY pass 1 length 36
[ 2483.725687] scsi 1:0:0:1: scsi scan: INQUIRY successful with code 0x0
[ 2483.729171] scsi 1:0:0:1: scsi scan: INQUIRY pass 2 length 109
[ 2483.732481] scsi 1:0:0:1: scsi scan: INQUIRY successful with code 0x0
[ 2483.735911] scsi 1:0:0:1: Direct-Access     IBM      2145             0000 PQ: 1 ANSI: 6
[ 2483.741282] scsi 1:0:0:1: Attached scsi generic sg2 type 0

$ tail /proc/scsi/scsi
Attached devices:
Host: scsi1 Channel: 00 Id: 00 Lun: 00
  Vendor: IBM      Model: 2145             Rev: 0000
  Type:   Direct-Access                    ANSI  SCSI revision: 06
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: IBM      Model: 2145             Rev: 0000
  Type:   Direct-Access                    ANSI  SCSI revision: 06
Host: scsi1 Channel: 00 Id: 00 Lun: 01
  Vendor: IBM      Model: 2145             Rev: 0000
  Type:   Direct-Access                    ANSI  SCSI revision: 06

$ lsscsi
[0:0:0:0]    disk    IBM      2145             0000  /dev/sdb
[1:0:0:0]    disk    IBM      2145             0000  /dev/sda
[1:0:0:1]    disk    IBM      2145             0000  -

Signed-off-by: Li Zhong <lizhongfs@gmail.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: megaraid_sas: Make megasas_host_device_list_query() static

Fix sparse warning:

drivers/scsi/megaraid/megaraid_sas_base.c:4652:1: warning:
 symbol 'megasas_host_device_list_query' was not declared. Should it be static?

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: ufs: Print real incorrect request response code

If UFS device responds an unknown request response code, we can not know
what it was via logs because the code is replaced by "DID_ERROR << 16"
before log printing.

Fix this to provide precise request response code information for easier
issue breakdown.

Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
Reviewed-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxgb4i: fix incorrect spelling "reveive" -> "receive"

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

scsi: libcxgbi: update route finding logic

To support vlan and bridge devices first find route using ifindex 0, if
route is not found through net device associated with input scsi host then
find route using ifindex of net device.

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

scsi: libcxgbi: find cxgbi device by MAC address

If cxgbi_device_find_by_netdev() returns NULL then find cxgbi device by MAC
address.

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

scsi: core: don't hold device refcount in IO path

scsi_device's refcount is always grabbed in IO path.

Turns out it isn't necessary, because blk_queue_cleanup() will drain any
in-flight IOs, then cancel timeout/requeue work, and SCSI's requeue_work is
canceled too in __scsi_remove_device().

Also scsi_device won't go away until blk_cleanup_queue() is done.

So don't hold the refcount in IO path, especially the refcount isn't
required in IO path since blk_queue_enter() / blk_queue_exit() is
introduced in the legacy block layer.

Cc: Dongli Zhang <dongli.zhang@oracle.com>
Cc: James Smart <james.smart@broadcom.com>
Cc: Bart Van Assche <bart.vanassche@wdc.com>
Cc: linux-scsi@vger.kernel.org,
Cc: Martin K . Petersen <martin.petersen@oracle.com>,
Cc: Christoph Hellwig <hch@lst.de>,
Cc: James E . J . Bottomley <jejb@linux.vnet.ibm.com>,
Cc: jianchao wang <jianchao.w.wang@oracle.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix read offset in qla24xx_load_risc_flash()

This patch fixes regression introduced by commit f8f97b0c5b7f ("scsi:
qla2xxx: Cleanups for NVRAM/Flash read/write path") where flash read/write
routine cleanup left out code which resulted into checksum failure leading
to use-after-free stack during driver load.

Following stack trace is seen in the log file

qla2xxx [0000:00:00.0]-0005: : QLogic Fibre Channel HBA Driver: 10.01.00.16-k.
qla2xxx [0000:00:0b.0]-001d: : Found an ISP2532 irq 11 iobase 0x0000000000f47f03.
qla2xxx [0000:00:0b.0]-00cd:8: ISP Firmware failed checksum.
qla2xxx [0000:00:0b.0]-00cf:8: Setup chip ****FAILED****.
qla2xxx [0000:00:0b.0]-00d6:8: Failed to initialize adapter - Adapter flags 2.
==================================================================
BUG: KASAN: use-after-free in __list_del_entry_valid+0x15/0xd0
Read of size 8 at addr ffff8880ca05a490 by task modprobe/857

CPU: 0 PID: 857 Comm: modprobe Not tainted 5.1.0-rc1-dbg+ #4
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
Call Trace:
  dump_stack+0x86/0xca
  print_address_description+0x6c/0x234
  ? __list_del_entry_valid+0x15/0xd0
  kasan_report.cold.3+0x1b/0x34
  ? __list_del_entry_valid+0x15/0xd0
  ? __kmem_cache_shutdown.cold.95+0xf5/0x176
  ? __list_del_entry_valid+0x15/0xd0
  __asan_load8+0x54/0x90
  __list_del_entry_valid+0x15/0xd0
  dma_pool_destroy+0x4f/0x260
  ? dma_free_attrs+0xb4/0xd0
  qla2x00_mem_free+0x529/0xcc0 [qla2xxx]
  ? kobject_put+0xdb/0x230
  qla2x00_probe_one+0x2b5e/0x45f0 [qla2xxx]
  ? qla2xxx_pci_error_detected+0x210/0x210 [qla2xxx]
  ? match_held_lock+0x20/0x240
  ? find_held_lock+0xca/0xf0
  ? mark_held_locks+0x86/0xb0
  ? _raw_spin_unlock_irqrestore+0x52/0x60
  ? __pm_runtime_resume+0x5b/0xb0
  ? lockdep_hardirqs_on+0x185/0x260
  ? _raw_spin_unlock_irqrestore+0x52/0x60
  ? trace_hardirqs_on+0x24/0x130
  ? preempt_count_sub+0x13/0xc0
  ? _raw_spin_unlock_irqrestore+0x3d/0x60
  pci_device_probe+0x154/0x1e0
  really_probe+0x17d/0x540
  ? device_driver_attach+0x90/0x90
  driver_probe_device+0x113/0x170
  ? device_driver_attach+0x90/0x90
  device_driver_attach+0x88/0x90
  __driver_attach+0xb5/0x190
  bus_for_each_dev+0xf8/0x160
  ? subsys_dev_iter_exit+0x10/0x10
  ? kasan_check_read+0x11/0x20
  ? preempt_count_sub+0x13/0xc0
  ? _raw_spin_unlock+0x2c/0x50
  driver_attach+0x26/0x30
  bus_add_driver+0x238/0x2f0
  driver_register+0xd7/0x150
  __pci_register_driver+0xd5/0xe0
  ? 0xffffffffa06c8000
  qla2x00_module_init+0x208/0x254 [qla2xxx]
  do_one_initcall+0xc0/0x3c9
  ? trace_event_raw_event_initcall_finish+0x150/0x150
  ? __kasan_kmalloc.constprop.5+0xc7/0xd0
  ? kasan_unpoison_shadow+0x35/0x50
  ? kasan_poison_shadow+0x2f/0x40
  ? __asan_register_globals+0x5a/0x70
  do_init_module+0x103/0x330
  load_module+0x36df/0x3b70
  ? fsnotify+0x611/0x640
  ? module_frob_arch_sections+0x20/0x20
  ? kernel_read+0x74/0xa0
  ? kasan_check_write+0x14/0x20
  ? kernel_read_file+0x25e/0x320
  ? do_mmap+0x42c/0x6c0
  __do_sys_finit_module+0x133/0x1c0
  ? __do_sys_finit_module+0x133/0x1c0
  ? __do_sys_init_module+0x210/0x210
  ? fput_many+0x1b/0xc0
  ? fput+0xe/0x10
  ? do_syscall_64+0x14/0x210
  ? entry_SYSCALL_64_after_hwframe+0x49/0xbe
  __x64_sys_finit_module+0x3e/0x50
  do_syscall_64+0x72/0x210
  entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7f8bd5c03219
Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 47 fc 0c 00 f7 d8 64 89 01 48
RSP: 002b:00007fff9d11de98 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
RAX: ffffffffffffffda RBX: 000055ef21596b50 RCX: 00007f8bd5c03219
RDX: 0000000000000000 RSI: 000055ef21596570 RDI: 0000000000000004
RBP: 000055ef21596570 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000004 R11: 0000000000000246 R12: 0000000000000000
R13: 000055ef21596c80 R14: 0000000000040000 R15: 000055ef21596b50

Allocated by task 857:
  save_stack+0x43/0xd0
  __kasan_kmalloc.constprop.5+0xc7/0xd0
  kasan_kmalloc+0x9/0x10
  kmem_cache_alloc_trace+0x144/0x300
  dma_pool_create+0xb5/0x3b0
  qla2x00_mem_alloc+0xb98/0x1ad0 [qla2xxx]
  qla2x00_probe_one+0xe28/0x45f0 [qla2xxx]
  pci_device_probe+0x154/0x1e0
  really_probe+0x17d/0x540
  driver_probe_device+0x113/0x170
  device_driver_attach+0x88/0x90
  __driver_attach+0xb5/0x190
  bus_for_each_dev+0xf8/0x160
  driver_attach+0x26/0x30
  bus_add_driver+0x238/0x2f0
  driver_register+0xd7/0x150
  __pci_register_driver+0xd5/0xe0
  qla2x00_module_init+0x208/0x254 [qla2xxx]
  do_one_initcall+0xc0/0x3c9
  do_init_module+0x103/0x330
  load_module+0x36df/0x3b70
  __do_sys_finit_module+0x133/0x1c0
  __x64_sys_finit_module+0x3e/0x50
  do_syscall_64+0x72/0x210
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

Freed by task 857:
  save_stack+0x43/0xd0
  __kasan_slab_free+0x139/0x190
  kasan_slab_free+0xe/0x10
  kfree+0xf0/0x2c0
  dma_pool_destroy+0x24c/0x260
  qla2x00_mem_free+0x529/0xcc0 [qla2xxx]
  qla2x00_free_device+0x167/0x1b0 [qla2xxx]
  qla2x00_probe_one+0x2b28/0x45f0 [qla2xxx]
  pci_device_probe+0x154/0x1e0
  really_probe+0x17d/0x540
  driver_probe_device+0x113/0x170
  device_driver_attach+0x88/0x90
  __driver_attach+0xb5/0x190
  bus_for_each_dev+0xf8/0x160
  driver_attach+0x26/0x30
  bus_add_driver+0x238/0x2f0
  driver_register+0xd7/0x150
  __pci_register_driver+0xd5/0xe0
  qla2x00_module_init+0x208/0x254 [qla2xxx]
  do_one_initcall+0xc0/0x3c9
  do_init_module+0x103/0x330
  load_module+0x36df/0x3b70
  __do_sys_finit_module+0x133/0x1c0
  __x64_sys_finit_module+0x3e/0x50
  do_syscall_64+0x72/0x210
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

The buggy address belongs to the object at ffff8880ca05a400
  which belongs to the cache kmalloc-192 of size 192
The buggy address is located 144 bytes inside of
  192-byte region [ffff8880ca05a400, ffff8880ca05a4c0)
The buggy address belongs to the page:
page:ffffea0003281680 count:1 mapcount:0 mapping:ffff88811bf03380 index:0x0 compound_mapcount: 0
flags: 0x4000000000010200(slab|head)
raw: 4000000000010200 0000000000000000 0000000c00000001 ffff88811bf03380
raw: 0000000000000000 0000000080200020 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
  ffff8880ca05a380: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
  ffff8880ca05a400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>ffff8880ca05a480: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
                          ^
  ffff8880ca05a500: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  ffff8880ca05a580: 00 00 00 00 00 00 00 00 fc fc fc fc fc fc fc fc
==================================================================

Fixes: f8f97b0c5b7f ("scsi: qla2xxx: Cleanups for NVRAM/Flash read/write path")
Reported-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Move qla2x00_set_fcport_state() from a .h into a .c file

The qla2x00_set_fcport_state() function is not in the hot path so move its
definition from a .h into a .c file.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove two superfluous casts

Casting a void pointer into another pointer before assigning the pointer to
a variable is not useful. Hence remove such casts.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove qla_tgt_cmd.data_work and qla_tgt_cmd.data_work_free

The 'data_work' and 'data_work_free' member variables are set but never
used. Hence remove both member variables. See also commit 6bcbb3174caa
("qla2xxx: Fix incorrect tcm_qla2xxx_free_cmd use during TMR ABORT (v2)").

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Move the <linux/io-64-nonatomic-lo-hi.h> include directive

The <linux/io-64-nonatomic-lo-hi.h> header file is included because of the
readq() macro. Since that macro is only used in qla_nx.c, move that include
statement into qla_nx.c.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Declare qla24xx_build_scsi_crc_2_iocbs() static

Since qla24xx_build_scsi_crc_2_iocbs() is only used inside a single source
file, declare this function static.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Move the port_state_str[] definition from a .h to a .c file

Reduce the size of the qla2xxx kernel module by moving an array definition
from a .h into a .c file.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Insert spaces where required

Improve source code readability by inserting spaces where these are
required according to the coding standard. This patch only inserts
whitespace and does not make any other changes.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Fix formatting of pointer types

Improve source code readability by following the Linux kernel coding style
for pointer types. This patch only changes whitespace.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Leave a blank line after declarations

This patch improves readability of the qla2xxx source code.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Use tabs to indent code

Most but not all code in the qla2xxx driver uses tabs for indentation.
Make the qla2xxx code easier to read by using tabs consistently for
indentation. This patch improves conformance with the Linux kernel coding
style.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: libsas: Print expander PHY indexes in decimal

Currently we print expander PHY indexes in a mix of decimal and hex.

It is more consistent and also more convenient to read decimal, so
make this change.

We use width of 2 for expander and 1 for root PHYs prints.

Some lines which were needlessly spilling multiple lines are unified.

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

scsi: libsas: Do discovery on empty PHY to update PHY info

When we discover the PHY is empty in sas_rediscover_dev(), the PHY
information (like negotiated linkrate) is not updated.

As such, for a user examining sysfs for that PHY, they would see
incorrect values:

root@(none)$ cd /sys/class/sas_phy/phy-0:0:20
root@(none)$ more negotiated_linkrate
3.0 Gbit
root@(none)$ echo 0 > enable
root@(none)$ more negotiated_linkrate
3.0 Gbit

So fix this, simply discover the PHY again, even though we know it's empty;
in the above example, this gives us:

root@(none)$ more negotiated_linkrate
Phy disabled

We must do this after unregistering the device associated with the PHY
(in sas_unregister_devs_sas_addr()).

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

scsi: libsas: Inject revalidate event for root port event

According to the SAS spec, an expander device shall transmit BROADCAST
(CHANGE) from at least one phy in each expander port other than the
expander port that is the cause for transmitting BROADCAST (CHANGE).

As such, for when the link is lost for a root PHY attached to an expander
PHY, we get no broadcast event.

This causes an issue for libsas, in that we will not revalidate the domain
for these events.

As a solution, for when a root PHY is formed or deformed from a root port,
insert a broadcast event to trigger a domain revalidation.

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

scsi: libsas: Improve vague log in SAS rediscovery

When an expander PHY which was part of a wideport disconnects, we would see
a log like this from sas_rediscover():

[   39.695554] sas: phy20 part of wide port with phy16

Here, phy20 is the PHY that disconnected, and phy16 is the lowest indexed
member PHY of the wideport.

The log implies the phy20 is still part of the wideport with phy16, so is
misleading or, at least, vague.

Improve the logs in SAS rediscovery by removing this log and adding a log
in sas_rediscover_dev() to tell what's really going on.

While we're at it, also make the logs in sas_find_bcast_dev() more
informative (and more consistent with the reset of the expander logs).

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

scsi: libsas: Try to retain programmed min linkrate for SATA min pathway unmatch fixing

Currently for fixing the linkrate matching during discovery such that the
linkrate of a SATA PHY does not exceed min pathway to initiator, we set the
SATA PHY programmed min linkrate to the same value as the programmed max
linkrate.

This is unnecessary, and we should be able to keep the same programmed min
linkrate if it is already lower than this new max programmed linkrate.

This patch makes that change.

In effect, this will not make much difference since we generally will
negotiate a linkrate at the programmed max linkrate, and the programmed min
linkrate will have no impact.

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

scsi: libsas: Stop hardcoding SAS address length

Many times we use 8 for SAS address length, while we already have a macro
for this - SAS_ADDR_SIZE.

Replace instances of this with the macro. However, don't touch the SAS
address array sizes sas.h, as these are defined according to the SAS spec.

Some missing whitespaces are also added, and whitespace indentation
in sas_hash_addr() is also fixed (see sas_hash_addr()).

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

scsi: hisi_sas: Some misc tidy-up

Do some minor tidy-up.

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: Don't fail IT nexus reset for Open Reject timeout

Currently we call hisi_sas_softreset_ata_disk() in
hisi_sas_I_T_nexus_reset().

If this fails for open reject reason, there is no reason to fail the IT
nexus reset, so only fail for TMF_RESP_FUNC_FAILED.

Some other strings spilled over multiple lines are reunited.

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: Don't hard reset disk during controller reset

In the function of hisi_sas_init_device(), we added ops->hardreset() to
clear affiliation of STP target port or handle [STP pending] state.

Function hisi_sas_init_device() will be called when a device is found or
during controller reset. At controller reset, we call
hisi_sas_init_device() to re-init the disks, so calling hardreset() is
unnecessary and it also will cause some delay at controller reset.

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: Support all RAS events with MSI interrupts

This patch is to switch HW all error handling from PCI AER to MSI interrupt
due to non-standard PCI implementation. All HW errors which were being
reported through PCI AER can be reported through MSI interrupt also.

Do two things to complete the switch:

1. Notify FW to switch to MSI handling through ACPI DSM.

2. Add MSI handling for some hw errors, ECC errors and poison errors (we
   also call some of them AXI reuser error). They were handled only through
   PCI AER before.

For old FW reporting PCI AER events, the PCI AER handler will see that the
driver on longer support AER, and will leave the device in offlined state,
which is safe.

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

scsi: hisi_sas: allocate different SAS address for directly attached situation

In commit 8b8d66531555 ("scsi: hisi_sas: make SAS address of SATA disks
unique"), we ensured that each SATA disk in the system has a unique SAS
address, even if it is fake. That was for v2 hw.

Add this for v3 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: Adjust the printk format of functions hisi_sas_init_device()

In function hisi_sas_init_device(), the log is as follows when error for
hardreset:

  hisi_sas_v3_hw 0000:74:02.0: SATA disk hardreset fail: 0xffffffed

Actually if hardreset failed, its return value is negative, so change the
print format from %x to %d.

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 for setting the PHY linkrate when disconnected

In commit efdcad62e7b8 ("scsi: hisi_sas: Set PHY linkrate when
disconnected"), we use the sas_phy_data.enable flag to track whether the
PHY was enabled or not, so that we know if we should set the PHY negotiated
linkrate at SAS_LINK_RATE_UNKNOWN or SAS_PHY_DISABLED.

However, it is not proper to use sas_phy_data.enable, since it is only set
when libsas attempts to set the PHY disabled/enabled; hence, it may not
even have an initial value.

As a solution to this problem, introduce hisi_sas_phy.enable to track
whether the PHY is enabled or not, so that we can set the negotiated
linkrate properly when the PHY comes down.

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

scsi: hisi_sas: Remedy inconsistent PHY down state in software

Currently there are two scenarioes which may cause PHY state of hardware
(which is 0) is inconsistent with the state held in software:

- Unplug SAS wire before get_phys_state when SAS controller reset, then the
  interrupts of phy down are ignored, phy state is 0 before reset, and it
  also gets 0 after reset, so phy down doesn't occur even if unplugged SAS
  wire;

- For v3 hw later version, it will close bus when 2 bit ECC error occurs.
  So if unplug SAS wire at that time, interrupts of phy down also not
  occur. So at last it will cause host reset. It also get phy state 0
  before and after reset, the same issue occurs.

To solve it, use hisi_sas_phy_down() directly in rescan topology function.

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: add host reset interface for test

Add host reset interface to make it easier for testing the host reset
feature.

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>

Merge branch '5.1/scsi-fixes' into 5.2/merge

We have a few submissions for 5.2 that depend on fixes merged post
5.1-rc1. Merge the fixes branch into queue.

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

scsi: target/iscsi: Make sure PDU processing continues if parsing a command fails

Currently the iSCSI target driver sends a CHECK CONDITION code back to the
initiator if the immediate data buffer is too large but it does not discard
that immediate data buffer. The result is that the iSCSI target driver
attempts to parse the immediate data itself as iSCSI PDUs and that all
further iSCSI communication fails. Fix this by receiving and discarding too
large immediate data buffers.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/iscsi: Make iscsit_map_iovec() more robust

Make the code for mapping an iovec more robust by checking the bounds of
the allocated iovec. This patch avoids that the following crash occurs if a
map attempt is made that exceeds the bounds of the iovec that is being
mapped:

BUG: unable to handle kernel NULL pointer dereference at 00000000
00000014
RIP: 0010:iscsit_map_iovec+0x120/0x190 [iscsi_target_mod]
Call Trace:
 iscsit_get_rx_pdu+0x8a2/0xe00 [iscsi_target_mod]
 iscsi_target_rx_thread+0x6e/0xa0 [iscsi_target_mod]
 kthread+0x109/0x140

Cc: Mike Christie <mchristi@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/iscsi: Handle too large immediate data buffers correctly

Since target_alloc_sgl() and iscsit_allocate_iovecs() allocate buffer space
for se_cmd.data_length bytes and since that number can be smaller than the
iSCSI Expected Data Transfer Length (EDTL), ensure that the iSCSI target
driver does not attempt to receive more bytes than what fits in the receive
buffer. Always receive the full immediate data buffer such that the iSCSI
target driver does not attempt to parse immediate data as an iSCSI PDU.

Note: the current code base only calls iscsit_get_dataout() if the size of
the immediate data buffer does not exceed the buffer size derived from the
SCSI CDB. See also target_cmd_size_check().

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/iscsi: Only send R2T if needed

If an initiator submits more immediate data than the size derived from the
SCSI CDB, do not send any R2T to the initiator. This scenario is triggered
by the libiscsi test ALL.iSCSIResiduals.WriteVerify16Residuals if the iSCSI
target driver is modified to discard too large immediate data buffers
instead of trying to parse these as an iSCSI PDU. This patch avoids that a
negative xfer_len value is passed to iscsit_add_r2t_to_list() if too large
immediate data buffers are handled correctly.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/iscsi: Detect conn_cmd_list corruption early

Certain behavior of an initiator can cause the target driver to send both a
reject and a SCSI response. If that happens two target_put_sess_cmd() calls
will occur without the command having been removed from conn_cmd_list. In
other words, conn_cmd_list will get corrupted once the freed memory is
reused. Although the Linux kernel can detect list corruption if list
debugging is enabled, in this case the context in which list corruption is
detected is not related to the context that caused list corruption.  Hence
add WARN_ON() statements that report the context that is causing list
corruption.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/core: Make the XCOPY setup code easier to read by inlining two functions

The target_xcopy_setup_pt_port() and target_xcopy_init_pt_lun() functions
obfuscate what is really going on. Hence inline these two functions. This
patch does not change any functionality.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/core: Simplify LUN initialization in XCOPY implementation

Other than removing a few pr_debug() statements, this patch does not change
any functionality.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/core: Remove a set-but-not-used member variable from the XCOPY implementation

This patch does not change any functionality.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/core: Rework the SPC-2 reservation handling code

Instead of tracking the initiator that established an SPC-2 reservation,
track the session through which the SPC-2 reservation has been
established. This patch does not change any functionality.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: target/core: Fix a race condition in the LUN lookup code

The rcu_dereference(deve->se_lun) expression occurs twice in the LUN lookup
functions. Since these expressions are not serialized against deve->se_lun
assignments each of these expressions may yield a different result. Avoid
that the wrong LUN pointer is stored in se_cmd by reading deve->se_lun only
once.

Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Fixes: 29a05deebf6c ("target: Convert se_node_acl->device_list[] to RCU hlist") # v4.10
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_transport_fc: nvme: display FC-NVMe port roles

Currently the FC-NVMe driver is leverating the SCSI FC transport class to
access the remote ports. Which means that all FC-NVMe remote ports will be
visible to the fc transport layer, but due to missing definitions the port
roles will always be 'unknown'.  This patch adds the missing definitions to
the fc transport class to that the port roles are correctly displayed.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: James Smart <james.smart@broadcom.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Reviewed-by: Giridhar Malavali <gmalavali@marvell.com>
Reviewed-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: cxgbi: remove redundant __kfree_skb call on skb and free cst->atid

The error return path via label rel_resource checks for a non-null skb
before free'ing it.  However, skb is always null at this exit path, so the
null check and the free are redundant and can be removed.  Removing this
allows the original goto's to rel_resource to be cleaned up; the first can
be replaced by a return of -EINVAL, the second can be replaced by a more
appropriate -ENOMEM return and fix a memory leak by freeing csk->atid.

Addresses-Coverity: ("Logically Dead Code")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: tcm_qla2xxx: Minimize #include directives

Only include those header files that are needed by the code in this source
file.

[mkp: include build fix from Bart for SPARC]

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: csiostor: fix missing data copy in csio_scsi_err_handler()

If scsi cmd sglist is not suitable for DDP then csiostor driver uses
preallocated buffers for DDP, because of this data copy is required from
DDP buffer to scsi cmd sglist before calling ->scsi_done().

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

scsi: qla2xxx: Unregister resources in the opposite order of the registration order

Make sure that resources are only unregistered after the users of these
resources have been unregistered. Only unregister the character device if
registration of it succeeded.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Unregister chrdev if module initialization fails

If module initialization fails after the character device has been
registered, unregister the character device. Additionally, avoid
duplicating error path code.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <giridhar.malavali@qlogic.com>
Fixes: 6a03b4cd78f3 ("[SCSI] qla2xxx: Add char device to increase driver use count") # v2.6.35.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Use get/put_unaligned where appropriate

This patch makes the code easier to read but does not change any
functionality.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Make qla2x00_process_response_queue() easier to read

Make qla2x00_process_response_queue() easier to read by splitting this
function. This patch does not change any functionality.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Reduce the number of forward declarations

Move the SCSI host template definition after the definition of the
functions that it references. Remove the forward declarations that became
unnecessary by that change. This patch does not change any functionality.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Declare local symbols static

This patch avoids that sparse complains that a declaration is missing.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Use ARRAY_SIZE() in the definition of QLA_LAST_SPEED

Make the code easier to read by avoiding to use a hard-coded constant. This
patch does not change any functionality.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Remove a comment that refers to the SCSI host lock

Since qla2xxx_queuecommand() no longer takes the SCSI host lock, remove a
comment that refers to the host lock.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: Change a stack variable into a static const variable

Make qla2x00_get_host_fabric_name() slightly faster by changing a stack
array into a static array. Declare that array const now that the
wwn_to_u64() argument pointer has been declared const.

Cc: Himanshu Madhani <hmadhani@marvell.com>
Cc: Giridhar Malavali <gmalavali@marvell.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: scsi_transport_fc: Declare wwn_to_u64() argument const

Since the wwn_to_u64() function does not change its input, make its
argument const.

Cc: James Smart <james.smart@broadcom.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Steffen Maier <maier@linux.vnet.ibm.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

scsi: qla2xxx: fix spelling mistake "alredy" -> "already"

There is a spelling mistake in a ql_log message. Fix it.

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

scsi: lpfc: add support for posting FC events on FPIN reception

This patch adds support to recognize FPIN ELS's that are received.  When
one is received, the fc transport will be called to handle the the FPIN.

Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>