This patch fixes regression added by commit d7459527
("scsi: qla2xxx: Add multiple queue pair functionality.").

When driver is not able to get reqeusted IRQs from the system, driver will
attempt tp clean up memory before failing hardware probe. During this cleanup,
driver assigns NULL value to the pointer which has not been allocated by
driver yet. This results in a NULL pointer access.

Log file will show following message and stack trace

qla2xxx [0000:a3:00.1]-00c7:21: MSI-X: Failed to enable support, giving up -- 32/-1.
qla2xxx [0000:a3:00.1]-0037:21: Falling back-to MSI mode --1.
qla2xxx [0000:a3:00.1]-003a:21: Failed to reserve interrupt 821 already in use.
BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<ffffffffc010c4b6>] qla2x00_probe_one+0x18b6/0x2730 [qla2xxx]
PGD 0
Oops: 0002 [#1] SMP

Fixes: d7459527 ("scsi: qla2xxx: Add multiple queue pair functionality.").
Cc: <stable@vger.kernel.org> # 4.10
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

for session deletion, replace sess_lock with work_lock.
Under certain case sess_lock is not feasiable to acquire.
The lock is needed temporarily to make sure a single
call to schedule of the work element.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This patch adds check for pending work event before queueing
relogin work to prevent redundant work to be active at the
same time.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

add ability to use gpnft/gnnft to handle RSCN.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Migrate switch registration commands: RFTID, RFFID, RNNID and RSNN_NN
out of mailbox interface to reduce fabric scan bottle neck.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

- add "async" gpn_ft, gnn_ft, gfpn_id, gnn_id switch commands.
- For 8G and newer adapters, use async commands when it comes to
fabric scan to reduce bottle neck.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

When connection type is N_Port to N_Port (point-to-point), there
is a possibilty where initiator will not send PLOGI request and
will directly send PRLI. In N2N connection the port has higher
port name sends the PLOGI but not allow to send PRLI if is a
target mode. Only initiator is allowed to send PRLI.

Current driver code deletes old session when it receives PLOGI
request. If we will not receive PLOGI request then we will not
delete old session and create new session. Add check for N2N
with PRLI receive only and trigger cleanup. For this case, the
cleanup requires individual cmd abort instead of using implicit
logout as a broad stroke flush.
Signed-off-by: NKrishna Kant <krishna.kant@purestorage.com>
Signed-off-by: NAlexei Potashnik <alexei@purestorage.com>
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add ability to send Implicit PRLO to flush IOs
from FW back to driver.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add option to tell FW to reserve 1/2 of emergency exchanges for ELS.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

DPC thread can stall during switch scan due to slow switch response.
This will stall other work element that needs attention. Moving work
element processing and relogin logic out of DPC thread and into its
own work queue.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

As part of chip reset, all commands from all QPairs are
flushed. This patch fixes code to use Q Pair lock for flush
instead of using old hardware_lock.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Max boundary for exchange off load is 32k exchanges. If a system
is unable to allocate large memory buffer to support this feature,
then driver will reduce the number of exchanges down to a value
system can support.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Current code uses Stop Firmware MB cmd to stop the chip before
driver unload.  This will leave the laser in its current state.
This give the illusion of this adapter is still alive.
For 8G & newer adapters, use chip reset to stop the chip and
bring down the laser.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Fix system crash due to NULL pointer access.

qlt_plogi_ack_t and fc_port structures were not properly bound before
calling qlt_plogi_ack_unref().

RIP: 0010:qlt_plogi_ack_unref+0xa1/0x150 [qla2xxx]
Call Trace:
qla24xx_create_new_sess+0xb1/0x320 [qla2xxx]
qla2x00_do_work+0x123/0x260 [qla2xxx]
qla2x00_iocb_work_fn+0x30/0x40 [qla2xxx]
process_one_work+0x1f3/0x530
worker_thread+0x4e/0x480
kthread+0x10c/0x140

Fixes: 726b8548 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org> # 4.10+
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NGiridhar Malavali <giridhar.malavali@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Current driver design schedules relogin process via DPC thread every 1
second. In a large fabric, this DPC thread tries to schedule too many
jobs and might get overloaded. As a result of this processing of DPC
thread, it can schedule relogin earlier than 1 second.

Fixes: 726b8548 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org> # 4.10+
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

If user swaps one target port for another target port for same switch
port, the new target port is not being recognized by the driver. Current
code assumes that old Target port has recovered from link down. The fix
will ask switch what is the WWPN of a specific NportID (GPNID) rather
than assuming it's the same Target port which has came back.

Fixes: 726b8548 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org> # 4.10+
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

GPNID is triggered by RSCN. For multiple RSCNs of the same affected
NPORT ID, serialize the GPNID to prevent confusion.

Fixes: 726b8548 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org> # 4.10+
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Fix co-existence between Block MQ and Target Mode. Block MQ and
initiator mode requires midlayer queue mapping to check for IRQ to be
affinitized. For target mode, it's not the case.

Fixes: 09620eeb ("scsi: qla2xxx: Add debug knob for user control workload")
Cc: <stable@vger.kernel.org> # 4.12+
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Move session delete from system work queue to driver's work queue for in
time processing.

Fixes: 726b8548 ("qla2xxx: Add framework for async fabric discovery")
Cc: <stable@vger.kernel.org> # 4.10+
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

In preparation for unconditionally passing the struct timer_list pointer to
all timer callbacks, switch to using the new timer_setup() and from_timer()
to pass the timer pointer explicitly.

Cc: qla2xxx-upstream@qlogic.com
Cc: "James E.J. Bottomley" <jejb@linux.vnet.ibm.com>
Cc: "Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-scsi@vger.kernel.org
Signed-off-by: NKees Cook <keescook@chromium.org>
Reviewed-by: NMartin K. Petersen <martin.petersen@oracle.com>
Tested-by: NBart Van Assche <Bart.VanAssche@wdc.com>

Avoid that the following is reported while loading the qla2xxx
kernel module:

BUG: using smp_processor_id() in preemptible [00000000] code: modprobe/783
caller is debug_smp_processor_id+0x17/0x20
CPU: 7 PID: 783 Comm: modprobe Not tainted 4.14.0-rc8-dbg+ #2
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
Call Trace:
 dump_stack+0x8e/0xce
 check_preemption_disabled+0xe3/0xf0
 debug_smp_processor_id+0x17/0x20
 qla2x00_probe_one+0xf43/0x26c0 [qla2xxx]
 pci_device_probe+0xca/0x140
 driver_probe_device+0x2e2/0x440
 __driver_attach+0xa3/0xe0
 bus_for_each_dev+0x5f/0x90
 driver_attach+0x19/0x20
 bus_add_driver+0x1c0/0x260
 driver_register+0x5b/0xd0
 __pci_register_driver+0x63/0x70
 qla2x00_module_init+0x1d6/0x222 [qla2xxx]
 do_one_initcall+0x3c/0x163
 do_init_module+0x55/0x1eb
 load_module+0x20a2/0x2890
 SYSC_finit_module+0xd7/0xf0
 SyS_finit_module+0x9/0x10
 entry_SYSCALL_64_fastpath+0x23/0xc2

Fixes: commit 8abfa9e2 ("scsi: qla2xxx: Add function call to qpair for door bell")
Signed-off-by: NBart Van Assche <bart.vanassche@wdc.com>
Cc: Quinn Tran <quinn.tran@cavium.com>
Cc: Himanshu Madhani <himanshu.madhani@cavium.com>
Cc: <stable@vger.kernel.org>
Acked-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

On error, kthread_create() returns an errno-encoded pointer, not NULL.
The routine qla2x00_probe_one() detects the error case and jumps to
probe_failed, but has already assigned the return value from
kthread_create() to ha->dpc_thread.  Then probe_failed checks to see if
ha->dpc_thread is not NULL before doing cleanup on it. Since in the
error case this is also not NULL, it ends up trying to access an invalid
task pointer.

Solution is to assign NULL to ha->dpc_thread in the error path to avoid
kthread cleanup in that case.
Signed-off-by: NDouglas Miller <dougmill@linux.vnet.ibm.com>
Acked-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

commit a9e170e2 ("scsi: qla2xxx: Fix uninitialized work element")
moved initializiation of work element earlier in the probe to fix call
stack. However, it still leaves a window where interrupt can be
generated before work element is initialized. Fix that window by
initializing work element before we are requesting IRQs.

[mkp: fixed typos]

Fixes: a9e170e2 ("scsi: qla2xxx: Fix uninitialized work element")
Cc: <stable@vger.kernel.org> # 4.13
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Based on the FC4 type, login will proceed to either FCP or FC-NVMe
remote ports.
Signed-off-by: NGiridhar Malavali <giridhar.malavali@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

In some environments, user can choose to not enable SCSI-MQ but wants to
use FC-NVMe feature of the driver. Since driver relies on Q-Pairs to
allocate FC-NVMe resources, use existing module parameter to create
Q-Pairs when FC-NVMe is enabled.
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NEwan D. Milne <emilne@redhat.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

[mkp: fixed whitespace]
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Fixes following stack trace

kernel: Call Trace:
kernel: dump_stack+0x63/0x84
kernel: __warn+0xd1/0xf0
kernel: warn_slowpath_null+0x1d/0x20
kernel: __queue_work+0x37a/0x420
kernel: queue_work_on+0x27/0x40
kernel: queue_work+0x14/0x20 [qla2xxx]
kernel: schedule_work+0x13/0x20 [qla2xxx]
kernel: qla2x00_post_work+0xab/0xb0 [qla2xxx]
kernel: qla2x00_post_aen_work+0x3b/0x50 [qla2xxx]
kernel: qla2x00_async_event+0x20d/0x15d0 [qla2xxx]
kernel: ? lock_timer_base+0x7d/0xa0
kernel: qla24xx_intr_handler+0x1da/0x310 [qla2xxx]
kernel: qla2x00_poll+0x36/0x60 [qla2xxx]
kernel: qla2x00_mailbox_command+0x659/0xec0 [qla2xxx]
kernel: ? proc_create_data+0x7a/0xd0
kernel: qla25xx_init_rsp_que+0x15b/0x240 [qla2xxx]
kernel: ? request_irq+0x14/0x20 [qla2xxx]
kernel: qla25xx_create_rsp_que+0x256/0x3c0 [qla2xxx]
kernel: qla2xxx_create_qpair+0x2af/0x5b0 [qla2xxx]
kernel: qla2x00_probe_one+0x1107/0x1c30 [qla2xxx]

Fixes: ec7193e2 ("qla2xxx: Fix delayed response to command for loop mode/direct connect.")
Cc: <stable@vger.kernel.org> # 4.13
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Call Trace:
 [<ffffffff81341687>] dump_stack+0x6b/0xa4
 [<ffffffff810c3e30>] ? print_irqtrace_events+0xd0/0xe0
 [<ffffffff8109e3c3>] ___might_sleep+0x183/0x240
 [<ffffffff8109e4d2>] __might_sleep+0x52/0x90
 [<ffffffff811fe17b>] kmem_cache_alloc_trace+0x5b/0x300
 [<ffffffff810c666b>] ? __lock_acquired+0x30b/0x420
 [<ffffffffa0733c28>] qla2x00_alloc_fcport+0x38/0x2a0 [qla2xxx]
 [<ffffffffa07217f4>] ? qla2x00_do_work+0x34/0x2b0 [qla2xxx]
 [<ffffffff816cc82b>] ? _raw_spin_lock_irqsave+0x7b/0x90
 [<ffffffffa072169a>] ? qla24xx_create_new_sess+0x3a/0x160 [qla2xxx]
 [<ffffffffa0721723>] qla24xx_create_new_sess+0xc3/0x160 [qla2xxx]
 [<ffffffff810c91ed>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffffa07218f8>] qla2x00_do_work+0x138/0x2b0 [qla2xxx]
Signed-off-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

If there are IO's running and host reset or chip reset is triggered,
IO's can fail due to rport time out. During chip reset recovery process,
driver notifies the transport layer that remote port no longer exist, by
calling fc_remote_port_delete(). When this function is called, it
actually delays deletion by "blocking" it. It sets the remote port state
to "FC_PORTSTATE_BLOCKED" and sets FC_RPORT_DEVLOSS_PENDING.  When
driver tries to abort the command by calling its abort handler function,
abort handler will wait until remote port state is blocked state or wait
for dev_loss_tmo time. Due to this blocking, rport times out and results
in an IO failure. This patch adds a check for any active reset process
before calling abort handler function.

[mkp: typo]
Signed-off-by: NSawan Chandak <sawan.chandak@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

ql2xmaxqdepth is the module parameter that seeds the per target
queue depth in the Scsi midlayer (sdev->queue_depth). Performance
testing revealed that increasing this value would improve IOPS
numbers under certain workloads.
Signed-off-by: NMichael Hernandez <michael.hernandez@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NGiridhar Malavali <giridhar.malavali@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NJoe Carnuccio <joe.carnuccio@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NJoe Carnuccio <joe.carnuccio@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Calling sp->free() ensures the context-correct free routine is called.
Signed-off-by: NJoe Carnuccio <joe.carnuccio@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

SFP can come in 2 formats: short range/SR and long range/LR.  For LR,
user the can increase the number of Buffer to Buffer credits between end
points via Cavium's command line tool.  By default, FW uses a lower BB
Credit value optimized for SR.  This patch will read the SFP for each
link up event and during chip reset sequence. If the SFP type and
setting are mismatch, then the chip is reset 1 time to use the
appropriate setting.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Update ha->wq max_active from 1 to default. MQ interrupts are queued up
via this work queue. This allows interrupts to be process in parrallel,
instead of serialized by the work queue.
Signed-off-by: NQuinn Tran <quinn.tran@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Driver will perform following

- Set PFLG_DRIVER_REMOVING flag and do not disable PCIe error reporting
  during adapter shutdown.
- If PCIe device count is already zero, return correct error type from
  PCI error interface.
- When device is offline, return correct error type from PCIe error
  path.
- If there is board disable thread active during shutdown and PCIe
  device count is zero then cancel scheduling board disable thread
  during shutdown and return.
Signed-off-by: NSawan Chandak <sawan.chandak@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

The driver held spinlocks during callbacks for NVME errors which
resulted in a deadlock because recovery LS cmds needed the same lock.
Signed-off-by: NDuane Grigsby <duane.grigsby@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add support to the driver to set the exchange threshold value for
the number of outstanding AENs.
Signed-off-by: NDuane Grigsby <duane.grigsby@cavium.com>
Signed-off-by: NDarren Trapp <darren.trapp@cavium.com>
Signed-off-by: NAnil Gurumurthy <anil.gurumurthy@cavium.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@cavium.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>