Need to make ktime samples more accurate

If ktime is turned on in the middle of an IO, the max calculation could
be misleading. Base sampling on the start time of the IO as opposed to
ktime_on.

Make ISR ktime timestamps be from when CQE is read instead of EQE.
Added additional sanity checks when deciding whether to accept an IO
sample or not.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Local Reject/Invalid RPI errors seen during discovery.

Temporary RPI cleanup was occurring regardless of SLI rev. It's only
necessary on SLI-4.

Adjust the test for whether cleanup is necessary.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Warning messages when NVME_TARGET_FC not defined on ppc builds

The lpfc_nvmet_replenish_context() function is only meaningful when NVME
target mode enabled. Surround the function body with ifdefs for target
mode enablement.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

In a link bounce scenario, a condition can occur where the discovery
engine swaps an ndlp structure (address change for an nport). While the
swap was successfully executed by the discovery engine, the driver did
not properly detect a change in the ndlp bound to the nvme rport.  This
error resulted in the nvme host transport issuing an IO to the correct
nvme rport, but the lpfc driver addressed a ndlp with an NLP_UNUSED
status and failed the io. This resulting it it looking like there were
missing namespaces and applications failed due to io errors.

To fix, in lpfc_nvme_register_rport, rework the "rebind" case to break
the nvme rport<->ndlp association when the ndlp already has an
nrport. Then rebind the rport to the correct ndlp data and backpointers.

[mkp: typo]
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

The driver crashes when attempting to use a freed ndpl pointer.

The pci_remove_one handler runs on a separate kernel thread. The order
of the removal is starting by freeing all of the ndlps and then
disabling interrupts. In between these two events the driver can still
receive an ELS and process it. When it tries to use the ndlp pointer
will be NULL

Change the order of the pci_remove_one vs disable interrupts so that
interrupts are disabled before the ndlp's are freed.

Cc: <stable@vger.kernel.org> # 4.12+
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

During pci hot plug, the kernel crashes in a list_add_call

The lookup by tag function will return null if the IOCB is out of range
or does not have the on txcmplq flag set.

Fix: Check for null return from lookup by tag.

Cc: <stable@vger.kernel.org> # 4.12+
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

During pci hot plug, the kernel crashes in timer management code.

The sli4 remove_one handler is not stoping the timers as it starts to
remove the port so that it can be swapped.

Fix: Stop the timers early in the handler routine.

Note: Fix in SLI-4 only. SLI-3 already stopped the timers properly.

Cc: <stable@vger.kernel.org> # 4.12+
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Use *_pool_zalloc rather than *_pool_alloc followed by memset with 0.
Found by coccinelle spatch "api/alloc/pool_zalloc-simple.cocci"
Signed-off-by: NThomas Meyer <thomas@m3y3r.de>
Acked-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

The pointer eqe is always non-null inside the while loop, so the check
to see if eqe is NULL is redudant and hence can be removed.

Detected by CoverityScan CID#1248693 ("Logically Dead Code")
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Acked-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

This is an interesting regression with gcc-8, showing a harmless warning
for correct code:

In file included from include/linux/kernel.h:13:0,
                 ...
                 from drivers/scsi/lpfc/lpfc_debugfs.c:23:
include/linux/printk.h:301:2: error: 'eq' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  printk(KERN_ERR pr_fmt(fmt), ##__VA_ARGS__)
  ^~~~~~
In file included from drivers/scsi/lpfc/lpfc_debugfs.c:58:0:
drivers/scsi/lpfc/lpfc_debugfs.h:451:31: note: 'eq' was declared here

I managed to reduce the warning into a small test case for gcc-8 that I
reported in the gcc bugzilla[1].

As a workaround, this changes the logic to move the two assignments of
'eq' out of the conditions and instead make the index conditional.  This
works for all configurations I tried and avoids adding a bogus
initialization.
Acked-by: NJames Smart <james.smart@broadcom.com>
Link: [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81958Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

The only reference to lpfc_nvmet_replenish_context() is inside of an
disabled:

drivers/scsi/lpfc/lpfc_nvmet.c:1457:1: error: 'lpfc_nvmet_replenish_context' defined but not used [-Werror=unused-function]

This replaces the preprocessor conditional with a C condition, so the
compiler can see that the function is intentionally unused.

Fixes: 9a38e4f1c82f ("scsi: lpfc: Fix MRQ > 1 context list handling")
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Update driver version to 11.4.0.3
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

cc1: warnings being treated as errors
drivers/scsi/lpfc/lpfc_init.c: In function 'lpfc_get_wwpn':
drivers/scsi/lpfc/lpfc_init.c:3253: error: integer constant is too large for 'long' type
Signed-off-by: NMaurizio Lombardi <mlombard@redhat.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Add Buffer to buffer credit recovery support to the driver.  This is a
negotiated feature with the peer that allows for both sides to detect
dropped RRDY's and FC Frames and recover credit.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Change hw queue binding messages to info - not error.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Port issue was fixed, the hbacmd reset would take more than 8 minutes to
complete.

There were conflicting NVME SGL posting/reposting responsibilities
between lpfc_online()/lpfc_sli4_hba_setup() and
lpfc_nvme_create_localport().  The lpfc_online() causes a REPOST on
existing NVME SGLs which is not released during the fc port reset.
However, lpfc_nvme_create_localport() wants to allocate new NVME buffers
and post them. Both cancelled out each other which had a side effect of
hosing the mailbox handling that was used to remove the sgl lists -
causing multiple 60s mbx timeouts.

Fix by preserving all SGL lists over the fc port reset.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

The nonrecovery occurred because the lpfc nvme initiator function did
not reestablish its localport creation with the nvme host transport in
lpfc_oneline.  Because of that, an NVME rport binding could not take
place.

Corrected by recreating the localport in the adapter reset recovery
routine.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

If the nvmet_fc transport breaks an io into multiple sequences, the
driver will improperly set the relative offset on the 2nd through N
sequences.

Correct by properly formatting the hw cmd so the relative offset is
picked up from the hw cmd.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Various oops including cpu LOCKUPs were seen.

For asynchronously received ius where the driver must assign exchange
resources, the resources were on a single get (free) list and put list
(finished, waiting to be put on get list). As all cpus are sharing the
lists, an interrupt for a receive frame may have to wait for all the
other cpus to place their done work onto the put list before it can
acquire the lock to pull from the list.

Fix by breaking the resource lists into per-cpu lists or at least more
than 1 list with cpu's sharing the lists). A cpu would allocate from the
free list for its own cpu, and put its done work on the its own put list
- avoiding the contention. As cpu load may vary, when empty, a cpu may
grab from another cpu, thereby changing resource distribution.  But
searching for a resource only occurs on 1 or a few cpus until a single
resource can be allocated. if the condition reoccurs, it starts looking
at a different cpu.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Various oops being seen on being in the ISR too long and cpu lockups,
when under heavy load.

The amount of work being posted off of completion queues kept the ISR
running almost all the time

Correct the issue by limiting the amount of work per iteration.

[mkp: typo]
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

When using fabric-assigned WWNs, the switch doesn't like copy of the
FLOGI payload, which includes valid VVL bits, to be used as the FDISC
payload.

Rather than wait for corrected switch firmware, ensure the VVL bits are
marked invalid on FDISCs.

[mkp: typo]
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

A race condition was found whereby the initiator would receive the RSCN
for a new NVME device before it had a chance to register its FC4 support
with the fabric. Thus, when queried by the initiator, it would see that
the target supported FC-NVME.

Corrected by making the assumption that the target always supports
FC-NVME thus a PRLI is sent. It's ok for the target to reject it.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

In adapter reset tests, an oops was seen with a NULL pointer in
lpfc_free_rq_buffer+0x20/0x60

The driver is failing to properly repost the nvmet sgl list when
recovering from the reset. Thus the driver eventually trys to walk an
errant buffer list.

Corrected the sgl buffer recovery as well as strengthening the
initialization of the bufferlist.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

After lip, the driver sometimes would have two rports for the same
device, allowing the namespaces to be duplicated by nvme.

In lpfc_plogi_confirm_nport() the driver was not swapping the nrport
maintained by the ndlp's undergoing address swapping. This allowed the
2nd rport to sneak in as it was considered a separate device.

This patch adds the fixes to Swap the nrport in each ndlp and take care
of the reference counts on the ndlps similar to FCP rports.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

After link bounce in a NVME Pt2Pt config, the driver managed to map the
same nport twice, resulting in multiple device nodes for the same
namespace.

In Pt2Pt, the driver must send PRLI's for both (scsi) FCP and NVME
rather than using fabric aids. The driver was inconsistent on handling
various PRLI completions, especially rejects, which had reject codes
cross the different protocol PRLI completions.

Fixed to perform the following: if nvmet mode (fc port can only be a
nvme target) - rejects all unsolicitly FCP PRLI's. Never issues a FCP
PRLI.

The multiple protocol PRLI's are sent simultaneously. However, driver
will now only state transition after both PRLI's are complete. New flags
were added to aid tracking the responses from the different PRLI's.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Modify driver return error codes to align with host nvme transport.

Driver isn't returning Exxx error codes to properly reflect out of
resource or connectivity conditions (-EBUSY), yet there were hard error
conditions returning -EBUSY.

Ensure the following situations return the proper return code:

 - Temporary failures or temporary resource availability: -EBUSY

 - Connectivity issues: -ENODEV

All others are treated as hard errors and return an -Exxx value that
indicates the type of error.

Also, lpfc_sli4_issue_wqe() was modified to not translate error from
-Exxx to WQE state.  This allows lpfc_nvme_fcp_io_submit() routine to
just return whatever -E value was returned from other routines.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Transitioned some informational discovery messages to now always be
displayed when log_verbose is set.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

lpfc oops when it discovers a NVME target but is configured for SCSI
only operation. Oops is in lpfc_nvme_register_port+0x33/0x300.

The localport is not valid so it should not have been referenced.

Added validity check for localport
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

When the switch blade is pulled out then plugged back in, the driver
does not issue a PLOGI to the target

When the switch blade is pulled out, it does not reset the link. The
driver ends up issuing a LOGO to the target, and finally sees devloss.
Since the driver believes that a LOGO is outstanding, it does not issue
a PLOGI to the target upon link up

Correct by placing the ndlp in UNUSED state When devloss happens in
LOGO_ISSUE state.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

The driver does not discover targets when in loop mode.

The NLP type is correctly getting set when a fabric connection is
detected but, not for loop. The unknown NLP type means that the driver
does not issue a PRLI when in loop topology. Thus target discovery
fails.

Fix by checking the topology during discovery.  If it is loop, set the
NLP FC4 type to FCP.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Message "0271 Illegal State Transition: node" seen in logs, all luns are
unuseable for that target.

A window exists in the rcv_plogi path where if the state is plogi issue
but the driver has not issued a plogi, then two reglogins will be sent
for the same RPI. The first one to complete will advance the state to
prli issue the second one will be detected as an illegal state, and
leave the node in an unusable state.

Correct the completion routine for the PLOGI ACC that detects the state
change when the driver starts discovery on the node again and drop the
REGLOGIN mailbox command.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Remove variable assignments. The value stored in local variable _rc_ is
overwritten at line 2448:rc = lpfc_sli4_bsg_set_link_diag_state(phba,
0); before it can be used.

Addresses-Coverity-ID: 1226935
Signed-off-by: NGustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Currently, calls to nvmet_fc_rcv_fcp_req() always copied the
FC-NVME cmd iu to a temporary buffer before returning, allowing
the driver to immediately repost the buffer to the hardware.

To address timing conditions on queue element structures vs async
command reception, the nvmet_fc transport occasionally may need to
hold on to the command iu buffer for a short period. In these cases,
the nvmet_fc_rcv_fcp_req() will return a special return code
(-EOVERFLOW). In these cases, the LLDD must delay until the new
defer_rcv lldd callback is called before recycling the buffer back
to the hw.

This patch adds support for the new nvmet_fc transport defer_rcv
callback and recognition of the new error code when passing commands
to the transport.
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

The PCI pool API is deprecated. This commit replaces the PCI pool old
API by the appropriate function with the DMA pool API. It also updates
some comments, accordingly.
Signed-off-by: NRomain Perier <romain.perier@collabora.com>
Reviewed-by: NPeter Senna Tschudin <peter.senna@collabora.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Signed-off-by: NDmitry Monakhov <dmonakhov@openvz.org>
Reviewed-by: NMartin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

If lpfc_nvmet_unsol_fcp_issue_abort() fails then we accidentally
increment "tgtp->xmt_abort_rsp_error" and then two lines later we
increment it a second time.

Fixes: 547077a4 ("scsi: lpfc: Adding additional stats counters for nvme.")
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Acked-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

We're calling spin_lock_irq() multiple times, the problem is that on the
first spin_unlock_irq() then we will re-enable IRQs and we don't want
that.

Fixes: 966bb5b7 ("scsi: lpfc: Break up IO ctx list into a separate get and put list")
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Acked-by: NJames Smart <james.smart@broadcom.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

A change in remote port removal introduced a spurious put which can
cause a premature structure teardown. The affects were most notable when
the driver attempted to unload as a null pointer would be hit.

Fix by removing the unnecessary put.
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

On link down, transport is calling driver to abort outstanding ios.
Driver erroneously rejects the abort if the port indicates it isn't
logged in - which will be the case after the link down. Thus, the io
can't clean up. This prevents reconnection at the transport level.

Fix by allowing abort to proceed.
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>

Set lpfc driver revision to 11.4.0.1
Signed-off-by: NDick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NHannes Reinecke <hare@suse.com>
Signed-off-by: NMartin K. Petersen <martin.petersen@oracle.com>