RSCN processing in qla2xxx driver can run in parallel with ELS/IO
processing. As such the decision to remove disappeared fc port's
session could be stale, because a new login sequence has occurred
since and created a brand new session.

Previous mechanism of dealing with this by delaying deletion request
was prone to erroneous deletions if the event that was supposed to
cancel the deletion never arrived or has been delayed in processing.

New mechanism relies on a time-like generation counter to serialize
RSCN updates relative to ELS/IO updates.

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NAlexei Potashnik <alexei@purestorage.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

cancel any commands from initiator's s_id that are still waiting
on qla_tgt_wq when PLOGI arrives.

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NAlexei Potashnik <alexei@purestorage.com>
Acked-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

- keep qla_tgt_sess object on the session list until it's freed

- modify use of sess->deleted flag to differentiate delayed
  session deletion that can be cancelled from irreversible one:
  QLA_SESS_DELETION_PENDING vs QLA_SESS_DELETION_IN_PROGRESS

- during IN_PROGRESS deletion all newly arrived commands and TMRs will
  be rejected, existing commands and TMRs will be terminated when
  given by the core to the fabric or simply dropped if session logout
  has already happened (logout terminates all existing exchanges)

- new PLOGI will initiate deletion of the following sessions
  (unless deletion is already IN_PROGRESS):
  - with the same port_name (with logout)
  - different port_name, different loop_id but the same port_id
    (with logout)
  - different port_name, different port_id, but the same loop_id
    (without logout)

- additionally each new PLOGI will store imm notify iocb in the
  same port_name session being deleted. When deletion process
  completes this iocb will be acked. Only the most recent PLOGI
  iocb is stored. The older ones will be terminated when replaced.

- new PRLI will initiate deletion of the following sessions
  (unless deletion is already IN_PROGRESS):
  - different port_name, different port_id, but the same loop_id
   (without logout)

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NAlexei Potashnik <alexei@purestorage.com>
Acked-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Since cmds go into qla_tgt_wq and TMRs don't, it's possible that TMR
like TASK_ABORT can be queued over the cmd for which it was meant.
To avoid this race, use a per-port list to keep track of cmds that
are enqueued to qla_tgt_wq but not yet processed. When a TMR arrives,
iterate through this list and remove any cmds that match the TMR.
This patch supports TASK_ABORT and LUN_RESET.

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NSwapnil Nagle <swapnil.nagle@purestorage.com>
Signed-off-by: NAlexei Potashnik <alexei@purestorage.com>
Acked-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

To fix some issues talking to ESX, this patch modifies the qla2xxx driver
so that it never logs into remote ports.  This has the side effect of
getting rid of the "rports" entirely, which means we never log out of
initiators and never tear down sessions when an initiator goes away.

This is mostly OK, except that we can run into trouble if we have
initiator A assigned FC address X:Y:Z by the fabric talking to us, and
then initiator A goes away.  Some time (could be a long time) later,
initiator B comes along and also gets FC address X:Y:Z (which is
available again, because initiator A is gone).  If initiator B starts
talking to us, then we'll still have the session for initiator A, and
since we look up incoming IO based on the FC address X:Y:Z, initiator B
will end up using ACLs for initiator A.

Fix this by:

 1. Handling RSCN events somewhat differently; instead of completely
    skipping the processing of fcports, we look through the list, and if
    an fcport disappears, we tell the target code the tear down the
    session and tell the HBA FW to release the N_Port handle.

 2. Handling "port down" events by flushing all of our sessions.  The
    firmware was already releasing the N_Port handle but we want the
    target code to drop all the sessions too.

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NRoland Dreier <roland@purestorage.com>
Signed-off-by: NAlexei Potashnik <alexei@purestorage.com>
Acked-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NKanoj Sarcar <kanoj.sarcar@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Remove unnecessary msleep from qlt_send_term_exchange as it
adds latency of 250 msec while sending terminate exchange to
an aborted task.

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NGiridhar Malavali <giridhar.malavali@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Adjust debug flag to match debug comment.
Signed-off-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Request IOCB queue element(s) is reserved during
good path IO.  Under error condition such as unable
to allocate IOCB handle condition, the IOCB count
that was reserved is not released.

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

After updating the consumer index of ATIO Q, a read is
required to flush the write to the adapter register.
Signed-off-by: NQuinn Tran <quinn.tran@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: NGiridhar Malavali <giridhar.malavali@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

[ Upstream commit ef86cb20 ]

This patch fixes a kernel panic for qla2xxx Target core
Module driver introduced by a fix in the qla2xxx initiator code.

Commit ef86cb20 ("qla2xxx: Mark port lost when we receive an RSCN for it.")
introduced the regression for qla2xxx Target driver.

Stack trace will have following signature

 --- <NMI exception stack> ---
[ffff88081faa3cc8] _raw_spin_lock_irqsave at ffffffff815b1f03
[ffff88081faa3cd0] qlt_fc_port_deleted at ffffffffa096ccd0 [qla2xxx]
[ffff88081faa3d20] qla2x00_schedule_rport_del at ffffffffa0913831[qla2xxx]
[ffff88081faa3d50] qla2x00_mark_device_lost at ffffffffa09159c5[qla2xxx]
[ffff88081faa3db0] qla2x00_async_event at ffffffffa0938d59 [qla2xxx]
[ffff88081faa3e30] qla24xx_msix_default at ffffffffa093a326 [qla2xxx]
[ffff88081faa3e90] handle_irq_event_percpu at ffffffff810a7b8d
[ffff88081faa3ee0] handle_irq_event at ffffffff810a7d32
[ffff88081faa3f10] handle_edge_irq at ffffffff810ab6b9
[ffff88081faa3f30] handle_irq at ffffffff8100619c
[ffff88081faa3f70] do_IRQ at ffffffff815b4b1c
 --- <IRQ stack> ---

Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: NSaurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: NHimanshu Madhani <himanshu.madhani@qlogic.com>
Reviewed-by: NNicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

As a follow up to ce31c1b0 - there are
still a few LIO match_int() calls that don't check the return value.
Propagate errors rather than using the potentially uninitialised result.
Signed-off-by: NDavid Disseldorp <ddiss@suse.de>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Here are some more instances where we are returning 0 from a configfs
store function, the unintended result of which is likely infinite retries
from userspace.
Signed-off-by: NAndy Grover <agrover@redhat.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

See https://bugzilla.redhat.com/show_bug.cgi?id=1240687

Returning 0 from a configfs store function results in infinite retries.
Reported-by: NYanko Kaneti <yaneti@declera.com>
Signed-off-by: NAndy Grover <agrover@redhat.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

API compliance scanning with coccinelle flagged:
./drivers/scsi/qla2xxx/tcm_qla2xxx.c:407:2-29:
         WARNING: timeout is HZ dependent

This was introduced in 'commit 75f8c1f6 ("[SCSI] tcm_qla2xxx: Add >=
24xx series fabric module for target-core")'. wait_for_completion_timeout()
expects a timeout in jiffies so the numeric constant makes the effective
timeout HZ dependent. Resolved by converting it to CONST * HZ.
Signed-off-by: NNicholas Mc Guire <hofrat@osadl.org>
Acked-by: NNilesh Javali <nilesh.javali@qlogic.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Since a RAM backend device is not really a rotational device,
we set it as is_nonrot=1 which will be forwarded in VPD page 0xb1
(block device characteristics) response.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Cc: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

LIO supports protection types 1,3 so setting a hard-coded SPT=3
is fine for now.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Triggered a compilation warning.

Fixes: 2650d71e target: move transport ID handling to the core
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

When split BAR is enabled, the driver needs to dump out the split BAR
registers rather than the original 64bit BAR registers.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

The unsafe doorbell and scratchpad access should display reason when
WARN is called.  Otherwise we get a stack dump without any explanation.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Printouts driver name and version to indicate what is being loaded.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Benchmarking showed a significant performance increase with the MTU size
to 64k instead of 16k.  Change the driver default to 64k.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Instead of using the platform code names, use the correct platform names
to identify the respective Intel NTB hardware.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Disable DMA usage by default, since the CPU provides much better
performance with write combining.  Provide a module parameter to enable
DMA usage when offloading the memcpy is preferred.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Changing the memory window BAR mappings to write combining significantly
boosts the performance.  We will also use memcpy that uses non-temporal
store, which showed performance improvement when doing non-cached
memcpys.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Allocate memory for the NUMA node of the NTB device.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Allocate memory and request the DMA channel for the same NUMA node as
the NTB device.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

When the ntb transport is connecting and waiting for the peer, the debug
console receives lots of debug level messages about the remote qp link
status being down.  Rate limit those messages.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

This is a simple debugging driver that enables the doorbell and
scratch pad registers to be read and written from the debugfs.  This
tool enables more complicated debugging to be scripted from user space.
This driver may be used to test that your ntb hardware and drivers are
functioning at a basic level.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

This is a simple ping pong driver that exercises the scratch pads and
doorbells of the ntb hardware.  This driver may be used to test that
your ntb hardware and drivers are functioning at a basic level.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Add module parameters for the addresses to be used in B2B topology.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Reset the link stats when the link goes down.  In particular, the TX and
RX index and count must be reset, or else the TX side will be sending
packets to the RX side where the RX side is not expecting them.  Reset
all the stats, to be consistent.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

On link down, don't advance RX index to the next entry.  The next entry
should never be valid after receiving the link down flag.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

The same message "qp %d: Link Down\n" was printed at two locations in
ntb_transport.  Change the messages so they are distinct.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Set errata flags for the specific device IDs to which they apply,
instead of the whole Xeon hardware class.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Link training should be enabled in the driver probe for root port mode.
We should not have to wait for transport to be loaded for this to
happen.  Otherwise the ntb device will not show up on the transparent
bridge side of the link.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

The transport was writing and then reading the peer scratch pad,
essentially reading what it just wrote instead of exchanging any
information with the peer.  The transport expects the peer values to be
the same as the local values, so this issue was not obvious.
Signed-off-by: NDave Jiang <dave.jiang@intel.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>

Change ntb_hw_intel to use the new NTB hardware abstraction layer.

Split ntb_transport into its own driver.  Change it to use the new NTB
hardware abstraction layer.
Signed-off-by: NAllen Hubbe <Allen.Hubbe@emc.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>