[SCSI] libiscsi: Reduce locking contention in fast path

Replace the session lock with two locks, a forward lock and
a backwards lock named frwd_lock and back_lock respectively.

The forward lock protects resources that change while sending a
request to the target, such as cmdsn, queued_cmdsn, and allocating
task from the commands' pool with kfifo_out.

The backward lock protects resources that change while processing
a response or in error path, such as cmdsn_exp, cmdsn_max, and
returning tasks to the commands' pool with kfifo_in.

Under a steady state fast-path situation, that is when one
or more processes/threads submit IO to an iscsi device and
a single kernel upcall (e.g softirq) is dealing with processing
of responses without errors, this patch eliminates the contention
between the queuecommand()/request response/scsi_done() flows
associated with iscsi sessions.

Between the forward and the backward locks exists a strict locking
hierarchy. The mutual exclusion zone protected by the forward lock can
enclose the mutual exclusion zone protected by the backward lock but not
vice versa.

For example, in iscsi_conn_teardown or in iscsi_xmit_data when there is
a failure and __iscsi_put_task is called, the backward lock is taken while
the forward lock is still taken. On the other hand, if in the RX path a nop
is to be sent, for example in iscsi_handle_reject or __iscsi_complete_pdu
than the forward lock is released and the backward lock is taken for the
duration of iscsi_send_nopout, later the backward lock is released and the
forward lock is retaken.

libiscsi_tcp uses two kernel fifos the r2t pool and the r2t queue.

The insertion and deletion from these queues didn't corespond to the
assumption taken by the new forward/backwards session locking paradigm.

That is, in iscsi_tcp_clenup_task which belongs to the RX (backwards)
path, r2t is taken out from r2t queue and inserted to the r2t pool.
In iscsi_tcp_get_curr_r2t which belong to the TX (forward) path, r2t
is also inserted to the r2t pool and another r2t is pulled from r2t
queue.

Only in iscsi_tcp_r2t_rsp which is called in the RX path but can requeue
to the TX path, r2t is taken from the r2t pool and inserted to the r2t
queue.

In order to cope with this situation, two spin locks were added,
pool2queue and queue2pool. The former protects extracting from the
r2t pool and inserting to the r2t queue, and the later protects the
extracing from the r2t queue and inserting to the r2t pool.

Signed-off-by: Shlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
[minor fix up to apply cleanly and compile fix]
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] libiscsi: Restructure iscsi_tcp r2t response logic

Restructure the iscsi_tcp_r2t_rsp routine in order to avoid allocating
r2t from r2tpool.queue and returning it back in case the parameters
rhdr->data_length and or rhdr->data_offset prohibit the requing.

Since the values of these parameters are known prior to the allocation,
we can pre-check and thus avoid futile allocations.

[jejb: checkpatch fixes]
Signed-off-by: Shlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] cxgb4i: Use cxgb4_select_ntuple to correctly calculate ntuple fields

Fixed calculates wrong tuple values on T5 adapter: switch to use the exported
API cxgb4_select_ntuple() from cxgb4 base driver.

Signed-off-by: Karen Xie <kxie@chelsio.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid_sas: check return value for megasas_get_pd_list()

When megasas_get_pd_list() fails we cannot detect any drives,
so we should be checking the return value accordingly.

[jejb: checkpatch fix]
Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid_sas_fusion: Return correct error value in megasas_get_ld_map_info()

When no HBA is found we should be returning '-ENXIO' to be consistent
with the other return values.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid_sas_fusion: correctly pass queue info pointer

The pointer to the queue info structure is potentially
a 64-bit value, so we should be using the correct macros
to set the values in the init frame.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] isci: update version to 1.2

The version of isci driver has not been updated for 2 years.
It was 83 isci commits ago. Suspend/resume support has been implemented
and many bugs have been fixed since 1.1. Now update the version to 1.2.

Signed-off-by: Lukasz Dorau <lukasz.dorau@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Maciej Patelczyk <maciej.patelczyk@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] bfa: set correct command return code

For various error conditions the bfa driver just returns
'DID_ERROR', which carries no information at all about the
actual source of error.
This patch updates the error handling to return a correct
error code, depending on the type of error occurred.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: Vijaya Mohan Guvva <vmohan@brocade.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Update driver version to 5.04.00-k4

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fix sparse warnings

Fix following sparse warnings:-
 drivers/scsi/qla4xxx/ql4_os.c:2109:33: warning: cast truncates bits from constant value (ffff7fff becomes 7fff)
 drivers/scsi/qla4xxx/ql4_os.c:2306:33: warning: cast truncates bits from constant value (ffff7fff becomes 7fff)

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Handle IPv6 AEN notifications

Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Update print statements in func qla4xxx_do_dpc()

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Update print statements in func qla4xxx_eh_abort()

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Update print statements in qla4xxx_mailbox_command()

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Updated print for device login, logout path

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Remove unused code from qla4xxx_set_ifcb()

Removing unused code as FW does not need any value in mbox-5.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fix failure of mbox 0x31

Issue:
While unloading driver MBOX 0x31 fail as DDB logout (MBOX 0x56)
operation is not completed.

Fix:
Wait for DDB Logout completion before MBOX 0x31

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Reduce rom-lock contention during reset recovery.

Issue:
Driver holds rom-lock for too long during reset recovery.

During adapter reset testing, it was found that the driver
holds the rom-lock for too long, because of which other
drivers fail to acquire the rom-lock, leading to reset
failures.
The primary cause is, in the bootstrap code, while
holding the rom-lock, the driver checks if the peg is
halted, causing a 2 second contention.

Fix:
When a reset recovery starts, the driver deduces the cause, and
sets appropriate flags in watchdog & recover_adapter routines.
This flag should be used to determine if bootstrap is invoked
from probe or reset context, reducing the rom-lock footprint of
the drivers.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Driver not able to collect minidump for ISP84xx

Issue:
minidump data collection fails as driver reports data mismatch

Fix:
When the driver encounters a new entry type that it cannot process,
it should just skip the entry and adjust the total buffer size by
subtracting the skipped bytes from it. This is to ensure that there
is no data mismatch because of the new entries.

Signed-off-by: Tej Parkash <tej.parkash@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fix pending IO completion in reset path before initiating chip reset

Issue:
Pending IO wait does not complete after triggering Graceful reset,
causing ack timeout and call traces.

Fix:
1. Reducing the IO command wait timeout before triggering reset,
   as logically also timeout should be less than reset timeout (10sec).
2. Moving the abort IO after chip reset, because only after
   chip reset, driver owns the IO otherwise it is with firmware and can
   still revert back with response.

Signed-off-by: Tej Parkash <tej.parkash@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fix processing response queue during probe

Issue:
While booting with kdump kernel, driver receive IOCB interrupts
for which it is not ready which results in processing them
before init_firmware during driver probe

Fix:
Two steps solution
1. Make driver ready to process the interrupt before interupts
   handlers is registered.
2. Stop driver processing iocb interrupts if not generated as per
   firmware protocol i.e R2H bit set

Signed-off-by: Tej Parkash <tej.parkash@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fix failure of IDC Time Extend mailbox command

Issue:
Mailbox command 0x102 (IDC Time Extend) failure seen while applying
changes to iface using iscsiadm

Fix:
Added fix to extend IDC timeout only for ISP84xx when IDC ACK
needs to be posted and disable ACB mailbox command completion
is in intermediate state.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Clear DDB index map upon connection close failure

Issue:
qla4xxx Unable to clear DDB indices when logout fails due to
failure of connection close mbox command.

Root cause:
If login to session fail, iscsiadm make call to destroy_session.
qla4xxx driver does not free ddb index map before free_ddb()

Fix:
Clear DDB Index map before free_ddb in "destroy_session"
in case of connection close mailbox command failure with 4005h.

Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Return correct error status from func qla4xxx_request_irqs()

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fixed AER reset sequence for ISP83xx/ISP84xx

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Correctly handle msleep_interruptible

If waiting for signals was interrupted then the device was put to
FAILED state. Use msleep instead of msleep_interruptible to handle
this correctly.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Rename ACB_STATE macros with IP_ADDRSTATE macros

Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Improve loopback failure messages

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Use IDC_CTRL bit1 directly instead of AF_83XX_NO_FWDUMP flag.

Removed AF_83XX_NO_FWDUMP flag and directly checking IDC_CTRL bit1
while taking minidump, to check for graceful reset.

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Fix comments in code

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Print WARN_ONCE() if iSCSI function presence bit removed

Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: ISP8xxx: Correct retry of adapter initialization

Issue:

For ISP8xxx, adapter initialization is not retried if
qla4xxx_initialize_adapter fails.

Fix:

If qla4xxx_initialize_adapter fails, first check if failure is due to IRQs not
attached in order to skip retrial, then free the IRQs and then retry
initializing the adapter.

Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi : Bump the driver version

Signed-off-by: John Soni Jose <sony.john-n@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi : Fix DMA Out of SW-IOMMU space error

Setting DMA bit mask 64 and roll back to 32 if not supported.

Signed-off-by: Minh Tran <minhduc.tran@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi: Fix scsi_cmnd leakage in driver.

scsi_cmnd n io_task was not NULL when
 - Link goes down while IO was happening and session is closed.
 - Task for which TMF was sent.

Signed-off-by: John Soni Jose <sony.john-n@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi: Fix the session cleanup when reboot/shutdown happens

In iSCSI Boot scenario, when machine is reboot/shutdown phase
the active sessions are not closed. Driver queue cleanup is
done as part of unload and device is disabled.

Sessions are still active, iSCSI commands are issued from
session which comes to driver, as driver cleanup and device
disabled there is kernel stack dump with errors.

Fix is invoking iscsi_session_failure with ISCSI_ERR_INVALID_HOST
on all the active sessions when shutdown routine is called.

Signed-off-by: John Soni Jose <sony.john-n@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi: Fix doorbell format for EQ/CQ/RQ s per SLI spec.

The doorbel format has been updated to support additonal functionalities
of SKH-R adapter. These changes are made such that older FW also works fine.

Signed-off-by: John Soni Jose <sony.john-n@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi: Fix port speed typo in driver.

The 100Mbps port speed macro used was not proper.

Signed-off-by: John Soni Jose <sony.john-n@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] be2iscsi: Fix handling timed out MBX completion from FW

When an MBX command timeout happens,the resources associated with
the MBX command were freed. If FW were to give the response to
host after the timeout value set by driver then driver crashes as the MBX Cmd
resources were already freed.

This patch fixes this issue by maintaing a state flag for each of
the MBX command posted/timedout/completed.

Signed-off-by: John Soni Jose <sony.john-n@emulex.com>
Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: Add support for a few HP Storage controllers

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add HP/3PAR vendor id to pci_ids.h

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa add sysfs debug switch for raid map debugging messages

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: improve error messages for driver initiated commands

On encountering unexpected error conditions from driver initiated
commands, print something useful like CDB and sense data rather than
something useless like the kernel virtual address of the command buffer.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: only do device rescan for certain events

Do no rescan on every events -- way too many rescans are
triggered if we don't filter the events.  Limit rescans
to be triggered by the following set of events:

 * controller state change
 * enclosure hot plug
 * physical drive state change
 * logical drive state change
 * redundant controller state change
 * accelerated io enabled/disabled
 * accelerated io configuration change

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: when switching out of accel mode await only accel command completions

Don't wait for *all* commands to complete, only for accelerated mode
commands.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add controller base data-at-rest encryption compatibility ioaccel2

Add controller-based data-at-rest encryption compatibility
to ioaccel2 path (HP SSD Smart Path).

Encryption feature requires driver to supply additional fields
for encryption enable, tweak index, and data encryption key index
in the ioaccel2 request structure.

Encryption enable flag and data encryption key index come from
raid_map data structure from raid offload command.

During ioaccel2 submission, check device structure's raid map to see if
encryption is enabled for the device. If so, call new function below.

Add function set_encrypt_ioaccel2 to set encryption flag, data encryption key
index, and calculate tweak value from request's logical block address.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: update source file copyrights

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: retry certain ioaccel error cases on the RAID path

Change the handling of HP SSD Smart Path errors with status:
  0x02 CHECK CONDITION
  0x08 BUSY
  0x18 RESERVATION CONFLICT
  0x40 TASK ABORTED
So that they get retried on the RAID Path.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: do not inquire for unsupported ioaccel status vpd page

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: allow VPD page zero to be queried

Code was confused and assumed that page zero was not
VPD page and all non-zero pages were VPD pages.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: rescan devices on ioaccel2 error

Allow driver to schedule a rescan whenever a request fails on the ioaccel2 path.
This eliminates the possibility of driver getting stuck in non-ioaccel mode.

IOaccel mode (HP SSD Smart Path) is disabled by driver upon error detection.
Driver relied on idea that request would be retried through normal path, and a
subsequent error would occur on that path, and be processed by controller
firmware.  As part of that process, controller disables ioaccel mode and later
reinstates it, signalling driver to change modes.

In some error cases, the error will not duplicate on the standard path,
so the driver could get stuck in non-ioaccel mode.
To avoid that, we allow driver to request a rescan during the next run of the
rescan thread.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: allow user to disable accelerated i/o path

Allow SSD Smart Path for a controller to be disabled by
the user, regardless of settings in controller firmware
or array configuration.

To disable:     echo 0 > /sys/class/scsi_host/host<id>/acciopath_status
To re-enable:   echo 1 > /sys/class/scsi_host/host<id>/acciopath_status
To check state: cat /sys/class/scsi_host/host<id>/acciopath_status

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: complete the ioaccel raidmap code

Load balance across members of a N-way mirror set, and
handle the meta-RAID levels: R10, R50, R60.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: make device update copy the raid map also

Otherwise we could wind up using incorrect raid map data, and
then very bad things would likely happen.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add task management for ioaccel mode 2

Underlying firmware cannot handle task abort on accelerated path (SSD Smart Path).
Change abort requests for accelerated path commands to physical target reset.
Send reset request on normal IO path.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Mike Miller <michael.miller@canonical.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: teach hpsa_device_reset to do either target or lun reset

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: get ioaccel mode 2 i/o working

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Joe Handzik <Joseph.T.Handzik@hp.com>
Signed-off-by: Mike Miller <michael.miller@canonical.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: initialize controller to perform io accelerator mode 2

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: get physical device handles for io accel mode 2 as well as mode 1

Signed-off-by: Mike MIller <michael.miller@canonical.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: do ioaccel mode 2 resource allocations

Signed-off-by: Mike Miller <michael.miller@canonical.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: Acknowledge controller events in ioaccell mode 2 as well as mode 1

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add ioaccel mode 2 structure definitions

Signed-off-by: Mike Miller <michael.miller@canonical.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: complain if physical or logical aborts are not supported

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add hp_ssd_smart_path_enabled sysfs attribute

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: do not rescan controllers known to be locked up

* Do not check event bits on locked up controllers to
  see if they need to be rescanned.
* Do not initiate any device rescans on controllers
  which are known to be locked up.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: poll controller to detect device change event

For shared SAS configurations, hosts need to poll Smart Arrays
periodically in order to be able to detect configuration changes
such as logical drives being added or removed from remote hosts.
A register on the controller indicates when such events have
occurred, and the driver polls the register via a workqueue
and kicks off a rescan of devices if such an event is detected.
Additionally, changes to logical drive raid offload eligibility
are autodetected in this way.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: update raid offload status on device rescan

When rescanning for logical drives, store information about whather
raid offload is enabled for each logical drive, and update the driver's
internal record of this.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add ioaccell mode 1 RAID offload support.

This enables sending i/o's destined for RAID logical drives
which can be serviced by a single physical disk down a different,
faster i/o path directly to physical drives for certain logical
volumes on SSDs bypassing the Smart Array RAID stack for a
performance improvement.

Signed-off-by: Matt Gates <matthew.gates@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Mike Miller <michael.miller@canonical.com>
Signed-off-by: Don Brace <brace@beardog.cce.hp.com>
Signed-off-by: Joe Handzik <joseph.t.handzik@hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: fix task management for mode-1 ioaccell path

For "mode 1" io accelerated commands, the command tag is in
a different location than for commands that go down the normal
RAID path, so the abort handler needs to take this into account.

Signed-off-by: Scott Teel <scott.teel@hp.com>
Signed-off-by: Mike Miller <michael.miller@canonical.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: only allow REQ_TYPE_FS to use fast path

When commands sent down the "fast path" fail, they must be re-tried down the
normal RAID path.  We do this by kicking i/o's back to the scsi mid layer with
a DID_SOFT_ERROR status, which causes them to be retried.  This won't work for
SG_IO's and other non REQ_TYPE_FS i/o's which could get kicked all the way back
to the application, which may have no idea that the command needs resubmitting
and likely no way to resubmit it in such a way the that driver can recognize it
as a resubmit and send it down the normal RAID path.  So we just always send
non REQ_TYPE_FS i/o's down the normal RAID path, never down the "fast path".

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: add support for 'fastpath' i/o

For certain i/o's to certain devices (unmasked physical disks) we
can bypass the RAID stack firmware and do the i/o to the device
directly and it will be faster.

Signed-off-by: Matt Gates <matthew.gates@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: mark last scatter gather element as the last

This is normally optional, but for SSD Smart Path support (in
subsequent patches) it is required.

Signed-off-by: Matt Gates <matthew.gates@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: use extended report luns command for HP SSD SmartPath

There is an extended report luns command which contains
additional information about physical devices.  In particular
we need to get the physical device handle so we can use an
alternate i/o path for fast physical devices like SSDs so
we can speed up certain i/o's by bypassing the RAID stack
code in the controller firmware.

Signed-off-by: Matt Gates <matthew.gates@hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] pm80xx: Spinlock fix

spin_lock_irqsave for the HBA lock is called in one function where flag
is local to that function. Another function is called from the first
function where lock has to be released using spin_unlock_irqrestore for
calling task_done of libsas. In the second function also flag is declared
and used. For calling task_done there is no need to enable the irq. So
instead of using spin_lock_irqsave and spin_unlock_irqrestore, spin_lock
and spin_unlock is used now. This also avoids passing the flags across all
the functions where HBA lock is being used. Also removed redundant code.

Reported-by: Jason Seba <jason.seba42@gmail.com>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Suresh Thiagarajan <Suresh.Thiagarajan@pmcs.com>
Signed-off-by: Viswas G <viswas.g@pmcs.com>
Acked-by: Jack Wang <xjtuwjp@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: fixup MSI-X registration

Commit 254f796b9f22b1944c64caabc356a56caaa2facd updated
the driver to use 16 MSI-X vectors, despite the fact that
older controllers would provide only 4.
This was causing MSI-X registration to drop down to INTx
mode. But as the controller support performant mode, the
initialisation will become confused and cause the machine
to stall during boot.

This patch fixes up the MSI-X registration to re-issue
the pci_enable_msix() call with the correct number of
MSI-X vectors. With that the hpsa driver continues to
works on older controllers like the P200.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Acked-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] aacraid: kdump fix

This patch fixes kernel panic issue while booting into the kdump kernel.

We have triggered crash and kdump vmcore was successful. No issues seen while
booting into the OS.

Signed-off-by: Mahesh Rajashekhara <Mahesh.Rajashekhara@pmcs.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid: missing bounds check in mimd_to_kioc()

pthru32->dataxferlen comes from the user so we need to check that it's
not too large so we don't overflow the buffer.

Reported-by: Nico Golde <nico@ngolde.de>
Reported-by: Fabian Yamaguchi <fabs@goesec.de>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Sumit Saxena <sumit.saxena@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] iscsi: fix wrong order of opcode and itt in iscsi_handle_reject prompt

This patch makes reject messages show right value for opcode and itt, which
is converse previously.

Signed-off-by: Vaughan Cao <vaughan.cao@oracle.com>
Acked-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] bfa: Fix smatch warnings

Fixed following smatch warnings in bfa.
drivers/scsi/bfa/bfa_ioc.c:3882 bfa_sfp_show_comp() error: memcpy()
'des' too small (64 vs 248)
drivers/scsi/bfa/bfa_ioc.c:6859 bfa_flash_status_read() warn: unsigned
'status' is never less than zero.
drivers/scsi/bfa/bfa_ioc.c:6881 bfa_flash_status_read() warn: unsigned
'status' is never less than zero.
drivers/scsi/bfa/bfa_ioc.c:6917 bfa_flash_read_start() warn: unsigned
'status' is never less than zero.
drivers/scsi/bfa/bfa_ioc.c:7043 bfa_flash_raw_read() warn: unsigned
'status' is never less than zero.

Signed-off-by: Vijaya Mohan Guvva <vmohan@brocade.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] bfa: fix missing unlock on error in bfad_iocmd_cfg_trunk()

Add the missing unlock before return from function bfad_iocmd_cfg_trunk()
in the error handling case.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Vijaya Mohan Guvva <vmohan@brocade.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] scsi_error: disable eh_deadline if no host_reset_handler is set

When the host template doesn't declare an eh_host_reset_handler
the eh_deadline mechanism is pointless and will set the
device to offline. So disable eh_deadline if no
eh_host_reset_handler is present.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] dual scan thread bug fix

In the highly unusual case where two threads are running concurrently through
the scanning code scanning the same target, we run into the situation where
one may allocate the target while the other is still using it.  In this case,
because the reap checks for STARGET_CREATED and kills the target without
reference counting, the second thread will do the wrong thing on reap.

Fix this by reference counting even creates and doing the STARGET_CREATED
check in the final put.

Tested-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org # delay backport for 2 months for field testing
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fix our current target reap infrastructure

This patch eliminates the reap_ref and replaces it with a proper kref.
On last put of this kref, the target is removed from visibility in
sysfs.  The final call to scsi_target_reap() for the device is done from
__scsi_remove_device() and only if the device was made visible.  This
ensures that the target disappears as soon as the last device is gone
rather than waiting until final release of the device (which is often
too long).

Reviewed-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org # delay backport by 2 months for field testing
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] pm80xx: Enable BAR shift to avoid BIOS conflict with MPI space for ATTO pm8001 based HBAs.

Signed-off-by: Bradley Grove <bgrove@attotech.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] pm80xx: Read saved WWN from NVMD for ATTO pm8001 based HBAs.

Signed-off-by: Bradley Grove <bgrove@attotech.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] esas2r: fix potential format string flaw

This makes sure format strings cannot leak into the printk call via the
constructed buffer.

Signed-off-by: Kees Cook <keescook@chromium.org>
Acked-by: Bradley Grove <bgrove@attotech.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc: use NULL instead of 0 for pointer

sparse says:
drivers/scsi/lpfc/lpfc_sli.c:16547:37: warning:
 Using plain integer as NULL pointer

Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
Acked-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc: remove self-assignments

Report from coverity

Reported-by: Dave Jones <davej@redhat.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Update lpfc version to driver version 8.3.45

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Fixed crash during driver unload.

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Fixed driver error messages after firmware download

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Fixed missing initialization for task management IOCBs

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Fix sysfs buffer overrun in read of lpfc_fcp_cpu_map for 128 CPUs.

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Incorporate changes to use reason in change_queue_depth function.

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Incorporated support of a low-latency io path

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.45: Added dport mailbox pass through support.

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.44: Update lpfc version to driver version 8.3.44

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.44: Fixed unassigned variable in ELS timeout message

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.44: Fixed incorrect allocation of iDiags directories/files in debugfs

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] lpfc 8.3.44: Fix kernel panics from corrupted ndlp list

Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>