[SCSI] Merge tag 'fcoe-02-19-13' into for-linus

FCoE Updates for 3.9

Signed-off-by: James Bottomley <JBottomley@Parallels.com>

Merge tag 'uapi-20121219' into for-linus

UAPI Disintegration 2012-12-19

This is the remaining SCSI part of the UAPI

Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ufs: Separate PCI code into glue driver

This patch separates PCI code from ufshcd.c and makes it as a
core driver module and adds a new file ufshcd-pci.c as PCI glue
driver.

[jejb: strip __devinit and devexit_p()]
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Namjae Jeon <linkinjeon@gmail.com>
Reviewed-by: Subhash Jadavani <subhashj@codeaurora.org>
Tested-by: Maya Erez <merez@codeaurora.org>
Signed-off-by: Vinayak Holikatti <vinholikatti@gmail.com>
Signed-off-by: Santosh Yaraganavi <santoshsy@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ufs: Segregate PCI Specific Code

This patch segregates the PCI specific code in ufshcd.c to make it
ready for splitting into core ufs driver and PCI glue driver. Also
copyright header modification to remove extra warranty disclaim.

Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Namjae Jeon <linkinjeon@gmail.com>
Reviewed-by: Subhash Jadavani <subhashj@codeaurora.org>
Tested-by: Maya Erez <merez@codeaurora.org>
Signed-off-by: Vinayak Holikatti <vinholikatti@gmail.com>
Signed-off-by: Santosh Yaraganavi <santoshsy@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] scsi: fix lpfc build when wmb() is defined as mb()

On architectures where wmb() is defined as mb(), a build error
happens since there is also a variable named 'mb' in lpfc_sli.c's
lpfc_sli_issue_mbox_s3() function.  Rename the variable to 'mbx'
to prevent the build error.

drivers/scsi/lpfc/lpfc_sli.c: error: called object 'mb' is not a function

Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Acked-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] storvsc: Handle dynamic resizing of the device

Handle LUN size changes by re-scanning the device.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] storvsc: Restructure error handling code on command completion

In preparation for handling additional sense codes, restructure and cleanup
the error handling code in the command completion code path.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] storvsc: avoid usage of WRITE_SAME

Set scsi_device->no_write_same because the host does not support it.
Also blacklist WRITE_SAME to avoid (and log) accident usage.

If the guest uses the ext4 filesystem, storvsc hangs while it prints
these messages in an endless loop:
...
[  161.459523] hv_storvsc vmbus_0_1: cmd 0x41 scsi status 0x2 srb status 0x6
[  161.462157] sd 2:0:0:0: [sda]
[  161.463135] Sense Key : No Sense [current]
[  161.464983] sd 2:0:0:0: [sda]
[  161.465899] Add. Sense: No additional sense information
[  161.468211] hv_storvsc vmbus_0_1: cmd 0x41 scsi status 0x2 srb status 0x6
[  161.475766] sd 2:0:0:0: [sda]
[  161.476728] Sense Key : No Sense [current]
[  161.478284] sd 2:0:0:0: [sda]
[  161.479441] Add. Sense: No additional sense information
...

This happens with a guest running on Windows Server 2012, but happens to
work while running on Windows Server 2008. WRITE_SAME isnt really
supported by both versions, so disable the command usage globally.

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] aacraid: suppress two GCC warnings

Building src.o for a 32 bit system triggers two GCC warnings:
    drivers/scsi/aacraid/src.c: In function ‘aac_src_deliver_message’:
    drivers/scsi/aacraid/src.c:410:3: warning: right shift count >= width of type [enabled by default]
    drivers/scsi/aacraid/src.c:434:2: warning: right shift count >= width of type [enabled by default]

These warnings are caused by a right shift of 32. Use upper_32_bits() to
suppress them.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Cc: Mahesh Rajashekhara <Mahesh_Rajashekhara@pmc-sierra.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] hpsa: check for dma_mapping_error in hpsa_passthru ioctls

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

[SCSI] hpsa: reorganize error handling in hpsa_passthru_ioctl

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

[SCSI] hpsa: check for dma_mapping_error in hpsa_map_sg_chain_block

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

[SCSI] hpsa: Check for dma_mapping_error for all code paths using fill_cmd

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

[SCSI] hpsa: Check for dma_mapping_error in hpsa_map_one

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

[SCSI] dc395x: uninitialized variable in device_alloc()

This bug was introduced back in bitkeeper days in 2003.  We use
"dcb->dev_mode" before it has been initialized.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Oliver Neukum <oliver@neukum.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] Fix range check in scsi_host_dif_capable()

The range checking from fe542396 was bad. We would still end up walking
beyond the array as Type 3 is defined to be 4 in the protection
bitmask. Instead use ARRAY_SIZE() for the range check.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] storvsc: Initialize the sglist

Properly initialize scatterlist before using it.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Cc: stable@vger.kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] mpt2sas: Add support for OEM specific controller

Defined SSDID & HW vendor brand strings. Added entries for SSDID within the
function that prints the brand string.

Signed-off-by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Fix oops while resetting an ipr adapter

When resetting an ipr adapter, we use scsi_block_requests to
block any new commands from scsi core, and then unblock after
the reset. When hotplug removing an adapter, we shut it down
and go through this same code, but we've seen issues with
scsi_unblock_requests running after the adapter's memory has
been freed. There is really no need to block/unblock when
the adapter is being removed, so this patch skips the
block/unblock and will immediately fail any commands that
happen to make it to queuecommand while the adapter is
being shutdown.

Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fnic: Fnic Trace Utility

Fnic Trace utility is a tracing functionality built directly into fnic driver
to trace events. The benefit that trace buffer brings to fnic driver is the
ability to see what it happening inside the fnic driver. It also provides the
capability to trace every IO event inside fnic driver to debug panics, hangs
and potentially IO corruption issues. This feature makes it easy to find
problems in fnic driver and it also helps in tracking down strange bugs in a
more manageable way. Trace buffer is shared across all fnic instances for
this implementation.

Signed-off-by: Hiral Patel <hiralpat@cisco.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fnic: New debug flags and debug log messages

Added new fnic debug flags for identifying IO state at every stage of IO while
debugging and also added more log messages for better debugging capability.

Signed-off-by: Sesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: Hiral Patel <hiralpat@cisco.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fnic: fnic driver may hit BUG_ON on device reset

The issue was observed when LUN Reset is issued through IOCTL or sg_reset
utility.

fnic driver issues LUN RESET to firmware. On successful completion of device
reset, driver cleans up all the pending IOs that were issued prior to device
reset. These pending IOs are expected to be in ABTS_PENDING state. This works
fine, when the device reset operation resulted from midlayer, but not when
device reset was triggered from IOCTL path as the pending IOs were not in
ABTS_PENDING state. execution path hits panic if the pending IO is not in
ABTS_PENDING state.

Changes:
The fix replaces BUG_ON check in fnic_clean_pending_aborts() with marking
pending IOs as ABTS_PENDING if they were not in ABTS_PENDING state and skips
if they were already in ABTS_PENDING state. An extra check is added to validate
the abort status of the commands after a delay of 2 * E_D_TOV using a
helper function. The helper function returns 1 if it finds any pending IO in
ABTS_PENDING state, belong to the LUN on which device reset was issued else 0.
With this, device reset operation returns success only if the helper funciton
returns 0, otherwise it returns failure.

Other changes:
- Removed code in fnic_clean_pending_aborts() that returns failure if it finds
  io_req NULL, instead of returning failure added code to continue with next io
- Added device reset flags for debugging in fnic_terminate_rport_io,
  fnic_rport_exch_reset, and fnic_clean_pending_aborts

Signed-off-by: Narsimhulu Musini <nmusini@cisco.com>
Signed-off-by: Hiral Patel <hiralpat@cisco.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fnic: Fix SGEs limit

Driver allows IOs with more SGEs than max SGEs supported by Palo. The current
max SGEs supported by the fnic driver is 1024. The current register settings
on Palo supports a max of 256 only. Palo would return any IO with more than
256 SGEs with an error indicating INVALID_SGLS. Fnic driver should limit the
max supported SGLs in the driver to 256 to avoid this error.

Signed-off-by: Sesidhar Baddela <sebaddel@cisco.com>
Signed-off-by: Hiral Patel <hiralpat@cisco.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] fnic: fixing issues in device and firmware reset code

1. Handling overlapped firmware resets
     This fix serialize multiple firmware resets to avoid situation where fnic
     device fails to come up for link up event, when firmware resets are issued
     back to back. If there are overlapped firmware resets are issued,
     the firmware reset operation checks whether there is any firmware reset in
     progress, if so it polls for its completion in a loop with 100ms delay.

2. Handling device reset timeout
     fnic_device_reset code has been modified to handle Device reset timeout:
     - Issue terminate on device reset timeout.
     - Introduced flags field (one of the scratch fields in scsi_cmnd).
     With this, device reset request would have DEVICE_RESET flag set for other
     routines to determine the type of the request.
     Also modified fnic_terminate_rport_io, fnic_rport_exch_rset, completion
     routines to handle SCSI commands with DEVICE_RESET flag.

3. LUN/Device Reset hangs when issued through IOCTL using utilities like
   sg_reset.
     Each SCSI command is associated with a valid tag, fnic uses this tag to
     retrieve associated scsi command on completion. the LUN/Device Reset issued
     through IOCTL resulting into a SCSI command that is not associated with a
     valid tag. So fnic fails to retrieve associated scsi command on completion,
     which causes hang. This fix allocates tag, associates it with the
     scsi command and frees the tag, when the operation completed.

4. Preventing IOs during firmware reset.
     Current fnic implementation allows IO submissions during firmware reset.
     This fix synchronizes IO submissions and firmware reset operations.
     It ensures that IOs issued to fnic prior to reset will be issued to the
     firmware before firmware reset.

Signed-off-by: Narsimhulu Musini <nmusini@cisco.com>
Signed-off-by: Hiral Patel <hiralpat@cisco.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] pm80xx: fix for memory region free

All memory regions are allocated based on variables total_len
and alignment but free was based on element_size.

Signed-off-by: Sakthivel K <DL.Team.PMC.SPC.Linux.open.source@pmcs.com>
Signed-off-by: Anand Kumar S <DL.Team.PMC.SPC.Linux.open.source@pmcs.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid_sas: Version and Changelog update

This patch updates the megaraid_sas driver version and updates
Documentation/scsi/ChangeLog.megaraid_sas.

Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid_sas: Dont load DevHandle unless FastPath enabled

The following patch for megaraid_sas will fix an issue where the
driver should not be loading the DevHandle unless FastPath is enabled.
 If FastPath was not enabled, this means the hardware raid map
validation failed for some reason, or the map was corrupted, which
could mean the DevHandle could be invalid.

Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] megaraid_sas: Add 4k FastPath DIF support

The following patch for megaraid_sas will allow Fastpath T10PI/DIF
frame builds to work with 4k sector size.

Signed-off-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Wait for IDC complete event to finish loopback operation.

Wait for the IDC complete AEN before returning the loopback operation back to
the application to make sure the port is put back into normal operations.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Integrate generic card temperature with mezz card temperature.

Give priority to I2C thermal.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Don't process state change aen for reset owner.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Don't process RSCNs for a vport on the same physical adapter.

Currently,the driver is processes RSCNs for each new NPIV ports that is created.
Processing the RSCN includes a name server query to see what has changed at the
name server side.  The name server query is performed by the physical port and
each virtual port on the physical adapter (since the RSCN is passed to each
virtual port for processing).  As the number of virtual ports being created
increases, this causes a lot of traffic and busies the firmware.

Processing the RSCN for a virtual port we already have a priori knowledge of is
not necessary so check the 24-bit fabric ID of the RSCN entry and skip
processing it if the RSCN is for a virtual port we already know about.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Add setting of driver version string for vendor application.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Correction of comment in MBC opcode defines.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Correct list-iteration bug in Report-ID Acquisition codes.

Code in qla24xx_report_id_acquisition() incorrectly assumed that
upon completion of list iteration (with no match), the 'pos'
(vp) variable passed to list_for_each_entry() would be set to
NULL.  In this context, if the firmware were to return an
unrecognized vp_idx, the follow-on assignments to vp-members
could result in corruption of the structure.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] Revert "[SCSI] qla2xxx: Optimize existing port name server query matching."

This reverts commit c0822b63ccbf3b019059f384c290b080cae859f1.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] Revert "[SCSI] qla2xxx: Avoid losing any fc ports when loop id's are exhausted."

This reverts commit 4dc77c36f86c2dc4e3f483146d33b64d12c0da3f.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: silence two GCC warnings.

Compiling qla_gs.o (part of the qla2xxx module) triggers two GCC
warnings:
    drivers/scsi/qla2xxx/qla_gs.c: In function ‘qla2x00_fdmi_rhba’:
    drivers/scsi/qla2xxx/qla_gs.c:1339:7: warning: array subscript is above array bounds [-Warray-bounds]
    drivers/scsi/qla2xxx/qla_gs.c: In function ‘qla2x00_fdmi_register’:
    drivers/scsi/qla2xxx/qla_gs.c:1663:15: warning: array subscript is above array bounds [-Warray-bounds]

It seems that the sequence of a strcpy followed by a strlen confuses GCC
when it is keeping track of array bounds here. (It is not clear to me
which array triggers this warning and by how much GCC thinks the
subscript is above its bounds. Neither is it clear to me why comparable
code in these two functions doesn't trigger this warning.)

An easy way to silence these warnings is to use preprocessor macros and
strncpy, as that apparently gives GCC enough information to keep track
of array bounds.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Update the copyright information.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Display the lock owner on lock acquire failure.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Move loopback mode reset after chip reset check.

If we need to do a chip reset because of a serious loopback error don't try to
reset the loopback mode on the port as the mailbox command will timeout.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Extra loopback error handling for ISP83xx.

Add the following error handling for loopback diagnostic mode with ISP83xx:

1. If we do not receive an MBA_DCBX_COMPLETE after our initial set port
configuration command, try to reset the port back into normal operation.
If that fails, take a FCoE dump and then reset the chip.
2. After completing the loopback diagnostic operation, if the reset of the port
back into normal operation fails then reset the port so we take a FCoE dump
and then reset the chip.
3. When we receive an IDC notification and the requested operation is loopback
extend the loop down timer so the link does not appear to down for an
extended period of time.

[jejb: fix checkpatch issue]
Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Unload hangs after issuing BSG commands to vport.

BSG code path increments ref count in the send path, but does not
decrement in the return path leading to hang during unload of the
driver.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Do MPI reset only for ISP81xx.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Do link initialization on get loop id failure.

To avoid continually doing ISP resets when get loop id fails to
obtain the adapter loop id, first try to do a link initialization.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Update the FTP site references in the driver sources.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Reject loopback request if one is already in progress.

If another function on the port has initiated a loopback operation do not
process the current request.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Print thermal unsupported message.

When the Read SFP mailbox command fails on the thermal device,
print a message explaining that thermal is not supported.

Signed-off-by: Joe Carnuccio <joe.carnuccio@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Avoid null pointer dereference in shutdown routine.

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Get VPD information from common location for CNA.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Correct race in loop_state assignment during reset handling.

There's a subtle race in the loop/bus-reset handling whereby a
VHA's loop-state can get incorrectly set to 'down' after the
loop-reset and firmware's completion of link re-negotiation.  The
original code incorrectly assumes that firmware AENs would arrive
only after mailbox-command execution to initiate the link-flap.

Here's a good case with the old code (AENs arrive after
mailbox-command completion):

qla2xxx [0000:03:00.1]-8012:91: BUS RESET ISSUED nexus=91:0:4.
qla2xxx [0000:03:00.1]-287d:91: FCPort state transitioned from ONLINE to LOST - portid=010100.
qla2xxx [0000:03:00.1]-580e:91: Asynchronous P2P MODE received.
qla2xxx [0000:03:00.1]-287d:91: FCPort state transitioned from ONLINE to LOST - portid=010400.
qla2xxx [0000:03:00.1]-802b:91: BUS RESET SUCCEEDED nexus=91:0:4.
qla2xxx [0000:03:00.1]-480b:91: Reset marker scheduled.
qla2xxx [0000:03:00.1]-5812:91: Port database changed ffff 0006 0000.
qla2xxx [0000:03:00.1]-505f:91: Link is operational (4 Gbps).
qla2xxx [0000:03:00.1]-480c:91: Reset marker end.
qla2xxx [0000:03:00.1]-480f:91: Loop resync scheduled.
qla2xxx [0000:03:00.1]-8837:91: F/W Ready - OK.
qla2xxx [0000:03:00.1]-883a:91: fw_state=3 (7, 0, 0, 0) curr time=170b8f315.
qla2xxx [0000:03:00.1]-280e:91: HBA in F P2P topology.
qla2xxx [0000:03:00.1]-2812:91: qla2x00_configure_hba success
qla2xxx [0000:03:00.1]-2814:91: Configure loop -- dpc flags = 0x5260.

notice how the 'Port database changed' (8014) arrived after the
bus-reset handler completed 'BUS RESET SUCCEEDED'.

Now, here's a failing case with the old code (AENs arrive before
mailbox-command completion):

qla2xxx [0000:03:00.1]-8012:91: BUS RESET ISSUED nexus=91:0:0.
qla2xxx [0000:03:00.1]-580e:91: Asynchronous P2P MODE received.
qla2xxx [0000:03:00.1]-287d:91: FCPort state transitioned from ONLINE to LOST - portid=010100.
qla2xxx [0000:03:00.1]-287d:91: FCPort state transitioned from ONLINE to LOST - portid=010400.
qla2xxx [0000:03:00.1]-4800:91: DPC handler sleeping.
qla2xxx [0000:03:00.1]-5812:91: Port database changed ffff 0006 0000.
qla2xxx [0000:03:00.1]-505f:91: Link is operational (4 Gbps).
qla2xxx [0000:03:00.1]-802b:91: BUS RESET SUCCEEDED nexus=91:0:0.
qla2xxx [0000:03:00.1]-480b:91: Reset marker scheduled.
qla2xxx [0000:03:00.1]-480c:91: Reset marker end.
qla2xxx [0000:03:00.1]-480f:91: Loop resync scheduled.
qla2xxx [0000:03:00.1]-8837:91: F/W Ready - OK.
qla2xxx [0000:03:00.1]-883a:91: fw_state=3 (7, 0, 0, 0) curr time=170be9eb2.
qla2xxx [0000:03:00.1]-280e:91: HBA in F P2P topology.
qla2xxx [0000:03:00.1]-2812:91: qla2x00_configure_hba success
qla2xxx [0000:03:00.1]-2814:91: Configure loop -- dpc flags = 0x5260.
qla2xxx [0000:03:00.1]-281e:91: Needs RSCN update and loop transition.
qla2xxx [0000:03:00.1]-286a:91: qla2x00_configure_loop *** FAILED ***.
qla2xxx [0000:03:00.1]-4810:91: Loop resync end.
qla2xxx [0000:03:00.1]-4800:91: DPC handler sleeping.

This race would ultimately lead to devices go unexpectedly
offline until another link-flap or chip-reset would cause driver
re-discovery to take place.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Correction to the message ids.

Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Prevent enabling target mode for unsupported HBAs.

Signed-off-by: Arun Easi <arun.easi@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Allow ISP81xx to create ATIO queues.

Signed-off-by: Arun Easi <arun.easi@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Enable target mode support for ISP83xx.

Signed-off-by: Arun Easi <arun.easi@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Ramp down queue depth for attached SCSI devices when driver resources are low.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla2xxx: Determine the number of outstanding commands based on available resources.

Base the number of outstanding requests the driver will keep track of on the
available resources instead of being hard-coded.

Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] libosd: check for kzalloc() failure

There wasn't any error handling for this kzalloc().

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: don't free NULL dma pool

The error path calls dma_pool_free() on this path but "chap_table" is
NULL and "chap_dma" is uninitialized.  It's cleaner to just return
directly.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] mvsas: fixed timeout issue when removing module

Root cause is libsas will clear asd_sas_port phy_mask value in sas_port_deform
after triggering destruct workqueue, but the workqueue will send sync cmd and
still need phy_mask value.  Now, mvsas using asd_sas_phy setting instead of
asd_sas_port setting.

Signed-off-by: James Bottomley <JBottomley@Parallels.com>

libfcoe: Check for unusable FCFs before looking for conflicting FCFs

When there are multiple FCFs in the fabric, and one of them becomes
unavailable, the fabric name for the unavailable FCF becomes 0 along
with FIP_FL_AVAIL getting reset. In this case, FCF selection logic does
not select any FCF as it first checks for conflicting FCFs (since fabric
name is 0, it fails the condition), instead of first checking if it is
usable or not. Fix it by first checking if FCF is usable and skip that
FCF, and go to the next one in the list to check if it can be selected.

Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>

libfc: XenServer fails to mount root filesystem.

schedule_delayed_work() is using msec instead of jiffies. On PLOGI
reject from target, remote port retry is scheduled @ 20 sec instead
of 2sec(FC_DEF_E_D_TOV).
XenServer dom0 kernel is configured with CONFIG_HZ=100Hz

Signed-off-by: Krishna Mohan <krmohan@cisco.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>

libfcoe: Handle CVL while waiting to select an FCF

When a CVL is received while we wait to select best FCF, we drop it
without handling it. This causes initiator and the switch to go
out-of-sync. Initiator proceeds selecting one of the FCFs and tries to
send FIP FLOGI. However the switch may reject the FLOGI, as it has
cleared its internal state, and expects the initiator to start FIP
discovery protocol. Fix this condition by resetting the fcoe
controller.

Signed-off-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Reviewed-by: Yi Zou <yi.zou@intel.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>

[SCSI] aacraid: 1024 max outstanding command support for Series 7 and above

Signed-off-by: Mahesh Rajashekhara <Mahesh_Rajashekhara@pmc-sierra.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] bnx2fc: adjust duplicate test

Delete successive tests to the same location.  The code tested the result
of a previous allocation, that itself was already tested.  It is changed to
test the result of the most recent allocation.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@s exists@
local idexpression y;
expression x,e;
@@

*if ( \(x == NULL\|IS_ERR(x)\|y != 0\) )
 { ... when forall
   return ...; }
... when != \(y = e\|y += e\|y -= e\|y |= e\|y &= e\|y++\|y--\|&y\)
    when != \(XT_GETPAGE(...,y)\|WMI_CMD_BUF(...)\)
*if ( \(x == NULL\|IS_ERR(x)\|y != 0\) )
 { ... when forall
   return ...; }
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Bhanu Prakash Gollapudi <bprakash@broadcom.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Update driver version to 5.03.00-k4

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

[SCSI] qla4xxx: Fix return code for qla4xxx_session_get_param.

While reading sysfs chap_in_idx and chap_out_idx attribute, if chap
authentication is not set we get an error as "Invalid argument".

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

[SCSI] qla4xxx: wait for boot target login response during probe.

Boot from SAN fails in some cases where boot target login response
is slow. In this case, probe finishes early and triggers init, which
doesn't find remote disk to mount root which leads to BFS failure.

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

[SCSI] qla4xxx: Added support for force firmware dump

This patch force driver to collect firmware dump using following command:-
"echo 2 > /sys/class/scsi_host/hostX/device/fw_dump"

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

[SCSI] qla4xxx: Re-register IRQ handler while retrying initialize of adapter

Problem:
If initialization of adapter fails for some reason, then the interrupt handlers
are released. The interrupt handlers are not registered again when initialization
of adapter is retried.

Solution:
Re-register the interrupt handler when adapter initialization is retried.

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

[SCSI] qla4xxx: Throttle active IOCBs to firmware limits

Problem:
QLA4xxx firmware may assert if given more IOCBs than it can handle.

Solution:
The driver was updated to throttle the number of active IOCBs based
on the total number of IOCB buffers received from get_firmware_status
mbx_sts[2].

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

[SCSI] qla4xxx: Remove unnecessary code from qla4xxx_init_local_data

Removed unnecessary calls to qla4xxx_get_firmware_status from
function qla4xxx_init_local_data

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

[SCSI] qla4xxx: Quiesce driver activities while loopback

Quiesce all different activities performed by driver upon the link events
while loopback diagnostics is in progress.

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

[SCSI] qla4xxx: Rename MBOX_ASTS_IDC_NOTIFY to MBOX_ASTS_IDC_REQUEST_NOTIFICATION

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

[SCSI] qla4xxx: Add spurious interrupt messages under debug level 2

The spurious interrupt messages are seen when interrupt lines are
shared between different adapters or modules. Example, interrupt
lines are shared between qlcnic and qla4xxx and USB modules, then the
console is flooded with flurry of spurious interrupt messages
which are expected in such scenario.

Hence put these messages under debug level 2 for INTx interrupt mode.

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

[SCSI] cxgb4i: Remove the scsi host device when removing device

When doing a hotplug remove of a cxgb4 device, there are still dandling
symlinks at /sys/class/scsi_host/hostX to the removed PCI device. The
upper layer device may also try to send data, which may crash the
system.

The DETACH message from the lower level driver is sent to the ULD when
the device is removed, when the scsi host should be removed from the
system, avoiding any problems.

After this patch, there are no more dangling symlinks and many attempts
to crash the system while there is SCSI activity and removing the device
have failed. Adding the device back again works as expected, with the
scsi hosts showing up again.

Based on a patch by Karen Xie.

Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Cc: Karen Xie <kxie@chelsio.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] bfa: fix strncpy() limiter in bfad_start_ops()

The closing parenthesis is in the wrong place so it takes the sizeof a
pointer instead of the sizeof the buffer minus one.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Anil Gurumurthy <agurumur@brocade.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] qla4xxx: Update driver version to 5.03.00-k3

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

[SCSI] qla4xxx: Correct the validation to check in get_sys_info mailbox

Check mbox_sts[3] instead of mbox_sts[4] for ISP83xx to validate size
of data returned

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: Pass correct function param to qla4_8xxx_rd_direct

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] lpfc 8.3.37: Update lpfc version for 8.3.37 driver release

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

[SCSI] lpfc 8.3.37: Fixed infinite loop in lpfc_sli4_fcf_rr_next_index_get.

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

[SCSI] lpfc 8.3.37: Fixed crash due to SLI Port invalid resource count

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

[SCSI] lpfc 8.3.37: Provide support for FCoE protocol dual-chute (ULP) operation

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

[SCSI] lpfc 8.3.37: Fixed stale ndlp state when the node is marked for deferred removal.

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

[SCSI] lpfc 8.3.37: Fix potential memory corruption bug

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

[SCSI] lpfc 8.3.37: Fixed no-context ABTS failed with BA_RJT

Fixed no-context ABTS received on unsolicited receive queue failed with BA_RJT

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

[SCSI] lpfc 8.3.37: Removed use of NOP mailboxes for interrupt verification

Removed use of NOP mailboxes for interrupt verification in pci_probe_one_s4

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

[SCSI] lpfc 8.3.37: Fixed exhausted retry for plogi to nameserver.

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

[SCSI] lpfc 8.3.37: Fixed ELS_REC received on the unsolicited receive queue

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

[SCSI] lpfc 8.3.37: Provide support for change_queue_type

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

[SCSI] ipr: Fix sparse error in ipr driver

This patch fixes the following sparse error:

 CHECK   drivers/scsi/ipr.c
spinlock.h:147:9: warning: context imbalance in 'ipr_reset_reload' - unexpected unlock

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Driver version 2.6.0

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Implement block iopoll

This patch implements blk iopoll in ipr driver for performance improvement.

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Reduce lock contention

This patch reduces lock contention while implementing distributed
completion processing.

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Add support for MSI-X and distributed completion

The new generation IBM SAS Controllers will support MSI-X interrupts and
Distributed Completion Processing features. This patch add these support
in ipr device driver.

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Resource path error logging cleanup

The resource path as displayed by the ipr driver is the
location string identifying a location on the SAS fabric.
This patch adds the SCSI host number such that error logs
can be more easily correlated in multiple adapter configurations.

Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Handler ID memory allocation failure at module load time

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] ipr: Add sereral new CCIN definitions for new adapters support

Add the appropriate definitions and table entries for new adapter support.

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

[SCSI] csiostor: Don't leak mem or fail to release firmware in csio_hw_flash_config()

If kzalloc() or csio_hw_check_fwconfig() fail we may leave the
csio_hw_flash_config() function without freeing allocated memory or
firmware. This should take care of the leaks.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Acked-by: Naresh Kumar Inna <naresh@chelsio.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>