a jump to 'free_memory' is apparently missing
Signed-off-by: NTomas Henzl <thenzl@redhat.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

This reverts commit 11206081.

Based on comments from Mike Christie and Sony John-N.

 During heavy IO in multipath environment with many active sessions
 and port-bouncing happening, there is a race condition because of
 which beiscsi_prcess_cqe() gets called for a connection whose
 endpoint is freed.

 Checking endpoint reference for a connection before processing in
 beiscsi_process_cq().
Signed-off-by: NMinh Tran <minhduc.tran@emulex.com>
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

 EQ teardown should happen only after all CQ are destroyed.
 In some FW config, adapter goes into a freeze state. This
 fix moves teardown of MCC-Q before the EQ teardown happens.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

 From: Dan Carpenter [mailto:dan.carpenter@oracle.com]
 Sent: Friday, March 28, 2014 1:42 AM
 Subject: re: [SCSI] be2iscsi: Fix handling timed out MBX completion from FW

 Hello Jayamohan Kallickal,

 The patch 1957aa7f: "[SCSI] be2iscsi: Fix handling timed out MBX completion from FW" from Jan 29, 2014, leads to the following static checker warning:

 drivers/scsi/be2iscsi/be_main.c:5581 beiscsi_dev_probe()
 error: memset() '&phba->ctrl.ptag_state[i]->tag_mem_state' too small (24 vs 32)

 drivers/scsi/be2iscsi/be_main.c
 5576          for (i = 0; i < MAX_MCC_CMD; i++) {
 5577                  init_waitqueue_head(&phba->ctrl.mcc_wait[i + 1]);
 5578                  phba->ctrl.mcc_tag[i] = i + 1;
 5579                  phba->ctrl.mcc_numtag[i + 1] = 0;
 5580                  phba->ctrl.mcc_tag_available++;
 5581                  memset(&phba->ctrl.ptag_state[i].tag_mem_state, 0,
 5582                         sizeof(struct beiscsi_mcc_tag_state));
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 Probably this this be change to sizeof(struct be_dma_mem struct)?  It looks like we are corrupting memory a bit here.

 5583          }

 regards,
 dan carpenter
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Signed-off-by: NMinh Tran <minhduc.tran@emulex.com>
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

 Before probe for function was completed, iSCSI Daemon had initiated login
 to target while OS was coming up. The targets which had node.startup=automatic,
 login process was initiated.Since function specific initialization was still in
 progress this lead to kernel panic.

 Fixed the issue by moving iscsi_host_add() call after adapter initialization
 is done.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

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

Setting DMA bit mask 64 and roll back to 32 if not supported.
Signed-off-by: NMinh Tran <minhduc.tran@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

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: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

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: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

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: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

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: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

This was a debugging measure to toggle enabled/disabled
when testing. But for real production setups, it's not
safe to toggle this setting without either reloading
drivers of quiescing IO first. Neither of which the toggle
enforces.

Additionally, it makes drivers deal with the conditional
state.

Remove it completely. It's up to the driver whether iopoll
is enabled or not.
Signed-off-by: NJens Axboe <axboe@fb.com>

https://bugzilla.kernel.org/show_bug.cgi?id=67091

Cc: Jayamohan Kallickal <Jayamohan.Kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

If certain configuration it is possible that ICD range is not page-aligned.
SGL posting failed in these configuration and driver load was not success.
This fix aligns ICD range values and SGL posting for IO is done.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Signed-off-by: NMinh Tran <minhduc.tran@emulex.com>
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

When iSCSI stack invokes Abort or Reset handlers, the aborted tasks
Invalid Bit in WRB needs to be set. Else FW will not be aware of
the command invalidated which leads to BAD_WRB error posted by FW.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

The MSIx to be created for SKH-R adapter should be based on
eq_count returned by get_fw_config.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Display CID available on each iSCSI Fn which can be used to
offload a connection. The display is split across available CID
on each chute.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

The connection is offload to each chute in a round-robin manner
if both the chute is loaded with iSCSI protocol
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Configuration parameters returns the number of connection that
can be offloaded one each chute.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Configuration parameters return number of CID each chute supports. The WRB_Q
is created for the passed CID count. If both the Chute has iSCSI Protocol then
WRB_Q creation is in a round robin mechanism.

For BE-X family iSCSI protocol is loaded only on single chute.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Initialization of SGL and related PAGE posting is to be done for the chute.
Based on configuration value of each Chute,SGL initialization and page posting
is done.

For BE-X family iSCSI protocol is loaded only on single chute
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Template HDR is created for each chute which has iSCSI Protocol loaded.
For BE-X family iSCSI protocol is loaded only on single chute.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

DEF_Q[HDR/DATA] is created on the chute on which iSCSI Protocol is loaded.
When a connection is offloaded, the DEF_Q HDR/Data ID needs to be passed.
FW posts ASYNC message received from target on the passed DEF_Q. Connection
can be offloaded on any of the chute so DEF_Q is created on each Chute.

Change in the ASYNC path initialization based on the configuration parameters
returned for each chute.

For BE-X family iSCSI protocol is loaded only on single chute.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

On the adapter iSCSI protocol can be loaded on either one or both
the CHUTE.Check on which CHUTE iSCSI Protocol is loaded and get
configuration parameters based on which driver initization is done.

For BE-X family iSCSI protocol is loaded only on single chute.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

The timeout set in MBX_CMD is 100sec and the ready bit checking in BMBX
mode is done for 4sec. After 4sec the task is scheduled out for 5 secs
to avoid kernel soft lockup stack trace. The loop of 4sec ready bit check
and then schedule out is done until the following conditon occur
- The Ready Bit is Set
- The timeout set in MBX_CMD expires
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

The default pdu is a common resource and needs to be protected
while manipulating it.
Signed-off-by: NMinh Tran <minhduc.tran@emulex.com>
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Storing MAC ADDR of each function in it's priv structure to
avoid issuing MAC_ADDR get IOCTL. Based on a flag set/unset
it's decided if MAC_ADDR is stored in priv structure or IOCTL
needs to be issued.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Allocating memory in the Host which will be used by the
TOE functionality during Session Offload. This fix will
allow performance improvement as adapter memory contention
will be reduced.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

For the workqueue creation interfaces that do not expect format strings,
make sure they cannot accidently be parsed that way.  Additionally, clean
up calls made with a single parameter that would be handled as a format
string.  Many callers are passing potentially dynamic string content, so
use "%s" in those cases to avoid any potential accidents.
Signed-off-by: NKees Cook <keescook@chromium.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Command Window value from the CQE was used to calculate the
max_cmdsn for that session.The command window value extracted
for SKH-R adapter was not proper. The value was extracted from
BE adapter completion event. Fixed the issue by getting the
cmd_wnd value from SKH-R CQE.

The exp_cmdsn and max_cmdsn values were not converted to BE format
before calling the __iscsi_complete_pdu(). Fixed the issue of converting
to BE format.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

The driver creates "NAPI" context per core which is fine,
however the above routine declares the ret variable as static!
Thus there is only one instance of this variable!
When this routine is called from more than one thread of execution,
than the result is unpredictable.

         static unsigned int ret;
         .....

         ret = beiscsi_process_cq(pbe_eq);
                 <--------Another thread can enter here and change "ret".
         if (ret < budget) {
                ....
         }
                 <--------Another thread can enter here and change "ret".
         return ret;

Fix - remove the "static"
Signed-off-by: NShlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

Number of CID assigned to a function from adapter can be dynamic. The CID count
for each function was fixed number before. Code Fix done so that adapters with
fixed/dynamic CID count will work with the driver.
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

When target send a NOP-IN with valid TTT, driver issues a NOP-OUT
and the task was not freed from driver. The task list available for
the session used to run out, and as no more task list were available
no more iSCSI commands were exchanged on that session.
This patches fixed the issue, by calling iscsi_put_task.
Signed-off-by: NMinh Tran <minhduc.tran@emulex.com>
Signed-off-by: NJohn Soni Jose <sony.john-n@emulex.com>
Signed-off-by: NJayamohan Kallickal <jayamohan.kallickal@emulex.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>