- 01 Mar, 2016 12 commits
-
-
Finn Thain authored
The list structures and related logic used in the NCR5380 driver mean that a command cannot be queued twice (i.e. can't appear on more than one queue and can't appear on the same queue more than once). The abort handler must forget the command so that the mid-layer can re-use it. E.g. the ML may send it back to the LLD via via scsi_eh_get_sense(). Fix this and also fix two error paths, so that commands get forgotten iff completed. Fixes: 8b00c3d5 ("ncr5380: Implement new eh_abort_handler") Tested-by:
Michael Schmitz <schmitzmic@gmail.com> Cc: <stable@vger.kernel.org> # 4.5 Signed-off-by:
Finn Thain <fthain@telegraphics.com.au> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com>
-
Finn Thain authored
Calling NCR5380_select() from the abort handler causes various problems. Firstly, it means potentially re-entering NCR5380_select(). Secondly, it means that the lock is released, which permits the EH handlers to be re-entered. The combination results in crashes. Don't do it. Fixes: 8b00c3d5 ("ncr5380: Implement new eh_abort_handler") Reported-and-tested-by:
-
Finn Thain authored
The calls to NCR5380_transfer_pio() for DATA IN and DATA OUT phases will modify cmd->SCp.this_residual, cmd->SCp.ptr and cmd->SCp.buffer. That works as long as EH does not intervene, which became possible in atari_NCR5380.c when I changed the locking to bring it closer to NCR5380.c. If error recovery aborts the command, the scsi_cmnd in question and its buffer will be returned to the mid-layer. So the transfer has to cease, but it can't be stopped by the initiator because the target controls the bus phase. The problem does not arise if the lock is not released. That was fine for atari_scsi, because it implements DMA. For the other drivers, we have to release the lock and re-enable interrupts for long PIO data transfers. The solution is to split the transfer into small chunks. In between chunks the main loop releases the lock and re-enables interrupts. Thus interrupts can be serviced and eh_bus_reset_handler can intervene if need be. This fixes an oops in NCR5380_transfer_pio() that can happen when the EH abort handler is invoked during DATA IN or DATA OUT phase. Fixes: 11d2f63b ("ncr5380: Change instance->host_lock to hostdata->lock") Reported-and-tested-by:
-
Finn Thain authored
Commands subject to exception handling are to be returned to the scsi mid-layer. Make sure that the various command pointers and command lists in the low-level driver are correctly cleansed of affected commands. This fixes some bugs that I accidentally introduced in v4.5-rc1 including the removal of INIT_LIST_HEAD for the 'autosense' and 'disconnected' command lists, and the possible NULL pointer dereference in NCR5380_bus_reset() that was reported by Dan Carpenter. hostdata->sensing may also point to an affected command so this pointer also has to be cleared. The abort handler calls complete_cmd() to take care of this; let's have the bus reset handler do the same. The issue queue may also contain an affected command. If so, remove it. This also follows the abort handler logic. Reported-by:
Dan Carpenter <dan.carpenter@oracle.com> Fixes: 62717f53 ("ncr5380: Implement new eh_bus_reset_handler") Tested-by:
-
Nicholas Krause authored
This adds the missing error check and path for if the call to the function hwi_init_controller fails as this error path was clearly missed when writing beiscsi_eeh_resume and thus we must add it now in order to be able to handle this nonrecoverable failing function call gracefully in beiscsi_eeh_resume. Signed-off-by:
Nicholas Krause <xerofoify@gmail.com> Reviewed-by:
Jitendra Bhivare <jitendra.bhivare@broadcom.com> Signed-off-by:
-
Usha Ketineni authored
Do not reset fip selection time for every advertisement in fcoe_ctlr_recv_adv() but set it only once for the first validated FCF. Otherwise FCF selection won't happen when the advertisements consistently arrive with sub FCOE_CTLR_START_DELAY periodicity. Tested-by:
Narendra K <narendra_k@dell.com> Acked-by:
Neil Horman <nhorman@tuxdriver.com> Reviewed-by:
Johannes Thumshirn <jthumshirn@suse.de> Acked-by:
Vasu Dev <vasu.dev@intel.com> Signed-off-by:
Usha Ketineni <usha.k.ketineni@intel.com> Signed-off-by:
-
John Garry authored
Signed-off-by:
John Garry <john.garry@huawei.com> Reviewed-by:
Hannes Reinecke <hare@suse.de> Signed-off-by:
-
John Garry authored
In high-datarate aging tests, it is found that the SCSI framework can periodically issue lu resets as some commands timeout. Response TASK SET FULL and SAS_QUEUE_FULL may be returned many times for the same command, causing the timeouts. The SAS_QUEUE_FULL errors come from TRANS_TX_CREDIT_TIMEOUT_ERR, TRANS_TX_CLOSE_NORMAL_ERR, and TRANS_TX_ERR_FRAME_TXED errors. They do not mean that the queue is full in the host, but rather it is equivalent to meaning the queue is full for the sdev. To overcome this, the queue depth for the sdev is reduced to 64 (from 256, set in sas_slave_configure()). Normally error code SAS_QUEUE_FULL will result in the sdev queue depth falling, but it falls too slowly during high-datarate tests and commands timeout before it has fallen to an adequete level from original value. Signed-off-by:
-
John Garry authored
When TRANS_TX_ERR_FRAME_TXED error occurs in a slot, the command should be re-attempted. This error is equivalent to meaning that the queue is full in the sdev (and not the host). A superflous debug statement is also removed in the slot complete handler. Signed-off-by:
-
John Garry authored
When TRANS_TX_CREDIT_TIMEOUT_ERR or TRANS_TX_CLOSE_NORMAL_ERR error occur in a slot, the command should be re-attempted. This error is equivalent to meaning that the queue is full in the sdev (and not the host). Signed-off-by:
Hannes Reinecke <hare@suse.com> Signed-off-by:
-
John Garry authored
Add a function to abort a slot (task) in the target device and then cleanup and complete the task. The function is called from work queue context as it cannot be called from the context where it is triggered (interrupt). Flag hisi_sas_slot.abort is added as the flag used in the slot error handler to indicate whether the slot needs to be aborted in the sdev prior to cleanup and finish. Signed-off-by:
-
John Garry authored
In hisi_sas_exec_internal_tmf_task(), the check for SAM_STAT_GOOD is replaced with TMF_RESP_FUNC_COMPLETE, which is a genuine tmf response code. SAM_STAT_GOOD and TMF_RESP_FUNC_COMPLETE have the same value, so this is why it worked before. Signed-off-by:
-
- 26 Feb, 2016 3 commits
-
-
Alison Schofield authored
struct timeval will overflow on 32-bit systems in y2038 and is being removed from the kernel. Replace the use of struct timeval and do_gettimeofday() with ktime_get_real_seconds() which provides a 64-bit seconds value and is y2038 safe. gdth driver requires changes in two areas: 1) gdth_store_event() loads two u32 timestamp fields for ioctl GDTIOCTL_EVENT These timestamp fields are part of struct gdth_evt_str used for passing event data to userspace. At the first instance of an event we do (first_stamp=last_stamp="current time"). If that same event repeats, we do (last_stamp="current time") AND increment same_count to indicate how many times the event has repeated since first_stamp. This patch replaces the use of timeval and do_gettimeofday() with ktime_get_real_seconds() cast to u32 to extend the timestamp fields to y2106. Beyond y2106, the userspace tools (ie. RAID controller monitors) can work around the time rollover and this driver would still not need to change. Alternative: The alternative approach is to introduce a new ioctl in gdth with the u32 time fields defined as u64. This would require userspace changes now, but not in y2106. 2) gdth_show_info() calculates elapsed time using u32 first_stamp It is adding events with timestamps to a seq_file. Timestamps are calculated as the "current time" minus the first_stamp. This patch replaces the use of timeval and do_gettimeofday() with ktime_get_real_seconds() cast to u32 to calculate the timestamp. This elapsed time calculation is safe even when the time wraps (beyond y2106) due to how unsigned subtraction works. A comment has been added to the code to indicate this safety. Alternative: This piece itself doesn't warrant an alternative, but if we do introduce a new structure & ioctl with u64 timestamps, this would change accordingly. Signed-off-by:Alison Schofield <amsfield22@gmail.com> Reviewed-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Sudip Mukherjee authored
The variable is_ver1 is always true and so OSD_CAP_LEN can never be used. Reported by Coverity. Signed-off-by:
Sudip Mukherjee <sudip.mukherjee@codethink.co.uk> Reviewed-by:
Matthew R. Ochs <mrochs@linux.vnet.ibm.com> Acked-by:
Boaz harrosh <ooo@elecrozaur.com> Signed-off-by:
-
Sudip Mukherjee authored
parport_claim() can fail and we should be checking if we were able to claim the port. Signed-off-by:
-
- 24 Feb, 2016 25 commits
-
-
Charles authored
Add S3/S4 support, add .suspend and .resume function in pci_driver. In .suspend handler, driver send S3/S4 signal to the device. Signed-off-by:
Charles Chiou <charles.chiou@tw.promise.com> Reviewed-by:
-
Charles authored
1. Add hotplug support. Pegasus support surprise removal. To this end, I use return_abnormal_state function to return DID_NO_CONNECT for all commands which sent to driver. 2. Remove stex_hba_stop in stex_remove because we cannot send command to device after hotplug. 3. Add new device status: MU_STATE_STOP, MU_STATE_NOCONNECT, MU_STATE_STOP. MU_STATE_STOP is currently not referenced. MU_STATE_NOCONNECT represent that device is plugged out from the host. 4. Use return_abnormal_function() to substitute part of code in stex_do_reset. Signed-off-by:
-
Charles authored
Pegasus is a high performace hardware RAID solution designed to unleash the raw power of Thunderbolt technology. 1. Add code to distinct SuperTrack and Pegasus series by sub device ID. It should support backward compatibility. 2. Change the driver version. Signed-off-by:
-
Don Brace authored
Reviewed-by:
Kevin Barnett <kevin.barnett@microsemi.com> Reviewed-by:
Justin Lindley <justin.lindley@microsemi.com> Reviewed-by:
Don Brace <don.brace@microsemi.com> Signed-off-by:
-
Don Brace authored
Reviewed-by:
-
Don Brace authored
This patch depends on patch - commit ac10a3e4ed64 ("Export function scsi_scan.c:sanitize_inquiry_string") Suggested-by:Mahesh Rajashekhara <mahesh.rajashekhara@microsemi.com> Reviewed-by:
Scott Teel <scott.teel@microsemi.com> Reviewed-by:
-
Don Brace authored
An oops can occur when submitting ioaccel2 commands when the phys_disk pointer is NULL in hpsa_scsi_ioaccel_raid_map. Happens when there are configuration changes during I/O operations. If the phys_disk pointer is NULL, send the command down the RAID path. Reviewed-by:
-
Don Brace authored
Aborts were not being sent down to HBA devices Reviewed-by:
-
Don Brace authored
The bitmap was changed after this definition was added to the driver. Correcting the bitmap definition. Reviewed-by:
-
Don Brace authored
Reviewed-by:
-
Don Brace authored
Stop annoying "Error, could not get enclosure information" messages. Reviewed-by:
-
Hannes Reinecke authored
[mkp: Fixed merge due to patches 20-22 of series being postponed] Reviewed-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Hannes Reinecke authored
If a device needs to be rescanned the device_handler might need to be rechecked, too. So add a 'rescan' callback to the device handler and call it upon scsi_rescan_device(). The rescan callback will be invoked from the Unit Attention handling of ASC/ASCQ 3F 03 (INQUIRY DATA HAS CHANGED). Reviewed-by:
Bart Van Assche <bart.vanassche@sandisk.com> Reviewed-by:
-
Hannes Reinecke authored
Sending a 'REPORT TARGET PORT GROUP' command is a costly operation, as the array has to gather information about all ports. So instead of using RTPG to poll for a status update when a port is in transitioning we should be sending a TEST UNIT READY, and wait for the sense code to report success. Reviewed-by:
Ewan Milne <emilne@redhat.com> Reviewed-by:
-
Hannes Reinecke authored
When we read in the target port group state we should be updating all affected port groups, otherwise we risk running out of sync. Reviewed-by:
-
Hannes Reinecke authored
When we receive a unit attention code of 'ALUA state changed' we should recheck the state, as it might be due to an implicit ALUA state transition. This allows us to return NEEDS_RETRY instead of ADD_TO_MLQUEUE, allowing to terminate the retries after a certain time. At the same time a workqueue item might already be queued, which should be started immediately to avoid any delays. Reviewed-by:
-
Hannes Reinecke authored
Add a new blacklist flag BLIST_SYNC_ALUA to instruct the alua device handler to use synchronous command submission for ALUA commands. Reviewed-by:
-
Hannes Reinecke authored
Some arrays may only capable of handling one STPG at a time, so this patch adds a singlethreaded workqueue for STPGs to be submitted synchronously. Reviewed-by:
-
Hannes Reinecke authored
The current ALUA device_handler has two drawbacks: - We're sending a 'SET TARGET PORT GROUP' command to every LUN, disregarding the fact that several LUNs might be in a port group and will be automatically switched whenever _any_ LUN within that port group receives the command. - Whenever a LUN is in 'transitioning' mode we cannot block I/O to that LUN, instead the controller has to abort the command. This leads to increased traffic across the wire and heavy load on the controller during switchover. With this patch the RTPG handling is moved to a per-portgroup workqueue. This reduces the number of 'REPORT TARGET PORT GROUP' and 'SET TARGET PORT GROUPS' sent to the controller as we're sending them now per port group, and not per device as previously. It also allows us to block I/O to any LUN / port group found to be in 'transitioning' ALUA mode, as the workqueue item will be requeued until the controller moves out of transitioning. Reviewed-by:
-
Hannes Reinecke authored
The 'relative port' field is not used, and might get stale when the port group changes. So remove the field altogether. Reviewed-by:
-
Hannes Reinecke authored
When the optimize_stpg module option is set we should just set it once during port_group allocation. Doing so allows us to override it later with device specific settings. Reviewed-by:
-
Hannes Reinecke authored
revert commit a8e5a2d5 ("[SCSI] scsi_dh_alua: ALUA handler attach should succeed while TPG is transitioning") This reverts commit a8e5a2d5 Obsoleted by the next patch. Reviewed-by:
-
Hannes Reinecke authored
Rework alua_check_vpd() to use scsi_vpd_get_tpg() and move the port group selection into the function, too. With that we can simplify alua_initialize() to just call alua_check_tpgs() and alua_check_vpd(); Reviewed-by:
-
Hannes Reinecke authored
Use scsi_vpd_lun_id() to assign a unique device identification to the alua port group structure. Reviewed-by:
-
Hannes Reinecke authored
The port group needs to be a separate structure as several LUNs might belong to the same group. Reviewed-by:
-
