- 20 Jun, 2018 29 commits
-
-
Bart Van Assche authored
scsi: mpt3sas: Split _base_reset_handler(), mpt3sas_scsih_reset_handler() and mpt3sas_ctl_reset_handler() Split each of these functions in three functions - one function per reset phase. This patch does not change any functionality but makes the code easier to read. Note: it is much easier to review the git diff -w output after having applied this patch than by reviewing the patch itself. Signed-off-by:
Bart Van Assche <bart.vanassche@wdc.com> Cc: Sathya Prakash <sathya.prakash@broadcom.com> Cc: Chaitra P B <chaitra.basappa@broadcom.com> Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com>
-
Bart Van Assche authored
Since ioc->shost_recovery is set after ioc->reset_in_progress_mutex is obtained, if concurrent resets are issued there is a short time during which ioc->reset_in_progress_mutex is locked and ioc->shost_recovery == 0. Avoid that this can cause trouble by unconditionally locking ioc->shost_recovery. Signed-off-by:
-
Bart Van Assche authored
This patch avoids that smatch complains about a double unlock on ioc->transport_cmds.mutex. Fixes: 651a0136 ("scsi: scsi_transport_sas: switch to bsg-lib for SMP passthrough") Signed-off-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Bart Van Assche authored
Make _base_build_nvme_prp() easier to read by introducing a structure to access NVMe command fields. Signed-off-by:
-
Bart Van Assche authored
This patch avoids that gcc complains about switch/case fall-through when building with W=1. Signed-off-by:
-
Bart Van Assche authored
This patch does not change any functionality. Signed-off-by:
-
Bart Van Assche authored
Modify the indentation such that smatch no longer complains about inconsistent indenting. Signed-off-by:
-
Jason Yan authored
Commit 2623c7a5 ("libata: add refcounting to ata_host") v4.17+ introduced refcounting to ata_host and will increase or decrease the refcount when adding or deleting transport ATA port. Now the ata host for libsas is embedded in domain_device, and the ->kref member is not initialized. Afer we add ata transport class, ata_host_get() will be called when adding transport ATA port and a warning will be triggered as below: refcount_t: increment on 0; use-after-free. WARNING: CPU: 2 PID: 103 at lib/refcount.c:153 refcount_inc+0x40/0x48 ...... Call trace: refcount_inc+0x40/0x48 ata_host_get+0x10/0x18 ata_tport_add+0x40/0x120 ata_sas_tport_add+0xc/0x14 sas_ata_init+0x7c/0xc8 sas_discover_domain+0x380/0x53c process_one_work+0x12c/0x288 worker_thread+0x58/0x3f0 kthread+0xfc/0x128 ret_from_fork+0x10/0x18 And also when removing transport ATA port ata_host_put() will be called and another similar warning will be triggered. If the refcount decreased to zero, the ata host will be freed. But this ata host is only part of domain_device, it cannot be freed directly. So we have to change this embedded static ata host to a dynamically allocated ata host and initialize the ->kref member. To use ata_host_get() and ata_host_put() in libsas, we need to move the declaration of these functions to the public libata.h and export them. Fixes: b6240a4d ("scsi: libsas: add transport class for ATA devices") Signed-off-by:
Jason Yan <yanaijie@huawei.com> CC: John Garry <john.garry@huawei.com> CC: Taras Kondratiuk <takondra@cisco.com> CC: Tejun Heo <tj@kernel.org> Acked-by:
Tejun Heo <tj@kernel.org> Signed-off-by:
-
Matthew Wilcox authored
With its one user gone, remove the library code. Signed-off-by:
Matthew Wilcox <willy@infradead.org> Reviewed-by:
Jens Axboe <axboe@kernel.dk> Signed-off-by:
-
Matthew Wilcox authored
The sbitmap and the percpu_ida perform essentially the same task, allocating tags for commands. The sbitmap outperforms the percpu_ida as documented here: https://lkml.org/lkml/2014/4/22/553 The sbitmap interface is a little harder to use, but being able to remove the percpu_ida code and getting better performance justifies the additional complexity. Signed-off-by:
-
Matthew Wilcox authored
Introduce target_free_tag() and convert all drivers to use it. Signed-off-by:
-
Tomas Winkler authored
ufshcd_dump_regs should use memcpy_fromio to read host registers instead of directly accessing using memcpy. The same function is utilized in ufs-qcom. Elminite compilation warning drivers/scsi/ufs/ufshcd.c:356:9: warning: incorrect type in argument 6 (different address spaces) drivers/scsi/ufs/ufshcd.c:356:9: expected void const *buf drivers/scsi/ufs/ufshcd.c:356:9: got void [noderef] <asn:2>*mmio_base Signed-off-by:
Tomas Winkler <tomas.winkler@intel.com> Reviewed-by:
Avri Altman <avri.altman@wdc.com> Signed-off-by:
-
Jeff Moyer authored
commit a45b599a ("scsi: sg: allocate with __GFP_ZERO in sg_build_indirect()") changed the call to alloc_pages to always use __GFP_ZERO. Just above that, though, there was this: if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SYS_RAWIO)) gfp_mask |= __GFP_ZERO; And there's only one user of the gfp_mask. Just or in the __GFP_ZERO flag at the top of the function and be done with it. Signed-off-by:
Jeff Moyer <jmoyer@redhat.com> Acked-by:
Douglas Gilbert <dgilbert@interlog.com> Signed-off-by:
-
Sebastian Andrzej Siewior authored
In commit d2ba5675 ("[SCSI] qla2xxx: Disable local-interrupts while polling for RISC status.") added a local_irq_disable() before invoking the ->intr_handler callback. The function, which was used in this callback, did not disable interrupts while acquiring the spin_lock so a deadlock was possible and this change was one possible solution. The function in question was qla2300_intr_handler() and is using spin_lock_irqsave() since commit 43fac4d9 ("[SCSI] qla2xxx: Resolve a performance issue in interrupt"). I checked all other ->intr_handler callbacks and all of them use the irqsave variant so it is safe to remove the local_irq_save() block now. Cc: qla2xxx-upstream@qlogic.com Signed-off-by:
Sebastian Andrzej Siewior <bigeasy@linutronix.de> Acked-by:
Himanshu Madhani <himanshu.madhani@cavium.com> Signed-off-by:
-
Johannes Thumshirn authored
This will make subsequent refactoring easier to handle. Note: this patch is nowhere checkpatch clean. Signed-off-by:
Johannes Thumshirn <jthumshirn@suse.de> Reviewed-by:
-
Zhouyang Jia authored
When ioremap_nocache fails, the lack of error-handling code may cause unexpected results. This patch adds error-handling code after calling ioremap_nocache. Signed-off-by:
Zhouyang Jia <jiazhouyang09@gmail.com> Reviewed-by:
Manish Rangankar <Manish.Rangankar@cavium.com> Signed-off-by:
-
Sebastian Andrzej Siewior authored
Since commit 312d3e56 ("[SCSI] libsas: remove ata_port.lock management duties from lldds") the sas_ata_qc_issue() function unlocks the ata_port.lock and disables interrupts before doing so. That lock is always taken with disabled interrupts so at this point, the interrupts are already disabled. There is no need to disable the interrupts before the unlock operation because they are already disabled. Restoring the interrupt state later does not change anything because they were disabled and remain disabled. Therefore remove the operations which do not change the behaviour. Fixes: 312d3e56 ("[SCSI] libsas: remove ata_port.lock management duties from lldds") Signed-off-by:
Dan Williams <dan.j.williams@intel.com> Reviewed-by:
John Garry <john.garry@huawei.com> Signed-off-by:
-
Wen Xiong authored
This patch adds formatting error overlay 0x41 to improve debug capabilities. Signed-off-by:
Wen Xiong <wenxiong@linux.vnet.ibm.com> Acked-by:
Brian King <brking@linux.vnet.ibm.com> Signed-off-by:
-
John Garry authored
Update CFG_1US_TIMER_TRSH and CON_CFG_DRIVER settings. Signed-off-by:
-
John Garry authored
The init is missed for hisi_sas_phy spinlock, so add it. Signed-off-by:
-
Xiang Chen authored
Currently the driver spends much time allocating and freeing the slot DMA buffer for command delivery/completion. To boost the performance, pre-allocate the buffers for all IPTT. The downside of this approach is that we are reallocating all buffer memory upfront, so hog memory which we may not need. However, the current method - DMA buffer pool - also caches all buffers and does not free them until the pool is destroyed, so is not exactly efficient either. On top of this, since the slot DMA buffer is slightly bigger than a 4K page, we need to allocate 2x4K pages per buffer (for 4K page kernel), which is quite wasteful. For 64K page size this is not such an issue. So, for the 4K page case, in order to make memory usage more efficient, pre-allocating larger blocks of DMA memory for the buffers can be more efficient. To make DMA memory usage most efficient, we would choose a single contiguous DMA memory block, but this could use up all the DMA memory in the system (when CMA enabled and no IOMMU), or we may just not be able to allocate a DMA buffer large enough when no CMA or IOMMU. To decide the block size we use the LCM (least common multiple) of the buffer size and the page size. We roundup(64) to ensure the LCM is not too large, even though a little memory may be wasted per block. So, with this, the total memory requirement is about is about 17MB for 4096 max IPTT. Previously (for 4K pages case), it would be 32MB (for all slots allocated). With this change, the relative increase of IOPS for bs=4K read when PAGE_SIZE=4K and PAGE_SIZE=64K is as follows: IODEPTH 4K PAGE_SIZE 64K PAGE_SIZE 32 56% 47% 64 53% 44% 128 64% 43% 256 67% 45% Signed-off-by:Xiang Chen <chenxiang66@hisilicon.com> Signed-off-by:
-
Xiaofei Tan authored
In host reset, we use TMF or soft-reset to re-init device, and if success, we will release all LLDD resources of this device. If the init fails - maybe because the device was removed or link has not come up - then do not release the LLDD resources, but rather rely on SCSI EH to handle the timeout for these resources later on. But if clear nexus ha calls host reset, which is the last effort of SCSI EH, we should release all LLDD remain resources. Because SCSI EH will release all tasks after clear nexus ha. Before release, we do I_T nexus reset to try to clear target remain IOs. Signed-off-by:
Xiaofei Tan <tanxiaofei@huawei.com> Signed-off-by:
-
Xiaofei Tan authored
During reset, we don't want PHY events reported to libsas for PHYs which were previously attached prior to reset. So check hisi_hba->flags for HISI_SAS_RESET_BIT to filter PHY events during reset. Signed-off-by:
-
Xiaofei Tan authored
After soft_reset() for host reset, we should not be allowed to send commands to the HW before the PHYs have come up and the port ids have been refreshed. Prior to this point, any commands cannot be successfully completed. This exclusion is achieved by grabbing the host reset semaphore. Signed-off-by:
-
Xiaofei Tan authored
There is a possible conflict when a device is removed and host reset occurs concurrently. The reason is that then the device is notified as gone, we try to clear the ITCT, which is notified via an interrupt. The dev gone function pends on this event with a completion, which is completed when the ITCT interrupt occurs. But host reset will disable all interrupts, the wait_for_completion() may wait indefinitely. This patch adds an semaphore to synchronise this two processes. The semaphore is taken by the host reset as the basis of synchronising. Signed-off-by:
-
Xiaofei Tan authored
There are many BROADCAST primitives generated by the host. We are only interested in BROADCAST (CHANGE) primitives currently, so only process this. We have applied this processing for v2 hw before, and it is also needed for v3 hw. Signed-off-by:
-
Xiang Chen authored
This patch replaces the usage of dma_alloc_coherent() with the managed version, dmam_alloc_coherent(), hereby reducing replicated code. Signed-off-by:
-
Ketan Mukadam authored
Update Broadcom copyright markings in all modified files. Signed-off-by:
Ketan Mukadam <ketan.mukadam@broadcom.com> Reviewed-by:
Jitendra Bhivare <jitendra.bhivare@broadcom.com> Signed-off-by:
-
Jitendra Bhivare authored
FW needs NULL terminated driver version string. Signed-off-by:
-
- 19 Jun, 2018 11 commits
-
-
Shivasharan S authored
Signed-off-by:
Shivasharan S <shivasharan.srikanteshwara@broadcom.com> Signed-off-by:
-
Shivasharan S authored
Firmware provides drive specific target reset and task abort timeout values. Driver needs to use these timeout values during task management calls. If FW does not provide these values, fall back to using earlier default timeout of 50 seconds for TM. [mkp: clarified comment] Signed-off-by:
Sumit Saxena <sumit.saxena@broadcom.com> Signed-off-by:
-
Shivasharan S authored
While driver is waiting for adapter to become operational, if a kill adapter is issued, driver can bail out from the wait loop immediately rather than waiting for the entire 180 seconds. Signed-off-by:
-
Shivasharan S authored
There is a possibility that firmware on the controller was upgraded before system was suspended. During resume, driver needs to read updated controller properties. Signed-off-by:
-
Shivasharan S authored
If MR_DCMD_CTRL_GET_INFO DCMD timed out, return failure rather than doing kill adapter. Signed-off-by:
-
Chaitra P B authored
Updated driver version to "26.100.00.00" Signed-off-by:
Chaitra P B <chaitra.basappa@broadcom.com> Signed-off-by:
-
Chaitra P B authored
scsi: mpt3sas: As per MPI-spec, use combined reply queue for SAS3.5 controllers when HBA supports more than 16 MSI-x vectors. Presently driver is using combined reply queue feature when MSI-x vectors > 8 for both SAS3 and SAS3.5 controllers. But as per MPI-spec, 1. For SAS3 controllers, driver should use combined reply queue when HBA supports more than 8 MSI-x vectors. 2. For SAS3.5 controllers, driver should use combined reply queue when HBA supports more than 16 MSI-x vectors. Modified driver code to use combined reply queue for SAS3 controllers when HBA supports > 8 MSI-x vectors and for SAS3.5 controllers when HBA supports > 16 MSI-x vectors. Signed-off-by:
-
Chaitra P B authored
scsi: mpt3sas: Fix, False timeout prints for ioctl and other internal commands during controller reset. When an ioctl is sent to FW, and if there is a controller reset issued before ioctl gets completed, then in controller reset path all the pending ioctl commands are terminated from "mpt3sas_ctl_reset_handler" function. This will wake up the waiting ioctl commands in ioctl path and print timeouts which are actually not timeouts. Introduced "mpt3sas_base_check_cmd_timeout" function to check and print whether command got timed out (or) terminated due to Host reset. Signed-off-by:
-
Chaitra P B authored
While configuring of NVMe device handling, _pcie_device structure member was accessed after its reference count is decremented/put. Hence modified code to access member of _pcie_device structure before its reference count is decremented/put. Signed-off-by:
-
Chaitra P B authored
In _scsih_scan_finished driver should mark port_enable_cmd.status as 'command not used', instead base_cmds.status was marked as 'command not used'. Update code to mark port_enable_cmd.status as 'command not used'. Signed-off-by:
-
Chaitra P B authored
scsi: mpt3sas: Don't abort I/Os issued to NVMe drives while processing Async Broadcast primitive event. Linux driver when receives Broadcast Asynchronous Event Notification (BAEN) from the controller firmware, checks all pending I/Os at the driver level and issues query task, abort task TMs. This is done in the driver to handle drives which are connected with multiple initiators and undergoing target resets. In the BAEN handling code, the I/Os issued to NVMe drives are also handled and query task and abort task TMs are issued, which are not necessary as there is no multi-initiator and no BAEN concept with NVMe drives. Hence when the driver checks for pending I/Os it skips NVMe drives at this moment. Signed-off-by:
-
