diff options
author | James Smart <james.smart@broadcom.com> | 2020-10-20 22:27:12 +0200 |
---|---|---|
committer | Martin K. Petersen <martin.petersen@oracle.com> | 2020-10-27 02:42:38 +0100 |
commit | e7dab164a9aa457f89d4528452bdfc3e15ac98b6 (patch) | |
tree | a1ffb6bf4493e37d5ed9fd0ea2c8d55caf736893 /drivers/scsi/lpfc/lpfc_init.c | |
parent | scsi: lpfc: Fix invalid sleeping context in lpfc_sli4_nvmet_alloc() (diff) | |
download | linux-e7dab164a9aa457f89d4528452bdfc3e15ac98b6.tar.xz linux-e7dab164a9aa457f89d4528452bdfc3e15ac98b6.zip |
scsi: lpfc: Fix scheduling call while in softirq context in lpfc_unreg_rpi
The following call trace was seen during HBA reset testing:
BUG: scheduling while atomic: swapper/2/0/0x10000100
...
Call Trace:
dump_stack+0x19/0x1b
__schedule_bug+0x64/0x72
__schedule+0x782/0x840
__cond_resched+0x26/0x30
_cond_resched+0x3a/0x50
mempool_alloc+0xa0/0x170
lpfc_unreg_rpi+0x151/0x630 [lpfc]
lpfc_sli_abts_recover_port+0x171/0x190 [lpfc]
lpfc_sli4_abts_err_handler+0xb2/0x1f0 [lpfc]
lpfc_sli4_io_xri_aborted+0x256/0x300 [lpfc]
lpfc_sli4_sp_handle_abort_xri_wcqe.isra.51+0xa3/0x190 [lpfc]
lpfc_sli4_fp_handle_cqe+0x89/0x4d0 [lpfc]
__lpfc_sli4_process_cq+0xdb/0x2e0 [lpfc]
__lpfc_sli4_hba_process_cq+0x41/0x100 [lpfc]
lpfc_cq_poll_hdler+0x1a/0x30 [lpfc]
irq_poll_softirq+0xc7/0x100
__do_softirq+0xf5/0x280
call_softirq+0x1c/0x30
do_softirq+0x65/0xa0
irq_exit+0x105/0x110
do_IRQ+0x56/0xf0
common_interrupt+0x16a/0x16a
With the conversion to blk_io_poll for better interrupt latency in normal
cases, it introduced this code path, executed when I/O aborts or logouts
are seen, which attempts to allocate memory for a mailbox command to be
issued. The allocation is GFP_KERNEL, thus it could attempt to sleep.
Fix by creating a work element that performs the event handling for the
remote port. This will have the mailbox commands and other items performed
in the work element, not the irq. A much better method as the "irq" routine
does not stall while performing all this deep handling code.
Ensure that allocation failures are handled and send LOGO on failure.
Additionally, enlarge the mailbox memory pool to reduce the possibility of
additional allocation in this path.
Link: https://lore.kernel.org/r/20201020202719.54726-3-james.smart@broadcom.com
Fixes: 317aeb83c92b ("scsi: lpfc: Add blk_io_poll support for latency improvment")
Cc: <stable@vger.kernel.org> # v5.9+
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Diffstat (limited to 'drivers/scsi/lpfc/lpfc_init.c')
-rw-r--r-- | drivers/scsi/lpfc/lpfc_init.c | 46 |
1 files changed, 27 insertions, 19 deletions
diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c index b6090357e8a5..edfefff0214a 100644 --- a/drivers/scsi/lpfc/lpfc_init.c +++ b/drivers/scsi/lpfc/lpfc_init.c @@ -5958,18 +5958,21 @@ lpfc_sli4_async_grp5_evt(struct lpfc_hba *phba, void lpfc_sli4_async_event_proc(struct lpfc_hba *phba) { struct lpfc_cq_event *cq_event; + unsigned long iflags; /* First, declare the async event has been handled */ - spin_lock_irq(&phba->hbalock); + spin_lock_irqsave(&phba->hbalock, iflags); phba->hba_flag &= ~ASYNC_EVENT; - spin_unlock_irq(&phba->hbalock); + spin_unlock_irqrestore(&phba->hbalock, iflags); + /* Now, handle all the async events */ + spin_lock_irqsave(&phba->sli4_hba.asynce_list_lock, iflags); while (!list_empty(&phba->sli4_hba.sp_asynce_work_queue)) { - /* Get the first event from the head of the event queue */ - spin_lock_irq(&phba->hbalock); list_remove_head(&phba->sli4_hba.sp_asynce_work_queue, cq_event, struct lpfc_cq_event, list); - spin_unlock_irq(&phba->hbalock); + spin_unlock_irqrestore(&phba->sli4_hba.asynce_list_lock, + iflags); + /* Process the asynchronous event */ switch (bf_get(lpfc_trailer_code, &cq_event->cqe.mcqe_cmpl)) { case LPFC_TRAILER_CODE_LINK: @@ -6001,9 +6004,12 @@ void lpfc_sli4_async_event_proc(struct lpfc_hba *phba) &cq_event->cqe.mcqe_cmpl)); break; } + /* Free the completion event processed to the free pool */ lpfc_sli4_cq_event_release(phba, cq_event); + spin_lock_irqsave(&phba->sli4_hba.asynce_list_lock, iflags); } + spin_unlock_irqrestore(&phba->sli4_hba.asynce_list_lock, iflags); } /** @@ -6630,6 +6636,8 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba) /* This abort list used by worker thread */ spin_lock_init(&phba->sli4_hba.sgl_list_lock); spin_lock_init(&phba->sli4_hba.nvmet_io_wait_lock); + spin_lock_init(&phba->sli4_hba.asynce_list_lock); + spin_lock_init(&phba->sli4_hba.els_xri_abrt_list_lock); /* * Initialize driver internal slow-path work queues @@ -6641,8 +6649,6 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba) INIT_LIST_HEAD(&phba->sli4_hba.sp_queue_event); /* Asynchronous event CQ Event work queue list */ INIT_LIST_HEAD(&phba->sli4_hba.sp_asynce_work_queue); - /* Fast-path XRI aborted CQ Event work queue list */ - INIT_LIST_HEAD(&phba->sli4_hba.sp_fcp_xri_aborted_work_queue); /* Slow-path XRI aborted CQ Event work queue list */ INIT_LIST_HEAD(&phba->sli4_hba.sp_els_xri_aborted_work_queue); /* Receive queue CQ Event work queue list */ @@ -10173,26 +10179,28 @@ lpfc_sli4_cq_event_release(struct lpfc_hba *phba, static void lpfc_sli4_cq_event_release_all(struct lpfc_hba *phba) { - LIST_HEAD(cqelist); - struct lpfc_cq_event *cqe; + LIST_HEAD(cq_event_list); + struct lpfc_cq_event *cq_event; unsigned long iflags; /* Retrieve all the pending WCQEs from pending WCQE lists */ - spin_lock_irqsave(&phba->hbalock, iflags); - /* Pending FCP XRI abort events */ - list_splice_init(&phba->sli4_hba.sp_fcp_xri_aborted_work_queue, - &cqelist); + /* Pending ELS XRI abort events */ + spin_lock_irqsave(&phba->sli4_hba.els_xri_abrt_list_lock, iflags); list_splice_init(&phba->sli4_hba.sp_els_xri_aborted_work_queue, - &cqelist); + &cq_event_list); + spin_unlock_irqrestore(&phba->sli4_hba.els_xri_abrt_list_lock, iflags); + /* Pending asynnc events */ + spin_lock_irqsave(&phba->sli4_hba.asynce_list_lock, iflags); list_splice_init(&phba->sli4_hba.sp_asynce_work_queue, - &cqelist); - spin_unlock_irqrestore(&phba->hbalock, iflags); + &cq_event_list); + spin_unlock_irqrestore(&phba->sli4_hba.asynce_list_lock, iflags); - while (!list_empty(&cqelist)) { - list_remove_head(&cqelist, cqe, struct lpfc_cq_event, list); - lpfc_sli4_cq_event_release(phba, cqe); + while (!list_empty(&cq_event_list)) { + list_remove_head(&cq_event_list, cq_event, + struct lpfc_cq_event, list); + lpfc_sli4_cq_event_release(phba, cq_event); } } |