The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@xxxxxxxxxxxxxxx>.
Possible dependencies:
68ad83188d78 ("scsi: qla2xxx: Fix crash when I/O abort times out")
31e6cdbe0eae ("scsi: qla2xxx: Implement ref count for SRB")
d4523bd6fd5d ("scsi: qla2xxx: Refactor asynchronous command initialization")
2cabf10dbbe3 ("scsi: qla2xxx: Fix hang on NVMe command timeouts")
e3d2612f583b ("scsi: qla2xxx: Fix use after free in debug code")
9efea843a906 ("scsi: qla2xxx: edif: Add detection of secure device")
dd30706e73b7 ("scsi: qla2xxx: edif: Add key update")
fac2807946c1 ("scsi: qla2xxx: edif: Add extraction of auth_els from the wire")
84318a9f01ce ("scsi: qla2xxx: edif: Add send, receive, and accept for auth_els")
7878f22a2e03 ("scsi: qla2xxx: edif: Add getfcinfo and statistic bsgs")
7ebb336e45ef ("scsi: qla2xxx: edif: Add start + stop bsgs")
d94d8158e184 ("scsi: qla2xxx: Add heartbeat check")
f7a0ed479e66 ("scsi: qla2xxx: Fix crash in PCIe error handling")
2ce35c0821af ("scsi: qla2xxx: Fix use after free in bsg")
5777fef788a5 ("scsi: qla2xxx: Consolidate zio threshold setting for both FCP & NVMe")
960204ecca5e ("scsi: qla2xxx: Simplify if statement")
a04658594399 ("scsi: qla2xxx: Wait for ABTS response on I/O timeouts for NVMe")
dbf1f53cfd23 ("scsi: qla2xxx: Implementation to get and manage host, target stats and initiator port")
707531bc2626 ("scsi: qla2xxx: If fcport is undergoing deletion complete I/O with retry")
605e74025f95 ("scsi: qla2xxx: Move sess cmd list/lock to driver")
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 68ad83188d782b2ecef2e41ac245d27e0710fe8e Mon Sep 17 00:00:00 2001
From: Arun Easi <aeasi@xxxxxxxxxxx>
Date: Tue, 29 Nov 2022 01:26:34 -0800
Subject: [PATCH] scsi: qla2xxx: Fix crash when I/O abort times out
While performing CPU hotplug, a crash with the following stack was seen:
Call Trace:
qla24xx_process_response_queue+0x42a/0x970 [qla2xxx]
qla2x00_start_nvme_mq+0x3a2/0x4b0 [qla2xxx]
qla_nvme_post_cmd+0x166/0x240 [qla2xxx]
nvme_fc_start_fcp_op.part.0+0x119/0x2e0 [nvme_fc]
blk_mq_dispatch_rq_list+0x17b/0x610
__blk_mq_sched_dispatch_requests+0xb0/0x140
blk_mq_sched_dispatch_requests+0x30/0x60
__blk_mq_run_hw_queue+0x35/0x90
__blk_mq_delay_run_hw_queue+0x161/0x180
blk_execute_rq+0xbe/0x160
__nvme_submit_sync_cmd+0x16f/0x220 [nvme_core]
nvmf_connect_admin_queue+0x11a/0x170 [nvme_fabrics]
nvme_fc_create_association.cold+0x50/0x3dc [nvme_fc]
nvme_fc_connect_ctrl_work+0x19/0x30 [nvme_fc]
process_one_work+0x1e8/0x3c0
On abort timeout, completion was called without checking if the I/O was
already completed.
Verify that I/O and abort request are indeed outstanding before attempting
completion.
Fixes: 71c80b75ce8f ("scsi: qla2xxx: Do command completion on abort timeout")
Reported-by: Marco Patalano <mpatalan@xxxxxxxxxx>
Tested-by: Marco Patalano <mpatalan@xxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Arun Easi <aeasi@xxxxxxxxxxx>
Signed-off-by: Nilesh Javali <njavali@xxxxxxxxxxx>
Link: https://lore.kernel.org/r/20221129092634.15347-1-njavali@xxxxxxxxxxx
Signed-off-by: Martin K. Petersen <martin.petersen@xxxxxxxxxx>
diff --git a/drivers/scsi/qla2xxx/qla_init.c b/drivers/scsi/qla2xxx/qla_init.c
index ce4c5d728407..8d9ecabb1aac 100644
--- a/drivers/scsi/qla2xxx/qla_init.c
+++ b/drivers/scsi/qla2xxx/qla_init.c
@@ -110,6 +110,7 @@ static void qla24xx_abort_iocb_timeout(void *data)
struct qla_qpair *qpair = sp->qpair;
u32 handle;
unsigned long flags;
+ int sp_found = 0, cmdsp_found = 0;
if (sp->cmd_sp)
ql_dbg(ql_dbg_async, sp->vha, 0x507c,
@@ -124,18 +125,21 @@ static void qla24xx_abort_iocb_timeout(void *data)
spin_lock_irqsave(qpair->qp_lock_ptr, flags);
for (handle = 1; handle < qpair->req->num_outstanding_cmds; handle++) {
if (sp->cmd_sp && (qpair->req->outstanding_cmds[handle] ==
- sp->cmd_sp))
+ sp->cmd_sp)) {
qpair->req->outstanding_cmds[handle] = NULL;
+ cmdsp_found = 1;
+ }
/* removing the abort */
if (qpair->req->outstanding_cmds[handle] == sp) {
qpair->req->outstanding_cmds[handle] = NULL;
+ sp_found = 1;
break;
}
}
spin_unlock_irqrestore(qpair->qp_lock_ptr, flags);
- if (sp->cmd_sp) {
+ if (cmdsp_found && sp->cmd_sp) {
/*
* This done function should take care of
* original command ref: INIT
@@ -143,8 +147,10 @@ static void qla24xx_abort_iocb_timeout(void *data)
sp->cmd_sp->done(sp->cmd_sp, QLA_OS_TIMER_EXPIRED);
}
- abt->u.abt.comp_status = cpu_to_le16(CS_TIMEOUT);
- sp->done(sp, QLA_OS_TIMER_EXPIRED);
+ if (sp_found) {
+ abt->u.abt.comp_status = cpu_to_le16(CS_TIMEOUT);
+ sp->done(sp, QLA_OS_TIMER_EXPIRED);
+ }
}
static void qla24xx_abort_sp_done(srb_t *sp, int res)
