From: Bart Van Assche bvanassche@acm.org
[ Upstream commit 8dd9593cc07ad7d999bef81b06789ef873a94881 ]
I'm not sure how this happened but the patch that was intended to fix abort handling was incomplete. This patch fixes that patch as follows:
- If aborting the SCSI command failed, wait until the SCSI command completes.
- Return SUCCESS instead of FAILED if an abort attempt races with SCSI command completion.
- Since qla2xxx_eh_abort() increments the sp reference count by calling sp_get(), decrement the sp reference count before returning.
Cc: Himanshu Madhani hmadhani@marvell.com Fixes: 219d27d7147e ("scsi: qla2xxx: Fix race conditions in the code for aborting SCSI commands") Signed-off-by: Bart Van Assche bvanassche@acm.org Tested-by: Himanshu Madhani hmadhani@marvell.com Reviewed-by: Himanshu Madhani hmadhani@marvell.com Signed-off-by: Martin K. Petersen martin.petersen@oracle.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/scsi/qla2xxx/qla_os.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+)
diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c index 82f6ae4dcfc0b..0f51387ceebdf 100644 --- a/drivers/scsi/qla2xxx/qla_os.c +++ b/drivers/scsi/qla2xxx/qla_os.c @@ -1274,6 +1274,7 @@ static int qla2xxx_eh_abort(struct scsi_cmnd *cmd) { scsi_qla_host_t *vha = shost_priv(cmd->device->host); + DECLARE_COMPLETION_ONSTACK(comp); srb_t *sp; int ret; unsigned int id; @@ -1309,6 +1310,7 @@ qla2xxx_eh_abort(struct scsi_cmnd *cmd) return SUCCESS; }
+ /* Get a reference to the sp and drop the lock. */ if (sp_get(sp)){ /* ref_count is already 0 */ spin_unlock_irqrestore(qpair->qp_lock_ptr, flags); @@ -1336,6 +1338,23 @@ qla2xxx_eh_abort(struct scsi_cmnd *cmd) sp->done(sp, DID_ABORT << 16); ret = SUCCESS; break; + case QLA_FUNCTION_PARAMETER_ERROR: { + /* Wait for the command completion. */ + uint32_t ratov = ha->r_a_tov/10; + uint32_t ratov_j = msecs_to_jiffies(4 * ratov * 1000); + + WARN_ON_ONCE(sp->comp); + sp->comp = ∁ + if (!wait_for_completion_timeout(&comp, ratov_j)) { + ql_dbg(ql_dbg_taskm, vha, 0xffff, + "%s: Abort wait timer (4 * R_A_TOV[%d]) expired\n", + __func__, ha->r_a_tov); + ret = FAILED; + } else { + ret = SUCCESS; + } + break; + } default: /* * Either abort failed or abort and completion raced. Let @@ -1345,6 +1364,8 @@ qla2xxx_eh_abort(struct scsi_cmnd *cmd) break; }
+ sp->comp = NULL; + atomic_dec(&sp->ref_count); ql_log(ql_log_info, vha, 0x801c, "Abort command issued nexus=%ld:%d:%llu -- %x.\n", vha->host_no, id, lun, ret);