scsi: pm80xx: Fix TMF task completion race condition

The TMF timeout timer may trigger at the same time when the response from a
controller is being handled. When this happens the SAS task may get freed
before the response processing is finished.

Fix this by calling complete() only when SAS_TASK_STATE_DONE is not set.

A similar race condition was fixed in commit b90cd6f2b905 ("scsi: libsas:
fix a race condition when smp task timeout")

Link: https://lore.kernel.org/r/20210707185945.35559-1-ipylypiv@google.com
Reviewed-by: Vishakha Channapattan <vishakhavc@google.com>
Acked-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Igor Pylypiv <ipylypiv@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

diff --git a/drivers/scsi/pm8001/pm8001_sas.c b/drivers/scsi/pm8001/pm8001_sas.c
index 48548a95327b5ca354ae6de99f811d4506673f0a..32e60f0c3b1483bdeb2f6aba6015ef7506a25387 100644 (file)
--- a/drivers/scsi/pm8001/pm8001_sas.c
+++ b/drivers/scsi/pm8001/pm8001_sas.c
@@ -684,8 +684,7 @@ int pm8001_dev_found(struct domain_device *dev)
 
 void pm8001_task_done(struct sas_task *task)
 {
-       if (!del_timer(&task->slow_task->timer))
-               return;
+       del_timer(&task->slow_task->timer);
        complete(&task->slow_task->completion);
 }
 
@@ -693,9 +692,14 @@ static void pm8001_tmf_timedout(struct timer_list *t)
 {
        struct sas_task_slow *slow = from_timer(slow, t, timer);
        struct sas_task *task = slow->task;
+       unsigned long flags;
 
-       task->task_state_flags |= SAS_TASK_STATE_ABORTED;
-       complete(&task->slow_task->completion);
+       spin_lock_irqsave(&task->task_state_lock, flags);
+       if (!(task->task_state_flags & SAS_TASK_STATE_DONE)) {
+               task->task_state_flags |= SAS_TASK_STATE_ABORTED;
+               complete(&task->slow_task->completion);
+       }
+       spin_unlock_irqrestore(&task->task_state_lock, flags);
 }
 
 #define PM8001_TASK_TIMEOUT 20
@@ -748,13 +752,10 @@ static int pm8001_exec_internal_tmf_task(struct domain_device *dev,
                }
                res = -TMF_RESP_FUNC_FAILED;
                /* Even TMF timed out, return direct. */
-               if ((task->task_state_flags & SAS_TASK_STATE_ABORTED)) {
-                       if (!(task->task_state_flags & SAS_TASK_STATE_DONE)) {
-                               pm8001_dbg(pm8001_ha, FAIL,
-                                          "TMF task[%x]timeout.\n",
-                                          tmf->tmf);
-                               goto ex_err;
-                       }
+               if (task->task_state_flags & SAS_TASK_STATE_ABORTED) {
+                       pm8001_dbg(pm8001_ha, FAIL, "TMF task[%x]timeout.\n",
+                                  tmf->tmf);
+                       goto ex_err;
                }
 
                if (task->task_status.resp == SAS_TASK_COMPLETE &&
@@ -834,12 +835,9 @@ pm8001_exec_internal_task_abort(struct pm8001_hba_info *pm8001_ha,
                wait_for_completion(&task->slow_task->completion);
                res = TMF_RESP_FUNC_FAILED;
                /* Even TMF timed out, return direct. */
-               if ((task->task_state_flags & SAS_TASK_STATE_ABORTED)) {
-                       if (!(task->task_state_flags & SAS_TASK_STATE_DONE)) {
-                               pm8001_dbg(pm8001_ha, FAIL,
-                                          "TMF task timeout.\n");
-                               goto ex_err;
-                       }
+               if (task->task_state_flags & SAS_TASK_STATE_ABORTED) {
+                       pm8001_dbg(pm8001_ha, FAIL, "TMF task timeout.\n");
+                       goto ex_err;
                }
 
                if (task->task_status.resp == SAS_TASK_COMPLETE &&