From: John Garry <john.garry@huawei.com>
To: <jejb@linux.vnet.ibm.com>, <martin.petersen@oracle.com>
Cc: <linuxarm@huawei.com>, <linux-scsi@vger.kernel.org>,
<linux-kernel@vger.kernel.org>,
Xiang Chen <chenxiang66@hisilicon.com>,
"John Garry" <john.garry@huawei.com>
Subject: [PATCH 3/7] scsi: hisi_sas: Fix the race between IO completion and timeout for SMP/internal IO
Date: Mon, 24 Sep 2018 23:06:30 +0800 [thread overview]
Message-ID: <1537801594-207139-4-git-send-email-john.garry@huawei.com> (raw)
In-Reply-To: <1537801594-207139-1-git-send-email-john.garry@huawei.com>
From: Xiang Chen <chenxiang66@hisilicon.com>
If SMP/internal IO times out, we will possibly free the task immediately.
However if the IO actually completes at the same time, the IO completion
may refer to task which have been freed.
So to solve the issue, flush the tasklet to finish IO completion before
free'ing slot/task.
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
---
drivers/scsi/hisi_sas/hisi_sas_main.c | 55 +++++++++++++++++++++++++++++------
1 file changed, 46 insertions(+), 9 deletions(-)
diff --git a/drivers/scsi/hisi_sas/hisi_sas_main.c b/drivers/scsi/hisi_sas/hisi_sas_main.c
index 3b95a7a..416f2c0 100644
--- a/drivers/scsi/hisi_sas/hisi_sas_main.c
+++ b/drivers/scsi/hisi_sas/hisi_sas_main.c
@@ -956,8 +956,7 @@ static int hisi_sas_control_phy(struct asd_sas_phy *sas_phy, enum phy_func func,
static void hisi_sas_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);
}
@@ -966,13 +965,17 @@ static void hisi_sas_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;
+ bool is_completed = true;
spin_lock_irqsave(&task->task_state_lock, flags);
- if (!(task->task_state_flags & SAS_TASK_STATE_DONE))
+ if (!(task->task_state_flags & SAS_TASK_STATE_DONE)) {
task->task_state_flags |= SAS_TASK_STATE_ABORTED;
+ is_completed = false;
+ }
spin_unlock_irqrestore(&task->task_state_lock, flags);
- complete(&task->slow_task->completion);
+ if (!is_completed)
+ complete(&task->slow_task->completion);
}
#define TASK_TIMEOUT 20
@@ -1023,10 +1026,18 @@ static int hisi_sas_exec_internal_tmf_task(struct domain_device *device,
if ((task->task_state_flags & SAS_TASK_STATE_ABORTED)) {
if (!(task->task_state_flags & SAS_TASK_STATE_DONE)) {
struct hisi_sas_slot *slot = task->lldd_task;
+ struct hisi_sas_cq *cq =
+ &hisi_hba->cq[slot->dlvry_queue];
dev_err(dev, "abort tmf: TMF task timeout and not done\n");
- if (slot)
+ if (slot) {
+ /*
+ * flush tasklet to avoid free'ing task
+ * before using task in IO completion
+ */
+ tasklet_kill(&cq->tasklet);
slot->task = NULL;
+ }
goto ex_err;
} else
@@ -1402,6 +1413,17 @@ static int hisi_sas_abort_task(struct sas_task *task)
spin_lock_irqsave(&task->task_state_lock, flags);
if (task->task_state_flags & SAS_TASK_STATE_DONE) {
+ struct hisi_sas_slot *slot = task->lldd_task;
+ struct hisi_sas_cq *cq;
+
+ if (slot) {
+ /*
+ * flush tasklet to avoid free'ing task
+ * before using task in IO completion
+ */
+ cq = &hisi_hba->cq[slot->dlvry_queue];
+ tasklet_kill(&cq->tasklet);
+ }
spin_unlock_irqrestore(&task->task_state_lock, flags);
rc = TMF_RESP_FUNC_COMPLETE;
goto out;
@@ -1457,12 +1479,19 @@ static int hisi_sas_abort_task(struct sas_task *task)
/* SMP */
struct hisi_sas_slot *slot = task->lldd_task;
u32 tag = slot->idx;
+ struct hisi_sas_cq *cq = &hisi_hba->cq[slot->dlvry_queue];
rc = hisi_sas_internal_task_abort(hisi_hba, device,
HISI_SAS_INT_ABT_CMD, tag);
if (((rc < 0) || (rc == TMF_RESP_FUNC_FAILED)) &&
- task->lldd_task)
- hisi_sas_do_release_task(hisi_hba, task, slot);
+ task->lldd_task) {
+ /*
+ * flush tasklet to avoid free'ing task
+ * before using task in IO completion
+ */
+ tasklet_kill(&cq->tasklet);
+ slot->task = NULL;
+ }
}
out:
@@ -1828,9 +1857,17 @@ static int hisi_sas_query_task(struct sas_task *task)
if ((task->task_state_flags & SAS_TASK_STATE_ABORTED)) {
if (!(task->task_state_flags & SAS_TASK_STATE_DONE)) {
struct hisi_sas_slot *slot = task->lldd_task;
-
- if (slot)
+ struct hisi_sas_cq *cq =
+ &hisi_hba->cq[slot->dlvry_queue];
+
+ if (slot) {
+ /*
+ * flush tasklet to avoid free'ing task
+ * before using task in IO completion
+ */
+ tasklet_kill(&cq->tasklet);
slot->task = NULL;
+ }
dev_err(dev, "internal task abort: timeout and not done.\n");
res = -EIO;
goto exit;
--
1.9.1
next prev parent reply other threads:[~2018-09-24 15:09 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-24 15:06 [PATCH 0/7] hisi_sas: Misc bugfixes and an optimisation patch John Garry
2018-09-24 15:06 ` [PATCH 1/7] scsi: hisi_sas: Feed back linkrate(max/min) when re-attached John Garry
2018-09-24 15:06 ` [PATCH 2/7] scsi: hisi_sas: Move evaluation of hisi_hba in hisi_sas_task_prep() John Garry
2018-09-24 15:06 ` John Garry [this message]
2018-09-24 15:06 ` [PATCH 4/7] scsi: hisi_sas: Free slot later in slot_complete_vx_hw() John Garry
2018-09-24 15:06 ` [PATCH 5/7] scsi: hisi_sas: unmask interrupts ent72 and ent74 John Garry
2018-09-24 15:06 ` [PATCH 6/7] scsi: hisi_sas: Use block layer tag instead for IPTT John Garry
2018-09-24 15:06 ` [PATCH 7/7] scsi: hisi_sas: Update v3 hw AIP_LIMIT and CFG_AGING_TIME register values John Garry
2018-10-04 15:30 ` [PATCH 0/7] hisi_sas: Misc bugfixes and an optimisation patch John Garry
2018-10-11 1:58 ` Martin K. Petersen
2018-10-11 6:36 ` Christoph Hellwig
2018-10-11 9:59 ` John Garry
2018-10-11 10:15 ` Christoph Hellwig
2018-10-11 13:12 ` John Garry
2018-10-11 13:32 ` Ming Lei
2018-10-11 14:07 ` John Garry
2018-10-11 23:36 ` Ming Lei
2018-10-12 9:02 ` John Garry
2018-10-12 13:30 ` Ming Lei
2018-10-12 10:47 ` John Garry
2018-10-16 4:28 ` Martin K. Petersen
2018-10-16 8:28 ` John Garry
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1537801594-207139-4-git-send-email-john.garry@huawei.com \
--to=john.garry@huawei.com \
--cc=chenxiang66@hisilicon.com \
--cc=jejb@linux.vnet.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=linuxarm@huawei.com \
--cc=martin.petersen@oracle.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®