evie/android_kernel_oneplus_msm8998
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

block: test-iosched: fix spinlock recursion

Browse source 
spin_lock_irq() / spin_unlock_irq() is used so interrupts are
enabled after unlocking the spinlock. However, it is not guaranteed
they were enabled before.
This change uses the proper irqsave / irqrestore variants instead.
Without it, a spinlock recursion on the scsi request completion path
is possible if completion interrupt occurs when used for UFS testing.

Change-Id: I25a9bf6faaa2bbfedc807111fbcb32276cccea2f
Signed-off-by: Gilad Broner <gbroner@codeaurora.org>
This commit is contained in:
Gilad Broner 2015-03-31 16:17:25 +03:00 • committed by David Keitel
parent b5bec249a9
commit a27d5a2f83
1 changed files with 13 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

21
block/test-iosched.c
View file

@ -167,6 +167,7 @@ int test_iosched_add_unique_test_req(struct test_iosched *tios,
struct request *rq;
int rw_flags;
struct test_request *test_rq;
unsigned long flags;
if (!tios)
return -ENODEV;
@ -230,10 +231,10 @@ int test_iosched_add_unique_test_req(struct test_iosched *tios,
"%s: added request %d to the test requests list, type = %d",
__func__, test_rq->req_id, req_unique);
spin_lock_irq(tios->req_q->queue_lock);
spin_lock_irqsave(tios->req_q->queue_lock, flags);
list_add_tail(&test_rq->queuelist, &tios->test_queue);
tios->test_count++;
spin_unlock_irq(tios->req_q->queue_lock);
spin_unlock_irqrestore(tios->req_q->queue_lock, flags);
return 0;
}
@ -408,14 +409,15 @@ int test_iosched_add_wr_rd_test_req(struct test_iosched *tios,
int pattern, rq_end_io_fn *end_req_io)
{
struct test_request *test_rq = NULL;
unsigned long flags;
test_rq = test_iosched_create_test_req(tios, is_err_expcted, direction,
start_sec, num_bios, pattern, end_req_io);
if (test_rq) {
spin_lock_irq(tios->req_q->queue_lock);
spin_lock_irqsave(tios->req_q->queue_lock, flags);
list_add_tail(&test_rq->queuelist, &tios->test_queue);
tios->test_count++;
spin_unlock_irq(tios->req_q->queue_lock);
spin_unlock_irqrestore(tios->req_q->queue_lock, flags);
return 0;
}
return -ENODEV;
@ -1106,6 +1108,7 @@ static int test_init_queue(struct request_queue *q, struct elevator_type *e)
const char *blk_dev_name;
int ret;
bool found = false;
unsigned long flags;
eq = elevator_alloc(q, e);
if (!eq)
@ -1168,9 +1171,9 @@ static int test_init_queue(struct request_queue *q, struct elevator_type *e)
}
}
spin_lock_irq(q->queue_lock);
spin_lock_irqsave(q->queue_lock, flags);
q->elevator = eq;
spin_unlock_irq(q->queue_lock);
spin_unlock_irqrestore(q->queue_lock, flags);
return 0;
@ -1209,14 +1212,16 @@ static void test_exit_queue(struct elevator_queue *e)
void test_iosched_add_urgent_req(struct test_iosched *tios,
struct test_request *test_rq)
{
unsigned long flags;
if (!tios)
return;
spin_lock_irq(&tios->lock);
spin_lock_irqsave(&tios->lock, flags);
test_rq->rq->cmd_flags |= REQ_URGENT;
list_add_tail(&test_rq->queuelist, &tios->urgent_queue);
tios->urgent_count++;
spin_unlock_irq(&tios->lock);
spin_unlock_irqrestore(&tios->lock, flags);
}
EXPORT_SYMBOL(test_iosched_add_urgent_req);