2020-10-29 17:03:45

by Steven Price

Subject: [PATCH] drm/panfrost: Don't corrupt the queue mutex on open/close

The mutex within the panfrost_queue_state should have the lifetime of
the queue, however it was erroneously initialised/destroyed during
panfrost_job_{open,close} which is called every time a client
opens/closes the drm node.

Move the initialisation/destruction to panfrost_job_{init,fini} where it
belongs.

Fixes: 1a11a88cfd9a ("drm/panfrost: Fix job timeout handling")
Signed-off-by: Steven Price <[email protected]>
Reviewed-by: Boris Brezillon <[email protected]>
---
drivers/gpu/drm/panfrost/panfrost_job.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index cfb431624eea..145ad37eda6a 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -595,6 +595,8 @@ int panfrost_job_init(struct panfrost_device *pfdev)
}

for (j = 0; j < NUM_JOB_SLOTS; j++) {
+ mutex_init(&js->queue[j].lock);
+
js->queue[j].fence_context = dma_fence_context_alloc(1);

ret = drm_sched_init(&js->queue[j].sched,
@@ -625,8 +627,10 @@ void panfrost_job_fini(struct panfrost_device *pfdev)

job_write(pfdev, JOB_INT_MASK, 0);

- for (j = 0; j < NUM_JOB_SLOTS; j++)
+ for (j = 0; j < NUM_JOB_SLOTS; j++) {
drm_sched_fini(&js->queue[j].sched);
+ mutex_destroy(&js->queue[j].lock);
+ }

}

@@ -638,7 +642,6 @@ int panfrost_job_open(struct panfrost_file_priv *panfrost_priv)
int ret, i;

for (i = 0; i < NUM_JOB_SLOTS; i++) {
- mutex_init(&js->queue[i].lock);
sched = &js->queue[i].sched;
ret = drm_sched_entity_init(&panfrost_priv->sched_entity[i],
DRM_SCHED_PRIORITY_NORMAL, &sched,
@@ -657,7 +660,6 @@ void panfrost_job_close(struct panfrost_file_priv *panfrost_priv)

for (i = 0; i < NUM_JOB_SLOTS; i++) {
drm_sched_entity_destroy(&panfrost_priv->sched_entity[i]);
- mutex_destroy(&js->queue[i].lock);
/* Ensure any timeouts relating to this client have completed */
flush_delayed_work(&js->queue[i].sched.work_tdr);
}
--
2.20.1

2020-10-30 09:22:02

by Boris Brezillon

Subject: Re: [PATCH] drm/panfrost: Don't corrupt the queue mutex on open/close

On Thu, 29 Oct 2020 17:00:47 +0000
Steven Price <[email protected]> wrote:

> The mutex within the panfrost_queue_state should have the lifetime of
> the queue, however it was erroneously initialised/destroyed during
> panfrost_job_{open,close} which is called every time a client
> opens/closes the drm node.
>
> Move the initialisation/destruction to panfrost_job_{init,fini} where it
> belongs.
>

Queued to drm-misc-next.

Thanks,

Boris

> Fixes: 1a11a88cfd9a ("drm/panfrost: Fix job timeout handling")
> Signed-off-by: Steven Price <[email protected]>
> Reviewed-by: Boris Brezillon <[email protected]>
> ---
> drivers/gpu/drm/panfrost/panfrost_job.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
> index cfb431624eea..145ad37eda6a 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_job.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_job.c
> @@ -595,6 +595,8 @@ int panfrost_job_init(struct panfrost_device *pfdev)
> }
>
> for (j = 0; j < NUM_JOB_SLOTS; j++) {
> + mutex_init(&js->queue[j].lock);
> +
> js->queue[j].fence_context = dma_fence_context_alloc(1);
>
> ret = drm_sched_init(&js->queue[j].sched,
> @@ -625,8 +627,10 @@ void panfrost_job_fini(struct panfrost_device *pfdev)
>
> job_write(pfdev, JOB_INT_MASK, 0);
>
> - for (j = 0; j < NUM_JOB_SLOTS; j++)
> + for (j = 0; j < NUM_JOB_SLOTS; j++) {
> drm_sched_fini(&js->queue[j].sched);
> + mutex_destroy(&js->queue[j].lock);
> + }
>
> }
>
> @@ -638,7 +642,6 @@ int panfrost_job_open(struct panfrost_file_priv *panfrost_priv)
> int ret, i;
>
> for (i = 0; i < NUM_JOB_SLOTS; i++) {
> - mutex_init(&js->queue[i].lock);
> sched = &js->queue[i].sched;
> ret = drm_sched_entity_init(&panfrost_priv->sched_entity[i],
> DRM_SCHED_PRIORITY_NORMAL, &sched,
> @@ -657,7 +660,6 @@ void panfrost_job_close(struct panfrost_file_priv *panfrost_priv)
>
> for (i = 0; i < NUM_JOB_SLOTS; i++) {
> drm_sched_entity_destroy(&panfrost_priv->sched_entity[i]);
> - mutex_destroy(&js->queue[i].lock);
> /* Ensure any timeouts relating to this client have completed */
> flush_delayed_work(&js->queue[i].sched.work_tdr);
> }