[PATCH 0/1] drm/amd: more VRR race fixes during IRQ handling

[PATCH 0/1] drm/amd: more VRR race fixes during IRQ handling

[tjakobi]

From: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>

Hello,

this is a follow-up for these (already merged) patches:
https://lore.kernel.org/all/cover.1725269643.git.tjakobi@math.uni-bielefeld.de/

A user on the GitLab issue tracker pointed out that the problem
also manifests itself on DCE110 hardware.

This is a straight port of DCE10/DCE35 code.

With best wishes,
Tobias

Tobias Jakobi (1):
  drm/amd/display: handle nulled pipe context in DCE110's set_drr()

 .../amd/display/dc/hwss/dce110/dce110_hwseq.c | 21 ++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

-- 
2.44.2

[PATCH 1/1] drm/amd/display: handle nulled pipe context in DCE110's set_drr()

[tjakobi]

From: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>

As set_drr() is called from IRQ context, it can happen that the
pipe context has been nulled by dc_state_destruct().

Apply the same protection here that is already present for
dcn35_set_drr() and dcn10_set_drr(). I.e. fetch the tg pointer
first (to avoid a race with dc_state_destruct()), and then
check the local copy before using it.

Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3142
Fixes: 06ad7e164256 ("drm/amd/display: Destroy DC context while keeping DML and DML2")
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
---
 .../amd/display/dc/hwss/dce110/dce110_hwseq.c | 21 ++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c
index 0d3ea291eeee..666dfc6d192e 100644
--- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c
@@ -1970,13 +1970,20 @@ static void set_drr(struct pipe_ctx **pipe_ctx,
 	 * as well.
 	 */
 	for (i = 0; i < num_pipes; i++) {
-		pipe_ctx[i]->stream_res.tg->funcs->set_drr(
-			pipe_ctx[i]->stream_res.tg, &params);
-
-		if (adjust.v_total_max != 0 && adjust.v_total_min != 0)
-			pipe_ctx[i]->stream_res.tg->funcs->set_static_screen_control(
-					pipe_ctx[i]->stream_res.tg,
-					event_triggers, num_frames);
+		/* dc_state_destruct() might null the stream resources, so fetch tg
+		 * here first to avoid a race condition. The lifetime of the pointee
+		 * itself (the timing_generator object) is not a problem here.
+		 */
+		struct timing_generator *tg = pipe_ctx[i]->stream_res.tg;
+
+		if ((tg != NULL) && tg->funcs) {
+			if (tg->funcs->set_drr)
+				tg->funcs->set_drr(tg, &params);
+			if (adjust.v_total_max != 0 && adjust.v_total_min != 0)
+				if (tg->funcs->set_static_screen_control)
+					tg->funcs->set_static_screen_control(
+						tg, event_triggers, num_frames);
+		}
 	}
 }
 
-- 
2.44.2

Re: [PATCH 1/1] drm/amd/display: handle nulled pipe context in DCE110's set_drr()

[Alex Deucher]

Acked-by: Alex Deucher <alexander.deucher@amd.com>

And applied.

Thanks!

On Mon, Sep 16, 2024 at 9:03 AM <tjakobi@math.uni-bielefeld.de> wrote:
>
> From: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
>
> As set_drr() is called from IRQ context, it can happen that the
> pipe context has been nulled by dc_state_destruct().
>
> Apply the same protection here that is already present for
> dcn35_set_drr() and dcn10_set_drr(). I.e. fetch the tg pointer
> first (to avoid a race with dc_state_destruct()), and then
> check the local copy before using it.
>
> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3142
> Fixes: 06ad7e164256 ("drm/amd/display: Destroy DC context while keeping DML and DML2")
> Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
> ---
>  .../amd/display/dc/hwss/dce110/dce110_hwseq.c | 21 ++++++++++++-------
>  1 file changed, 14 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c
> index 0d3ea291eeee..666dfc6d192e 100644
> --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c
> +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c
> @@ -1970,13 +1970,20 @@ static void set_drr(struct pipe_ctx **pipe_ctx,
>          * as well.
>          */
>         for (i = 0; i < num_pipes; i++) {
> -               pipe_ctx[i]->stream_res.tg->funcs->set_drr(
> -                       pipe_ctx[i]->stream_res.tg, &params);
> -
> -               if (adjust.v_total_max != 0 && adjust.v_total_min != 0)
> -                       pipe_ctx[i]->stream_res.tg->funcs->set_static_screen_control(
> -                                       pipe_ctx[i]->stream_res.tg,
> -                                       event_triggers, num_frames);
> +               /* dc_state_destruct() might null the stream resources, so fetch tg
> +                * here first to avoid a race condition. The lifetime of the pointee
> +                * itself (the timing_generator object) is not a problem here.
> +                */
> +               struct timing_generator *tg = pipe_ctx[i]->stream_res.tg;
> +
> +               if ((tg != NULL) && tg->funcs) {
> +                       if (tg->funcs->set_drr)
> +                               tg->funcs->set_drr(tg, &params);
> +                       if (adjust.v_total_max != 0 && adjust.v_total_min != 0)
> +                               if (tg->funcs->set_static_screen_control)
> +                                       tg->funcs->set_static_screen_control(
> +                                               tg, event_triggers, num_frames);
> +               }
>         }
>  }
>
> --
> 2.44.2
>