Skip to content

Commit a3cc326

Browse files
tobiasjakobialexdeucher
authored andcommitted
drm/amd/display: Avoid race between dcn10_set_drr() and dc_state_destruct()
dc_state_destruct() nulls the resource context of the DC state. The pipe context passed to dcn10_set_drr() is a member of this resource context. If dc_state_destruct() is called parallel to the IRQ processing (which calls dcn10_set_drr() at some point), we can end up using already nulled function callback fields of struct stream_resource. The logic in dcn10_set_drr() already tries to avoid this, by checking tg against NULL. But if the nulling happens exactly after the NULL check and before the next access, then we get a race. Avoid this by copying tg first to a local variable, and then use this variable for all the operations. This should work, as long as nobody frees the resource pool where the timing generators live. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3142 Fixes: 06ad7e1 ("drm/amd/display: Destroy DC context while keeping DML and DML2") Signed-off-by: Tobias Jakobi <[email protected]> Tested-by: Raoul van Rüschen <[email protected]> Tested-by: Christopher Snowhill <[email protected]> Reviewed-by: Harry Wentland <[email protected]> Tested-by: Sefa Eyeoglu <[email protected]> Signed-off-by: Mario Limonciello <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
1 parent 760e3c8 commit a3cc326

File tree

1 file changed

+12
-8
lines changed

1 file changed

+12
-8
lines changed

drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c

Lines changed: 12 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -3212,15 +3212,19 @@ void dcn10_set_drr(struct pipe_ctx **pipe_ctx,
32123212
* as well.
32133213
*/
32143214
for (i = 0; i < num_pipes; i++) {
3215-
if ((pipe_ctx[i]->stream_res.tg != NULL) && pipe_ctx[i]->stream_res.tg->funcs) {
3216-
if (pipe_ctx[i]->stream_res.tg->funcs->set_drr)
3217-
pipe_ctx[i]->stream_res.tg->funcs->set_drr(
3218-
pipe_ctx[i]->stream_res.tg, &params);
3215+
/* dc_state_destruct() might null the stream resources, so fetch tg
3216+
* here first to avoid a race condition. The lifetime of the pointee
3217+
* itself (the timing_generator object) is not a problem here.
3218+
*/
3219+
struct timing_generator *tg = pipe_ctx[i]->stream_res.tg;
3220+
3221+
if ((tg != NULL) && tg->funcs) {
3222+
if (tg->funcs->set_drr)
3223+
tg->funcs->set_drr(tg, &params);
32193224
if (adjust.v_total_max != 0 && adjust.v_total_min != 0)
3220-
if (pipe_ctx[i]->stream_res.tg->funcs->set_static_screen_control)
3221-
pipe_ctx[i]->stream_res.tg->funcs->set_static_screen_control(
3222-
pipe_ctx[i]->stream_res.tg,
3223-
event_triggers, num_frames);
3225+
if (tg->funcs->set_static_screen_control)
3226+
tg->funcs->set_static_screen_control(
3227+
tg, event_triggers, num_frames);
32243228
}
32253229
}
32263230
}

0 commit comments

Comments
 (0)