From: Wayne Lin <Wayne.Lin@xxxxxxx>
[Why]
Find out few locks problems while doing secure display. They are
following few parts:
1. crc_rd_work_lock in amdgpu_dm_crtc_handle_crc_window_irq() should
also use spin_lock_irqsave instead of spin_lock_irq.
2. In crc_win_update_set(), crc_rd_work_lock should be grabbed after
obtaining lock event_lock. Otherwise, will cause deadlock by conflicting
the lock order in amdgpu_dm_crtc_handle_crc_window_irq()
3. flush_work() in crc_win_update_set() is no need and will cause
deadlock since amdgpu_dm_crtc_notify_ta_to_read() also tries to grab
lock crc_rd_work_lock.
[How]
Fix above problems.
Signed-off-by: Wayne Lin <Wayne.Lin@xxxxxxx>
Reviewed-by: Solomon Chiu <Solomon.Chiu@xxxxxxx>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c | 10 +++++-----
.../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c | 6 ++----
2 files changed, 7 insertions(+), 9 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
index 3adbaf50a558..c6d6baab106e 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
@@ -433,7 +433,7 @@ void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc)
struct amdgpu_device *adev = NULL;
struct crc_rd_work *crc_rd_wrk = NULL;
struct crc_params *crc_window = NULL, tmp_window;
- unsigned long flags;
+ unsigned long flags1, flags2;
struct crtc_position position;
uint32_t v_blank;
uint32_t v_back_porch;
@@ -447,7 +447,7 @@ void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc)
adev = drm_to_adev(crtc->dev);
drm_dev = crtc->dev;
- spin_lock_irqsave(&drm_dev->event_lock, flags);
+ spin_lock_irqsave(&drm_dev->event_lock, flags1);
stream_state = acrtc->dm_irq_params.stream;
cur_crc_src = acrtc->dm_irq_params.crc_src;
timing_out = &stream_state->timing;
@@ -508,10 +508,10 @@ void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc)
if (acrtc->dm_irq_params.crc_window.skip_frame_cnt == 0) {
if (adev->dm.crc_rd_wrk) {
crc_rd_wrk = adev->dm.crc_rd_wrk;
- spin_lock_irq(&crc_rd_wrk->crc_rd_work_lock);
+ spin_lock_irqsave(&crc_rd_wrk->crc_rd_work_lock, flags2);
crc_rd_wrk->phy_inst =
stream_state->link->link_enc_hw_inst;
- spin_unlock_irq(&crc_rd_wrk->crc_rd_work_lock);
+ spin_unlock_irqrestore(&crc_rd_wrk->crc_rd_work_lock, flags2);
schedule_work(&crc_rd_wrk->notify_ta_work);
}
} else {
@@ -522,7 +522,7 @@ void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc)
}
cleanup:
- spin_unlock_irqrestore(&drm_dev->event_lock, flags);
+ spin_unlock_irqrestore(&drm_dev->event_lock, flags1);
}
void amdgpu_dm_crtc_secure_display_resume(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
index 6d839d3fb6a3..b8644f49e0f2 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
@@ -2695,14 +2695,12 @@ static int crc_win_update_set(void *data, u64 val)
struct crc_rd_work *crc_rd_wrk = adev->dm.crc_rd_wrk;
if (val) {
- spin_lock_irq(&crc_rd_wrk->crc_rd_work_lock);
spin_lock_irq(&adev_to_drm(adev)->event_lock);
+ spin_lock_irq(&crc_rd_wrk->crc_rd_work_lock);
if (crc_rd_wrk && crc_rd_wrk->crtc) {
old_crtc = crc_rd_wrk->crtc;
old_acrtc = to_amdgpu_crtc(old_crtc);
- flush_work(&adev->dm.crc_rd_wrk->notify_ta_work);
}
-
new_acrtc = to_amdgpu_crtc(new_crtc);
if (old_crtc && old_crtc != new_crtc) {
@@ -2720,8 +2718,8 @@ static int crc_win_update_set(void *data, u64 val)
new_acrtc->dm_irq_params.crc_window.skip_frame_cnt = 0;
crc_rd_wrk->crtc = new_crtc;
}
- spin_unlock_irq(&adev_to_drm(adev)->event_lock);
spin_unlock_irq(&crc_rd_wrk->crc_rd_work_lock);
+ spin_unlock_irq(&adev_to_drm(adev)->event_lock);
}
return 0;
--
2.29.0
_______________________________________________
amd-gfx mailing list
amd-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
