* [PATCH 1/2] drm/radeon: Don't hang in radeon_flip_work_func on disabled crtc. (v2)
2016-02-19 1:06 radeon/amdgpu pageflip regression fixes for Linux 4.4+ Mario Kleiner
@ 2016-02-19 1:06 ` Mario Kleiner
2016-02-19 1:16 ` Michel Dänzer
2016-02-19 1:06 ` [PATCH 2/2] drm/amdgpu: Don't hang in amdgpu_flip_work_func on disabled crtc Mario Kleiner
1 sibling, 1 reply; 5+ messages in thread
From: Mario Kleiner @ 2016-02-19 1:06 UTC (permalink / raw)
To: dri-devel; +Cc: Alex Deucher, Michel Dänzer, stable, =mario.kleiner.de
This fixes a regression introduced in Linux 4.4.
Limit the amount of time radeon_flip_work_func can
delay programming a page flip, by both limiting the
maximum amount of time per wait cycle and the maximum
number of wait cycles. Continue the flip if the limit
is exceeded, even if that may result in a visual or
timing glitch.
This is to prevent a hang of page flips, as reported
in fdo bug #93746: Disconnecting a DisplayPort display
in parallel to a kms pageflip getting queued can cause
the following hang of page flips and thereby an unusable
desktop:
1. kms pageflip ioctl() queues pageflip -> queues execution
of radeon_flip_work_func.
2. Hotunplug of display causes the driver to DPMS OFF
the unplugged display. Display engine shuts down,
scanout no longer moves, but stays at its resting
position at start line of vblank.
3. radeon_flip_work_func executes while crtc is off, and
due to the non-moving scanout position, the new flip
delay code introduced into Linux 4.4 by
commit 5b5561b3660d ("drm/radeon: Fixup hw vblank counter/ts..")
enters an infinite wait loop.
4. After reconnecting the display, the pageflip continues
to hang in 3. and the display doesn't update its view
of the desktop.
This patch fixes the Linux 4.4 regression from fdo bug #93746
<https://bugs.freedesktop.org/show_bug.cgi?id=93746>
v2: Skip wait immediately if !radeon_crtc->enabled, as
suggested by Michel.
Reported-by: Bernd Steinhauser <linux@bernd-steinhauser.de>
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Tested-by: Bernd Steinhauser <linux@bernd-steinhauser.de>
Cc: <stable@vger.kernel.org> # 4.4+
Cc: Michel Dänzer <michel.daenzer@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/radeon/radeon_display.c | 17 +++++++++++++++--
1 file changed, 15 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c
index b3bb923..1fab4b9 100644
--- a/drivers/gpu/drm/radeon/radeon_display.c
+++ b/drivers/gpu/drm/radeon/radeon_display.c
@@ -403,7 +403,8 @@ static void radeon_flip_work_func(struct work_struct *__work)
struct drm_crtc *crtc = &radeon_crtc->base;
unsigned long flags;
int r;
- int vpos, hpos, stat, min_udelay;
+ int vpos, hpos, stat, min_udelay = 0;
+ unsigned repcnt = 4;
struct drm_vblank_crtc *vblank = &crtc->dev->vblank[work->crtc_id];
down_read(&rdev->exclusive_lock);
@@ -454,7 +455,7 @@ static void radeon_flip_work_func(struct work_struct *__work)
* In practice this won't execute very often unless on very fast
* machines because the time window for this to happen is very small.
*/
- for (;;) {
+ while (radeon_crtc->enabled && repcnt--) {
/* GET_DISTANCE_TO_VBLANKSTART returns distance to real vblank
* start in hpos, and to the "fudged earlier" vblank start in
* vpos.
@@ -472,10 +473,22 @@ static void radeon_flip_work_func(struct work_struct *__work)
/* Sleep at least until estimated real start of hw vblank */
spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
min_udelay = (-hpos + 1) * max(vblank->linedur_ns / 1000, 5);
+ if (min_udelay > vblank->framedur_ns / 2000) {
+ /* Don't wait ridiculously long - something is wrong */
+ repcnt = 0;
+ break;
+ }
usleep_range(min_udelay, 2 * min_udelay);
spin_lock_irqsave(&crtc->dev->event_lock, flags);
};
+ if (!repcnt)
+ DRM_DEBUG_DRIVER("Delay problem on crtc %d: min_udelay %d, "
+ "framedur %d, linedur %d, stat %d, vpos %d, "
+ "hpos %d\n", work->crtc_id, min_udelay,
+ vblank->framedur_ns / 1000,
+ vblank->linedur_ns / 1000, stat, vpos, hpos);
+
/* do the flip (mmio) */
radeon_page_flip(rdev, radeon_crtc->crtc_id, work->base);
--
1.9.1
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
^ permalink raw reply related [flat|nested] 5+ messages in thread* [PATCH 2/2] drm/amdgpu: Don't hang in amdgpu_flip_work_func on disabled crtc.
2016-02-19 1:06 radeon/amdgpu pageflip regression fixes for Linux 4.4+ Mario Kleiner
2016-02-19 1:06 ` [PATCH 1/2] drm/radeon: Don't hang in radeon_flip_work_func on disabled crtc. (v2) Mario Kleiner
@ 2016-02-19 1:06 ` Mario Kleiner
1 sibling, 0 replies; 5+ messages in thread
From: Mario Kleiner @ 2016-02-19 1:06 UTC (permalink / raw)
To: dri-devel; +Cc: Alex Deucher, Michel Dänzer, stable, =mario.kleiner.de
This fixes a regression introduced in Linux 4.4.
This is a port of the same fix for radeon-kms in the
patch "drm/radeon: Don't hang in radeon_flip_work_func
on disabled crtc. (v2)"
Limit the amount of time amdgpu_flip_work_func can
delay programming a page flip, by both limiting the
maximum amount of time per wait cycle and the maximum
number of wait cycles. Continue the flip if the limit
is exceeded, even if that may result in a visual or
timing glitch.
This is to prevent a hang of page flips, as reported
in fdo bug #93746: Disconnecting a DisplayPort display
in parallel to a kms pageflip getting queued can cause
the following hang of page flips and thereby an unusable
desktop:
1. kms pageflip ioctl() queues pageflip -> queues execution
of amdgpu_flip_work_func.
2. Hotunplug of display causes the driver to DPMS OFF
the unplugged display. Display engine shuts down,
scanout no longer moves, but stays at its resting
position at start line of vblank.
3. amdgpu_flip_work_func executes while crtc is off, and
due to the non-moving scanout position, the new flip
delay code introduced into Linux 4.4 by
commit 8e36f9d33c13 ("drm/amdgpu: Fixup hw vblank counter/ts..")
enters an infinite wait loop.
4. After reconnecting the display, the pageflip continues
to hang in 3. and the display doesn't update its view
of the desktop.
This patch fixes the Linux 4.4 regression from fdo bug #93746
<https://bugs.freedesktop.org/show_bug.cgi?id=93746>
Reported-by: Bernd Steinhauser <linux@bernd-steinhauser.de>
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: <stable@vger.kernel.org> # 4.4+
Cc: Michel Dänzer <michel.daenzer@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 18 +++++++++++++++---
1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index acd066d0..8297bc3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -72,8 +72,8 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
struct drm_crtc *crtc = &amdgpuCrtc->base;
unsigned long flags;
- unsigned i;
- int vpos, hpos, stat, min_udelay;
+ unsigned i, repcnt = 4;
+ int vpos, hpos, stat, min_udelay = 0;
struct drm_vblank_crtc *vblank = &crtc->dev->vblank[work->crtc_id];
amdgpu_flip_wait_fence(adev, &work->excl);
@@ -96,7 +96,7 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
* In practice this won't execute very often unless on very fast
* machines because the time window for this to happen is very small.
*/
- for (;;) {
+ while (amdgpuCrtc->enabled && repcnt--) {
/* GET_DISTANCE_TO_VBLANKSTART returns distance to real vblank
* start in hpos, and to the "fudged earlier" vblank start in
* vpos.
@@ -114,10 +114,22 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
/* Sleep at least until estimated real start of hw vblank */
spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
min_udelay = (-hpos + 1) * max(vblank->linedur_ns / 1000, 5);
+ if (min_udelay > vblank->framedur_ns / 2000) {
+ /* Don't wait ridiculously long - something is wrong */
+ repcnt = 0;
+ break;
+ }
usleep_range(min_udelay, 2 * min_udelay);
spin_lock_irqsave(&crtc->dev->event_lock, flags);
};
+ if (!repcnt)
+ DRM_DEBUG_DRIVER("Delay problem on crtc %d: min_udelay %d, "
+ "framedur %d, linedur %d, stat %d, vpos %d, "
+ "hpos %d\n", work->crtc_id, min_udelay,
+ vblank->framedur_ns / 1000,
+ vblank->linedur_ns / 1000, stat, vpos, hpos);
+
/* do the flip (mmio) */
adev->mode_info.funcs->page_flip(adev, work->crtc_id, work->base);
/* set the flip status */
--
1.9.1
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
^ permalink raw reply related [flat|nested] 5+ messages in thread