linux/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c, branch v4.6

drm/amdgpu: fix leaking fence in the pageflip code

2016-04-01T14:08:38Z

This fixes a memory leak when we can't register the callback on a fence. Signed-off-by: Christian König Reviewed-by: Alex Deucher Cc: stable@vger.kernel.org Signed-off-by: Alex Deucher

Merge drm-fixes into drm-next.

2016-03-13T23:46:02Z

Nouveau wanted this to avoid some worse conflicts when I merge that.

drm/amdgpu: Fix error handling in amdgpu_flip_work_func.

2016-03-05T17:31:45Z

The patch e1d09dc0ccc6: "drm/amdgpu: Don't hang in amdgpu_flip_work_func on disabled crtc." from Feb 19, 2016, leads to the following static checker warning, as reported by Dan Carpenter in https://lists.freedesktop.org/archives/dri-devel/2016-February/101987.html drivers/gpu/drm/amd/amdgpu/amdgpu_display.c:127 amdgpu_flip_work_func() warn: should this be 'repcnt == -1' drivers/gpu/drm/amd/amdgpu/amdgpu_display.c:136 amdgpu_flip_work_func() error: double unlock 'spin_lock:&crtc->dev->event_lock' drivers/gpu/drm/amd/amdgpu/amdgpu_display.c:136 amdgpu_flip_work_func() error: double unlock 'irqsave:flags' This patch fixes both reported problems: Change post-decrement of repcnt to pre-decrement, so it can't underflow anymore, but still performs up to three repetitions - three is the maximum one could expect in practice. Move the spin_unlock_irqrestore to where it actually belongs. Reviewed-by: Michel Dänzer Reported-by: Dan Carpenter Signed-off-by: Mario Kleiner Cc: # 4.4+ Cc: Michel Dänzer Cc: Alex Deucher Signed-off-by: Alex Deucher

drm/amdgpu: Don't hang in amdgpu_flip_work_func on disabled crtc.

2016-02-19T23:15:30Z

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 Reported-by: Bernd Steinhauser Signed-off-by: Mario Kleiner Cc: # 4.4+ Cc: Michel Dänzer Cc: Alex Deucher Reviewed-by: Michel Dänzer Signed-off-by: Alex Deucher

drm/amdgpu: remove page flip work queue v3

2016-02-16T22:25:38Z

Just use the system queue now that we don't block any more. v2: handle DAL as well. v3: agd: split DAL changes out Signed-off-by: Christian König Reviewed-by: Mykola Lysenko Acked-by: Alex Deucher (v1)

drm/amdgpu: stop blocking for page filp fences

2016-02-16T22:24:07Z

Just register an callback and reschedule the work item if necessary. Signed-off-by: Christian König Acked-by: Alex Deucher

drm/amdgpu: stop calling amdgpu_gpu_reset from the flip code

2016-02-16T22:23:31Z

We don't return -EDEADLK any more. Signed-off-by: Christian König Acked-by: Alex Deucher

drm/amdgpu: Move MMIO flip out of spinlocked region

2016-02-10T21:08:55Z

Prior actual MMIO flip we need to acquire DAL mutex to guard our target state which get modified on reset mode. Assign page flip status before actual flip to handle the possible race condition with interrupt. Signed-off-by: Vitaly Prosyak Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher

Back merge tag 'v4.4-rc4' into drm-next

2015-12-08T01:04:26Z

We've picked up a few conflicts and it would be nice to resolve them before we move onwards.

drm/amdgpu: Fixup hw vblank counter/ts for new drm_update_vblank_count() (v3)

2015-12-04T20:15:07Z

commit 4dfd6486 "drm: Use vblank timestamps to guesstimate how many vblanks were missed" introduced in Linux 4.4-rc1 makes the drm core more fragile to drivers which don't update hw vblank counters and vblank timestamps in sync with firing of the vblank irq and essentially at leading edge of vblank. This exposed a problem with radeon-kms/amdgpu-kms which do not satisfy above requirements: The vblank irq fires a few scanlines before start of vblank, but programmed pageflips complete at start of vblank and vblank timestamps update at start of vblank, whereas the hw vblank counter increments only later, at start of vsync. This leads to problems like off by one errors for vblank counter updates, vblank counters apparently going backwards or vblank timestamps apparently having time going backwards. The net result is stuttering of graphics in games, or little hangs, as well as total failure of timing sensitive applications. See bug #93147 for an example of the regression on Linux 4.4-rc: https://bugs.freedesktop.org/show_bug.cgi?id=93147 This patch tries to align all above events better from the viewpoint of the drm core / of external callers to fix the problem: 1. The apparent start of vblank is shifted a few scanlines earlier, so the vblank irq now always happens after start of this extended vblank interval and thereby drm_update_vblank_count() always samples the updated vblank count and timestamp of the new vblank interval. To achieve this, the reporting of scanout positions by radeon_get_crtc_scanoutpos() now operates as if the vblank starts radeon_crtc->lb_vblank_lead_lines before the real start of the hw vblank interval. This means that the vblank timestamps which are based on these scanout positions will now update at this earlier start of vblank. 2. The driver->get_vblank_counter() function will bump the returned vblank count as read from the hw by +1 if the query happens after the shifted earlier start of the vblank, but before the real hw increment at start of vsync, so the counter appears to increment at start of vblank in sync with the timestamp update. 3. Calls from vblank irq-context and regular non-irq calls are now treated identical, always simulating the shifted vblank start, to avoid inconsistent results for queries happening from vblank irq vs. happening from drm_vblank_enable() or vblank_disable_fn(). 4. The radeon_flip_work_func will delay mmio programming a pageflip until the start of the real vblank iff it happens to execute inside the shifted earlier start of the vblank, so pageflips now also appear to execute at start of the shifted vblank, in sync with vblank counter and timestamp updates. This to avoid some races between updates of vblank count and timestamps that are used for swap scheduling and pageflip execution which could cause pageflips to execute before the scheduled target vblank. The lb_vblank_lead_lines "fudge" value is calculated as the size of the display controllers line buffer in scanlines for the given video mode: Vblank irq's are triggered by the line buffer logic when the line buffer refill for a video frame ends, ie. when the line buffer source read position enters the hw vblank. This means that a vblank irq could fire at most as many scanlines before the current reported scanout position of the crtc timing generator as the number of scanlines the line buffer can maximally hold for a given video mode. This patch has been successfully tested on a RV730 card with DCE-3 display engine and on a evergreen card with DCE-4 display engine, in single-display and dual-display configuration, with different video modes. A similar patch is needed for amdgpu-kms to fix the same problem. Limitations: - Maybe replace the udelay() in the flip_work_func() by a suitable usleep_range() for a bit better efficiency? Will try that. - Line buffer sizes in pixels are hard-coded on < DCE-4 to a value i just guessed to be high enough to work ok, lacking info on the true sizes atm. Probably fixes: fdo#93147 Port of Mario's radeon fix to amdgpu. Signed-off-by: Alex Deucher (v1) Reviewed-by: Mario Kleiner (v2) Refine amdgpu_flip_work_func() for better efficiency. In amdgpu_flip_work_func, replace the busy waiting udelay(5) with event lock held by a more performance and energy efficient usleep_range() until at least predicted true start of hw vblank, with some slack for scheduler happiness. Release the event lock during waits to not delay other outputs in doing their stuff, as the waiting can last up to 200 usecs in some cases. Also small fix to code comment and formatting in that function. (v2) Signed-off-by: Mario Kleiner (v3) Fix crash in crtc disabled case