pesintta / vdr-plugin-vaapidevice

VDR VAAPIDevice Plug-in
9 stars 12 forks source link

"Sharpening Support" reported but using it, will crash GPU #39

Open Quantomax opened 6 years ago

Quantomax commented 6 years ago

Hi,

I've a problem with sharpening support on Intel J3160 (CherryView 2): If I enable Sharpening, e.g. switch from 0 -> 1, I'll get something like a frozen not-deinterlaced flickering picture (seems like three frames joined), crashing GPU some moments later. Need to detach/attach to resume.

vainfo: error: XDG_RUNTIME_DIR not set in the environment. libva info: VA-API version 1.0.0 libva info: va_getDriverName() returns 0 libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so libva info: Found init function __vaDriverInit_1_0 libva info: va_openDriver() returns 0 vainfo: VA-API version: 1.0 (libva 2.0.0) vainfo: Driver version: Intel i965 driver for Intel(R) CherryView - 2.0.0 vainfo: Supported profile and entrypoints VAProfileMPEG2Simple : VAEntrypointVLD VAProfileMPEG2Simple : VAEntrypointEncSlice VAProfileMPEG2Main : VAEntrypointVLD VAProfileMPEG2Main : VAEntrypointEncSlice VAProfileH264ConstrainedBaseline: VAEntrypointVLD VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice VAProfileH264Main : VAEntrypointVLD VAProfileH264Main : VAEntrypointEncSlice VAProfileH264High : VAEntrypointVLD VAProfileH264High : VAEntrypointEncSlice VAProfileH264MultiviewHigh : VAEntrypointVLD VAProfileH264MultiviewHigh : VAEntrypointEncSlice VAProfileH264StereoHigh : VAEntrypointVLD VAProfileH264StereoHigh : VAEntrypointEncSlice VAProfileVC1Simple : VAEntrypointVLD VAProfileVC1Main : VAEntrypointVLD VAProfileVC1Advanced : VAEntrypointVLD VAProfileNone : VAEntrypointVideoProc VAProfileJPEGBaseline : VAEntrypointVLD VAProfileJPEGBaseline : VAEntrypointEncPicture VAProfileVP8Version0_3 : VAEntrypointVLD VAProfileHEVCMain : VAEntrypointVLD

Logs: [21959] VDR version 2.3.8 started [21959] loading plugin: /usr/lib/vdr/plugins/libvdr-vaapidevice.so.2.3.8 video/vaapi: libva 1.0 (Intel i965 driver for Intel(R) CherryView - 2.0.0) initialized video/vaapi: background-color is unsupported video/vaapi: supports video processing video/vaapi: supports unscaled osd [vaapidevice] ready .. codec: YUV 420 supported Using entrypoint for vpp: 10 video/vaapi: can't destroy postproc context! can't destroy config! video/vaapi: noise reduction supported video/vaapi: 0,00 - 1,00 ++ 0,03 = 0,50 Enabling denoise filter (pos = 0) video/vaapi: deinterlacing supported video/vaapi: bob deinterlace supported motion adaptive deinterlace supported motion compensated deinterlace supported Enabling Deint (pos = 1) video/vaapi: sharpening supported video/vaapi: 0,00 - 1,00 ++ 0,03 = 0,50 Enabling sharpening filter (pos = 0) video/vaapi: enabling color balance filters Supported color balance filter count: 4 Hue (-180,00 - 180,00 ++ 1,00 = 0,00) (pos = 2) Saturation (0,00 - 10,00 ++ 0,10 = 1,00) (pos = 2) Brightness (-100,00 - 100,00 ++ 1,00 = 0,00) (pos = 2) Contrast (0,00 - 10,00 ++ 0,10 = 1,00) (pos = 2) video/vaapi: skin tone enhancement supported video/vaapi: 0,00 - 4,00 ++ 1,00 = 0,00 Enabling skin tone filter (pos = 3) Allocating 1 forward reference surfaces for postprocessing Allocating 0 backward reference surfaces for postprocessing video: display buffer empty, duping frame (46/0) 0 Surface 0 query status = 0x6: invalid VASurfaceID Surface 0 query status = 0x6: invalid VASurfaceID ...

While switching on sharpening via menu video/vaapi: gpu hung -4294949ms 78208

kernel: [53772.319113] [drm] GPU HANG: ecode 8:0:0x85ddfffb, in vdr [21959], reason: Hang on rcs0, action: reset kernel: [53772.319118] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. kernel: [53772.319120] [drm] Please file a new bug report on bugs.freedesktop.org against DRI -> DRM/Intel kernel: [53772.319121] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. kernel: [53772.319122] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. kernel: [53772.319123] [drm] GPU crash dump saved to /sys/class/drm/card0/error kernel: [53772.319136] i915 0000:00:02.0: Resetting rcs0 after gpu hang kernel: [53776.454505] asynchronous wait on fence i915:vdr[21959]/0:38eff timed out kernel: [53780.294596] i915 0000:00:02.0: Resetting rcs0 after gpu hang kernel: [53788.294494] i915 0000:00:02.0: Resetting rcs0 after gpu hang kernel: [53796.294490] i915 0000:00:02.0: Resetting rcs0 after gpu hang ` Kernel is a 4.14.10.

rofafor commented 6 years ago

IMO, this looks like a bug in Intel drivers - you should make an issue for them instead.

pesintta commented 6 years ago

You can also run this simple sharpening test tool I've written some years ago when I reported similar bug to Intel devs: https://bugs.freedesktop.org/attachment.cgi?id=110316

It may make the crash easier to reproduce or not, YMMV.