Skip to contentSkip to conversation

libva to play h264 on GMA4500HD

DISCUSSION Alexey Fisher 3 participants 4 messages Jul 12, 2010 – Jul 13, 2010

lore
Share & tools
lore
#0 lore raw

Hallo all,

i trying to test current libva, but have problem to play H264 video. If i'm correct this board ( intel DG45ID (GMA4500HD), Vendor: 0x8086,
Device: 0x2e22, Revision: 0x03 (A3)) should support it, mpeg2 videos are played correctly.

mplayer return this error:
libavformat file format detected.
[lavf] stream 0: video (h264), -vid 0
[lavf] stream 1: audio (aac), -aid 0, -alang eng
VIDEO: [H264] 976x544 24bpp 25.000 fps 2455.9 kbps (299.8 kbyte/s) Clip info:
major_brand: isom
minor_version: 1
compatible_brands: isom
libva: libva version 0.31.1
libva: va_getDriverName() returns 0
libva: Trying to open /usr/lib/dri/i965_drv_video.so
libva: va_openDriver() returns 0
==========================================================================
Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family [VD_FFMPEG] VA API accelerated codec.
Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264)
==========================================================================
==========================================================================
Opening audio decoder: [faad] AAC (MPEG2/4 Advanced Audio Coding)
AUDIO: 44100 Hz, 2 ch, s16le, 127.9 kbit/9.07% (ratio: 15991->176400) Selected audio codec: [faad] afm: faad (FAAD AAC (MPEG-2/MPEG-4 Audio))
==========================================================================
[AO OSS] audio_setup: Can't open audio device /dev/dsp: No such file or directory
AO: [alsa] 44100Hz 2ch s16le (2 bytes per sample)
Starting playback...
Unsupported PixelFormat 61
[VD_FFMPEG] Trying pixfmt=1.
Movie-Aspect is 1.79:1 - prescaling to correct movie aspect.
VO: [vaapi] 976x544 => 976x544 H.264 VA-API Acceleration [VD_FFMPEG] XVMC-accelerated MPEG-2.
mplayer: i965_avc_ildb.c:621: i965_avc_ildb_decode_init: Assertion `avc_ildb_context->urb.cs_start + avc_ildb_context->urb.num_cs_entries * avc_ildb_context->urb.size_cs_entry <= ((((&i965->intel)->device_id == 0x0042) || ((&i965->intel)->device_id == 0x0046)) ? 1024 :
(((&i965->intel)->device_id == 0x2E02 || (&i965->intel)->device_id == 0x2E12 || (&i965->intel)->device_id == 0x2E22 || (&i965->intel)->device_id == 0x2E32) || ((&i965->intel)->device_id == 0x2A42)) ? 384 : 256)' failed.

=================================================================

i use libva from git master, ( last commit e9c00fc43e "Fix vaGetImage() bounds checking code."), libdrm 2.4.21 from ubuntu ppa. And ubuntu kernel 2.6.35-7-generic.

#3 lore raw
Quoted 11 lines · Pedro Ribeiro

2010/7/12 Niccolò Belli <[email protected]>:

Quoted 8 lines

I fear there is still no support for gma45...

Darkbasic
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Support should come in Q3 according to the devs.

I just digged to the code. Now i a bit disappointed, at the time i choice this board, intel made difference between GMA4500 and GMA4500HD. In the code there is no difference. it is just one of the boards form kernels_gen4 with not really "hardware" acceleration:
i965_media_h264.c

        if (IS_IRONLAKE(i965->intel.device_id)) {
            h264_avc_kernels = h264_avc_kernels_gen5;
            avc_mc_kernel_offset = avc_mc_kernel_offset_gen5;
            intra_kernel_header = &intra_kernel_header_gen5;
            i965_h264_context->use_avc_hw_scoreboard = 1;
            i965_h264_context->use_hw_w128 = 1;
        } else {
            h264_avc_kernels = h264_avc_kernels_gen4;
            avc_mc_kernel_offset = avc_mc_kernel_offset_gen4;
            intra_kernel_header = &intra_kernel_header_gen4;
            i965_h264_context->use_avc_hw_scoreboard = 0;   <===
            i965_h264_context->use_hw_w128 = 0;             <===
        }

Any way, there is always some thing better... About the bug, i see gen4 is enabled to do h264. Devs put assert here to prevent some thing bad - thank you.

Here is what i get:
i965_avc_ildb.c:620
============================================

	assert(avc_ildb_context->urb.cs_start +
		avc_ildb_context->urb.num_cs_entries *
		avc_ildb_context->urb.size_cs_entry <=
		URB_SIZE((&i965->intel)));

===============================================

and here is why i get it:
avc_ildb_context->urb.cs_start = 640
avc_ildb_context->urb.num_cs_entries = 1
avc_ildb_context->urb.size_cs_entry = 1
URB_SIZE((&i965->intel)) = 384

URB_SIZE is definded here:
i965_defines.h
========================================
#define URB_SIZE(intel) (IS_IRONLAKE(intel->device_id) ? 1024 :

\
                                 IS_G4X(intel->device_id) ? 384 : 256)

===============================================

After it I tried to do some thing bad and set

avc_ildb_context->urb.size_vfe_entry = 380;

so i got GPU freeze. Now i'm happy :)

Hope this info help you.

Regards,
Alexey

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx