[mythtv-users] Via CLE266 (via-slice) HW acceleration problems(segfault)

[Ivan A. Beveridge]

On Wed, Jun 30, 2004 at 11:51:21PM +0100, Ivan A. Beveridge wrote:
> On Wed, Jun 30, 2004 at 09:58:28PM +0200, Kenneth Aafløy wrote:

> > Me too! The messages above is printed very early in the loading of 
> > via_v4l_drv, so I'm suspecting that you might have a pciid not present in the 
> > driver. 
> 
> Ha - this looks promising (in a warped kind of way ;)
> 
> http://forums.viaarena.com/messageview.cfm?catid=28&threadid=55911&STARTPAGE=2
> 
> It mentions this thread:
> http://forums.viaarena.com/messageview.cfm?catid=28&threadid=56180&FTVAR_MSGDBTABLE=

> It looks like I need to either fiddle around (as mentioned in the 2nd
> URL), or jump to 2.6.7 (and no further) .... I'd rather not go back to 
> 2.4 kernel.

I've tried doing as mentioned in the URL (build the drm from a cvs
checkout). Dmesg shows this on unloading & loading:
======================================
video:/usr/local/src/drm/linux# dmesg
[drm] Module unloaded
[drm] Initialized via 1.3.0 20040411 on minor 0: VIA Technologies, Inc.
VT8623 [Apollo CLE266] integrated CastleRock graphics
via_v4l: Device name:(VIA Technologies, Inc. VT8623 [Apollo CLE266]
integrated CastleRock graphics), slot_name:0000:01:00.0 (0x1106:0x3122)
Revision:17
via_v4l: registered video device (0) corresponds to /dev/video0
via_v4l: Device name:(VIA Technologies, Inc. VT8623 [Apollo CLE266]
integrated CastleRock graphics), slot_name:0000:01:00.0 (0x1106:0x3122)
Revision:17
via_v4l: registered video device (1) corresponds to /dev/video1
via_v4l: Device name:(VIA Technologies, Inc. VT8623 [Apollo CLE266]
integrated CastleRock graphics), slot_name:0000:01:00.0 (0x1106:0x3122)
Revision:17
via_v4l: registered video device (2) corresponds to /dev/video2
agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0.
agpgart: Putting AGP V2 device at 0000:00:00.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:01:00.0 into 4x mode
via_v4l: Revision:17
======================================

v4l-info /dev/video2 looks good. The X log is now showing V4L is
working:
======================================
<snip>
VIAScreenInit : V4L Enabled : fd = 5
(--) VIA(0): mapping framebuffer @ 0xd8000000 with size 0x2000000
(==) VIA(0): Write-combining range (0xd8000000,0x2000000)
(--) VIA(0): Frame buffer start: 0x402c9000, free start: 0x195000 end:
0x2000000
(--) VIA(0): mapping MMIO @ 0xdc000000 with size 0x9000
(--) VIA(0): mapping BitBlt MMIO @ 0xdc200000 with size 0x10000
<snip>
======================================


However, turning on CLE266 hardware compression in mythfrontend causes a
segfault (although at a different place). This is the output of
mythfrontend (running "mythfrontent -v all"):

======================================
2004-07-01 17:04:20 mythfrontend version: 0.15.20040525-1 www.mythtv.org
2004-07-01 17:04:20 Enabled verbose msgs : important general playback
mythtv: could not connect to socket
mythtv: No such file or directory
lirc_init failed for mythtv, see preceding messages
2004-07-01 17:04:21 Registering Internal as a media playback plugin.
2004-07-01 17:04:21 Registering MythDVD DVD Media Handler as a media handler
2004-07-01 17:04:21 Registering MythDVD VCD Media Handler as a media handler
2004-07-01 17:06:53 Connecting to backend server: 127.0.0.1:6543 (try 1 of 5)
2004-07-01 17:06:53 Using protocol version 8
2004-07-01 17:06:53 Using protocol version 8
2004-07-01 17:08:20 Using protocol version 8
2004-07-01 17:08:21 Output filters for this recording are: ''
2004-07-01 17:08:22 Interlaced: -1  video_height: 576  fps: 25
2004-07-01 17:08:22 Estimated bitrate = 192000
2004-07-01 17:08:22 Interlaced: -1  video_height: 576  fps: -1
Input #0, mpeg, from 'rbuf://127.0.0.1:6543/media/cache/mythtv/tv/ringbuf1.nuv':
  Stream #0.0: Video: mpegvideo_via, 704x576, 25.00 fps
  Stream #0.1: Audio: mp2, 48000 Hz, stereo, 192 kb/s
2004-07-01 17:08:22 Partial position map found
2004-07-01 17:08:22 Opening OSS audio device '/dev/dsp'.
2004-07-01 17:08:22 Audio fragment size: 4096
2004-07-01 17:08:22 Over/underscan. V: 0, H: 0, XOff: 0, YOff: 0
Segmentation fault
======================================

I'll try to run it again and get further details. What is off use - a
backtrace? Anything else?

Cheers


Ivan
-- 
Ivan Beveridge	<ivan at dreamtime.org>