[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#251307: xserver-xfree86: A problem with the new ATI driver: XVideo apps crash X server (on ATI Range IIc (Mach64) card)

tag 251307 + moreinfo upstream
retitle 251307 xserver-xfree86: [ati/atimisc] XVideo apps crash X server; complaints of unresolved symbol xf86XVCopyYUV12ToPacked on 3D Rage IIC AGP rev 122

On Fri, May 28, 2004 at 12:50:28AM +0300, Modestas Vainius wrote:
> Package: xserver-xfree86
> Version: 4.3.0.dfsg.1-2
> Severity: important
> Tags: sid
> Hello,
> First of all, thank you for including the new ATI driver. I have used
> the driver from http://gatos.sf.net for a year or so to get Xv support
> for my video card. Finally, the xserver-xfree86 is supposed to support
> XVideo on my pretty old video card out of the box.

Yes, I was hoping this would make some people happy.

> However, there is one problem currently. It seems that any XVideo app
> (I have tested mplayer and kaffeine (xine)) brings the X server down.
> /var/log/XFree86.0.log.old provides good hints though. I have attached this
> file it doesn't get included in the report automatically.
> /var/log/XFree86.0.log.old has the log of the X server instance which was 
> crashed by mplayer. /var/log/XFree86.0.log (automatically included)
> is from the currently running X server instance. I think, the cause of
> my problem is the unresolved symbol:
> Symbol xf86XVCopyYUV12ToPacked from module /usr/X11R6/lib/modules/drivers/atimisc_drv.o is unresolved!
> which yields the following error on crash:
> 		This should not happen!
> 	        An unresolved function was called!

This is quite strange.  According to objdump:

% objdump --dynamic-syms /usr/X11R6/bin/XFree86 | grep xf86XVCopyYUV12ToPacked
10070394 g    DF .text  000001ac  Base        xf86XVCopyYUV12ToPacked

What does this command say for you?

I've looked over the sources and this is not a symbol that has to be
imported; it, along with the rest of XVideo appears to be in the part of
the XFree86 X server that is resident in memory all the time (else it
wouldn't appear in objdump).

For that matter:

% objdump --dynamic-syms /usr/X11R6/bin/XFree86 | grep xf86XV
1006c808 g    DF .text  00000080  Base        xf86XVRegisterGenericAdaptorDriver
10070394 g    DF .text  000001ac  Base        xf86XVCopyYUV12ToPacked
1006c9c0 g    DF .text  0000002c  Base        xf86XVQueryOffscreenImages
1006c9ec g    DF .text  00000024  Base        xf86XVAllocateVideoAdaptorRec
1006fd98 g    DF .text  000005fc  Base        xf86XVClipVideoHelper
1006ca30 g    DF .text  00000248  Base        xf86XVScreenInit
10070540 g    DF .text  0000009c  Base        xf86XVCopyPacked
1006ca10 g    DF .text  00000020  Base        xf86XVFreeVideoAdaptorRec
1006c888 g    DF .text  00000108  Base        xf86XVListGenericAdaptors
1006fbdc g    DF .text  000001bc  Base        xf86XVFillKeyHelper
1006c990 g    DF .text  00000030  Base        xf86XVRegisterOffscreenImages

> Feel free to ask for more information. You may send patches to me.
> I will test them shortly!

Try using xserver-xfree86-dbg instead.  What happens then?

If the symbol were truly unresolved, the package build would fail when the
gigantic, static XFree86 debug binary were linked.  But that doesn't happen.

G. Branden Robinson                |       If atheism is a religion, then
Debian GNU/Linux                   |       health is a disease.
branden@debian.org                 |       -- Clark Adams
http://people.debian.org/~branden/ |

Attachment: signature.asc
Description: Digital signature

Reply to: