Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 113760 - xorg-x11-6.8.99+ can't find opengl with both hands and a mirror.
Summary: xorg-x11-6.8.99+ can't find opengl with both hands and a mirror.
Status: RESOLVED DUPLICATE of bug 110506
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: New packages (show other bugs)
Hardware: x86 Linux
: High major (vote)
Assignee: Gentoo X packagers
URL: http://pastebin.ca/31485
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-11-27 21:30 UTC by Chandra Barnett
Modified: 2005-11-29 11:47 UTC (History)
0 users

See Also:
Package list:
Runtime testing required: ---


Attachments
New xorg failure log (Xorg.0.log,12.17 KB, text/plain)
2005-11-28 01:49 UTC, Chandra Barnett
Details
xorg.conf which generated the prior failure (xorg.conf.tiny,1.46 KB, text/plain)
2005-11-28 01:50 UTC, Chandra Barnett
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Chandra Barnett 2005-11-27 21:30:06 UTC
My video card (Intel 915GM) is not supported by xorg-6.8.2, so I am trying to
get the unstable xorg builds to work. Both 6.8.99.15 and 7.0.0_RC2 are having
the exact same issue finding the OpenGL interface which they installed
themselves. I cannot load the glx module due to the unresolved symbol
"glTexSubImage3D", which should be provided by the OpenGL interface. eselect (or
opengl-update for 6.8.99.15) tell me I am using the correct interface, and all
the files that should be in /usr/lib/opengl/xorg-x11 seem to be there, with
appropriate symlinks in /usr/lib, but no go. I tried explicitly loading libGL,
and xorg informs me that it is not a valid module.

My xorg.conf is at http://pastebin.ca/31467 and the error log is at
http://pastebin.ca/31485 . In switching from 6.8.99.15 to 7.0.0_RC2, I installed
all of the OpenGL stuff clean, as in, moved everything opengl-related in
/usr/lib and /usr/lib/xorg/modules into hidden directories and emerged
--emptytree to make sure everything was rebuilt. The problem was still identical
in both cases.

I can start xorg ok with no acceleration and 4-bit colour by falling back to the
vesa driver. If I had wanted to do that I'd have stuck with xorg-6.8.2...

Reproducible: Always
Steps to Reproduce:
1. Install xorg-x11-6.8.99.15.ebuild or xorg-x11-7.0.0_RC2.ebuild.
2. Set up xorg.conf as above.
3. Run startx.

Actual Results:  
X failed to load glx, thus failed to load dri, thus failed to load i810 and
exited with no screens found. Logged as above.

Expected Results:  
Should have started X with the i810 driver.
Comment 1 Donnie Berkholz (RETIRED) gentoo-dev 2005-11-27 22:28:36 UTC
Relevant log chunks:

(II) Loading /usr/lib/xorg/modules/libextmod.so
dlopen: /usr/lib/xorg/modules/libextmod.so: undefined symbol: XTestJumpPointer

(II) LoadModule: "glx"
(II) Loading /usr/lib/xorg/modules/extensions/libglx.so
dlopen: /usr/lib/xorg/modules/extensions/libglx.so: undefined symbol:
glTexSubImage3D

(II) Loading /usr/lib/xorg/modules/extensions/libdri.so
dlopen: /usr/lib/xorg/modules/extensions/libdri.so: undefined symbol:
__glXgetActiveScreen

(II) Loading /usr/lib/xorg/modules/drivers/i810_drv.so
dlopen: /usr/lib/xorg/modules/drivers/i810_drv.so: undefined symbol: DRICloseScreen


Get rid of that Load "GL" line in your xorg.conf, same for "drm" and "vgahw".
Delete all ModulePath entries from xorg.conf (This is very important). Then try
again and please attach your new xorg.conf and X log rather than pastebin, to
ensure they're preserved.
Comment 2 Chandra Barnett 2005-11-28 01:49:21 UTC
Created attachment 73707 [details]
New xorg failure log

Here there were no ModulePath entries and I did not attempt to explicitly load
libGL, libdrm, or libvgahw.
Comment 3 Chandra Barnett 2005-11-28 01:50:15 UTC
Created attachment 73708 [details]
xorg.conf which generated the prior failure
Comment 4 Joshua Baergen (RETIRED) gentoo-dev 2005-11-28 08:28:13 UTC
I see you're running a hardened kernel.  Are you running a hardened system, and,
if so, did you build X with vanilla gcc?
Comment 5 Chandra Barnett 2005-11-28 23:22:00 UTC
(In reply to comment #4)
> I see you're running a hardened kernel.  Are you running a hardened system, and,
> if so, did you build X with vanilla gcc?

I do have USE=hardened set globally, but I compiled gcc with this flag as well.
There oughtn't be any USE flag mismatches at least in xorg's dependencies. I
emerged it --emptytree the last few times I reinstalled it...

I don't have any of the alternate security models from the hardened-sources
enabled, if that's what you're getting at.
Comment 6 Adam Jackson 2005-11-29 11:42:58 UTC
(In reply to comment #5)
> I do have USE=hardened set globally, but I compiled gcc with this flag as well.
> There oughtn't be any USE flag mismatches at least in xorg's dependencies. I
> emerged it --emptytree the last few times I reinstalled it...

The point is that Xorg simply can not be linked correctly when the linker does
'-z now', which the hardened compiler will do by default.

the xorg ebuilds really need to just inject -z nonow into LDFLAGS, unconditionally.
Comment 7 Joshua Baergen (RETIRED) gentoo-dev 2005-11-29 11:47:30 UTC
Thanks Adam, I tried to post the same thing at the same time :P

*** This bug has been marked as a duplicate of 110506 ***