glesx.so init causes sig 11

[blaz]

Just swapped in an new HD2600 XT (AGP version) and tried installing the latest driver using: ati-driver-installer-8-01-x86.x86_64.run
on Debian's 2.6.22-3-686 kernel in "testing".

Install log looks good. And, after reboot, X loads up nicely until the very end when glesx.so causes a signal 11 during initialization:
...
9: /usr/lib/xorg/modules//glesx.so [0xb581aa1c]
10: /usr/lib/xorg/modules//glesx.so(esutInit+0x74) [0xb5819294]
11: /usr/lib/xorg/modules//glesx.so [0xb5813f6d]
12: /usr/lib/xorg/modules//glesx.so(GlesxExtensionInit+0x94) [0xb5814234]
13: /usr/bin/X11/X(InitExtensions+0x85) [0x80f6885]
14: /usr/bin/X11/X(main+0x2bf) [0x8076c6f]
15: /lib/i686/cmov/libc.so.6(__libc_start_main+0xe0) [0xb7d24450]
16: /usr/bin/X11/X(FontFileCompleteXLFD+0x1e9) [0x80761b1]

xorg.conf is minimal, deleted all my old stuff and used "aticonfig --initial".

Searching around here for other problems with this card and glesx.so, I tried turning off a couple of options:
Option "Textured2D" "off"
Option "TexturedXrender" "off"
but they didn't help.

Is this worth a bug report? Does ATI pay any attention to them?

[Swoopy]

I'm amazed you got this far at all.
I have the same chipset AGP card and all I get is a hang when trying to use the fglrx version you mention (or any other for that matter).
I use x86_64 Ubuntu/7.10 on a dual core AMD4800+ system with a VIA K8T800 chipset-based motherboard.

What are the specs of your CPU / motherboard / Linux distro?

-edit- P.S. I have 4 GB of RAM (the max my board will support) and suspect my driver hang might be due to a chipset or MTRR memory mapping issue.
I tried all AGP aperture settings that my mainboard BIOS will allow.

[givemesugarr]

I'm amazed you got this far at all.
I have the same chipset AGP card and all I get is a hang when trying to use the fglrx version you mention (or any other for that matter).
I use x86_64 Ubuntu/7.10 on a dual core AMD4800+ system with a VIA K8T800 chipset-based motherboard.

What are the specs of your CPU / motherboard / Linux distro?

-edit- P.S. I have 4 GB of RAM (the max my board will support) and suspect my driver hang might be due to a chipset or MTRR memory mapping issue.
I tried all AGP aperture settings that my mainboard BIOS will allow.

if you think that it's a mtrr problem then add:
Option "mtrr" "off"
to your device section in xorg.conf.

[blaz]

I'm amazed you got this far at all.
I have the same chipset AGP card and all I get is a hang when trying to use the fglrx version you mention (or any other for that matter).
I use x86_64 Ubuntu/7.10 on a dual core AMD4800+ system with a VIA K8T800 chipset-based motherboard.

What are the specs of your CPU / motherboard / Linux distro?

-edit- P.S. I have 4 GB of RAM (the max my board will support) and suspect my driver hang might be due to a chipset or MTRR memory mapping issue.
I tried all AGP aperture settings that my mainboard BIOS will allow.

Debian "testing" with Debian's 2.6.22-686 kernel or my own 2.6.23.14 kernel. Same CPU as you but with an nvidia NForce3 chipset. But agpgart uses the amd64_agp module, the cpu based agp that interfaces with chipset based agp. When I tried using a chipset only based module (nvidia_agp) with agpgart, fglrx wouldn't hang: there would just be a message about no "internal" agp support for 2.6 kernels and there would be no 3-D. Never tried any bios settings. Here's agp messages in log for V2.6.23.14:
kernel: Linux agpgart interface v0.102
kernel: agpgart: Detected AGP bridge 0
kernel: agpgart: Setting up Nforce3 AGP.
kernel: agpgart: AGP aperture is 64M @ 0xf8000000

But now I've got this new glesx.so problem.:rolleyes:

[blaz]

Looks like this bug has already bit someone:
http://ati.cchtml.com/show_bug.cgi?id=989
I tried the 256 MB agp aperture and no longer get this segfault. But it looks like I now some other problems. Increasing the aperture size doesn't really fix things, just moves the problem around. Time to look at radeonhd I guess.