nvidia_driver does not work if compiled with gcc 4.7.0 (Bug #376)


Added by Florian Franzmann over 2 years ago. Updated over 2 years ago.


Status:New Start date:03/28/2012
Priority:High Due date:
Assignee:Sukneet Basuta % Done:

0%

Category:-
Target version:-
Grimoire:Test

Description

nvidia_driver version is 295.33, starting X fails like this:

root@narya:gcc [0]> startx
hostname: Unknown host
xauth: file /root/.serverauth.18074 does not exist

X.Org X Server 1.8.1
Release Date: 2010-05-11
X Protocol Version 11, Revision 0
Build Operating System: Linux 3.3.0 i686
Current Operating System: Linux narya 3.3.0 #2 SMP PREEMPT Mon Mar 26 12:46:08 CEST 2012 i686
Kernel command line: auto BOOT_IMAGE=3.3.0 ro root=907 libata.atapi_enabled=1
Build Date: 28 March 2012 09:08:29PM

Current version of pixman: 0.20.2
Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
Markers: (--) probed, () from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/var/log/Xorg.0.log", Time: Wed Mar 28 21:12:37 2012
(==) Using config file: "/etc/X11/xorg.conf"
(==) Using system config directory "/usr/share/X11/xorg.conf.d"
(EE) Failed to load module "mouse" (module does not exist, 0)
(EE) Failed to load module "kbd" (module does not exist, 0)
(EE) NVIDIA: Failed to initialize default colormap
(EE) NVIDIA:
Aborting **

Fatal server error:
AddScreen/ScreenInit failed for driver 0

Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
Please also check the log file at "/var/log/Xorg.0.log" for additional information.

xinit: giving up
xinit: unable to connect to X server: Connection refused
xinit: server error


Related issues

blocks Grimoire - Bug #422: [devel-gcc-4.7] Master bug New 08/03/2012
blocked by Grimoire - Bug #423: [devel-xorg-modular] Master bug New 03/17/2013

Associated revisions

Revision 514a2688
Added by Florian Franzmann over 2 years ago

Revert "Revert "Revert "Revert "gcc: => 4.7.0""""

This reverts commit b60d994e946c6460098488236d296984fe129efa.

xorg 1.12.1 is unaffected by bug #376

History

Updated by Ladislav Hagara over 2 years ago

No problem here (ia32):

$ gaze version gcc linux nvidia_driver xorg-server pixman
Grimoire Section Spell Grimoire Version Installed Version
-------- ------- ----- ---------------- -----------------
test gnu gcc 4.6.3 4.7.0
test kernels linux 3.3.1 3.3.1
z-rejected z-kernels nvidia_driver 295.33 295.33
test xorg-xserver xorg-server 1.8.1 1.12.0
test xorg-lib pixman 0.20.2 0.24.4

Updated by Florian Franzmann over 2 years ago

I have linux 3.3.1, nvidia_driver 295.33, xorg-server 1.8.1 and pixmand 0.20.2. Did you compile xorg-server with gcc 4.7.0 or 4.6.3?

Updated by David Haley over 2 years ago

Florian Franzmann wrote:

(EE) NVIDIA: Failed to initialize default colormap
(EE) NVIDIA: * Aborting *

Fatal server error:
AddScreen/ScreenInit failed for driver 0

Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
Please also check the log file at "/var/log/Xorg.0.log" for additional information.

xinit: giving up
xinit: unable to connect to X server: Connection refused
xinit: server error

I am actually experiencing the same thing. I can state that all spells that are listed below have been compiled with gcc 4.7.0. Also below are the current versions of the spells that seem to have been discussed above:

:~$ gaze version gcc linux nvidia_driver xorg-server pixman
Grimoire Section Spell Grimoire Version Installed Version
-------- ------- ----- ---------------- -----------------
test gnu gcc 4.6.3 4.7.0
test kernels linux 3.3.2 3.3.2
z-rejected z-kernels nvidia_driver 295.40 295.40
test xorg-xserver xorg-server 1.8.1 1.8.1
test xorg-lib pixman 0.22.2 0.22.2

If any more informaiton is needed or I need to test something please let me know.

Updated by Florian Franzmann over 2 years ago

The revert of the update to gcc 4.7 has been reverted in 8e2aee773f40bf6f41c62e40fab274e3cb9787a9. Why? I just checked and the problem persists. 8e2aee773f40bf6f41c62e40fab274e3cb9787a9 should be reverted in test.
xorg-server 1.12.1 doesn't seem to have this problem, so maybe we should update gcc to 4.7 in devel-xorg-modular? Having gcc 4.7 in test is unacceptable at the moment.

  • Priority changed from High to Urgent

Updated by Florian Franzmann over 2 years ago

  • Assignee set to Sukneet Basuta

Updated by Sukneet Basuta over 2 years ago

Sorry, I'm stupid. I didn't realize I recommitted gcc 4.7. I did it to cast it, but then forgot about it.
I really need to be more careful when I commit things.

Feel free to revert it again, or I'll do it when I get back home.

I'll have a look into this though.

Updated by Sukneet Basuta over 2 years ago

Re-Reverted gcc 4.7 with commit b60d994e946c6460098488236d296984fe129efa

I have to ask though, why is gcc 4.7 unacceptable in test? This bug withstanding, what else is an issue?
Despite being largely ineffective, isn't the general policy to create a new branch major software version changes?

Updated by Florian Franzmann over 2 years ago

It is unacceptable until it has been thoroughly tested and all serious bugs have been fixed. At the very least gcc 4.7 should be able to do a sorcery rebuild on a system with lots of spells installed.

  • Priority changed from Urgent to High

Updated by Sukneet Basuta over 2 years ago

Fair enough, but where are supposed to test it? No grimoire branch has gcc 4.7. Granted, I could just create one.

I understand that many of you expect test to be mostly stable, but very few people seemingly test branches, so they are largely ineffective, and the name "test" implies its for testing. I can totally understand Ladislav's frustration.

Perhaps what we need is a something like a rolling-release system for people that expect test to be mostly stable. Or I believe there was a devel grimoire before I was around, perhaps we can bring that back. I'm not sure why you guys got rid of it though. I'm sure there was a good reason.

Sukneet Basuta wrote:

Despite being largely ineffective, isn't the general policy to create a new branch for major software version changes?

Added a for in there.

Updated by Thomas Orgis over 2 years ago

The devel grimoire vanished because it was an extra level of nuisance to get things done. Long-lived branches face the same fate.

Back then, the grimoire process was largely stalled; there was a long phase without stable releases. The effort to get things going again involved dropping devel grimoire and starting the regular schedule of snapping stable-rc from test and letting that become the next stable grimoire in a timely manner. We settled on a list of core spells that have to be tested before going stable, with the intention of extending that list in future (Xorg was not part of this list yet, AFAIR).

The migration to git with its promise of inflationary branching and merging brought the idea of task-specific devel grimoires in form of branches. What they need is some attention/pressure on folks to get them tested and merged ...

Also available in: Atom PDF