Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 70317 - Possible to include __VMALLOC_RESERVE workaround in nvidia-kernel-1.0.5336 ?
Summary: Possible to include __VMALLOC_RESERVE workaround in nvidia-kernel-1.0.5336 ?
Status: RESOLVED FIXED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: [OLD] Core system (show other bugs)
Hardware: x86 Linux
: High normal
Assignee: Gentoo X packagers
URL: http://www.linuxforum.com/forums/inde...
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2004-11-06 19:23 UTC by Lindsay Haisley
Modified: 2004-11-07 15:12 UTC (History)
0 users

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


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Lindsay Haisley 2004-11-06 19:23:50 UTC
It looks as if the 2.6.9-gentoo-r1 kernel doesn't play nice with nvidia-kernel 1.0.5336-r4.  The nvidia module wouldn't install in this kernel and gave the error:

nvidia: Unknown symbol __VMALLOC_RESERVE

A workable fix, albeit perhaps not the ideal one, is provided by Andreas Klauer on The Linux Forum.  

-----------------------------

Andreas Klauer  	
Posted: Oct 27 2004, 11:13 PM
Quote Post
	
bigbob wrote:

  QUOTE
  dmesg is: nvidia: module license 'NVIDIA' taints kernel.
  nvidia: Unknown symbol __VMALLOC_RESERVE

For x86 systems, add the following in arch/i386/mm/init.c in line 44:

EXPORT_SYMBOL(__VMALLOC_RESERVE);

Then recompile the kernel. Fixed the problem for me. It would probably be
better to patch the nvidia kernel glue code instead, but I can't do that.

-- 
* Andreas Klauer | ~'' L@@k at Nerd Boy!

-----------------

This fix worked for me.  I tried upgrading to (unstable) nvidia-kernel-1.0.6111-r2, which doesn't appear to have the symbol problem, however I get system lockups with it, and haven't been able to work around this.  I'm still using xfree rather than xorg-x11 which may account for this problem, but that's not the subject of this report.
Comment 1 Daniel Drake (RETIRED) gentoo-dev 2004-11-07 01:47:15 UTC
This isn't something we are prepared to put in the kernel, since its an nvidia issue. Best bet would be to see if the nvidia ebuild maintainers can provide a workaround...
Comment 2 Andrew Bevitt 2004-11-07 03:11:39 UTC
We can put the same patch into 5336 if necessary, but as todays commits made
1.0.6111-r3 stable and
1.0.6629 unstable and
masked all other versions its kinda almost a non-event, but as i said in the commits, if things go terribly wrong we could unmask...

Lindsay : Can you try out 6629 for us see how you go.
Comment 3 Lindsay Haisley 2004-11-07 09:44:57 UTC
I'll give it a shot and let you know.  1.0.6111 froze the keyboard and screen on startup and never even gave me the nvidia logo.  I note that the warning re. using 4K stack size was missing from 1.0.6111 so I tried kernels with both 4K and 8K stack sizes (the pre-logo lockup was the symptom I got w. 1.0.5336-r4 and a 4K stack).  Neither worked.
Comment 4 Lindsay Haisley 2004-11-07 12:30:39 UTC
OK, everything appears to be copesetic, so far.  I upgraded from xfree to stable xorg-x11 (since portage was complaining about header file locations) and installed nvidia-kernel-1.0.6629, which works fine.  I pulled the patch out of /usr/src/linux/arch/i386/mm/init.c, built and installed a new kernel, and everything came up OK.  I get neither a complaint from modprobe about __VMALLOC_RESERVE nor a lock-up when X starts.  

Mark this bug as fixed if you wish.  Thanks!
Comment 5 Andrew Bevitt 2004-11-07 15:12:14 UTC
Thanks :) its nice to see 6629 works