Re: Locking problem: fb_info->lock and mm->mmap_sem

Andrew Morton napisa³(a):

On Sun, 10 May 2009 20:34:55 +0200 Krzysztof Helt >krzysztof.h1@poczta.fm>
wrote:

quoted

I can prepare the patch which contains changes
equivalent to revert of these fixes.

This version does not show any problems. 

That sounds like an OK plan to me.  Doing this via separate manual
reversion patches would make things easier to maintain, assuming that
this is a 2.6.31 activity.

I suppose it should go into the 2.6.30 as it fixes a current problem.

It might be that the old bkl-based code was simply buggy, because
lock_kernel() gets secretly unlocked by so many operations.  It doesn't
matter, really.

I suppose that the BKL worked because it was recursive.

Restoring the lock_kernel()s would make people sad.  Do you have a plan
to de-BKL the fbdev code via other means?  That wasn't clear from your
email.

Not at the moment. I would like to fix some issues pending (e.g.
nvidia problem with suspend).

I think it was very close to de-bKL with my mm_lock patch
as the last issue I found can only happen if there is more
than one framebuffer (probe and release at the same time).

Regards,
Krzysztof

----------------------------------------------------------------------
Do wygrania 100 podwójnych zaproszeñ!
Sprawdz >> http://link.interia.pl/f2160 





------------------------------------------------------------------------------
The NEW KODAK i700 Series Scanners deliver under ANY circumstances! Your
production scanning environment may not be a perfect world - but thanks to
Kodak, there's a perfect scanner to get the job done! With the NEW KODAK i700
Series Scanner you'll get full speed at 300 dpi even with all image 
processing features enabled. http://p.sf.net/sfu/kodak-com