http://oprofile.sourceforge.net/oprofile-2.5.html
as usual. This release has two significant changes :
o We got the task release stuff horribly wrong for CLONE_DETACHED tasks
as used by nptl, leading to instant oopsen. I think I have fixed this.
One drawback is that the do_exit()-related code is now an oprofile blind
spot, but I can't see a better solution
o Use _raw forms for the spinlock we sleep whilst holding to avoid
CONFIG_PREEMPT complaining. I have lightly tested preempt on SMP with
the new patch, and it seems to work
regards
john
--
"That's just kitten-eating wrong."
- Richard Henderson
On Sun, Oct 13, 2002 at 08:38:38PM +0100, John Levon wrote:
> http://oprofile.sourceforge.net/oprofile-2.5.html
sourceforge seems to be down, it's also available at
http://movementarian.org/oprofile-2.5-all.diff.bz2
regards
john
--
"That's just kitten-eating wrong."
- Richard Henderson