Subject : Hitting WARN_ON in hw_breakpoint code
Submitter : Paul Mackerras <firstname.lastname@example.org>
Date : 2010-06-23 12:57
Message-ID : 20100623125740.GA3368@brick.ozlabs.ibm.com
References : http://marc.info/?l=linux-kernel&m=127729789113432&w=2
This entry is being used for tracking a regression from 2.6.34. Please don't
close it until the problem is fixed in the mainline.
On Friday, July 09, 2010, Frederic Weisbecker wrote:
> On Thu, Jul 08, 2010 at 06:34:25PM -0700, Linus Torvalds wrote:
> > On Thu, Jul 8, 2010 at 4:33 PM, Rafael J. Wysocki <email@example.com> wrote:
> > > Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=16284
> > > Subject : Hitting WARN_ON in hw_breakpoint code
> > > Submitter : Paul Mackerras <firstname.lastname@example.org>
> > > Date : 2010-06-23 12:57 (16 days old)
> > This has "I have a fix, will post it very soon." in the thread from
> > Frederic, but I'm not seeing anything else. Frederic?
> Right. In fact it wasn't a regression. The per task breakpoint reservation
> design was broken from the beginning and this warning has revealed the
> problem. This only touched perf, and it did since perf support breakpoints.
> Fortunately ptrace wasn't concerned by this problem, even not by side effects
> of this.
> The fix is invasive as it's a rewrite of a (little) part of the breakpoint
> reservation. And since the symptom is only a warning and also breakpoints
> never released from the constraint table (just a counter, no memory leak),
> the fix is headed for 2.6.36.
> It is ready in tip:/perf/core:
> I think this ticket can be safely closed.