Messages in this thread | | | Subject | Re: [Bug #11035] System hangs on 2.6.26-rc8 | From | Nigel Cunningham <> | Date | Fri, 18 Jul 2008 17:17:32 +1000 |
| |
Hi.
On Fri, 2008-07-18 at 09:11 +0200, Ingo Molnar wrote: > * Roman Mindalev <lists@r000n.net> wrote: > > > I done bisection. > > Result below: > > > > 8f46924600e30b140445f5b84abe9b80d2fff5fb is first bad commit > > commit 8f46924600e30b140445f5b84abe9b80d2fff5fb > > Author: Ingo Molnar <mingo@elte.hu> > > Date: Wed Jan 30 13:34:09 2008 +0100 > > > > x86: enable CONFIG_DEBUG_PAGEALLOC more widely > > > > make CONFIG_DEBUG_PAGEALLOC universally available. > > > > CONFIG_HIBERNATION and CONFIG_HUGETLBFS was disabling it, for no > > particular reason. > > as far as i can see you see a lockup under certain circumstances, right? > > this debug option catches use-after-free and other types of invalid > memory accesses. When it catches a bug the kernel most likely crashes > and produces a backlog. Because you are in graphical mode that is not > visible. > > This would possibly be debuggable if you set up netconsole logging to > another system on a local LAN - see > Documentation/networking/netconsole.txt. > > Vegard - would it be possible to make DEBUG_PAGEALLOC faults single-shot > and non-fatal, just like kmemcheck does it? That way people would see a > nice kernel message instead of an immediate crash. That means we'd have > to find a reliable filter for DEBUG_PAGEALLOC-provoked pagefaults though > ...
Not that it matters now, but that original commit message was wrong - CONFIG_HIBERNATION used to be incompatible with CONFIG_DEBUG_PAGEALLOC because [u]swsusp didn't (until very recently) handle the fact that DEBUG_SLAB unmaps empty pages on x86.
Regards,
Nigel
| |