Re: [PATCH 5/5] x86,mm: make pagefault killable

From: KOSAKI Motohiro
Date: Wed Mar 23 2011 - 04:10:17 EST


> On Tue, 22 Mar 2011 20:09:29 +0900 (JST)
> KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx> wrote:
>
> > When oom killer occured, almost processes are getting stuck following
> > two points.
> >
> > 1) __alloc_pages_nodemask
> > 2) __lock_page_or_retry
> >
> > 1) is not much problematic because TIF_MEMDIE lead to make allocation
> > failure and get out from page allocator. 2) is more problematic. When
> > OOM situation, Zones typically don't have page cache at all and Memory
> > starvation might lead to reduce IO performance largely. When fork bomb
> > occur, TIF_MEMDIE task don't die quickly mean fork bomb may create
> > new process quickly rather than oom-killer kill it. Then, the system
> > may become livelock.
> >
> > This patch makes pagefault interruptible by SIGKILL.
> >
> > Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx>
> > ---
> > arch/x86/mm/fault.c | 9 +++++++++
> > include/linux/mm.h | 1 +
> > mm/filemap.c | 22 +++++++++++++++++-----
> > 3 files changed, 27 insertions(+), 5 deletions(-)
> >
> > diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
> > index 20e3f87..797c7d0 100644
> > --- a/arch/x86/mm/fault.c
> > +++ b/arch/x86/mm/fault.c
> > @@ -1035,6 +1035,7 @@ do_page_fault(struct pt_regs *regs, unsigned long error_code)
> > if (user_mode_vm(regs)) {
> > local_irq_enable();
> > error_code |= PF_USER;
> > + flags |= FAULT_FLAG_KILLABLE;
> > } else {
> > if (regs->flags & X86_EFLAGS_IF)
> > local_irq_enable();
> > @@ -1138,6 +1139,14 @@ good_area:
> > }
> >
> > /*
> > + * Pagefault was interrupted by SIGKILL. We have no reason to
> > + * continue pagefault.
> > + */
> > + if ((flags & FAULT_FLAG_KILLABLE) && (fault & VM_FAULT_RETRY) &&
> > + fatal_signal_pending(current))
> > + return;
> > +
>
> Hmm? up_read(&mm->mmap_sem) ?

When __lock_page_or_retry() return 0, It call up_read(mmap_sem) in this
function.

I agree this is strange (or ugly). but I don't want change this spec in
this time.



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/