Re: [Problem] kernel hangs at boot (bisected 892d208bcf)

[Catalin Marinas <catalin.marinas@arm.com>]

On Fri, Jan 20, 2012 at 02:20:24AM +0000, Josh Boyer wrote:
> On Thu, Jan 19, 2012 at 3:52 PM, Dirk Gouders
> <gouders@et.bocholt.fh-gelsenkirchen.de> wrote:
> >> I did some more testing and it turns out that I have to revert a single
> >> line of commit b66930052abf2 to make the kernel boot:
> >>
> >> diff --git a/mm/kmemleak.c b/mm/kmemleak.c
> >> index c833add..d8bbb68 100644
> >> --- a/mm/kmemleak.c
> >> +++ b/mm/kmemleak.c
> >> @@ -1710,6 +1710,7 @@ static void kmemleak_disable(void)
> >>                 return;
> >>
> >>         /* stop any memory operation tracing */
> >> +       atomic_set(&kmemleak_early_log, 0);
> >>         atomic_set(&kmemleak_enabled, 0);
> >>
> >>         /* check whether it is too early for a kernel thread */
> >>
> >> I started that kernel with initcall_debug and attach the dmesg output.
> >
> > Hi Catlin,
> >
> > I want to note that in my config CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF is
> > set which causes kmemleak_disable() to be called in kmemleak_init().
> 
> I also had CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y in my .config.  Disabling that
> option allowed the machine to boot just fine.  Dirk seems to be onto something
> here.

It definitely looks like a kmemleak problem with
CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y. I managed to reproduce it on ARM as
well.

I'll have a look and get back shortly.

Thanks.

-- 
Catalin