2009-12-10 08:51:06

by Ingo Molnar

[permalink] [raw]
Subject: upstream regression: slob hang


FYI, the attached config hangs in the SLOB code, on 32-bit x86 laptop
(dual core). No log available from that box.

Ingo


Attachments:
(No filename) (124.00 B)
config (72.41 kB)
Download all attachments

2009-12-10 16:22:12

by Matt Mackall

[permalink] [raw]
Subject: Re: upstream regression: slob hang

On Thu, 2009-12-10 at 09:49 +0100, Ingo Molnar wrote:
> FYI, the attached config hangs in the SLOB code, on 32-bit x86 laptop
> (dual core). No log available from that box.

Is this one of your randconfigs? I take it switching from SLOB to SLUB
fixes it?

--
http://selenic.com : development and support for Mercurial and Linux

2009-12-10 16:24:54

by Ingo Molnar

[permalink] [raw]
Subject: Re: upstream regression: slob hang


* Matt Mackall <[email protected]> wrote:

> On Thu, 2009-12-10 at 09:49 +0100, Ingo Molnar wrote:
>
> > FYI, the attached config hangs in the SLOB code, on 32-bit x86
> > laptop (dual core). No log available from that box.
>
> Is this one of your randconfigs? [...]

Yeah, a bootable one.

> [...] I take it switching from SLOB to SLUB fixes it?

havent tried it. If you cannot reproduce it with the config i sent i can
try to dig deeper. That box is not particularly easy to debug.

Ingo

2009-12-14 18:27:09

by Pekka Enberg

[permalink] [raw]
Subject: Re: upstream regression: slob hang

Hi Ingo,

Ingo Molnar wrote:
> FYI, the attached config hangs in the SLOB code, on 32-bit x86 laptop
> (dual core). No log available from that box.

Boots fine on my 32-bit x86 laptop.

Pekka

2009-12-14 18:36:07

by Matt Mackall

[permalink] [raw]
Subject: Re: upstream regression: slob hang

On Mon, 2009-12-14 at 20:26 +0200, Pekka Enberg wrote:
> Hi Ingo,
>
> Ingo Molnar wrote:
> > FYI, the attached config hangs in the SLOB code, on 32-bit x86 laptop
> > (dual core). No log available from that box.
>
> Boots fine on my 32-bit x86 laptop.

Thanks, Pekka. I've been too busy to experiment with this. Ingo,
anything interesting about your laptop? How much memory do you have?

--
http://selenic.com : development and support for Mercurial and Linux

2009-12-15 06:45:22

by Pekka Enberg

[permalink] [raw]
Subject: Re: upstream regression: slob hang

Ingo Molnar wrote:
> > > FYI, the attached config hangs in the SLOB code, on 32-bit x86 laptop
> > > (dual core). No log available from that box.

On Mon, 2009-12-14 at 20:26 +0200, Pekka Enberg wrote:
> > Boots fine on my 32-bit x86 laptop.

On Mon, 2009-12-14 at 12:36 -0600, Matt Mackall wrote:
> Thanks, Pekka. I've been too busy to experiment with this. Ingo,
> anything interesting about your laptop? How much memory do you have?

Side note: SLOB hasn't really changed for a while so I suspect Ingo is
hitting some once in a blue moon type of old bug.

2009-12-15 07:42:52

by Ingo Molnar

[permalink] [raw]
Subject: Re: upstream regression: slob hang


* Pekka Enberg <[email protected]> wrote:

> Ingo Molnar wrote:
> > > > FYI, the attached config hangs in the SLOB code, on 32-bit x86 laptop
> > > > (dual core). No log available from that box.
>
> On Mon, 2009-12-14 at 20:26 +0200, Pekka Enberg wrote:
> > > Boots fine on my 32-bit x86 laptop.
>
> On Mon, 2009-12-14 at 12:36 -0600, Matt Mackall wrote:
> > Thanks, Pekka. I've been too busy to experiment with this. Ingo,
> > anything interesting about your laptop? How much memory do you have?
>
> Side note: SLOB hasn't really changed for a while so I suspect Ingo is
> hitting some once in a blue moon type of old bug.

That's possible - although as of late i'm generally finding new regressions,
the old once-in-a-blue-moon races i've triggered already. (the roughly half a
million random bootups i've done in the last 2 years in -tip testing went a
long way in rooting them out)

I'll try to figure this one out in a few days hopefully - the laptop is not
available for testing right now for other reasons.

Ingo