Re: [PATCH 0/2] ia64: possible module unwind table optimisation

From: Phil Carmody
Date: Wed Sep 15 2010 - 18:23:32 EST


On 15/09/10 19:28 +0200, ext Tony Luck wrote:
> On Wed, Sep 15, 2010 at 5:48 AM, Phil Carmody
> <ext-phil.2.carmody@xxxxxxxxx> wrote:
> >>   for (table = unw.tables; table; table = table->next)
> >>
> >> before we find the right table to look?
> >
> > Yes, that's the one. Sorry I wasn't more explicit.
>
> My usual workload (building new Linux kernels - what else :-) doesn't
> seem to generate
> any unwind requests. I reached the loop just 58 times while booting,
> and only did 23
> steps past the first item [didn't track a histogram, so I don't know
> whether that was a
> single 23 step lookup, or 23 one-step ones]. Building the kernel
> (make -j32) didn't add
> to either count.
>
> Workload ideas?

Turn on kmemleak, or anything else which repeatedly runs up the stack just
for the fun of gathering backtraces. Kmemleak's what caused us to notice
the issue in our ARM-based environment.

Phil
--
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/