Re: [PATCH 3.15 33/37] Fix gcc-4.9.0 miscompilation of load_balance() in scheduler

[Alexei Starovoitov]

On Wed, Aug 6, 2014 at 1:14 AM, Jakub Jelinek <jakub@xxxxxxxxxx> wrote:
> On Tue, Aug 05, 2014 at 03:36:39PM -0700, Linus Torvalds wrote:
d>> I don't understand how you guys can be so cavalier about a compiler
>> bug that has already resulted in actual real problems. You bring up
>
> I have no problem with a -fno-var-tracking-assignments workaround for
> compilers that have the PR61801 wrong-code bug.  What I have problem with
> is with disabling it even for compilers that have that bug fixed.
> That is in essence disabling a useful feature just because it could have
> other bugs.  If my memory serves me well, PR61801 is the only wrong-code
> I remember caused by -fvar-tracking-assignments during the 5 years since
> it has been introduced into gcc.  Sure, there have been several
> -fcompare-debug bugs, where we generated slightly different code between
> -g and -g0, and as you mentioned we have one still pending (Vladimir is
> working on it right now), but that is mainly relevant to the case where

I think gcc guys are taking a wrong lesson out of this. kernel doesn't
care too much whether gcc produces the same binary with -g and -g0.
kernel developers also don't care about amount debug info for variables,
but they care about hard to find compiler bugs. In this case sched2
mishap around debug_insn was a symptom. The root cause is lack
of attention to -mno-red-zone. Kernel is not another user space program
where data/control flow analysis is all compiler need to make things right.
-mno-red-zone lesson exposes lack of 'interrupt' concept in compiler.
I think there has to be some infra put in place to make sure that it's not
just a scheduling barrier. Otherwise next bug will pop much sooner
than 5 years and it will not be related to debug info at all.
In this sense Steven's perl script to detect red-zone violations did more
to re-enable var-tracking than -fcompare-debug fixes.
--
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/