Bug#854194: valgrind: segfaults on MIPS Cavium Octeon boards

James Cowgill <[email protected]> Tue, 7 Feb 2017 13:03:38 +0000

Newsgroups	gmane.linux.debian.ports.mips
Message-ID	<[email protected]>

Control: reopen -1
Control: retitle -1 valgrind: segfaults on MIPS Cavium Octeon boards
Control: severity -1 important

Hi,

On Tue, 7 Feb 2017 17:20:44 +0800 YunQiang Su <[email protected]> wrote:
> On Sun, 05 Feb 2017 00:39:49 +0200 Adrian Bunk <[email protected]> wrote:
> > Package: valgrind
> > Version: 1:3.12.0~svn20160714-1+b1
> > Severity: serious
> >
> 
> It seems that 1:3.12.0-1.1 doesn't have this problem.
> I also test it on Loongson 3A.
> 
> root@thor:/# valgrind /usr/bin/hello
> ==2085== Memcheck, a memory error detector
> ==2085== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
> ==2085== Using Valgrind-3.12.0 and LibVEX; rerun with -h for copyright info
> ==2085== Command: /usr/bin/hello
> ==2085==
> 
> VEX: Unsupported baseline
>      Found: Loongson-baseline
> Cannot continue. Good-bye

Well this just means that valgrind refuses to run on Loongson. The code
which triggers this bug hasn't yet run at this point.

However, if I hack /proc/cpuinfo so that it pretends to be an Octeon
machine, then valgrind does work correctly on Loongson 3A machines (at
least ls works). I can also get valgrind to work correctly on the CI20.

It seems this bug is Octeon specific. I'm downgrading the bug on that
basis since it doesn't affect mipsel in general. It just so happens that
the majority of the buildds and the 2 mips porterboxes are Octeons.

Thanks,
James

signature.asc (application/pgp-signature, 819 B)