Re: [Valgrind-users] valgrind error on fedora 19 AMD64

[Jerry B. <je...@ge...>]

Hi Mark,
Thanks for the note. Here's the version I've been using:

valgrind-3.8.1-15.fc19.x86_64

I wasn't able to find a more recent fedora package.

We run on servers at rackspace. There is an older generation of cpu's and
valgrind works find on them. It is the newer (about a year-old) generation
on which valgrind doesn't work. This was a big surprise, since I think they
are stock processors. Might there be something in the VM/hypervisor layer
which is not translating correctly?

I did just go the svn route. The results look similar.

 valgrind]$ ll /usr/bin/valgrind
-rwxr-xr-x. 1 root root 54427 Oct 16 17:44 /usr/bin/valgrind

valgrind]$ valgrind ls -l
==21888== Memcheck, a memory error detector
==21888== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
==21888== Using Valgrind-3.9.0.SVN and LibVEX; rerun with -h for copyright
info
==21888== Command: ls -l
==21888==

vex: priv/main_main.c:338 (LibVEX_Translate): Assertion
`are_valid_hwcaps(VexArchAMD64, vta->archinfo_host.hwcaps)' failed.
vex storage: T total 0 bytes allocated
vex storage: P total 0 bytes allocated

valgrind: the 'impossible' happened:
   LibVEX called failure_exit().
==21888==    at 0x3804FBAC: report_and_quit (m_libcassert.c:260)
==21888==    by 0x3804FDB1: vgPlain_core_panic_at (m_libcassert.c:350)
==21888==    by 0x3804FDDA: vgPlain_core_panic (m_libcassert.c:360)
==21888==    by 0x38067E32: failure_exit (m_translate.c:731)
==21888==    by 0x380FA1F8: vex_assert_fail (main_util.c:219)
==21888==    by 0x380F90AF: LibVEX_Translate (main_main.c:338)
==21888==    by 0x3806A24E: vgPlain_translate (m_translate.c:1602)
==21888==    by 0x3809B596: vgPlain_scheduler (scheduler.c:1004)
==21888==    by 0x380AA6FC: run_a_thread_NORETURN (syswrap-linux.c:103)

sched status:
  running_tid=1

Thread 1: status = VgTs_Runnable
==21888==    at 0x315C601420: ??? (in /usr/lib64/ld-2.17.so)
==21888==    by 0x1: ???
==21888==    by 0xFFF0005CA: ???
==21888==    by 0xFFF0005CD: ???


Note: see also the FAQ in the source distribution.
It contains workarounds to several common problems.
In particular, if Valgrind aborted or crashed after
identifying problems in your program, there's a good chance
that fixing those problems will prevent Valgrind aborting or
crashing, especially if it happened in m_mallocfree.c.

If that doesn't help, please report this bug to: www.valgrind.org

In the bug report, send all the above text, the valgrind
version, and what OS and version you are using.  Thanks.

Thanks.
Jerry


On Wed, Oct 16, 2013 at 6:55 AM, Mark Wielaard <mj...@re...> wrote:

> On Tue, 2013-10-15 at 14:48 -0700, Jerry Blakley wrote:
> > I'm running valgrind on fedora 19. I checked libvex code which
> > executes the hwcaps code, but can't find anything in particular.
>
>
> > vex: priv/main_main.c:319 (LibVEX_Translate): Assertion
> > `are_valid_hwcaps(VexArchAMD64, vta->archinfo_host.hwcaps)' failed.
>
> This will happen when valgrind/VEX doesn't understand the (combination
> of) cpuid capabilities of your CPU.
>
> Specifically show_hwcaps_amd64 () in VEX/priv/main_main.c should return
> a valid string describing the configuration (not NULL).
>
> The version in fedora valgrind 3.8.1 only handles fixed combinations of
> flags, the version in valgrind SVN is more flexible.
>
> > flags : fpu de tsc msr pae cx8 cmov pat clflush mmx fxsr sse sse2 ht
> > syscall nx mmxext fxsr_opt lm rep_good nopl pni pclmulqdq ssse3 fma
> > cx16 sse4_1 sse4_2 popcnt aes f16c hypervisor lahf_lm cmp_legacy
> > extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch xop fma4 tce
> > tbm perfctr_core perfctr_nb arat cpb hw_pstate
>
> If I am reading that right, you don't have AVX, RDTSCP, BMI or AVX. You
> do have SSE3 (pni), XC16 and LZCNT (abm)
>
> Which should give you:
>
>       case VEX_HWCAPS_AMD64_SSE3 | VEX_HWCAPS_AMD64_CX16
>            | VEX_HWCAPS_AMD64_LZCNT:
>          return "amd64-sse3-cx16-lzcnt";
>
> So I am probably missing some detail/flag.
>
> Which fedora valgrind package version is this exactly?
> You can tell by running "rpm -q valgrind".
> Have you tried valgrind build from SVN?
>
> Thanks,
>
> Mark
>
>