|
From: Vishal <vis...@gm...> - 2013-08-02 05:22:53
|
> Hi Vishal,
>
> How about the other experiment where you build Valgrind with the default
optimization
> level turned off? i.e. start over (do a: make clean) and then do a: make
CFLAGS="-O0 -g"
>
> Also, I tried the following out:
>
> ----
> anmol:/proj/ppc/DT/labhome/anmol/valgrind-3.8.1> git diff
coregrind/m_machine.c
> diff --git a/coregrind/m_machine.c b/coregrind/m_machine.c
> index 6a77057..122674a 100644
> --- a/coregrind/m_machine.c
> +++ b/coregrind/m_machine.c
> <at> <at> -1465,6 +1465,8 <at> <at> void
VG_(machine_ppc64_set_clszB)( Int szB )
> {
> vg_assert(hwcaps_done);
>
> + VG_(printf)("\nmachine_ppc64_set_clszB() (post-
entry)/(vai.ppc_cache_line_szB: %d, szB:
> %d)\n", vai.ppc_cache_line_szB, szB);
> +
> /* Either the value must not have been set yet (zero) or we can
> tolerate it being set to the same value multiple times, as the
> stack scanning logic in m_main is a bit stupid. */
> <at> <at> -1473,6 +1475,7 <at> <at> void
VG_(machine_ppc64_set_clszB)( Int szB )
>
> vg_assert(szB == 32 || szB == 64 || szB == 128);
> vai.ppc_cache_line_szB = szB;
> + VG_(printf)("\nmachine_ppc64_set_clszB() (pre-
exit)/(vai.ppc_cache_line_szB: %d, szB: %d)\n",
> vai.ppc_cache_line_szB, szB);
> }
> anmol:/proj/ppc/DT/labhome/anmol/valgrind-3.8.1> ./bin/valgrind --leak-
check=full ~/hello
>
> machine_ppc64_set_clszB() (post-entry)/(vai.ppc_cache_line_szB: 0, szB:
128)
>
> machine_ppc64_set_clszB() (pre-exit)/(vai.ppc_cache_line_szB: 128, szB:
128)
>
> machine_ppc64_set_clszB() (post-entry)/(vai.ppc_cache_line_szB: 128, szB:
128)
>
> machine_ppc64_set_clszB() (pre-exit)/(vai.ppc_cache_line_szB: 128, szB:
128)
>
> /* Usual Valgrind output... */
> anmol:/proj/ppc/DT/labhome/anmol/valgrind-3.8.1>
> ----
>
> If you could do the similar experiment for your Valgrind, we'll learn
something about what is going on.
>
> Regards,
> Anmol.
Hi Anmol,
Here's the dump you asked for.
machine_ppc64_set_clszB() (post-entry)/(vai.ppc_cache_line_szB: 0, szB: 128)
machine_ppc64_set_clszB() (pre-exit)/(vai.ppc_cache_line_szB: 128, szB: 128)
machine_ppc64_set_clszB() (post-entry)/(vai.ppc_cache_line_szB: 128, szB:
32)
valgrind: m_machine.c:1382 (vgPlain_machine_ppc32_set_clszB): Assertion
'vai.ppc_cache_line_szB == 0 || vai.ppc_cache_line_szB == szB' failed.
==12374== at 0x38035274: ??? (in
/workspace/sw/vishald/etc/lib/valgrind/memcheck-ppc32-linux)
sched status:
running_tid=0
I took a look at https://bugs.kde.org/show_bug.cgi?id=306587 as you
suggested and it seems to be the same issue. The patch there is the same I
tried in my installation. But just want to make sure it is correct.
Warm Regards,
Vishal
|