Update valgrind to fix one problem with 64 bit code, and a regression running...

Update valgrind to fix one problem with 64 bit code, and a regression running wine.
BUG=valgrind_210481, valgrind_205541
TEST=valgrind net_unittests --gtest_filter=ProxyResolverV8Test.LoadLog ; valgrind wine's unit tests and see no HEAP_CreateSubHeap valgrind warnings

[dank, 2009-10-20 23:14:36]

Timur, I'd like to update to a new revision
of valgrind, too, and ditch the '-redzone' suffix.
Can you provide revision numbers and a
tsan patch to match, if needed?

[Timur Iskhodzhanov, 2009-10-21 09:33:28]

The VEX patch doesn't help if you run net_unittests with
--gtest_filter=ProxyResolverV8*.

Under Memcheck:
[==========] Running 15 tests from 2 test cases.
[----------] Global test environment set-up.
[----------] 14 tests from ProxyResolverV8Test
[ RUN      ] ProxyResolverV8Test.Direct
[       OK ] ProxyResolverV8Test.Direct (7008 ms)
[ RUN      ] ProxyResolverV8Test.ReturnEmptyString
/usr/local/google/users/timurrrr/chromium/src/sconsbuild/Debug/net_unittests(_ZN10StackTraceC1Ev+0x1d)[0xba1eff]
/usr/local/google/users/timurrrr/chromium/src/sconsbuild/Debug/net_unittests[0x9781b3]
/lib/libc.so.6[0x8ab0100]
[0x10b66acc]

Under ThreadSanitizer:
[==========] Running 15 tests from 2 test cases.
[----------] Global test environment set-up.
[----------] 14 tests from ProxyResolverV8Test
[ RUN      ] ProxyResolverV8Test.Direct
[       OK ] ProxyResolverV8Test.Direct (9145 ms)
[ RUN      ] ProxyResolverV8Test.ReturnEmptyString
[       OK ] ProxyResolverV8Test.ReturnEmptyString (1995 ms)
[ RUN      ] ProxyResolverV8Test.Basic
[       OK ] ProxyResolverV8Test.Basic (5216 ms)
[ RUN      ] ProxyResolverV8Test.BadReturnType
#
# Fatal error in v8/src/handles-inl.h, line 48
# CHECK(location_ != __null) failed
#

Can this be a bug in V8, not Valgrind?

http://codereview.chromium.org/306020/diff/1/3
File tools/valgrind/build-valgrind-for-chromium.sh (right):

http://codereview.chromium.org/306020/diff/1/3#newcode19
Line 19: TSAN_SVN_REV=1129
1178 should be fine

http://codereview.chromium.org/306020/diff/1/4
File tools/valgrind/vbug205541.patch (right):

http://codereview.chromium.org/306020/diff/1/4#newcode4
Line 4: --- memcheck/mc_main.c	(revision 10877)
why not 10880?

http://codereview.chromium.org/306020/diff/1/4#newcode11
Line 11: +      MC_(make_mem_defined)(a, len);
In this case the condition above doesn't make sense.
However, I accept this as a workaround

http://codereview.chromium.org/306020/diff/1/2
File tools/valgrind/vbug210481.patch (right):

http://codereview.chromium.org/306020/diff/1/2#newcode4
Line 4: --- VEX/priv/guest_amd64_toIR.c	(Revision 1924)
why not 1913?

[Timur Iskhodzhanov, 2009-10-21 10:10:35]

It seems like V8 problem is ALSO related to the abscence of Valgrind annotations
for amd64 (only ia32 was annotated her http://codereview.chromium.org/13612 for
some reason).
Valgrind is running OK with --smc-check=all WITH your VEX patch.
With Valgrind annotations added to x64, your VEX patch but without
--smc-check=all it fails.

Investigating.

[Timur Iskhodzhanov, 2009-10-21 10:23:22]

Sorry, I was applying the annotations the wrong way for amd64. With properly
applied annotations AND your VEX patch everything works fine.

[not_the_right_dank, 2009-10-21 13:45:37]

OK.   Do we need a patch somewhere to get the annotations right?

On Wed, Oct 21, 2009 at 3:23 AM,  <timurrrr@chromium.org> wrote:
>
> Sorry, I was applying the annotations the wrong way for amd64. With properly
> applied annotations AND your VEX patch everything works fine.
>
> http://codereview.chromium.org/306020
>

[not_the_right_dank, 2009-10-21 13:47:37]

Never mind, I read your link.

Can you provide new revision numbers for valgrind and vex with a new matchi=
ng
tsan patch (if needed)?  I'd like to update as part of this patch.

On Wed, Oct 21, 2009 at 6:45 AM, Dan Kegel <dank@kegel.com> wrote:
> OK. =A0 Do we need a patch somewhere to get the annotations right?
>
> On Wed, Oct 21, 2009 at 3:23 AM, =A0<timurrrr@chromium.org> wrote:
>>
>> Sorry, I was applying the annotations the wrong way for amd64. With prop=
erly
>> applied annotations AND your VEX patch everything works fine.
>>
>> http://codereview.chromium.org/306020
>>
>

[not_the_right_dank, 2009-10-21 16:34:50]

Aw, I went and committed as is.  We can update to a new valgrind later.

On Wed, Oct 21, 2009 at 6:46 AM, Dan Kegel <dank@kegel.com> wrote:
> Never mind, I read your link.
>
> Can you provide new revision numbers for valgrind and vex with a new matc=
hing
> tsan patch (if needed)? =A0I'd like to update as part of this patch.
>
> On Wed, Oct 21, 2009 at 6:45 AM, Dan Kegel <dank@kegel.com> wrote:
>> OK. =A0 Do we need a patch somewhere to get the annotations right?
>>
>> On Wed, Oct 21, 2009 at 3:23 AM, =A0<timurrrr@chromium.org> wrote:
>>>
>>> Sorry, I was applying the annotations the wrong way for amd64. With pro=
perly
>>> applied annotations AND your VEX patch everything works fine.
>>>
>>> http://codereview.chromium.org/306020
>>>
>>
>

[stuartmorgan, 2016-02-09 16:11:37]

Description was changed from

==========
Update valgrind to fix one problem with 64 bit code, and a regression running
wine.
BUG=valgrind_210481, valgrind_205541
TEST=valgrind net_unittests --gtest_filter=ProxyResolverV8Test.LoadLog ;
valgrind wine's unit tests and see no HEAP_CreateSubHeap valgrind warnings
==========

to

==========
Update valgrind to fix one problem with 64 bit code, and a regression running
wine.
BUG=valgrind_210481, valgrind_205541
TEST=valgrind net_unittests --gtest_filter=ProxyResolverV8Test.LoadLog ;
valgrind wine's unit tests and see no HEAP_CreateSubHeap valgrind warnings
==========