This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: libsanitizer merge from upstream r208536


[plain text]

On Wed, May 14, 2014 at 2:42 AM, Andrew Pinski <pinskia@gmail.com> wrote:
> On Mon, May 12, 2014 at 4:20 AM, Konstantin Serebryany
> <konstantin.s.serebryany@gmail.com> wrote:
>> This is the first libsanitizer merge in 4.10 (the last merge was in
>> December 2013).
>>
>> Tested on Ubuntu 12.04 like this:
>> rm -rf */{*/,}libsanitizer && make -j 50
>> make -j 40 -C gcc check-g{cc,++}
>> RUNTESTFLAGS='--target_board=unix\{-m32,-m64\} asan.exp' && \
>> make -j 40 -C gcc check-g{cc,++}
>> RUNTESTFLAGS='--target_board=unix\{-m32,-m64\} tsan.exp' && \
>> make -j 40 -C gcc check
>> RUNTESTFLAGS='--target_board=unix\{-m32,-m64\} ubsan.exp' && \
>> echo PASS
>>
>> 5 months' worth of changes may break any platform we are not testing ourselves
>> (that includes Ubuntu 12.04, 13.10, 14.04, Mac 10.9, Windows 7, Android ARM),
>> please help us test this patch on your favorite platform.
>
> One theme I have been noticing in the libsanitizer code is that it has
> all of the knowledge of glibc when it comes to syscalls but makes some
> bad assumptions dealing with some structures.  For an example on MIPS
> n64: stat and stat64 are different layout between user and kernel.
> Don't ask, it is a historical mistake.
>
> Also it really forces target maintainers to maintain both glibc and
> libsanitizer code.  ILP32 on aarch64 has some similar issues as x32
> and MIPS n32.
> Even Linus is pushing new 32bit targets to go the route of 64bit time
> which is means libsanitizer code becomes even more complex.
This is indeed a PITA.
It is caused by our need to avoid even greater PITA of including
system headers into the sanitizer files
(we allow system #includes only in a handful of files).
>
> Is there a way to reduce this maintenance overhead?  Can we please
> split the target specific structures out to a header file which can
> then be included.  This will reduce the overall messy #if's all of the
> code and help to add new targets.

The problem is far from trivial.
I will be more than happy to receive help in form of patches.

--kcc

>
> Thanks,
> Andrew Pinski
>
>
>>
>> Expected ChangeLog entries:
>> =============== gcc/testsuite/ChangeLog
>> 2014-05-XX  Kostya Serebryany  <kcc@google.com>
>>
>>         * c-c++-common/tsan/mutexset1.c: Update the test to match
>>         upstream r208536.
>>         * g++.dg/asan/symbolize-callback-1.C: Delete the deprecated test.
>>
>> =============== libsanitizer/ChangeLog
>> 2014-05-XX  Kostya Serebryany  <kcc@google.com>
>>
>>         * All source files: Merge from upstream r208536.
>>         * asan/Makefile.am (asan_files): Added new files.
>>         * asan/Makefile.in: Regenerate.
>>         * tsan/Makefile.am (tsan_files): Added new files.
>>         * tsan/Makefile.in: Regenerate.
>>         * sanitizer_common/Makefile.am (sanitizer_common_files): Added
>> new files.
>>         * sanitizer_common/Makefile.in: Regenerate.
>>
>> --kcc


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]