This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Fix for PR/39548
This starts to getting weird -- I compared the code (for the bad
function) generated from your test case and the one reproducible for
me, the code look almost identical to the second copyrename pass ---
after which the code looks quite different -- this indicates some
possible non-determinstic behavior in copyrename. I will take a look
at it when I find the time.
David
On Thu, Mar 26, 2009 at 10:18 AM, Eric Botcazou <ebotcazou@adacore.com> wrote:
>> The problem does not reproduce with this version. Anyway I will take
>> a look a little more.
>
> g++ -S -fno-exceptions -fno-tree-vrp -O2 -fprofile-generate pr39548.C
> pr39548.C: In member function 'void foo_14::hmmmmh_29(blah_46)':
> pr39548.C:874: error: definition in block 29 does not dominate use in block 30
> for SSA_NAME: D.57875_368 in statement:
> prephitmp.2487_329 = PHI <pretmp.2486_444(4), D.57875_368(30), D.47407_44(31)>
> PHI argument
> D.57875_368
> for PHI node
> prephitmp.2487_329 = PHI <pretmp.2486_444(4), D.57875_368(30), D.47407_44(31)>
> pr39548.C:874: internal compiler error: verify_ssa failed
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <http://gcc.gnu.org/bugs.html> for instructions.
>
> eric@hermes:~/build/gcc/native> g++ -v
> Using built-in specs.
> Target: x86_64-suse-linux
> Configured with: /home/eric/svn/gcc/configure
> x86_64-suse-linux --prefix=/home/eric/install/gcc --enable-languages=c,c++ --enable-__cxa_atexit --disable-nls --disable-libmudflap
> Thread model: posix
> gcc version 4.4.0 20090326 (experimental) [trunk revision 145085] (GCC)
>
> Make sure not to sure the patched compiler...
>
> --
> Eric Botcazou
>