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: [PATCH] Fix the one entry mem{{,p}cpy,move,set} optimization aliasing issues (PR middle-end/29272, take 2)


On 10/9/06, Janis Johnson <janis187@us.ibm.com> wrote:
On Mon, Oct 09, 2006 at 10:01:30AM -0400, Jakub Jelinek wrote:
> Ok to commit this then?
> Let's keep discussing how to improve it after it gets committed,
> so that the trunk isn't broken for too long.
>
> 2006-10-09  Jakub Jelinek  <jakub@redhat.com>
>
>       PR middle-end/29272
>       * builtins.c (var_decl_component_p): New function.
>       (fold_builtin_memset, fold_builtin_memory_op): Restrict
>       single entry optimization to variables and components thereof.
>
>       * gcc.c-torture/execute/20060930-2.c: New test.

Since the mem{{,p}cpy,move,set} optimization went in I have occasionally
tried to come up with a minimized testcase for runtime failures in
FreePOOMA tests, and have tried patches that have been posted to fix for
various problems in that area.

This patch fixes FreePOOMA, which is to say that the tests that normally
pass now work on powerpc-linux when compiled with "-O2 -funroll-loops".
Yippee!

Yeah - I only know it's an interaction between std::list and src/Utilities/Pool* and some heavy inlining ;)

Richard.


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