This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug target/80479] [7/8 Regression] strcmp() produces valgrind errors on ppc64le
- From: "acsawdey at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Wed, 26 Apr 2017 13:04:02 +0000
- Subject: [Bug target/80479] [7/8 Regression] strcmp() produces valgrind errors on ppc64le
- Auto-submitted: auto-generated
- References: <bug-80479-4@http.gcc.gnu.org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80479
acsawdey at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |munroesj at gcc dot gnu.org,
| |segher at gcc dot gnu.org,
| |wschmidt at gcc dot gnu.org
--- Comment #13 from acsawdey at gcc dot gnu.org ---
The tests that I have done show that the inline code is 2-3x faster than glibc
(depending on which glibc you are testing). The two big factors are that we can
generate better code because we frequently know the alignment of args in gcc,
and getting rid of the function call overhead. So no, there isn't any magic
from this that can be transplanted into the library function.
The addi 31,31,1 is part of the test case not the strcmp expansion. I missed it
when I was taking out the extraneous ops. The subf. could be changed to xor. as
you say.