x86: Add cmpmemsi for -minline-all-stringops
We used to expand memcmp to "repz cmpsb" via cmpstrnsi. It was changed
by
commit
9b0f6f5e511ca512e4faeabc81d2fd3abad9b02f
Author: Nick Clifton <nickc@redhat.com>
Date: Fri Aug 12 16:26:11 2011 +0000
builtins.c (expand_builtin_memcmp): Do not use cmpstrnsi pattern.
* builtins.c (expand_builtin_memcmp): Do not use cmpstrnsi
pattern.
* doc/md.texi (cmpstrn): Note that the comparison stops if both
fetched bytes are zero.
(cmpstr): Likewise.
(cmpmem): Note that the comparison does not stop if both of the
fetched bytes are zero.
Duplicate the cmpstrn pattern for cmpmem. The only difference is that
the length argument of cmpmem is guaranteed to be less than or equal to
lengths of 2 memory areas. Since "repz cmpsb" can be much slower than
memcmp function implemented with vector instruction, see
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=43052
expand cmpmem to "repz cmpsb" only for -minline-all-stringops.
gcc/
PR target/95151
* config/i386/i386-expand.c (ix86_expand_cmpstrn_or_cmpmem): New
function.
* config/i386/i386-protos.h (ix86_expand_cmpstrn_or_cmpmem): New
prototype.
* config/i386/i386.md (cmpmemsi): New pattern.
gcc/testsuite/
PR target/95151
* gcc.target/i386/pr95151-1.c: New test.
* gcc.target/i386/pr95151-2.c: Likewise.
* gcc.target/i386/pr95151-3.c: Likewise.
* gcc.target/i386/pr95151-4.c: Likewise.