diff options
| author | Richard Biener <rguenther@suse.de> | 2021-01-13 13:48:31 +0100 |
|---|---|---|
| committer | Richard Biener <rguenther@suse.de> | 2021-01-13 14:51:08 +0100 |
| commit | 285fa338b06b804e72997c4d876ecf08a9c083af (patch) | |
| tree | 97ed22c7c131348831676bde0bcb4d1dc9346891 /gcc/match.pd | |
| parent | a2d04f3d2c90155bea035d7527a49bc4a6a3397b (diff) | |
| download | gcc-285fa338b06b804e72997c4d876ecf08a9c083af.zip gcc-285fa338b06b804e72997c4d876ecf08a9c083af.tar.gz gcc-285fa338b06b804e72997c4d876ecf08a9c083af.tar.bz2 | |
tree-optimization/92645 - avoid harmful early BIT_FIELD_REF canonicalization
This avoids canonicalizing BIT_FIELD_REF <T1> (a, <sz>, 0) to
(T1)a on integer typed a. This confuses the vectorizer SLP matching.
With this delayed to after vector lowering the testcase in PR92645
from Skia is now finally optimized to reasonable assembly.
2021-01-13 Richard Biener <rguenther@suse.de>
PR tree-optimization/92645
* match.pd (BIT_FIELD_REF to conversion): Delay canonicalization
until after vector lowering.
* gcc.target/i386/pr92645-7.c: New testcase.
* gcc.dg/tree-ssa/ssa-fre-54.c: Adjust.
* gcc.dg/pr69047.c: Likewise.
Diffstat (limited to 'gcc/match.pd')
| -rw-r--r-- | gcc/match.pd | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/gcc/match.pd b/gcc/match.pd index c286a54..60c383d 100644 --- a/gcc/match.pd +++ b/gcc/match.pd @@ -6075,6 +6075,8 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT) /* Low-parts can be reduced to integral conversions. ??? The following doesn't work for PDP endian. */ || (BYTES_BIG_ENDIAN == WORDS_BIG_ENDIAN + /* But only do this after vectorization. */ + && canonicalize_math_after_vectorization_p () /* Don't even think about BITS_BIG_ENDIAN. */ && TYPE_PRECISION (TREE_TYPE (@0)) % BITS_PER_UNIT == 0 && TYPE_PRECISION (type) % BITS_PER_UNIT == 0 |