OpenE2K/gcc
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

tree-optimization/92645 - avoid harmful early BIT_FIELD_REF canonicalization

Browse Source 
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.
This commit is contained in:
Richard Biener 2021-01-13 13:48:31 +01:00
parent a2d04f3d2c
commit 285fa338b0
4 changed files with 31 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
gcc/match.pd
View File

@ -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

4
gcc/testsuite/gcc.dg/pr69047.c
View File

@ -1,5 +1,5 @@
/* { dg-do compile } */
/* { dg-options "-O -fdump-tree-cddce1" } */
/* { dg-options "-O -fdump-tree-forwprop4" } */
__UINT8_TYPE__
f(__UINT16_TYPE__ b)
@ -15,4 +15,4 @@ f(__UINT16_TYPE__ b)
return a;
}
/* { dg-final { scan-tree-dump "_\[0-9\]+ = \\(\[^)\]+\\) b" "cddce1" } } */
/* { dg-final { scan-tree-dump "_\[0-9\]+ = \\(\[^)\]+\\) b" "forwprop4" } } */

6
gcc/testsuite/gcc.dg/tree-ssa/ssa-fre-54.c
View File

@ -1,6 +1,6 @@
/* { dg-do run } */
/* { dg-require-effective-target int32plus } */
/* { dg-options "-O -fdump-tree-fre1 -fdump-tree-dse1" } */
/* { dg-options "-O -fdump-tree-forwprop4 -fdump-tree-dse1" } */
extern void abort (void);
@ -51,6 +51,6 @@ int main()
return 0;
}
/* { dg-final { scan-tree-dump "\\(char\\) i_" "fre1" } } */
/* { dg-final { scan-tree-dump "\\(short int\\) i_" "fre1" } } */
/* { dg-final { scan-tree-dump "\\(char\\) i_" "forwprop4" } } */
/* { dg-final { scan-tree-dump "\\(short int\\) i_" "forwprop4" } } */
/* { dg-final { scan-tree-dump-not "u.i =" "dse1" } } */

24
gcc/testsuite/gcc.target/i386/pr92645-7.c Normal file
View File

@ -0,0 +1,24 @@
/* { dg-do compile } */
/* { dg-require-effective-target lp64 } */
/* { dg-options "-O3 -msse2" } */
typedef long v2di __attribute__((vector_size(16)));
typedef int v4si __attribute__((vector_size(16)));
void bar (v4si *p, __int128_t *q)
{
union { __int128_t a; v4si b; } u;
u.a = *q;
(*p)[0] = u.b[0];
(*p)[1] = u.b[2];
(*p)[2] = u.b[1];
(*p)[3] = u.b[3];
}
/* The function should end up with sth like
[v]pshufd $216, (%esi), %xmm0
[v]movdqa %xmm0, (%edi)
ret
recognized by SLP vectorization involving an existing "vector". */
/* { dg-final { scan-assembler-not "punpck" } } */
/* { dg-final { scan-assembler-times "pshufd" 1 } } */