gcc.dg/tree-ssa/sra-12.c is skipped on a bunch of targets, including AArch64,
because the default max-scalarization-size depends on MOVE_RATIO, and on those
targets thus ends up being too small for SRA to optimize the testcase. Recently
I noticed that the test has been failing for some time on ARM too. This patch
fixes the test on ARM, AArch64, avr, and sh, and by extension I believe also on
nds32, although I haven't managed to build a nds32 compiler to check.
There is an argument that instead we should skip the test on ARM too; or rather,
since at least ARM and AArch64 would like the test to pass, we should xfail it
on those platforms until we have time to experiment with the threshold/param for
SRA. I hope to do some more investigation on that front as part of (or followup
to) PR/63679.
Is this OK for trunk?
Cheers,
Alan
gcc/testsuite/ChangeLog:
* gcc.dg/tree-ssa/sra-12.c: Enable test on all targets; add --param
sra-max-scalarization-size-Ospeed.
---
gcc/testsuite/gcc.dg/tree-ssa/sra-12.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c
b/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c
index 427f7af..139c7d1 100644
--- a/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c
+++ b/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c
@@ -1,7 +1,6 @@
/* Verify that SRA total scalarization will not be confused by padding. */
-/* Test skipped for targets with small (often default) MOVE_RATIO. */
/* { dg-do compile } */
-/* { dg-options "-O1 -fdump-tree-release_ssa" } */
+/* { dg-options "-O1 --param sra-max-scalarization-size-Ospeed=16
-fdump-tree-release_ssa" } */
struct S
{
@@ -21,4 +20,4 @@ int foo (struct S *p)
*p = l;
}
-/* { dg-final { scan-tree-dump-times "l;" 0 "release_ssa" { target { !
"aarch64*-*-* avr*-*-* nds32*-*-* sh*-*-*" } } } } */
+/* { dg-final { scan-tree-dump-times "l;" 0 "release_ssa" } } */
--
1.9.1
