Message ID | 20110707183309.GB3584@virgil.arch.suse.de |
---|---|
State | New |
Headers | show |
On Jul 7, 2011, at 11:33 AM, Martin Jambor wrote: > on avr targets, the aggregate that we test to be totally scalarized > away is deemed to be to big for it - this is determined by the value > of MOVE_RATIO which is very low unless redefined by the target. This > is analogous situation to what we had with > testsuite/gcc.dg/tree-ssa/pr42585.C a year ago, the solution there was > also to skip the test. > > And that's exactly what the patch below does, verified by running > make -k check RUNTESTFLAGS="tree-ssa.exp=sra-12.c" on i686-linux and > an avr cross. Because of the pr42585.C precedent I consider it > obvious and will commit it tomorrow unless someone objects. Ok, but please add a comment to it with a short quip about MOVE_RATIO being too small. The idea is that is someone else does a port to a machine similar in nature, they can say, yes, we are a small MOVE_RATIO type of port and I'll just skip it too. Thanks.
Index: src/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c =================================================================== --- src.orig/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c +++ src/gcc/testsuite/gcc.dg/tree-ssa/sra-12.c @@ -21,5 +21,5 @@ int foo (struct S *p) *p = l; } -/* { dg-final { scan-tree-dump-times "l;" 0 "release_ssa"} } */ +/* { dg-final { scan-tree-dump-times "l;" 0 "release_ssa" { target { ! "avr*-*-*" } } } } */ /* { dg-final { cleanup-tree-dump "release_ssa" } } */