On Mon, Jan 18, 2016 at 9:28 AM, Bin Cheng <bin.ch...@arm.com> wrote: > Hi, > Turns out the check on number of iv_uses is still too large on target hppa. > It only supports small offset in REG+offset addressing mode for floating > point load/store. Even with this restriction, the grouped version is better > than before, so I am going to further relax the check condition for it. > Test run on HPPA. Applied as an obvious change. > > Thanks, > bin > > gcc/testsuite/ChangeLog > 2016-01-18 Bin Cheng <bin.ch...@arm.com> > > PR tree-optimization/66797 > * gcc.c-torture/execute/pr65447.c: Relax check condition. > Hmm, the patch is at revision 232497.
Index: gcc/testsuite/gcc.dg/tree-ssa/pr65447.c =================================================================== --- gcc/testsuite/gcc.dg/tree-ssa/pr65447.c (revision 232496) +++ gcc/testsuite/gcc.dg/tree-ssa/pr65447.c (revision 232497) @@ -50,4 +50,4 @@ } /* We should groups address type IV uses. */ -/* { dg-final { scan-tree-dump-not "\\nuse 5\\n" "ivopts" } } */ +/* { dg-final { scan-tree-dump-not "\\nuse 21\\n" "ivopts" } } */ Index: gcc/testsuite/ChangeLog =================================================================== --- gcc/testsuite/ChangeLog (revision 232496) +++ gcc/testsuite/ChangeLog (revision 232497) @@ -1,3 +1,8 @@ +2016-01-18 Bin Cheng <bin.ch...@arm.com> + + PR tree-optimization/66797 + * gcc.c-torture/execute/pr65447.c: Relax check condition. + 2016-01-18 Richard Biener <rguent...@suse.de> PR tree-optimization/69170 Thanks, bin