Re: Committed: Fix sh regression for target/55160

[Oleg Endo]

On Thu, 2012-11-01 at 02:31 -0400, Joern Rennecke wrote:
> gen_doloop_end_split creates a pattern that sets pc, hence emit_jump_insn
> has to be used instead of emit_insn.
> 
> Committed as obvious.

I'd like to add a test case for this.

Attached patch was tested with
make -k check-gcc RUNTESTFLAGS="sh.exp=pr55160.c --target_board=sh-sim
\{-m2/-ml,-m2/-mb,-m2a/-mb,-m4/-ml,-m4/-mb,-m4a/-ml,-m4a/-mb}"

OK to install?

Cheers,
Oleg


testsuite/ChangeLog:

        PR target/55160
        * gcc.target/sh/pr55160.c: New.

Index: gcc/testsuite/gcc.target/sh/pr55160.c
===================================================================
--- gcc/testsuite/gcc.target/sh/pr55160.c	(revision 0)
+++ gcc/testsuite/gcc.target/sh/pr55160.c	(revision 0)
@@ -0,0 +1,25 @@
+/* Check that the decrement-and-test instruction is generated.  */
+/* { dg-do compile { target "sh*-*-*" } } */
+/* { dg-options "-O1" } */
+/* { dg-skip-if "" { "sh*-*-*" } { "-m5*"} { "" } }  */
+/* { dg-final { scan-assembler-times "dt\tr" 2 } } */
+
+int
+test_00 (int* x, int c)
+{
+  int s = 0;
+  int i;
+  for (i = 0; i < c; ++i)
+    s += x[i];
+  return s;
+}
+
+int
+test_01 (int* x, int c)
+{
+  int s = 0;
+  int i;
+  for (i = 0; i < c; ++i)
+    s += *--x;
+  return s;
+}