gcc.dg/sms-3.c and gcc.dg/sms-6.c fail on powerpc when -fsched-pressure is
used. The -fsched-pressure option changes the value returned by
rs6000_issue_rate which in turn affects the computed initiation interval in the
SMS code and leads to failure to modulo schedule the single loop in sms-3.c and
1 of the 3 loops in sms-6.c. This patch adds that option for powerpc and is a
precursor to a patch I'll submit shortly that enables -fsched-pressure by
default for the rs6000 port.
I have verified the updated tests pass both before and after my patch to enable
-fsched-pressure is applied. Ok for trunk?
-Pat
testsuite/ChangeLog:
2016-12-20 Pat Haugen <pthau...@us.ibm.com>
* gcc.dg/sms-3.c: Add -fno-sched-pressure for powerpc.
* gcc.dg/sms-6.c: Likewise.
Index: gcc.dg/sms-3.c
===================================================================
--- gcc.dg/sms-3.c (revision 243651)
+++ gcc.dg/sms-3.c (working copy)
@@ -1,5 +1,6 @@
/* { dg-do run } */
/* { dg-options "-O2 -fmodulo-sched -funroll-loops -fdump-rtl-sms --param sms-min-sc=1 -fmodulo-sched-allow-regmoves" } */
+/* { dg-options "-O2 -fmodulo-sched -funroll-loops -fdump-rtl-sms --param sms-min-sc=1 -fmodulo-sched-allow-regmoves -fno-sched-pressure" { target powerpc*-*-* } } */
extern void abort (void);
Index: gcc.dg/sms-6.c
===================================================================
--- gcc.dg/sms-6.c (revision 243651)
+++ gcc.dg/sms-6.c (working copy)
@@ -1,7 +1,7 @@
/* { dg-do run } */
/* { dg-require-effective-target size32plus } */
/* { dg-options "-O2 -fmodulo-sched -fdump-rtl-sms --param sms-min-sc=1" } */
-/* { dg-options "-O2 -fmodulo-sched -fdump-rtl-sms --param sms-min-sc=1 -fmodulo-sched-allow-regmoves" { target powerpc*-*-* } } */
+/* { dg-options "-O2 -fmodulo-sched -fdump-rtl-sms --param sms-min-sc=1 -fmodulo-sched-allow-regmoves -fno-sched-pressure" { target powerpc*-*-* } } */
extern void abort (void);
