Message ID | 4D9F24FB.7060901@codesourcery.com |
---|---|
State | New |
Headers | show |
On 8 April 2011 16:08, Janis Johnson <janisjo@codesourcery.com> wrote: > Test gcc.target/arm/sync-1.c specifies -march=armv7-a and fails > execution for multilibs whose hardware or simulator doesn't support that > architecture. The test uses __sync_fetch_and_add, which GCC doesn't > support for all arm targets. Using the -march option allows it to > compile if the multilib options don't conflict, but the test fails at > runtime, often with hangs, if the simulator used for testing doesn't > support atomic builtins. This patch removes the -march option and > restricts the test to targets that support atomic builtins. > > I don't yet have an FSF copyright assignment (it's currently in the > hands of the FSF) but this patch is small enough to not require one. If > this is OK someone else will need to appy it. This is OK. Ramana >
Index: gcc/testsuite/gcc.target/arm/sync-1.c =================================================================== --- gcc/testsuite/gcc.target/arm/sync-1.c (revision 320350) +++ gcc/testsuite/gcc.target/arm/sync-1.c (working copy) @@ -1,5 +1,5 @@ -/* { dg-do run } */ -/* { dg-options "-O2 -march=armv7-a" } */ +/* { dg-do run { target sync_int_long } } */ +/* { dg-options "-O2" } */ volatile int mem;