Hi, While working on an aarch64 patch, I ran into wrong code produced by my patch. I am not ready to submit the patch yet but since I reduced the testcase and there was no testcase in the testsuite yet, I thought I commit the testcase. This testcase is reduced from aarch64_float_const_representable_p in aarch64.c.
Thanks, Andrew Pinski ChangeLog: * gcc.c-torture/execute/20141125-1.c: New testcase.
Index: testsuite/gcc.c-torture/execute/20141125-1.c =================================================================== --- testsuite/gcc.c-torture/execute/20141125-1.c (revision 0) +++ testsuite/gcc.c-torture/execute/20141125-1.c (revision 0) @@ -0,0 +1,17 @@ +int f(long long a) __attribute__((noinline,noclone)); +int f(long long a) +{ + if (a & 0x3ffffffffffffffull) + return 1; + return 1024; +} + +int main(void) +{ + if(f(0x48375d8000000000ull) != 1) + __builtin_abort (); + if (f(0xfc00000000000000ull) != 1024) + __builtin_abort (); + return 0; +} +