https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102150
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Ever confirmed|0 |1
Status|UNCONFIRMED |NEW
Last reconfirmed| |2025-02-11
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note the original testcase no longer produces the trap; I have not looked into
why.
But a better testcase comes from
https://github.com/llvm/llvm-project/pull/126807:
```
#define V
int g();
void h(long long);
void f()
{
long long a;
int el2 = g();
if(el2)
asm V ("mrs $0, SPSR_EL2": "=r"(a));
else
asm V ("mrs $0, SPSR_EL3": "=r"(a));
h(a);
}
```
Right now if V is defined to empty, ifcvt (ce1) pulls up one of the inline-asm
to before the branch.
