__atomic_add_fetch adds a value to some memory, and returns the result.
If there is no direct support for this, expand_builtin_atomic_fetch_op
is asked to implement this as __atomic_fetch_add (which returns the
original value of the mem), followed by the addition. Now, the
__atomic_add_fetch could have been a tail call, but we shouldn't
perform the __atomic_fetch_add as a tail call: following code would
not be executed, and in fact thrown away because there is a barrier
after tail calls.
This fixes it.
Tested on powerpc64-linux {-m32,-m64}. Is this okay for trunk?
Segher
2017-05-28 Segher Boessenkool <seg...@kernel.crashing.org>
PR middle-end/80902
* builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
a call, force the call to not be a tail call.
---
gcc/builtins.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/gcc/builtins.c b/gcc/builtins.c
index 4f6c9c4..3a70693 100644
--- a/gcc/builtins.c
+++ b/gcc/builtins.c
@@ -6079,6 +6079,12 @@ expand_builtin_atomic_fetch_op (machine_mode mode, tree
exp, rtx target,
gcc_assert (TREE_OPERAND (addr, 0) == fndecl);
TREE_OPERAND (addr, 0) = builtin_decl_explicit (ext_call);
+ /* If we will emit code after the call, the call can not be a tail call.
+ If it is emitted as a tail call, a barrier is emitted after it, and
+ then all trailing code is removed. */
+ if (!ignore)
+ CALL_EXPR_TAILCALL (exp) = 0;
+
/* Expand the call here so we can emit trailing code. */
ret = expand_call (exp, target, ignore);
--
1.9.3
