As discussed in D118021 <https://reviews.llvm.org/D118021>, `clang -m32` on
Solaris/sparcv9 currently incorrectly doesn't inline atomics on 8-byte
operands, unlike `gcc`. With the workaround in that patch in place, we're
left with may undefined references to `__sync_val_compare_and_swap_8`,
which isn't provided by `libatomic`. This reference is due to the use of
`__sync_val_compare_and_swap` in `sanitizer_atomic_clang.h`'s
`atomic_compare_exchange_strong`. As is already done in
`scudo/standalone/atomic_helpers.h`, using `__atomic_compare_exchange`
instead avoids this problem.
Tested on `sparcv9-sun-solaris2.11`, `amd64-pc-solaris2.11`, and
`x86_64-pc-linux-gnu`.
Differential Revision: https://reviews.llvm.org/D118024
inline bool atomic_compare_exchange_strong(volatile T *a, typename T::Type *cmp,
typename T::Type xchg,
memory_order mo) {
- typedef typename T::Type Type;
- Type cmpv = *cmp;
- Type prev;
- prev = __sync_val_compare_and_swap(&a->val_dont_use, cmpv, xchg);
- if (prev == cmpv) return true;
- *cmp = prev;
- return false;
+ // Transitioned from __sync_val_compare_and_swap to support targets like
+ // SPARC V8 that cannot inline atomic cmpxchg. __atomic_compare_exchange
+ // can then be resolved from libatomic. __ATOMIC_SEQ_CST is used to best
+ // match the __sync builtin memory order.
+ return __atomic_compare_exchange(&a->val_dont_use, cmp, &xchg, false,
+ __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST);
}
template<typename T>