[Bug target/95750] [x86] Use dummy atomic insn instead of mfence in __atomic_thread_fence(seq_cst)

[andysem at mail dot ru]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95750

--- Comment #5 from andysem at mail dot ru ---
> Please also read [1] why we avoid -4(%%esp).

I believe, memcheck would complain because the value at -4(%%rsp) would be
uninitialized. This is unfortunate. The workaround could be to initialize it
prior to the atomic operation. This would avoid data dependencies with the
surrounding code.