Fw: [PATCH 1/2] BZ#13743: PowerPC - Add a new header for platform specific functions

[Richard Henderson]

On 03/09/12 14:46, Steven Munroe wrote:
> How do you feel about the data stream control register and getting a
> more complete set (than __sync_synchronize()) of memory barriers? *GCC
> or GLIBC)?

That's __atomic_thread_fence (memory_model) now.

Yes, we don't currently have a way to explicitly say "read barrier"
or "write barrier" atm, but those could (1) be introduced as new
memory models or (2) be inferred via optimization from the surrounding
atomic builtins.


r~