On 07/31/2012 05:09 AM, Richard Guenther wrote:
On
Thus, the bad news is that it's hard for the middle-end to
recover alignment of a memory access that is represented as
a builtin function call that takes addresses as parameters
(which also makes them address-taken and thus possibly aliased).
Didn't Andrew have some patches to introduce a GIMPLE_ATOMIC
eventually side-stepping this issue (maybe that used addresses, too)?
yes, but Im not sure I'm going to be able to gimple atomic in for
4.8... Im trying to make sure the C11 stuff gets in then its back to
gimple atomic... you never know tho... I'm just concerned about
introducing it that late in the cycle... I'll reconsider the plan in
the next week or two.. maybe I can get gimple atomic in first then the
C11 stuff... put on a mega-push in august...
Andrew