On 27 March 2014 10:25, Renato Golin <renato.go...@linaro.org> wrote: > On 27 March 2014 10:12, Andreas Schwab <sch...@suse.de> wrote: >> Can't you use __builtin_frame_address (0) instead? > > That would give me the frame pointer, not the stack pointer, and the > user would have to calculate manually the offset to get the actual > stack pointer, which would be target-specific, possibly making it even > worse. Is that what you meant?
I think this intrinsic makes no more guarantees than __builtin_frame_address. In neither case can you derive the actual, current stack pointer with certainty. As far as I'm aware, the users don't actually care about precision (they'd better not!).The difference is that this one doesn't force a frame pointer to be created, which can be good for optimised code. Cheers. Tim.
