[tree-ssa] RFC: Never gimplify ASM_EXPRs

[Paul Koning]

>>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:

 Andrew> Paul Koning writes:
 >> >>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:
 >> ...
 Andrew> Not really; a prefetch instruction is more of a suggestion
 Andrew> than an access, is it not?
 >> NO!  A prefetch instruction, in those architectures where it is
 >> implemented, is NOT a suggestion, it's an order to do real work.

 Andrew> I think we both understand what prefetch does.  The crucial
 Andrew> difference here is that the RHS must *not* be fully evaluated
 Andrew> -- if the operand specifies an invalid memory address, no
 Andrew> exception occurs, and the instruction has no effect.

That's true, and your point is exactly correct.  Linus made
essentiallly the same point when he talked about memory accesses with
side effects and things like that.

       paul