Re: PATCH: Fix 20000724-1.c

[Mark Mitchell <mark at codesourcery dot com>]

>>>>> "Richard" == Richard Henderson <rth@redhat.com> writes:

    Richard> The actual flaw is that we've lost track of the fact that
    Richard> there is an outstanding reference to a stack address in a
    Richard> register, and we decide that we can deallocate the stack
    Richard> frame before that reference is used.

Ah.

Well, I guess the change to regmove.c still makes sense, since if
someone did explicitly clobber the stack pointer, we should notice
it.  Right?

But, yes, I agree with your analysis.  So, we need to notice the
clobber of the memory, and not combine across that?

Or, I see that Jakub refers to a patch from Jan that does something
different, in that it puts the combinations in conservative locations?
Might something else might later move them around again?

Anyhow, what do you suggest?

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com