This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH]: Fix PR c/12372
On Wed, Oct 29, 2003 at 03:20:29PM +0100, Eric Botcazou wrote:
> * flow.c (propagate_one_insn) [CALL_INSN]: Initialize
> sibcall_p on block entry. Invalidate MEMs related to the
> frame pointer for const or pure sibcalls.
First, it's overkill, but I guess that's ok as long as we
remember that and fix it later.
Second, it invalidates everything on the list, guaranteed.
A better way to apply this hammer is to check EDGE_SIBCALL
in init_propagate_block_info before adding them in the
Incidentally, that condition is growig large enough it
probably ought to be pulled out to its own function...