On Thu, Apr 12, 2012 at 9:22 AM, Alan Modra <amo...@gmail.com> wrote:

> I tried that.  It doesn't work without something else in the insn to
> stop rtl-dce deleting it, so you may as well use SETs.  But thanks for
> the prod in the right direction.  We do get slightly better results
> when the regs are not hidden away in an UNSPEC, for instance
> non-stack writes/reads are seen by the alias oracle to not conflict
> with the epilogue frame deallocation.
> Bootstrapped etc. powerpc-linux.  OK to apply, David?
>        PR target/52828
>        * config/rs6000/rs6000.c (rs6000_emit_stack_tie): Rewrite with
>        tie regs on destination of sets.  Delete forward declaration.
>        (rs6000_emit_stack_reset): Update rs6000_emit_stack_tie calls.
>        (rs6000_emit_prologue): Likewise.
>        (rs6000_emit_epilogue): Likewise.  Use in place of gen_frame_tie
>        and gen_stack_tie.
>        (is_mem_ref): Use tie_operand to recognise stack ties.
>        * config/rs6000/predicates.md (tie_operand): New.
>        * config/rs6000/rs6000.md (UNSPEC_TIE): Delete.
>        (restore_stack_block): Generate new stack tie rtl.
>        (restore_stack_nonlocal): Likewise.
>        (stack_tie): Update.
>        (frame_tie): Delete.

This probably is getting close to the best we can do with GCC's RTL
alias analysis infrastructure.

This version is okay, but I also want to give Richi and Olivier an
opportunity to comment if they still have any concerns.

Thanks, David

Reply via email to