Re: [PATCH IRA] update_equiv_regs fails to set EQUIV reg-note for pseudo with more than one definition

[Jeff Law <law@redhat.com>]

On 09/27/14 08:48, Felix Yang wrote:
> Thanks for the explaination.
> I have changed the loop_depth into a short interger hoping that we can
> save some memory :-)
Thanks.

> Attached please find the updated patch. Bootstrapped and reg-tested on
> x86_64-suse-linux.
> Please do a final revew once the assignment is ready.
>
> As for the new list walking interface, I choose the function
> "no_equiv" and tried the "checked cast" way.
> The bad news is that GCC failed to bootstrap with the following change:
>
> Index: ira.c
> ===================================================================
> --- ira.c (revision 215536)
> +++ ira.c (working copy)
> @@ -3242,12 +3242,12 @@ no_equiv (rtx reg, const_rtx store ATTRIBUTE_UNUSE
>     void *data ATTRIBUTE_UNUSED)
>   {
>     int regno;
> -  rtx list;
> +  rtx_insn_list *list;
>
>     if (!REG_P (reg))
>       return;
>     regno = REGNO (reg);
> -  list = reg_equiv[regno].init_insns;
> +  list = as_a <rtx_insn_list *> (reg_equiv[regno].init_insns);
>     if (list == const0_rtx)
>       return;
>     reg_equiv[regno].init_insns = const0_rtx;
> @@ -3258,9 +3258,9 @@ no_equiv (rtx reg, const_rtx store ATTRIBUTE_UNUSE
>       return;
>     ira_reg_equiv[regno].defined_p = false;
>     ira_reg_equiv[regno].init_insns = NULL;
> -  for (; list; list =  XEXP (list, 1))
> +  for (; list; list = list->next ())
>       {
> -      rtx insn = XEXP (list, 0);
> +      rtx_insn *insn = list->insn ();
>         remove_note (insn, find_reg_note (insn, REG_EQUIV, NULL_RTX));
>       }
>   }
Yea.  I'm going to post a patch shortly to go ahead with this 
conversion.  There's a couple issues that come into play.

First const0_rtx is not an INSN, so we *really* don't want it in the 
INSN field of an INSN_LIST.  That's probably the ICE you're seeing.

const0_rtx is being used to mark pseudos which we've already determined 
can't have a valid equivalence.  So we just need a different marker. 
That different marker must be embeddable in an INSN_LIST node.   The 
easiest is just a NULL insn ;-)

The other tests for the const0_rtx marker in ira.c need relatively 
trivial updating.  And in the end we don't need the checked cast at all ;-)



Jeff