Re: [PATCH] More REG_EQ* notes cleanups

[Grigory Zagorodnev <grigory_zagorodnev at linux dot intel dot com>]

Roger Sayle wrote:
Hi, Roger!
> wondering whether you could test whether reverting just the local-alloc.c
> I think the clean-up of REG_EQ* notes is a good thing, so to avoid
> reverting the entire patch, it'd be good to narrow down the cause/location
> of the problem.
Yes. It is enough to back out local-alloc.c changes only.

As for a minimal reproducer, it would take some extra time to prepare 
something suitable. Here is a single line cut from benchmark source on 
that compiler breaks. Corresponding assembly code is listed below.

fprintf (stdout, " %d(%d)", kb->hmm_hist[i], (k*100)/kb->nfr);

<<< good one
        idivl   32(%ecx)
        movl    $.LC7, 4(%esp)
        movl    %eax, 12(%esp)
        movl    stdout, %eax
        movl    %eax, (%esp)
        call    fprintf

>>> bad one
        idivl   32(%ecx)
        movl    %eax, -24628(%ebp)
        movl    stdout, %eax
        movl    $.LC7, 4(%esp)
        movl    %eax, (%esp)
        call    fprintf

- Grigory