On Tue, Dec 19, 2017 at 04:29:53PM -0200, Alexandre Oliva wrote:
> On Dec 15, 2017, Jakub Jelinek <ja...@redhat.com> wrote:
>
> > We want to verify there are no statements before labels.
>
> Erhm, actually...
>
> We already verify that in gimple_verify_flow_info.
>
> Is there any point in having such redundant (?) verification?
Ah, you're right. The reason I've added was that you've added
if (is_gimple_debug (gsi_stmt (gsi)))
continue;
there and so it didn't catch those. But now you're reverting that
and so it should be effective again.
Has the patch you've posted passed bootstrap/regtest?
I'll try to eyeball it in detail tomorrow (or if you have spare cycles,
just try to verify it against revision before SFNs were introduced)
that the affected spots are reverted to the old code unless needed for SFNs
without markers in between bbs.
Jakub
