Re: [PATCH] backwards threader: handle SSA names occurring in abnormal PHIs

On Thu, Aug 13, 2026 at 08:40:38AM +0200, Richard Biener wrote:

> Hmm, there's no reason to not perform analysis on SSA names marked
> as abnormal.  You'd not get any useful control expression for the
> edges, but the copy part holds.  What you have to be careful with
> is not disrupting coalescing during transforms because where
> SSA names on abnormal edges (thus in PHI args) are special is that
> you cannot insert a copy there, so we have to allocate the incoming
> value and the PHI result to the same register.
> 
> So I wonder why gimple_range_ssa_p cares about abnormals at all.
> 
> What can be problematic is duplicating a PHI where the copy
> inherits an abnormal incoming edge - extra copies of abnormally
> marked SSA names are to be avoided.
> 
> So I wonder if you can split this up and remove the abnormal
> restriction from range analysis (and see what breaks - I'm curious).

Surprisingly (for me anyhow), nothing.  Also, the consumers of this
info (substitute_and_fold and tree-vrp.cc) seem to already have guards
for abnormal PHIs.

Could these restrictions be an inherited
restriction from when we had ASSERT_EXPR copies to represent ranges?

Your suggestion reduces my two patches to one line a piece :).

I'll post as a follow-up.

Thanks.
Aldy