https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71374

--- Comment #5 from Vladimir Makarov <vmakarov at gcc dot gnu.org> ---
The problem is in that p88 does not conflict with a new reload pseudo created
to match p88 and the 3rd output (p91).  They do not conflict as the reload
pseudo has the same value as p88.

So the solution is to provide a different value for the reload pseudo.

I'll have the patch on Friday but it needs some performance testing.  So
probably the patch will be committed on the next week or this Friday if I am
very lucky.

Reply via email to