https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83084
--- Comment #4 from Jakub Jelinek ---
Ah, that might be it. I'd say UNSPEC_VOLATILE should never appear in
DEBUG_INSNs. I'll have a look where it is introduced and why.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83084
--- Comment #3 from Segher Boessenkool ---
Is this because the var_location contains an unspec_volatile? Is that a
target bug, then?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83084
--- Comment #2 from Jakub Jelinek ---
The sched1 dump differences are:
// rs6000_sched_reorder :
// rs6000_variable_issue (more = 7) = 0
-;; 0--> b 1: i10031 [scratch]=unspec[[scratch]] 36
:DU_both_power8,LSU_power8:GENERAL_REGS
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83084
Jakub Jelinek changed:
What|Removed |Added
CC||aoliva at gcc dot gnu.org,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83084
Jakub Jelinek changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last reconfirmed|
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83084
Richard Biener changed:
What|Removed |Added
Target Milestone|--- |7.3