https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69238
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2016-01-13
CC| |jakub at gcc dot gnu.org,
| |meissner at gcc dot gnu.org,
| |vmakarov at gcc dot gnu.org
Ever confirmed|0 |1
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Very likely a dup of PR69195. Again, similarly to the other PR, there are
numerous loads followed by numerous stores, pseudo 242 is REG_DEAD on the last
load, then there is some unrelated load into pseudo 298 and then stores.
Now, IRA decides because of the live info that both pseudo 242 and 298 are put
into r4, but then reload (the testcase works with -mlra, why can't finally the
powerpc* port switch away from reload :( ?!) chooses to materialize the loads
next to the corresponding stores, and uses there the r4 [242] eventhough it has
been clobbered already by the load to r4 [298].
