https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85034
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Priority|P3 |P2
Status|UNCONFIRMED |NEW
Last reconfirmed| |2018-03-22
CC| |jakub at gcc dot gnu.org,
| |vmakarov at gcc dot gnu.org
Target Milestone|--- |6.5
Summary|-O1 internal compiler |[6/7/8 Regression] -O1
|error: in |internal compiler error: in
|elimination_costs_in_insn, |elimination_costs_in_insn,
|at reload1.c:3633 |at reload1.c:3633
Ever confirmed|0 |1
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Simplified testcase:
void
foo ()
{
volatile float a;
struct S { char a; } b;
asm volatile ("" : "=r" (a) : "0Ir" (b));
}
which ICEs also in C and doesn't need lambdas.
Started to ICE in between r126325 (which still works) and r126402 (which ICEs
with:
pr85034.c: In function ‘foo’:
pr85034.c:7: error: unrecognizable insn:
(insn 17 15 6 2 pr85034.c:6 (set (reg:SF 60)
(const_int 0 [0x0])) -1 (nil))
pr85034.c:7: internal compiler error: in extract_insn, at recog.c:1991
) and since most likely r160260 it is the ICE in the elimination_costs_in_insn.
