https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71678
Bug ID: 71678
Summary: [avr] ICE from switch / case on long long (casesi +
DImode)
Product: gcc
Version: 5.2.1
Status: UNCONFIRMED
Keywords: ice-on-valid-code
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: gjl at gcc dot gnu.org
Target Milestone: ---
Target: avr
Created attachment 38778
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=38778&action=edit
ice-casesi.c: C test case
$ avr-gcc -mmcu=atmega128 ice-casesi.c -Os -S -fno-tree-switch-conversion
unsigned char foo (long long x)
{
unsigned char y = 0;
switch (x)
{
case 0: y = 67; break;
case 1: y = 20; break;
case 2: y = 109; break;
case 3: y = 33; break;
case 4: y = 44; break;
case 5: y = 37; break;
case 6: y = 10; break;
case 7: y = 11; break;
}
return y;
}
will throw an ICE:
ice-casesi.c: In function 'foo':
ice-casesi.c:16:1: error: unrecognizable insn:
}
^
(insn 55 54 56 4 (parallel [
(set (reg:HI 47)
(minus:HI (subreg:HI (subreg:SI (reg:DI 44) 0) 0)
(reg:HI 45)))
(clobber (scratch:QI))
]) ice-casesi.c:4 -1
(nil))
ice-casesi.c:16:1: internal compiler error: in extract_insn, at recog.c:2287
ice-casesi.c:16:1: internal compiler error: Segmentation fault
Problem is the casesi expander and its explicit SUBREG. Here casesi from
current trunk, but the define_expand looks the same sice r31935:
(define_expand "casesi"
[(parallel [(set (match_dup 6)
(minus:HI (subreg:HI (match_operand:SI 0 "register_operand"
"") 0)
(match_operand:HI 1 "register_operand" "")))
(clobber (scratch:QI))])
...
I used avr-gcc 5.2.1 but affected are all versions of avr-gcc (maybe with a
slightly different test case with more cases or so).
