https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69171
--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
On x86_64-apple-darwin15, between r232062 and r232092, when compiling
gcc.c-torture/compile/SFset.c with -O1 I get
opt/gcc/_clean/gcc/testsuite/gcc.c-torture/compile/SFset.c: In function
'indreg0limm1_set':
/opt/gcc/_clean/gcc/testsuite/gcc.c-torture/compile/SFset.c:73:1: error:
unrecognizable insn:
{indreg0 = limm1; }
^
(insn 28 27 19 2 (set (reg:V4SF 21 xmm0 [104])
(vec_merge:V4SF (vec_duplicate:V4SF (float:SF (mem/u/c:SI (const:DI
(unspec:DI [
(symbol_ref:DI ("glob1") [flags 0x1000]
<var_decl 0x14373a1b0 glob1>)
] UNSPEC_GOTPCREL)) [3 S4 A8])))
(reg:V4SF 21 xmm0 [104])
(const_int 1 [0x1])))
/opt/gcc/_clean/gcc/testsuite/gcc.c-torture/compile/SFset.c:73 -1
(nil))
/opt/gcc/_clean/gcc/testsuite/gcc.c-torture/compile/SFset.c:73:1: internal
compiler error: in extract_insn, at recog.c:2286
Is this the same bug?
