[Bug target/81375] New: [6/7/8 Regression] unrecognizable insn

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81375

            Bug ID: 81375
           Summary: [6/7/8 Regression] unrecognizable insn
           Product: gcc
           Version: 7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
                CC: hjl at gnu dot org
  Target Milestone: ---
              Host: x86_64-linux-gnu
            Target: x86_64-linux-gnu
             Build: x86_64-linux-gnu

It's a crappy scenario, but still it ICEs:

gfortran
/home/marxin/Programming/gcc/gcc/testsuite/gfortran.dg/vect/fast-math-vect-8.f90
-march=lakemont -mfpmath=387+sse -m32
/home/marxin/Programming/gcc/gcc/testsuite/gfortran.dg/vect/fast-math-vect-8.f90:60:0:

     return

Error: unrecognizable insn:
(insn 78 77 79 6 (set (reg:SF 102 [ _16 ])
        (div:SF (reg:SF 218)
            (reg:SF 101 [ _15 ])))
"/home/marxin/Programming/gcc/gcc/testsuite/gfortran.dg/vect/fast-math-vect-8.f90":31
-1
     (nil))
/home/marxin/Programming/gcc/gcc/testsuite/gfortran.dg/vect/fast-math-vect-8.f90:60:0:
internal compiler error: in extract_insn, at recog.c:2311

Started with r229082, before that we did:

f951: Error: bad value (lakemont) for -march= switch
f951: Warning: SSE instruction set disabled, using 387 arithmetics