I wrote http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49411, and I have some general questions:
1. I left it as normal P3, is that appropriate for an obvious ICE like this? 2. Should I open another bug for 4.6? 3. What's the best/right way to specify testcases in GCC bugs? And the specific question about how to fix it: I see in gcc/config/i386/sse.md that 'define_insn "xop_rotr<mode>3"' matches operand 2 with "const_0_to<sserotatemax>_operand"; doesn't that preclude the -1 from matching? -- Quentin
