This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: grokfield miscompiled with -O9
On Sat, Jan 08, 2000 at 05:52:22PM -0200, Alexandre Oliva wrote:
> The real problem was that insn 40 shouldn't have been emitted like
> that. I believe subregs should never be used to initialize a
> register like that.
No, that's fine. Quoting from the documentation on SUBREG:
Storing in a non-paradoxical `subreg' has undefined results for
bits belonging to the same word as the `subreg'. This laxity makes
it easier to generate efficient code for such instructions.
So we've identified a bug in expand_compound_operation, or
one of the non-zero-bits functions. I'll look into it with
one of yall's test cases.
> * toplev.c (rest_of_compilation): Initialize cse_not_expected as
> in prepare_function_start().
This is fine.
r~