This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix PR optimization/11210
On Mon, 2003-06-30 at 01:22, Eric Botcazou wrote:
> So the function is able to look through NOP_EXPRs to find the correct bitpos,
> but not the correct bitsize! What's the rationale for this semantics?
I see your point now. You are right, this is a problem. pbitsize is
set at the beginning, and then we strip off NOP_EXPRs that can change
the bitsize, but we don't update pbitsize. This is a bug. This means
the function probably only works correct if you strip off NOP_EXPRs, but
it is likely that all of the callers are already doing this. This helps
explain why we got funny results in decode_field_reference when we tried
using STRIP_SIGN_NOPS.
This code goes all of the way back to gcc-2.0, so the problem has been
there for a very long time.
So a ??? comment does make sense here, but it should be specific about
the pbitsize problem.
Jim