This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PR 13722 candidate fix
On Fri, 2004-01-23 at 15:23, Zack Weinberg wrote:
> unadjustible POST_MODIFY were being emitted in the wrong place). So,
> if you wouldn't mind trying the appended revision (changes only
> ia64.c)?
FYI you got the PR number wrong again in the attachment file name.
A -O0 c/ada bootstrap works. A -O1 c/ada bootstrap does not. So there
is the possibility here of doing a binary search to narrow down the
problem to one file, and with a little hackery, we could narrow down the
problem to one function. This doesn't require any heavy lifting, but
will be time consuming.
I'm going to try disabling auto-inc addressing types, to see if maybe
only one of them is broken.
I think it would also be helpful to have debug_rtx() calls in
ia64_split_tmode_move, so I can see the actual before and after rtx
during the ada build. I might be able to spot a problem that way. I
may try doing this.
--
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com