This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Merge results as of 2003-05-06
- From: Diego Novillo <dnovillo at redhat dot com>
- To: Jeff Law <law at redhat dot com>
- Cc: "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>,Jason Merrill <jason at redhat dot com>
- Date: 08 May 2003 07:33:23 -0400
- Subject: Re: [tree-ssa] Merge results as of 2003-05-06
- Organization: Red Hat Canada
- References: <200305080304.h483451f010077@speedy.slc.redhat.com>
On Wed, 2003-05-07 at 23:04, law@redhat.com wrote:
> In message <20030507130003.GB7705@tornado.toronto.redhat.com>, Diego Novillo wr
> ites:
> >On Tue, May 06, 2003 at 10:24:46PM -0600, law@redhat.com wrote:
> >
> >> When/how precisely was the FUNCTION_DECL removed?!?
> >>
> >No clue yet. It doesn't look like it should be hard to fix,
> >though. I'll start working on it after I commit the merge.
> I'm pretty sure I've found this. Ironically, this was the same bug I
> was tracking in my work on the unused variable elimination code.
>
Yeah, I found it in the BIND_EXPR merging we do in
remove_useless_stms_and_vars. But I like your solution better.
My solution allowed the BIND_EXPRs to be merged and adjusted the BLOCK_*
for the inner BIND_EXPR. However, Jason told me on IRC that
manipulating the BLOCK_* that way was not advisable.
Besides the fix involved a call to chainon() to adjust BLOCK_CHAIN, so
it had the potential of getting expensive for little gain.
Diego.