This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: enable checking for stage1 compiler
- From: Daniel Jacobowitz <drow at false dot org>
- To: Nathan Sidwell <nathan at codesourcery dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>,Mark Mitchell <mark at codesourcery dot com>
- Date: Mon, 13 Sep 2004 10:31:02 -0400
- Subject: Re: enable checking for stage1 compiler
- References: <4145A648.6040300@codesourcery.com>
On Mon, Sep 13, 2004 at 02:53:12PM +0100, Nathan Sidwell wrote:
> This patch builds the stage1 compiler with checking enabled, regardless
> of the final compiler's checking status.
>
> The change to ssa_verify is necessary to avoid a comparison failure.
> Without it checking compiler and unchecking compilers end up with dominance
> graphs with differently ordered child nodes.
>
> booted & tested on i686-pc-linux-gnu, ok?
Thinking about this... this means that stage1 (used to build stage2)
will be --enable-checking and stage2 (used to build stage3) will be
--disable-checking. So to bootstrap the --enable-checking and
--disable-checking compilers need to produce exactly the same code.
Obviously (as in your next patch) this is desirable. But I think that
the beginning of stage3 isn't a good time for it, since we haven't been
testing that. I bet it's going to turn up a lot of similar problems...
Maybe I'm just being paranoid again.
--
Daniel Jacobowitz