This is the mail archive of the
mailing list for the GCC project.
Re: --disable-checking bootstrap failure
- From: Nathan Sidwell <nathan at codesourcery dot com>
- To: Eric Christopher <echristo at redhat dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, Diego Novillo <dnovillo at redhat dot com>, stevenb at suse dot de
- Date: Fri, 10 Sep 2004 19:14:17 +0100
- Subject: Re: --disable-checking bootstrap failure
- Organization: Codesourcery LLC
- References: <4141D527.email@example.com> <firstname.lastname@example.org>
Eric Christopher wrote:
On Fri, 2004-09-10 at 09:24, Nathan Sidwell wrote:
I have tracked the --disable-checking bootstrap failure to the assertification
of tree-cfg.c. However, the bug is not a misapplied assert, or
an assert with sideeffects. It must reside elsewhere.
I attach a patch which reinserts the check with an abort, but this is
obviously papering over the problem.
I've attached a preprocessed tree-cfg.i file, with this patch applied
(line 24410). Removing that line causes miscompilation of the file
on (at least) i686-pc-linux-gnu, with '-O2 -g'.
It's getting to the end of the day here, and I'm going to have another
learning curve figuring out where to look next. anyone want to help?
I've got a little bit so I looked some and tracked it down to -O1
-fno-tree-dce that compiles and -O1 gets stuck in a loop. So it's a
conversion somewhere in dce that's causing it. I'll look a bit more, but
wanted to update.
compiling that .i I posted (with or without the offending line), using
a checking compiler gets me different ICES with -O1 with/without
Nathan Sidwell :: http://www.codesourcery.com :: CodeSourcery LLC
email@example.com :: http://www.planetfall.pwp.blueyonder.co.uk