This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: snapshot still won't bootstrap with BOOT_CFLAGS="-O4 -g"...
- To: mrs at wrs dot com, egcs at cygnus dot com
- Subject: Re: snapshot still won't bootstrap with BOOT_CFLAGS="-O4 -g"...
- From: N8TM at aol dot com
- Date: Mon, 28 Dec 1998 17:37:44 EST
In a message dated 12/28/98 2:12:03 PM Pacific Standard Time, mrs@wrs.com
writes:
<< people that want something to work (anything,
included bootstraping with different flags), should test it often, and
complain when it breaks, and track down who broke it >>
I have been getting the same failure attempting to bootstrap with '-Os
-march=pentiumpro' on several platforms for over 2 months now. I have debated
how loud to complain, don't know to whom to complain, and find today that some
mail readers have not been able to read my linux testsuite reports (and maybe
my Unix ones?). Lately, I have simply been stating the combinations of FLAGS
used to build, hoping that if anyone is interested they will notice that
bootstrap works only with '-O -march=pentiumpro' while the other flags may be
set one notch higher. In fact, recent snapshots do well at '-O2
-march=pentiumpro' except for the bootstrapping problem. It seems to be that
simply changing the FLAGS to a common architecture without "raising" the
optimization level ought to be safer than this.
Likewise, everyone has been reporting an excessive number of the same
testsuite failures on hppa* for some weeks, due to an unrecognized jump_insn.
It would be nice to know that whoever understands this is working on the
problem. At least the one which interrupted the bootstrap has gone away.
Would it help to clean up the test_summary reports (remove '\r' and the like)
before taking them out of linux/unix through Windoze style mail programs?
None of the Windoze mail programs will take the full suggested subject string
without truncating it or spilling it into the text, so I have been leaving out
the part in parentheses.