[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: where to report bootstrap bugs?

From: Ted Zlatanov
Subject: Re: where to report bootstrap bugs?
Date: Wed, 02 Jan 2008 07:30:33 -0600
User-agent: Gnus/5.110007 (No Gnus v0.7) Emacs/22.1 (gnu/linux)

On Mon, 31 Dec 2007 14:15:10 -0800 Dan Nicolaescu <address@hidden> wrote: 

DN> Ted Zlatanov <address@hidden> writes:
>> On Mon, 31 Dec 2007 22:14:25 +0100 Michael Albinus <address@hidden> wrote: 
MA> Ted Zlatanov <address@hidden> writes:
>> >> Is there a more appropriate place to ask about `make bootstrap' problems
>> >> from a CVS checkout, if emacs-devel is not it?
>> >> 
>> >> After 1.5 months since I submitted a problem report to emacs-devel (I
>> >> can't do a bug report because the bootstrap won't build) I haven't heard
>> >> anything.  While I may be the only one having the problem, my Ubuntu
>> >> installation is hardly unusual so other users could be running into this
>> >> as well.
MA> I don't remember your report, but from time to time I need a "make
MA> distclean; ./configure; make bootstrap". Maybe this is sufficient for
MA> you as well?  I'm also running Ubuntu.
>> I thought I was going crazy, so I did this, plus a clean checkout to a
>> completely new directory...  Nothing worked.  I traced the problem to a
>> test in fileio.c, and I wonder if it's a system library issue.  See my
>> report, which I reposted.

DN> You can try compiling with -O0 to try exclude compiler bugs...

Ah, thanks for the hint!  I looked through my CFLAGS and found that the
bug happens with CFLAGS '-malign-double' which I had turned on together
with other optimization flags like -O9, -ffast-math, and -funroll-loops.
None of the others had an effect (and they shouldn't), but this one
broke the bootsrap build.  I kept looking at the compiler output while
debugging instead of looking at the compiler itself, that's why I missed
this so far.  Can anyone else verify this is the problem?  For me, it
has fixed the build to remove the -malign-double flag.

I don't know if -malign-double in the CFLAGS should be considered a user
screwup or an actual Emacs bug since it breaks the bootstrap (at least
on i686 32-bit).  Maybe the bootstrap process should catch this flag and
at least complain about it.  In any case, thanks for your help!


reply via email to

[Prev in Thread] Current Thread [Next in Thread]