This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: New GCC 3.0 build failure on FreeBSD 4.3
- To: mark at codesourcery dot com, pfeifer at dbai dot tuwien dot ac dot at
- Subject: Re: New GCC 3.0 build failure on FreeBSD 4.3
- From: Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>
- Date: Mon, 11 Jun 2001 17:32:30 -0500 (CDT)
- CC: gcc-patches at gcc dot gnu dot org
- Reply-to: rittle at labs dot mot dot com
>> As of today, GCC 3.0 fails to build on FreeBSD 4.3:
>> I'm really getting worried about this kind of breaking happening *that*
>> late in the release process!
> I am looking at this right on. You can be sure that it was working
> fine Friday night at 6PM CDT. But it started breaking in automatic
> builds which started at 10PM PDT that same night.
Non-fatal warning seen in log from bootstrap early Friday night:
[...]/gcc-3.0/libiberty/strerror.c:468: warning: type qualifiers for \
`sys_nerr' conflict with previous decl
Became fatal later that night:
[...]/gcc/libiberty/strerror.c:468: conflicting types for `sys_nerr'
/usr/include/stdio.h:224: previous declaration of `sys_nerr'
If I were a betting man, I'd say this patch broke us:
< 2001-06-08 Mark Mitchell <mark@codesourcery.com>
<
< * c-decl.c (grokdeclarator): Make the TREE_TYPE for declarations
< accurately reflect its cv-qualification.
< * c-typeck.c (type_lists_compatible_p): Ignore the top-levl
< cv-qualifiers on function types.
<
Mark, how do you want this fixed?
I will be able to work on this tonight until 11PM CDT. Until I hear
from you, I will try to figure out why HAVE_SYS_ERRLIST is not defined
when libiberty/strerror.c is compiled since libiberty attempts to hide
the OS version of the symbol when that macro is defined (and there is
a nice comment explaining that BSD 4.4 ``might declare sys_errlist in
a way that the compiler might consider incompatible with our later
declaration, perhaps by using const attributes.''.
Regards,
Loren