Date: Mon, 21 Dec 1998 12:02:41 -0800 (PST) From: Matthew Dillon <dillon@apollo.backplane.com> To: Bruce Evans <bde@zeta.org.au> Cc: cvs-all@FreeBSD.ORG, cvs-committers@FreeBSD.ORG Subject: Re: cvs commit: src Makefile.inc1 Message-ID: <199812212002.MAA47149@apollo.backplane.com> References: <199812211533.CAA13838@godzilla.zeta.org.au>
next in thread | previous in thread | raw e-mail | index | archive | help
:> Modified files:
:> . Makefile.inc1
:> Log:
:> Elf (and a.out?) rebuild of bootstrap tools appeared to be broken due to
:> some tools requiring header files, such as errno.h, that are softlinks.
:> The mini installation of include files in Makefile.inc1 wasn't doing
:> the job, so it has been ripped out and replaced with a true make of the
:> /usr/src/include/ directory (make all and make beforeinstall). I think
:> the original idea was to save time by not installing unnecessary header
:> files, but it doesn't really save all that much time.
:
:The original idea was to make the bootstrap actually work -- `make all'
:can't be run in src/include because relevant tools like rpcgen haven't
:been built yet.
I don't think a complete bootstrap would work anyway... we already depend
on the compiler and various utilities being properly installed prior to
the tools build :-) rpcgen hasn't changed in ages, it's probably *more*
stable then gcc right now.
:The problem seems to be that cpio creates bogus links and there aren't
:enough fixups. The correct fix may be to back out the large but
:incomplete fixups in rev.1.185-1.186 of src/Makefile. I've never used
:it, and the mtree breakage that inspired it (following symlinks) seems
:to be fixed in another way (nochange directives in BSD.usr.dist).
The problem is not cpio, it's the fact that it's trying to duplicate
the tree from /usr/src/include and that tree is missing a huge number of
header files which are copied or softlinked in /usr/src/include's Makefile.
:> Also, I have moved a NOCLEAN conditional to cover rebuilding the object
:> tree 'Rebuilding the ${OBJFORMAT} obj tree' section. This may or may not
:> be correct but it appears to function properly. If it is not correct we
:> need to find another way to avoid scanning every single file in the
:> entire source hierarchy for make restarts.
:
:This is not correct. It breaks the initial build of the obj tree (if NOCLEAN
:is set initially).
:
:Bruce
Hmm.. you can actually run the initial buildworld with -DNOCELAN set ?
I suppose we can add a new -D directive. In fact, it would be nice if
we could also avoid rebuilding the tools every time we try to start up
a make from where it died but that is less critical (it doesn't take very
much time to rebuild the tools).
How about a NOREBUILD directive to replace the NOCLEAN extension I did.
Any objections? NOREBUILD would automatically set NOCLEAN and NOTOOLS
as well as surpress the rebuilding of the object tree.
-Matt
Matthew Dillon Engineering, HiWay Technologies, Inc. & BEST Internet
Communications & God knows what else.
<dillon@backplane.com> (Please include original email in any response)
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199812212002.MAA47149>