Date: Thu, 21 Dec 2000 09:35:42 +0200 From: Neil Blakey-Milner <nbm@mithrandr.moria.org> To: Marcel Moolenaar <marcel@cup.hp.com> Cc: Warner Losh <imp@village.org>, Assar Westerlund <assar@FreeBSD.ORG>, Dag-Erling Smorgrav <des@ofug.org>, cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG, marcel@FreeBSD.ORG Subject: Re: cvs commit: src/sys/libkern strlcat.c strlcpy.c src/sys/sys libkern.h src/sys/conf files Message-ID: <20001221093542.A78846@mithrandr.moria.org> In-Reply-To: <3A412117.64454559@cup.hp.com>; from marcel@cup.hp.com on Wed, Dec 20, 2000 at 01:13:59PM -0800 References: <xzpitohzu34.fsf@flood.ping.uio.no> <xzpae9tztba.fsf@flood.ping.uio.no> <5lsnnlkcf7.fsf@assaris.sics.se> <xzp66khzsdn.fsf@flood.ping.uio.no> <5l4s01ka92.fsf@assaris.sics.se> <200012181724.KAA91757@harmony.village.org> <20001219081616.D54775@gsmx07.alcatel.com.au> <20001220142934.I644@ringworld.oblivion.bg> <20001220171617.A49980@mithrandr.moria.org> <3A412117.64454559@cup.hp.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed 2000-12-20 (13:13), Marcel Moolenaar wrote:
> Neil Blakey-Milner wrote:
> >
> > On Wed 2000-12-20 (14:29), Peter Pentchev wrote:
> > > Hmm.. quite off-topic, but - is this so on -current? (I'll be able to
> > > test this in a matter of hours, refetching my -current source tree)
> > > It most definitely is NOT correct for 4-stable - the kernel build
> > > requires 'nm' (IIRC) from /usr/obj; if not 'nm', then something else
> > > from the build toolchain.
>
> -stable isn't different from -current in that respect. To allow for
> cross-building, a nm(1) suitable for the target is built and installed
> in the object tree. This nm(1) is found by setting PATH to include the
> object tree. As long as you're not trying to cross-build a kernel then
> the nm(1) installed on your system works fine. Note that building a
> -current kernel on a -stable machine is also considered a cross-build.
>
> > It's "equivalently broken" because we allow the use of gcc, ld, and
> > friends off the installed filesystem if they're not in the obj tree by
> > virtue of having a PATH that allows this.
>
> This is a design decision. Explain why you think it's brokenness.
>
> And as with a lot of things, it gets lost under piles of other things to
> look at :-)
>
> > I suppose this is a nudge to look at it again - we really need to fix
> > this.
>
> I'm afraid you have to fill me in again. it's been too long :-/
I agree with the design decision to make it buildable without a
populated object tree. It'd be silly any other way, really.
The problem is that "nm" uses "objformat" to find the correct binary to
run, and it uses OBJFORMAT_PATH to find that out. In the kernel
environment, OBJFORMAT_PATH is set to (something like)
${.OBJDIR}/usr/libexec. I propose that becomes
"${.OBJDIR}/usr/libexec:/usr/libexec", for the failover effect of PATH
to be preserved.
Neil
--
Neil Blakey-Milner
nbm@mithrandr.moria.org
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?20001221093542.A78846>