Subject: Re: problems trying to build today's -current
To: Frederick Bruckman <firstname.lastname@example.org>
From: Steven M. Bellovin <email@example.com>
Date: 02/14/2003 13:52:10
In message <Pine.NEB.firstname.lastname@example.org>, Fred
erick Bruckman writes:
>On Fri, 14 Feb 2003, Alan Barrett wrote:
>> On Fri, 14 Feb 2003, Steve Bellovin wrote:
>> > Using build.sh kernel=... I get this:
>> > cc -O -o nbmake *.o
>> > ===> DESTDIR path:
>> > ===> TOOLDIR path: /usr/src/tools/obj/tools.NetBSD-1.6M-i386
>> > ===> Created /usr/src/tools/obj/tools.NetBSD-1.6M-i386/bin/nbmake
>> > ===> Updated /usr/src/tools/obj/tools.NetBSD-1.6M-i386/bin/nbmake-i386
>> > ===> Building kernel without building new tools
>> build.sh classifies this as pilot error. The present code is similar
>> to what I wrote some time ago, and I didn't know a good way, within the
>> existing framework, to check whether building tools was necessary.
>> > I assume I need a new config; I would think, though, that the toolchain
>> > would ((make that "should") do that for me, given the usual warnings
>> > about not running a new userland without first upgrading the kernel.
>> Yes, I agree that build.sh "should" do that for you, but it doesn't.
>> This could be fixed by enhancing the framework to touch various marker
>> files at various points, and to compare timestamps of the marker files
>> at various other points.
>I wouldn't like that. Usually, you need to update the tools a maximum
>of one time after updating the sources, and then you build different
>kernels, sets, etc. with the same tools. Only the user knows if he
>just updated all his sources, or if he just tweaked his kernel config.
>> The instructions in the BUILDING file try to make it clear that it's
>> your responsibility to run "build.sh tools" before "build.sh kernel=FOO"
>> (or to run them both together, as "build.sh tools kernel=FOO").
It's odd -- I didn't see the warning line in the text I cut-and-pasted.
It seemed to be building much of the toolchain (at least I think it
was; I no longer have that log), so I didn't look more closely. My
apologies for the confusion I caused.
(And my next problem was solved by doing today's second cvs update...)
--Steve Bellovin, http://www.research.att.com/~smb (me)
http://www.wilyhacker.com (2nd edition of "Firewalls" book)