Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 16 Sep 2004 12:55:43 -0600
From:      John Nielsen <lists@jnielsen.net>
To:        freebsd-current@freebsd.org
Cc:        current@freebsd.org
Subject:   Re: Signal 4 installing kernel/world built for different cputype
Message-ID:  <200409161255.43859.lists@jnielsen.net>
In-Reply-To: <53FDD37E-080F-11D9-A3FD-003065ABFD92@mac.com>
References:  <1095355065.4149cab9b58ff@webmail.jnielsen.net> <53FDD37E-080F-11D9-A3FD-003065ABFD92@mac.com>

next in thread | previous in thread | raw e-mail | index | archive | help
--Boundary-00=_vGeSBeMhWOB7Tn7
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

On Thursday 16 September 2004 12:36 pm, Charles Swiger wrote:
> On Sep 16, 2004, at 1:17 PM, John Nielsen wrote:
> > I use a build machine to build world and kernels for other, slower
> > machines.  The build machine is an Athlon Thunderbird running
> > 5.3-BETA4,
> > and I'd like to upgrade the other machines to 5.3 as well.  I have
> > compiled everything that runs on the build system with
> > CPUTYPE?=athlon-tbird.
> >
> > When I do a buildworld/buildkernel for a pentium machine with
> > CPUTYPE=i586
> > on make's command line, it looks like I get incompatible binaries in
> > obj/usr/src/i386/legacy/usr/bin.
>
> This is not surprising.  You need to set your CPUTYPE to the lowest
> common denominator of CPUs which you want the compiled code to run on.

Perhaps I wasn't clear.  I maintain separate obj directories for each 
cputype, so code compiled for athlon-tbird goes under obj/tbird, and code 
compiled for i586 goes under obj/586.  I'm attaching my build script and my 
install scripts for "stealth" (the thunderbird build machine) and 
"neo" (one of the pentium machines).  For major updates I would of course 
reboot with the new kernel before installing world--I mainly use the 
install scripts for minor updates.

I have watched the build output enough to know that the correct cpu and arch 
flags are being used, and except for the directory I mentioned above, 
correct code appears to be generated.  The main thing I want to know at the 
moment is how I could verify that the correct flags are used to build 
install and friends in the i386/legacy directory, since I already know that 
the binaries generated in that directory do not work correctly and that 
others do.

JN


--Boundary-00=_vGeSBeMhWOB7Tn7--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200409161255.43859.lists>