Re: Rough go moving from 5.4-RELEASE to 6.X



On Tuesday 08 May 2007 16:43, Lane wrote:
On Monday 07 May 2007 20:58, RW wrote:
On Mon, 7 May 2007 20:07:09 -0500

Lane <lane@xxxxxxxxxxxxxx> wrote:
I'm trying to upgrade from 5.4 to ANYTHING and I'm confounded by

I've tried with (in /etc/cvsupfile) "tag=" each of RELENG_5,
RELENG_5_5, RELENG_6, and several others. It even fails when I try
to just get the source to RELENG_5_4.

Did you check UPDATING to see that you actually had the source
for RELENG_5_4?

Yes. I can cvsup or "make update" to any valid release. I can verify
this by, as I mentioned, wiping out /usr/src and then running cvsup.

Have you tried using an empty make.conf file?

Yes. Build still fails.

I can

make buildkernel
make installkernel

and reboot into 5.4, 5.5, 6.0, 6.1 or 6.2

The errors occur only when I use

make buildworld

You do "rm -r /usr/obj", everytime right?


There is this note in /usr/src/UPDATING, which I am currently tracking
as the root of my problems. However it is a bit of a stretch since I
built this system AFTER 7/2004 ... still ... it may be the root ... But
then that doesn't explain why the kernel installs ... UGH! I'm weary
...

20040728:
System compiler has been upgraded to GCC 3.4.2-pre. As with any
major compiler upgrade, there are several issues to be aware of. GCC
3.4.x has broken C++ ABI compatibility with previous releases yet again
and users will have to rebuild all their C++ programs with the new
compiler. A new unit-at-a-time optimization mode, which is default in
this compiler release, is more aggressive in removing unused static
symbols. This is the likely cause of 'make buildworld' breakages with
non-default CFLAGS where optimization level is set to -O2 or higher.

With the upgrade of the system compiler, the kernel has been
upgraded to match the new system compiler. This makes it impossible to
build a new kernel with the old compiler. Upgrade your system via make
buildworld and make kernel (see below) to fix this problem.

Did you try building 5.4-RELEASE? Just to rule out the possibility
of a broken update mechanism?

Just a few wild guesses...

Nikos
_______________________________________________
freebsd-questions@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscribe@xxxxxxxxxxx"



Relevant Pages

  • Kernel upgrade query - want 48-bit LBA
    ... I've never done a kernel upgrade, ... accessing all of the new hard disk. ... stuff - gcc and, I guess, various support libraries for the compiler. ...
    (comp.os.linux)
  • Re: Rough go moving from 5.4-RELEASE to 6.X
    ... System compiler has been upgraded to GCC 3.4.2-pre. ... With the upgrade of the system compiler, the kernel has been upgraded ...
    (freebsd-questions)
  • Re: upgrade 5.3b7
    ... > it can be compiled with the newer system compiler only. ... > installing the world, then the newer compiler ... > build) the kernel without the installworld you are basically using the old ... > the system compiler with an updated version. ...
    (freebsd-questions)
  • RE: upgrade 5.3b7
    ... Subject: upgrade 5.3b7 ... the new kernel. ... It does *not* say you have to do an installworld. ... it can be compiled with the newer system compiler only. ...
    (freebsd-questions)
  • amd64 bitops fix for -Os
    ... filesystem, the kernel mostly hangs. ... When optimizing for speed, the generated code is such that the flags ... Obviously the asm statement must not rely on the compiler setting up ... -inline long find_first_zero_bit(const unsigned long * addr, ...
    (Linux-Kernel)