On 10 June 2015 at 01:22, Ken Moffat <[email protected]> wrote:

> On Tue, Jun 09, 2015 at 03:56:34PM -0700, Paul Rogers wrote:
> > My desktop has two identical twin boxes.  I've been building my
> > (B)LFS-7.2 on one of them.  I think it's ready to be clonable,
> > distributable.  I packaged it up (using the as-built binaries tarballs
> > made by my package manager immediately after compiling each package (a
> > basic process I've used through 4 previous builds and dozens of
> > clones)), and installed it on the other box.  It runs.  It has the
> > original LFS 3.5.2 kernel, and before I "got into" BLFS I'd patched the
> > kernel up to 3.9.11.  But both kernels were configured for starting on a
> > generic box, with the intent that once "up" one would soon rebuild the
> > kernel for the specific box.  I'm to that point.  I brought the working
> > config file from the development box, and get the following:
> >
> > ___8<...
> >   CC      arch/x86/kernel/acpi/cstate.o
> >   LD      arch/x86/kernel/acpi/built-in.o
> >   CC      arch/x86/kernel/apic/apic.o
> > arch/x86/kernel/apic/apic.c: In function 'lapic_next_deadline':
> > arch/x86/kernel/apic/apic.c:475:2: error: 'MSR_IA32_TSC_DEADLINE'
> > undeclared (first use in this function)
> > arch/x86/kernel/apic/apic.c:475:2: note: each undeclared identifier is
> > reported only once for each function it appears in
> > make[3]: *** [arch/x86/kernel/apic/apic.o] Error 1
> > make[2]: *** [arch/x86/kernel/apic] Error 2
> > make[1]: *** [arch/x86/kernel] Error 2
> > make: *** [arch/x86] Error 2
> >
> If I've parsed that correctly, you are now trying to build with a
> .config specific to the second machine ?  If that is not the case,
> I've no idea but google might have a pointer (try searching for
> error: MSR_IA32_TSC_DEADLINE undeclared
> ).
>
> Daily, I try to sort-of keep up with the kernel list, and one of the
> things I *frequently* see is messages about undefined identifiers -
> mostly for linux-next, but some get through to released kernels.
> And the root cause is that a different .config exposes a kernel
> dependency which the person who made the change to the kernel code
> had not tested.
>
> If it was me, I would probably not be trying to roll-out such an old
> system.  But if I needed to, I would be using a stable kernel.  In
> this case, 3.10.current sounds like the way to go (it is still
> maintained).  I forget the details now, but at some point the udev
> options changed : I'm > 99% certain that I have moved systems built
> with 3.8 and 3.9 kernels to 3.10, but less sure if 3.14 was ok
> there.  And even with 3.10 it is possible that I had to "pay
> attention" to the help (for a sysfs change, I suppose).
>
> For preference, I would probably try 3.14.latest, see if it boots,
> and if not try 3.10.latest.  If neither boot, you have probably
> decided to build an "uncommon" config and (arguably) got something
> wrong (been there, done that, most of the scars have healed ;)
>
> Although I have switched my server to 3.18, which is still receiving
> maintenance, I note that updates seem to be a bit slower than for
> 3.14 and 3.10 - with luck, when I upgrade my server I'll be able to
> use 4.1 which should suffice for longer than I plan to keep that
> system.
>

I'm not sure that it's a kernel version problem, but I can only speak from
from my own experience. I have a hybrid box which started out as LFS 7.2,
and over time I've added various packages, both from the BLFS book and
other packages not in the book.  Its had a number of different kernel
versions, and now It's running kernel 4.03 without any (kernel) issues.
Personally, I never use kernel patches to update the kernel; I always
re-build from scratch, but using the same .config where possible.  I think
that's the safer way to do things.

Richard
-- 
http://lists.linuxfromscratch.org/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to