help: why can't my hand-rolled kernel mount an LVM root FS?

[Robert P. J. Day]

On Wed, 7 Jul 2010, Andy Whitcroft wrote:

> On Wed, Jul 07, 2010 at 03:49:34AM -0400, Robert P. J. Day wrote:
> > On Tue, 6 Jul 2010, manoj.iyer at canonical.com wrote:
> >
> > > Maverick will be a lot closer to mainline kernel compared to lucid.
> > >
> > > On Tue, 6 Jul 2010, Robert P. J. Day wrote:
> > >
> > > > On Tue, 6 Jul 2010, manoj.iyer at canonical.com wrote:
> > > >
> > > >> You could diff the config files to see what you are missing.
> > > >>
> > > >> Also, look here for help on building Ubuntu kernels:
> > > >> https://wiki.ubuntu.com/Kernel/Dev
> > > >> http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/
> > > >
> > > >  ok, i finally have a kernel that boots off of my LVM root fs, by
> > > > following the instructions here:
> > > >
> > > >  https://wiki.ubuntu.com/KernelTeam/GitKernelBuild
> > > >
> > > > the section, "Using Ubuntu Kernel Configuration," so i'm not going to
> > > > mess with success, except to ask, could i use the ubuntu-maverick git
> > > > tree instead of ubuntu-lucid?  thanks.
> >
> >   confirmation that something worked, and a question.
> >
> >   first, i followed the instructions at
> > https://wiki.ubuntu.com/KernelTeam/GitKernelBuild for building a
> > fairly recent kernel and it worked, i got my
> > "custom"-versioned packages, installed them, booted and came up under
> > my new kernel.  excellent.  that's the confirmation.
> >
> >   i then tried the same thing using the "ubuntu-maverick" git pull
> > instead and, got all the way to the step of compiling the new kernel
> > and, very soon into the process (after having selected a version
> > string of "-rday" instead of "-custom") got:
> >
> >
> > ====== making target debian/stamp/build/kernel [new prereqs: vars]======
> > This is kernel package version 12.032.
> > The changelog says we are creating 2.6.35-rc4-rday
> > However, I thought the version is 2.6.35-rc4-rday+
> > exit 1
> > make: *** [debian/stamp/build/kernel] Error 1
> >
> >
> >   i'm used to seeing that "+" suffix as a version control identifier
> > (or something like that).  so ... is there a reason that using the
> > lucid tree would work fine, but the maverick tree produces the above?
> >
> >   the lucid tree is probably just fine for my purposes, now i'm just
> > curious.  and i clearly have much reading to do.
> >
>
> That is a bug in the upstream kernel code, I have a patch in -rc4 which
> should mitigate that.

  ah, thanks, i was starting to go a bit nuts trying to figure out
what i did wrong.  i'm assuming it has to do with this snippet from
scripts/setlocalversion:

                        # running further git commands
                        if $short; then
                                echo "+"
                                return
                        fi

which tacks on a "+" based on whether this is a tagged commit, but
that's not taken into account elsewhere?  or something like that.

rday

-- 

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA

        Top-notch, inexpensive online Linux/OSS/kernel courses
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================