Re: New kernel unable to mount/see a whole HD
On Mon, 2006-01-02 at 10:03 -0700, Paul E Condon wrote:
> On Mon, Jan 02, 2006 at 09:53:38AM -0600, Michael Martinell wrote:
> >
> > On Mon, January 2, 2006 9:39 am, J.F. Gratton wrote:
> > > (I must first start by apologizing if you've seen this post twice in
> > > 12hrs.. I've had problems here with my smtp; not sure it went well)
> > >
> > > Hello,
> > >
> > > I've been unable to access any partition on /dev/hda since I've compiled
> > > my own kernel 2.6.14.5 . I'm currently running 2.6.12-10.
> > >
> > > I want to get rid of initrd and compile my own kernels (I've done it for
> > > a long time, it's just that since Deb 3.0 rX I have been a bit lazy and
> > > let the OS install new ones when new ones were available).
> > >
> > > Now, on /dev/hda I have all my "foreign" OS partitions (ntfs and vfat
> > > filesystems, as well as UFS -for solaris).
> > >
> > > For the life of me, I can't see any pertinent changes between my two
> > > kernel configs (provided in attachments) that might give the slightest
> > > clue as to why kernel-2.6.12 (dpkg-provided) will mount /dev/hda* and
> > > why kernel-2.6.14.5 (user-compiled) won't.
> > >
> > > The console won't spew any error messages concerning the missing vfat
> > > partitions. If I manually try to mount them (say.. mount -t
> > > vfat /dev/hdb6 /mnt/temp-mountpoint), I get a "/dev/hda6 : device busy".
> > > So I guess it "knows" that /dev/hdb6 exists, right ?
> > >
> > > It's not a question of filesystems not being included in modules/kernel;
> > > there are vfat partitions on /dev/hdb and those are being seen and
> > > mounted. I insist on the fact that *everything* works just fine with
> > > 2.6.12 but not with 2.6.14.5.
> > >
> > > Anyhow.. I'm stumped, can't see why it won't go ok with 2.6.15.4, if
> > > anyone can help me out with the files attached, just go ahead, please :)
> > >
> > > Regards, and best wishes for 2k6 !
> > >
> > > -- Jeff
> > >
> > >
> > >
> >
> > This may seem a bit obvious, but did you try to copy config-currentkernel that
> > relates to your old install (from /boot), rename it to .config and put it into
> > your source path for the new kernel. That should give you the exact same
> > config that you are currently using. Then you could just run the configure
> > program (make menuconfig) for the kernel to select or unselect any other
> > features. Then save the new .config file. In this way you can verify that
> > your previously known good config is the one you are using. I have complied
> > lots of kernel's, however have only had problems when I try to create a
> > .config from scratch.
> >
>
> As an infrequent builder of special kernels, I have found it useful to insert
> another step into the above procedure. Before I make my changes to .config, I
> do a trial run of compiling. That is to say, I compile, as best I can, the same
> kernel as is given in the Debian distribution. If I can't do this, I know I am
> doing something wrong, independent of any changes I hope to make. At this stage
> I catch dumb mistakes that I frequently make when working from memory.
>
> HTH
>
> --
> Paul E Condon
> pecondon@mesanetworks.net
>
Well I tried your (Michael and Paul's) ways (which, btw used to be the
way I built kernels; somehow I forgot it this time; as I said it has
been a bit of time since I've built one.. Around v2.6.9, I think). It
did not work.
A bit of info I did not have -because I did not think of it- before:
cfdisk /dev/hda shows the drive and partitions. It just won't mount
them !! mount -t vfat /dev/hdaX will get me a "device busy". It _is_
annoying.
The only real differences between the distro-provided kernel and the one
I want to build is basically I want some amd64 specific tweaks, I want
to get rid of the initrd and prevent compiling of many useless modules;
nothing really esoteric per se.
I _am_ stumped...
-- Jeff
Reply to: