Re: linux-2.6.31.6 doesn't boot

2009-11-29 Thread Bruce Dubbs
Alberto Hernando wrote:
 Hi.
 
 I've tried it, and:
 
 root:/boot/grub# grub-setup -r '(hd0,8)'
 No device is specified.
 
 The same. I've also tried several things using slashes and still the same.
 Is there really something wrong with using lilo? I mean, how can I be sure
 that grub will be able to boot a kernel that lilo can't? I'm more used to
 use lilo than grub.

Certainly you can use lilo if you want.  You can also use almost any 
distro on a CD and have it install some version of grub.  You can build 
the 32-bit version of grub on a 32-bit OS and install that.  It's your 
distro.

LFS is about learning and in this case we are learning about grub.

Now that I look at it, I remember a message about problems with 
installing grub2 on a partition.  What I think you should be doing is 
installing it on the MBR.

Try

   grub-setup -r '(hd0)'

   -- Bruce

-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: linux-2.6.31.6 doesn't boot

2009-11-29 Thread Ken Moffat
2009/11/29 Alberto Hernando pajaro...@gmail.com:
 Hi.

 I'm building LFS-6.5 (-svn, actually), and I can't make it boot. I've done
 the same in two machines, an old athlon amd and a pentium-iv. Kernel builds
 fine in both systems, but none of them can boot. The error is the same, VFS
 can't find a valid root system, please add a valid root option. I'm using
 lilo. I've made sure that ext3 is built inside the kernel, and I'm using old
 configs from 2.6.26 and 2.6.28. I think both configs are ok, and /etc/fstab
 is as the book says. I can't find the mistake.

At some point (I thought it was *before* 2.6.28, but maybe I'm mistaken)
the kernel option for IDE disks changed (as part of the move to libata).
Unfortunately, all my systems use SATA drives, so I'm not entirely sure
which option is relevant.

On my current kernel (2.6.32-rc) I have
CONFIG_ATA=y
CONFIG_ATA_SFF=y # I don't recall what that does
CONFIG_SATA_VIA=y (the driver I'm using).

If you have a parallel ATA disk (40 or 80-way cable)
you'll need one or more of the CONFIG_PATA selections
to provide the driver.

This is one of the reasons I prefer to build a kernel on an
existing installed system first - it lets me confirm that what
I have chosen in the .config will let me boot (or not!).

Also, with  libata drivers you will need to specify /dev/sdXn
in /etc/fstab, instead of /dev/hdXn - for an older system,
mounting by label [man e2label] is helpful for the change.

Meanwhile, lilo is unlikely to be the problem (it still works for
me).

ĸen
-- 
After tragedy, and farce, OMG poneys!
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: build problem LFS 6.5

2009-11-29 Thread Baho Utot
linux fan wrote:
 On 11/28/09, Baho Utot wrote:

   
 make test | tee $CURRDIR/06.58.Check.log
 make install | tee $CURRDIR/06.58.Install.log
 
 Logging helps so much when things go wrong.

 As already mentioned

   
 Scripting an LFS build is *hard*
 

 [putolin]
 I only suggest this due to amount of time spent and it failing.
 You can debrief what jhalfs did with its scripts in
 lfs-commands/chapterXX/* and maybe detect what went wrong in handmade script.
   
hmmm my scripts build LFS correctly every time. Spent zero time with 
them failing.
But then I am not the OP.

-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: errors at 6.7. Linux-2.6.30.2 API Headers

2009-11-29 Thread knothead00
 I assume you mean Chapter 5.6, but we don't really reference the pdf
 pages.

sorry --  i meant when i try to install (in section) 6.7.
Linux-2.6.30.2 API Headers
i get the errors.


 root:/sources/linux-2.6.30.2# make headers_check
   CHK     include/linux/version.h
   UPD     include/linux/version.h
   HOSTCC  scripts/basic/fixdep
 scripts/basic/fixdep.c: In function 'traps':
 scripts/basic/fixdep.c:377: warning: dereferencing type-punned pointer
 will break strict-aliasing rules
 scripts/basic/fixdep.c:379: warning: dereferencing type-punned pointer
 will break strict-aliasing rules
 gcc: error trying to exec 'as': execvp: No such file or directory

 Well, you don't have the full binutils installed.  It can't find the
 assembler, as.


 What i decided to do is revert to the snapshot with my vmware
machine.  This snapshot goes back to the finish of 5.10. GCC-4.4.1 -
Pass 2 with successful testing of the compiling and linking of the
toolchain.

Any suggestion to verify a successful install of binutils to prevent this again?

Thanks.
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: errors at 6.7. Linux-2.6.30.2 API Headers

2009-11-29 Thread Bruce Dubbs
knothea...@gmail.com wrote:
 I assume you mean Chapter 5.6, but we don't really reference the pdf
 pages.
 
 sorry --  i meant when i try to install (in section) 6.7.
 Linux-2.6.30.2 API Headers
 i get the errors.

 root:/sources/linux-2.6.30.2# make headers_check
   CHK include/linux/version.h
   UPD include/linux/version.h
   HOSTCC  scripts/basic/fixdep
 scripts/basic/fixdep.c: In function 'traps':
 scripts/basic/fixdep.c:377: warning: dereferencing type-punned pointer
 will break strict-aliasing rules
 scripts/basic/fixdep.c:379: warning: dereferencing type-punned pointer
 will break strict-aliasing rules
 gcc: error trying to exec 'as': execvp: No such file or directory
 Well, you don't have the full binutils installed.  It can't find the
 assembler, as.
 
 
  What i decided to do is revert to the snapshot with my vmware
 machine.  This snapshot goes back to the finish of 5.10. GCC-4.4.1 -
 Pass 2 with successful testing of the compiling and linking of the
 toolchain.
 
 Any suggestion to verify a successful install of binutils to prevent this 
 again?

There are several ways, but I think the easiest is to check /usr/bin and 
/usr/lib for the programs and libraries specified at

6.12.2. Contents of Binutils

   -- Bruce
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: linux-2.6.31.6 doesn't boot

2009-11-29 Thread linux fan
On 11/29/09, Alberto Hernando wrote:
 ...  The error is the same, VFS
 can't find a valid root system, please add a valid root option. I
 .. and /etc/fstab is as the book says.

If /etc/fstab is as the book says (verbatim), it won't work.
You must interpolate
/dev/xxx /fff  defaults1 1
replacing xxx and fff respectively.
Your other post suggests that it would be:
/dev/hda8/ext3  defaults1 1

The VFS can't find ... suggests that the boot loader (grub/lilo)
done its job and the kernel was loading until it choked on the
filesystem fstab told it,

I think
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: linux-2.6.31.6 doesn't boot

2009-11-29 Thread Bruce Dubbs
Alberto Hernando wrote:
 Hi.
 
 I've tried it, and:
 
 root:/boot/grub# grub-setup -r '(hd0,8)'
 No device is specified.

I gave incorrect advice before.  Try:

   grub-setup '(hd0)'

or

   grub-setup /dev/sda

The -r should not be there.  The syntax is:

   grub-setup [OPTION]... DEVICE

If you use -r, then that is an option and DEVICE is still required.
For instance, look at the entry:

menuentry GNU/Linux, Linux 2.6.30.2-lfs65 {
 insmod ext2
 set root=(hd0,1)
 linux   /linux-2.6.30.2-lfs65 root=/dev/sda5 ro
}

In this example, the root partition is /dev/sda1 and is mounted as /boot.

If no 'set root' instruction is specified, it will use the default.  The 
default is set by the -r option or if not specified at all it will be 
guessed.   How you figure this all out is not really obvious.  I 
couldn't find any documentation about it, so I read the code.  :)

Note that the term root is overloaded and the 'set root' refers to the 
partition where grub should search for files which is something quite 
different from the root= parameter on the line starting with 'linux' 
which is the partition that the linux kernel mounts as /.

However both these 'root's can point to the same place.  It would look
something like:

menuentry GNU/Linux, Linux 2.6.30.2-lfs65 {
 insmod ext2
 set root=(hd0,5)
 linux   /boot/linux-2.6.30.2-lfs65 root=/dev/sda5 ro
}

This is the case of a standard LFS build where /boot is just another 
directory in /.  As I wrote in the book, Using the current lfs 
partition will also work, but configuration for multiple systems is more 
difficult.

   -- Bruce
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page


Re: linux-2.6.31.6 doesn't boot

2009-11-29 Thread linux fan
On 11/29/09, linux fan wrote:

 The VFS can't find ... suggests that the boot loader (grub/lilo)
 done its job and the kernel was loading until it choked on the
 filesystem fstab told it,

If the message is exactly:
Kernel panic -not syncing VFS: unable to mount root fs on unknown-block(0,0)

Then it is the root= parameter on the linux/kernel line that is wrong
-- while the kernel is loading, it gets to the point where it wants to
use that parameter, but it is bogus so there is nothing else to do but
panic.

I just tried it accidentally and it gave that message.
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page