[Qemu-devel] Re: requirements of root filesystem image when passed to qemu

[Peter Teoh]

On Fri, Jun 13, 2008 at 1:14 AM, Peter Teoh <address@hidden> wrote:
> First, after installing debian (8.04) on a qemu image - debian.cow - I
> then used it to mount the latest kernel:
>
> qemu -kernel /boot/vmlinuz-2.6.26-rc5 -hda debian.cow -append
> "root=/dev/hda console=ttyS0" -nographic
>
> [    5.493332] 0300    1331200 hda driver: ide-disk
> [    5.494978]   0301     248976 hda1
> [    5.496151]   0302    1076355 hda2
> [    5.497403] No filesystem could mount root, tried:  ext3 ext2 iso9660
> [    5.499552] Kernel panic - not syncing: VFS: Unable to mount root
> fs on unknown-block(3,0)
>
> But when I downloaded the root filesystem image at:
>
> http://bellard.org/qemu/download.html
>
> for linux-0.2.img, and execute:
>
> qemu -kernel /boot/vmlinuz-2.6.26-rc5 -hda linux-0.2.img -append
> "root=/dev/hda console=ttyS0" -nographic
>
> everything went through to the root prompt level, even though the
> linux-0.2.img image has its own kernel version (0.2).   And I noticed
> that (using "file"):
>
> linux-0.2.img: x86 boot sector, LInux i386 boot LOader
> debian.cow: QEMU Copy-On-Write disk image version 1, size 0 + 1363148800
>
> Next, using another rootfs image:
>
> file FedoraCore5-x86-root_fs
> FedoraCore5-x86-root_fs: x86 boot sector
>
> I run qemu successfully.
>
> Next I entered ubuntu installation image:
>
> ubuntu-8.04-desktop-i386.iso: ISO 9660 CD-ROM filesystem data UDF
> filesystem data (unknown version, id 'NSR01') 'Ubuntu 8.04 i386
>       ' (bootable)
>
> qemu -kernel /boot/vmlinuz-2.6.26-rc5 -hda
> ubuntu-8.04-desktop-i386.iso -append "root=/dev/hda console=ttyS0"
> -nographic
>
> And the result is:
>
> [    6.136373] Unable to load NLS charset utf8
> [    6.144019] VFS: Mounted root (iso9660 filesystem) readonly.
> [    6.144826] debug: unmapping init memory c0795000..c07d4000
> [    6.145575] Write protecting the kernel text: 2352k
> [    6.146417] Write protecting the kernel read-only data: 1068k
> [    6.147388] Warning: unable to open an initial console.
> [    6.148323] Kernel panic - not syncing: No init found.  Try passing
> init= option to kernel.
>
> So what is the requirement for -hda then?   1.   presence of
> bootsector image 2.   presence of /sbin/init???
>
> Next I created the busybox rootfs image following this instruction:
>
> http://dev.gentoo.org/~solar/busybox/bbemu
>
> And run qemu as before, just substituting only the -hda option, it result in:
>
> [    5.564894] debug: unmapping init memory c0795000..c07d4000
> [    5.565539] Write protecting the kernel text: 2352k
> [    5.565891] Write protecting the kernel read-only data: 1068k
> [    5.656106] mount used greatest stack depth: 2240 bytes left
> Bummer, could not run '/bin/sh': No such file or directory
> Bummer, could not run '/bin/sh': No such file or directory
> Bummer, could not run '/etc/init.d/rcS': No such file or directory
> Bummer, could not run '/bin/sh': No such file or directory
> Bummer, could not run '/bin/sh': No such file or directory
> Bummer, could not run '/bin/sh': No such file or directory
>
> Endless repetition here..
>
> Why?
>
> Why the difference?   (ie, why can't I mount the latest kernel on the
> debian image)  How to resolve the error as shown above?
>
> --

That was the last stage of booting - /sbin/init was executed, and
endless repetition is due to the inittab's respawn mechanism.   and
instead of /bin/sh, which not sure was not created by bbemu, i changed
it to /bin/msh, and it goes into a prompt.

Thanks.



-- 
Regards,
Peter Teoh