Using the debian rootfs with the development vm

[ukscone]

i just grabbed the debian rootfs and installed it in the development vm so that it can be used with scratchbox2.

there are simpler ways (use the mount offset option) to do this but this is the way I am used to.

sudo losetup -o 80740352 /dev/loop0 debian6-17-02-2012.img

sudo mkdir /mnt/debrootfs

sudo mount /dev/loop0 /media/debrootfs

then use your favourite method to copy whole directory trees of files from one place to another to copy all the files and dirs in debrootfs to a directory in $HOME/raspberry_pi_development

say raspi_debian

and then

cd $HOME/raspberry_pi_development/

sudo chown -R raspberry raspi_debian/*

sudo chgrp -R raspberry raspi_debian/*

sudo chmod -R 777  raspi_debian/*

cd raspi_debian

sb2-init raspberry_deb $HOME/raspberry_pi_development/arm-2011.03/bin/arm-none-linux-gnueabi-gcc

and that's pretty much it.

[ukscone]

oh and don't worry your little heads about where I got the 80740352 from it's just magic

Looks great mate

Will have to try that out.

Where did you dl from in the end?

[ukscone]

grabbed it using bittorrent in my windows host and setup a shared folder between the vm and windohs!

oh and don"t worry your little heads about where I got the 80740352 from it"s just magic

It"s the sector in the image where file system starts times the sector size (157696*512)

[ukscone]

Thorium said:

oh and don"t worry your little heads about where I got the 80740352 from it"s just magic

It"s the sector in the image where file system starts times the sector size (157696*512)

correct



and another interesting offset is 1048576 which is the offset to the fat partition

ukscone said:

i just grabbed the debian rootfs and installed it in the development vm so that it can be used with scratchbox2.

there are simpler ways (use the mount offset option) to do this but this is the way I am used to.

sudo losetup -o 80740352 /dev/loop0 debian6-17-02-2012.img

sudo mkdir /mnt/debrootfs

sudo mount /dev/loop0 /media/debrootfs

then use your favourite method to copy whole directory trees of files from one place to another to copy all the files and dirs in debrootfs to a directory in $HOME/raspberry_pi_development

say raspi_debian

and then

cd $HOME/raspberry_pi_development/

sudo chown -R raspberry raspi_debian/*

sudo chgrp -R raspberry raspi_debian/*

sudo chmod -R 777  raspi_debian/*

cd raspi_debian

sb2-init raspberry_deb $HOME/raspberry_pi_development/arm-2011.03/bin/arm-none-linux-gnueabi-gcc

and that's pretty much it.

I've not used virtual machines in Linux before so this is a bit new to me. I have Lubuntu 11.10.

On the third line

     sudo mount /dev/loop0 /media/debrootfs

I get

     mount: mount point /media/debrootfs does not exist

Any idea what I am doing wrong?

Hi grumpyoldgit,

You mount media into a directory. If the directory /media/debrootfs doesn't exist then you'll probably get that message above.

You should just need to do a 'mkdir /media/debrootfs/'

Hope this helps,

Liam.

The more simple way to mount the image, as mentioned by ukscone:

($ sudo mkdir /media/debrootfs)

($ sudo losetup -d /dev/loop0)

$ sudo mount -o loop=/dev/loop0,offset=80740352 debian6-17-02-2012.img /media/debrootfs/

The magic numbers (divided/multiplied by 512, the sector size) can be extracted from the image file using `fdisk`:

$ fdisk -l -u debian6-17-02-2012.img

*edit* trying again

If you want to extract the single partitions to use them with an ARM emulator (qemu) and not to mount them:

For the FAT-part (where the GPU, boot and kernel stuff is):

dd bs=512 if=debian6-17-02-2012.img of=debian6-17-02-2012-fat.img skip=2048 count=76800

For the ext4 (rootfs)-part:

dd bs=512 if=debian6-17-02-2012.img of=debian6-17-02-2012-rootfs.img skip=157696 count=1628160

jannis said:

If you want to extract the single partitions to use them with an ARM emulator (qemu) and not to mount them:

dd bs=512 if=debian6-17-02-2012.img of=debian6-17-02-2012-rootfs.img skip=157696 count=1628160

You do not actually need to extract the partitions for qemu, since the operating system understands partitions just as it would on real hardware.  In fact, it may be better not to extract the partitions, so that you can easily access the swap partition, and can return the image to an SD card later.

If you do wish to extract the partitions it is worth noting that they are all exact multiples of 1MiB in both size and position, as is the modern practice.  This allows a much more efficient block size on the 'dd's:

dd if=debian6-17-02-2012.img of=d6boot.raw bs=1M skip=1 count=75
dd if=debian6-17-02-2012.img of=d6root.raw bs=1M skip=77 count=1590
dd if=debian6-17-02-2012.img of=d6swap.raw bs=1M skip=1668 count=191

jojopi said:

You do not actually need to extract the partitions for qemu, since the operating system understands partitions just as it would on real hardware.  In fact, it may be better not to extract the partitions, so that you can easily access the swap partition, and can return the image to an SD card later.

True, you don't have to extract the partitions. But you can It was rather meant for reference.

If you do wish to extract the partitions it is worth noting that they are all exact multiples of 1MiB in both size and position, as is the modern practice.  This allows a much more efficient block size on the 'dd's:

dd if=debian6-17-02-2012.img of=d6boot.raw bs=1M skip=1 count=75
dd if=debian6-17-02-2012.img of=d6root.raw bs=1M skip=77 count=1590
dd if=debian6-17-02-2012.img of=d6swap.raw bs=1M skip=1668 count=191

Didn't notice this, though. Thanks for pointing it out