Hoefle Marco wrote: > Right now I got the emdebian-arm.tgz which has a size of 25 MByte > uncompressed which means it will fit into the flash of the XScale PXA270 > board. > In the emdebian how to it is mentioned that a debian like system is required > to update the romfs. This is not the case for me. We have uboot running > together > with this kernel version: > > Hit any key to stop autoboot: 0 > > **** Try to boot from Flash **** > > ## Booting image at 00040000 ... > > Image Name: Linux Kernel Image > > Image Type: ARM Linux Kernel Image (uncompressed) > > Data Size: 2367256 Bytes = 2.3 MB > > Load Address: a0008000 > > Entry Point: a0008000 > > Verifying Checksum ... OK > > OK > > > > Starting kernel ... > > > > Linux version 2.6.22.18-ELinOS-469 (j...@hrwks7001-l) (gcc version 3.4.4 > (ELinO8 > CPU: XScale-PXA270 [69054117] revision 7 (ARMv5TE), cr=0000397f > > > The flash is partitioned into: > 0x00000000-0x00040000 : "Bootloader" > > 0x00040000-0x00400000 : "Kernel" > > 0x00400000-0x02000000 : "Filesystem" > > I suppose it won't work to replace the data on 0x00400000 with the packed > .tgz data. Do you have any hints on how to proceed? > Thanks, > Marco > > Hi Marco,
does your board have a network interface? If so I would suggest that you use a NFS mounted root filesystem for your first attempts and initial debugging. That is easier than writing it to flash as you can just modify files on your desktop / laptop and the changes are "seen" by the board. For this to work you'll need NFS compiled into your kernel. You may need to reconfigure and recompile your kernel if this is not the case. To do this: 1) aptitude install nfs-kernel-server 2) Unpack the emdebian-arm.tgz somewhere (I'm assuming /home/marco/nfsroot for the rest). You'll need to do this as root so it can create device nodes etc. 3) Edit /etc/exports and add a line like: /home/marco/nfsroot (rw,no_root_squash) 4) Restart the nfs server (invoke-rc.d nfs-kernel-server restart) [as root] 5) Set your board's kernel command line to something like root=/dev/nfs ip=_targetIp_:_hostIp_:_gateway_:_netmask_ nfsroot=/home/marco/nfsroot where: _targetIp_ = ip address for your board _hostIp_ = ip address of your desktop / laptop A more detailed explanation of all this can be found at http://www.faqs.org/docs/Linux-mini/NFS-Root.html Regarding actually flashing the root filesystem (once it works with NFS): As you suspect you can't just put the tgz into flash - you need to create a filesystem on the flash. For a good overview of embedded filesystems take a look at http://free-electrons.com/docs/filesystems/ I would probably use jffs2 but ubi looks good (though I haven't tried it and its only available from 2.6.27). The basic idea is that on your desktop you: 1) create a file that will contain the image 2) use losetup to create a block device refering to that file 3) use the block2mtd module to simulate a mtd device on top of the above 4) mount jffs on this simulated mtd 5) copy your root filesystem onto it [see page 18 of the document above for the actual commands] Once you've done all that the image file will contain what you have to write to flash. You'll have to use uboot to flash it (for example after transfering by tftp). Another way is to boot the board with a NFS root filesystem and then write the flash directly from linux on the device (page 17) Hope this helps a little btw the free-electrons site I referenced above has lots of useful info for embedded linux. Martin -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
