[ARMedslack] sheevaplug: flashing initrd, consequently no uboot upgrade required

[Thierry MERLE]

On Sat, 19 Sep 2009 17:24:44 +0200
Thierry MERLE <thierry.merle at free.fr> wrote:

> Hello,
> OK succeeded in flashing an initrd instead of the rootfs, and obviously booting on it.
> After some SP brickings (I even erased uboot by mistake, and I will never ever do that again), I managed to put initrd on flash and boot with it.
> In fact you should follow these directives: http://www.openplug.org/plugwiki/index.php/Flashing_an_Operating_System_onto_the_Plug
> But with some modifications:
> flash_erase /dev/mtdblock1
> flash_eraseall /dev/mtd1
> flash_erase /dev/mtdblock2
> flash_eraseall /dev/mtd2
> cat uImage-kirkwood > /dev/mtdblock1
> cat uinitrd-kirkwood > /dev/mtdblock2
> uImage-kirwood and uinitrd-kirkwood are found in armedslack-12.2 distribution.
> 
> Then reboot and stop uboot in its automatic process with [ENTER]
> setenv bootcmd 'nand read.e 0x800000 0x100000 0x400000; nand read.e 0x01100000 0x500000 0x500000'
sorry, the complete command is:
setenv bootcmd 'nand read.e 0x800000 0x100000 0x400000; nand read.e 0x01100000 0x500000 0x500000; bootm 0x800000 0x01100000'
in order to boot automatically after loading into RAM.

> #This will load in RAM the kernel, and the initrd image (5MB)
> setenv bootargs 'console=ttyS0,115200 ro root=/dev/sda1'
> #My USB disk has a root FS on sda1, in fact this is my NSLU2 installation.
> setenv arcNumber 2097
> setenv mainlineLinux yes
> saveenv
> boot
> 
> And you are done.
> I am not sure the flash_erase* commands are necessary but I experienced so many write errors that using them is the safest way to have something working.
> 
> Well, I lied a bit, I did not flash the armedslack kernel but rebuilt a 2.6.31 kernel with the few patches to make it work on the SP (http://sheeva.with-linux.com/sheeva/index.php?dir=2.6.31%2Fpatches%2F)
> I had kernel panics with the armedslack kernel, specifically when I
> tried to configure Ethernet. I removed the kernel module from the initrd (mv643xx_eth)
> so the detection is made after, and no more kernel panic. This is
> probably a Thierry's problem.
> 
> Cheers,
> Thierry
> _______________________________________________
> ARMedslack mailing list
> ARMedslack at lists.armedslack.org
> http://lists.armedslack.org/mailman/listinfo/armedslack


-- 
                   ,
                   }\      °
           _    .-`  `-.  o
           \`'./     (o)\
            ) >   )))    } O
           /_.'\     / \/
       jgs      `'---'=`     http://thierry.merle.free.fr/