[debian] Re: How stupid I was to choose vfat for /boot

jidanni at jidanni.org jidanni at jidanni.org
Wed Nov 19 22:55:08 CET 2008


[Fixed addresses in header.]
>>>>> "JB" == Joachim Breitner <nomeata at debian.org> writes:

JB> Am Dienstag, den 18.11.2008, 09:54 +0800 schrieb jidanni at jidanni.org:
>> Nonetheless,
>> $ strings env.orig
>> bootargs_base=rootfstype=jffs2 root=/dev/mtdblock6 console=ttySAC2,115200 console=tty0 loglevel=8 regular_boot
>> bootcmd=setenv bootargs ${bootargs_base} ${mtdparts} rootfstype=ext2 root=/dev/mmcblk0p2 rootdelay=5; mmcinit; ext2load mmc 1 0x32000000 ${sd_image_name}; bootm 0x32000000
JB> [..]

>> Which is different than the definitive:
>> 
>> GTA02v6 # printenv
>> boot_menu_timeout=60
>> bootargs_base=rootfstype=jffs2 root=/dev/mtdblock6 console=ttySAC2,115200 console=tty0 loglevel=4 regular_boot
>> bootcmd=setenv bootargs ${bootargs_base} ${mtdparts}; nand read.e 0x32000000 kernel 0x200000; bootm 0x32000000

>> Moral of story: GTA02v6 is different. Anyways, glory be, I haven't
>> "bricked" it yet.

JB> This is really strage: When you download the environment from the
JB> device, it contains the ext2 modifications, but when you print it in
JB> uboot, it does not? Is that from NAND or form NOR uboot?

The printenv must be from NOR, as I got to it via the NOR button
presses mentioned on http://wiki.openmoko.org/wiki/Qi

JB> But no idea really what’s wrong here.

So dfu-util -a u-boot_env -U env.orig
must be talking to NAND. Yes I recall reading on the wiki that there
was a difference between NAND and NOR. And NOR is where I usually end
up because that's what the buttons I push are, according to
http://wiki.openmoko.org/wiki/Qi .

Wait, man dfu-util says
       -U, --upload FILE
              Read firmware from device into FILE.

              Note: Upload support is currently broken.

So of course the -U info must be wacko.

>>>>> "WA" == Werner Almesberger <werner at openmoko.org> writes:

WA> Andy and I have pretty much given up on u-boot and have switched
WA> to Qi or are in the process of doing so, but Harald may still be
WA> interested in fixing this kind of issues in u-boot.

I see. I would be dumb to continue trying to figure out uboot.

OK, could at least on other person on smartphones-userland install Qi
following the instructions on http://wiki.openmoko.org/wiki/Qi , and
report back that I will not brick my machine.

WA> Also, please make sure you use the latest u-boot. There have been
WA> some fixes about a month ago.

Never mind. I will try Qi like you said.

So I will reformat my first partition to ext2, put back the kernel
there, (copied to uImage.bin).
Wait, if anything goes wrong:
  • The current (Oct. 12, 2008) 2.6.27 kernel does not boot from the build-in flash NAND. It only works from the uSD boot
    on ext3. If one tries to boot it from the flash, then AUX red LED blinks ~4/sec. This means kernel panic...

This might mean that even with pushing the buttons, there may be no
way for me to boot the kernel the Openmoko team installed on NAND when
I spent the whole day with them in Taipei in 9/2008 (engineers  now do
not have time to talk to visitors (who are not buying anything?) as of
10/2008)

So maybe if I mess up my uSD card, I will not be able to start the
machine again.

Also if Qi turns into a disaster, the http://wiki.openmoko.org/wiki/Qi
webpage should mention how to restore uboot.
It should say to go to
http://wiki.openmoko.org/wiki/Uboot#Bootloader_binary
and then download the right version, and then do what?
Maybe something like dfu-util -a u-boot -R -D ...



More information about the openmoko-kernel mailing list