Unbriking Kindle DX

[xms]

Originally Posted by hawhill

That said, I'm not sure how the filesystems are laid out on the KDX b004. I guess from the sizes that sdx1 is rootfs, sdx2 is /var/... and the third one is the user store.

Yes it is so, sdx2 == /var/local

Originally Posted by hawhill

From the logs you have posted, the initramfs can not mount the root file system. So don't concentrate on the rest of the partitions yet, but rather check the root file system. Can you mount it on the host machine?

Yes i can see all partitions both in windows and linux,
but linux can't recognize it as ext3 as i understand (mount is not working)..

Originally Posted by hawhill

Before messing with it, I would suggest to make a full dump of everything, i.e. the full block device (not just single partitions). And try to run e2fsck on the root file system then.

Thanks for advice!

[knc1]

Originally Posted by xms

p.s if you mean only about sdX name is wrong sure i will replace it with real name;
else i didn't understand clear what did you mean..

thanks

sd<anything> is the name of a device holding a file system - I.E: The filesystem layer above whatever the storage layer happens to be.

mmc<anything> is the name of an underlying storage layer.

The mmc<whateve>.image file is a copy of what is recorded on the bare storage media.

Under Linux, you can use the loop dev. to mount that image file, allowing you to access the file system that it represents.

[hawhill]

Well, in fact, when it comes to the user partition, it's a bit more difficult. That's because the user partition carries a partition header within it, like if the partition was another full disk. At least that is the case for my K3. They probably did that because Windows and other OS can cope better with "Hard Disks" (devices with a partition table) than with "Floppy Disks" (devices without a partition table, containing one filesystem that spans the disk). Plus, in the worst case the user will reformat it, maybe forcing it to get a partition table. This is because the Kindle presents this (internal) partition as a full disk externally via USB.

However, when the full MMC space is exportet as in this recovery mode, the user partition now suddenly carries the file system starting at a later offset.

All this isn't relevant for the OP, of course, since the error indicates clearly that the root file system is the culprit. Running an ext3 file system check should bring more information (e2fsck is the tool needed for that, it does ext3 despite its name).

[xms]

My Kindle restored!

It was needed just to overwrite rootfs with the dump (using "dd").

All like described here:
http://yifan.lu/2011/04/02/recoverin...rupt-kindle-2/

except i didn't use custom kernel because have password
(received by Kindle diagnostic tool: http://www.siralex.info/wp-content/p...nload.php?id=9)

also you can ask anyone with same model to do the image for you with just simple script

Spoiler:

#!/bin/sh

# diff OTA patch script

_FUNCTIONS=/etc/rc.d/functions
[ -f ${_FUNCTIONS} ] && . ${_FUNCTIONS}


MSG_SLLVL_D="debug"
MSG_SLLVL_I="info"
MSG_SLLVL_W="warn"
MSG_SLLVL_E="err"
MSG_SLLVL_C="crit"
MSG_SLNUM_D=0
MSG_SLNUM_I=1
MSG_SLNUM_W=2
MSG_SLNUM_E=3
MSG_SLNUM_C=4
MSG_CUR_LVL=/var/local/system/syslog_level

logmsg()
{
local _NVPAIRS
local _FREETEXT
local _MSG_SLLVL
local _MSG_SLNUM

_MSG_LEVEL=$1
_MSG_COMP=$2

{ [ $# -ge 4 ] && _NVPAIRS=$3 && shift ; }

_FREETEXT=$3

eval _MSG_SLLVL=\${MSG_SLLVL_$_MSG_LEVEL}
eval _MSG_SLNUM=\${MSG_SLNUM_$_MSG_LEVEL}

local _CURLVL

{ [ -f $MSG_CUR_LVL ] && _CURLVL=`cat $MSG_CUR_LVL` ; } || _CURLVL=1

if [ $_MSG_SLNUM -ge $_CURLVL ]; then
/usr/bin/logger -p local4.$_MSG_SLLVL -t "ota_install" "$_MSG_LEVEL def:$_MSG_COMP:$_NVPAIRS:$_FREETEXT"
fi

[ "$_MSG_LEVEL" != "D" ] && echo "ota_install: $_MSG_LEVEL def:$_MSG_COMP:$_NVPAIRS:$_FREETEXT"
}

if [ -z "${_PERCENT_COMPLETE}" ]; then
export _PERCENT_COMPLETE=0
fi

update_percent_complete()
{
_PERCENT_COMPLETE=$((${_PERCENT_COMPLETE} + $1))
update_progressbar ${_PERCENT_COMPLETE}
}

update_percent_complete 10

dd if=/dev/mmcblk0 of=/mnt/us/mmcblk0_7688.dmp bs=512 count=7688
update_percent_complete 20
dd if=/dev/mmcblk0p1 of=/mnt/us/mmcblk0p1.dmp
update_percent_complete 70
dd if=/dev/mmcblk0p2 of=/mnt/us/mmcblk0p2.dmp
update_percent_complete 90
dd if=/dev/mmcblk0p3 of=/mnt/us/mmcblk0p3.dmp
update_progressbar 100

return 0

packed into update.bin package with KindleTool (taken from this forum)
NB: jailbreak installed is needed!

p.s. seems it was possible to do the restore even without disassembling kindle and serial console, just packing rootfs.tar.gz into restore package..
(didn't test this)

Thanks for all advices!

[knc1]

Maybe GM will move that image link to the pastebin list for us before it gets reported to a moderator (direct links to files containing copyrighted materials are not allowed here).

[twobob]

Binned

[geekmaster]

Originally Posted by knc1

Maybe GM will move that image link to the pastebin list for us before it gets reported to a moderator (direct links to files containing copyrighted materials are not allowed here).

What link?

[knc1]

Originally Posted by geekmaster

What link?

My bad.
I did not check the contents of the posted link.

[equinox1]

Just FYI.
Recently I bricked my DX trying to update my old 2.1.1 firmware to something new. After few unsuccessful attempts to get rootfs.img for my DX (B004) on this forum I decided to go another way:

I downloaded torrent with 3.2.1 firmware hack for 2/DX, extracted rootfs.img and kernel using Kindletools and successfully loaded rootfs using dd. Then I flashed an uimage kernel from the same package (another update*.bin) using uboot console and Ymodem (prg_kernel_serial). After that I successfully started Kindle using bootm command.

So, if somebody has bricked Kindle 2/DX - probably 3.2.1 version is the best option. 3G and web browser works OK. The only problem with 3.2.1 is slow return from sleep mode (sometimes). Since now I don't use sleep at all, switching DX completely off (just holding slider for 5 sec).

PS: you will need a serial console to export MMC0 via USB and flash kernel via Y-Modem.

[geekmaster]

Originally Posted by equinox1

...
So, if somebody has bricked Kindle 2/DX and tired asking people to dump their DX - 3.2.1 version is your way...

There *IS* a DX firmware image set in the pastebin. But your information is useful as an alternative.

[cavorite]

My DXG appears to be stuck in a boot loop. I can interrupt the loop at two different points and either get into u-boot or the recovery console.
In the recovery console, my (correctly generated for my serial number) password is not accepted, so I cannot export MMC0 via USB.

If I flash an extracted 3.2.1 kernel through u-boot/Ymodem as described, will that let me use the recovery console (with my old password or ??)

If so, I think I could then complete the job by copying rootfs.img onto the DXG with dd.


Will this work, or is there a better way -- any help is appreciated!

[geekmaster]

As I understand it, the recovery console does not use the DES password hash, and may need the version with 4 hex digits. Try that and report back.

[cavorite]

Excellent suggestion but it's a moot point for me till I clear my new problem. Actually, this was evidently happening before I made my previous post, probably because I was working from older terminal capture files and was stuck on the password problem in my mind.

My DXG has bad info in MMCBLK0p1. This appears to be the cause of the boot loop and likely the cause of me now not being able to get into the recovery menu at all.

Correct boot shows this: (part 1 of 2)

/dev/mmcblk0p4:
CHS=4/16/109695 size=3594485760 bytes
flag type      first       last  lba first   lba size
Partition p1:
0x00 0x0b         16 <large>            16    7020464
    CHS: 0/1/1 - <large>
Partition p2:
Partition p3:
Partition p4:

My boot shows:

/dev/mmcblk0p4:
CHS=4/16/109695 size=3594485760 bytes
flag type      first       last  lba first   lba size
Partition p1:
0x00 0x0b         16 <large>            16    7020EXT3-fs: error loading journal.
464
    CHS: 0/1/1 - <large>
Partition p2:
Partition p3:
ParEXT3-fs: error loading journal.
Partition p4:

Correct boot: (part 2)

kinit: do_mounts
kinit: name_to_dev_t(/dev/mmcblk0p1) = mmcblk0p1(179,1)
kinit: root_dev = mmcblk0p1(179,1)
kinit: /dev/root appears to be a ext3 filesystem
kinit: trying to mount /dev/root on /root with type ext3
kinit: Mounted root (ext3 filesystem) readonly.

My boot:

kinit: do_mounts
kinit: name_to_dev_t(/dev/mmcblk0p1) = mmcblk0p1(179,1)
kinit: root_dev = mmcblk0p1(179,1)
kinit: /dev/root appears to be a ext3 filesystem
kinit: trying to mount /dev/root on /root with type ext3
kinit: failed to identify filesystem /dev/root, trying all
kinit: trying to mount /dev/root on /root with type ext3
kinit: trying to mount /dev/root on /root with type cramfs
kinit: trying to mount /dev/root on /root with type msdos
kinit: trying to mount /dev/root on /root with type vfat
kinit: Unable to mount root fs on device mmcblk0p1(179,1)

After this, the boot continues to go off the rails, eventually terminates and restarts. If I press enter to get the recovery menu it is ignored. Can still get into u-boot. Will flashing a replacement kernel correct my problem, or at least get me to a stage where I can correct it?

[NiLuJe]

@cavorite: That looks like a bad partition/FS/partition table, I don't think switching Kernel will magically fix your partition ;o).

[knc1]

You should be able to fix it from U-Boot.
What those messages mean is the kernel can not find the root file system to mount.
Kernel is doing exactly what it is intended to do - leave it alone.