Actually, ixtab's instructions seem to agree with GM's getkernels script (at least on my Touch, I'll check on a PW2).
Beware of the blocksize! GM (and Amazon's tools, IIRC) uses a blocksize of 1024, while ixtab's instructions use 4096. Every offset *should* still be a multiple of 1024, but take that into account in your maths
.
EDIT (on a PW rnning 5.4.5.1):
Code:
dd if=/dev/mmcblk0 of=main_kernel.img bs=1024 skip=260 count=2370
(260/4 == 65, for other maths-impaired individuals like me
)
And the only difference between what getkernels dumps and what FW 5.4.5.1 bundles is a bunch of extra crap bytes at the end:
uImage:
Code:
002504f0 cc 04 25 00 00 00 00 00 00 00 00 00 00 00 00 00 |..%.............|
00250500 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff |................|
*
00260000
main_kernel.img:
Code:
002504f0 cc 04 25 00 00 00 00 00 00 00 00 00 00 00 00 00 |..%.............|
00250500 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff |................|
*
00250800
Code:
File: ‘uImage’
Size: 2490368 Blocks: 4864 IO Block: 4096 regular file
Device: fc02h/64514d Inode: 2813702 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 1000/ niluje) Gid: ( 1000/ niluje)
Access: 2014-08-12 00:00:11.000000000 +0200
Modify: 2014-08-05 14:01:33.000000000 +0200
Change: 2014-08-12 00:00:11.152355325 +0200
Birth: -
File: ‘main_kernel.img’
Size: 2426880 Blocks: 4744 IO Block: 4096 regular file
Device: fc02h/64514d Inode: 2821612 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 1000/ niluje) Gid: ( 1000/ niluje)
Access: 2014-11-30 17:45:04.637692139 +0100
Modify: 2014-11-30 17:45:04.829689738 +0100
Change: 2014-11-30 17:45:04.829689738 +0100
Birth: -
----
So, yup, I'd stay ixtab's instructions are still correct
. Did I get that right, @knc1?