SamyGO

eee, u right
and

needs a rebuild.


Denny

Isn`t a time to flash full -console enabled kernel?

juuso wrote:Isn`t a time to flash full -console enabled kernel?

Why not?

1. First, try to patch the kernel in memory to make sure it works as needed: passes all characters from serial input.
2. Dump 32-64 bytes near the patch point.
3. Look into dumps from /dev/bml0/5 or /dev/bml0/7 (firmware partitions with kernel image - find the active one - running).
4. Find the same byte sequence, make sure that it is unique, and change the same bytes as you changed in memory.
5. Calculate new hash for patched kernel and write it to the correct place in /dev/bml0/9 or /dev/bml0/10.
6. Copy all other parts: mtd_exe, mtd_appdata and rootfs, because you will need to flash everything into currently inactive partition.
7. Check hashes.
8. Only when you are sure that everything is OK, start flashing.
9. Check that everything was flashed correctly including hashes.
10. Toggle partitions and reboot.

Steps 3-7 should be done on a PC. Did I forget something?

welcome back ,

wortex u want to try now something ?
i have an good idea


Denny

off course he wants

yeah , sure, this need some time anyway the best think is at end of script (rc.local) to make again swap of parition let say, short before starting exeDSP , if this change goes bad.

ok, it is crazy idea but give us maybee posibility to do rw in mtd_exe in runtime and we must be carefull with releases > 3009.x

folow :

in /etc/rc.local @ line 69 > mount -t rfs -r $EXE_MOUNT /mtd_exe/ to remove -r maybee adding -o codepage=utf8
in cmac calculation, write just 0x10 bytes as file len and calculate for this len cmac , store it in need /dev/bml
this will give always same value as we just hash only 16 bytes of the stl device and authuld will eat this 16 bytes as authuld can also not know how big is realy fat file . means, we can modify mtd_exe anytime.

the same can i am 99.9999% sure for other squashfd paritions be done.

what do you think ?


Denny