EMMC Dumps with different hashes

[chrizzler]

Hey guys,

because my MCU has a blackscreen, i wanted to swap my emmc.
So I tried to dump an image of my emmc, which gave me 0 errors with ddrescue.

I dumped the image several times, but each dump has a different hash.

When I try to make an dump of each partition, I always get the same hashes for partition 1 and 3, but never for 2 and 4.

Would really appreciate, if someone could give me advice

 

[chrizzler]

Btw: All partitions of the emmc were mountable with my usb allsocket159+169

 

[DOCAL]

Your sequence wasn't clear, but mounting some filesystems will cause minor changes in the metadata.

I'd suggest taking multiple dumps back to back without mounting anything, then do a binary diff between them and see what the differences are. If you see that some blocks that differ are all zero, those could just be bad reads.

Out of interest, have you looked to see what the life span counter shows for the emmc (if the ones that Tesla use support that).
Under linux the mmc command would let you query that. Something like...

mmc extcsd read /dev/mmcblk0

Results will probably include 2 types, A (SLC) and B (MLC), numbers are in 10s of percent used. So 0x1 = 0-10% used, 2 = 11-20% used.

 

[chrizzler]

So I did a comparism of my dumped images with cmp and also compared the images with a Hexeditor on Windows.
They differ at many points and the differences aren't just zeros, so no bad reads i guess.

Unfortunately i can't run mmc commands to read because my allsockets shows the emmc as /dev/sd*.