Different results each boot dumping spi

[bm16ton]

If i dump the spi and reboot dump spi again i get different results in files. Seems usually ill have large segments of all f's. In the few files i have it seems random so far so ill dump more and look for a pattern. Not sure if it applies or not but the machine does use recoveryAB and so my 32mb bios is dumped as 64mb file. If i split the files the two parts are always identical. Not sure if this all happens because of some security thing, or a glitch in the app. Oh one more piece, if i dont reboot i so far will get the exact same dump, and a warm reset seems to produce less f filled segments. Thank you for writing this software and the blog posts! Very helpful.

[n3k]

This could happen because of the non-volatile region. That said, there are 2 ways of extracting the flash: the memory mapped way or via spi-controller commands. The version that is in the CLI uses the memory mapped approach. There are examples in the PoCs on how to do it via spi-controller commands that you can use to double check.

[bm16ton]

Oooh im on it. I wasnt sure why it would very so much in time to rip the spi flash, and had a foolish idea that maybe other processes where somehow causing issues. So i made a bare initrd with the kernel module and userland app. And even after a reboot i was able to get identical files. This of course proves nothing and requires more testing, but i like your suggestion much better. Off topic but any resources on the PSB/PSP fuses you would suggest? My machine has firmware anti rollback enabled.The very little i can find where its mentioned, makes it seem if i set a PSB fuse(offset 0x10994,bit 25) and enable PSB (bit 24) i can have it ignore the FAR protection. But im unsure because its all mentioned in passing and sometimes using double negatives. With only one chance to get it right, its nerve wrecking. Also the disable secure debug ive seen with double negatives wondering if it makes secure debug available or locks it up. Sorry for the off topic banter. Just resources are so slim.

…

On Sat, Aug 24, 2024, 4:07 PM Enrique Elias Nissim ***@***.***> wrote: This could happen because of the non-volatile region. That said, there are 2 ways of extracting the flash: the memory mapped way or via spi-controller commands. The version that is in the CLI uses the memory mapped approach. There are examples in the PoCs on how to do it via spi-controller commands that you can use to double check. — Reply to this email directly, view it on GitHub <#14 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADAWMPZ4PU5QWNV22DYDUFTZTDRZDAVCNFSM6AAAAABNBSPIQWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGMBYGUZDGMRZGU> . You are receiving this because you authored the thread.Message ID: ***@***.***>