Improve cramfs CRC check

[qkaiser]

We observed a sample from Cisco where the CRC value was stored in a different endianness. This is non standard but we decided to support it anyway.

We also added a check for old cramfs format. If it's an old format, the CRC check always returns true because it's not supported.

More information about CRC validation and old format: https://github.com/secularbird/cramfs/blob/master/cramfsck.c

Closes #394

[martonilles]

can we somehow identify those firmware and just disable in those cases?

[qkaiser]

can we somehow identify those firmware and just disable in those cases?

We actually need to keep the crc check, it was an endianness issue: computed_crc=0x3af2f56e header_crc=0x6ef5f23a

[martonilles]

would it be possible to add cases for this in the integration tests?

• swapped crc
• omitted crc

[qkaiser]

I added test cases, putting this here for future reference:

dd if=fruits.cramfs_be bs=1 skip=32 count=4 of=crc.bin conv=swab
cp fruits.cramfs_be fruits.crc_swap.cramfs_be
dd if=crc.bin of=fruits.crc_swap.cramfs_be bs=1 skip=32 count=4 conv=notrunc

Don't know why people call it "disk destroyer" :(

[qkaiser]

would it be possible to add cases for this in the integration tests?

• swapped crc
• omitted crc

Created test cases for both situations.