[PATCH 0/6] block: add support for REQ_OP_VERIFY

Johannes Thumshirn Johannes.Thumshirn at wdc.com
Wed Jul 13 02:36:40 PDT 2022


On 13.07.22 11:14, Chaitanya Kulkarni wrote:
>> I'd suggest calculating some kind of checksum, even something like a
>> SHA-1 of the contents would be worth having.  It doesn't need to be
>> crypto-secure; just something the host can verify the device didn't spoof.
> I did not understand exactly what you mean here.

I _think_ what Willy wants to say here is, we need some kind of "out-of-band"
checksums to verify the device is not lying to us.

Something like the checksums for each data block that i.e. btrfs has. On read,
we're verifying the calculated checksum of the payload with the saved one and
if they're not matching (for whatever reason) return -EIO.

As the device doesn't know the location of the data checksum, it can't spoof it
for the host.

(Excuse me for my btrfs centric view on this topic, but that's what I know and
what I'm used to)



More information about the Linux-nvme mailing list