Hi all,
We're using the DM8168 crypto accelerator with OCF to perform AES encryption and SHA1 authentication in our application.
Sometimes approximately about 1 out of 50 SHA1 results seems to be wrong.
We noticed that this problem happens even with a test program that performs SHA1 with a known key and input text, producing a known result, thus ruling out any problems in feeding key or input from our application.
The same input and key hashed several times produces different results. This problem doesn't happen when a similar test is performed using AES.
Has anybody come across this or a similar problem?
BTW, I followed the steps in this wiki to get the drivers and OCF framework installed .
OCF version: Tried both OCF linux 20120127 (latest) and the previous version (20110720).
Regards,
Sundar