Overflowing CCM scratchpad

RE: Overflowing CCM scratchpad

David Anthony — Wed, 19 Jun 2019 10:01:41 GMT

Thanks !!

RE: Overflowing CCM scratchpad

Einar Thorsrud — Wed, 19 Jun 2019 07:00:07 GMT

Hi,

I am glad you found the issue., and thanks for letting me know the details. I have reported this internally so that the PS can be updated, properly defining MAXPACKETSIZE.

RE: Overflowing CCM scratchpad

David Anthony — Tue, 18 Jun 2019 14:04:12 GMT

Hi,

This question was originally posted a few days ago.

Since then I have solved the issue.

The root of the problem was the amount of space allocated for the scratchpad.

According to the CCM chapter of the product specification:

"A space of 43 bytes, or (16 + MAXPACKETSIZE) bytes, whatever is largest, must be reserved in RAM."

As there is no definition of MAXPACKETSIZE to be found in the CCM chapter, I made an assumption.

I took MAXPACKETSIZE to mean the maximum packet size I intended to use which was:

64 + header + length + RFU + MIC = 64 + 1 + 1 + 1 + 4 = 71.

This turned out to be inadequate and caused scratchpad overflows.

Trial and error showed that MAXPACKETSIZE actually means the maximum theoretical packet size. This gives:

MAXPACKETSIZE = 255 + header + length + RFU + MIC = 255 + 1 + 1 + 1 + 4 = 262

Using this value of MAXPACKETSIZE to calculate the required scratchpad size results in no overflows.

The requirements for the size of the scratch data area could perhaps be a little clearer in the product specification.

Thanks.

RE: Overflowing CCM scratchpad

Einar Thorsrud — Mon, 17 Jun 2019 11:11:56 GMT

Hi,

I am lacking a bit of information in order to understand your test better. Can you show more of your code, particularly how scratchpad is declared? How do you know that there has been an overflow? Did you, for instance, memset it to a known pattern before doing the CCM decrypt operation?