What is actual size required for scratch area for CCM on nrf52?

[will]

Hello:

I read the documentation and errata for Rev 1 of the nrf52. I am using the Rigado BMD300 module (thus the nrf52) without a softdevice attempting to do on the fly encryption/decryption. I have noticed that the scratch area required by the nrf52 is much greater than the 43 bytes mentioned in the documentation. It appears it is more on the order of 67 words (67 * 4 octets).

Has anyone else experienced this before?

Thanks!

[bjorn-spockeli]

Yes, we have seen this before and it may be an error in the Product Specification. It has been pointed out that the number of bytes must be reserved in RAM should be 16 + MAXPACKETSIZE bytes.

I have reported this internally and will update my answer as soon as I hear back from R&D.

-Bjørn

---

UPDATE 13.04.16:

Confirmed by R&D that the number of bytes that must be reserved in RAM should be 16 + MAXPACKETSIZE bytes. This will be fixed in the in the next revison of the nRF52832 Product Specification.

[will]

Thanks Bjorn. It would be good to know the exact amount.

[Kevin]

Hello, I got the same problem and I'm encrypting packet of length 13. The scratch pointer is written on 267 bytes for me too. Is it normal? Can you confirm that SCRATCH DATA area size correspond to maximum value of length field = 255 (in the case i am using a 8bit length field) + 16 bytes? you can see more information on my problem on my post : devzone.nordicsemi.com/.../

ps: i am using NRF52832-QFAA80XXXX

[bjorn-spockeli]

Yes, that is correct. When using 8-bit mode of CCM (MODE = LENGTH_EXTENDED), the size of the SCRATCH area will be 267 bytes.

[Kevin]

Thank you Bjorn!

[Torsten Robitzki]

I found this thread, because I was running in the same Problem. BTW: the term MAXPACKETSIZE is used exactly once in the documentation. So I understood it to be the maxium packet size, not the maximum value that could be stored in the length field.