AES Decryption on STM32W

[mingjong]

Posted on June 04, 2012 at 02:59

Hi:

  From simpleMAC & datasheet, it mentioned the STM32W has AES encryption h/w module & API.

  But how about AES decryption?

  It takes long time to fulfill this function in software. And i think it is quit strange if now h/w aes decryption implemented?

#whoosh! #lmgtfy #whoosh! #stm32w-aes

[mingjong]

Posted on June 15, 2012 at 01:56

I am so surprise!

DOES ST TRY TO HIDE THE ISSUE THAT THE CHIP DON'T HAVE AES DECRYPTION!!

DON'T CHOOSE ESCAPE, 

Give me a clear answer, does the chip is failed to implement AES decryption h/w?

Again, i am disappointed about ST's attitude toward customer's question.

[frankmeyer9]

Posted on June 15, 2012 at 08:24

You might have got something wrong here.

Albeit this is, in fact, an ST related forum, it is NOT manned by ST people, comparable to a help desk. Actually, they rarely monitor this forum.

Your complaint might be valid, but you better contact ST or a sales representative directly.

[harry_rostovtsev]

Posted on August 01, 2013 at 01:42

Edit: I'm an idiot.

[Tesla DeLorean]

Posted on August 01, 2013 at 02:10

Look at the STM32F2xx_StdPeriph_Examples for how to encrypt and decrypt with the AES module.  The basic code looks like this (cut and pasted from their example):

Not to point out the obvious, but he's complaining about the STM32W108 not the F2xx parts.

[adrian]

Posted on August 01, 2013 at 07:59

Define slow?  Now define what would be an acceptable decryption speed.

Are you sure your AES implementation is as optimised as possible?  

I know the processor you're using has limited clock rates, but I've just recently been doing crypto stuff on the F207, and I was able to shave half a second off decryption with a simple optimisation of a function.  We do AES256 decryption and that takes virtually zero time to decrypt a block of data, so much so that I wasn't convinced it was doing it correctly when I first started seeing the numbers.

[Tesla DeLorean]

Posted on August 01, 2013 at 09:01

Before we wade into a long discussion, this thread is over a year old. Mr Jou is ticked off because he looked under the hood and found half the engine missing. I've looked over the materials for the STM32W parts, and the library, and it would appear that in only encrypts. I can understand his frustration about his problem, and at the lack of response to his question.

The AES software libraries from ST require a lot of hoops to be jumped through, limits on exportation, etc. I find this all rather odd because I can find solutions on the internets that work quite effectively, and can be readily optimized/targeted.

The slowness being complained about is somewhat more hard to quantify, but I'll observe that software solutions are significantly slower than hardware solutions, and optimized ones take up a lot more space/resources.

May be the hardware is there but doesn't work, or is missing to address some export limitations/concerns, which I find confusing for a device that is designed, built and tested outside the USA. I recall several eBay vendors in China offering F215 and F417 based boards.

[jeremy239955]

Posted on June 03, 2014 at 20:08

Adrian, I was planning on using the H/W encryption on a 217, but was just found out I'm stuck with a 207.

Wondering if you could share where you got your S/W implementation for AES.

Thanks much.

-Jeremy

[Tesla DeLorean]

Posted on June 03, 2014 at 21:25

As I mentioned ST does have a software library, but it requires as many flaming hopes and export controls as the hardware.

What amounts of data are we talking about, and AES-128, 192 or 256?