Anyone familiar with Cortex-M23?

[technix]

I managed to get a few samples of ATSAML11D16A - probably the first Cortex-M23+TrustZone chip ever, and I plan to cook a mini dev kit up for it. Is there any specific things I should test for? Is there any suggestion on designing the dev kit?

What I have in mind for the dev kit is a small board breaking all signals except clocks and SWD out to a 0.6in DIP style package. The clock pins goes to matching crystals, the SWD pins goes to a debug connector, and throw in an onboard USB to serial converter for power and quick communication, a reset button and a MCU-controlled LED.

[ataradov]

If you are doing a complete breakout board, then bring out the clock and SWD pins to the headers as well. Dedicated clock pins may not be needed for all applications, but spare pins may be useful. SWD pins may also be used as GPIO pins after programming.

On a hardware level M23 is not that different from other Cortex devices. 

[technix]

If you are doing a complete breakout board, then bring out the clock and SWD pins to the headers as well. Dedicated clock pins may not be needed for all applications, but spare pins may be useful. SWD pins may also be used as GPIO pins after programming.

If I break those out, I have to add 0R jumpers to allow isolation, especially on the clocks which does not like long dangling stubs.

On a hardware level M23 is not that different from other Cortex devices.

New architecture, new ISA, TZ is a big thing AFAIK.

[ataradov]

If I break those out, I have to add 0R jumpers to allow isolation, especially on the clocks which does not like long dangling stubs.

For a small board, it is really not that important. Don't overthink it.

New architecture, new ISA, TZ is a big thing AFAIK.

But it has no bearing on hardware design. There is nothing extra you need to do on the board to support TZ.

[technix]

If I break those out, I have to add 0R jumpers to allow isolation, especially on the clocks which does not like long dangling stubs.

For a small board, it is really not that important. Don't overthink it.

Will it still be small when plugged into a breadboard?

New architecture, new ISA, TZ is a big thing AFAIK.

But it has no bearing on hardware design. There is nothing extra you need to do on the board to support TZ.

There are some sayings flying around that you need proper system design to prevent TZ from being side channel compromised.

[ataradov]

Will it still be small when plugged into a breadboard?

It all will probably still work unless the breadboard is extremely bad. The recommendation you are talking about is correct for industrial application. It is almost never a problem in a lab environment. It is ultimately up to you, but I consider spare pins to be move valuable than external crystal in almost all applications. The device has all the high speed clocks you need and they are accurate enough for most uses. If you are really worried, put 0 Ohm jumpers.

There are some sayings flying around that you need proper system design to prevent TZ from being side channel compromised.

If I were to try to hack your devices, I would not do it in your well-designed board. I would remove the IC and put it on my breakout board, where I can create conditions suitable for hacking.

[technix]

Will it still be small when plugged into a breadboard?

It all will probably still work unless the breadboard is extremely bad. The recommendation you are talking about is correct for industrial application. It is almost never a problem in a lab environment. It is ultimately up to you, but I consider spare pins to be move valuable than external crystal in almost all applications. The device has all the high speed clocks you need and they are accurate enough for most uses. If you are really worried, put 0 Ohm jumpers.

Well if this is the case I think I will design the board to be similar to a general-purpose 24-pin breakout board, but with stuff going on at the back side of the board. There is still one pin I wouldn't break out though - there is a built-in buck converter for core voltage and one associated inductor pin, which is connected internally. That physical pin would be used for VBUS though, since this board will contain an onboard USB to serial adapter.

[westfw]

TZ is a big thing AFAIK.

Yeah, right.  For all those secure applications that people have been trying to implement on cheap ARMa with 16k of flash, only to be thwarted by the inherent HW architecture limitations of the CM0 core...

[ataradov]

M23 is nice on its own. It is much more powerful than M0+. TZ is nice to have for some applications, but it is hard to use, since more design and planning is needed, so I don't expect a lot of people will end up using it.

Specifically with SAM L11, the nice thing (if you are a device manufacturer) is that you can completely lock the device. Even erase will not work unless you know the key. So getting off the shelf hardware and making your own firmware would be impossible. Sucks for hacker community though.