SiFive HiFive1 (RISC-V based microcontroller)

[hamster_nz]

I got a SiFive HiFive1 RISC-V devboard for Christmas - the hint to my wife worked

https://www.sifive.com/products/hifive1/

The HiFive1 is an Arduino-Compatible development kit featuring the Freedom E310, the industry’s first commercially available RISC-V SoC.

I spent last night building a new Ubuntu VM, then cloning and building the GitHub repos for the tool chain. A few hours later it now says "Hello World" - not much for a micro-controller with 16 MB of (offchip) flash.

Just wanted to say it all worked as advertised - it's open source all the way down to the micro-controller's HDL.

[ataradov]

Waiting on mine from KS campaign.

[andersm]

it's open source all the way down to the micro-controller's HDL.

Got mine yesterday, haven't gotten further than reading the documentation. But it's only the processor core that's open source, not the whole chip.

[ataradov]

But it's only the processor core that's open source, not the whole chip.

What do you mean? The whole thing is built from Rocket Chip Generator, which includes the peripherals as well.

[andersm]

My mistake then (though in practice this is impossible to verify, which is why I think all the freedom/trust/etc. talk is mostly empty propaganda. For an end-user like me, all that changes is who you have to place your trust in.)

[hamster_nz]

My mistake then (though in practice this is impossible to verify, which is why I think all the freedom/trust/etc. talk is mostly empty propaganda. For an end-user like me, all that changes is who you have to place your trust in.)

It does allow you to review the design, and check for oversights and/or h/w bugs even when you don't have the resources to make your own chips.

This is no worse than the level of trust you place in your compiler and libraries to not have code-generation errors or include some unexpected secret extra payload in your executable. There is no way I could review the whole GCC and C standard library code base looking for issues, but it is comforting to know that I could if I really wanted to.

The truly paranoid could decap the die and compare against masks and/or HDL, which should cost far less than actually spinning your own chip - a bit like disassembling all your libraries to check them against the original source...

[ataradov]

It does allow you to review the design, and check for oversights and/or h/w bugs even when you don't have the resources to make your own chips.

Except that you don't know if what is baked inside the silicon is the same stuff as in the code. You trust them to give you an actual code

The truly paranoid could decap the die and compare against masks and/or HDL, which should cost far less than actually spinning your own chip - a bit like disassembling all your libraries to check them against the original source...

There were a number of nasty things described on one of the CCC presentations, where electrically and logically correct layout allowed for some manipulations.

[brucehoult]

it's open source all the way down to the micro-controller's HDL.

Got mine yesterday, haven't gotten further than reading the documentation. But it's only the processor core that's open source, not the whole chip.

The entire SoC is open source.

The HiFive1 board contains non-open components. An example is the 128 Mb flash memory, though that is fairly trivial and doesn't contain a lot of IP. The more notorious one is the FTDI USB chip. I think there simply wasn't another good open option at the moment, especially as they wanted to support JTAG via USB for debugging and for programming the flash.

Even RMS started the FSF with just emacs, bison, ld and gdb usable. (gcc came soon after)

My Founder's Edition HiFIve1 finally arrived in Moscow on Friday and I managed to wrestle it quickly from the hands of Russia Post despite the language barrier :-) Sadly all my Arduino accessories are in NZ, so I'll have to find some more here.