I'm not aware of state of Haskell compilers for ARM platform, but I'm an embedded developer. As far as I know, you can't run Cortex-M3 code on Raspberry Pi, because latest is Cortex-A8 and I believe M3 and A8 are binary incompatible. Regards, Alexey 2015-02-06 4:18 GMT+02:00 Mike Meyer :

Lately, I've not been doing much Haskell because I've buried in embedded systems work. The code is all C, with only a smidgen of C++ (for which I am grateful).

One of the things I've been toying with is turning part of the on-chip configuration information a haskel functions. They'd even be pure code, and would be a lot more expressive than what's there now.

The target platforms are running Cortex M3 CPUs. I know that jhc/ajhc will generate code for those CPUs, but I really need at least a subset of the compiler running on the target processor with no pre-installed RTOS. Has that been done? Is it likely to even be possible?

As part of the project, I'll need to install both a cross-compiler on my Unix box, and the the compiler (or parts of it) on the M3. Can someone provide pointers to tutorials on that? My google-foo wasn't powerful enough to find any.

I'm thinking about using either an Aruino Due (as it has the M3 CPU on it) as a testbed platform. So tutorials targeted for that platform would be particularly appreciated.

Since I hopefully have the eyes of people experienced with these things, should I be considering a raspberry pi or something similar instead of the Due? Basically, will I be able to run code built for the M3 CPU on the pi CPUs?

thanks, Mike

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

Try this demo on Cortex-M3: https://github.com/ajhc/demo-cortex-m3/tree/master/stm32f3-discovery I only have Cortex-M0+ and Cortex-M4 in my disposition but I think the above should run on your hardware. You would likely need to modify the linker script stm32f3.ld for your memory layout. BTW, Raspberry Pi is ARM11 (ARM architecture v6) and not Cortex-M3 (arm architecture v7). On Raspberry Pi you can run bare metal binaries on as little as 32KiB RAM in a similar way to the above demo, with a GCC cross-compiler rather than in Linux under LLVM. I understand your question was about bare metal support for Haskell and not LLVM. --Vladimir On 6 February 2015 at 02:18, Mike Meyer wrote:

Lately, I've not been doing much Haskell because I've buried in embedded systems work. The code is all C, with only a smidgen of C++ (for which I am grateful).

One of the things I've been toying with is turning part of the on-chip configuration information a haskel functions. They'd even be pure code, and would be a lot more expressive than what's there now.

The target platforms are running Cortex M3 CPUs. I know that jhc/ajhc will generate code for those CPUs, but I really need at least a subset of the compiler running on the target processor with no pre-installed RTOS. Has that been done? Is it likely to even be possible?

As part of the project, I'll need to install both a cross-compiler on my Unix box, and the the compiler (or parts of it) on the M3. Can someone provide pointers to tutorials on that? My google-foo wasn't powerful enough to find any.

I'm thinking about using either an Aruino Due (as it has the M3 CPU on it) as a testbed platform. So tutorials targeted for that platform would be particularly appreciated.

Since I hopefully have the eyes of people experienced with these things, should I be considering a raspberry pi or something similar instead of the Due? Basically, will I be able to run code built for the M3 CPU on the pi CPUs?

thanks, Mike

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe