Support a heap memory region

[japaric]

The obvious place to place this region is right after the .data + .bss region. Should we go ahead with that or should we support placing it at a custom address?

[whitequark]

That sounds good as a first approximation, and doesn't have any critical issues I can think of.

[japaric]

FWIW, you can already create a heap with the current version of this crate. The _edata symbol already points to the end of .data so you can use it as the start of the heap.

Here are some instructions on how to use the alloc-cortex-m crate to allocate a Vec. The instructions assume a cortex-m-quickstart setup.

I think the next steps for this issue would be

• create a proper _heap_start symbol that is just an alias of _edata
• write documentation about the _heap_start symbol here
• add an example to cortex-m-quickstart that shows how to initialize and use an allocator

[whitequark]

Yeah, I've seen alloc-cortex-m. I actually have no need in these instructions (at least right now) because all of my code (especially my TCP/IP stack) is designed to be heap-less :)

[japaric]

I have seen some linker scripts declare a .heap section in the linker script. Has Anyone any need for that? I can't think of any use for such section myself.

[whitequark]

It is quite useless.