The unreachable instruction is actually a block terminator so may only
be inserted at the end of a basic block. This is not always the case for
type asserts, so don't insert this instruction.
LLVM will hopefully optimize it anyway when it realizes anything after
the panic() call is unreachable.
This is kind of dirty with that huge list of linker params, but it works
and it produces smaller object files (probably because GCC is better
optimized for size).
This increases code size on the nrf, but it is a fixed increase and can
hopefully be reduced in the future.
The Makefile gets a lot smaller with this which is a huge advantage
(less build time complexity).
This is the last critical part of the C runtime.
Code size is reduced by 4 bytes for examples/blinky2 (probably due to
inlining) and is unchanged for examples/test.
To get more compatibility with the go command, implement a similar
command line interface (with "tinygo build <package>" etc.).
Other than that, an all-round cleanup of command parsing.
This specifically fixes unix alloc(): previously when allocation fails
it would (recursively) call alloc() again to create an interface due to
lacking escape analysis.
Also, all other cases shouldn't try to allocate just because something
bad happens at runtime.
TODO: implement escape analysis.
Because a few things were left unimplemented it only happened to kind-of
work before in my test cases.
This commit should complete interface-to-interface type asserts.
Add 'optimized-out' parameters to functions. This might make debugging
slightly easier by showing parameter names.
Sadly I din't get setting variables to work. It might be a bug in
CreateExpression.
Not setting the correct debug location at function entry leads to
instructions getting the address of the previously emitted function,
which is of course wrong. LLVM complains about it when validating a
module.
Don't store addresses in the values of registers, this leads to problems
with char arrays (among others). Instead, do it like it's done in C with
raw addresses cast to struct pointers.
This commit also splits gen-device.py, as AVR and ARM have very
different ideas of what a register is. It's easier to just keep them
separate.
These files don't really belong in this repository. It's better to
generate them automatically from a source, like the one provided by the
avr-rust project. So a new command `make gen-device-avr` has been
provided for this purpose.
Ayke van Laethem
Ron Evans