This commit parallelizes almost everything that can currently be
parallelized. With that, it also introduces a framework for easily
parallelizing other parts of the compiler.
Code for baremetal targets already compiles slightly faster because it
can parallelize the compilation of supporting assembly files. However,
the speedup is especially noticeable when libraries (compiler-rt,
picolibc) also need to be compiled: they will be compiled in parallel
next to the Go files using all available cores. On my dual core laptop
(4 cores if you count hyperthreading) this cuts compilation time roughly
in half when compiling something for a Cortex-M board after running
`tinygo clean`.
This works around some UB in LLVM, where an out-of-bounds conversion would produce a poison value.
The selected behavior is saturating, except that NaN is mapped to the minimum value.
This commit finally introduces unit tests for the compiler, to check
whether input Go code is converted to the expected output IR.
To make this necessary, a few refactors were needed. Hopefully these
refactors (to compile a program package by package instead of all at
once) will eventually become standard, so that packages can all be
compiled separate from each other and be cached between compiles.
This should make exported names a bit more consistent.
I believe there was a bug report for this issue, but I can't easily find
it. In any case, I think it's an important improvement to match the
behavior of the Go toolchain.
On WebAssembly it is possible to grow the heap with the memory.grow
instruction. This commit implements this feature and with that also
removes the -heap-size flag that was reportedly broken (I haven't
verified that). This should make it easier to use TinyGo for
WebAssembly, where there was no good reason to use a fixed heap size.
This commit has no effect on baremetal targets with optimizations
enabled.
This commit swaps the layout of the heap. Previously, the metadata was
at the start and the data blocks (the actual heap memory) followed
after. This commit swaps those, so that the heap area starts with the
data blocks followed by the heap metadata.
This arrangement is not very relevant for baremetal targets that always
have all RAM allocated, but it is an important improvement for other
targets such as WebAssembly where growing the heap is possible but
starting with a small heap is a good idea. Because the metadata lives at
the end, and because the metadata does not contain pointers, it can
easily be moved. The data itself cannot be moved as the conservative GC
does not know all the pointer locations, plus moving the data could be
very expensive.
This commit changes the number of wait states for the stm32f103 chip to
2 instead of 4. This gets it back in line with the datasheet, but it
also has the side effect of breaking I2C. Therefore, another (seemingly
unrelated) change is needed: the i2cTimeout constant must be increased
to a higher value to adjust to the lower flash wait states - presumably
because the lower number of wait states allows the chip to run code
faster.
Before this change, the compiler could panic with the following message:
panic: 20 not an Int
That of course doesn't make much sense. But it apparently is expected
behavior, see https://github.com/golang/go/issues/43165 for details.
This commit fixes this issue by converting the constant to an integer if
needed.
During a run of interp, some memory (for example, memory allocated
through runtime.alloc) may not have a known LLVM type. This memory is
alllocated by creating an i8 array.
This does not necessarily work, as i8 has no alignment requirements
while the allocated object may have allocation requirements. Therefore,
the resulting global may have an alignment that is too loose.
This works on some microcontrollers but notably does not work on a
Cortex-M0 or Cortex-M0+, as all load/store operations must be aligned.
This commit fixes this by setting the alignment of untyped memory to the
maximum alignment. The determination of "maximum alignment" is not
great but should get the job done on most architectures.
For a full explanation, see interp/README.md. In short, this rewrite is
a redesign of the partial evaluator which improves it over the previous
partial evaluator. The main functional difference is that when
interpreting a function, the interpretation can be rolled back when an
unsupported instruction is encountered (for example, an actual unknown
instruction or a branch on a value that's only known at runtime). This
also means that it is no longer necessary to scan functions to see
whether they can be interpreted: instead, this package now just tries to
interpret it and reverts when it can't go further.
This new design has several benefits:
* Most errors coming from the interp package are avoided, as it can
simply skip the code it can't handle. This has long been an issue.
* The memory model has been improved, which means some packages now
pass all tests that previously didn't pass them.
* Because of a better design, it is in fact a bit faster than the
previous version.
This means the following packages now pass tests with `tinygo test`:
* hash/adler32: previously it would hang in an infinite loop
* math/cmplx: previously it resulted in errors
This also means that the math/big package can be imported. It would
previously fail with a "interp: branch on a non-constant" error.
Because the parentHandle parameter wasn't always set to the right value,
the coroutine lowering pass would sometimes panic with "trying to make
exported function async" even though there was no exported function
involved. Therefore, it should unconditionally be set to avoid this.
The parent function doesn't always have the parentHandle function
parameter set because it can only be set after defining a function, not
when it is only declared.
Previously, EmitPointerPack would generate an out-of-bounds read from an
alloca. This commit fixes that by creating an alloca of the appropriate
size instead of using the size of the to-be-packed data (which might be
smaller than a pointer).
I discovered this error while working on a rewrite of the interp
package, which checks for out-of-bounds reads and writes. There I
discovered this issue when the image package was compiled.
This newer peripheral supports DMA (through EasyDMA) and should
generally be faster. Importantly for some operations: interrupts (within
255 byte buffers) will not interfere with the SPI transfer.
The nrf52 series is all very similar and copying the code only makes it
harder to maintain the code or to add more chips in the nrf52 series
(for example, the nrf52833 as used in the micro:bit v2).
This commit also has a small improvement regarding pins: it now includes
chip-level pin names (P0.00, P0.01, etc) to the machine package.
Be able to run `tinygo gdb -target=arduino examples/serial` and debug a
program with the power of a real debugger.
Note that this only works on LLVM 11 because older versions have a bug
in the AVR backend that cause it to produce invalid debug information:
https://reviews.llvm.org/D74213.
This commit switches to LLVM 11 for builds with LLVM linked statically
(e.g. `make`). It does not yet switch the default for builds dynamically
linked to LLVM, that should be done in a later change.
This commit also changes to use the default host toolchain (probably
GCC) instead of Clang as the default compiler in CI. There were some
issues with Clang 3.8 in CI and hopefully this will fix it.
Additionally it updates the way LLVM is built on Windows, with
-DLLVM_ENABLE_PIC=OFF (which should have been used all along). This
change makes it possible to revert a hack to build libclang manually and
instead uses the libclang static library like on all other operating
systems, simplifying the Makefile.
I have chosed to call this implementation `esp8266` instead of `xtensa`
as it has been written specifically for the ESP8266 and there are no
other Xtensa chips with the CALL0 ABI (no windowing) that I know of. The
only other related chip is the ESP32, which does implement register
windowing and thus needs a very different implementation.
This has been a *lot* of work, trying to understand the Xtensa windowed
registers ABI. But in the end I managed to come up with a very simple
implementation that so far seems to work very well.
I tested this with both blinky examples (with blinky2 slightly edited)
and ./testdata/coroutines.go to verify that it actually works.
Most development happened on the ESP32 QEMU fork from Espressif
(https://github.com/espressif/qemu/wiki) but I also verified that it
works on a real ESP32.
The Cortex-M target isn't much changed, but much of the logic for the
AVR stack switcher that was previously in assembly has now been moved to
Go to make it more maintainable and in fact smaller in code size. Three
functions (tinygo_getCurrentStackPointer, tinygo_switchToTask,
tinygo_switchToScheduler) have been changed to one: tinygo_swapTask.
This reduction in assembly code should make the code more maintainable
and should make it easier to port stack switching to other
architectures.
I've also moved the assembly files to src/internal/task, which seems
like a more appropriate location to me.
Unfortunately, CircleCI doesn't seem to provide Debian stretch builds
with Go 1.15. We should be using Debian stretch (an older distro) to
make sure the tinygo binary runs on as many Linux systems as possible
(including older ones), and I think using Go 1.14 for these builds is
unfortunate but the better tradeoff.
