This is a big commit that changes the way runtime type information is stored in
the binary. Instead of compressing it and storing it in a number of sidetables,
it is stored similar to how the Go compiler toolchain stores it (but still more
compactly).
This has a number of advantages:
* It is much easier to add new features to reflect support. They can simply
be added to these structs without requiring massive changes (especially in
the reflect lowering pass).
* It removes the reflect lowering pass, which was a large amount of hard to
understand and debug code.
* The reflect lowering pass also required merging all LLVM IR into one
module, which is terrible for performance especially when compiling large
amounts of code. See issue 2870 for details.
* It is (probably!) easier to reason about for the compiler.
The downside is that it increases code size a bit, especially when reflect is
involved. I hope to fix some of that in later patches.
These instructions sometimes pop up in LLVM 15, but they never occured
in LLVM 14. Implementing them is relatively straightforward: simply
generalize the code that already exists for llvm.ICmp interpreting.
A number of llvm.Const* functions (in particular extractvalue and
insertvalue) were removed in LLVM 15, so we have to use a builder
instead. This builder will create the same constant values, it simply
uses a different API.
This change triggers a revert whenever a basic block runs instructions at runtime twice.
As a result, a loop body with runtime-only instructions will no longer be unrolled.
This should help some extreme cases where loops can be expanded into hundreds or thousands of instructions.
Bug:
1. fn.locals[v.value] returns 0 (the default value) if v.value is not
part of the fn.locals map.
2. locals[fn.locals[v.value]] then returns the first local value, which
is usually non-nil
3. This incorrect value is then used as the operand value.
The manifestation of this convoluted bug was
https://github.com/tinygo-org/tinygo/issues/2842. It didn't occur more
often probably because it only seems to happen in practice with inline
assembly.
Fixes https://github.com/tinygo-org/tinygo/issues/2842
Previously, a type assertion on a nil interface would result in an out-of-bounds operand access, as we assumed it was a ptrtoint.
This would usually result in an undefined value which happens not to have the same name as the asserted type (and therefore the assertion fails as expected).
However, with an LLVM build with asserts, LLVM throws an assertion error:
```
interp.test: /home/niaow/go/src/github.com/tinygo-org/tinygo/llvm-project/llvm/include/llvm/IR/User.h:170: llvm::Value* llvm::User::getOperand(unsigned int) const: Assertion `i < NumUserOperands && "getOperand() out of range!"' failed.
```
This change handles a type code of 0 specially.
This removes the parentHandle argument from the internal calling convention.
It was formerly used to implment coroutines.
Now that coroutines have been removed, it is no longer necessary.
This change prevents interp from trying to execute goroutine starts or checks.
This fixes a bug where a goroutine started by an init function would run before the init function.
This commit will use the memory layout information for heap allocations
added in the previous commit to determine LLVM types, instead of
guessing their types based on the content. This fixes a bug in which
recursive data structures (such as doubly linked lists) would result in
a compiler stack overflow due to infinite recursion.
Not all heap allocations have a memory layout yet, but this can be
incrementally fixed in the future. So far, this commit should fix
(almost?) all cases of this stack overflow issue.
This commit simplifies the IR a little bit: instead of calling
pseudo-functions runtime.interfaceImplements and
runtime.interfaceMethod, real declared functions are being called that
are then defined in the interface lowering pass. This should simplify
the interaction between various transformation passes. It also reduces
the number of lines of code, which is generally a good thing.
The interp package is in many cases able to execute map functions in the
runtime directly. This is probably slower than adding special support
for them in the interp package and also doesn't cover all cases (most
importantly, map keys that contain pointers) but removing this code also
removes a large amount of code that needs to be maintained and is
susceptible to hard-to-find bugs.
As a side effect, this resulted in different output of the
testdata/map.go test because the test relied on the existing iteration
order of TinyGo maps. I've updated the test to not rely on this test,
making the output compatible with what the Go toolchain would output.
I've discovered a bug in the implementation of the PHI instruction in
the interp package. This commit fixes the bug.
I've found this issue while investigating an issue with maps after
running interp per package.
A switch statement is not normally emitted by the compiler package, but
LLVM function passes may convert a series of if/else pairs to a switch
statement. A future change will run function passes in the package
compile phase, so the interp package (which is also run after all
modules are merged together) will need to deal with these new switch
statements.
Sometimes, LLVM may rename named structs when merging modules.
Therefore, we can't rely on typecodeID structs to retain their struct
names.
This commit changes the interface lowering pass to not rely on these
names. The interp package does however still rely on this name, but I
hope to fix that in the future.
This results in a significant speedup in some cases. For example, this
runs over twice as fast with a warm cache:
tinygo build -o test.elf ./testdata/stdlib.go
This should help a lot with edit-compile-test cycles, that typically
only modify a single package.
This required some changes to the interp package to deal with globals
created in a previous run of the interp package and to deal with
external globals (that can't be loaded from or stored to).
This commit replaces a number of panics with returning an error value as
a result of changing the toLLVMValue method signature. This should make
it easier to diagnose issues.
This matches the main Go implementation and (among others) fixes a
compatibility issue with the encoding/json package. The encoding/json
package compares reflect.Type variables against nil, which does not work
as long as reflect.Type is of integer type.
This also adds a reflect.RawType() function (like reflect.Type()) that
makes it easier to avoid working with interfaces in the runtime package.
It is internal only, but exported to let the runtime package use it.
This change introduces a small code size increase when working with the
reflect package, but I've tried to keep it to a minimum. Most programs
that don't make extensive use of the reflect package (and don't use
package like fmt) should not be impacted by this.
This is necessary so that when reflect.Type is converted from a concrete
type to an interface type, the errors package can still be interpreted.
Without this change, basically every program would grow in size by a few
bytes.
The errors package has a call like this in the package initializer. This
commit adds support for running it at compile time, avoiding the call at
runtime.
This doesn't always help (the call is already optimized away in many
small programs) but it does help to shave off some binary size in larger
programs. Perhaps more importantly, it will avoid a penalty in code size
when the reflect package will convert reflect.Type from a regular type
to an interface type.
Previously there was code to avoid impossible type asserts but it wasn't
great and in fact was too aggressive when combined with reflection.
This commit improves this by checking all types that exist in the
program that may appear in an interface (even struct fields and the
like) but without creating runtime.typecodeID objects with the type
assert. This has two advantages:
* As mentioned, it optimizes impossible type asserts away.
* It allows methods on types that were only asserted on (in
runtime.typeAssert) but never used in an interface to be optimized
away using GlobalDCE. This may have a cascading effect so that other
parts of the code can be further optimized.
This sometimes massively improves code size and mostly negates the code
size regression of the previous commit.
This distinction was useful before when reflect wasn't properly
supported. Back then it made sense to only include method sets that were
actually used in an interface. But now that it is possible to get to
other values (for example, by extracting fields from structs) and it is
possible to turn them back into interfaces, it is necessary to preserve
all method sets that can possibly be used in the program in a type
assert, interface assert or interface method call.
In the future, this logic will need to be revisited again when
reflect.New or reflect.Zero gets implemented.
Code size increases a bit in some cases, but usually in a very limited
way (except for one outlier in the drivers smoke tests). The next commit
will improve the situation significantly.
GetElementPtr would not work on values that weren't pointers. Because
fixed addresses (often used in memory-mapped I/O) are integers rather
than pointers in interp, it would return an error.
This resulted in the teensy40 target not compiling correctly since the
interp package rewrite. This commit should fix that.
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.
