update go-ffmpreg to v0.2.5 (pulls in latest tetratelabs/wazero) (#3203)

This commit is contained in:
kim 2024-08-15 00:08:55 +00:00 committed by GitHub
parent 6fe96a5611
commit 09f24e0446
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
75 changed files with 1772 additions and 1913 deletions

4
go.mod
View file

@ -12,7 +12,7 @@ require (
codeberg.org/gruf/go-debug v1.3.0
codeberg.org/gruf/go-errors/v2 v2.3.2
codeberg.org/gruf/go-fastcopy v1.1.3
codeberg.org/gruf/go-ffmpreg v0.2.4
codeberg.org/gruf/go-ffmpreg v0.2.5
codeberg.org/gruf/go-iotools v0.0.0-20240710125620-934ae9c654cf
codeberg.org/gruf/go-kv v1.6.4
codeberg.org/gruf/go-list v0.0.0-20240425093752-494db03d641f
@ -56,7 +56,7 @@ require (
github.com/superseriousbusiness/oauth2/v4 v4.3.2-SSB.0.20230227143000-f4900831d6c8
github.com/tdewolff/minify/v2 v2.20.37
github.com/technologize/otel-go-contrib v1.1.1
github.com/tetratelabs/wazero v1.7.3
github.com/tetratelabs/wazero v1.8.0
github.com/tomnomnom/linkheader v0.0.0-20180905144013-02ca5825eb80
github.com/ulule/limiter/v3 v3.11.2
github.com/uptrace/bun v1.2.1

8
go.sum
View file

@ -52,8 +52,8 @@ codeberg.org/gruf/go-fastcopy v1.1.3 h1:Jo9VTQjI6KYimlw25PPc7YLA3Xm+XMQhaHwKnM7x
codeberg.org/gruf/go-fastcopy v1.1.3/go.mod h1:GDDYR0Cnb3U/AIfGM3983V/L+GN+vuwVMvrmVABo21s=
codeberg.org/gruf/go-fastpath/v2 v2.0.0 h1:iAS9GZahFhyWEH0KLhFEJR+txx1ZhMXxYzu2q5Qo9c0=
codeberg.org/gruf/go-fastpath/v2 v2.0.0/go.mod h1:3pPqu5nZjpbRrOqvLyAK7puS1OfEtQvjd6342Cwz56Q=
codeberg.org/gruf/go-ffmpreg v0.2.4 h1:9NR0a5a0RjiIpyQgsqmHen6oadABADv04BWt7dr9kuE=
codeberg.org/gruf/go-ffmpreg v0.2.4/go.mod h1:oPMfBkOK7xmR/teT/dKW6SeMFpRos9ceR/OuUrxBfcQ=
codeberg.org/gruf/go-ffmpreg v0.2.5 h1:suQJ8VdWLkqUhDhHJEdOHMdLqvPisUxcAJiKrxjc6KQ=
codeberg.org/gruf/go-ffmpreg v0.2.5/go.mod h1:sViRI0BYK2B8PJw4BrOg7DquPD71mZjDfffRAFcDtvk=
codeberg.org/gruf/go-iotools v0.0.0-20240710125620-934ae9c654cf h1:84s/ii8N6lYlskZjHH+DG6jyia8w2mXMZlRwFn8Gs3A=
codeberg.org/gruf/go-iotools v0.0.0-20240710125620-934ae9c654cf/go.mod h1:zZAICsp5rY7+hxnws2V0ePrWxE0Z2Z/KXcN3p/RQCfk=
codeberg.org/gruf/go-kv v1.6.4 h1:3NZiW8HVdBM3kpOiLb7XfRiihnzZWMAixdCznguhILk=
@ -552,8 +552,8 @@ github.com/tdewolff/test v1.0.11-0.20240106005702-7de5f7df4739 h1:IkjBCtQOOjIn03
github.com/tdewolff/test v1.0.11-0.20240106005702-7de5f7df4739/go.mod h1:XPuWBzvdUzhCuxWO1ojpXsyzsA5bFoS3tO/Q3kFuTG8=
github.com/technologize/otel-go-contrib v1.1.1 h1:wZH9aSPNWZWIkEh3vfaKfMb15AJ80jJ1aVj/4GZdqIw=
github.com/technologize/otel-go-contrib v1.1.1/go.mod h1:dCN/wj2WyUO8aFZFdIN+6tfJHImjTML/8r2YVYAy3So=
github.com/tetratelabs/wazero v1.7.3 h1:PBH5KVahrt3S2AHgEjKu4u+LlDbbk+nsGE3KLucy6Rw=
github.com/tetratelabs/wazero v1.7.3/go.mod h1:ytl6Zuh20R/eROuyDaGPkp82O9C/DJfXAwJfQ3X6/7Y=
github.com/tetratelabs/wazero v1.8.0 h1:iEKu0d4c2Pd+QSRieYbnQC9yiFlMS9D+Jr0LsRmcF4g=
github.com/tetratelabs/wazero v1.8.0/go.mod h1:yAI0XTsMBhREkM/YDAK/zNou3GoiAce1P6+rp/wQhjs=
github.com/tidwall/btree v0.0.0-20191029221954-400434d76274 h1:G6Z6HvJuPjG6XfNGi/feOATzeJrfgTNJY+rGrHbA04E=
github.com/tidwall/btree v0.0.0-20191029221954-400434d76274/go.mod h1:huei1BkDWJ3/sLXmO+bsCNELL+Bp2Kks9OLyQFkzvA8=
github.com/tidwall/buntdb v1.1.2 h1:noCrqQXL9EKMtcdwJcmuVKSEjqu1ua99RHHgbLTEHRo=

View file

@ -10,6 +10,10 @@
)
type Args struct {
// Optional further module configuration function.
// (e.g. to mount filesystem dir, set env vars, etc).
Config func(wazero.ModuleConfig) wazero.ModuleConfig
// Standard FDs.
Stdin io.Reader
Stdout io.Writer
@ -17,10 +21,6 @@ type Args struct {
// CLI args.
Args []string
// Optional further module configuration function.
// (e.g. to mount filesystem dir, set env vars, etc).
Config func(wazero.ModuleConfig) wazero.ModuleConfig
}
type Instantiator struct {

View file

@ -1,7 +1,7 @@
gofumpt := mvdan.cc/gofumpt@v0.5.0
gofumpt := mvdan.cc/gofumpt@v0.6.0
gosimports := github.com/rinchsan/gosimports/cmd/gosimports@v0.3.8
golangci_lint := github.com/golangci/golangci-lint/cmd/golangci-lint@v1.55.2
golangci_lint := github.com/golangci/golangci-lint/cmd/golangci-lint@v1.60.0
asmfmt := github.com/klauspost/asmfmt/cmd/asmfmt@v1.3.2
# sync this with netlify.toml!
hugo := github.com/gohugoio/hugo@v0.115.2
@ -20,22 +20,6 @@ main_packages := $(sort $(foreach f,$(dir $(main_sources)),$(if $(findstring ./,
go_test_options ?= -timeout 300s
ensureCompilerFastest := -ldflags '-X github.com/tetratelabs/wazero/internal/integration_test/vs.ensureCompilerFastest=true'
.PHONY: bench
bench:
@go build ./internal/integration_test/bench/...
@# Don't use -test.benchmem as it isn't accurate when comparing against CGO libs
@for d in vs/time vs/wasmedge vs/wasmtime ; do \
cd ./internal/integration_test/$$d ; \
go test -bench=. . -tags='wasmedge' $(ensureCompilerFastest) ; \
cd - ;\
done
bench_testdata_dir := internal/integration_test/bench/testdata
.PHONY: build.bench
build.bench:
@tinygo build -o $(bench_testdata_dir)/case.wasm -scheduler=none --no-debug -target=wasi $(bench_testdata_dir)/case.go
.PHONY: test.examples
test.examples:
@go test $(go_test_options) ./examples/... ./imports/assemblyscript/example/... ./imports/emscripten/... ./imports/wasi_snapshot_preview1/example/...
@ -183,7 +167,7 @@ build.spectest.threads:
.PHONY: test
test:
@go test $(go_test_options) $$(go list ./... | grep -vE '$(spectest_v1_dir)|$(spectest_v2_dir)')
@go test $(go_test_options) ./...
@cd internal/version/testdata && go test $(go_test_options) ./...
@cd internal/integration_test/fuzz/wazerolib && CGO_ENABLED=0 WASM_BINARY_PATH=testdata/test.wasm go test ./...
@ -194,17 +178,6 @@ coverage: ## Generate test coverage
@go test -coverprofile=coverage.txt -covermode=atomic --coverpkg=$(coverpkg) $(main_packages)
@go tool cover -func coverage.txt
.PHONY: spectest
spectest:
@$(MAKE) spectest.v1
@$(MAKE) spectest.v2
spectest.v1:
@go test $(go_test_options) $$(go list ./... | grep $(spectest_v1_dir))
spectest.v2:
@go test $(go_test_options) $$(go list ./... | grep $(spectest_v2_dir))
golangci_lint_path := $(shell go env GOPATH)/bin/golangci-lint
$(golangci_lint_path):
@ -214,7 +187,7 @@ golangci_lint_goarch ?= $(shell go env GOARCH)
.PHONY: lint
lint: $(golangci_lint_path)
@GOARCH=$(golangci_lint_goarch) CGO_ENABLED=0 $(golangci_lint_path) run --timeout 5m
@GOARCH=$(golangci_lint_goarch) CGO_ENABLED=0 $(golangci_lint_path) run --timeout 5m -E testableexamples
.PHONY: format
format:

View file

@ -1,6 +1,6 @@
# wazero: the zero dependency WebAssembly runtime for Go developers
[![WebAssembly Core Specification Test](https://github.com/tetratelabs/wazero/actions/workflows/spectest.yaml/badge.svg)](https://github.com/tetratelabs/wazero/actions/workflows/spectest.yaml) [![Go Reference](https://pkg.go.dev/badge/github.com/tetratelabs/wazero.svg)](https://pkg.go.dev/github.com/tetratelabs/wazero) [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Go Reference](https://pkg.go.dev/badge/github.com/tetratelabs/wazero.svg)](https://pkg.go.dev/github.com/tetratelabs/wazero) [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
WebAssembly is a way to safely run code compiled in other languages. Runtimes
execute WebAssembly Modules (Wasm), which are most often binaries with a `.wasm`

View file

@ -151,9 +151,13 @@ type Module interface {
// ExportedFunction returns a function exported from this module or nil if it wasn't.
//
// Note: The default wazero.ModuleConfig attempts to invoke `_start`, which
// in rare cases can close the module. When in doubt, check IsClosed prior
// to invoking a function export after instantiation.
// # Notes
// - The default wazero.ModuleConfig attempts to invoke `_start`, which
// in rare cases can close the module. When in doubt, check IsClosed prior
// to invoking a function export after instantiation.
// - The semantics of host functions assumes the existence of an "importing module" because, for example, the host function needs access to
// the memory of the importing module. Therefore, direct use of ExportedFunction is forbidden for host modules.
// Practically speaking, it is usually meaningless to directly call a host function from Go code as it is already somewhere in Go code.
ExportedFunction(name string) Function
// ExportedFunctionDefinitions returns all the exported function

View file

@ -179,6 +179,9 @@ type HostFunctionBuilder interface {
// are deferred until Compile.
// - Functions are indexed in order of calls to NewFunctionBuilder as
// insertion ordering is needed by ABI such as Emscripten (invoke_*).
// - The semantics of host functions assumes the existence of an "importing module" because, for example, the host function needs access to
// the memory of the importing module. Therefore, direct use of ExportedFunction is forbidden for host modules.
// Practically speaking, it is usually meaningless to directly call a host function from Go code as it is already somewhere in Go code.
type HostModuleBuilder interface {
// Note: until golang/go#5860, we can't use example tests to embed code in interface godocs.
@ -341,12 +344,24 @@ func (b *hostModuleBuilder) Compile(ctx context.Context) (CompiledModule, error)
return c, nil
}
// hostModuleInstance is a wrapper around api.Module that prevents calling ExportedFunction.
type hostModuleInstance struct{ api.Module }
// ExportedFunction implements api.Module ExportedFunction.
func (h hostModuleInstance) ExportedFunction(name string) api.Function {
panic("calling ExportedFunction is forbidden on host modules. See the note on ExportedFunction interface")
}
// Instantiate implements HostModuleBuilder.Instantiate
func (b *hostModuleBuilder) Instantiate(ctx context.Context) (api.Module, error) {
if compiled, err := b.Compile(ctx); err != nil {
return nil, err
} else {
compiled.(*compiledModule).closeWithModule = true
return b.r.InstantiateModule(ctx, compiled, NewModuleConfig())
m, err := b.r.InstantiateModule(ctx, compiled, NewModuleConfig())
if err != nil {
return nil, err
}
return hostModuleInstance{m}, nil
}
}

View file

@ -24,6 +24,13 @@
// All implementations are in wazero.
// - Instances of this can be reused across multiple runtimes, if configured
// via RuntimeConfig.
// - The cache check happens before the compilation, so if multiple Goroutines are
// trying to compile the same module simultaneously, it is possible that they
// all compile the module. The design here is that the lock isn't held for the action "Compile"
// but only for checking and saving the compiled result. Therefore, we strongly recommend that the embedder
// does the centralized compilation in a single Goroutines (or multiple Goroutines per Wasm binary) to generate cache rather than
// trying to Compile in parallel for a single module. In other words, we always recommend to produce CompiledModule
// share it across multiple Goroutines to avoid trying to compile the same module simultaneously.
type CompilationCache interface{ api.Closer }
// NewCompilationCache returns a new CompilationCache to be passed to RuntimeConfig.

View file

@ -495,7 +495,20 @@ type ModuleConfig interface {
WithFSConfig(FSConfig) ModuleConfig
// WithName configures the module name. Defaults to what was decoded from
// the name section. Empty string ("") clears any name.
// the name section. Duplicate names are not allowed in a single Runtime.
//
// Calling this with the empty string "" makes the module anonymous.
// That is useful when you want to instantiate the same CompiledModule multiple times like below:
//
// for i := 0; i < N; i++ {
// // Instantiate a new Wasm module from the already compiled `compiledWasm` anonymously without a name.
// instance, err := r.InstantiateModule(ctx, compiledWasm, wazero.NewModuleConfig().WithName(""))
// // ....
// }
//
// See the `concurrent-instantiation` example for a complete usage.
//
// Non-empty named modules are available for other modules to import by name.
WithName(string) ModuleConfig
// WithStartFunctions configures the functions to call after the module is

View file

@ -0,0 +1,19 @@
package experimental
import (
"context"
"github.com/tetratelabs/wazero/api"
"github.com/tetratelabs/wazero/internal/expctxkeys"
)
// ImportResolver is an experimental func type that, if set,
// will be used as the first step in resolving imports.
// See issue 2294.
// If the import name is not found, it should return nil.
type ImportResolver func(name string) api.Module
// WithImportResolver returns a new context with the given ImportResolver.
func WithImportResolver(ctx context.Context, resolver ImportResolver) context.Context {
return context.WithValue(ctx, expctxkeys.ImportResolverKey{}, resolver)
}

View file

@ -1962,17 +1962,11 @@ func pathSymlinkFn(_ context.Context, mod api.Module, params []uint64) experimen
return dir.FS.Symlink(
// Do not join old path since it's only resolved when dereference the link created here.
// And the dereference result depends on the opening directory's file descriptor at that point.
bufToStr(oldPathBuf),
unsafe.String(&oldPathBuf[0], int(oldPathLen)),
newPathName,
)
}
// bufToStr converts the given byte slice as string unsafely.
func bufToStr(buf []byte) string {
// TODO: use unsafe.String after flooring Go 1.20.
return *(*string)(unsafe.Pointer(&buf))
}
// pathUnlinkFile is the WASI function named PathUnlinkFileName which unlinks a
// file.
//

View file

@ -68,9 +68,7 @@ func pollOneoffFn(_ context.Context, mod api.Module, params []uint64) sys.Errno
}
outBuf, ok := mem.Read(out, nsubscriptions*32)
// zero-out all buffer before writing
for i := range outBuf {
outBuf[i] = 0
}
clear(outBuf)
if !ok {
return sys.EFAULT

View file

@ -154,11 +154,6 @@ func (t *Table[Key, Item]) Range(f func(Key, Item) bool) {
// Reset clears the content of the table.
func (t *Table[Key, Item]) Reset() {
for i := range t.masks {
t.masks[i] = 0
}
var zero Item
for i := range t.items {
t.items[i] = zero
}
clear(t.masks)
clear(t.items)
}

View file

@ -26,11 +26,14 @@
type (
controlFrame struct {
frameID uint32
// originalStackLen holds the number of values on the stack
// originalStackLenWithoutParam holds the number of values on the stack
// when Start executing this control frame minus params for the block.
originalStackLenWithoutParam int
blockType *wasm.FunctionType
kind controlFrameKind
// originalStackLenWithoutParamUint64 is almost the same as originalStackLenWithoutParam
// except that it holds the number of values on the stack in uint64.
originalStackLenWithoutParamUint64 int
blockType *wasm.FunctionType
kind controlFrameKind
}
controlFrames struct{ frames []controlFrame }
)
@ -157,9 +160,11 @@ type compiler struct {
enabledFeatures api.CoreFeatures
callFrameStackSizeInUint64 int
stack []unsignedType
currentFrameID uint32
controlFrames controlFrames
unreachableState struct {
// stackLenInUint64 is the length of the stack in uint64.
stackLenInUint64 int
currentFrameID uint32
controlFrames controlFrames
unreachableState struct {
on bool
depth int
}
@ -341,6 +346,7 @@ funcIndex := c.next
c.pc = 0
c.currentOpPC = 0
c.currentFrameID = 0
c.stackLenInUint64 = 0
c.unreachableState.on, c.unreachableState.depth = false, 0
if err := c.compile(sig, code.Body, code.LocalTypes, code.BodyOffsetInCodeSection); err != nil {
@ -449,10 +455,11 @@ func (c *compiler) handleInstruction() error {
// Create a new frame -- entering this block.
frame := controlFrame{
frameID: c.nextFrameID(),
originalStackLenWithoutParam: len(c.stack) - len(bt.Params),
kind: controlFrameKindBlockWithoutContinuationLabel,
blockType: bt,
frameID: c.nextFrameID(),
originalStackLenWithoutParam: len(c.stack) - len(bt.Params),
originalStackLenWithoutParamUint64: c.stackLenInUint64 - bt.ParamNumInUint64,
kind: controlFrameKindBlockWithoutContinuationLabel,
blockType: bt,
}
c.controlFrames.push(frame)
@ -473,10 +480,11 @@ func (c *compiler) handleInstruction() error {
// Create a new frame -- entering loop.
frame := controlFrame{
frameID: c.nextFrameID(),
originalStackLenWithoutParam: len(c.stack) - len(bt.Params),
kind: controlFrameKindLoop,
blockType: bt,
frameID: c.nextFrameID(),
originalStackLenWithoutParam: len(c.stack) - len(bt.Params),
originalStackLenWithoutParamUint64: c.stackLenInUint64 - bt.ParamNumInUint64,
kind: controlFrameKindLoop,
blockType: bt,
}
c.controlFrames.push(frame)
@ -515,8 +523,9 @@ func (c *compiler) handleInstruction() error {
// Create a new frame -- entering if.
frame := controlFrame{
frameID: c.nextFrameID(),
originalStackLenWithoutParam: len(c.stack) - len(bt.Params),
frameID: c.nextFrameID(),
originalStackLenWithoutParam: len(c.stack) - len(bt.Params),
originalStackLenWithoutParamUint64: c.stackLenInUint64 - bt.ParamNumInUint64,
// Note this will be set to controlFrameKindIfWithElse
// when else opcode found later.
kind: controlFrameKindIfWithoutElse,
@ -543,7 +552,7 @@ func (c *compiler) handleInstruction() error {
// If it is currently in unreachable, and the non-nested if,
// reset the stack so we can correctly handle the else block.
top := c.controlFrames.top()
c.stack = c.stack[:top.originalStackLenWithoutParam]
c.stackSwitchAt(top)
top.kind = controlFrameKindIfWithElse
// Re-push the parameters to the if block so that else block can use them.
@ -572,7 +581,7 @@ func (c *compiler) handleInstruction() error {
// Reset the stack manipulated by the then block, and re-push the block param types to the stack.
c.stack = c.stack[:frame.originalStackLenWithoutParam]
c.stackSwitchAt(frame)
for _, t := range frame.blockType.Params {
c.stackPush(wasmValueTypeTounsignedType(t))
}
@ -601,7 +610,7 @@ func (c *compiler) handleInstruction() error {
return nil
}
c.stack = c.stack[:frame.originalStackLenWithoutParam]
c.stackSwitchAt(frame)
for _, t := range frame.blockType.Results {
c.stackPush(wasmValueTypeTounsignedType(t))
}
@ -628,7 +637,7 @@ func (c *compiler) handleInstruction() error {
// We need to reset the stack so that
// the values pushed inside the block.
dropOp := newOperationDrop(c.getFrameDropRange(frame, true))
c.stack = c.stack[:frame.originalStackLenWithoutParam]
c.stackSwitchAt(frame)
// Push the result types onto the stack.
for _, t := range frame.blockType.Results {
@ -3505,6 +3514,11 @@ func (c *compiler) stackPeek() (ret unsignedType) {
return
}
func (c *compiler) stackSwitchAt(frame *controlFrame) {
c.stack = c.stack[:frame.originalStackLenWithoutParam]
c.stackLenInUint64 = frame.originalStackLenWithoutParamUint64
}
func (c *compiler) stackPop() (ret unsignedType) {
// No need to check stack bound
// as we can assume that all the operations
@ -3512,11 +3526,13 @@ func (c *compiler) stackPop() (ret unsignedType) {
// at module validation phase.
ret = c.stack[len(c.stack)-1]
c.stack = c.stack[:len(c.stack)-1]
c.stackLenInUint64 -= 1 + int(unsignedTypeV128&ret>>2)
return
}
func (c *compiler) stackPush(ts unsignedType) {
c.stack = append(c.stack, ts)
c.stackLenInUint64 += 1 + int(unsignedTypeV128&ts>>2)
}
// emit adds the operations into the result.
@ -3565,7 +3581,7 @@ func (c *compiler) emitDefaultValue(t wasm.ValueType) {
// of the n-th local.
func (c *compiler) localDepth(index wasm.Index) int {
height := c.localIndexToStackHeightInUint64[index]
return c.stackLenInUint64(len(c.stack)) - 1 - int(height)
return c.stackLenInUint64 - 1 - height
}
func (c *compiler) localType(index wasm.Index) (t wasm.ValueType) {
@ -3592,14 +3608,7 @@ func (c *compiler) getFrameDropRange(frame *controlFrame, isEnd bool) inclusiveR
} else {
start = frame.blockType.ResultNumInUint64
}
var end int
if frame.kind == controlFrameKindFunction {
// On the function return, we eliminate all the contents on the stack
// including locals (existing below of frame.originalStackLen)
end = c.stackLenInUint64(len(c.stack)) - 1
} else {
end = c.stackLenInUint64(len(c.stack)) - 1 - c.stackLenInUint64(frame.originalStackLenWithoutParam)
}
end := c.stackLenInUint64 - 1 - frame.originalStackLenWithoutParamUint64
if start <= end {
return inclusiveRange{Start: int32(start), End: int32(end)}
} else {
@ -3607,17 +3616,6 @@ func (c *compiler) getFrameDropRange(frame *controlFrame, isEnd bool) inclusiveR
}
}
func (c *compiler) stackLenInUint64(ceil int) (ret int) {
for i := 0; i < ceil; i++ {
if c.stack[i] == unsignedTypeV128 {
ret += 2
} else {
ret++
}
}
return
}
func (c *compiler) readMemoryArg(tag string) (memoryArg, error) {
c.result.UsesMemory = true
alignment, num, err := leb128.LoadUint32(c.body[c.pc+1:])

View file

@ -3901,14 +3901,9 @@ func() {
case operationKindV128Dot:
x2Hi, x2Lo := ce.popValue(), ce.popValue()
x1Hi, x1Lo := ce.popValue(), ce.popValue()
ce.pushValue(
uint64(uint32(int32(int16(x1Lo>>0))*int32(int16(x2Lo>>0))+int32(int16(x1Lo>>16))*int32(int16(x2Lo>>16)))) |
(uint64(uint32(int32(int16(x1Lo>>32))*int32(int16(x2Lo>>32))+int32(int16(x1Lo>>48))*int32(int16(x2Lo>>48)))) << 32),
)
ce.pushValue(
uint64(uint32(int32(int16(x1Hi>>0))*int32(int16(x2Hi>>0))+int32(int16(x1Hi>>16))*int32(int16(x2Hi>>16)))) |
(uint64(uint32(int32(int16(x1Hi>>32))*int32(int16(x2Hi>>32))+int32(int16(x1Hi>>48))*int32(int16(x2Hi>>48)))) << 32),
)
lo, hi := v128Dot(x1Hi, x1Lo, x2Hi, x2Lo)
ce.pushValue(lo)
ce.pushValue(hi)
frame.pc++
case operationKindV128ITruncSatFromF:
hi, lo := ce.popValue(), ce.popValue()
@ -4584,3 +4579,18 @@ func (ce *callEngine) callGoFuncWithStack(ctx context.Context, m *wasm.ModuleIns
ce.stack = ce.stack[0 : len(ce.stack)-shrinkLen]
}
}
// v128Dot performs a dot product of two 64-bit vectors.
// Note: for some reason (which I suspect is due to a bug in Go compiler's regalloc),
// inlining this function causes a bug which happens **only when** we run with -race AND arm64 AND Go 1.22.
func v128Dot(x1Hi, x1Lo, x2Hi, x2Lo uint64) (uint64, uint64) {
r1 := int32(int16(x1Lo>>0)) * int32(int16(x2Lo>>0))
r2 := int32(int16(x1Lo>>16)) * int32(int16(x2Lo>>16))
r3 := int32(int16(x1Lo>>32)) * int32(int16(x2Lo>>32))
r4 := int32(int16(x1Lo>>48)) * int32(int16(x2Lo>>48))
r5 := int32(int16(x1Hi>>0)) * int32(int16(x2Hi>>0))
r6 := int32(int16(x1Hi>>16)) * int32(int16(x2Hi>>16))
r7 := int32(int16(x1Hi>>32)) * int32(int16(x2Hi>>32))
r8 := int32(int16(x1Hi>>48)) * int32(int16(x2Hi>>48))
return uint64(uint32(r1+r2)) | (uint64(uint32(r3+r4)) << 32), uint64(uint32(r5+r6)) | (uint64(uint32(r7+r8)) << 32)
}

View file

@ -69,7 +69,7 @@ type Compiler interface {
AllocateVReg(typ ssa.Type) regalloc.VReg
// ValueDefinition returns the definition of the given value.
ValueDefinition(ssa.Value) *SSAValueDefinition
ValueDefinition(ssa.Value) SSAValueDefinition
// VRegOf returns the virtual register of the given ssa.Value.
VRegOf(value ssa.Value) regalloc.VReg
@ -79,13 +79,13 @@ type Compiler interface {
// MatchInstr returns true if the given definition is from an instruction with the given opcode, the current group ID,
// and a refcount of 1. That means, the instruction can be merged/swapped within the current instruction group.
MatchInstr(def *SSAValueDefinition, opcode ssa.Opcode) bool
MatchInstr(def SSAValueDefinition, opcode ssa.Opcode) bool
// MatchInstrOneOf is the same as MatchInstr but for multiple opcodes. If it matches one of ssa.Opcode,
// this returns the opcode. Otherwise, this returns ssa.OpcodeInvalid.
//
// Note: caller should be careful to avoid excessive allocation on opcodes slice.
MatchInstrOneOf(def *SSAValueDefinition, opcodes []ssa.Opcode) ssa.Opcode
MatchInstrOneOf(def SSAValueDefinition, opcodes []ssa.Opcode) ssa.Opcode
// AddRelocationInfo appends the relocation information for the function reference at the current buffer offset.
AddRelocationInfo(funcRef ssa.FuncRef)
@ -126,10 +126,7 @@ type compiler struct {
nextVRegID regalloc.VRegID
// ssaValueToVRegs maps ssa.ValueID to regalloc.VReg.
ssaValueToVRegs [] /* VRegID to */ regalloc.VReg
// ssaValueDefinitions maps ssa.ValueID to its definition.
ssaValueDefinitions []SSAValueDefinition
// ssaValueRefCounts is a cached list obtained by ssa.Builder.ValueRefCounts().
ssaValueRefCounts []int
ssaValuesInfo []ssa.ValueInfo
// returnVRegs is the list of virtual registers that store the return values.
returnVRegs []regalloc.VReg
varEdges [][2]regalloc.VReg
@ -206,15 +203,10 @@ func (c *compiler) setCurrentGroupID(gid ssa.InstructionGroupID) {
// assignVirtualRegisters assigns a virtual register to each ssa.ValueID Valid in the ssa.Builder.
func (c *compiler) assignVirtualRegisters() {
builder := c.ssaBuilder
refCounts := builder.ValueRefCounts()
c.ssaValueRefCounts = refCounts
c.ssaValuesInfo = builder.ValuesInfo()
need := len(refCounts)
if need >= len(c.ssaValueToVRegs) {
c.ssaValueToVRegs = append(c.ssaValueToVRegs, make([]regalloc.VReg, need+1)...)
}
if need >= len(c.ssaValueDefinitions) {
c.ssaValueDefinitions = append(c.ssaValueDefinitions, make([]SSAValueDefinition, need+1)...)
if diff := len(c.ssaValuesInfo) - len(c.ssaValueToVRegs); diff > 0 {
c.ssaValueToVRegs = append(c.ssaValueToVRegs, make([]regalloc.VReg, diff+1)...)
}
for blk := builder.BlockIteratorReversePostOrderBegin(); blk != nil; blk = builder.BlockIteratorReversePostOrderNext() {
@ -225,40 +217,26 @@ func (c *compiler) assignVirtualRegisters() {
typ := p.Type()
vreg := c.AllocateVReg(typ)
c.ssaValueToVRegs[pid] = vreg
c.ssaValueDefinitions[pid] = SSAValueDefinition{BlockParamValue: p, BlkParamVReg: vreg}
c.ssaTypeOfVRegID[vreg.ID()] = p.Type()
}
// Assigns each value to a virtual register produced by instructions.
for cur := blk.Root(); cur != nil; cur = cur.Next() {
r, rs := cur.Returns()
var N int
if r.Valid() {
id := r.ID()
ssaTyp := r.Type()
typ := r.Type()
vReg := c.AllocateVReg(typ)
c.ssaValueToVRegs[id] = vReg
c.ssaValueDefinitions[id] = SSAValueDefinition{
Instr: cur,
N: 0,
RefCount: refCounts[id],
}
c.ssaTypeOfVRegID[vReg.ID()] = ssaTyp
N++
}
for _, r := range rs {
id := r.ID()
ssaTyp := r.Type()
vReg := c.AllocateVReg(ssaTyp)
c.ssaValueToVRegs[id] = vReg
c.ssaValueDefinitions[id] = SSAValueDefinition{
Instr: cur,
N: N,
RefCount: refCounts[id],
}
c.ssaTypeOfVRegID[vReg.ID()] = ssaTyp
N++
}
}
}
@ -299,8 +277,12 @@ func (c *compiler) Init() {
}
// ValueDefinition implements Compiler.ValueDefinition.
func (c *compiler) ValueDefinition(value ssa.Value) *SSAValueDefinition {
return &c.ssaValueDefinitions[value.ID()]
func (c *compiler) ValueDefinition(value ssa.Value) SSAValueDefinition {
return SSAValueDefinition{
V: value,
Instr: c.ssaBuilder.InstructionOfValue(value),
RefCount: c.ssaValuesInfo[value.ID()].RefCount,
}
}
// VRegOf implements Compiler.VRegOf.
@ -319,7 +301,7 @@ func (c *compiler) TypeOf(v regalloc.VReg) ssa.Type {
}
// MatchInstr implements Compiler.MatchInstr.
func (c *compiler) MatchInstr(def *SSAValueDefinition, opcode ssa.Opcode) bool {
func (c *compiler) MatchInstr(def SSAValueDefinition, opcode ssa.Opcode) bool {
instr := def.Instr
return def.IsFromInstr() &&
instr.Opcode() == opcode &&
@ -328,7 +310,7 @@ func (c *compiler) MatchInstr(def *SSAValueDefinition, opcode ssa.Opcode) bool {
}
// MatchInstrOneOf implements Compiler.MatchInstrOneOf.
func (c *compiler) MatchInstrOneOf(def *SSAValueDefinition, opcodes []ssa.Opcode) ssa.Opcode {
func (c *compiler) MatchInstrOneOf(def SSAValueDefinition, opcodes []ssa.Opcode) ssa.Opcode {
instr := def.Instr
if !def.IsFromInstr() {
return ssa.OpcodeInvalid

View file

@ -9,7 +9,7 @@
func (c *compiler) Lower() {
c.assignVirtualRegisters()
c.mach.SetCurrentABI(c.GetFunctionABI(c.ssaBuilder.Signature()))
c.mach.ExecutableContext().StartLoweringFunction(c.ssaBuilder.BlockIDMax())
c.mach.StartLoweringFunction(c.ssaBuilder.BlockIDMax())
c.lowerBlocks()
}
@ -20,12 +20,11 @@ func (c *compiler) lowerBlocks() {
c.lowerBlock(blk)
}
ectx := c.mach.ExecutableContext()
// After lowering all blocks, we need to link adjacent blocks to layout one single instruction list.
var prev ssa.BasicBlock
for next := builder.BlockIteratorReversePostOrderBegin(); next != nil; next = builder.BlockIteratorReversePostOrderNext() {
if prev != nil {
ectx.LinkAdjacentBlocks(prev, next)
c.mach.LinkAdjacentBlocks(prev, next)
}
prev = next
}
@ -33,8 +32,7 @@ func (c *compiler) lowerBlocks() {
func (c *compiler) lowerBlock(blk ssa.BasicBlock) {
mach := c.mach
ectx := mach.ExecutableContext()
ectx.StartBlock(blk)
mach.StartBlock(blk)
// We traverse the instructions in reverse order because we might want to lower multiple
// instructions together.
@ -76,7 +74,7 @@ func (c *compiler) lowerBlock(blk ssa.BasicBlock) {
default:
mach.LowerInstr(cur)
}
ectx.FlushPendingInstructions()
mach.FlushPendingInstructions()
}
// Finally, if this is the entry block, we have to insert copies of arguments from the real location to the VReg.
@ -84,7 +82,7 @@ func (c *compiler) lowerBlock(blk ssa.BasicBlock) {
c.lowerFunctionArguments(blk)
}
ectx.EndBlock()
mach.EndBlock()
}
// lowerBranches is called right after StartBlock and before any LowerInstr call if
@ -93,23 +91,24 @@ func (c *compiler) lowerBlock(blk ssa.BasicBlock) {
//
// See ssa.Instruction IsBranching, and the comment on ssa.BasicBlock.
func (c *compiler) lowerBranches(br0, br1 *ssa.Instruction) {
ectx := c.mach.ExecutableContext()
mach := c.mach
c.setCurrentGroupID(br0.GroupID())
c.mach.LowerSingleBranch(br0)
ectx.FlushPendingInstructions()
mach.FlushPendingInstructions()
if br1 != nil {
c.setCurrentGroupID(br1.GroupID())
c.mach.LowerConditionalBranch(br1)
ectx.FlushPendingInstructions()
mach.FlushPendingInstructions()
}
if br0.Opcode() == ssa.OpcodeJump {
_, args, target := br0.BranchData()
_, args, targetBlockID := br0.BranchData()
argExists := len(args) != 0
if argExists && br1 != nil {
panic("BUG: critical edge split failed")
}
target := c.ssaBuilder.BasicBlock(targetBlockID)
if argExists && target.ReturnBlock() {
if len(args) > 0 {
c.mach.LowerReturns(args)
@ -118,24 +117,25 @@ func (c *compiler) lowerBranches(br0, br1 *ssa.Instruction) {
c.lowerBlockArguments(args, target)
}
}
ectx.FlushPendingInstructions()
mach.FlushPendingInstructions()
}
func (c *compiler) lowerFunctionArguments(entry ssa.BasicBlock) {
ectx := c.mach.ExecutableContext()
mach := c.mach
c.tmpVals = c.tmpVals[:0]
data := c.ssaBuilder.ValuesInfo()
for i := 0; i < entry.Params(); i++ {
p := entry.Param(i)
if c.ssaValueRefCounts[p.ID()] > 0 {
if data[p.ID()].RefCount > 0 {
c.tmpVals = append(c.tmpVals, p)
} else {
// If the argument is not used, we can just pass an invalid value.
c.tmpVals = append(c.tmpVals, ssa.ValueInvalid)
}
}
c.mach.LowerParams(c.tmpVals)
ectx.FlushPendingInstructions()
mach.LowerParams(c.tmpVals)
mach.FlushPendingInstructions()
}
// lowerBlockArguments lowers how to pass arguments to the given successor block.
@ -152,12 +152,12 @@ func (c *compiler) lowerBlockArguments(args []ssa.Value, succ ssa.BasicBlock) {
src := args[i]
dstReg := c.VRegOf(dst)
srcDef := c.ssaValueDefinitions[src.ID()]
if srcDef.IsFromInstr() && srcDef.Instr.Constant() {
srcInstr := c.ssaBuilder.InstructionOfValue(src)
if srcInstr != nil && srcInstr.Constant() {
c.constEdges = append(c.constEdges, struct {
cInst *ssa.Instruction
dst regalloc.VReg
}{cInst: srcDef.Instr, dst: dstReg})
}{cInst: srcInstr, dst: dstReg})
} else {
srcReg := c.VRegOf(src)
// Even when the src=dst, insert the move so that we can keep such registers keep-alive.

View file

@ -1,221 +0,0 @@
package backend
import (
"fmt"
"math"
"github.com/tetratelabs/wazero/internal/engine/wazevo/ssa"
"github.com/tetratelabs/wazero/internal/engine/wazevo/wazevoapi"
)
type ExecutableContext interface {
// StartLoweringFunction is called when the lowering of the given function is started.
// maximumBlockID is the maximum value of ssa.BasicBlockID existing in the function.
StartLoweringFunction(maximumBlockID ssa.BasicBlockID)
// LinkAdjacentBlocks is called after finished lowering all blocks in order to create one single instruction list.
LinkAdjacentBlocks(prev, next ssa.BasicBlock)
// StartBlock is called when the compilation of the given block is started.
// The order of this being called is the reverse post order of the ssa.BasicBlock(s) as we iterate with
// ssa.Builder BlockIteratorReversePostOrderBegin and BlockIteratorReversePostOrderEnd.
StartBlock(ssa.BasicBlock)
// EndBlock is called when the compilation of the current block is finished.
EndBlock()
// FlushPendingInstructions flushes the pending instructions to the buffer.
// This will be called after the lowering of each SSA Instruction.
FlushPendingInstructions()
}
type ExecutableContextT[Instr any] struct {
CurrentSSABlk ssa.BasicBlock
// InstrPool is the InstructionPool of instructions.
InstructionPool wazevoapi.Pool[Instr]
asNop func(*Instr)
setNext func(*Instr, *Instr)
setPrev func(*Instr, *Instr)
// RootInstr is the root instruction of the executable.
RootInstr *Instr
labelPositionPool wazevoapi.Pool[LabelPosition[Instr]]
NextLabel Label
// LabelPositions maps a label to the instructions of the region which the label represents.
LabelPositions []*LabelPosition[Instr]
OrderedBlockLabels []*LabelPosition[Instr]
// PerBlockHead and PerBlockEnd are the head and tail of the instruction list per currently-compiled ssa.BasicBlock.
PerBlockHead, PerBlockEnd *Instr
// PendingInstructions are the instructions which are not yet emitted into the instruction list.
PendingInstructions []*Instr
// SsaBlockIDToLabels maps an SSA block ID to the label.
SsaBlockIDToLabels []Label
}
func NewExecutableContextT[Instr any](
resetInstruction func(*Instr),
setNext func(*Instr, *Instr),
setPrev func(*Instr, *Instr),
asNop func(*Instr),
) *ExecutableContextT[Instr] {
return &ExecutableContextT[Instr]{
InstructionPool: wazevoapi.NewPool[Instr](resetInstruction),
asNop: asNop,
setNext: setNext,
setPrev: setPrev,
labelPositionPool: wazevoapi.NewPool[LabelPosition[Instr]](resetLabelPosition[Instr]),
NextLabel: LabelInvalid,
}
}
func resetLabelPosition[T any](l *LabelPosition[T]) {
*l = LabelPosition[T]{}
}
// StartLoweringFunction implements ExecutableContext.
func (e *ExecutableContextT[Instr]) StartLoweringFunction(max ssa.BasicBlockID) {
imax := int(max)
if len(e.SsaBlockIDToLabels) <= imax {
// Eagerly allocate labels for the blocks since the underlying slice will be used for the next iteration.
e.SsaBlockIDToLabels = append(e.SsaBlockIDToLabels, make([]Label, imax+1)...)
}
}
func (e *ExecutableContextT[Instr]) StartBlock(blk ssa.BasicBlock) {
e.CurrentSSABlk = blk
l := e.SsaBlockIDToLabels[e.CurrentSSABlk.ID()]
if l == LabelInvalid {
l = e.AllocateLabel()
e.SsaBlockIDToLabels[blk.ID()] = l
}
end := e.allocateNop0()
e.PerBlockHead, e.PerBlockEnd = end, end
labelPos := e.GetOrAllocateLabelPosition(l)
e.OrderedBlockLabels = append(e.OrderedBlockLabels, labelPos)
labelPos.Begin, labelPos.End = end, end
labelPos.SB = blk
}
// EndBlock implements ExecutableContext.
func (e *ExecutableContextT[T]) EndBlock() {
// Insert nop0 as the head of the block for convenience to simplify the logic of inserting instructions.
e.insertAtPerBlockHead(e.allocateNop0())
l := e.SsaBlockIDToLabels[e.CurrentSSABlk.ID()]
e.LabelPositions[l].Begin = e.PerBlockHead
if e.CurrentSSABlk.EntryBlock() {
e.RootInstr = e.PerBlockHead
}
}
func (e *ExecutableContextT[T]) insertAtPerBlockHead(i *T) {
if e.PerBlockHead == nil {
e.PerBlockHead = i
e.PerBlockEnd = i
return
}
e.setNext(i, e.PerBlockHead)
e.setPrev(e.PerBlockHead, i)
e.PerBlockHead = i
}
// FlushPendingInstructions implements ExecutableContext.
func (e *ExecutableContextT[T]) FlushPendingInstructions() {
l := len(e.PendingInstructions)
if l == 0 {
return
}
for i := l - 1; i >= 0; i-- { // reverse because we lower instructions in reverse order.
e.insertAtPerBlockHead(e.PendingInstructions[i])
}
e.PendingInstructions = e.PendingInstructions[:0]
}
func (e *ExecutableContextT[T]) Reset() {
e.labelPositionPool.Reset()
e.InstructionPool.Reset()
for i := range e.LabelPositions {
e.LabelPositions[i] = nil
}
e.PendingInstructions = e.PendingInstructions[:0]
e.OrderedBlockLabels = e.OrderedBlockLabels[:0]
e.RootInstr = nil
e.SsaBlockIDToLabels = e.SsaBlockIDToLabels[:0]
e.PerBlockHead, e.PerBlockEnd = nil, nil
e.NextLabel = LabelInvalid
}
// AllocateLabel allocates an unused label.
func (e *ExecutableContextT[T]) AllocateLabel() Label {
e.NextLabel++
return e.NextLabel
}
func (e *ExecutableContextT[T]) GetOrAllocateLabelPosition(l Label) *LabelPosition[T] {
if len(e.LabelPositions) <= int(l) {
e.LabelPositions = append(e.LabelPositions, make([]*LabelPosition[T], int(l)+1-len(e.LabelPositions))...)
}
ret := e.LabelPositions[l]
if ret == nil {
ret = e.labelPositionPool.Allocate()
ret.L = l
e.LabelPositions[l] = ret
}
return ret
}
func (e *ExecutableContextT[T]) GetOrAllocateSSABlockLabel(blk ssa.BasicBlock) Label {
if blk.ReturnBlock() {
return LabelReturn
}
l := e.SsaBlockIDToLabels[blk.ID()]
if l == LabelInvalid {
l = e.AllocateLabel()
e.SsaBlockIDToLabels[blk.ID()] = l
}
return l
}
func (e *ExecutableContextT[T]) allocateNop0() *T {
i := e.InstructionPool.Allocate()
e.asNop(i)
return i
}
// LinkAdjacentBlocks implements backend.Machine.
func (e *ExecutableContextT[T]) LinkAdjacentBlocks(prev, next ssa.BasicBlock) {
prevLabelPos := e.LabelPositions[e.GetOrAllocateSSABlockLabel(prev)]
nextLabelPos := e.LabelPositions[e.GetOrAllocateSSABlockLabel(next)]
e.setNext(prevLabelPos.End, nextLabelPos.Begin)
}
// LabelPosition represents the regions of the generated code which the label represents.
type LabelPosition[Instr any] struct {
SB ssa.BasicBlock
L Label
Begin, End *Instr
BinaryOffset int64
}
// Label represents a position in the generated code which is either
// a real instruction or the constant InstructionPool (e.g. jump tables).
//
// This is exactly the same as the traditional "label" in assembly code.
type Label uint32
const (
LabelInvalid Label = 0
LabelReturn Label = math.MaxUint32
)
// String implements backend.Machine.
func (l Label) String() string {
return fmt.Sprintf("L%d", l)
}

View file

@ -14,7 +14,6 @@
// CompileGoFunctionTrampoline implements backend.Machine.
func (m *machine) CompileGoFunctionTrampoline(exitCode wazevoapi.ExitCode, sig *ssa.Signature, needModuleContextPtr bool) []byte {
ectx := m.ectx
argBegin := 1 // Skips exec context by default.
if needModuleContextPtr {
argBegin++
@ -25,7 +24,7 @@ func (m *machine) CompileGoFunctionTrampoline(exitCode wazevoapi.ExitCode, sig *
m.currentABI = abi
cur := m.allocateNop()
ectx.RootInstr = cur
m.rootInstr = cur
// Execution context is always the first argument.
execCtrPtr := raxVReg
@ -272,7 +271,7 @@ func (m *machine) CompileGoFunctionTrampoline(exitCode wazevoapi.ExitCode, sig *
cur = m.revertRBPRSP(cur)
linkInstr(cur, m.allocateInstr().asRet())
m.encodeWithoutSSA(ectx.RootInstr)
m.encodeWithoutSSA(m.rootInstr)
return m.c.Buf()
}
@ -347,10 +346,8 @@ func (m *machine) storeReturnAddressAndExit(cur *instruction, execCtx regalloc.V
// CompileStackGrowCallSequence implements backend.Machine.
func (m *machine) CompileStackGrowCallSequence() []byte {
ectx := m.ectx
cur := m.allocateNop()
ectx.RootInstr = cur
m.rootInstr = cur
cur = m.setupRBPRSP(cur)
@ -379,7 +376,7 @@ func (m *machine) CompileStackGrowCallSequence() []byte {
cur = m.revertRBPRSP(cur)
linkInstr(cur, m.allocateInstr().asRet())
m.encodeWithoutSSA(ectx.RootInstr)
m.encodeWithoutSSA(m.rootInstr)
return m.c.Buf()
}

View file

@ -17,16 +17,6 @@ type instruction struct {
kind instructionKind
}
// Next implements regalloc.Instr.
func (i *instruction) Next() regalloc.Instr {
return i.next
}
// Prev implements regalloc.Instr.
func (i *instruction) Prev() regalloc.Instr {
return i.prev
}
// IsCall implements regalloc.Instr.
func (i *instruction) IsCall() bool { return i.kind == call }
@ -36,9 +26,6 @@ func (i *instruction) IsIndirectCall() bool { return i.kind == callIndirect }
// IsReturn implements regalloc.Instr.
func (i *instruction) IsReturn() bool { return i.kind == ret }
// AddedBeforeRegAlloc implements regalloc.Instr.
func (i *instruction) AddedBeforeRegAlloc() bool { return i.addedBeforeRegAlloc }
// String implements regalloc.Instr.
func (i *instruction) String() string {
switch i.kind {
@ -651,26 +638,14 @@ func resetInstruction(i *instruction) {
*i = instruction{}
}
func setNext(i *instruction, next *instruction) {
i.next = next
}
func setPrev(i *instruction, prev *instruction) {
i.prev = prev
}
func asNop(i *instruction) {
i.kind = nop0
}
func (i *instruction) asNop0WithLabel(label backend.Label) *instruction { //nolint
func (i *instruction) asNop0WithLabel(label label) *instruction { //nolint
i.kind = nop0
i.u1 = uint64(label)
return i
}
func (i *instruction) nop0Label() backend.Label {
return backend.Label(i.u1)
func (i *instruction) nop0Label() label {
return label(i.u1)
}
type instructionKind byte
@ -1161,7 +1136,7 @@ func (i *instruction) asJmp(target operand) *instruction {
return i
}
func (i *instruction) jmpLabel() backend.Label {
func (i *instruction) jmpLabel() label {
switch i.kind {
case jmp, jmpIf, lea, xmmUnaryRmR:
return i.op1.label()

View file

@ -130,9 +130,9 @@ func (m *machine) lowerAddendsToAmode(x, y addend, offBase uint32) *amode {
}
}
func (m *machine) lowerAddend(x *backend.SSAValueDefinition) addend {
if x.IsFromBlockParam() {
return addend{x.BlkParamVReg, 0, 0}
func (m *machine) lowerAddend(x backend.SSAValueDefinition) addend {
if !x.IsFromInstr() {
return addend{m.c.VRegOf(x.V), 0, 0}
}
// Ensure the addend is not referenced in multiple places; we will discard nested Iadds.
op := m.c.MatchInstrOneOf(x, addendsMatchOpcodes[:])

View file

@ -16,18 +16,13 @@
// NewBackend returns a new backend for arm64.
func NewBackend() backend.Machine {
ectx := backend.NewExecutableContextT[instruction](
resetInstruction,
setNext,
setPrev,
asNop,
)
return &machine{
ectx: ectx,
m := &machine{
cpuFeatures: platform.CpuFeatures,
regAlloc: regalloc.NewAllocator(regInfo),
regAlloc: regalloc.NewAllocator[*instruction, *labelPosition, *regAllocFn](regInfo),
spillSlots: map[regalloc.VRegID]int64{},
amodePool: wazevoapi.NewPool[amode](nil),
labelPositionPool: wazevoapi.NewIDedPool[labelPosition](resetLabelPosition),
instrPool: wazevoapi.NewPool[instruction](resetInstruction),
constSwizzleMaskConstIndex: -1,
constSqmulRoundSatIndex: -1,
constI8x16SHLMaskTableIndex: -1,
@ -41,23 +36,46 @@ func NewBackend() backend.Machine {
constExtAddPairwiseI16x8uMask1Index: -1,
constExtAddPairwiseI16x8uMask2Index: -1,
}
m.regAllocFn.m = m
return m
}
type (
// machine implements backend.Machine for amd64.
machine struct {
c backend.Compiler
ectx *backend.ExecutableContextT[instruction]
stackBoundsCheckDisabled bool
instrPool wazevoapi.Pool[instruction]
amodePool wazevoapi.Pool[amode]
cpuFeatures platform.CpuFeatureFlags
regAlloc regalloc.Allocator
regAllocFn *backend.RegAllocFunction[*instruction, *machine]
regAlloc regalloc.Allocator[*instruction, *labelPosition, *regAllocFn]
regAllocFn regAllocFn
regAllocStarted bool
// labelPositionPool is the pool of labelPosition. The id is the label where
// if the label is less than the maxSSABlockID, it's the ssa.BasicBlockID.
labelPositionPool wazevoapi.IDedPool[labelPosition]
// nextLabel is the next label to be allocated. The first free label comes after maxSSABlockID
// so that we can have an identical label for the SSA block ID, which is useful for debugging.
nextLabel label
// rootInstr is the first instruction of the function.
rootInstr *instruction
// currentLabelPos is the currently-compiled ssa.BasicBlock's labelPosition.
currentLabelPos *labelPosition
// orderedSSABlockLabelPos is the ordered list of labelPosition in the generated code for each ssa.BasicBlock.
orderedSSABlockLabelPos []*labelPosition
// returnLabelPos is the labelPosition for the return block.
returnLabelPos labelPosition
// perBlockHead and perBlockEnd are the head and tail of the instruction list per currently-compiled ssa.BasicBlock.
perBlockHead, perBlockEnd *instruction
// pendingInstructions are the instructions which are not yet emitted into the instruction list.
pendingInstructions []*instruction
// maxSSABlockID is the maximum ssa.BasicBlockID in the current function.
maxSSABlockID label
spillSlotSize int64
spillSlots map[regalloc.VRegID]int64
currentABI *backend.FunctionABI
@ -67,8 +85,11 @@ func NewBackend() backend.Machine {
labelResolutionPends []labelResolutionPend
// jmpTableTargets holds the labels of the jump table targets.
jmpTableTargets [][]uint32
consts []_const
// jmpTableTargetNext is the index to the jmpTableTargets slice to be used for the next jump table.
jmpTableTargetsNext int
consts []_const
constSwizzleMaskConstIndex, constSqmulRoundSatIndex,
constI8x16SHLMaskTableIndex, constI8x16LogicalSHRMaskTableIndex,
@ -79,9 +100,10 @@ func NewBackend() backend.Machine {
}
_const struct {
lo, hi uint64
_var []byte
label *labelPosition
lo, hi uint64
_var []byte
label label
labelPos *labelPosition
}
labelResolutionPend struct {
@ -90,22 +112,73 @@ func NewBackend() backend.Machine {
// imm32Offset is the offset of the last 4 bytes of the instruction.
imm32Offset int64
}
labelPosition = backend.LabelPosition[instruction]
)
func (m *machine) getOrAllocateConstLabel(i *int, _var []byte) backend.Label {
type (
// label represents a position in the generated code which is either
// a real instruction or the constant InstructionPool (e.g. jump tables).
//
// This is exactly the same as the traditional "label" in assembly code.
label uint32
// labelPosition represents the regions of the generated code which the label represents.
// This implements regalloc.Block.
labelPosition struct {
// sb is not nil if this corresponds to a ssa.BasicBlock.
sb ssa.BasicBlock
// cur is used to walk through the instructions in the block during the register allocation.
cur,
// begin and end are the first and last instructions of the block.
begin, end *instruction
// binaryOffset is the offset in the binary where the label is located.
binaryOffset int64
}
)
// String implements backend.Machine.
func (l label) String() string {
return fmt.Sprintf("L%d", l)
}
func resetLabelPosition(l *labelPosition) {
*l = labelPosition{}
}
const labelReturn = math.MaxUint32
func ssaBlockLabel(sb ssa.BasicBlock) label {
if sb.ReturnBlock() {
return labelReturn
}
return label(sb.ID())
}
// getOrAllocateSSABlockLabelPosition returns the labelPosition for the given basic block.
func (m *machine) getOrAllocateSSABlockLabelPosition(sb ssa.BasicBlock) *labelPosition {
if sb.ReturnBlock() {
m.returnLabelPos.sb = sb
return &m.returnLabelPos
}
l := ssaBlockLabel(sb)
pos := m.labelPositionPool.GetOrAllocate(int(l))
pos.sb = sb
return pos
}
func (m *machine) getOrAllocateConstLabel(i *int, _var []byte) label {
index := *i
if index == -1 {
label := m.allocateLabel()
l, pos := m.allocateLabel()
index = len(m.consts)
m.consts = append(m.consts, _const{
_var: _var,
label: label,
_var: _var,
label: l,
labelPos: pos,
})
*i = index
}
return m.consts[index].label.L
return m.consts[index].label
}
// Reset implements backend.Machine.
@ -120,18 +193,20 @@ func (m *machine) Reset() {
}
m.stackBoundsCheckDisabled = false
m.ectx.Reset()
m.regAllocFn.Reset()
m.regAlloc.Reset()
m.labelPositionPool.Reset()
m.instrPool.Reset()
m.regAllocStarted = false
m.clobberedRegs = m.clobberedRegs[:0]
m.spillSlotSize = 0
m.maxRequiredStackSizeForCalls = 0
m.perBlockHead, m.perBlockEnd, m.rootInstr = nil, nil, nil
m.pendingInstructions = m.pendingInstructions[:0]
m.orderedSSABlockLabelPos = m.orderedSSABlockLabelPos[:0]
m.amodePool.Reset()
m.jmpTableTargets = m.jmpTableTargets[:0]
m.jmpTableTargetsNext = 0
m.constSwizzleMaskConstIndex = -1
m.constSqmulRoundSatIndex = -1
m.constI8x16SHLMaskTableIndex = -1
@ -146,8 +221,63 @@ func (m *machine) Reset() {
m.constExtAddPairwiseI16x8uMask2Index = -1
}
// ExecutableContext implements backend.Machine.
func (m *machine) ExecutableContext() backend.ExecutableContext { return m.ectx }
// StartLoweringFunction implements backend.Machine StartLoweringFunction.
func (m *machine) StartLoweringFunction(maxBlockID ssa.BasicBlockID) {
m.maxSSABlockID = label(maxBlockID)
m.nextLabel = label(maxBlockID) + 1
}
// LinkAdjacentBlocks implements backend.Machine.
func (m *machine) LinkAdjacentBlocks(prev, next ssa.BasicBlock) {
prevPos, nextPos := m.getOrAllocateSSABlockLabelPosition(prev), m.getOrAllocateSSABlockLabelPosition(next)
prevPos.end.next = nextPos.begin
}
// StartBlock implements backend.Machine.
func (m *machine) StartBlock(blk ssa.BasicBlock) {
m.currentLabelPos = m.getOrAllocateSSABlockLabelPosition(blk)
labelPos := m.currentLabelPos
end := m.allocateNop()
m.perBlockHead, m.perBlockEnd = end, end
labelPos.begin, labelPos.end = end, end
m.orderedSSABlockLabelPos = append(m.orderedSSABlockLabelPos, labelPos)
}
// EndBlock implements ExecutableContext.
func (m *machine) EndBlock() {
// Insert nop0 as the head of the block for convenience to simplify the logic of inserting instructions.
m.insertAtPerBlockHead(m.allocateNop())
m.currentLabelPos.begin = m.perBlockHead
if m.currentLabelPos.sb.EntryBlock() {
m.rootInstr = m.perBlockHead
}
}
func (m *machine) insertAtPerBlockHead(i *instruction) {
if m.perBlockHead == nil {
m.perBlockHead = i
m.perBlockEnd = i
return
}
i.next = m.perBlockHead
m.perBlockHead.prev = i
m.perBlockHead = i
}
// FlushPendingInstructions implements backend.Machine.
func (m *machine) FlushPendingInstructions() {
l := len(m.pendingInstructions)
if l == 0 {
return
}
for i := l - 1; i >= 0; i-- { // reverse because we lower instructions in reverse order.
m.insertAtPerBlockHead(m.pendingInstructions[i])
}
m.pendingInstructions = m.pendingInstructions[:0]
}
// DisableStackCheck implements backend.Machine.
func (m *machine) DisableStackCheck() { m.stackBoundsCheckDisabled = true }
@ -155,23 +285,17 @@ func (m *machine) DisableStackCheck() { m.stackBoundsCheckDisabled = true }
// SetCompiler implements backend.Machine.
func (m *machine) SetCompiler(c backend.Compiler) {
m.c = c
m.regAllocFn = backend.NewRegAllocFunction[*instruction, *machine](m, c.SSABuilder(), c)
m.regAllocFn.ssaB = c.SSABuilder()
}
// SetCurrentABI implements backend.Machine.
func (m *machine) SetCurrentABI(abi *backend.FunctionABI) {
m.currentABI = abi
}
func (m *machine) SetCurrentABI(abi *backend.FunctionABI) { m.currentABI = abi }
// RegAlloc implements backend.Machine.
func (m *machine) RegAlloc() {
rf := m.regAllocFn
for _, pos := range m.ectx.OrderedBlockLabels {
rf.AddBlock(pos.SB, pos.L, pos.Begin, pos.End)
}
m.regAllocStarted = true
m.regAlloc.DoAllocation(rf)
m.regAlloc.DoAllocation(&rf)
// Now that we know the final spill slot size, we must align spillSlotSize to 16 bytes.
m.spillSlotSize = (m.spillSlotSize + 15) &^ 15
}
@ -184,49 +308,54 @@ func (m *machine) InsertReturn() {
// LowerSingleBranch implements backend.Machine.
func (m *machine) LowerSingleBranch(b *ssa.Instruction) {
ectx := m.ectx
switch b.Opcode() {
case ssa.OpcodeJump:
_, _, targetBlk := b.BranchData()
_, _, targetBlkID := b.BranchData()
if b.IsFallthroughJump() {
return
}
jmp := m.allocateInstr()
target := ectx.GetOrAllocateSSABlockLabel(targetBlk)
if target == backend.LabelReturn {
target := ssaBlockLabel(m.c.SSABuilder().BasicBlock(targetBlkID))
if target == labelReturn {
jmp.asRet()
} else {
jmp.asJmp(newOperandLabel(target))
}
m.insert(jmp)
case ssa.OpcodeBrTable:
index, target := b.BrTableData()
m.lowerBrTable(index, target)
index, targetBlkIDs := b.BrTableData()
m.lowerBrTable(index, targetBlkIDs)
default:
panic("BUG: unexpected branch opcode" + b.Opcode().String())
}
}
func (m *machine) addJmpTableTarget(targets []ssa.BasicBlock) (index int) {
// TODO: reuse the slice!
labels := make([]uint32, len(targets))
for j, target := range targets {
labels[j] = uint32(m.ectx.GetOrAllocateSSABlockLabel(target))
func (m *machine) addJmpTableTarget(targets ssa.Values) (index int) {
if m.jmpTableTargetsNext == len(m.jmpTableTargets) {
m.jmpTableTargets = append(m.jmpTableTargets, make([]uint32, 0, len(targets.View())))
}
index = m.jmpTableTargetsNext
m.jmpTableTargetsNext++
m.jmpTableTargets[index] = m.jmpTableTargets[index][:0]
for _, targetBlockID := range targets.View() {
target := m.c.SSABuilder().BasicBlock(ssa.BasicBlockID(targetBlockID))
m.jmpTableTargets[index] = append(m.jmpTableTargets[index], uint32(ssaBlockLabel(target)))
}
index = len(m.jmpTableTargets)
m.jmpTableTargets = append(m.jmpTableTargets, labels)
return
}
var condBranchMatches = [...]ssa.Opcode{ssa.OpcodeIcmp, ssa.OpcodeFcmp}
func (m *machine) lowerBrTable(index ssa.Value, targets []ssa.BasicBlock) {
func (m *machine) lowerBrTable(index ssa.Value, targets ssa.Values) {
_v := m.getOperand_Reg(m.c.ValueDefinition(index))
v := m.copyToTmp(_v.reg())
targetCount := len(targets.View())
// First, we need to do the bounds check.
maxIndex := m.c.AllocateVReg(ssa.TypeI32)
m.lowerIconst(maxIndex, uint64(len(targets)-1), false)
m.lowerIconst(maxIndex, uint64(targetCount-1), false)
cmp := m.allocateInstr().asCmpRmiR(true, newOperandReg(maxIndex), v, false)
m.insert(cmp)
@ -255,23 +384,22 @@ func (m *machine) lowerBrTable(index ssa.Value, targets []ssa.BasicBlock) {
jmpTable := m.allocateInstr()
targetSliceIndex := m.addJmpTableTarget(targets)
jmpTable.asJmpTableSequence(targetSliceIndex, len(targets))
jmpTable.asJmpTableSequence(targetSliceIndex, targetCount)
m.insert(jmpTable)
}
// LowerConditionalBranch implements backend.Machine.
func (m *machine) LowerConditionalBranch(b *ssa.Instruction) {
exctx := m.ectx
cval, args, targetBlk := b.BranchData()
cval, args, targetBlkID := b.BranchData()
if len(args) > 0 {
panic(fmt.Sprintf(
"conditional branch shouldn't have args; likely a bug in critical edge splitting: from %s to %s",
exctx.CurrentSSABlk,
targetBlk,
m.currentLabelPos.sb,
targetBlkID,
))
}
target := exctx.GetOrAllocateSSABlockLabel(targetBlk)
target := ssaBlockLabel(m.c.SSABuilder().BasicBlock(targetBlkID))
cvalDef := m.c.ValueDefinition(cval)
switch m.c.MatchInstrOneOf(cvalDef, condBranchMatches[:]) {
@ -1272,9 +1400,9 @@ func (m *machine) lowerVconst(dst regalloc.VReg, lo, hi uint64) {
}
load := m.allocateInstr()
constLabel := m.allocateLabel()
m.consts = append(m.consts, _const{label: constLabel, lo: lo, hi: hi})
load.asXmmUnaryRmR(sseOpcodeMovdqu, newOperandMem(m.newAmodeRipRel(constLabel.L)), dst)
l, pos := m.allocateLabel()
m.consts = append(m.consts, _const{label: l, labelPos: pos, lo: lo, hi: hi})
load.asXmmUnaryRmR(sseOpcodeMovdqu, newOperandMem(m.newAmodeRipRel(l)), dst)
m.insert(load)
}
@ -1473,21 +1601,24 @@ func (m *machine) lowerExitIfTrueWithCode(execCtx regalloc.VReg, cond ssa.Value,
jmpIf.asJmpIf(condFromSSAIntCmpCond(c).invert(), newOperandLabel(l))
}
func (m *machine) tryLowerBandToFlag(x, y *backend.SSAValueDefinition) (ok bool) {
var target *backend.SSAValueDefinition
func (m *machine) tryLowerBandToFlag(x, y backend.SSAValueDefinition) (ok bool) {
var target backend.SSAValueDefinition
var got bool
if x.IsFromInstr() && x.Instr.Constant() && x.Instr.ConstantVal() == 0 {
if m.c.MatchInstr(y, ssa.OpcodeBand) {
target = y
got = true
}
}
if y.IsFromInstr() && y.Instr.Constant() && y.Instr.ConstantVal() == 0 {
if m.c.MatchInstr(x, ssa.OpcodeBand) {
target = x
got = true
}
}
if target == nil {
if !got {
return false
}
@ -1522,7 +1653,7 @@ func (m *machine) allocateExitInstructions(execCtx, exitCodeReg regalloc.VReg) (
return
}
func (m *machine) lowerExitWithCode(execCtx regalloc.VReg, code wazevoapi.ExitCode) (afterLabel backend.Label) {
func (m *machine) lowerExitWithCode(execCtx regalloc.VReg, code wazevoapi.ExitCode) (afterLabel label) {
exitCodeReg := rbpVReg
saveRsp, saveRbp, setExitCode := m.allocateExitInstructions(execCtx, exitCodeReg)
@ -1819,9 +1950,9 @@ func (m *machine) lowerCall(si *ssa.Instruction) {
// callerGenVRegToFunctionArg is the opposite of GenFunctionArgToVReg, which is used to generate the
// caller side of the function call.
func (m *machine) callerGenVRegToFunctionArg(a *backend.FunctionABI, argIndex int, reg regalloc.VReg, def *backend.SSAValueDefinition, stackSlotSize int64) {
func (m *machine) callerGenVRegToFunctionArg(a *backend.FunctionABI, argIndex int, reg regalloc.VReg, def backend.SSAValueDefinition, stackSlotSize int64) {
arg := &a.Args[argIndex]
if def != nil && def.IsFromInstr() {
if def.IsFromInstr() {
// Constant instructions are inlined.
if inst := def.Instr; inst.Constant() {
m.insertLoadConstant(inst, reg)
@ -1904,25 +2035,20 @@ func (m *machine) InsertMove(dst, src regalloc.VReg, typ ssa.Type) {
// Format implements backend.Machine.
func (m *machine) Format() string {
ectx := m.ectx
begins := map[*instruction]backend.Label{}
for _, pos := range ectx.LabelPositions {
begins := map[*instruction]label{}
for l := label(0); l < m.nextLabel; l++ {
pos := m.labelPositionPool.Get(int(l))
if pos != nil {
begins[pos.Begin] = pos.L
begins[pos.begin] = l
}
}
irBlocks := map[backend.Label]ssa.BasicBlockID{}
for i, l := range ectx.SsaBlockIDToLabels {
irBlocks[l] = ssa.BasicBlockID(i)
}
var lines []string
for cur := ectx.RootInstr; cur != nil; cur = cur.next {
for cur := m.rootInstr; cur != nil; cur = cur.next {
if l, ok := begins[cur]; ok {
var labelStr string
if blkID, ok := irBlocks[l]; ok {
labelStr = fmt.Sprintf("%s (SSA Block: %s):", l, blkID)
if l <= m.maxSSABlockID {
labelStr = fmt.Sprintf("%s (SSA Block: blk%d):", l, l)
} else {
labelStr = fmt.Sprintf("%s:", l)
}
@ -1935,9 +2061,9 @@ func (m *machine) Format() string {
}
for _, vc := range m.consts {
if vc._var == nil {
lines = append(lines, fmt.Sprintf("%s: const [%d %d]", vc.label.L, vc.lo, vc.hi))
lines = append(lines, fmt.Sprintf("%s: const [%d %d]", vc.label, vc.lo, vc.hi))
} else {
lines = append(lines, fmt.Sprintf("%s: const %#x", vc.label.L, vc._var))
lines = append(lines, fmt.Sprintf("%s: const %#x", vc.label, vc._var))
}
}
return "\n" + strings.Join(lines, "\n") + "\n"
@ -1945,18 +2071,14 @@ func (m *machine) Format() string {
func (m *machine) encodeWithoutSSA(root *instruction) {
m.labelResolutionPends = m.labelResolutionPends[:0]
ectx := m.ectx
bufPtr := m.c.BufPtr()
for cur := root; cur != nil; cur = cur.next {
offset := int64(len(*bufPtr))
if cur.kind == nop0 {
l := cur.nop0Label()
if int(l) >= len(ectx.LabelPositions) {
continue
}
if pos := ectx.LabelPositions[l]; pos != nil {
pos.BinaryOffset = offset
pos := m.labelPositionPool.Get(int(l))
if pos != nil {
pos.binaryOffset = offset
}
}
@ -1973,7 +2095,7 @@ func (m *machine) encodeWithoutSSA(root *instruction) {
switch p.instr.kind {
case jmp, jmpIf, lea:
target := p.instr.jmpLabel()
targetOffset := ectx.LabelPositions[target].BinaryOffset
targetOffset := m.labelPositionPool.Get(int(target)).binaryOffset
imm32Offset := p.imm32Offset
jmpOffset := int32(targetOffset - (p.imm32Offset + 4)) // +4 because RIP points to the next instruction.
binary.LittleEndian.PutUint32((*bufPtr)[imm32Offset:], uint32(jmpOffset))
@ -1985,33 +2107,33 @@ func (m *machine) encodeWithoutSSA(root *instruction) {
// Encode implements backend.Machine Encode.
func (m *machine) Encode(ctx context.Context) (err error) {
ectx := m.ectx
bufPtr := m.c.BufPtr()
var fn string
var fnIndex int
var labelToSSABlockID map[backend.Label]ssa.BasicBlockID
var labelPosToLabel map[*labelPosition]label
if wazevoapi.PerfMapEnabled {
fn = wazevoapi.GetCurrentFunctionName(ctx)
labelToSSABlockID = make(map[backend.Label]ssa.BasicBlockID)
for i, l := range ectx.SsaBlockIDToLabels {
labelToSSABlockID[l] = ssa.BasicBlockID(i)
labelPosToLabel = make(map[*labelPosition]label)
for i := 0; i <= m.labelPositionPool.MaxIDEncountered(); i++ {
pos := m.labelPositionPool.Get(i)
labelPosToLabel[pos] = label(i)
}
fnIndex = wazevoapi.GetCurrentFunctionIndex(ctx)
}
m.labelResolutionPends = m.labelResolutionPends[:0]
for _, pos := range ectx.OrderedBlockLabels {
for _, pos := range m.orderedSSABlockLabelPos {
offset := int64(len(*bufPtr))
pos.BinaryOffset = offset
for cur := pos.Begin; cur != pos.End.next; cur = cur.next {
pos.binaryOffset = offset
for cur := pos.begin; cur != pos.end.next; cur = cur.next {
offset := int64(len(*bufPtr))
switch cur.kind {
case nop0:
l := cur.nop0Label()
if pos := ectx.LabelPositions[l]; pos != nil {
pos.BinaryOffset = offset
if pos := m.labelPositionPool.Get(int(l)); pos != nil {
pos.binaryOffset = offset
}
case sourceOffsetInfo:
m.c.AddSourceOffsetInfo(offset, cur.sourceOffsetInfo())
@ -2026,22 +2148,16 @@ func (m *machine) Encode(ctx context.Context) (err error) {
}
if wazevoapi.PerfMapEnabled {
l := pos.L
var labelStr string
if blkID, ok := labelToSSABlockID[l]; ok {
labelStr = fmt.Sprintf("%s::SSA_Block[%s]", l, blkID)
} else {
labelStr = l.String()
}
l := labelPosToLabel[pos]
size := int64(len(*bufPtr)) - offset
wazevoapi.PerfMap.AddModuleEntry(fnIndex, offset, uint64(size), fmt.Sprintf("%s:::::%s", fn, labelStr))
wazevoapi.PerfMap.AddModuleEntry(fnIndex, offset, uint64(size), fmt.Sprintf("%s:::::%s", fn, l))
}
}
for i := range m.consts {
offset := int64(len(*bufPtr))
vc := &m.consts[i]
vc.label.BinaryOffset = offset
vc.labelPos.binaryOffset = offset
if vc._var == nil {
lo, hi := vc.lo, vc.hi
m.c.Emit8Bytes(lo)
@ -2059,7 +2175,7 @@ func (m *machine) Encode(ctx context.Context) (err error) {
switch p.instr.kind {
case jmp, jmpIf, lea, xmmUnaryRmR:
target := p.instr.jmpLabel()
targetOffset := ectx.LabelPositions[target].BinaryOffset
targetOffset := m.labelPositionPool.Get(int(target)).binaryOffset
imm32Offset := p.imm32Offset
jmpOffset := int32(targetOffset - (p.imm32Offset + 4)) // +4 because RIP points to the next instruction.
binary.LittleEndian.PutUint32(buf[imm32Offset:], uint32(jmpOffset))
@ -2068,7 +2184,7 @@ func (m *machine) Encode(ctx context.Context) (err error) {
// Each entry is the offset from the beginning of the jmpTableIsland instruction in 8 bytes.
targets := m.jmpTableTargets[p.instr.u1]
for i, l := range targets {
targetOffset := ectx.LabelPositions[backend.Label(l)].BinaryOffset
targetOffset := m.labelPositionPool.Get(int(l)).binaryOffset
jmpOffset := targetOffset - tableBegin
binary.LittleEndian.PutUint64(buf[tableBegin+int64(i)*8:], uint64(jmpOffset))
}
@ -2097,7 +2213,7 @@ func (m *machine) ResolveRelocations(refToBinaryOffset []int, binary []byte, rel
// CallTrampolineIslandInfo implements backend.Machine CallTrampolineIslandInfo.
func (m *machine) CallTrampolineIslandInfo(_ int) (_, _ int, _ error) { return }
func (m *machine) lowerIcmpToFlag(xd, yd *backend.SSAValueDefinition, _64 bool) {
func (m *machine) lowerIcmpToFlag(xd, yd backend.SSAValueDefinition, _64 bool) {
x := m.getOperand_Reg(xd)
y := m.getOperand_Mem_Imm32_Reg(yd)
cmp := m.allocateInstr().asCmpRmiR(true, y, x.reg(), _64)
@ -2140,7 +2256,7 @@ func (m *machine) lowerFcmpToFlags(instr *ssa.Instruction) (f1, f2 cond, and boo
// allocateInstr allocates an instruction.
func (m *machine) allocateInstr() *instruction {
instr := m.ectx.InstructionPool.Allocate()
instr := m.instrPool.Allocate()
if !m.regAllocStarted {
instr.addedBeforeRegAlloc = true
}
@ -2154,24 +2270,22 @@ func (m *machine) allocateNop() *instruction {
}
func (m *machine) insert(i *instruction) {
ectx := m.ectx
ectx.PendingInstructions = append(ectx.PendingInstructions, i)
m.pendingInstructions = append(m.pendingInstructions, i)
}
func (m *machine) allocateBrTarget() (nop *instruction, l backend.Label) { //nolint
pos := m.allocateLabel()
l = pos.L
func (m *machine) allocateBrTarget() (nop *instruction, l label) { //nolint
l, pos := m.allocateLabel()
nop = m.allocateInstr()
nop.asNop0WithLabel(l)
pos.Begin, pos.End = nop, nop
pos.begin, pos.end = nop, nop
return
}
func (m *machine) allocateLabel() *labelPosition {
ectx := m.ectx
l := ectx.AllocateLabel()
pos := ectx.GetOrAllocateLabelPosition(l)
return pos
func (m *machine) allocateLabel() (label, *labelPosition) {
l := m.nextLabel
pos := m.labelPositionPool.GetOrAllocate(int(l))
m.nextLabel++
return l, pos
}
func (m *machine) getVRegSpillSlotOffsetFromSP(id regalloc.VRegID, size byte) int64 {
@ -3185,22 +3299,22 @@ func (m *machine) lowerShuffle(x, y ssa.Value, lo, hi uint64, ret ssa.Value) {
}
}
xmaskLabel := m.allocateLabel()
m.consts = append(m.consts, _const{lo: xMask[0], hi: xMask[1], label: xmaskLabel})
ymaskLabel := m.allocateLabel()
m.consts = append(m.consts, _const{lo: yMask[0], hi: yMask[1], label: ymaskLabel})
xl, xmaskPos := m.allocateLabel()
m.consts = append(m.consts, _const{lo: xMask[0], hi: xMask[1], label: xl, labelPos: xmaskPos})
yl, ymaskPos := m.allocateLabel()
m.consts = append(m.consts, _const{lo: yMask[0], hi: yMask[1], label: yl, labelPos: ymaskPos})
xx, yy := m.getOperand_Reg(m.c.ValueDefinition(x)), m.getOperand_Reg(m.c.ValueDefinition(y))
tmpX, tmpY := m.copyToTmp(xx.reg()), m.copyToTmp(yy.reg())
// Apply mask to X.
tmp := m.c.AllocateVReg(ssa.TypeV128)
loadMaskLo := m.allocateInstr().asXmmUnaryRmR(sseOpcodeMovdqu, newOperandMem(m.newAmodeRipRel(xmaskLabel.L)), tmp)
loadMaskLo := m.allocateInstr().asXmmUnaryRmR(sseOpcodeMovdqu, newOperandMem(m.newAmodeRipRel(xl)), tmp)
m.insert(loadMaskLo)
m.insert(m.allocateInstr().asXmmRmR(sseOpcodePshufb, newOperandReg(tmp), tmpX))
// Apply mask to Y.
loadMaskHi := m.allocateInstr().asXmmUnaryRmR(sseOpcodeMovdqu, newOperandMem(m.newAmodeRipRel(ymaskLabel.L)), tmp)
loadMaskHi := m.allocateInstr().asXmmUnaryRmR(sseOpcodeMovdqu, newOperandMem(m.newAmodeRipRel(yl)), tmp)
m.insert(loadMaskHi)
m.insert(m.allocateInstr().asXmmRmR(sseOpcodePshufb, newOperandReg(tmp), tmpY))

View file

@ -12,7 +12,7 @@ func (m *machine) PostRegAlloc() {
}
func (m *machine) setupPrologue() {
cur := m.ectx.RootInstr
cur := m.rootInstr
prevInitInst := cur.next
// At this point, we have the stack layout as follows:
@ -130,14 +130,13 @@ func (m *machine) setupPrologue() {
// 3. Inserts the dec/inc RSP instruction right before/after the call instruction.
// 4. Lowering that is supposed to be done after regalloc.
func (m *machine) postRegAlloc() {
ectx := m.ectx
for cur := ectx.RootInstr; cur != nil; cur = cur.next {
for cur := m.rootInstr; cur != nil; cur = cur.next {
switch k := cur.kind; k {
case ret:
m.setupEpilogueAfter(cur.prev)
continue
case fcvtToSintSequence, fcvtToUintSequence:
m.ectx.PendingInstructions = m.ectx.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
if k == fcvtToSintSequence {
m.lowerFcvtToSintSequenceAfterRegalloc(cur)
} else {
@ -146,29 +145,29 @@ func (m *machine) postRegAlloc() {
prev := cur.prev
next := cur.next
cur := prev
for _, instr := range m.ectx.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
linkInstr(cur, next)
continue
case xmmCMov:
m.ectx.PendingInstructions = m.ectx.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
m.lowerXmmCmovAfterRegAlloc(cur)
prev := cur.prev
next := cur.next
cur := prev
for _, instr := range m.ectx.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
linkInstr(cur, next)
continue
case idivRemSequence:
m.ectx.PendingInstructions = m.ectx.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
m.lowerIDivRemSequenceAfterRegAlloc(cur)
prev := cur.prev
next := cur.next
cur := prev
for _, instr := range m.ectx.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
linkInstr(cur, next)

View file

@ -1,13 +1,226 @@
package amd64
import (
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend"
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend/regalloc"
"github.com/tetratelabs/wazero/internal/engine/wazevo/ssa"
)
// InsertMoveBefore implements backend.RegAllocFunctionMachine.
func (m *machine) InsertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
// regAllocFn implements regalloc.Function.
type regAllocFn struct {
ssaB ssa.Builder
m *machine
loopNestingForestRoots []ssa.BasicBlock
blockIter int
}
// PostOrderBlockIteratorBegin implements regalloc.Function.
func (f *regAllocFn) PostOrderBlockIteratorBegin() *labelPosition {
f.blockIter = len(f.m.orderedSSABlockLabelPos) - 1
return f.PostOrderBlockIteratorNext()
}
// PostOrderBlockIteratorNext implements regalloc.Function.
func (f *regAllocFn) PostOrderBlockIteratorNext() *labelPosition {
if f.blockIter < 0 {
return nil
}
b := f.m.orderedSSABlockLabelPos[f.blockIter]
f.blockIter--
return b
}
// ReversePostOrderBlockIteratorBegin implements regalloc.Function.
func (f *regAllocFn) ReversePostOrderBlockIteratorBegin() *labelPosition {
f.blockIter = 0
return f.ReversePostOrderBlockIteratorNext()
}
// ReversePostOrderBlockIteratorNext implements regalloc.Function.
func (f *regAllocFn) ReversePostOrderBlockIteratorNext() *labelPosition {
if f.blockIter >= len(f.m.orderedSSABlockLabelPos) {
return nil
}
b := f.m.orderedSSABlockLabelPos[f.blockIter]
f.blockIter++
return b
}
// ClobberedRegisters implements regalloc.Function.
func (f *regAllocFn) ClobberedRegisters(regs []regalloc.VReg) {
f.m.clobberedRegs = append(f.m.clobberedRegs[:0], regs...)
}
// LoopNestingForestRoots implements regalloc.Function.
func (f *regAllocFn) LoopNestingForestRoots() int {
f.loopNestingForestRoots = f.ssaB.LoopNestingForestRoots()
return len(f.loopNestingForestRoots)
}
// LoopNestingForestRoot implements regalloc.Function.
func (f *regAllocFn) LoopNestingForestRoot(i int) *labelPosition {
root := f.loopNestingForestRoots[i]
pos := f.m.getOrAllocateSSABlockLabelPosition(root)
return pos
}
// LowestCommonAncestor implements regalloc.Function.
func (f *regAllocFn) LowestCommonAncestor(blk1, blk2 *labelPosition) *labelPosition {
sb := f.ssaB.LowestCommonAncestor(blk1.sb, blk2.sb)
pos := f.m.getOrAllocateSSABlockLabelPosition(sb)
return pos
}
// Idom implements regalloc.Function.
func (f *regAllocFn) Idom(blk *labelPosition) *labelPosition {
sb := f.ssaB.Idom(blk.sb)
pos := f.m.getOrAllocateSSABlockLabelPosition(sb)
return pos
}
// SwapBefore implements regalloc.Function.
func (f *regAllocFn) SwapBefore(x1, x2, tmp regalloc.VReg, instr *instruction) {
f.m.swap(instr.prev, x1, x2, tmp)
}
// StoreRegisterBefore implements regalloc.Function.
func (f *regAllocFn) StoreRegisterBefore(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertStoreRegisterAt(v, instr, false)
}
// StoreRegisterAfter implements regalloc.Function.
func (f *regAllocFn) StoreRegisterAfter(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertStoreRegisterAt(v, instr, true)
}
// ReloadRegisterBefore implements regalloc.Function.
func (f *regAllocFn) ReloadRegisterBefore(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertReloadRegisterAt(v, instr, false)
}
// ReloadRegisterAfter implements regalloc.Function.
func (f *regAllocFn) ReloadRegisterAfter(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertReloadRegisterAt(v, instr, true)
}
// InsertMoveBefore implements regalloc.Function.
func (f *regAllocFn) InsertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
f.m.insertMoveBefore(dst, src, instr)
}
// LoopNestingForestChild implements regalloc.Function.
func (f *regAllocFn) LoopNestingForestChild(pos *labelPosition, i int) *labelPosition {
childSB := pos.sb.LoopNestingForestChildren()[i]
return f.m.getOrAllocateSSABlockLabelPosition(childSB)
}
// Succ implements regalloc.Block.
func (f *regAllocFn) Succ(pos *labelPosition, i int) *labelPosition {
succSB := pos.sb.Succ(i)
if succSB.ReturnBlock() {
return nil
}
return f.m.getOrAllocateSSABlockLabelPosition(succSB)
}
// Pred implements regalloc.Block.
func (f *regAllocFn) Pred(pos *labelPosition, i int) *labelPosition {
predSB := pos.sb.Pred(i)
return f.m.getOrAllocateSSABlockLabelPosition(predSB)
}
// BlockParams implements regalloc.Function.
func (f *regAllocFn) BlockParams(pos *labelPosition, regs *[]regalloc.VReg) []regalloc.VReg {
c := f.m.c
*regs = (*regs)[:0]
for i := 0; i < pos.sb.Params(); i++ {
v := c.VRegOf(pos.sb.Param(i))
*regs = append(*regs, v)
}
return *regs
}
// ID implements regalloc.Block.
func (pos *labelPosition) ID() int32 {
return int32(pos.sb.ID())
}
// InstrIteratorBegin implements regalloc.Block.
func (pos *labelPosition) InstrIteratorBegin() *instruction {
ret := pos.begin
pos.cur = ret
return ret
}
// InstrIteratorNext implements regalloc.Block.
func (pos *labelPosition) InstrIteratorNext() *instruction {
for {
if pos.cur == pos.end {
return nil
}
instr := pos.cur.next
pos.cur = instr
if instr == nil {
return nil
} else if instr.addedBeforeRegAlloc {
// Only concerned about the instruction added before regalloc.
return instr
}
}
}
// InstrRevIteratorBegin implements regalloc.Block.
func (pos *labelPosition) InstrRevIteratorBegin() *instruction {
pos.cur = pos.end
return pos.cur
}
// InstrRevIteratorNext implements regalloc.Block.
func (pos *labelPosition) InstrRevIteratorNext() *instruction {
for {
if pos.cur == pos.begin {
return nil
}
instr := pos.cur.prev
pos.cur = instr
if instr == nil {
return nil
} else if instr.addedBeforeRegAlloc {
// Only concerned about the instruction added before regalloc.
return instr
}
}
}
// FirstInstr implements regalloc.Block.
func (pos *labelPosition) FirstInstr() *instruction { return pos.begin }
// LastInstrForInsertion implements regalloc.Block.
func (pos *labelPosition) LastInstrForInsertion() *instruction {
return lastInstrForInsertion(pos.begin, pos.end)
}
// Preds implements regalloc.Block.
func (pos *labelPosition) Preds() int { return pos.sb.Preds() }
// Entry implements regalloc.Block.
func (pos *labelPosition) Entry() bool { return pos.sb.EntryBlock() }
// Succs implements regalloc.Block.
func (pos *labelPosition) Succs() int { return pos.sb.Succs() }
// LoopHeader implements regalloc.Block.
func (pos *labelPosition) LoopHeader() bool { return pos.sb.LoopHeader() }
// LoopNestingForestChildren implements regalloc.Block.
func (pos *labelPosition) LoopNestingForestChildren() int {
return len(pos.sb.LoopNestingForestChildren())
}
func (m *machine) insertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
typ := src.RegType()
if typ != dst.RegType() {
panic("BUG: src and dst must have the same type")
@ -26,8 +239,7 @@ func (m *machine) InsertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
linkInstr(cur, prevNext)
}
// InsertStoreRegisterAt implements backend.RegAllocFunctionMachine.
func (m *machine) InsertStoreRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
func (m *machine) insertStoreRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
if !v.IsRealReg() {
panic("BUG: VReg must be backed by real reg to be stored")
}
@ -61,8 +273,7 @@ func (m *machine) InsertStoreRegisterAt(v regalloc.VReg, instr *instruction, aft
return linkInstr(cur, prevNext)
}
// InsertReloadRegisterAt implements backend.RegAllocFunctionMachine.
func (m *machine) InsertReloadRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
func (m *machine) insertReloadRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
if !v.IsRealReg() {
panic("BUG: VReg must be backed by real reg to be stored")
}
@ -98,13 +309,7 @@ func (m *machine) InsertReloadRegisterAt(v regalloc.VReg, instr *instruction, af
return linkInstr(cur, prevNext)
}
// ClobberedRegisters implements backend.RegAllocFunctionMachine.
func (m *machine) ClobberedRegisters(regs []regalloc.VReg) {
m.clobberedRegs = append(m.clobberedRegs[:0], regs...)
}
// Swap implements backend.RegAllocFunctionMachine.
func (m *machine) Swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
func (m *machine) swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
if x1.RegType() == regalloc.RegTypeInt {
prevNext := cur.next
xc := m.allocateInstr().asXCHG(x1, newOperandReg(x2), 8)
@ -113,25 +318,24 @@ func (m *machine) Swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
} else {
if tmp.Valid() {
prevNext := cur.next
m.InsertMoveBefore(tmp, x1, prevNext)
m.InsertMoveBefore(x1, x2, prevNext)
m.InsertMoveBefore(x2, tmp, prevNext)
m.insertMoveBefore(tmp, x1, prevNext)
m.insertMoveBefore(x1, x2, prevNext)
m.insertMoveBefore(x2, tmp, prevNext)
} else {
prevNext := cur.next
r2 := x2.RealReg()
// Temporarily spill x1 to stack.
cur = m.InsertStoreRegisterAt(x1, cur, true).prev
cur = m.insertStoreRegisterAt(x1, cur, true).prev
// Then move x2 to x1.
cur = linkInstr(cur, m.allocateInstr().asXmmUnaryRmR(sseOpcodeMovdqa, newOperandReg(x2), x1))
linkInstr(cur, prevNext)
// Then reload the original value on x1 from stack to r2.
m.InsertReloadRegisterAt(x1.SetRealReg(r2), cur, true)
m.insertReloadRegisterAt(x1.SetRealReg(r2), cur, true)
}
}
}
// LastInstrForInsertion implements backend.RegAllocFunctionMachine.
func (m *machine) LastInstrForInsertion(begin, end *instruction) *instruction {
func lastInstrForInsertion(begin, end *instruction) *instruction {
cur := end
for cur.kind == nop0 {
cur = cur.prev
@ -146,8 +350,3 @@ func (m *machine) LastInstrForInsertion(begin, end *instruction) *instruction {
return end
}
}
// SSABlockLabel implements backend.RegAllocFunctionMachine.
func (m *machine) SSABlockLabel(id ssa.BasicBlockID) backend.Label {
return m.ectx.SsaBlockIDToLabels[id]
}

View file

@ -127,7 +127,7 @@ func (m *machine) lowerSqmulRoundSat(x, y, ret ssa.Value) {
tmpX := m.copyToTmp(xx.reg())
m.insert(m.allocateInstr().asXmmRmR(sseOpcodePmulhrsw, yy, tmpX))
m.insert(m.allocateInstr().asXmmRmR(sseOpcodePcmpeqd, newOperandReg(tmpX), tmp))
m.insert(m.allocateInstr().asXmmRmR(sseOpcodePcmpeqw, newOperandReg(tmpX), tmp))
m.insert(m.allocateInstr().asXmmRmR(sseOpcodePxor, newOperandReg(tmp), tmpX))
m.copyTo(tmpX, m.c.VRegOf(ret))

View file

@ -59,7 +59,7 @@ func (o *operand) format(_64 bool) string {
case operandKindImm32:
return fmt.Sprintf("$%d", int32(o.imm32()))
case operandKindLabel:
return backend.Label(o.imm32()).String()
return label(o.imm32()).String()
default:
panic(fmt.Sprintf("BUG: invalid operand: %s", o.kind))
}
@ -85,22 +85,22 @@ func (o *operand) imm32() uint32 {
return uint32(o.data)
}
func (o *operand) label() backend.Label {
func (o *operand) label() label {
switch o.kind {
case operandKindLabel:
return backend.Label(o.data)
return label(o.data)
case operandKindMem:
mem := o.addressMode()
if mem.kind() != amodeRipRel {
panic("BUG: invalid label")
}
return backend.Label(mem.imm32)
return label(mem.imm32)
default:
panic("BUG: invalid operand kind")
}
}
func newOperandLabel(label backend.Label) operand {
func newOperandLabel(label label) operand {
return operand{kind: operandKindLabel, data: uint64(label)}
}
@ -221,7 +221,7 @@ func (m *machine) newAmodeRegRegShift(imm32 uint32, base, index regalloc.VReg, s
return ret
}
func (m *machine) newAmodeRipRel(label backend.Label) *amode {
func (m *machine) newAmodeRipRel(label label) *amode {
ret := m.amodePool.Allocate()
*ret = amode{kindWithShift: uint32(amodeRipRel), imm32: uint32(label)}
return ret
@ -246,18 +246,18 @@ func (a *amode) String() string {
"%d(%s,%s,%d)",
int32(a.imm32), formatVRegSized(a.base, true), formatVRegSized(a.index, true), shift)
case amodeRipRel:
return fmt.Sprintf("%s(%%rip)", backend.Label(a.imm32))
return fmt.Sprintf("%s(%%rip)", label(a.imm32))
default:
panic("BUG: invalid amode kind")
}
}
func (m *machine) getOperand_Mem_Reg(def *backend.SSAValueDefinition) (op operand) {
if def.IsFromBlockParam() {
return newOperandReg(def.BlkParamVReg)
func (m *machine) getOperand_Mem_Reg(def backend.SSAValueDefinition) (op operand) {
if !def.IsFromInstr() {
return newOperandReg(m.c.VRegOf(def.V))
}
if def.SSAValue().Type() == ssa.TypeV128 {
if def.V.Type() == ssa.TypeV128 {
// SIMD instructions require strict memory alignment, so we don't support the memory operand for V128 at the moment.
return m.getOperand_Reg(def)
}
@ -272,9 +272,9 @@ func (m *machine) getOperand_Mem_Reg(def *backend.SSAValueDefinition) (op operan
return m.getOperand_Reg(def)
}
func (m *machine) getOperand_Mem_Imm32_Reg(def *backend.SSAValueDefinition) (op operand) {
if def.IsFromBlockParam() {
return newOperandReg(def.BlkParamVReg)
func (m *machine) getOperand_Mem_Imm32_Reg(def backend.SSAValueDefinition) (op operand) {
if !def.IsFromInstr() {
return newOperandReg(m.c.VRegOf(def.V))
}
if m.c.MatchInstr(def, ssa.OpcodeLoad) {
@ -287,9 +287,9 @@ func (m *machine) getOperand_Mem_Imm32_Reg(def *backend.SSAValueDefinition) (op
return m.getOperand_Imm32_Reg(def)
}
func (m *machine) getOperand_Imm32_Reg(def *backend.SSAValueDefinition) (op operand) {
if def.IsFromBlockParam() {
return newOperandReg(def.BlkParamVReg)
func (m *machine) getOperand_Imm32_Reg(def backend.SSAValueDefinition) (op operand) {
if !def.IsFromInstr() {
return newOperandReg(m.c.VRegOf(def.V))
}
instr := def.Instr
@ -323,24 +323,14 @@ func asImm32(val uint64, allowSignExt bool) (uint32, bool) {
return u32val, true
}
func (m *machine) getOperand_Reg(def *backend.SSAValueDefinition) (op operand) {
func (m *machine) getOperand_Reg(def backend.SSAValueDefinition) (op operand) {
var v regalloc.VReg
if def.IsFromBlockParam() {
v = def.BlkParamVReg
if instr := def.Instr; instr != nil && instr.Constant() {
// We inline all the constant instructions so that we could reduce the register usage.
v = m.lowerConstant(instr)
instr.MarkLowered()
} else {
instr := def.Instr
if instr.Constant() {
// We inline all the constant instructions so that we could reduce the register usage.
v = m.lowerConstant(instr)
instr.MarkLowered()
} else {
if n := def.N; n == 0 {
v = m.c.VRegOf(instr.Return())
} else {
_, rs := instr.Returns()
v = m.c.VRegOf(rs[n-1])
}
}
v = m.c.VRegOf(def.V)
}
return newOperandReg(v)
}

View file

@ -1,11 +0,0 @@
//go:build !tinygo
package amd64
import "reflect"
// setSliceLimits sets both Cap and Len for the given reflected slice.
func setSliceLimits(s *reflect.SliceHeader, limit uintptr) {
s.Len = int(limit)
s.Cap = int(limit)
}

View file

@ -1,11 +0,0 @@
//go:build tinygo
package amd64
import "reflect"
// setSliceLimits sets both Cap and Len for the given reflected slice.
func setSliceLimits(s *reflect.SliceHeader, limit uintptr) {
s.Len = limit
s.Len = limit
}

View file

@ -9,12 +9,14 @@
)
func stackView(rbp, top uintptr) []byte {
l := int(top - rbp)
var stackBuf []byte
{
// TODO: use unsafe.Slice after floor version is set to Go 1.20.
//nolint:staticcheck
hdr := (*reflect.SliceHeader)(unsafe.Pointer(&stackBuf))
hdr.Data = rbp
setSliceLimits(hdr, top-rbp)
hdr.Len = l
hdr.Cap = l
}
return stackBuf
}
@ -72,9 +74,9 @@ func GoCallStackView(stackPointerBeforeGoCall *uint64) []uint64 {
// | SizeInBytes |
// +-----------------+ <---- stackPointerBeforeGoCall
// (low address)
data := unsafe.Pointer(uintptr(unsafe.Pointer(stackPointerBeforeGoCall)) + 8)
data := unsafe.Add(unsafe.Pointer(stackPointerBeforeGoCall), 8)
size := *stackPointerBeforeGoCall / 8
return unsafe.Slice((*uint64)(data), int(size))
return unsafe.Slice((*uint64)(data), size)
}
func AdjustClonedStack(oldRsp, oldTop, rsp, rbp, top uintptr) {

View file

@ -182,9 +182,9 @@ func (m *machine) LowerReturns(rets []ssa.Value) {
// callerGenVRegToFunctionArg is the opposite of GenFunctionArgToVReg, which is used to generate the
// caller side of the function call.
func (m *machine) callerGenVRegToFunctionArg(a *backend.FunctionABI, argIndex int, reg regalloc.VReg, def *backend.SSAValueDefinition, slotBegin int64) {
func (m *machine) callerGenVRegToFunctionArg(a *backend.FunctionABI, argIndex int, reg regalloc.VReg, def backend.SSAValueDefinition, slotBegin int64) {
arg := &a.Args[argIndex]
if def != nil && def.IsFromInstr() {
if def.IsFromInstr() {
// Constant instructions are inlined.
if inst := def.Instr; inst.Constant() {
val := inst.Return()
@ -228,10 +228,9 @@ func (m *machine) callerGenFunctionReturnVReg(a *backend.FunctionABI, retIndex i
}
func (m *machine) resolveAddressModeForOffsetAndInsert(cur *instruction, offset int64, dstBits byte, rn regalloc.VReg, allowTmpRegUse bool) (*instruction, *addressMode) {
exct := m.executableContext
exct.PendingInstructions = exct.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
mode := m.resolveAddressModeForOffset(offset, dstBits, rn, allowTmpRegUse)
for _, instr := range exct.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
return cur, mode

View file

@ -14,7 +14,6 @@
// CompileGoFunctionTrampoline implements backend.Machine.
func (m *machine) CompileGoFunctionTrampoline(exitCode wazevoapi.ExitCode, sig *ssa.Signature, needModuleContextPtr bool) []byte {
exct := m.executableContext
argBegin := 1 // Skips exec context by default.
if needModuleContextPtr {
argBegin++
@ -26,7 +25,7 @@ func (m *machine) CompileGoFunctionTrampoline(exitCode wazevoapi.ExitCode, sig *
cur := m.allocateInstr()
cur.asNop0()
exct.RootInstr = cur
m.rootInstr = cur
// Execution context is always the first argument.
execCtrPtr := x0VReg
@ -244,7 +243,7 @@ func (m *machine) CompileGoFunctionTrampoline(exitCode wazevoapi.ExitCode, sig *
ret.asRet()
linkInstr(cur, ret)
m.encode(m.executableContext.RootInstr)
m.encode(m.rootInstr)
return m.compiler.Buf()
}
@ -302,20 +301,18 @@ func (m *machine) restoreRegistersInExecutionContext(cur *instruction, regs []re
}
func (m *machine) lowerConstantI64AndInsert(cur *instruction, dst regalloc.VReg, v int64) *instruction {
exct := m.executableContext
exct.PendingInstructions = exct.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
m.lowerConstantI64(dst, v)
for _, instr := range exct.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
return cur
}
func (m *machine) lowerConstantI32AndInsert(cur *instruction, dst regalloc.VReg, v int32) *instruction {
exct := m.executableContext
exct.PendingInstructions = exct.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
m.lowerConstantI32(dst, v)
for _, instr := range exct.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
return cur

View file

@ -36,18 +36,6 @@
instructionKind byte
)
func asNop0(i *instruction) {
i.kind = nop0
}
func setNext(i, next *instruction) {
i.next = next
}
func setPrev(i, prev *instruction) {
i.prev = prev
}
// IsCall implements regalloc.Instr IsCall.
func (i *instruction) IsCall() bool {
return i.kind == call
@ -63,21 +51,6 @@ func (i *instruction) IsReturn() bool {
return i.kind == ret
}
// Next implements regalloc.Instr Next.
func (i *instruction) Next() regalloc.Instr {
return i.next
}
// Prev implements regalloc.Instr Prev.
func (i *instruction) Prev() regalloc.Instr {
return i.prev
}
// AddedBeforeRegAlloc implements regalloc.Instr AddedBeforeRegAlloc.
func (i *instruction) AddedBeforeRegAlloc() bool {
return i.addedBeforeRegAlloc
}
type defKind byte
const (

View file

@ -12,7 +12,7 @@
// Encode implements backend.Machine Encode.
func (m *machine) Encode(ctx context.Context) error {
m.resolveRelativeAddresses(ctx)
m.encode(m.executableContext.RootInstr)
m.encode(m.rootInstr)
if l := len(m.compiler.Buf()); l > maxFunctionExecutableSize {
return fmt.Errorf("function size exceeds the limit: %d > %d", l, maxFunctionExecutableSize)
}

View file

@ -17,19 +17,18 @@
// LowerSingleBranch implements backend.Machine.
func (m *machine) LowerSingleBranch(br *ssa.Instruction) {
ectx := m.executableContext
switch br.Opcode() {
case ssa.OpcodeJump:
_, _, targetBlk := br.BranchData()
_, _, targetBlkID := br.BranchData()
if br.IsFallthroughJump() {
return
}
b := m.allocateInstr()
target := ectx.GetOrAllocateSSABlockLabel(targetBlk)
if target == labelReturn {
targetBlk := m.compiler.SSABuilder().BasicBlock(targetBlkID)
if targetBlk.ReturnBlock() {
b.asRet()
} else {
b.asBr(target)
b.asBr(ssaBlockLabel(targetBlk))
}
m.insert(b)
case ssa.OpcodeBrTable:
@ -40,7 +39,8 @@ func (m *machine) LowerSingleBranch(br *ssa.Instruction) {
}
func (m *machine) lowerBrTable(i *ssa.Instruction) {
index, targets := i.BrTableData()
index, targetBlockIDs := i.BrTableData()
targetBlockCount := len(targetBlockIDs.View())
indexOperand := m.getOperand_NR(m.compiler.ValueDefinition(index), extModeNone)
// Firstly, we have to do the bounds check of the index, and
@ -50,7 +50,7 @@ func (m *machine) lowerBrTable(i *ssa.Instruction) {
// subs wzr, index, maxIndexReg
// csel adjustedIndex, maxIndexReg, index, hs ;; if index is higher or equal than maxIndexReg.
maxIndexReg := m.compiler.AllocateVReg(ssa.TypeI32)
m.lowerConstantI32(maxIndexReg, int32(len(targets)-1))
m.lowerConstantI32(maxIndexReg, int32(targetBlockCount-1))
subs := m.allocateInstr()
subs.asALU(aluOpSubS, xzrVReg, indexOperand, operandNR(maxIndexReg), false)
m.insert(subs)
@ -61,24 +61,24 @@ func (m *machine) lowerBrTable(i *ssa.Instruction) {
brSequence := m.allocateInstr()
tableIndex := m.addJmpTableTarget(targets)
brSequence.asBrTableSequence(adjustedIndex, tableIndex, len(targets))
tableIndex := m.addJmpTableTarget(targetBlockIDs)
brSequence.asBrTableSequence(adjustedIndex, tableIndex, targetBlockCount)
m.insert(brSequence)
}
// LowerConditionalBranch implements backend.Machine.
func (m *machine) LowerConditionalBranch(b *ssa.Instruction) {
exctx := m.executableContext
cval, args, targetBlk := b.BranchData()
cval, args, targetBlkID := b.BranchData()
if len(args) > 0 {
panic(fmt.Sprintf(
"conditional branch shouldn't have args; likely a bug in critical edge splitting: from %s to %s",
exctx.CurrentSSABlk,
targetBlk,
m.currentLabelPos.sb,
targetBlkID,
))
}
target := exctx.GetOrAllocateSSABlockLabel(targetBlk)
targetBlk := m.compiler.SSABuilder().BasicBlock(targetBlkID)
target := ssaBlockLabel(targetBlk)
cvalDef := m.compiler.ValueDefinition(cval)
switch {
@ -791,7 +791,7 @@ func (m *machine) LowerInstr(instr *ssa.Instruction) {
default:
panic("TODO: lowering " + op.String())
}
m.executableContext.FlushPendingInstructions()
m.FlushPendingInstructions()
}
func (m *machine) lowerShuffle(rd regalloc.VReg, rn, rm operand, lane1, lane2 uint64) {

View file

@ -162,9 +162,9 @@ func (o operand) assignReg(v regalloc.VReg) operand {
//
// `mode` is used to extend the operand if the bit length is smaller than mode.bits().
// If the operand can be expressed as operandKindImm12, `mode` is ignored.
func (m *machine) getOperand_Imm12_ER_SR_NR(def *backend.SSAValueDefinition, mode extMode) (op operand) {
if def.IsFromBlockParam() {
return operandNR(def.BlkParamVReg)
func (m *machine) getOperand_Imm12_ER_SR_NR(def backend.SSAValueDefinition, mode extMode) (op operand) {
if !def.IsFromInstr() {
return operandNR(m.compiler.VRegOf(def.V))
}
instr := def.Instr
@ -179,9 +179,9 @@ func (m *machine) getOperand_Imm12_ER_SR_NR(def *backend.SSAValueDefinition, mod
// getOperand_MaybeNegatedImm12_ER_SR_NR is almost the same as getOperand_Imm12_ER_SR_NR, but this might negate the immediate value.
// If the immediate value is negated, the second return value is true, otherwise always false.
func (m *machine) getOperand_MaybeNegatedImm12_ER_SR_NR(def *backend.SSAValueDefinition, mode extMode) (op operand, negatedImm12 bool) {
if def.IsFromBlockParam() {
return operandNR(def.BlkParamVReg), false
func (m *machine) getOperand_MaybeNegatedImm12_ER_SR_NR(def backend.SSAValueDefinition, mode extMode) (op operand, negatedImm12 bool) {
if !def.IsFromInstr() {
return operandNR(m.compiler.VRegOf(def.V)), false
}
instr := def.Instr
@ -193,7 +193,7 @@ func (m *machine) getOperand_MaybeNegatedImm12_ER_SR_NR(def *backend.SSAValueDef
}
signExtended := int64(c)
if def.SSAValue().Type().Bits() == 32 {
if def.V.Type().Bits() == 32 {
signExtended = (signExtended << 32) >> 32
}
negatedWithoutSign := -signExtended
@ -208,9 +208,9 @@ func (m *machine) getOperand_MaybeNegatedImm12_ER_SR_NR(def *backend.SSAValueDef
// ensureValueNR returns an operand of either operandKindER, operandKindSR, or operandKindNR from the given value (defined by `def).
//
// `mode` is used to extend the operand if the bit length is smaller than mode.bits().
func (m *machine) getOperand_ER_SR_NR(def *backend.SSAValueDefinition, mode extMode) (op operand) {
if def.IsFromBlockParam() {
return operandNR(def.BlkParamVReg)
func (m *machine) getOperand_ER_SR_NR(def backend.SSAValueDefinition, mode extMode) (op operand) {
if !def.IsFromInstr() {
return operandNR(m.compiler.VRegOf(def.V))
}
if m.compiler.MatchInstr(def, ssa.OpcodeSExtend) || m.compiler.MatchInstr(def, ssa.OpcodeUExtend) {
@ -251,9 +251,9 @@ func (m *machine) getOperand_ER_SR_NR(def *backend.SSAValueDefinition, mode extM
// ensureValueNR returns an operand of either operandKindSR or operandKindNR from the given value (defined by `def).
//
// `mode` is used to extend the operand if the bit length is smaller than mode.bits().
func (m *machine) getOperand_SR_NR(def *backend.SSAValueDefinition, mode extMode) (op operand) {
if def.IsFromBlockParam() {
return operandNR(def.BlkParamVReg)
func (m *machine) getOperand_SR_NR(def backend.SSAValueDefinition, mode extMode) (op operand) {
if !def.IsFromInstr() {
return operandNR(m.compiler.VRegOf(def.V))
}
if m.compiler.MatchInstr(def, ssa.OpcodeIshl) {
@ -273,9 +273,9 @@ func (m *machine) getOperand_SR_NR(def *backend.SSAValueDefinition, mode extMode
}
// getOperand_ShiftImm_NR returns an operand of either operandKindShiftImm or operandKindNR from the given value (defined by `def).
func (m *machine) getOperand_ShiftImm_NR(def *backend.SSAValueDefinition, mode extMode, shiftBitWidth byte) (op operand) {
if def.IsFromBlockParam() {
return operandNR(def.BlkParamVReg)
func (m *machine) getOperand_ShiftImm_NR(def backend.SSAValueDefinition, mode extMode, shiftBitWidth byte) (op operand) {
if !def.IsFromInstr() {
return operandNR(m.compiler.VRegOf(def.V))
}
instr := def.Instr
@ -289,28 +289,18 @@ func (m *machine) getOperand_ShiftImm_NR(def *backend.SSAValueDefinition, mode e
// ensureValueNR returns an operand of operandKindNR from the given value (defined by `def).
//
// `mode` is used to extend the operand if the bit length is smaller than mode.bits().
func (m *machine) getOperand_NR(def *backend.SSAValueDefinition, mode extMode) (op operand) {
func (m *machine) getOperand_NR(def backend.SSAValueDefinition, mode extMode) (op operand) {
var v regalloc.VReg
if def.IsFromBlockParam() {
v = def.BlkParamVReg
if def.IsFromInstr() && def.Instr.Constant() {
// We inline all the constant instructions so that we could reduce the register usage.
v = m.lowerConstant(def.Instr)
def.Instr.MarkLowered()
} else {
instr := def.Instr
if instr.Constant() {
// We inline all the constant instructions so that we could reduce the register usage.
v = m.lowerConstant(instr)
instr.MarkLowered()
} else {
if n := def.N; n == 0 {
v = m.compiler.VRegOf(instr.Return())
} else {
_, rs := instr.Returns()
v = m.compiler.VRegOf(rs[n-1])
}
}
v = m.compiler.VRegOf(def.V)
}
r := v
switch inBits := def.SSAValue().Type().Bits(); {
switch inBits := def.V.Type().Bits(); {
case mode == extModeNone:
case inBits == 32 && (mode == extModeZeroExtend32 || mode == extModeSignExtend32):
case inBits == 32 && mode == extModeZeroExtend64:

View file

@ -3,6 +3,7 @@
import (
"context"
"fmt"
"math"
"strings"
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend"
@ -14,12 +15,33 @@
type (
// machine implements backend.Machine.
machine struct {
compiler backend.Compiler
executableContext *backend.ExecutableContextT[instruction]
currentABI *backend.FunctionABI
compiler backend.Compiler
currentABI *backend.FunctionABI
instrPool wazevoapi.Pool[instruction]
// labelPositionPool is the pool of labelPosition. The id is the label where
// if the label is less than the maxSSABlockID, it's the ssa.BasicBlockID.
labelPositionPool wazevoapi.IDedPool[labelPosition]
regAlloc regalloc.Allocator
regAllocFn *backend.RegAllocFunction[*instruction, *machine]
// nextLabel is the next label to be allocated. The first free label comes after maxSSABlockID
// so that we can have an identical label for the SSA block ID, which is useful for debugging.
nextLabel label
// rootInstr is the first instruction of the function.
rootInstr *instruction
// currentLabelPos is the currently-compiled ssa.BasicBlock's labelPosition.
currentLabelPos *labelPosition
// orderedSSABlockLabelPos is the ordered list of labelPosition in the generated code for each ssa.BasicBlock.
orderedSSABlockLabelPos []*labelPosition
// returnLabelPos is the labelPosition for the return block.
returnLabelPos labelPosition
// perBlockHead and perBlockEnd are the head and tail of the instruction list per currently-compiled ssa.BasicBlock.
perBlockHead, perBlockEnd *instruction
// pendingInstructions are the instructions which are not yet emitted into the instruction list.
pendingInstructions []*instruction
// maxSSABlockID is the maximum ssa.BasicBlockID in the current function.
maxSSABlockID label
regAlloc regalloc.Allocator[*instruction, *labelPosition, *regAllocFn]
regAllocFn regAllocFn
amodePool wazevoapi.Pool[addressMode]
@ -35,6 +57,8 @@
// jmpTableTargets holds the labels of the jump table targets.
jmpTableTargets [][]uint32
// jmpTableTargetNext is the index to the jmpTableTargets slice to be used for the next jump table.
jmpTableTargetsNext int
// spillSlotSize is the size of the stack slot in bytes used for spilling registers.
// During the execution of the function, the stack looks like:
@ -91,45 +115,132 @@
nextLabel label
offset int64
}
)
labelPosition = backend.LabelPosition[instruction]
label = backend.Label
type (
// label represents a position in the generated code which is either
// a real instruction or the constant InstructionPool (e.g. jump tables).
//
// This is exactly the same as the traditional "label" in assembly code.
label uint32
// labelPosition represents the regions of the generated code which the label represents.
// This implements regalloc.Block.
labelPosition struct {
// sb is not nil if this corresponds to a ssa.BasicBlock.
sb ssa.BasicBlock
// cur is used to walk through the instructions in the block during the register allocation.
cur,
// begin and end are the first and last instructions of the block.
begin, end *instruction
// binaryOffset is the offset in the binary where the label is located.
binaryOffset int64
}
)
const (
labelReturn = backend.LabelReturn
labelInvalid = backend.LabelInvalid
labelReturn label = math.MaxUint32
labelInvalid = labelReturn - 1
)
// String implements backend.Machine.
func (l label) String() string {
return fmt.Sprintf("L%d", l)
}
func resetLabelPosition(l *labelPosition) {
*l = labelPosition{}
}
// NewBackend returns a new backend for arm64.
func NewBackend() backend.Machine {
m := &machine{
spillSlots: make(map[regalloc.VRegID]int64),
executableContext: newExecutableContext(),
regAlloc: regalloc.NewAllocator(regInfo),
regAlloc: regalloc.NewAllocator[*instruction, *labelPosition, *regAllocFn](regInfo),
amodePool: wazevoapi.NewPool[addressMode](resetAddressMode),
instrPool: wazevoapi.NewPool[instruction](resetInstruction),
labelPositionPool: wazevoapi.NewIDedPool[labelPosition](resetLabelPosition),
}
m.regAllocFn.m = m
return m
}
func newExecutableContext() *backend.ExecutableContextT[instruction] {
return backend.NewExecutableContextT[instruction](resetInstruction, setNext, setPrev, asNop0)
func ssaBlockLabel(sb ssa.BasicBlock) label {
if sb.ReturnBlock() {
return labelReturn
}
return label(sb.ID())
}
// ExecutableContext implements backend.Machine.
func (m *machine) ExecutableContext() backend.ExecutableContext {
return m.executableContext
// getOrAllocateSSABlockLabelPosition returns the labelPosition for the given basic block.
func (m *machine) getOrAllocateSSABlockLabelPosition(sb ssa.BasicBlock) *labelPosition {
if sb.ReturnBlock() {
m.returnLabelPos.sb = sb
return &m.returnLabelPos
}
l := ssaBlockLabel(sb)
pos := m.labelPositionPool.GetOrAllocate(int(l))
pos.sb = sb
return pos
}
// LinkAdjacentBlocks implements backend.Machine.
func (m *machine) LinkAdjacentBlocks(prev, next ssa.BasicBlock) {
prevPos, nextPos := m.getOrAllocateSSABlockLabelPosition(prev), m.getOrAllocateSSABlockLabelPosition(next)
prevPos.end.next = nextPos.begin
}
// StartBlock implements backend.Machine.
func (m *machine) StartBlock(blk ssa.BasicBlock) {
m.currentLabelPos = m.getOrAllocateSSABlockLabelPosition(blk)
labelPos := m.currentLabelPos
end := m.allocateNop()
m.perBlockHead, m.perBlockEnd = end, end
labelPos.begin, labelPos.end = end, end
m.orderedSSABlockLabelPos = append(m.orderedSSABlockLabelPos, labelPos)
}
// EndBlock implements ExecutableContext.
func (m *machine) EndBlock() {
// Insert nop0 as the head of the block for convenience to simplify the logic of inserting instructions.
m.insertAtPerBlockHead(m.allocateNop())
m.currentLabelPos.begin = m.perBlockHead
if m.currentLabelPos.sb.EntryBlock() {
m.rootInstr = m.perBlockHead
}
}
func (m *machine) insertAtPerBlockHead(i *instruction) {
if m.perBlockHead == nil {
m.perBlockHead = i
m.perBlockEnd = i
return
}
i.next = m.perBlockHead
m.perBlockHead.prev = i
m.perBlockHead = i
}
// FlushPendingInstructions implements backend.Machine.
func (m *machine) FlushPendingInstructions() {
l := len(m.pendingInstructions)
if l == 0 {
return
}
for i := l - 1; i >= 0; i-- { // reverse because we lower instructions in reverse order.
m.insertAtPerBlockHead(m.pendingInstructions[i])
}
m.pendingInstructions = m.pendingInstructions[:0]
}
// RegAlloc implements backend.Machine Function.
func (m *machine) RegAlloc() {
rf := m.regAllocFn
for _, pos := range m.executableContext.OrderedBlockLabels {
rf.AddBlock(pos.SB, pos.L, pos.Begin, pos.End)
}
m.regAllocStarted = true
m.regAlloc.DoAllocation(rf)
m.regAlloc.DoAllocation(&m.regAllocFn)
// Now that we know the final spill slot size, we must align spillSlotSize to 16 bytes.
m.spillSlotSize = (m.spillSlotSize + 15) &^ 15
}
@ -146,13 +257,22 @@ func (m *machine) Reset() {
m.clobberedRegs = m.clobberedRegs[:0]
m.regAllocStarted = false
m.regAlloc.Reset()
m.regAllocFn.Reset()
m.spillSlotSize = 0
m.unresolvedAddressModes = m.unresolvedAddressModes[:0]
m.maxRequiredStackSizeForCalls = 0
m.executableContext.Reset()
m.jmpTableTargets = m.jmpTableTargets[:0]
m.jmpTableTargetsNext = 0
m.amodePool.Reset()
m.instrPool.Reset()
m.labelPositionPool.Reset()
m.pendingInstructions = m.pendingInstructions[:0]
m.perBlockHead, m.perBlockEnd, m.rootInstr = nil, nil, nil
m.orderedSSABlockLabelPos = m.orderedSSABlockLabelPos[:0]
}
// StartLoweringFunction implements backend.Machine StartLoweringFunction.
func (m *machine) StartLoweringFunction(maxBlockID ssa.BasicBlockID) {
m.maxSSABlockID = label(maxBlockID)
m.nextLabel = label(maxBlockID) + 1
}
// SetCurrentABI implements backend.Machine SetCurrentABI.
@ -168,12 +288,11 @@ func (m *machine) DisableStackCheck() {
// SetCompiler implements backend.Machine.
func (m *machine) SetCompiler(ctx backend.Compiler) {
m.compiler = ctx
m.regAllocFn = backend.NewRegAllocFunction[*instruction, *machine](m, ctx.SSABuilder(), ctx)
m.regAllocFn.ssaB = ctx.SSABuilder()
}
func (m *machine) insert(i *instruction) {
ectx := m.executableContext
ectx.PendingInstructions = append(ectx.PendingInstructions, i)
m.pendingInstructions = append(m.pendingInstructions, i)
}
func (m *machine) insertBrTargetLabel() label {
@ -183,18 +302,18 @@ func (m *machine) insertBrTargetLabel() label {
}
func (m *machine) allocateBrTarget() (nop *instruction, l label) {
ectx := m.executableContext
l = ectx.AllocateLabel()
l = m.nextLabel
m.nextLabel++
nop = m.allocateInstr()
nop.asNop0WithLabel(l)
pos := ectx.GetOrAllocateLabelPosition(l)
pos.Begin, pos.End = nop, nop
pos := m.labelPositionPool.GetOrAllocate(int(l))
pos.begin, pos.end = nop, nop
return
}
// allocateInstr allocates an instruction.
func (m *machine) allocateInstr() *instruction {
instr := m.executableContext.InstructionPool.Allocate()
instr := m.instrPool.Allocate()
if !m.regAllocStarted {
instr.addedBeforeRegAlloc = true
}
@ -251,7 +370,6 @@ func (m *machine) resolveAddressingMode(arg0offset, ret0offset int64, i *instruc
// resolveRelativeAddresses resolves the relative addresses before encoding.
func (m *machine) resolveRelativeAddresses(ctx context.Context) {
ectx := m.executableContext
for {
if len(m.unresolvedAddressModes) > 0 {
arg0offset, ret0offset := m.arg0OffsetFromSP(), m.ret0OffsetFromSP()
@ -265,35 +383,36 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
var fn string
var fnIndex int
var labelToSSABlockID map[label]ssa.BasicBlockID
var labelPosToLabel map[*labelPosition]label
if wazevoapi.PerfMapEnabled {
fn = wazevoapi.GetCurrentFunctionName(ctx)
labelToSSABlockID = make(map[label]ssa.BasicBlockID)
for i, l := range ectx.SsaBlockIDToLabels {
labelToSSABlockID[l] = ssa.BasicBlockID(i)
labelPosToLabel = make(map[*labelPosition]label)
for i := 0; i <= m.labelPositionPool.MaxIDEncountered(); i++ {
labelPosToLabel[m.labelPositionPool.Get(i)] = label(i)
}
fn = wazevoapi.GetCurrentFunctionName(ctx)
fnIndex = wazevoapi.GetCurrentFunctionIndex(ctx)
}
// Next, in order to determine the offsets of relative jumps, we have to calculate the size of each label.
var offset int64
for i, pos := range ectx.OrderedBlockLabels {
pos.BinaryOffset = offset
for i, pos := range m.orderedSSABlockLabelPos {
pos.binaryOffset = offset
var size int64
for cur := pos.Begin; ; cur = cur.next {
for cur := pos.begin; ; cur = cur.next {
switch cur.kind {
case nop0:
l := cur.nop0Label()
if pos := ectx.LabelPositions[l]; pos != nil {
pos.BinaryOffset = offset + size
if pos := m.labelPositionPool.Get(int(l)); pos != nil {
pos.binaryOffset = offset + size
}
case condBr:
if !cur.condBrOffsetResolved() {
var nextLabel label
if i < len(ectx.OrderedBlockLabels)-1 {
if i < len(m.orderedSSABlockLabelPos)-1 {
// Note: this is only used when the block ends with fallthrough,
// therefore can be safely assumed that the next block exists when it's needed.
nextLabel = ectx.OrderedBlockLabels[i+1].L
nextLabel = ssaBlockLabel(m.orderedSSABlockLabelPos[i+1].sb)
}
m.condBrRelocs = append(m.condBrRelocs, condBrReloc{
cbr: cur, currentLabelPos: pos, offset: offset + size,
@ -302,21 +421,14 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
}
}
size += cur.size()
if cur == pos.End {
if cur == pos.end {
break
}
}
if wazevoapi.PerfMapEnabled {
if size > 0 {
l := pos.L
var labelStr string
if blkID, ok := labelToSSABlockID[l]; ok {
labelStr = fmt.Sprintf("%s::SSA_Block[%s]", l, blkID)
} else {
labelStr = l.String()
}
wazevoapi.PerfMap.AddModuleEntry(fnIndex, offset, uint64(size), fmt.Sprintf("%s:::::%s", fn, labelStr))
wazevoapi.PerfMap.AddModuleEntry(fnIndex, offset, uint64(size), fmt.Sprintf("%s:::::%s", fn, labelPosToLabel[pos]))
}
}
offset += size
@ -330,7 +442,7 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
offset := reloc.offset
target := cbr.condBrLabel()
offsetOfTarget := ectx.LabelPositions[target].BinaryOffset
offsetOfTarget := m.labelPositionPool.Get(int(target)).binaryOffset
diff := offsetOfTarget - offset
if divided := diff >> 2; divided < minSignedInt19 || divided > maxSignedInt19 {
// This case the conditional branch is too huge. We place the trampoline instructions at the end of the current block,
@ -351,11 +463,11 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
}
var currentOffset int64
for cur := ectx.RootInstr; cur != nil; cur = cur.next {
for cur := m.rootInstr; cur != nil; cur = cur.next {
switch cur.kind {
case br:
target := cur.brLabel()
offsetOfTarget := ectx.LabelPositions[target].BinaryOffset
offsetOfTarget := m.labelPositionPool.Get(int(target)).binaryOffset
diff := offsetOfTarget - currentOffset
divided := diff >> 2
if divided < minSignedInt26 || divided > maxSignedInt26 {
@ -366,7 +478,7 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
case condBr:
if !cur.condBrOffsetResolved() {
target := cur.condBrLabel()
offsetOfTarget := ectx.LabelPositions[target].BinaryOffset
offsetOfTarget := m.labelPositionPool.Get(int(target)).binaryOffset
diff := offsetOfTarget - currentOffset
if divided := diff >> 2; divided < minSignedInt19 || divided > maxSignedInt19 {
panic("BUG: branch relocation for large conditional branch larger than 19-bit range must be handled properly")
@ -378,7 +490,7 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
targets := m.jmpTableTargets[tableIndex]
for i := range targets {
l := label(targets[i])
offsetOfTarget := ectx.LabelPositions[l].BinaryOffset
offsetOfTarget := m.labelPositionPool.Get(int(l)).binaryOffset
diff := offsetOfTarget - (currentOffset + brTableSequenceOffsetTableBegin)
targets[i] = uint32(diff)
}
@ -399,7 +511,7 @@ func (m *machine) resolveRelativeAddresses(ctx context.Context) {
)
func (m *machine) insertConditionalJumpTrampoline(cbr *instruction, currentBlk *labelPosition, nextLabel label) {
cur := currentBlk.End
cur := currentBlk.end
originalTarget := cbr.condBrLabel()
endNext := cur.next
@ -422,32 +534,27 @@ func (m *machine) insertConditionalJumpTrampoline(cbr *instruction, currentBlk *
cur = linkInstr(cur, br)
// Update the end of the current block.
currentBlk.End = cur
currentBlk.end = cur
linkInstr(cur, endNext)
}
// Format implements backend.Machine.
func (m *machine) Format() string {
ectx := m.executableContext
begins := map[*instruction]label{}
for _, pos := range ectx.LabelPositions {
for l := label(0); l < m.nextLabel; l++ {
pos := m.labelPositionPool.Get(int(l))
if pos != nil {
begins[pos.Begin] = pos.L
begins[pos.begin] = l
}
}
irBlocks := map[label]ssa.BasicBlockID{}
for i, l := range ectx.SsaBlockIDToLabels {
irBlocks[l] = ssa.BasicBlockID(i)
}
var lines []string
for cur := ectx.RootInstr; cur != nil; cur = cur.next {
for cur := m.rootInstr; cur != nil; cur = cur.next {
if l, ok := begins[cur]; ok {
var labelStr string
if blkID, ok := irBlocks[l]; ok {
labelStr = fmt.Sprintf("%s (SSA Block: %s):", l, blkID)
if l <= m.maxSSABlockID {
labelStr = fmt.Sprintf("%s (SSA Block: blk%d):", l, int(l))
} else {
labelStr = fmt.Sprintf("%s:", l)
}
@ -508,13 +615,17 @@ func (m *machine) frameSize() int64 {
return s
}
func (m *machine) addJmpTableTarget(targets []ssa.BasicBlock) (index int) {
// TODO: reuse the slice!
labels := make([]uint32, len(targets))
for j, target := range targets {
labels[j] = uint32(m.executableContext.GetOrAllocateSSABlockLabel(target))
func (m *machine) addJmpTableTarget(targets ssa.Values) (index int) {
if m.jmpTableTargetsNext == len(m.jmpTableTargets) {
m.jmpTableTargets = append(m.jmpTableTargets, make([]uint32, 0, len(targets.View())))
}
index = m.jmpTableTargetsNext
m.jmpTableTargetsNext++
m.jmpTableTargets[index] = m.jmpTableTargets[index][:0]
for _, targetBlockID := range targets.View() {
target := m.compiler.SSABuilder().BasicBlock(ssa.BasicBlockID(targetBlockID))
m.jmpTableTargets[index] = append(m.jmpTableTargets[index], uint32(target.ID()))
}
index = len(m.jmpTableTargets)
m.jmpTableTargets = append(m.jmpTableTargets, labels)
return
}

View file

@ -15,9 +15,7 @@ func (m *machine) PostRegAlloc() {
// setupPrologue initializes the prologue of the function.
func (m *machine) setupPrologue() {
ectx := m.executableContext
cur := ectx.RootInstr
cur := m.rootInstr
prevInitInst := cur.next
//
@ -196,21 +194,20 @@ func (m *machine) createFrameSizeSlot(cur *instruction, s int64) *instruction {
// 1. Removes the redundant copy instruction.
// 2. Inserts the epilogue.
func (m *machine) postRegAlloc() {
ectx := m.executableContext
for cur := ectx.RootInstr; cur != nil; cur = cur.next {
for cur := m.rootInstr; cur != nil; cur = cur.next {
switch cur.kind {
case ret:
m.setupEpilogueAfter(cur.prev)
case loadConstBlockArg:
lc := cur
next := lc.next
m.executableContext.PendingInstructions = m.executableContext.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
m.lowerLoadConstantBlockArgAfterRegAlloc(lc)
for _, instr := range m.executableContext.PendingInstructions {
for _, instr := range m.pendingInstructions {
cur = linkInstr(cur, instr)
}
linkInstr(cur, next)
m.executableContext.PendingInstructions = m.executableContext.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
default:
// Removes the redundant copy instruction.
if cur.IsCopy() && cur.rn.realReg() == cur.rd.RealReg() {
@ -432,11 +429,9 @@ func (m *machine) insertStackBoundsCheck(requiredStackSize int64, cur *instructi
// CompileStackGrowCallSequence implements backend.Machine.
func (m *machine) CompileStackGrowCallSequence() []byte {
ectx := m.executableContext
cur := m.allocateInstr()
cur.asNop0()
ectx.RootInstr = cur
m.rootInstr = cur
// Save the callee saved and argument registers.
cur = m.saveRegistersInExecutionContext(cur, saveRequiredRegs)
@ -458,16 +453,14 @@ func (m *machine) CompileStackGrowCallSequence() []byte {
ret.asRet()
linkInstr(cur, ret)
m.encode(ectx.RootInstr)
m.encode(m.rootInstr)
return m.compiler.Buf()
}
func (m *machine) addsAddOrSubStackPointer(cur *instruction, rd regalloc.VReg, diff int64, add bool) *instruction {
ectx := m.executableContext
ectx.PendingInstructions = ectx.PendingInstructions[:0]
m.pendingInstructions = m.pendingInstructions[:0]
m.insertAddOrSubStackPointer(rd, diff, add)
for _, inserted := range ectx.PendingInstructions {
for _, inserted := range m.pendingInstructions {
cur = linkInstr(cur, inserted)
}
return cur

View file

@ -3,18 +3,226 @@
// This file implements the interfaces required for register allocations. See backend.RegAllocFunctionMachine.
import (
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend"
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend/regalloc"
"github.com/tetratelabs/wazero/internal/engine/wazevo/ssa"
)
// ClobberedRegisters implements backend.RegAllocFunctionMachine.
func (m *machine) ClobberedRegisters(regs []regalloc.VReg) {
m.clobberedRegs = append(m.clobberedRegs[:0], regs...)
// regAllocFn implements regalloc.Function.
type regAllocFn struct {
ssaB ssa.Builder
m *machine
loopNestingForestRoots []ssa.BasicBlock
blockIter int
}
// Swap implements backend.RegAllocFunctionMachine.
func (m *machine) Swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
// PostOrderBlockIteratorBegin implements regalloc.Function.
func (f *regAllocFn) PostOrderBlockIteratorBegin() *labelPosition {
f.blockIter = len(f.m.orderedSSABlockLabelPos) - 1
return f.PostOrderBlockIteratorNext()
}
// PostOrderBlockIteratorNext implements regalloc.Function.
func (f *regAllocFn) PostOrderBlockIteratorNext() *labelPosition {
if f.blockIter < 0 {
return nil
}
b := f.m.orderedSSABlockLabelPos[f.blockIter]
f.blockIter--
return b
}
// ReversePostOrderBlockIteratorBegin implements regalloc.Function.
func (f *regAllocFn) ReversePostOrderBlockIteratorBegin() *labelPosition {
f.blockIter = 0
return f.ReversePostOrderBlockIteratorNext()
}
// ReversePostOrderBlockIteratorNext implements regalloc.Function.
func (f *regAllocFn) ReversePostOrderBlockIteratorNext() *labelPosition {
if f.blockIter >= len(f.m.orderedSSABlockLabelPos) {
return nil
}
b := f.m.orderedSSABlockLabelPos[f.blockIter]
f.blockIter++
return b
}
// ClobberedRegisters implements regalloc.Function.
func (f *regAllocFn) ClobberedRegisters(regs []regalloc.VReg) {
f.m.clobberedRegs = append(f.m.clobberedRegs[:0], regs...)
}
// LoopNestingForestRoots implements regalloc.Function.
func (f *regAllocFn) LoopNestingForestRoots() int {
f.loopNestingForestRoots = f.ssaB.LoopNestingForestRoots()
return len(f.loopNestingForestRoots)
}
// LoopNestingForestRoot implements regalloc.Function.
func (f *regAllocFn) LoopNestingForestRoot(i int) *labelPosition {
root := f.loopNestingForestRoots[i]
pos := f.m.getOrAllocateSSABlockLabelPosition(root)
return pos
}
// LowestCommonAncestor implements regalloc.Function.
func (f *regAllocFn) LowestCommonAncestor(blk1, blk2 *labelPosition) *labelPosition {
sb := f.ssaB.LowestCommonAncestor(blk1.sb, blk2.sb)
pos := f.m.getOrAllocateSSABlockLabelPosition(sb)
return pos
}
// Idom implements regalloc.Function.
func (f *regAllocFn) Idom(blk *labelPosition) *labelPosition {
sb := f.ssaB.Idom(blk.sb)
pos := f.m.getOrAllocateSSABlockLabelPosition(sb)
return pos
}
// SwapBefore implements regalloc.Function.
func (f *regAllocFn) SwapBefore(x1, x2, tmp regalloc.VReg, instr *instruction) {
f.m.swap(instr.prev, x1, x2, tmp)
}
// StoreRegisterBefore implements regalloc.Function.
func (f *regAllocFn) StoreRegisterBefore(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertStoreRegisterAt(v, instr, false)
}
// StoreRegisterAfter implements regalloc.Function.
func (f *regAllocFn) StoreRegisterAfter(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertStoreRegisterAt(v, instr, true)
}
// ReloadRegisterBefore implements regalloc.Function.
func (f *regAllocFn) ReloadRegisterBefore(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertReloadRegisterAt(v, instr, false)
}
// ReloadRegisterAfter implements regalloc.Function.
func (f *regAllocFn) ReloadRegisterAfter(v regalloc.VReg, instr *instruction) {
m := f.m
m.insertReloadRegisterAt(v, instr, true)
}
// InsertMoveBefore implements regalloc.Function.
func (f *regAllocFn) InsertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
f.m.insertMoveBefore(dst, src, instr)
}
// LoopNestingForestChild implements regalloc.Function.
func (f *regAllocFn) LoopNestingForestChild(pos *labelPosition, i int) *labelPosition {
childSB := pos.sb.LoopNestingForestChildren()[i]
return f.m.getOrAllocateSSABlockLabelPosition(childSB)
}
// Succ implements regalloc.Block.
func (f *regAllocFn) Succ(pos *labelPosition, i int) *labelPosition {
succSB := pos.sb.Succ(i)
if succSB.ReturnBlock() {
return nil
}
return f.m.getOrAllocateSSABlockLabelPosition(succSB)
}
// Pred implements regalloc.Block.
func (f *regAllocFn) Pred(pos *labelPosition, i int) *labelPosition {
predSB := pos.sb.Pred(i)
return f.m.getOrAllocateSSABlockLabelPosition(predSB)
}
// BlockParams implements regalloc.Function.
func (f *regAllocFn) BlockParams(pos *labelPosition, regs *[]regalloc.VReg) []regalloc.VReg {
c := f.m.compiler
*regs = (*regs)[:0]
for i := 0; i < pos.sb.Params(); i++ {
v := c.VRegOf(pos.sb.Param(i))
*regs = append(*regs, v)
}
return *regs
}
// ID implements regalloc.Block.
func (pos *labelPosition) ID() int32 {
return int32(pos.sb.ID())
}
// InstrIteratorBegin implements regalloc.Block.
func (pos *labelPosition) InstrIteratorBegin() *instruction {
ret := pos.begin
pos.cur = ret
return ret
}
// InstrIteratorNext implements regalloc.Block.
func (pos *labelPosition) InstrIteratorNext() *instruction {
for {
if pos.cur == pos.end {
return nil
}
instr := pos.cur.next
pos.cur = instr
if instr == nil {
return nil
} else if instr.addedBeforeRegAlloc {
// Only concerned about the instruction added before regalloc.
return instr
}
}
}
// InstrRevIteratorBegin implements regalloc.Block.
func (pos *labelPosition) InstrRevIteratorBegin() *instruction {
pos.cur = pos.end
return pos.cur
}
// InstrRevIteratorNext implements regalloc.Block.
func (pos *labelPosition) InstrRevIteratorNext() *instruction {
for {
if pos.cur == pos.begin {
return nil
}
instr := pos.cur.prev
pos.cur = instr
if instr == nil {
return nil
} else if instr.addedBeforeRegAlloc {
// Only concerned about the instruction added before regalloc.
return instr
}
}
}
// FirstInstr implements regalloc.Block.
func (pos *labelPosition) FirstInstr() *instruction { return pos.begin }
// LastInstrForInsertion implements regalloc.Block.
func (pos *labelPosition) LastInstrForInsertion() *instruction {
return lastInstrForInsertion(pos.begin, pos.end)
}
// Preds implements regalloc.Block.
func (pos *labelPosition) Preds() int { return pos.sb.Preds() }
// Entry implements regalloc.Block.
func (pos *labelPosition) Entry() bool { return pos.sb.EntryBlock() }
// Succs implements regalloc.Block.
func (pos *labelPosition) Succs() int { return pos.sb.Succs() }
// LoopHeader implements regalloc.Block.
func (pos *labelPosition) LoopHeader() bool { return pos.sb.LoopHeader() }
// LoopNestingForestChildren implements regalloc.Block.
func (pos *labelPosition) LoopNestingForestChildren() int {
return len(pos.sb.LoopNestingForestChildren())
}
func (m *machine) swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
prevNext := cur.next
var mov1, mov2, mov3 *instruction
if x1.RegType() == regalloc.RegTypeInt {
@ -32,12 +240,12 @@ func (m *machine) Swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
if !tmp.Valid() {
r2 := x2.RealReg()
// Temporarily spill x1 to stack.
cur = m.InsertStoreRegisterAt(x1, cur, true).prev
cur = m.insertStoreRegisterAt(x1, cur, true).prev
// Then move x2 to x1.
cur = linkInstr(cur, m.allocateInstr().asFpuMov128(x1, x2))
linkInstr(cur, prevNext)
// Then reload the original value on x1 from stack to r2.
m.InsertReloadRegisterAt(x1.SetRealReg(r2), cur, true)
m.insertReloadRegisterAt(x1.SetRealReg(r2), cur, true)
} else {
mov1 = m.allocateInstr().asFpuMov128(tmp, x1)
mov2 = m.allocateInstr().asFpuMov128(x1, x2)
@ -50,8 +258,7 @@ func (m *machine) Swap(cur *instruction, x1, x2, tmp regalloc.VReg) {
}
}
// InsertMoveBefore implements backend.RegAllocFunctionMachine.
func (m *machine) InsertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
func (m *machine) insertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
typ := src.RegType()
if typ != dst.RegType() {
panic("BUG: src and dst must have the same type")
@ -70,13 +277,7 @@ func (m *machine) InsertMoveBefore(dst, src regalloc.VReg, instr *instruction) {
linkInstr(cur, prevNext)
}
// SSABlockLabel implements backend.RegAllocFunctionMachine.
func (m *machine) SSABlockLabel(id ssa.BasicBlockID) backend.Label {
return m.executableContext.SsaBlockIDToLabels[id]
}
// InsertStoreRegisterAt implements backend.RegAllocFunctionMachine.
func (m *machine) InsertStoreRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
func (m *machine) insertStoreRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
if !v.IsRealReg() {
panic("BUG: VReg must be backed by real reg to be stored")
}
@ -100,8 +301,7 @@ func (m *machine) InsertStoreRegisterAt(v regalloc.VReg, instr *instruction, aft
return linkInstr(cur, prevNext)
}
// InsertReloadRegisterAt implements backend.RegAllocFunctionMachine.
func (m *machine) InsertReloadRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
func (m *machine) insertReloadRegisterAt(v regalloc.VReg, instr *instruction, after bool) *instruction {
if !v.IsRealReg() {
panic("BUG: VReg must be backed by real reg to be stored")
}
@ -134,8 +334,7 @@ func (m *machine) InsertReloadRegisterAt(v regalloc.VReg, instr *instruction, af
return linkInstr(cur, prevNext)
}
// LastInstrForInsertion implements backend.RegAllocFunctionMachine.
func (m *machine) LastInstrForInsertion(begin, end *instruction) *instruction {
func lastInstrForInsertion(begin, end *instruction) *instruction {
cur := end
for cur.kind == nop0 {
cur = cur.prev

View file

@ -14,7 +14,7 @@ func UnwindStack(sp, _, top uintptr, returnAddresses []uintptr) []uintptr {
var stackBuf []byte
{
// TODO: use unsafe.Slice after floor version is set to Go 1.20.
//nolint:staticcheck
hdr := (*reflect.SliceHeader)(unsafe.Pointer(&stackBuf))
hdr.Data = sp
hdr.Len = l
@ -78,13 +78,7 @@ func GoCallStackView(stackPointerBeforeGoCall *uint64) []uint64 {
// +-----------------+ <---- stackPointerBeforeGoCall
// (low address)
ptr := unsafe.Pointer(stackPointerBeforeGoCall)
data := (*uint64)(unsafe.Add(ptr, 16)) // skips the (frame_size, sliceSize).
size := *(*uint64)(unsafe.Add(ptr, 8))
var view []uint64
{
sh := (*reflect.SliceHeader)(unsafe.Pointer(&view))
sh.Data = uintptr(unsafe.Add(ptr, 16)) // skips the (frame_size, sliceSize).
sh.Len = int(size)
sh.Cap = int(size)
}
return view
return unsafe.Slice(data, size)
}

View file

@ -11,7 +11,24 @@
type (
// Machine is a backend for a specific ISA machine.
Machine interface {
ExecutableContext() ExecutableContext
// StartLoweringFunction is called when the compilation of the given function is started.
// The maxBlockID is the maximum ssa.BasicBlockID in the function.
StartLoweringFunction(maxBlockID ssa.BasicBlockID)
// LinkAdjacentBlocks is called after finished lowering all blocks in order to create one single instruction list.
LinkAdjacentBlocks(prev, next ssa.BasicBlock)
// StartBlock is called when the compilation of the given block is started.
// The order of this being called is the reverse post order of the ssa.BasicBlock(s) as we iterate with
// ssa.Builder BlockIteratorReversePostOrderBegin and BlockIteratorReversePostOrderEnd.
StartBlock(ssa.BasicBlock)
// EndBlock is called when the compilation of the current block is finished.
EndBlock()
// FlushPendingInstructions flushes the pending instructions to the buffer.
// This will be called after the lowering of each SSA Instruction.
FlushPendingInstructions()
// DisableStackCheck disables the stack check for the current compilation for debugging/testing.
DisableStackCheck()

View file

@ -1,321 +0,0 @@
package backend
import (
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend/regalloc"
"github.com/tetratelabs/wazero/internal/engine/wazevo/ssa"
)
// RegAllocFunctionMachine is the interface for the machine specific logic that will be used in RegAllocFunction.
type RegAllocFunctionMachine[I regalloc.InstrConstraint] interface {
// InsertMoveBefore inserts the move instruction from src to dst before the given instruction.
InsertMoveBefore(dst, src regalloc.VReg, instr I)
// InsertStoreRegisterAt inserts the instruction(s) to store the given virtual register at the given instruction.
// If after is true, the instruction(s) will be inserted after the given instruction, otherwise before.
InsertStoreRegisterAt(v regalloc.VReg, instr I, after bool) I
// InsertReloadRegisterAt inserts the instruction(s) to reload the given virtual register at the given instruction.
// If after is true, the instruction(s) will be inserted after the given instruction, otherwise before.
InsertReloadRegisterAt(v regalloc.VReg, instr I, after bool) I
// ClobberedRegisters is called when the register allocation is done and the clobbered registers are known.
ClobberedRegisters(regs []regalloc.VReg)
// Swap swaps the two virtual registers after the given instruction.
Swap(cur I, x1, x2, tmp regalloc.VReg)
// LastInstrForInsertion implements LastInstrForInsertion of regalloc.Function. See its comment for details.
LastInstrForInsertion(begin, end I) I
// SSABlockLabel returns the label of the given ssa.BasicBlockID.
SSABlockLabel(id ssa.BasicBlockID) Label
}
type (
// RegAllocFunction implements regalloc.Function.
RegAllocFunction[I regalloc.InstrConstraint, m RegAllocFunctionMachine[I]] struct {
m m
ssb ssa.Builder
c Compiler
// iter is the iterator for reversePostOrderBlocks
iter int
reversePostOrderBlocks []RegAllocBlock[I, m]
// labelToRegAllocBlockIndex maps label to the index of reversePostOrderBlocks.
labelToRegAllocBlockIndex [] /* Label to */ int
loopNestingForestRoots []ssa.BasicBlock
}
// RegAllocBlock implements regalloc.Block.
RegAllocBlock[I regalloc.InstrConstraint, m RegAllocFunctionMachine[I]] struct {
// f is the function this instruction belongs to. Used to reuse the regAllocFunctionImpl.predsSlice slice for Defs() and Uses().
f *RegAllocFunction[I, m]
sb ssa.BasicBlock
l Label
begin, end I
loopNestingForestChildren []ssa.BasicBlock
cur I
id int
cachedLastInstrForInsertion I
}
)
// NewRegAllocFunction returns a new RegAllocFunction.
func NewRegAllocFunction[I regalloc.InstrConstraint, M RegAllocFunctionMachine[I]](m M, ssb ssa.Builder, c Compiler) *RegAllocFunction[I, M] {
return &RegAllocFunction[I, M]{
m: m,
ssb: ssb,
c: c,
}
}
// AddBlock adds a new block to the function.
func (f *RegAllocFunction[I, M]) AddBlock(sb ssa.BasicBlock, l Label, begin, end I) {
i := len(f.reversePostOrderBlocks)
f.reversePostOrderBlocks = append(f.reversePostOrderBlocks, RegAllocBlock[I, M]{
f: f,
sb: sb,
l: l,
begin: begin,
end: end,
id: int(sb.ID()),
})
if len(f.labelToRegAllocBlockIndex) <= int(l) {
f.labelToRegAllocBlockIndex = append(f.labelToRegAllocBlockIndex, make([]int, int(l)-len(f.labelToRegAllocBlockIndex)+1)...)
}
f.labelToRegAllocBlockIndex[l] = i
}
// Reset resets the function for the next compilation.
func (f *RegAllocFunction[I, M]) Reset() {
f.reversePostOrderBlocks = f.reversePostOrderBlocks[:0]
f.iter = 0
}
// StoreRegisterAfter implements regalloc.Function StoreRegisterAfter.
func (f *RegAllocFunction[I, M]) StoreRegisterAfter(v regalloc.VReg, instr regalloc.Instr) {
m := f.m
m.InsertStoreRegisterAt(v, instr.(I), true)
}
// ReloadRegisterBefore implements regalloc.Function ReloadRegisterBefore.
func (f *RegAllocFunction[I, M]) ReloadRegisterBefore(v regalloc.VReg, instr regalloc.Instr) {
m := f.m
m.InsertReloadRegisterAt(v, instr.(I), false)
}
// ReloadRegisterAfter implements regalloc.Function ReloadRegisterAfter.
func (f *RegAllocFunction[I, M]) ReloadRegisterAfter(v regalloc.VReg, instr regalloc.Instr) {
m := f.m
m.InsertReloadRegisterAt(v, instr.(I), true)
}
// StoreRegisterBefore implements regalloc.Function StoreRegisterBefore.
func (f *RegAllocFunction[I, M]) StoreRegisterBefore(v regalloc.VReg, instr regalloc.Instr) {
m := f.m
m.InsertStoreRegisterAt(v, instr.(I), false)
}
// ClobberedRegisters implements regalloc.Function ClobberedRegisters.
func (f *RegAllocFunction[I, M]) ClobberedRegisters(regs []regalloc.VReg) {
f.m.ClobberedRegisters(regs)
}
// SwapBefore implements regalloc.Function SwapBefore.
func (f *RegAllocFunction[I, M]) SwapBefore(x1, x2, tmp regalloc.VReg, instr regalloc.Instr) {
f.m.Swap(instr.Prev().(I), x1, x2, tmp)
}
// PostOrderBlockIteratorBegin implements regalloc.Function PostOrderBlockIteratorBegin.
func (f *RegAllocFunction[I, M]) PostOrderBlockIteratorBegin() regalloc.Block {
f.iter = len(f.reversePostOrderBlocks) - 1
return f.PostOrderBlockIteratorNext()
}
// PostOrderBlockIteratorNext implements regalloc.Function PostOrderBlockIteratorNext.
func (f *RegAllocFunction[I, M]) PostOrderBlockIteratorNext() regalloc.Block {
if f.iter < 0 {
return nil
}
b := &f.reversePostOrderBlocks[f.iter]
f.iter--
return b
}
// ReversePostOrderBlockIteratorBegin implements regalloc.Function ReversePostOrderBlockIteratorBegin.
func (f *RegAllocFunction[I, M]) ReversePostOrderBlockIteratorBegin() regalloc.Block {
f.iter = 0
return f.ReversePostOrderBlockIteratorNext()
}
// ReversePostOrderBlockIteratorNext implements regalloc.Function ReversePostOrderBlockIteratorNext.
func (f *RegAllocFunction[I, M]) ReversePostOrderBlockIteratorNext() regalloc.Block {
if f.iter >= len(f.reversePostOrderBlocks) {
return nil
}
b := &f.reversePostOrderBlocks[f.iter]
f.iter++
return b
}
// LoopNestingForestRoots implements regalloc.Function LoopNestingForestRoots.
func (f *RegAllocFunction[I, M]) LoopNestingForestRoots() int {
f.loopNestingForestRoots = f.ssb.LoopNestingForestRoots()
return len(f.loopNestingForestRoots)
}
// LoopNestingForestRoot implements regalloc.Function LoopNestingForestRoot.
func (f *RegAllocFunction[I, M]) LoopNestingForestRoot(i int) regalloc.Block {
blk := f.loopNestingForestRoots[i]
l := f.m.SSABlockLabel(blk.ID())
index := f.labelToRegAllocBlockIndex[l]
return &f.reversePostOrderBlocks[index]
}
// InsertMoveBefore implements regalloc.Function InsertMoveBefore.
func (f *RegAllocFunction[I, M]) InsertMoveBefore(dst, src regalloc.VReg, instr regalloc.Instr) {
f.m.InsertMoveBefore(dst, src, instr.(I))
}
// LowestCommonAncestor implements regalloc.Function LowestCommonAncestor.
func (f *RegAllocFunction[I, M]) LowestCommonAncestor(blk1, blk2 regalloc.Block) regalloc.Block {
ret := f.ssb.LowestCommonAncestor(blk1.(*RegAllocBlock[I, M]).sb, blk2.(*RegAllocBlock[I, M]).sb)
l := f.m.SSABlockLabel(ret.ID())
index := f.labelToRegAllocBlockIndex[l]
return &f.reversePostOrderBlocks[index]
}
// Idom implements regalloc.Function Idom.
func (f *RegAllocFunction[I, M]) Idom(blk regalloc.Block) regalloc.Block {
builder := f.ssb
idom := builder.Idom(blk.(*RegAllocBlock[I, M]).sb)
if idom == nil {
panic("BUG: idom must not be nil")
}
l := f.m.SSABlockLabel(idom.ID())
index := f.labelToRegAllocBlockIndex[l]
return &f.reversePostOrderBlocks[index]
}
// ID implements regalloc.Block.
func (r *RegAllocBlock[I, m]) ID() int32 { return int32(r.id) }
// BlockParams implements regalloc.Block.
func (r *RegAllocBlock[I, m]) BlockParams(regs *[]regalloc.VReg) []regalloc.VReg {
c := r.f.c
*regs = (*regs)[:0]
for i := 0; i < r.sb.Params(); i++ {
v := c.VRegOf(r.sb.Param(i))
*regs = append(*regs, v)
}
return *regs
}
// InstrIteratorBegin implements regalloc.Block.
func (r *RegAllocBlock[I, m]) InstrIteratorBegin() regalloc.Instr {
r.cur = r.begin
return r.cur
}
// InstrIteratorNext implements regalloc.Block.
func (r *RegAllocBlock[I, m]) InstrIteratorNext() regalloc.Instr {
for {
if r.cur == r.end {
return nil
}
instr := r.cur.Next()
r.cur = instr.(I)
if instr == nil {
return nil
} else if instr.AddedBeforeRegAlloc() {
// Only concerned about the instruction added before regalloc.
return instr
}
}
}
// InstrRevIteratorBegin implements regalloc.Block.
func (r *RegAllocBlock[I, m]) InstrRevIteratorBegin() regalloc.Instr {
r.cur = r.end
return r.cur
}
// InstrRevIteratorNext implements regalloc.Block.
func (r *RegAllocBlock[I, m]) InstrRevIteratorNext() regalloc.Instr {
for {
if r.cur == r.begin {
return nil
}
instr := r.cur.Prev()
r.cur = instr.(I)
if instr == nil {
return nil
} else if instr.AddedBeforeRegAlloc() {
// Only concerned about the instruction added before regalloc.
return instr
}
}
}
// FirstInstr implements regalloc.Block.
func (r *RegAllocBlock[I, m]) FirstInstr() regalloc.Instr {
return r.begin
}
// EndInstr implements regalloc.Block.
func (r *RegAllocBlock[I, m]) EndInstr() regalloc.Instr {
return r.end
}
// LastInstrForInsertion implements regalloc.Block.
func (r *RegAllocBlock[I, m]) LastInstrForInsertion() regalloc.Instr {
var nil I
if r.cachedLastInstrForInsertion == nil {
r.cachedLastInstrForInsertion = r.f.m.LastInstrForInsertion(r.begin, r.end)
}
return r.cachedLastInstrForInsertion
}
// Preds implements regalloc.Block.
func (r *RegAllocBlock[I, m]) Preds() int { return r.sb.Preds() }
// Pred implements regalloc.Block.
func (r *RegAllocBlock[I, m]) Pred(i int) regalloc.Block {
sb := r.sb
pred := sb.Pred(i)
l := r.f.m.SSABlockLabel(pred.ID())
index := r.f.labelToRegAllocBlockIndex[l]
return &r.f.reversePostOrderBlocks[index]
}
// Entry implements regalloc.Block.
func (r *RegAllocBlock[I, m]) Entry() bool { return r.sb.EntryBlock() }
// Succs implements regalloc.Block.
func (r *RegAllocBlock[I, m]) Succs() int {
return r.sb.Succs()
}
// Succ implements regalloc.Block.
func (r *RegAllocBlock[I, m]) Succ(i int) regalloc.Block {
sb := r.sb
succ := sb.Succ(i)
if succ.ReturnBlock() {
return nil
}
l := r.f.m.SSABlockLabel(succ.ID())
index := r.f.labelToRegAllocBlockIndex[l]
return &r.f.reversePostOrderBlocks[index]
}
// LoopHeader implements regalloc.Block.
func (r *RegAllocBlock[I, m]) LoopHeader() bool {
return r.sb.LoopHeader()
}
// LoopNestingForestChildren implements regalloc.Block.
func (r *RegAllocBlock[I, m]) LoopNestingForestChildren() int {
r.loopNestingForestChildren = r.sb.LoopNestingForestChildren()
return len(r.loopNestingForestChildren)
}
// LoopNestingForestChild implements regalloc.Block.
func (r *RegAllocBlock[I, m]) LoopNestingForestChild(i int) regalloc.Block {
blk := r.loopNestingForestChildren[i]
l := r.f.m.SSABlockLabel(blk.ID())
index := r.f.labelToRegAllocBlockIndex[l]
return &r.f.reversePostOrderBlocks[index]
}

View file

@ -4,104 +4,100 @@
// These interfaces are implemented by ISA-specific backends to abstract away the details, and allow the register
// allocators to work on any ISA.
//
// TODO: the interfaces are not stabilized yet, especially x64 will need some changes. E.g. x64 has an addressing mode
// where index can be in memory. That kind of info will be useful to reduce the register pressure, and should be leveraged
// by the register allocators, like https://docs.rs/regalloc2/latest/regalloc2/enum.OperandConstraint.html
type (
// Function is the top-level interface to do register allocation, which corresponds to a CFG containing
// Blocks(s).
Function interface {
//
// I is the type of the instruction, and B is the type of the basic block.
Function[I Instr, B Block[I]] interface {
// PostOrderBlockIteratorBegin returns the first block in the post-order traversal of the CFG.
// In other words, the last blocks in the CFG will be returned first.
PostOrderBlockIteratorBegin() Block
PostOrderBlockIteratorBegin() B
// PostOrderBlockIteratorNext returns the next block in the post-order traversal of the CFG.
PostOrderBlockIteratorNext() Block
PostOrderBlockIteratorNext() B
// ReversePostOrderBlockIteratorBegin returns the first block in the reverse post-order traversal of the CFG.
// In other words, the first blocks in the CFG will be returned first.
ReversePostOrderBlockIteratorBegin() Block
ReversePostOrderBlockIteratorBegin() B
// ReversePostOrderBlockIteratorNext returns the next block in the reverse post-order traversal of the CFG.
ReversePostOrderBlockIteratorNext() Block
ReversePostOrderBlockIteratorNext() B
// ClobberedRegisters tell the clobbered registers by this function.
ClobberedRegisters([]VReg)
// LoopNestingForestRoots returns the number of roots of the loop nesting forest in a function.
LoopNestingForestRoots() int
// LoopNestingForestRoot returns the i-th root of the loop nesting forest in a function.
LoopNestingForestRoot(i int) Block
LoopNestingForestRoot(i int) B
// LowestCommonAncestor returns the lowest common ancestor of two blocks in the dominator tree.
LowestCommonAncestor(blk1, blk2 Block) Block
LowestCommonAncestor(blk1, blk2 B) B
// Idom returns the immediate dominator of the given block.
Idom(blk Block) Block
Idom(blk B) B
// LoopNestingForestChild returns the i-th child of the block in the loop nesting forest.
LoopNestingForestChild(b B, i int) B
// Pred returns the i-th predecessor of the block in the CFG.
Pred(b B, i int) B
// Succ returns the i-th successor of the block in the CFG.
Succ(b B, i int) B
// BlockParams returns the virtual registers used as the parameters of this block.
BlockParams(B, *[]VReg) []VReg
// Followings are for rewriting the function.
// SwapAtEndOfBlock swaps the two virtual registers at the end of the given block.
SwapBefore(x1, x2, tmp VReg, instr Instr)
// SwapBefore swaps the two virtual registers at the end of the given block.
SwapBefore(x1, x2, tmp VReg, instr I)
// StoreRegisterBefore inserts store instruction(s) before the given instruction for the given virtual register.
StoreRegisterBefore(v VReg, instr Instr)
StoreRegisterBefore(v VReg, instr I)
// StoreRegisterAfter inserts store instruction(s) after the given instruction for the given virtual register.
StoreRegisterAfter(v VReg, instr Instr)
StoreRegisterAfter(v VReg, instr I)
// ReloadRegisterBefore inserts reload instruction(s) before the given instruction for the given virtual register.
ReloadRegisterBefore(v VReg, instr Instr)
ReloadRegisterBefore(v VReg, instr I)
// ReloadRegisterAfter inserts reload instruction(s) after the given instruction for the given virtual register.
ReloadRegisterAfter(v VReg, instr Instr)
ReloadRegisterAfter(v VReg, instr I)
// InsertMoveBefore inserts move instruction(s) before the given instruction for the given virtual registers.
InsertMoveBefore(dst, src VReg, instr Instr)
InsertMoveBefore(dst, src VReg, instr I)
}
// Block is a basic block in the CFG of a function, and it consists of multiple instructions, and predecessor Block(s).
Block interface {
// Right now, this corresponds to a ssa.BasicBlock lowered to the machine level.
Block[I Instr] interface {
comparable
// ID returns the unique identifier of this block which is ordered in the reverse post-order traversal of the CFG.
ID() int32
// BlockParams returns the virtual registers used as the parameters of this block.
BlockParams(*[]VReg) []VReg
// InstrIteratorBegin returns the first instruction in this block. Instructions added after lowering must be skipped.
// Note: multiple Instr(s) will not be held at the same time, so it's safe to use the same impl for the return Instr.
InstrIteratorBegin() Instr
InstrIteratorBegin() I
// InstrIteratorNext returns the next instruction in this block. Instructions added after lowering must be skipped.
// Note: multiple Instr(s) will not be held at the same time, so it's safe to use the same impl for the return Instr.
InstrIteratorNext() Instr
InstrIteratorNext() I
// InstrRevIteratorBegin is the same as InstrIteratorBegin, but in the reverse order.
InstrRevIteratorBegin() Instr
InstrRevIteratorBegin() I
// InstrRevIteratorNext is the same as InstrIteratorNext, but in the reverse order.
InstrRevIteratorNext() Instr
InstrRevIteratorNext() I
// FirstInstr returns the fist instruction in this block where instructions will be inserted after it.
FirstInstr() Instr
// EndInstr returns the end instruction in this block.
EndInstr() Instr
FirstInstr() I
// LastInstrForInsertion returns the last instruction in this block where instructions will be inserted before it.
// Such insertions only happen when we need to insert spill/reload instructions to adjust the merge edges.
// At the time of register allocation, all the critical edges are already split, so there is no need
// to worry about the case where branching instruction has multiple successors.
// Therefore, usually, it is the nop instruction, but if the block ends with an unconditional branching, then it returns
// the unconditional branch, not the nop. In other words it is either nop or unconditional branch.
LastInstrForInsertion() Instr
LastInstrForInsertion() I
// Preds returns the number of predecessors of this block in the CFG.
Preds() int
// Pred returns the i-th predecessor of this block in the CFG.
Pred(i int) Block
// Entry returns true if the block is for the entry block.
Entry() bool
// Succs returns the number of successors of this block in the CFG.
Succs() int
// Succ returns the i-th successor of this block in the CFG.
Succ(i int) Block
// LoopHeader returns true if this block is a loop header.
LoopHeader() bool
// LoopNestingForestChildren returns the number of children of this block in the loop nesting forest.
LoopNestingForestChildren() int
// LoopNestingForestChild returns the i-th child of this block in the loop nesting forest.
LoopNestingForestChild(i int) Block
}
// Instr is an instruction in a block, abstracting away the underlying ISA.
Instr interface {
comparable
fmt.Stringer
// Next returns the next instruction in the same block.
Next() Instr
// Prev returns the previous instruction in the same block.
Prev() Instr
// Defs returns the virtual registers defined by this instruction.
Defs(*[]VReg) []VReg
// Uses returns the virtual registers used by this instruction.
@ -124,13 +120,5 @@
IsIndirectCall() bool
// IsReturn returns true if this instruction is a return instruction.
IsReturn() bool
// AddedBeforeRegAlloc returns true if this instruction is added before register allocation.
AddedBeforeRegAlloc() bool
}
// InstrConstraint is an interface for arch-specific instruction constraints.
InstrConstraint interface {
comparable
Instr
}
)

File diff suppressed because it is too large Load diff

View file

@ -46,52 +46,50 @@ func (rs RegSet) Range(f func(allocatedRealReg RealReg)) {
}
}
type regInUseSet [64]VReg
type regInUseSet[I Instr, B Block[I], F Function[I, B]] [64]*vrState[I, B, F]
func newRegInUseSet() regInUseSet {
var ret regInUseSet
func newRegInUseSet[I Instr, B Block[I], F Function[I, B]]() regInUseSet[I, B, F] {
var ret regInUseSet[I, B, F]
ret.reset()
return ret
}
func (rs *regInUseSet) reset() {
for i := range rs {
rs[i] = VRegInvalid
}
func (rs *regInUseSet[I, B, F]) reset() {
clear(rs[:])
}
func (rs *regInUseSet) format(info *RegisterInfo) string { //nolint:unused
func (rs *regInUseSet[I, B, F]) format(info *RegisterInfo) string { //nolint:unused
var ret []string
for i, vr := range rs {
if vr != VRegInvalid {
ret = append(ret, fmt.Sprintf("(%s->v%d)", info.RealRegName(RealReg(i)), vr.ID()))
if vr != nil {
ret = append(ret, fmt.Sprintf("(%s->v%d)", info.RealRegName(RealReg(i)), vr.v.ID()))
}
}
return strings.Join(ret, ", ")
}
func (rs *regInUseSet) has(r RealReg) bool {
return r < 64 && rs[r] != VRegInvalid
func (rs *regInUseSet[I, B, F]) has(r RealReg) bool {
return r < 64 && rs[r] != nil
}
func (rs *regInUseSet) get(r RealReg) VReg {
func (rs *regInUseSet[I, B, F]) get(r RealReg) *vrState[I, B, F] {
return rs[r]
}
func (rs *regInUseSet) remove(r RealReg) {
rs[r] = VRegInvalid
func (rs *regInUseSet[I, B, F]) remove(r RealReg) {
rs[r] = nil
}
func (rs *regInUseSet) add(r RealReg, vr VReg) {
func (rs *regInUseSet[I, B, F]) add(r RealReg, vr *vrState[I, B, F]) {
if r >= 64 {
return
}
rs[r] = vr
}
func (rs *regInUseSet) range_(f func(allocatedRealReg RealReg, vr VReg)) {
func (rs *regInUseSet[I, B, F]) range_(f func(allocatedRealReg RealReg, vr *vrState[I, B, F])) {
for i, vr := range rs {
if vr != VRegInvalid {
if vr != nil {
f(RealReg(i), vr)
}
}

View file

@ -1,43 +1,19 @@
package backend
import (
"github.com/tetratelabs/wazero/internal/engine/wazevo/backend/regalloc"
"github.com/tetratelabs/wazero/internal/engine/wazevo/ssa"
)
// SSAValueDefinition represents a definition of an SSA value.
type SSAValueDefinition struct {
// BlockParamValue is valid if Instr == nil
BlockParamValue ssa.Value
// BlkParamVReg is valid if Instr == nil
BlkParamVReg regalloc.VReg
V ssa.Value
// Instr is not nil if this is a definition from an instruction.
Instr *ssa.Instruction
// N is the index of the return value in the instr's return values list.
N int
// RefCount is the number of references to the result.
RefCount int
RefCount uint32
}
// IsFromInstr returns true if this definition is from an instruction.
func (d *SSAValueDefinition) IsFromInstr() bool {
return d.Instr != nil
}
func (d *SSAValueDefinition) IsFromBlockParam() bool {
return d.Instr == nil
}
func (d *SSAValueDefinition) SSAValue() ssa.Value {
if d.IsFromBlockParam() {
return d.BlockParamValue
} else {
r, rs := d.Instr.Returns()
if d.N == 0 {
return r
} else {
return rs[d.N-1]
}
}
}

View file

@ -554,17 +554,21 @@ func (c *callEngine) cloneStack(l uintptr) (newSP, newFP, newTop uintptr, newSta
// Copy the existing contents in the previous Go-allocated stack into the new one.
var prevStackAligned, newStackAligned []byte
{
//nolint:staticcheck
sh := (*reflect.SliceHeader)(unsafe.Pointer(&prevStackAligned))
sh.Data = c.stackTop - relSp
setSliceLimits(sh, relSp, relSp)
sh.Len = int(relSp)
sh.Cap = int(relSp)
}
newTop = alignedStackTop(newStack)
{
newSP = newTop - relSp
newFP = newTop - relFp
//nolint:staticcheck
sh := (*reflect.SliceHeader)(unsafe.Pointer(&newStackAligned))
sh.Data = newSP
setSliceLimits(sh, relSp, relSp)
sh.Len = int(relSp)
sh.Cap = int(relSp)
}
copy(newStackAligned, prevStackAligned)
return

View file

@ -275,7 +275,7 @@ func (c *Compiler) LowerToSSA() {
builder.DefineVariable(variable, value, entryBlock)
c.setWasmLocalVariable(wasm.Index(i), variable)
}
c.declareWasmLocals(entryBlock)
c.declareWasmLocals()
c.declareNecessaryVariables()
c.lowerBody(entryBlock)
@ -295,7 +295,7 @@ func (c *Compiler) setWasmLocalVariable(index wasm.Index, variable ssa.Variable)
}
// declareWasmLocals declares the SSA variables for the Wasm locals.
func (c *Compiler) declareWasmLocals(entry ssa.BasicBlock) {
func (c *Compiler) declareWasmLocals() {
localCount := wasm.Index(len(c.wasmFunctionTyp.Params))
for i, typ := range c.wasmFunctionLocalTypes {
st := WasmTypeToSSAType(typ)
@ -543,11 +543,11 @@ func (c *Compiler) initializeCurrentBlockKnownBounds() {
cb := &c.bounds[i][c.pointers[i]]
if cb.id != smallestID {
same = false
break
} else {
if cb.bound < minBound {
minBound = cb.bound
}
c.pointers[i]++
}
}
@ -555,14 +555,6 @@ func (c *Compiler) initializeCurrentBlockKnownBounds() {
// Absolute address cannot be used in the intersection since the value might be only defined in one of the predecessors.
c.recordKnownSafeBound(smallestID, minBound, ssa.ValueInvalid)
}
// Move pointer(s) for the smallest ID forward (if same, move all).
for i := 0; i < preds; i++ {
cb := &c.bounds[i][c.pointers[i]]
if cb.id == smallestID {
c.pointers[i]++
}
}
}
}
}

View file

@ -1538,8 +1538,7 @@ func (c *Compiler) lowerCurrentOpcode() {
builder.SetCurrentBlock(elseBlk)
case wasm.OpcodeBrTable:
labels := state.tmpForBrTable
labels = labels[:0]
labels := state.tmpForBrTable[:0]
labelCount := c.readI32u()
for i := 0; i < int(labelCount); i++ {
labels = append(labels, c.readI32u())
@ -1557,6 +1556,7 @@ func (c *Compiler) lowerCurrentOpcode() {
} else {
c.lowerBrTable(labels, index)
}
state.tmpForBrTable = labels // reuse the temporary slice for next use.
state.unreachable = true
case wasm.OpcodeNop:
@ -4068,13 +4068,14 @@ func (c *Compiler) lowerBrTable(labels []uint32, index ssa.Value) {
numArgs = len(f.blockType.Results)
}
targets := make([]ssa.BasicBlock, len(labels))
varPool := builder.VarLengthPool()
trampolineBlockIDs := varPool.Allocate(len(labels))
// We need trampoline blocks since depending on the target block structure, we might end up inserting moves before jumps,
// which cannot be done with br_table. Instead, we can do such per-block moves in the trampoline blocks.
// At the linking phase (very end of the backend), we can remove the unnecessary jumps, and therefore no runtime overhead.
currentBlk := builder.CurrentBlock()
for i, l := range labels {
for _, l := range labels {
// Args are always on the top of the stack. Note that we should not share the args slice
// among the jump instructions since the args are modified during passes (e.g. redundant phi elimination).
args := c.nPeekDup(numArgs)
@ -4082,17 +4083,17 @@ func (c *Compiler) lowerBrTable(labels []uint32, index ssa.Value) {
trampoline := builder.AllocateBasicBlock()
builder.SetCurrentBlock(trampoline)
c.insertJumpToBlock(args, targetBlk)
targets[i] = trampoline
trampolineBlockIDs = trampolineBlockIDs.Append(builder.VarLengthPool(), ssa.Value(trampoline.ID()))
}
builder.SetCurrentBlock(currentBlk)
// If the target block has no arguments, we can just jump to the target block.
brTable := builder.AllocateInstruction()
brTable.AsBrTable(index, targets)
brTable.AsBrTable(index, trampolineBlockIDs)
builder.InsertInstruction(brTable)
for _, trampoline := range targets {
builder.Seal(trampoline)
for _, trampolineID := range trampolineBlockIDs.View() {
builder.Seal(builder.BasicBlock(ssa.BasicBlockID(trampolineID)))
}
}

View file

@ -1,5 +1,3 @@
//go:build go1.21
package frontend
import (

View file

@ -1,17 +0,0 @@
//go:build !go1.21
// TODO: delete after the floor Go version is 1.21
package frontend
import (
"sort"
"github.com/tetratelabs/wazero/internal/engine/wazevo/ssa"
)
func sortSSAValueIDs(IDs []ssa.ValueID) {
sort.SliceStable(IDs, func(i, j int) bool {
return int(IDs[i]) < int(IDs[j])
})
}

View file

@ -16,6 +16,7 @@ func buildHostModuleOpaque(m *wasm.Module, listeners []experimental.FunctionList
binary.LittleEndian.PutUint64(ret[0:], uint64(uintptr(unsafe.Pointer(m))))
if len(listeners) > 0 {
//nolint:staticcheck
sliceHeader := (*reflect.SliceHeader)(unsafe.Pointer(&listeners))
binary.LittleEndian.PutUint64(ret[8:], uint64(sliceHeader.Data))
binary.LittleEndian.PutUint64(ret[16:], uint64(sliceHeader.Len))
@ -33,6 +34,7 @@ func buildHostModuleOpaque(m *wasm.Module, listeners []experimental.FunctionList
func hostModuleFromOpaque(opaqueBegin uintptr) *wasm.Module {
var opaqueViewOverSlice []byte
//nolint:staticcheck
sh := (*reflect.SliceHeader)(unsafe.Pointer(&opaqueViewOverSlice))
sh.Data = opaqueBegin
sh.Len = 32
@ -42,6 +44,7 @@ func hostModuleFromOpaque(opaqueBegin uintptr) *wasm.Module {
func hostModuleListenersSliceFromOpaque(opaqueBegin uintptr) []experimental.FunctionListener {
var opaqueViewOverSlice []byte
//nolint:staticcheck
sh := (*reflect.SliceHeader)(unsafe.Pointer(&opaqueViewOverSlice))
sh.Data = opaqueBegin
sh.Len = 32
@ -51,9 +54,11 @@ func hostModuleListenersSliceFromOpaque(opaqueBegin uintptr) []experimental.Func
l := binary.LittleEndian.Uint64(opaqueViewOverSlice[16:])
c := binary.LittleEndian.Uint64(opaqueViewOverSlice[24:])
var ret []experimental.FunctionListener
//nolint:staticcheck
sh = (*reflect.SliceHeader)(unsafe.Pointer(&ret))
sh.Data = uintptr(b)
setSliceLimits(sh, uintptr(l), uintptr(c))
sh.Len = int(l)
sh.Cap = int(c)
return ret
}
@ -62,6 +67,7 @@ func hostModuleGoFuncFromOpaque[T any](index int, opaqueBegin uintptr) T {
ptr := opaqueBegin + offset
var opaqueViewOverFunction []byte
//nolint:staticcheck
sh := (*reflect.SliceHeader)(unsafe.Pointer(&opaqueViewOverFunction))
sh.Data = ptr
sh.Len = 16

View file

@ -1,11 +0,0 @@
//go:build !tinygo
package wazevo
import "reflect"
// setSliceLimits sets both Cap and Len for the given reflected slice.
func setSliceLimits(s *reflect.SliceHeader, l, c uintptr) {
s.Len = int(l)
s.Cap = int(c)
}

View file

@ -1,11 +0,0 @@
//go:build tinygo
package wazevo
import "reflect"
// setSliceLimits sets both Cap and Len for the given reflected slice.
func setSliceLimits(s *reflect.SliceHeader, l, c uintptr) {
s.Len = l
s.Cap = c
}

View file

@ -34,9 +34,6 @@ type BasicBlock interface {
// The returned Value is the definition of the param in this block.
Param(i int) Value
// InsertInstruction inserts an instruction that implements Value into the tail of this block.
InsertInstruction(raw *Instruction)
// Root returns the root instruction of this block.
Root() *Instruction
@ -81,7 +78,7 @@ type BasicBlock interface {
rootInstr, currentInstr *Instruction
// params are Values that represent parameters to a basicBlock.
// Each parameter can be considered as an output of PHI instruction in traditional SSA.
params []Value
params Values
preds []basicBlockPredecessorInfo
success []*basicBlock
// singlePred is the alias to preds[0] for fast lookup, and only set after Seal is called.
@ -179,23 +176,23 @@ func (bb *basicBlock) ReturnBlock() bool {
// AddParam implements BasicBlock.AddParam.
func (bb *basicBlock) AddParam(b Builder, typ Type) Value {
paramValue := b.allocateValue(typ)
bb.params = append(bb.params, paramValue)
bb.params = bb.params.Append(&b.(*builder).varLengthPool, paramValue)
return paramValue
}
// addParamOn adds a parameter to this block whose value is already allocated.
func (bb *basicBlock) addParamOn(value Value) {
bb.params = append(bb.params, value)
func (bb *basicBlock) addParamOn(b *builder, value Value) {
bb.params = bb.params.Append(&b.varLengthPool, value)
}
// Params implements BasicBlock.Params.
func (bb *basicBlock) Params() int {
return len(bb.params)
return len(bb.params.View())
}
// Param implements BasicBlock.Param.
func (bb *basicBlock) Param(i int) Value {
return bb.params[i]
return bb.params.View()[i]
}
// Valid implements BasicBlock.Valid.
@ -208,8 +205,8 @@ func (bb *basicBlock) Sealed() bool {
return bb.sealed
}
// InsertInstruction implements BasicBlock.InsertInstruction.
func (bb *basicBlock) InsertInstruction(next *Instruction) {
// insertInstruction implements BasicBlock.InsertInstruction.
func (bb *basicBlock) insertInstruction(b *builder, next *Instruction) {
current := bb.currentInstr
if current != nil {
current.next = next
@ -221,12 +218,12 @@ func (bb *basicBlock) InsertInstruction(next *Instruction) {
switch next.opcode {
case OpcodeJump, OpcodeBrz, OpcodeBrnz:
target := next.blk.(*basicBlock)
target.addPred(bb, next)
target := BasicBlockID(next.rValue)
b.basicBlock(target).addPred(bb, next)
case OpcodeBrTable:
for _, _target := range next.targets {
target := _target.(*basicBlock)
target.addPred(bb, next)
for _, _target := range next.rValues.View() {
target := BasicBlockID(_target)
b.basicBlock(target).addPred(bb, next)
}
}
}
@ -268,7 +265,7 @@ func (bb *basicBlock) Tail() *Instruction {
// reset resets the basicBlock to its initial state so that it can be reused for another function.
func resetBasicBlock(bb *basicBlock) {
bb.params = bb.params[:0]
bb.params = ValuesNil
bb.rootInstr, bb.currentInstr = nil, nil
bb.preds = bb.preds[:0]
bb.success = bb.success[:0]
@ -310,8 +307,8 @@ func (bb *basicBlock) addPred(blk BasicBlock, branch *Instruction) {
// formatHeader returns the string representation of the header of the basicBlock.
func (bb *basicBlock) formatHeader(b Builder) string {
ps := make([]string, len(bb.params))
for i, p := range bb.params {
ps := make([]string, len(bb.params.View()))
for i, p := range bb.params.View() {
ps[i] = p.formatWithType(b)
}
@ -339,7 +336,9 @@ func (bb *basicBlock) validate(b *builder) {
if len(bb.preds) > 0 {
for _, pred := range bb.preds {
if pred.branch.opcode != OpcodeBrTable {
if target := pred.branch.blk; target != bb {
blockID := int(pred.branch.rValue)
target := b.basicBlocksPool.View(blockID)
if target != bb {
panic(fmt.Sprintf("BUG: '%s' is not branch to %s, but to %s",
pred.branch.Format(b), bb.Name(), target.Name()))
}
@ -349,14 +348,14 @@ func (bb *basicBlock) validate(b *builder) {
if bb.ReturnBlock() {
exp = len(b.currentSignature.Results)
} else {
exp = len(bb.params)
exp = len(bb.params.View())
}
if len(pred.branch.vs.View()) != exp {
panic(fmt.Sprintf(
"BUG: len(argument at %s) != len(params at %s): %d != %d: %s",
pred.blk.Name(), bb.Name(),
len(pred.branch.vs.View()), len(bb.params), pred.branch.Format(b),
len(pred.branch.vs.View()), len(bb.params.View()), pred.branch.Format(b),
))
}

View file

@ -1,5 +1,3 @@
//go:build go1.21
package ssa
import (

View file

@ -1,24 +0,0 @@
//go:build !go1.21
// TODO: delete after the floor Go version is 1.21
package ssa
import "sort"
func sortBlocks(blocks []*basicBlock) {
sort.SliceStable(blocks, func(i, j int) bool {
iBlk, jBlk := blocks[i], blocks[j]
if jBlk.ReturnBlock() {
return true
}
if iBlk.ReturnBlock() {
return false
}
iRoot, jRoot := iBlk.rootInstr, jBlk.rootInstr
if iRoot == nil || jRoot == nil { // For testing.
return true
}
return iBlk.rootInstr.id < jBlk.rootInstr.id
})
}

View file

@ -94,9 +94,9 @@ type Builder interface {
// Returns nil if there's no unseen BasicBlock.
BlockIteratorNext() BasicBlock
// ValueRefCounts returns the map of ValueID to its reference count.
// The returned slice must not be modified.
ValueRefCounts() []int
// ValuesInfo returns the data per Value used to lower the SSA in backend.
// This is indexed by ValueID.
ValuesInfo() []ValueInfo
// BlockIteratorReversePostOrderBegin is almost the same as BlockIteratorBegin except it returns the BasicBlock in the reverse post-order.
// This is available after RunPasses is run.
@ -129,20 +129,24 @@ type Builder interface {
// InsertZeroValue inserts a zero value constant instruction of the given type.
InsertZeroValue(t Type)
// BasicBlock returns the BasicBlock of the given ID.
BasicBlock(id BasicBlockID) BasicBlock
// InstructionOfValue returns the Instruction that produces the given Value or nil if the Value is not produced by any Instruction.
InstructionOfValue(v Value) *Instruction
}
// NewBuilder returns a new Builder implementation.
func NewBuilder() Builder {
return &builder{
instructionsPool: wazevoapi.NewPool[Instruction](resetInstruction),
basicBlocksPool: wazevoapi.NewPool[basicBlock](resetBasicBlock),
varLengthBasicBlockPool: wazevoapi.NewVarLengthPool[BasicBlock](),
varLengthPool: wazevoapi.NewVarLengthPool[Value](),
valueAnnotations: make(map[ValueID]string),
signatures: make(map[SignatureID]*Signature),
valueIDAliases: make(map[ValueID]Value),
redundantParameterIndexToValue: make(map[int]Value),
returnBlk: &basicBlock{id: basicBlockIDReturnBlock},
instructionsPool: wazevoapi.NewPool[Instruction](resetInstruction),
basicBlocksPool: wazevoapi.NewPool[basicBlock](resetBasicBlock),
varLengthBasicBlockPool: wazevoapi.NewVarLengthPool[BasicBlock](),
varLengthPool: wazevoapi.NewVarLengthPool[Value](),
valueAnnotations: make(map[ValueID]string),
signatures: make(map[SignatureID]*Signature),
returnBlk: &basicBlock{id: basicBlockIDReturnBlock},
}
}
@ -159,19 +163,16 @@ type builder struct {
currentBB *basicBlock
returnBlk *basicBlock
// variables track the types for Variable with the index regarded Variable.
variables []Type
// nextValueID is used by builder.AllocateValue.
nextValueID ValueID
// nextVariable is used by builder.AllocateVariable.
nextVariable Variable
valueIDAliases map[ValueID]Value
// valueAnnotations contains the annotations for each Value, only used for debugging.
valueAnnotations map[ValueID]string
// valueRefCounts is used to lower the SSA in backend, and will be calculated
// by the last SSA-level optimization pass.
valueRefCounts []int
// valuesInfo contains the data per Value used to lower the SSA in backend. This is indexed by ValueID.
valuesInfo []ValueInfo
// dominators stores the immediate dominator of each BasicBlock.
// The index is blockID of the BasicBlock.
@ -184,12 +185,10 @@ type builder struct {
loopNestingForestRoots []BasicBlock
// The followings are used for optimization passes/deterministic compilation.
instStack []*Instruction
valueIDToInstruction []*Instruction
blkStack []*basicBlock
blkStack2 []*basicBlock
ints []int
redundantParameterIndexToValue map[int]Value
instStack []*Instruction
blkStack []*basicBlock
blkStack2 []*basicBlock
redundantParams []redundantParam
// blockIterCur is used to implement blockIteratorBegin and blockIteratorNext.
blockIterCur int
@ -207,6 +206,34 @@ type builder struct {
zeros [typeEnd]Value
}
// ValueInfo contains the data per Value used to lower the SSA in backend.
type ValueInfo struct {
// RefCount is the reference count of the Value.
RefCount uint32
alias Value
}
// redundantParam is a pair of the index of the redundant parameter and the Value.
// This is used to eliminate the redundant parameters in the optimization pass.
type redundantParam struct {
// index is the index of the redundant parameter in the basicBlock.
index int
// uniqueValue is the Value which is passed to the redundant parameter.
uniqueValue Value
}
// BasicBlock implements Builder.BasicBlock.
func (b *builder) BasicBlock(id BasicBlockID) BasicBlock {
return b.basicBlock(id)
}
func (b *builder) basicBlock(id BasicBlockID) *basicBlock {
if id == basicBlockIDReturnBlock {
return b.returnBlk
}
return b.basicBlocksPool.View(int(id))
}
// InsertZeroValue implements Builder.InsertZeroValue.
func (b *builder) InsertZeroValue(t Type) {
if b.zeros[t].Valid() {
@ -256,7 +283,7 @@ func (b *builder) Init(s *Signature) {
sig.used = false
}
b.ints = b.ints[:0]
b.redundantParams = b.redundantParams[:0]
b.blkStack = b.blkStack[:0]
b.blkStack2 = b.blkStack2[:0]
b.dominators = b.dominators[:0]
@ -265,17 +292,11 @@ func (b *builder) Init(s *Signature) {
for v := ValueID(0); v < b.nextValueID; v++ {
delete(b.valueAnnotations, v)
delete(b.valueIDAliases, v)
b.valueRefCounts[v] = 0
b.valueIDToInstruction[v] = nil
b.valuesInfo[v] = ValueInfo{alias: ValueInvalid}
}
b.nextValueID = 0
b.reversePostOrderedBasicBlocks = b.reversePostOrderedBasicBlocks[:0]
b.doneBlockLayout = false
for i := range b.valueRefCounts {
b.valueRefCounts[i] = 0
}
b.currentSourceOffset = sourceOffsetUnknown
}
@ -355,7 +376,7 @@ func (b *builder) Idom(blk BasicBlock) BasicBlock {
// InsertInstruction implements Builder.InsertInstruction.
func (b *builder) InsertInstruction(instr *Instruction) {
b.currentBB.InsertInstruction(instr)
b.currentBB.insertInstruction(b, instr)
if l := b.currentSourceOffset; l.Valid() {
// Emit the source offset info only when the instruction has side effect because
@ -377,7 +398,7 @@ func (b *builder) InsertInstruction(instr *Instruction) {
}
r1 := b.allocateValue(t1)
instr.rValue = r1
instr.rValue = r1.setInstructionID(instr.id)
tsl := len(ts)
if tsl == 0 {
@ -386,20 +407,14 @@ func (b *builder) InsertInstruction(instr *Instruction) {
rValues := b.varLengthPool.Allocate(tsl)
for i := 0; i < tsl; i++ {
rValues = rValues.Append(&b.varLengthPool, b.allocateValue(ts[i]))
rn := b.allocateValue(ts[i])
rValues = rValues.Append(&b.varLengthPool, rn.setInstructionID(instr.id))
}
instr.rValues = rValues
}
// DefineVariable implements Builder.DefineVariable.
func (b *builder) DefineVariable(variable Variable, value Value, block BasicBlock) {
if b.variables[variable].invalid() {
panic("BUG: trying to define variable " + variable.String() + " but is not declared yet")
}
if b.variables[variable] != value.Type() {
panic(fmt.Sprintf("BUG: inconsistent type for variable %d: expected %s but got %s", variable, b.variables[variable], value.Type()))
}
bb := block.(*basicBlock)
bb.lastDefinitions[variable] = value
}
@ -426,20 +441,9 @@ func (b *builder) EntryBlock() BasicBlock {
// DeclareVariable implements Builder.DeclareVariable.
func (b *builder) DeclareVariable(typ Type) Variable {
v := b.allocateVariable()
iv := int(v)
if l := len(b.variables); l <= iv {
b.variables = append(b.variables, make([]Type, 2*(l+1))...)
}
b.variables[v] = typ
return v
}
// allocateVariable allocates a new variable.
func (b *builder) allocateVariable() (ret Variable) {
ret = b.nextVariable
v := b.nextVariable
b.nextVariable++
return
return v.setType(typ)
}
// allocateValue implements Builder.AllocateValue.
@ -475,8 +479,7 @@ func (b *builder) findValueInLinearPath(variable Variable, blk *basicBlock) Valu
// MustFindValue implements Builder.MustFindValue.
func (b *builder) MustFindValue(variable Variable) Value {
typ := b.definedVariableType(variable)
return b.findValue(typ, variable, b.currentBB)
return b.findValue(variable.getType(), variable, b.currentBB)
}
// findValue recursively tries to find the latest definition of a `variable`. The algorithm is described in
@ -504,7 +507,7 @@ func (b *builder) findValue(typ Type, variable Variable, blk *basicBlock) Value
return value
} else if blk.EntryBlock() {
// If this is the entry block, we reach the uninitialized variable which has zero value.
return b.zeros[b.definedVariableType(variable)]
return b.zeros[variable.getType()]
}
if pred := blk.singlePred; pred != nil {
@ -536,14 +539,13 @@ func (b *builder) findValue(typ Type, variable Variable, blk *basicBlock) Value
if uniqueValue != ValueInvalid {
// If all the predecessors have the same definition, we can use that value.
b.DefineVariable(variable, uniqueValue, blk)
b.alias(tmpValue, uniqueValue)
return uniqueValue
} else {
// Otherwise, add the tmpValue to this block as a parameter which may or may not be redundant, but
// later we eliminate trivial params in an optimization pass. This must be done before finding the
// definitions in the predecessors so that we can break the cycle.
blk.addParamOn(tmpValue)
blk.addParamOn(b, tmpValue)
// After the new param is added, we have to manipulate the original branching instructions
// in predecessors so that they would pass the definition of `variable` as the argument to
// the newly added PHI.
@ -566,8 +568,8 @@ func (b *builder) Seal(raw BasicBlock) {
for _, v := range blk.unknownValues {
variable, phiValue := v.variable, v.value
typ := b.definedVariableType(variable)
blk.addParamOn(phiValue)
typ := variable.getType()
blk.addParamOn(b, phiValue)
for i := range blk.preds {
pred := &blk.preds[i]
predValue := b.findValue(typ, variable, pred.blk)
@ -579,15 +581,6 @@ func (b *builder) Seal(raw BasicBlock) {
}
}
// definedVariableType returns the type of the given variable. If the variable is not defined yet, it panics.
func (b *builder) definedVariableType(variable Variable) Type {
typ := b.variables[variable]
if typ.invalid() {
panic(fmt.Sprintf("%s is not defined yet", variable))
}
return typ
}
// Format implements Builder.Format.
func (b *builder) Format() string {
str := strings.Builder{}
@ -689,15 +682,24 @@ func (b *builder) blockIteratorReversePostOrderNext() *basicBlock {
}
}
// ValueRefCounts implements Builder.ValueRefCounts.
func (b *builder) ValueRefCounts() []int {
return b.valueRefCounts
// ValuesInfo implements Builder.ValuesInfo.
func (b *builder) ValuesInfo() []ValueInfo {
return b.valuesInfo
}
// alias records the alias of the given values. The alias(es) will be
// eliminated in the optimization pass via resolveArgumentAlias.
func (b *builder) alias(dst, src Value) {
b.valueIDAliases[dst.ID()] = src
did := int(dst.ID())
if did >= len(b.valuesInfo) {
l := did + 1 - len(b.valuesInfo)
b.valuesInfo = append(b.valuesInfo, make([]ValueInfo, l)...)
view := b.valuesInfo[len(b.valuesInfo)-l:]
for i := range view {
view[i].alias = ValueInvalid
}
}
b.valuesInfo[did].alias = src
}
// resolveArgumentAlias resolves the alias of the arguments of the given instruction.
@ -722,10 +724,13 @@ func (b *builder) resolveArgumentAlias(instr *Instruction) {
// resolveAlias resolves the alias of the given value.
func (b *builder) resolveAlias(v Value) Value {
info := b.valuesInfo
l := ValueID(len(info))
// Some aliases are chained, so we need to resolve them recursively.
for {
if src, ok := b.valueIDAliases[v.ID()]; ok {
v = src
vid := v.ID()
if vid < l && info[vid].alias.Valid() {
v = info[vid].alias
} else {
break
}
@ -773,3 +778,13 @@ func (b *builder) LoopNestingForestRoots() []BasicBlock {
func (b *builder) LowestCommonAncestor(blk1, blk2 BasicBlock) BasicBlock {
return b.sparseTree.findLCA(blk1.ID(), blk2.ID())
}
// InstructionOfValue returns the instruction that produces the given Value, or nil
// if the Value is not produced by any instruction.
func (b *builder) InstructionOfValue(v Value) *Instruction {
instrID := v.instructionID()
if instrID <= 0 {
return nil
}
return b.instructionsPool.View(instrID - 1)
}

View file

@ -25,11 +25,13 @@ type Instruction struct {
v3 Value
vs Values
typ Type
blk BasicBlock
targets []BasicBlock
prev, next *Instruction
rValue Value
// rValue is the (first) return value of this instruction.
// For branching instructions except for OpcodeBrTable, they hold BlockID to jump cast to Value.
rValue Value
// rValues are the rest of the return values of this instruction.
// For OpcodeBrTable, it holds the list of BlockID to jump cast to Value.
rValues Values
gid InstructionGroupID
sourceOffset SourceOffset
@ -105,6 +107,9 @@ type Instruct
// Returns Value(s) produced by this instruction if any.
// The `first` is the first return value, and `rest` is the rest of the values.
func (i *Instruction) Returns() (first Value, rest []Value) {
if i.IsBranching() {
return ValueInvalid, nil
}
return i.rValue, i.rValues.View()
}
@ -2077,7 +2082,7 @@ func (i *Instruction) InvertBrx() {
}
// BranchData returns the branch data for this instruction necessary for backends.
func (i *Instruction) BranchData() (condVal Value, blockArgs []Value, target BasicBlock) {
func (i *Instruction) BranchData() (condVal Value, blockArgs []Value, target BasicBlockID) {
switch i.opcode {
case OpcodeJump:
condVal = ValueInvalid
@ -2087,17 +2092,17 @@ func (i *Instruction) BranchData() (condVal Value, blockArgs []Value, target Bas
panic("BUG")
}
blockArgs = i.vs.View()
target = i.blk
target = BasicBlockID(i.rValue)
return
}
// BrTableData returns the branch table data for this instruction necessary for backends.
func (i *Instruction) BrTableData() (index Value, targets []BasicBlock) {
func (i *Instruction) BrTableData() (index Value, targets Values) {
if i.opcode != OpcodeBrTable {
panic("BUG: BrTableData only available for OpcodeBrTable")
}
index = i.v
targets = i.targets
targets = i.rValues
return
}
@ -2105,7 +2110,7 @@ func (i *Instruction) BrTableData() (index Value, targets []BasicBlock) {
func (i *Instruction) AsJump(vs Values, target BasicBlock) *Instruction {
i.opcode = OpcodeJump
i.vs = vs
i.blk = target
i.rValue = Value(target.ID())
return i
}
@ -2130,7 +2135,7 @@ func (i *Instruction) AsBrz(v Value, args Values, target BasicBlock) {
i.opcode = OpcodeBrz
i.v = v
i.vs = args
i.blk = target
i.rValue = Value(target.ID())
}
// AsBrnz initializes this instruction as a branch-if-not-zero instruction with OpcodeBrnz.
@ -2138,15 +2143,16 @@ func (i *Instruction) AsBrnz(v Value, args Values, target BasicBlock) *Instructi
i.opcode = OpcodeBrnz
i.v = v
i.vs = args
i.blk = target
i.rValue = Value(target.ID())
return i
}
// AsBrTable initializes this instruction as a branch-table instruction with OpcodeBrTable.
func (i *Instruction) AsBrTable(index Value, targets []BasicBlock) {
// targets is a list of basic block IDs cast to Values.
func (i *Instruction) AsBrTable(index Value, targets Values) {
i.opcode = OpcodeBrTable
i.v = index
i.targets = targets
i.rValues = targets
}
// AsCall initializes this instruction as a call instruction with OpcodeCall.
@ -2531,7 +2537,8 @@ func (i *Instruction) Format(b Builder) string {
if i.IsFallthroughJump() {
vs[0] = " fallthrough"
} else {
vs[0] = " " + i.blk.(*basicBlock).Name()
blockId := BasicBlockID(i.rValue)
vs[0] = " " + b.BasicBlock(blockId).Name()
}
for idx := range view {
vs[idx+1] = view[idx].Format(b)
@ -2542,7 +2549,8 @@ func (i *Instruction) Format(b Builder) string {
view := i.vs.View()
vs := make([]string, len(view)+2)
vs[0] = " " + i.v.Format(b)
vs[1] = i.blk.(*basicBlock).Name()
blockId := BasicBlockID(i.rValue)
vs[1] = b.BasicBlock(blockId).Name()
for idx := range view {
vs[idx+2] = view[idx].Format(b)
}
@ -2551,8 +2559,8 @@ func (i *Instruction) Format(b Builder) string {
// `BrTable index, [label1, label2, ... labelN]`
instSuffix = fmt.Sprintf(" %s", i.v.Format(b))
instSuffix += ", ["
for i, target := range i.targets {
blk := target.(*basicBlock)
for i, target := range i.rValues.View() {
blk := b.BasicBlock(BasicBlockID(target))
if i == 0 {
instSuffix += blk.Name()
} else {
@ -2621,11 +2629,12 @@ func (i *Instruction) Format(b Builder) string {
instr := i.opcode.String() + instSuffix
var rvs []string
if rv := i.rValue; rv.Valid() {
rvs = append(rvs, rv.formatWithType(b))
r1, rs := i.Returns()
if r1.Valid() {
rvs = append(rvs, r1.formatWithType(b))
}
for _, v := range i.rValues.View() {
for _, v := range rs {
rvs = append(rvs, v.formatWithType(b))
}

View file

@ -112,7 +112,7 @@ func passDeadBlockEliminationOpt(b *builder) {
// This requires the reverse post-order traversal to be calculated before calling this function,
// hence passCalculateImmediateDominators must be called before this.
func passRedundantPhiEliminationOpt(b *builder) {
redundantParameterIndexes := b.ints[:0] // reuse the slice from previous iterations.
redundantParams := b.redundantParams[:0] // reuse the slice from previous iterations.
// TODO: this might be costly for large programs, but at least, as far as I did the experiment, it's almost the
// same as the single iteration version in terms of the overall compilation time. That *might be* mostly thanks to the fact
@ -128,10 +128,11 @@ func passRedundantPhiEliminationOpt(b *builder) {
_ = b.blockIteratorReversePostOrderBegin() // skip entry block!
// Below, we intentionally use the named iteration variable name, as this comes with inevitable nested for loops!
for blk := b.blockIteratorReversePostOrderNext(); blk != nil; blk = b.blockIteratorReversePostOrderNext() {
paramNum := len(blk.params)
params := blk.params.View()
paramNum := len(params)
for paramIndex := 0; paramIndex < paramNum; paramIndex++ {
phiValue := blk.params[paramIndex]
phiValue := params[paramIndex]
redundant := true
nonSelfReferencingValue := ValueInvalid
@ -162,55 +163,58 @@ func passRedundantPhiEliminationOpt(b *builder) {
}
if redundant {
b.redundantParameterIndexToValue[paramIndex] = nonSelfReferencingValue
redundantParameterIndexes = append(redundantParameterIndexes, paramIndex)
redundantParams = append(redundantParams, redundantParam{
index: paramIndex, uniqueValue: nonSelfReferencingValue,
})
}
}
if len(b.redundantParameterIndexToValue) == 0 {
if len(redundantParams) == 0 {
continue
}
changed = true
// Remove the redundant PHIs from the argument list of branching instructions.
for predIndex := range blk.preds {
var cur int
redundantParamsCur, predParamCur := 0, 0
predBlk := blk.preds[predIndex]
branchInst := predBlk.branch
view := branchInst.vs.View()
for argIndex, value := range view {
if _, ok := b.redundantParameterIndexToValue[argIndex]; !ok {
view[cur] = value
cur++
if len(redundantParams) == redundantParamsCur ||
redundantParams[redundantParamsCur].index != argIndex {
view[predParamCur] = value
predParamCur++
} else {
redundantParamsCur++
}
}
branchInst.vs.Cut(cur)
branchInst.vs.Cut(predParamCur)
}
// Still need to have the definition of the value of the PHI (previously as the parameter).
for _, redundantParamIndex := range redundantParameterIndexes {
phiValue := blk.params[redundantParamIndex]
onlyValue := b.redundantParameterIndexToValue[redundantParamIndex]
for i := range redundantParams {
redundantValue := &redundantParams[i]
phiValue := params[redundantValue.index]
// Create an alias in this block from the only phi argument to the phi value.
b.alias(phiValue, onlyValue)
b.alias(phiValue, redundantValue.uniqueValue)
}
// Finally, Remove the param from the blk.
var cur int
paramsCur, redundantParamsCur := 0, 0
for paramIndex := 0; paramIndex < paramNum; paramIndex++ {
param := blk.params[paramIndex]
if _, ok := b.redundantParameterIndexToValue[paramIndex]; !ok {
blk.params[cur] = param
cur++
param := params[paramIndex]
if len(redundantParams) == redundantParamsCur || redundantParams[redundantParamsCur].index != paramIndex {
params[paramsCur] = param
paramsCur++
} else {
redundantParamsCur++
}
}
blk.params = blk.params[:cur]
blk.params.Cut(paramsCur)
// Clears the map for the next iteration.
for _, paramIndex := range redundantParameterIndexes {
delete(b.redundantParameterIndexToValue, paramIndex)
}
redundantParameterIndexes = redundantParameterIndexes[:0]
redundantParams = redundantParams[:0]
}
if !changed {
@ -219,7 +223,7 @@ func passRedundantPhiEliminationOpt(b *builder) {
}
// Reuse the slice for the future passes.
b.ints = redundantParameterIndexes
b.redundantParams = redundantParams
}
// passDeadCodeEliminationOpt traverses all the instructions, and calculates the reference count of each Value, and
@ -231,11 +235,13 @@ func passRedundantPhiEliminationOpt(b *builder) {
// TODO: the algorithm here might not be efficient. Get back to this later.
func passDeadCodeEliminationOpt(b *builder) {
nvid := int(b.nextValueID)
if nvid >= len(b.valueRefCounts) {
b.valueRefCounts = append(b.valueRefCounts, make([]int, nvid-len(b.valueRefCounts)+1)...)
}
if nvid >= len(b.valueIDToInstruction) {
b.valueIDToInstruction = append(b.valueIDToInstruction, make([]*Instruction, nvid-len(b.valueIDToInstruction)+1)...)
if nvid >= len(b.valuesInfo) {
l := nvid - len(b.valuesInfo) + 1
b.valuesInfo = append(b.valuesInfo, make([]ValueInfo, l)...)
view := b.valuesInfo[len(b.valuesInfo)-l:]
for i := range view {
view[i].alias = ValueInvalid
}
}
// First, we gather all the instructions with side effects.
@ -255,14 +261,6 @@ func passDeadCodeEliminationOpt(b *builder) {
// The strict side effect should create different instruction groups.
gid++
}
r1, rs := cur.Returns()
if r1.Valid() {
b.valueIDToInstruction[r1.ID()] = cur
}
for _, r := range rs {
b.valueIDToInstruction[r.ID()] = cur
}
}
}
@ -283,28 +281,28 @@ func passDeadCodeEliminationOpt(b *builder) {
v1, v2, v3, vs := live.Args()
if v1.Valid() {
producingInst := b.valueIDToInstruction[v1.ID()]
producingInst := b.InstructionOfValue(v1)
if producingInst != nil {
liveInstructions = append(liveInstructions, producingInst)
}
}
if v2.Valid() {
producingInst := b.valueIDToInstruction[v2.ID()]
producingInst := b.InstructionOfValue(v2)
if producingInst != nil {
liveInstructions = append(liveInstructions, producingInst)
}
}
if v3.Valid() {
producingInst := b.valueIDToInstruction[v3.ID()]
producingInst := b.InstructionOfValue(v3)
if producingInst != nil {
liveInstructions = append(liveInstructions, producingInst)
}
}
for _, v := range vs {
producingInst := b.valueIDToInstruction[v.ID()]
producingInst := b.InstructionOfValue(v)
if producingInst != nil {
liveInstructions = append(liveInstructions, producingInst)
}
@ -352,34 +350,19 @@ func (b *builder) incRefCount(id ValueID, from *Instruction) {
if wazevoapi.SSALoggingEnabled {
fmt.Printf("v%d referenced from %v\n", id, from.Format(b))
}
b.valueRefCounts[id]++
info := &b.valuesInfo[id]
info.RefCount++
}
// passNopInstElimination eliminates the instructions which is essentially a no-op.
func passNopInstElimination(b *builder) {
if int(b.nextValueID) >= len(b.valueIDToInstruction) {
b.valueIDToInstruction = append(b.valueIDToInstruction, make([]*Instruction, int(b.nextValueID)-len(b.valueIDToInstruction)+1)...)
}
for blk := b.blockIteratorBegin(); blk != nil; blk = b.blockIteratorNext() {
for cur := blk.rootInstr; cur != nil; cur = cur.next {
r1, rs := cur.Returns()
if r1.Valid() {
b.valueIDToInstruction[r1.ID()] = cur
}
for _, r := range rs {
b.valueIDToInstruction[r.ID()] = cur
}
}
}
for blk := b.blockIteratorBegin(); blk != nil; blk = b.blockIteratorNext() {
for cur := blk.rootInstr; cur != nil; cur = cur.next {
switch cur.Opcode() {
// TODO: add more logics here.
case OpcodeIshl, OpcodeSshr, OpcodeUshr:
x, amount := cur.Arg2()
definingInst := b.valueIDToInstruction[amount.ID()]
definingInst := b.InstructionOfValue(amount)
if definingInst == nil {
// If there's no defining instruction, that means the amount is coming from the parameter.
continue

View file

@ -33,7 +33,7 @@ func passLayoutBlocks(b *builder) {
}
nonSplitBlocks = append(nonSplitBlocks, blk)
if i != len(b.reversePostOrderedBasicBlocks)-1 {
_ = maybeInvertBranches(blk, b.reversePostOrderedBasicBlocks[i+1])
_ = maybeInvertBranches(b, blk, b.reversePostOrderedBasicBlocks[i+1])
}
}
@ -111,7 +111,7 @@ func passLayoutBlocks(b *builder) {
}
fallthroughBranch := blk.currentInstr
if fallthroughBranch.opcode == OpcodeJump && fallthroughBranch.blk == trampoline {
if fallthroughBranch.opcode == OpcodeJump && BasicBlockID(fallthroughBranch.rValue) == trampoline.id {
// This can be lowered as fallthrough at the end of the block.
b.reversePostOrderedBasicBlocks = append(b.reversePostOrderedBasicBlocks, trampoline)
trampoline.visited = 1 // mark as inserted.
@ -157,7 +157,7 @@ func (b *builder) markFallthroughJumps() {
for i, blk := range b.reversePostOrderedBasicBlocks {
if i < l {
cur := blk.currentInstr
if cur.opcode == OpcodeJump && cur.blk == b.reversePostOrderedBasicBlocks[i+1] {
if cur.opcode == OpcodeJump && BasicBlockID(cur.rValue) == b.reversePostOrderedBasicBlocks[i+1].id {
cur.AsFallthroughJump()
}
}
@ -168,7 +168,7 @@ func (b *builder) markFallthroughJumps() {
// nextInRPO is the next block in the reverse post-order.
//
// Returns true if the branch is inverted for testing purpose.
func maybeInvertBranches(now *basicBlock, nextInRPO *basicBlock) bool {
func maybeInvertBranches(b *builder, now *basicBlock, nextInRPO *basicBlock) bool {
fallthroughBranch := now.currentInstr
if fallthroughBranch.opcode == OpcodeBrTable {
return false
@ -187,7 +187,8 @@ func maybeInvertBranches(now *basicBlock, nextInRPO *basicBlock) bool {
// So this block has two branches (a conditional branch followed by an unconditional branch) at the end.
// We can invert the condition of the branch if it makes the fallthrough more likely.
fallthroughTarget, condTarget := fallthroughBranch.blk.(*basicBlock), condBranch.blk.(*basicBlock)
fallthroughTarget := b.basicBlock(BasicBlockID(fallthroughBranch.rValue))
condTarget := b.basicBlock(BasicBlockID(condBranch.rValue))
if fallthroughTarget.loopHeader {
// First, if the tail's target is loopHeader, we don't need to do anything here,
@ -231,8 +232,8 @@ func maybeInvertBranches(now *basicBlock, nextInRPO *basicBlock) bool {
}
condBranch.InvertBrx()
condBranch.blk = fallthroughTarget
fallthroughBranch.blk = condTarget
condBranch.rValue = Value(fallthroughTarget.ID())
fallthroughBranch.rValue = Value(condTarget.ID())
if wazevoapi.SSALoggingEnabled {
fmt.Printf("inverting branches at %d->%d and %d->%d\n",
now.ID(), fallthroughTarget.ID(), now.ID(), condTarget.ID())
@ -275,7 +276,7 @@ func (b *builder) splitCriticalEdge(pred, succ *basicBlock, predInfo *basicBlock
// Replace originalBranch with the newBranch.
newBranch := b.AllocateInstruction()
newBranch.opcode = originalBranch.opcode
newBranch.blk = trampoline
newBranch.rValue = Value(trampoline.ID())
switch originalBranch.opcode {
case OpcodeJump:
case OpcodeBrz, OpcodeBrnz:
@ -303,7 +304,7 @@ func (b *builder) splitCriticalEdge(pred, succ *basicBlock, predInfo *basicBlock
trampoline.validate(b)
}
if len(trampoline.params) > 0 {
if len(trampoline.params.View()) > 0 {
panic("trampoline should not have params")
}

View file

@ -15,17 +15,31 @@
//
// Variable is useful to track the SSA Values of a variable in the source program, and
// can be used to find the corresponding latest SSA Value via Builder.FindValue.
//
// Higher 4-bit is used to store Type for this variable.
type Variable uint32
// String implements fmt.Stringer.
func (v Variable) String() string {
return fmt.Sprintf("var%d", v)
return fmt.Sprintf("var%d", v&0x0fffffff)
}
func (v Variable) setType(typ Type) Variable {
if v >= 1<<28 {
panic(fmt.Sprintf("Too large variable: %d", v))
}
return Variable(typ)<<28 | v
}
func (v Variable) getType() Type {
return Type(v >> 28)
}
// Value represents an SSA value with a type information. The relationship with Variable is 1: N (including 0),
// that means there might be multiple Variable(s) for a Value.
//
// Higher 32-bit is used to store Type for this value.
// 32 to 59-bit is used to store the unique identifier of the Instruction that generates this value if any.
// 60 to 63-bit is used to store Type for this value.
type Value uint64
// ValueID is the lower 32bit of Value, which is the pure identifier of Value without type info.
@ -33,7 +47,7 @@ func (v Variable) String() string {
const (
valueIDInvalid ValueID = math.MaxUint32
ValueInvalid Value = Value(valueIDInvalid)
ValueInvalid = Value(valueIDInvalid)
)
// Format creates a debug string for this Value using the data stored in Builder.
@ -54,7 +68,7 @@ func (v Value) formatWithType(b Builder) (ret string) {
if wazevoapi.SSALoggingEnabled { // This is useful to check live value analysis bugs.
if bd := b.(*builder); bd.donePostBlockLayoutPasses {
id := v.ID()
ret += fmt.Sprintf("(ref=%d)", bd.valueRefCounts[id])
ret += fmt.Sprintf("(ref=%d)", bd.valuesInfo[id].RefCount)
}
}
return ret
@ -67,7 +81,7 @@ func (v Value) Valid() bool {
// Type returns the Type of this value.
func (v Value) Type() Type {
return Type(v >> 32)
return Type(v >> 60)
}
// ID returns the valueID of this value.
@ -77,7 +91,20 @@ func (v Value) ID() ValueID {
// setType sets a type to this Value and returns the updated Value.
func (v Value) setType(typ Type) Value {
return v | Value(typ)<<32
return v | Value(typ)<<60
}
// setInstructionID sets an Instruction.id to this Value and returns the updated Value.
func (v Value) setInstructionID(id int) Value {
if id < 0 || uint(id) >= 1<<28 {
panic(fmt.Sprintf("Too large instruction ID: %d", id))
}
return v | Value(id)<<32
}
// instructionID() returns the Instruction.id of this Value.
func (v Value) instructionID() int {
return int(v>>32) & 0x0fffffff
}
// Values is a slice of Value. Use this instead of []Value to reuse the underlying memory.

View file

@ -5,9 +5,7 @@ func ResetMap[K comparable, V any](m map[K]V) map[K]V {
if m == nil {
m = make(map[K]V)
} else {
for v := range m {
delete(m, v)
}
clear(m)
}
return m
}

View file

@ -0,0 +1,6 @@
package expctxkeys
// ImportResolverKey is a context.Context Value key.
// Its associated value should be an ImportResolver.
// See issue 2294.
type ImportResolverKey struct{}

View file

@ -1,23 +0,0 @@
//go:build !(darwin || linux || freebsd) || tinygo
package platform
func remapCodeSegmentAMD64(code []byte, size int) ([]byte, error) {
b, err := mmapCodeSegmentAMD64(size)
if err != nil {
return nil, err
}
copy(b, code)
mustMunmapCodeSegment(code)
return b, nil
}
func remapCodeSegmentARM64(code []byte, size int) ([]byte, error) {
b, err := mmapCodeSegmentARM64(size)
if err != nil {
return nil, err
}
copy(b, code)
mustMunmapCodeSegment(code)
return b, nil
}

View file

@ -1,21 +0,0 @@
//go:build (darwin || linux || freebsd) && !tinygo
package platform
func remapCodeSegmentAMD64(code []byte, size int) ([]byte, error) {
return remapCodeSegment(code, size, mmapProtAMD64)
}
func remapCodeSegmentARM64(code []byte, size int) ([]byte, error) {
return remapCodeSegment(code, size, mmapProtARM64)
}
func remapCodeSegment(code []byte, size, prot int) ([]byte, error) {
b, err := mmapCodeSegment(size, prot)
if err != nil {
return nil, err
}
copy(b, code)
mustMunmapCodeSegment(code)
return b, nil
}

View file

@ -36,28 +36,6 @@ func MmapCodeSegment(size int) ([]byte, error) {
}
}
// RemapCodeSegment reallocates the memory mapping of an existing code segment
// to increase its size. The previous code mapping is unmapped and must not be
// reused after the function returns.
//
// This is similar to mremap(2) on linux, and emulated on platforms which do not
// have this syscall.
//
// See https://man7.org/linux/man-pages/man2/mremap.2.html
func RemapCodeSegment(code []byte, size int) ([]byte, error) {
if size < len(code) {
panic("BUG: RemapCodeSegment with size less than code")
}
if code == nil {
return MmapCodeSegment(size)
}
if runtime.GOARCH == "amd64" {
return remapCodeSegmentAMD64(code, size)
} else {
return remapCodeSegmentARM64(code, size)
}
}
// MunmapCodeSegment unmaps the given memory region.
func MunmapCodeSegment(code []byte) error {
if len(code) == 0 {
@ -65,17 +43,3 @@ func MunmapCodeSegment(code []byte) error {
}
return munmapCodeSegment(code)
}
// mustMunmapCodeSegment panics instead of returning an error to the
// application.
//
// # Why panic?
//
// It is less disruptive to the application to leak the previous block if it
// could be unmapped than to leak the new block and return an error.
// Realistically, either scenarios are pretty hard to debug, so we panic.
func mustMunmapCodeSegment(code []byte) {
if err := munmapCodeSegment(code); err != nil {
panic(err)
}
}

View file

@ -54,7 +54,6 @@ func decodeUTF8(r *bytes.Reader, contextFormat string, contextArgs ...interface{
return "", 0, fmt.Errorf("%s is not valid UTF-8", fmt.Sprintf(contextFormat, contextArgs...))
}
// TODO: use unsafe.String after flooring Go 1.20.
ret := *(*string)(unsafe.Pointer(&buf))
ret := unsafe.String(&buf[0], int(size))
return ret, size + uint32(sizeOfSize), nil
}

View file

@ -451,14 +451,14 @@ functionType := &m.TypeSection[m.FunctionSection[idx]]
return fmt.Errorf("read immediate: %w", err)
}
list := make([]uint32, nl)
sts.ls = sts.ls[:0]
for i := uint32(0); i < nl; i++ {
l, n, err := leb128.DecodeUint32(br)
if err != nil {
return fmt.Errorf("read immediate: %w", err)
}
num += n
list[i] = l
sts.ls = append(sts.ls, l)
}
ln, n, err := leb128.DecodeUint32(br)
if err != nil {
@ -511,7 +511,7 @@ functionType := &m.TypeSection[m.FunctionSection[idx]]
}
}
for _, l := range list {
for _, l := range sts.ls {
if int(l) >= len(controlBlockStack.stack) {
return fmt.Errorf("invalid l param given for %s", OpcodeBrTableName)
}
@ -2003,6 +2003,8 @@ funcType := &m.TypeSection[typeIndex]
type stacks struct {
vs valueTypeStack
cs controlBlockStack
// ls is the label slice that is reused for each br_table instruction.
ls []uint32
}
func (sts *stacks) reset(functionType *FunctionType) {
@ -2012,6 +2014,7 @@ func (sts *stacks) reset(functionType *FunctionType) {
sts.vs.maximumStackPointer = 0
sts.cs.stack = sts.cs.stack[:0]
sts.cs.stack = append(sts.cs.stack, controlBlock{blockType: functionType})
sts.ls = sts.ls[:0]
}
type controlBlockStack struct {

View file

@ -52,7 +52,8 @@ type MemoryInstance struct {
definition api.MemoryDefinition
// Mux is used in interpreter mode to prevent overlapping calls to atomic instructions,
// introduced with WebAssembly threads proposal.
// introduced with WebAssembly threads proposal, and in compiler mode to make memory modifications
// within Grow non-racy for the Go race detector.
Mux sync.Mutex
// waiters implements atomic wait and notify. It is implemented similarly to golang.org/x/sync/semaphore,
@ -227,6 +228,11 @@ func MemoryPagesToBytesNum(pages uint32) (bytesNum uint64) {
// Grow implements the same method as documented on api.Memory.
func (m *MemoryInstance) Grow(delta uint32) (result uint32, ok bool) {
if m.Shared {
m.Mux.Lock()
defer m.Mux.Unlock()
}
currentPages := m.Pages()
if delta == 0 {
return currentPages, true
@ -299,6 +305,7 @@ func PagesToUnitOfBytes(pages uint32) string {
// Uses atomic write to update the length of a slice.
func atomicStoreLengthAndCap(slice *[]byte, length uintptr, cap uintptr) {
//nolint:staticcheck
slicePtr := (*reflect.SliceHeader)(unsafe.Pointer(slice))
capPtr := (*uintptr)(unsafe.Pointer(&slicePtr.Cap))
atomic.StoreUintptr(capPtr, cap)
@ -308,6 +315,7 @@ func atomicStoreLengthAndCap(slice *[]byte, length uintptr, cap uintptr) {
// Uses atomic write to update the length of a slice.
func atomicStoreLength(slice *[]byte, length uintptr) {
//nolint:staticcheck
slicePtr := (*reflect.SliceHeader)(unsafe.Pointer(slice))
lenPtr := (*uintptr)(unsafe.Pointer(&slicePtr.Len))
atomic.StoreUintptr(lenPtr, length)

View file

@ -3,6 +3,7 @@
import (
"context"
"encoding/binary"
"errors"
"fmt"
"sync"
"sync/atomic"
@ -352,7 +353,7 @@ func (s *Store) instantiate(
return nil, err
}
if err = m.resolveImports(module); err != nil {
if err = m.resolveImports(ctx, module); err != nil {
return nil, err
}
@ -410,12 +411,22 @@ funcIdx := *module.StartSection
return
}
func (m *ModuleInstance) resolveImports(module *Module) (err error) {
func (m *ModuleInstance) resolveImports(ctx context.Context, module *Module) (err error) {
// Check if ctx contains an ImportResolver.
resolveImport, _ := ctx.Value(expctxkeys.ImportResolverKey{}).(experimental.ImportResolver)
for moduleName, imports := range module.ImportPerModule {
var importedModule *ModuleInstance
importedModule, err = m.s.module(moduleName)
if err != nil {
return err
if resolveImport != nil {
if v := resolveImport(moduleName); v != nil {
importedModule = v.(*ModuleInstance)
}
}
if importedModule == nil {
importedModule, err = m.s.module(moduleName)
if err != nil {
return err
}
}
for _, i := range imports {
@ -649,20 +660,20 @@ func (s *Store) GetFunctionTypeID(t *FunctionType) (FunctionTypeID, error) {
}
// CloseWithExitCode implements the same method as documented on wazero.Runtime.
func (s *Store) CloseWithExitCode(ctx context.Context, exitCode uint32) (err error) {
func (s *Store) CloseWithExitCode(ctx context.Context, exitCode uint32) error {
s.mux.Lock()
defer s.mux.Unlock()
// Close modules in reverse initialization order.
var errs []error
for m := s.moduleList; m != nil; m = m.next {
// If closing this module errs, proceed anyway to close the others.
if e := m.closeWithExitCode(ctx, exitCode); e != nil && err == nil {
// TODO: use multiple errors handling in Go 1.20.
err = e // first error
if err := m.closeWithExitCode(ctx, exitCode); err != nil {
errs = append(errs, err)
}
}
s.moduleList = nil
s.nameToModule = nil
s.nameToModuleCap = 0
s.typeIDs = nil
return
return errors.Join(errs...)
}

View file

@ -3,8 +3,6 @@
import (
"errors"
"fmt"
"github.com/tetratelabs/wazero/api"
)
// deleteModule makes the moduleName available for instantiation again.
@ -88,7 +86,7 @@ func (s *Store) registerModule(m *ModuleInstance) error {
}
// Module implements wazero.Runtime Module
func (s *Store) Module(moduleName string) api.Module {
func (s *Store) Module(moduleName string) *ModuleInstance {
m, err := s.module(moduleName)
if err != nil {
return nil

View file

@ -171,7 +171,6 @@ func (d *DWARFLines) Line(instructionOffset uint64) (ret []string) {
// Advance the line reader for the found position.
lineReader.Seek(ln.pos)
err = lineReader.Next(&le)
if err != nil {
// If we reach this block, that means there's a bug in the []line creation logic above.
panic("BUG: stored dwarf.LineReaderPos is invalid")

View file

@ -197,7 +197,13 @@ func (r *runtime) Module(moduleName string) api.Module {
if len(moduleName) == 0 {
return nil
}
return r.store.Module(moduleName)
m := r.store.Module(moduleName)
if m == nil {
return nil
} else if m.Source.IsHostModule {
return hostModuleInstance{m}
}
return m
}
// CompileModule implements Runtime.CompileModule

View file

@ -7,9 +7,6 @@
// sysParseable is only used here as we define "supported" as being able to
// parse `info.Sys()`. The above `go:build` constraints exclude 32-bit until
// that's requested.
//
// TODO: When Go 1.21 is out, use the "unix" build constraint (as 1.21 makes
// our floor Go version 1.19.
const sysParseable = false
func statFromFileInfo(info fs.FileInfo) Stat_t {

6
vendor/modules.txt vendored
View file

@ -30,7 +30,7 @@ codeberg.org/gruf/go-fastcopy
# codeberg.org/gruf/go-fastpath/v2 v2.0.0
## explicit; go 1.14
codeberg.org/gruf/go-fastpath/v2
# codeberg.org/gruf/go-ffmpreg v0.2.4
# codeberg.org/gruf/go-ffmpreg v0.2.5
## explicit; go 1.22.0
codeberg.org/gruf/go-ffmpreg/embed/ffmpeg
codeberg.org/gruf/go-ffmpreg/embed/ffprobe
@ -842,8 +842,8 @@ github.com/tdewolff/parse/v2/strconv
# github.com/technologize/otel-go-contrib v1.1.1
## explicit; go 1.17
github.com/technologize/otel-go-contrib/otelginmetrics
# github.com/tetratelabs/wazero v1.7.3
## explicit; go 1.20
# github.com/tetratelabs/wazero v1.8.0
## explicit; go 1.21
github.com/tetratelabs/wazero
github.com/tetratelabs/wazero/api
github.com/tetratelabs/wazero/experimental