main: release version 0.19.0

Signed-off-by: deadprogram <ron@hybridgroup.com>

.circleci/config.yml

@@ -68,14 +68,17 @@ commands:
     steps:
       - restore_cache:
           keys:
-            - llvm-source-11-v1
+            - llvm-source-11-v2
       - run:
           name: "Fetch LLVM source"
           command: make llvm-source
       - save_cache:
-          key: llvm-source-11-v1
+          key: llvm-source-11-v2
           paths:
-            - llvm-project
+            - llvm-project/clang/lib/Headers
+            - llvm-project/clang/include
+            - llvm-project/lld/include
+            - llvm-project/llvm/include
   build-wasi-libc:
     steps:
       - restore_cache:
@@ -154,12 +157,15 @@ commands:
       - llvm-source-linux
       - restore_cache:
           keys:
-            - llvm-build-11-linux-v2-assert
+            - llvm-build-11-linux-v4-assert
       - run:
           name: "Build LLVM"
           command: |
             if [ ! -f llvm-build/lib/liblldELF.a ]
             then
+              # fetch LLVM source
+              rm -rf llvm-project
+              make llvm-source
               # install dependencies
               sudo apt-get install cmake ninja-build
               # hack ninja to use less jobs
@@ -167,20 +173,22 @@ commands:
               chmod +x /go/bin/ninja
               # build!
               make ASSERT=1 llvm-build
+              find llvm-build -name CMakeFiles -prune -exec rm -r '{}' \;
             fi
       - save_cache:
-          key: llvm-build-11-linux-v2-assert
+          key: llvm-build-11-linux-v4-assert
           paths:
             llvm-build
-      - run: |
-          # Note: -p=2 limits parallelism to two jobs at a time, which is
-          # necessary to keep memory consumption down and avoid OOM (for a
-          # 2CPU/4GB executor).
-          GOFLAGS="-p=2" make ASSERT=1
+      - run: make ASSERT=1
       - build-wasi-libc
       - run:
           name: "Test TinyGo"
           command: make ASSERT=1 test
+          environment:
+            # Note: -p=2 limits parallelism to two jobs at a time, which is
+            # necessary to keep memory consumption down and avoid OOM (for a
+            # 2CPU/4GB executor).
+            GOFLAGS: -p=2
       - save_cache:
           key: go-cache-v2-{{ checksum "go.mod" }}-{{ .Environment.CIRCLE_BUILD_NUM }}
           paths:
@@ -217,12 +225,15 @@ commands:
       - llvm-source-linux
       - restore_cache:
           keys:
-            - llvm-build-11-linux-v2-noassert
+            - llvm-build-11-linux-v4-noassert
       - run:
           name: "Build LLVM"
           command: |
             if [ ! -f llvm-build/lib/liblldELF.a ]
             then
+              # fetch LLVM source
+              rm -rf llvm-project
+              make llvm-source
               # install dependencies
               sudo apt-get install cmake ninja-build
               # hack ninja to use less jobs
@@ -230,9 +241,10 @@ commands:
               chmod +x /go/bin/ninja
               # build!
               make llvm-build
+              find llvm-build -name CMakeFiles -prune -exec rm -r '{}' \;
             fi
       - save_cache:
-          key: llvm-build-11-linux-v2-noassert
+          key: llvm-build-11-linux-v4-noassert
           paths:
             llvm-build
       - build-wasi-libc
@@ -282,43 +294,50 @@ commands:
           variant: "macos"
       - restore_cache:
           keys:
-            - go-cache-macos-v2-{{ checksum "go.mod" }}-{{ .Environment.CIRCLE_PREVIOUS_BUILD_NUM }}
-            - go-cache-macos-v2-{{ checksum "go.mod" }}
+            - go-cache-macos-v3-{{ checksum "go.mod" }}-{{ .Environment.CIRCLE_PREVIOUS_BUILD_NUM }}
+            - go-cache-macos-v3-{{ checksum "go.mod" }}
       - restore_cache:
           keys:
-            - llvm-source-11-macos-v1
+            - llvm-source-11-macos-v3
       - run:
           name: "Fetch LLVM source"
           command: make llvm-source
       - save_cache:
-          key: llvm-source-11-macos-v1
+          key: llvm-source-11-macos-v3
           paths:
-            - llvm-project
+            - llvm-project/clang/lib/Headers
+            - llvm-project/clang/include
+            - llvm-project/lld/include
+            - llvm-project/llvm/include
       - restore_cache:
           keys:
-            - llvm-build-11-macos-v2
+            - llvm-build-11-macos-v5
       - run:
           name: "Build LLVM"
           command: |
             if [ ! -f llvm-build/lib/liblldELF.a ]
             then
+              # fetch LLVM source
+              rm -rf llvm-project
+              make llvm-source
               # install dependencies
               HOMEBREW_NO_AUTO_UPDATE=1 brew install cmake ninja
               # build!
               make llvm-build
+              find llvm-build -name CMakeFiles -prune -exec rm -r '{}' \;
             fi
       - save_cache:
-          key: llvm-build-11-macos-v2
+          key: llvm-build-11-macos-v5
           paths:
             llvm-build
       - restore_cache:
           keys:
-            - wasi-libc-sysroot-macos-v3
+            - wasi-libc-sysroot-macos-v4
       - run:
           name: "Build wasi-libc"
           command: make wasi-libc
       - save_cache:
-          key: wasi-libc-sysroot-macos-v3
+          key: wasi-libc-sysroot-macos-v4
           paths:
             - lib/wasi-libc/sysroot
       - run:
@@ -340,7 +359,7 @@ commands:
             tinygo version
       - run: make smoketest AVR=0
       - save_cache:
-          key: go-cache-macos-v2-{{ checksum "go.mod" }}-{{ .Environment.CIRCLE_BUILD_NUM }}
+          key: go-cache-macos-v3-{{ checksum "go.mod" }}-{{ .Environment.CIRCLE_BUILD_NUM }}
           paths:
             - ~/.cache/go-build
             - /go/pkg/mod
@@ -382,7 +401,7 @@ jobs:
       - build-linux
   build-macos:
     macos:
-      xcode: "10.1.0"
+      xcode: "11.1.0" # macOS 10.14
     steps:
       - build-macos
 

.github/workflows/build-tinygo-dev-docker.yml

@@ -0,0 +1,45 @@
+name: CI for tinygo-dev docker container
+on:
+  push:
+    branches: [ dev ]
+
+jobs:
+  push_to_registry:
+    name: Push Docker image to GHCR/Docker Hub
+    runs-on: ubuntu-latest
+    permissions:
+      packages: write
+      contents: read
+    steps:
+      - name: Check out the repo
+        uses: actions/checkout@v2
+        with:
+          submodules: recursive
+      - name: Docker meta
+        id: meta
+        uses: docker/metadata-action@v3
+        with:
+          images: |
+            tinygo/tinygo-dev
+            ghcr.io/${{ github.repository }}/tinygo-dev
+          tags: |
+            type=sha,format=long
+            type=raw,value=latest
+      - name: Log in to Docker Hub
+        uses: docker/login-action@v1
+        with:
+          username: ${{ secrets.DOCKER_HUB_USERNAME }}
+          password: ${{ secrets.DOCKER_HUB_ACCESS_TOKEN }}
+      - name: Log in to Github Container Registry
+        uses: docker/login-action@v1
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Build and push
+        uses: docker/build-push-action@v2
+        with:
+          context: .
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}

.gitignore

@@ -16,6 +16,8 @@ src/device/stm32/*.go
 src/device/stm32/*.s
 src/device/kendryte/*.go
 src/device/kendryte/*.s
+src/device/rp/*.go
+src/device/rp/*.s
 vendor
 llvm-build
 llvm-project

CHANGELOG.md

@@ -1,3 +1,191 @@
+0.19.0
+---
+
+* **command line**
+  - don't consider compile-only tests as failing
+  - add -test flag for `tinygo list`
+  - escape commands while printing them with the -x flag
+  - make flash-command portable and safer to use
+  - use `extended-remote` instead of `remote` in GDB
+  - detect specific serial port IDs based on USB vid/pid
+  - add a flag to the command line to select the serial implementation
+* **compiler**
+  - `cgo`: improve constant parser
+  - `compiler`: support chained interrupt handlers
+  - `compiler`: add support for running a builtin in a goroutine
+  - `compiler`: do not emit nil checks for loading closure variables
+  - `compiler`: skip context parameter when starting regular goroutine
+  - `compiler`: refactor method names
+  - `compiler`: add function and global section pragmas
+  - `compiler`: implement `syscall.rawSyscallNoError` in inline assembly
+  - `interp`: ignore inline assembly in markExternal
+  - `interp`: fix a bug in pointer cast workaround
+  - `loader`: fix testing a main package
+* **standard library**
+  - `crypto/rand`: replace this package with a TinyGo version
+  - `machine`: make USBCDC global a pointer
+  - `machine`: make UART objects pointer receivers
+  - `machine`: define Serial as the default output
+  - `net`: add initial support for net.IP
+  - `net`: add more net compatibility
+  - `os`: add stub for os.ReadDir
+  - `os`: add FileMode constants from Go 1.16
+  - `os`: add stubs required for net/http
+  - `os`: implement process related functions
+  - `reflect`: implement AppendSlice
+  - `reflect`: add stubs required for net/http
+  - `runtime`: make task.Data a 64-bit integer to avoid overflow
+  - `runtime`: expose memory stats
+  - `sync`: implement NewCond
+  - `syscall`: fix int type in libc version
+* **targets**
+  - `cortexm`: do not disable interrupts on abort
+  - `cortexm`: bump default stack size to 2048 bytes
+  - `nrf`: avoid heap allocation in waitForEvent
+  - `nrf`: don't trigger a heap allocation in SPI.Transfer
+  - `nrf52840`: add support for flashing with the BOSSA tool
+  - `rp2040`: add support for GPIO input
+  - `rp2040`: add basic support for ADC
+  - `rp2040`: gpio and adc pin definitions
+  - `rp2040`: implement UART
+  - `rp2040`: patch elf to checksum 2nd stage boot
+  - `stm32`: add PWM for most chips
+  - `stm32`: add support for pin interrupts
+  - `stm32f103`: add support for PinInputPullup / PinInputPulldown
+  - `wasi`: remove wasm build tag
+* **boards**
+  - `feather-rp2040`: add support for this board
+  - `feather-nrf52840-sense`: add board definition for this board
+  - `pca10059`: support flashing from Windows
+  - `nano-rp2040`: add this board
+  - `nano-33-ble`: add support for this board
+  - `pico`: add the Raspberry Pi Pico board with the new RP2040 chip
+  - `qtpy`: add pin for neopixels
+  - all: add definition for ws2812 for supported boards
+
+
+0.18.0
+---
+
+* **command line**
+  - drop support for Go 1.11 and 1.12
+  - throw an error when no target is specified on Windows
+  - improve error messages in `getDefaultPort()`, support for multiple ports
+  - remove `-cflags` and `-ldflags` flags
+  - implement `-ldflags="-X ..."`
+  - add `-print-allocs` flag that lets you print all heap allocations
+  - openocd commands in tinygo command line
+  - add `-llvm-features` parameter
+  - match `go test` output
+  - discover USB ports only, this will ignore f.ex. bluetooth
+  - use physicmal path instead of cached GOROOT in function getGoroot
+  - add goroot for snap installs
+* **compiler**
+  - `builder`: add support for `-opt=0`
+  - `builder`, `compiler`: compile and cache packages in parallel
+  - `builder`: run interp per package
+  - `builder`: cache C and assembly file outputs
+  - `builder`: add support for `-x` flag to print commands
+  - `builder`: add optsize attribute while building the package
+  - `builder`: run function passes per package
+  - `builder`: hard code Clang compiler
+  - `compiler`: do not use `llvm.GlobalContext()`
+  - `compiler`: remove SimpleDCE pass
+  - `compiler`: do not emit nil checks for `*ssa.Alloc` instructions
+  - `compiler`: merge `runtime.typecodeID` and runtime.typeInInterface
+  - `compiler`: do not check for impossible type asserts
+  - `compiler`: fix use of global context: `llvm.Int32Type()`
+  - `compiler`: add interface IR test
+  - `compiler`: fix lack of method name in interface matching
+  - `compiler`: fix "fragment covers entire variable" bug
+  - `compiler`: optimize string literals and globals
+  - `compiler`: decouple func lowering from interface type codes
+  - `compiler`: add function attributes to some runtime calls
+  - `compiler`: improve position information in error messages
+  - `cgo`: add support for CFLAGS in .c files
+  - `interp`: support GEP on fixed (MMIO) addresses
+  - `interp`: handle `(reflect.Type).Elem()`
+  - `interp`: add support for runtime.interfaceMethod
+  - `interp`: make toLLVMValue return an error instead of panicking
+  - `interp`: add support for switch statement
+  - `interp`: fix phi instruction
+  - `interp`: remove map support
+  - `interp`: support extractvalue/insertvalue with multiple operands
+  - `transform`: optimize string comparisons against ""
+  - `transform`: optimize `reflect.Type` `Implements()` method
+  - `transform`: fix bug in interface lowering when signatures are renamed
+  - `transform`: don't rely on struct name of `runtime.typecodeID`
+  - `transform`: use IPSCCP pass instead of the constant propagation pass
+  - `transform`: fix func lowering assertion failure
+  - `transform`: do not lower zero-sized alloc to alloca
+  - `transform`: split interface and reflect lowering
+* **standard library**
+  - `runtime`: add dummy debug package
+  - `machine`: fix data shift/mask in newUSBSetup
+  - `machine`: make `machine.I2C0` and similar objects pointers
+  - `machine`: unify usbcdc code
+  - `machine`: refactor PWM support
+  - `machine`: avoid heap allocations in USB code
+  - `reflect`: let `reflect.Type` be of interface type
+  - `reflect`: implement a number of stub functions
+  - `reflect`: check for access in the `Interface` method call
+  - `reflect`: fix `AssignableTo` and `Implements` methods
+  - `reflect`: implement `Value.CanAddr`
+  - `reflect`: implement `Sizeof` and `Alignof` for func values
+  - `reflect`: implement `New` function
+  - `runtime`: implement command line arguments in hosted environments
+  - `runtime`: implement environment variables for Linux
+  - `runtime`: improve timers on nrf, and samd chips
+* **targets**
+  - all: use -Qunused-arguments only for assembly files
+  - `atmega1280`: add PWM support
+  - `attiny`: remove dummy UART
+  - `atsamd21`: improve SPI
+  - `atsamd51`: fix PWM support in atsamd51p20
+  - `atsamd5x`: improve SPI
+  - `atsamd51`, `atsame5x`: unify samd51 and same5x
+  - `atsamd51`, `atsamd21`: fix `ADC.Get()` value at 8bit and 10bit
+  - `atsame5x`: add support for CAN
+  - `avr`: remove I2C stubs from attiny support
+  - `cortexm`: check for `arm-none-eabi-gdb` and `gdb-multiarch` commands
+  - `cortexm`: add `__isr_vector` symbol
+  - `cortexm`: disable FPU on Cortex-M4
+  - `cortexm`: clean up Cortex-M target files
+  - `fe310`: fix SPI read
+  - `gameboy-advance`: Fix RGBA color interpretation
+  - `nrf52833`: add PWM support
+  - `stm32l0`: use unified UART logic
+  - `stm32`: move f103 (bluepill) to common i2c code
+  - `stm32`: separate altfunc selection for UART Tx/Rx
+  - `stm32`: i2c implementation for F7, L5 and L4 MCUs
+  - `stm32`: make SPI CLK fast to fix data issue
+  - `stm32`: support SPI on L4 series
+  - `unix`: avoid possible heap allocation with `-opt=0`
+  - `unix`: use conservative GC by default
+  - `unix`: use the tasks scheduler instead of coroutines
+  - `wasi`: upgrade WASI version to wasi_snapshot_preview1
+  - `wasi`: darwin: support basic file io based on libc
+  - `wasm`: only export explicitly exported functions
+  - `wasm`: use WASI ABI for exit function
+  - `wasm`: scan globals conservatively
+* **boards**
+  - `arduino-mega1280`: add support for the Arduino Mega 1280
+  - `arduino-nano-new`: Add Arduino Nano w/ New Bootloader target
+  - `atsame54-xpro`: add initial support this board
+  - `feather-m4-can`: add initial support for this board
+  - `grandcentral-m4`: add board support for Adafruit Grand Central M4 (SAMD51)
+  - `lgt92`: update to new UART structure
+  - `microbit`: remove LED constant
+  - `microbit-v2`: add support for S113 SoftDevice
+  - `nucleol432`: add support for this board
+  - `nucleo-l031k6`: add this board
+  - `pca10059`: initial support for this board
+  - `qtpy`: fix msd-volume-name
+  - `qtpy`: fix i2c setting
+  - `teensy40`: move txBuffer allocation to UART declaration
+  - `teensy40`: add UART0 as alias for UART1
+
+
 0.17.0
 
 ---

Dockerfile

@@ -1,5 +1,5 @@
-# TinyGo base stage installs the most recent Go 1.15.x, LLVM 11 and the TinyGo compiler itself.
-FROM golang:1.15 AS tinygo-base
+# TinyGo base stage installs the most recent Go 1.16.x, LLVM 11 and the TinyGo compiler itself.
+FROM golang:1.16 AS tinygo-base
 
 RUN wget -O- https://apt.llvm.org/llvm-snapshot.gpg.key| apt-key add - && \
     echo "deb http://apt.llvm.org/buster/ llvm-toolchain-buster-11 main" >> /etc/apt/sources.list && \

Makefile

@@ -36,7 +36,7 @@ else
     LLVM_OPTION += '-DLLVM_ENABLE_ASSERTIONS=OFF'
 endif
 
-.PHONY: all tinygo test $(LLVM_BUILDDIR) llvm-source clean fmt gen-device gen-device-nrf gen-device-nxp gen-device-avr
+.PHONY: all tinygo test $(LLVM_BUILDDIR) llvm-source clean fmt gen-device gen-device-nrf gen-device-nxp gen-device-avr gen-device-rp
 
 LLVM_COMPONENTS = all-targets analysis asmparser asmprinter bitreader bitwriter codegen core coroutines coverage debuginfodwarf executionengine frontendopenmp instrumentation interpreter ipo irreader linker lto mc mcjit objcarcopts option profiledata scalaropts support target
 
@@ -71,7 +71,7 @@ else
 endif
 
 # Libraries that should be linked in for the statically linked Clang.
-CLANG_LIB_NAMES = clangAnalysis clangARCMigrate clangAST clangASTMatchers clangBasic clangCodeGen clangCrossTU clangDriver clangDynamicASTMatchers clangEdit clangFormat clangFrontend clangFrontendTool clangHandleCXX clangHandleLLVM clangIndex clangLex clangParse clangRewrite clangRewriteFrontend clangSema clangSerialization clangStaticAnalyzerCheckers clangStaticAnalyzerCore clangStaticAnalyzerFrontend clangTooling clangToolingASTDiff clangToolingCore clangToolingInclusions
+CLANG_LIB_NAMES = clangAnalysis clangAST clangASTMatchers clangBasic clangCodeGen clangCrossTU clangDriver clangDynamicASTMatchers clangEdit clangFormat clangFrontend clangFrontendTool clangHandleCXX clangHandleLLVM clangIndex clangLex clangParse clangRewrite clangRewriteFrontend clangSema clangSerialization clangTooling clangToolingASTDiff clangToolingCore clangToolingInclusions
 CLANG_LIBS = $(START_GROUP) $(addprefix -l,$(CLANG_LIB_NAMES)) $(END_GROUP) -lstdc++
 
 # Libraries that should be linked in for the statically linked LLD.
@@ -107,7 +107,7 @@ fmt-check:
 	@unformatted=$$(gofmt -l $(FMT_PATHS)); [ -z "$$unformatted" ] && exit 0; echo "Unformatted:"; for fn in $$unformatted; do echo "  $$fn"; done; exit 1
 
 
-gen-device: gen-device-avr gen-device-esp gen-device-nrf gen-device-sam gen-device-sifive gen-device-kendryte gen-device-nxp
+gen-device: gen-device-avr gen-device-esp gen-device-nrf gen-device-sam gen-device-sifive gen-device-kendryte gen-device-nxp gen-device-rp
 ifneq ($(STM32), 0)
 gen-device: gen-device-stm32
 endif
@@ -150,16 +150,19 @@ gen-device-stm32: build/gen-device-svd
 	./build/gen-device-svd -source=https://github.com/tinygo-org/stm32-svd lib/stm32-svd/svd src/device/stm32/
 	GO111MODULE=off $(GO) fmt ./src/device/stm32
 
+gen-device-rp: build/gen-device-svd
+	./build/gen-device-svd -source=https://github.com/posborne/cmsis-svd/tree/master/data/RaspberryPi lib/cmsis-svd/data/RaspberryPi/ src/device/rp/
+	GO111MODULE=off $(GO) fmt ./src/device/rp
 
 # Get LLVM sources.
-$(LLVM_PROJECTDIR)/README.md:
+$(LLVM_PROJECTDIR)/llvm:
 	git clone -b xtensa_release_11.0.0 --depth=1 https://github.com/tinygo-org/llvm-project $(LLVM_PROJECTDIR)
-llvm-source: $(LLVM_PROJECTDIR)/README.md
+llvm-source: $(LLVM_PROJECTDIR)/llvm
 
 # Configure LLVM.
 TINYGO_SOURCE_DIR=$(shell pwd)
 $(LLVM_BUILDDIR)/build.ninja: llvm-source
-	mkdir -p $(LLVM_BUILDDIR); cd $(LLVM_BUILDDIR); cmake -G Ninja $(TINYGO_SOURCE_DIR)/$(LLVM_PROJECTDIR)/llvm "-DLLVM_TARGETS_TO_BUILD=X86;ARM;AArch64;RISCV;WebAssembly" "-DLLVM_EXPERIMENTAL_TARGETS_TO_BUILD=AVR;Xtensa" -DCMAKE_BUILD_TYPE=Release -DLIBCLANG_BUILD_STATIC=ON -DLLVM_ENABLE_TERMINFO=OFF -DLLVM_ENABLE_ZLIB=OFF -DLLVM_ENABLE_LIBEDIT=OFF -DLLVM_ENABLE_Z3_SOLVER=OFF -DLLVM_ENABLE_OCAMLDOC=OFF -DLLVM_ENABLE_LIBXML2=OFF -DLLVM_ENABLE_PROJECTS="clang;lld" -DLLVM_TOOL_CLANG_TOOLS_EXTRA_BUILD=OFF $(LLVM_OPTION)
+	mkdir -p $(LLVM_BUILDDIR); cd $(LLVM_BUILDDIR); cmake -G Ninja $(TINYGO_SOURCE_DIR)/$(LLVM_PROJECTDIR)/llvm "-DLLVM_TARGETS_TO_BUILD=X86;ARM;AArch64;RISCV;WebAssembly" "-DLLVM_EXPERIMENTAL_TARGETS_TO_BUILD=AVR;Xtensa" -DCMAKE_BUILD_TYPE=Release -DLIBCLANG_BUILD_STATIC=ON -DLLVM_ENABLE_TERMINFO=OFF -DLLVM_ENABLE_ZLIB=OFF -DLLVM_ENABLE_LIBEDIT=OFF -DLLVM_ENABLE_Z3_SOLVER=OFF -DLLVM_ENABLE_OCAMLDOC=OFF -DLLVM_ENABLE_LIBXML2=OFF -DLLVM_ENABLE_PROJECTS="clang;lld" -DLLVM_TOOL_CLANG_TOOLS_EXTRA_BUILD=OFF -DCLANG_ENABLE_STATIC_ANALYZER=OFF -DCLANG_ENABLE_ARCMT=OFF $(LLVM_OPTION)
 
 # Build LLVM.
 $(LLVM_BUILDDIR): $(LLVM_BUILDDIR)/build.ninja
@@ -182,25 +185,28 @@ tinygo:
 test: wasi-libc
 	CGO_CPPFLAGS="$(CGO_CPPFLAGS)" CGO_CXXFLAGS="$(CGO_CXXFLAGS)" CGO_LDFLAGS="$(CGO_LDFLAGS)" $(GO) test -v -buildmode exe -tags byollvm ./builder ./cgo ./compileopts ./compiler ./interp ./transform .
 
+TEST_PACKAGES = \
+	container/heap \
+	container/list \
+	container/ring \
+	crypto/des \
+	encoding \
+	encoding/ascii85 \
+	encoding/base32 \
+	encoding/hex \
+	hash/adler32 \
+	hash/fnv \
+	hash/crc64 \
+	math \
+	math/cmplx \
+	text/scanner \
+	unicode/utf8 \
+
 # Test known-working standard library packages.
-# TODO: do this in one command, parallelize, and only show failing tests (no
-# implied -v flag).
+# TODO: parallelize, and only show failing tests (no implied -v flag).
 .PHONY: tinygo-test
 tinygo-test:
-	$(TINYGO) test container/heap
-	$(TINYGO) test container/list
-	$(TINYGO) test container/ring
-	$(TINYGO) test crypto/des
-	$(TINYGO) test encoding/ascii85
-	$(TINYGO) test encoding/base32
-	$(TINYGO) test encoding/hex
-	$(TINYGO) test hash/adler32
-	$(TINYGO) test hash/fnv
-	$(TINYGO) test hash/crc64
-	$(TINYGO) test math
-	$(TINYGO) test math/cmplx
-	$(TINYGO) test text/scanner
-	$(TINYGO) test unicode/utf8
+	$(TINYGO) test $(TEST_PACKAGES)
 
 .PHONY: smoketest
 smoketest:
@@ -224,9 +230,9 @@ smoketest:
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=pca10040            examples/mcp3008
 	@$(MD5SUM) test.hex
-	$(TINYGO) build -size short -o test.hex -target=microbit            examples/microbit-blink
+	$(TINYGO) build -size short -o test.hex -target=pca10040            examples/memstats
 	@$(MD5SUM) test.hex
-	$(TINYGO) build -size short -o test.hex -target=microbit-v2         examples/microbit-blink
+	$(TINYGO) build -size short -o test.hex -target=microbit            examples/microbit-blink
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=pca10040            examples/pininterrupt
 	@$(MD5SUM) test.hex
@@ -256,6 +262,10 @@ smoketest:
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=microbit-s110v8     examples/echo
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=microbit-v2         examples/microbit-blink
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=microbit-v2-s113v7  examples/microbit-blink
+	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=nrf52840-mdk        examples/blinky1
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=pca10031            examples/blinky1
@@ -288,6 +298,8 @@ smoketest:
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.gba -target=gameboy-advance     examples/gba-display
 	@$(MD5SUM) test.gba
+	$(TINYGO) build -size short -o test.hex -target=grandcentral-m4     examples/blinky1
+	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=itsybitsy-m4        examples/blinky1
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=feather-m4          examples/blinky1
@@ -324,9 +336,11 @@ smoketest:
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=feather-nrf52840  	examples/blinky1
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=feather-nrf52840-sense examples/blinky1
+	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=itsybitsy-nrf52840  examples/blinky1
 	@$(MD5SUM) test.hex
-	$(TINYGO) build -size short -o test.hex -target=qtpy  				examples/blinky1
+	$(TINYGO) build -size short -o test.hex -target=qtpy                examples/serial
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=teensy40            examples/blinky1
 	@$(MD5SUM) test.hex
@@ -334,6 +348,24 @@ smoketest:
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=p1am-100            examples/blinky1
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=atsame54-xpro       examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=atsame54-xpro       examples/can
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=feather-m4-can      examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=feather-m4-can      examples/caninterrupt
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=arduino-nano33      examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=pico                examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=nano-33-ble         examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=nano-rp2040         examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=feather-rp2040 		examples/blinky1
+	@$(MD5SUM) test.hex
 	# test pwm
 	$(TINYGO) build -size short -o test.hex -target=itsybitsy-m0        examples/pwm
 	@$(MD5SUM) test.hex
@@ -341,8 +373,6 @@ smoketest:
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=feather-m4          examples/pwm
 	@$(MD5SUM) test.hex
-	$(TINYGO) build -size short -o test.hex -target=pyportal            examples/pwm
-	@$(MD5SUM) test.hex
 ifneq ($(STM32), 0)
 	$(TINYGO) build -size short -o test.hex -target=bluepill            examples/blinky1
 	@$(MD5SUM) test.hex
@@ -354,6 +384,8 @@ ifneq ($(STM32), 0)
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=nucleo-f722ze       examples/blinky1
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=nucleo-l031k6       examples/blinky1
+	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=nucleo-l432kc       examples/blinky1
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=nucleo-l552ze       examples/blinky1
@@ -364,6 +396,8 @@ ifneq ($(STM32), 0)
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=stm32f4disco-1      examples/blinky1
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=stm32f4disco-1      examples/pwm
+	@$(MD5SUM) test.hex
 endif
 ifneq ($(AVR), 0)
 	$(TINYGO) build -size short -o test.hex -target=atmega1284p         examples/serial
@@ -374,6 +408,10 @@ ifneq ($(AVR), 0)
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=arduino -scheduler=tasks  examples/blinky1
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=arduino-mega1280    examples/blinky1
+	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=arduino-mega1280    examples/pwm
+	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=arduino-nano        examples/blinky1
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=digispark           examples/blinky1
@@ -400,6 +438,8 @@ endif
 	@$(MD5SUM) test.hex
 	$(TINYGO) build -size short -o test.hex -target=pca10040 -opt=1     examples/blinky1
 	@$(MD5SUM) test.hex
+	$(TINYGO) build -size short -o test.hex -target=pca10040 -serial=none examples/echo
+	@$(MD5SUM) test.hex
 	$(TINYGO) build             -o test.nro -target=nintendoswitch      examples/serial
 	@$(MD5SUM) test.nro
 	$(TINYGO) build -size short -o test.hex -target=pca10040 -opt=0     ./testdata/stdlib.go

README.md

@@ -43,15 +43,19 @@ See the [getting started instructions](https://tinygo.org/getting-started/) for
 
 You can compile TinyGo programs for microcontrollers, WebAssembly and Linux.
 
-The following 55 microcontroller boards are currently supported:
+The following 68 microcontroller boards are currently supported:
 
 * [Adafruit Circuit Playground Bluefruit](https://www.adafruit.com/product/4333)
 * [Adafruit Circuit Playground Express](https://www.adafruit.com/product/3333)
 * [Adafruit CLUE](https://www.adafruit.com/product/4500)
 * [Adafruit Feather M0](https://www.adafruit.com/product/2772)
 * [Adafruit Feather M4](https://www.adafruit.com/product/3857)
+* [Adafruit Feather M4 CAN](https://www.adafruit.com/product/4759)
 * [Adafruit Feather nRF52840 Express](https://www.adafruit.com/product/4062)
+* [Adafruit Feather nRF52840 Sense](https://www.adafruit.com/product/4516)
+* [Adafruit Feather RP2040](https://www.adafruit.com/product/4884)
 * [Adafruit Feather STM32F405 Express](https://www.adafruit.com/product/4382)
+* [Adafruit Grand Central M4](https://www.adafruit.com/product/4064)
 * [Adafruit ItsyBitsy M0](https://www.adafruit.com/product/3727)
 * [Adafruit ItsyBitsy M4](https://www.adafruit.com/product/3800)
 * [Adafruit ItsyBitsy nRF52840](https://www.adafruit.com/product/4481)
@@ -62,10 +66,14 @@ The following 55 microcontroller boards are currently supported:
 * [Adafruit PyPortal](https://www.adafruit.com/product/4116)
 * [Adafruit QT Py](https://www.adafruit.com/product/4600)
 * [Adafruit Trinket M0](https://www.adafruit.com/product/3500)
+* [Arduino Mega 1280](https://www.arduino.cc/en/Main/arduinoBoardMega/)
 * [Arduino Mega 2560](https://store.arduino.cc/arduino-mega-2560-rev3)
 * [Arduino MKR1000](https://store.arduino.cc/arduino-mkr1000-wifi)
 * [Arduino Nano](https://store.arduino.cc/arduino-nano)
-* [Arduino Nano33 IoT](https://store.arduino.cc/nano-33-iot)
+* [Arduino Nano 33 BLE](https://store.arduino.cc/nano-33-ble)
+* [Arduino Nano 33 BLE Sense](https://store.arduino.cc/nano-33-ble-sense)
+* [Arduino Nano 33 IoT](https://store.arduino.cc/nano-33-iot)
+* [Arduino Nano RP2040 Connect](https://store.arduino.cc/nano-rp2040-connect)
 * [Arduino Uno](https://store.arduino.cc/arduino-uno-rev3)
 * [Arduino Zero](https://store.arduino.cc/usa/arduino-zero)
 * [BBC micro:bit](https://microbit.org/)
@@ -77,11 +85,13 @@ The following 55 microcontroller boards are currently supported:
 * [Game Boy Advance](https://en.wikipedia.org/wiki/Game_Boy_Advance)
 * [Makerdiary nRF52840-MDK](https://wiki.makerdiary.com/nrf52840-mdk/)
 * [Makerdiary nRF52840-MDK USB Dongle](https://wiki.makerdiary.com/nrf52840-mdk-usb-dongle/)
+* [Microchip SAM E54 Xplained Pro](https://www.microchip.com/developmenttools/productdetails/atsame54-xpro)
 * [nice!nano](https://docs.nicekeyboards.com/#/nice!nano/)
 * [Nintendo Switch](https://www.nintendo.com/switch/)
 * [Nordic Semiconductor PCA10031](https://www.nordicsemi.com/eng/Products/nRF51-Dongle)
 * [Nordic Semiconductor PCA10040](https://www.nordicsemi.com/eng/Products/Bluetooth-low-energy/nRF52-DK)
 * [Nordic Semiconductor PCA10056](https://www.nordicsemi.com/Software-and-Tools/Development-Kits/nRF52840-DK)
+* [Nordic Semiconductor pca10059](https://www.nordicsemi.com/Software-and-tools/Development-Kits/nRF52840-Dongle)
 * [Particle Argon](https://docs.particle.io/datasheets/wi-fi/argon-datasheet/)
 * [Particle Boron](https://docs.particle.io/datasheets/cellular/boron-datasheet/)
 * [Particle Xenon](https://docs.particle.io/datasheets/discontinued/xenon-datasheet/)
@@ -90,12 +100,15 @@ The following 55 microcontroller boards are currently supported:
 * [PJRC Teensy 3.6](https://www.pjrc.com/store/teensy36.html)
 * [PJRC Teensy 4.0](https://www.pjrc.com/store/teensy40.html)
 * [ProductivityOpen P1AM-100](https://facts-engineering.github.io/modules/P1AM-100/P1AM-100.html)
+* [Raspberry Pi Pico](https://www.raspberrypi.org/products/raspberry-pi-pico/)
 * [Seeed Wio Terminal](https://www.seeedstudio.com/Wio-Terminal-p-4509.html)
 * [Seeed Seeeduino XIAO](https://www.seeedstudio.com/Seeeduino-XIAO-Arduino-Microcontroller-SAMD21-Cortex-M0+-p-4426.html)
 * [Seeed Sipeed MAix BiT](https://www.seeedstudio.com/Sipeed-MAix-BiT-for-RISC-V-AI-IoT-p-2872.html)
 * [SiFIve HiFive1](https://www.sifive.com/boards/hifive1)
 * [ST Micro "Nucleo" F103RB](https://www.st.com/en/evaluation-tools/nucleo-f103rb.html)
 * [ST Micro "Nucleo" F722ZE](https://www.st.com/en/evaluation-tools/nucleo-f722ze.html)
+* [ST Micro "Nucleo" L031K6](https://www.st.com/ja/evaluation-tools/nucleo-l031k6.html)
+* [ST Micro "Nucleo" L432KC](https://www.st.com/ja/evaluation-tools/nucleo-l432kc.html)
 * [ST Micro "Nucleo" L552ZE](https://www.st.com/en/evaluation-tools/nucleo-l552ze-q.html)
 * [ST Micro STM32F103XX "Bluepill"](https://stm32-base.org/boards/STM32F103C8T6-Blue-Pill)
 * [ST Micro STM32F407 "Discovery"](https://www.st.com/en/evaluation-tools/stm32f4discovery.html)

azure-pipelines.yml

@@ -28,7 +28,7 @@ jobs:
     - task: CacheBeta@0
       displayName: Cache LLVM build
       inputs:
-        key: llvm-build-11-windows-v4
+        key: llvm-build-11-windows-v5
         path: llvm-build
     - task: Bash@3
       displayName: Build LLVM

builder/build.go

@@ -12,7 +12,9 @@ import (
 	"errors"
 	"fmt"
 	"go/types"
+	"hash/crc32"
 	"io/ioutil"
+	"math/bits"
 	"os"
 	"path/filepath"
 	"runtime"
@@ -39,6 +41,11 @@ type BuildResult struct {
 	// The directory of the main package. This is useful for testing as the test
 	// binary must be run in the directory of the tested package.
 	MainDir string
+
+	// ImportPath is the import path of the main package. This is useful for
+	// correctly printing test results: the import path isn't always the same as
+	// the path listed on the command line.
+	ImportPath string
 }
 
 // packageAction is the struct that is serialized to JSON and hashed, to work as
@@ -52,14 +59,17 @@ type BuildResult struct {
 // key, avoiding the need for recompiling all dependencies when only the
 // implementation of an imported package changes.
 type packageAction struct {
-	ImportPath      string
-	CompilerVersion int // compiler.Version
-	InterpVersion   int // interp.Version
-	LLVMVersion     string
-	Config          *compiler.Config
-	CFlags          []string
-	FileHashes      map[string]string // hash of every file that's part of the package
-	Imports         map[string]string // map from imported package to action ID hash
+	ImportPath       string
+	CompilerVersion  int // compiler.Version
+	InterpVersion    int // interp.Version
+	LLVMVersion      string
+	Config           *compiler.Config
+	CFlags           []string
+	FileHashes       map[string]string // hash of every file that's part of the package
+	Imports          map[string]string // map from imported package to action ID hash
+	OptLevel         int               // LLVM optimization level (0-3)
+	SizeLevel        int               // LLVM optimization for size level (0-2)
+	UndefinedGlobals []string          // globals that are left as external globals (no initializer)
 }
 
 // Build performs a single package to executable Go build. It takes in a package
@@ -91,6 +101,7 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 		DefaultStackSize:   config.Target.DefaultStackSize,
 		NeedsStackObjects:  config.NeedsStackObjects(),
 		Debug:              config.Debug(),
+		LLVMFeatures:       config.LLVMFeatures(),
 	}
 
 	// Load the target machine, which is the LLVM object that contains all
@@ -127,20 +138,30 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 	var packageJobs []*compileJob
 	packageBitcodePaths := make(map[string]string)
 	packageActionIDs := make(map[string]string)
+	optLevel, sizeLevel, _ := config.OptLevels()
 	for _, pkg := range lprogram.Sorted() {
 		pkg := pkg // necessary to avoid a race condition
 
+		var undefinedGlobals []string
+		for name := range config.Options.GlobalValues[pkg.Pkg.Path()] {
+			undefinedGlobals = append(undefinedGlobals, name)
+		}
+		sort.Strings(undefinedGlobals)
+
 		// Create a cache key: a hash from the action ID below that contains all
 		// the parameters for the build.
 		actionID := packageAction{
-			ImportPath:      pkg.ImportPath,
-			CompilerVersion: compiler.Version,
-			InterpVersion:   interp.Version,
-			LLVMVersion:     llvm.Version,
-			Config:          compilerConfig,
-			CFlags:          pkg.CFlags,
-			FileHashes:      make(map[string]string, len(pkg.FileHashes)),
-			Imports:         make(map[string]string, len(pkg.Pkg.Imports())),
+			ImportPath:       pkg.ImportPath,
+			CompilerVersion:  compiler.Version,
+			InterpVersion:    interp.Version,
+			LLVMVersion:      llvm.Version,
+			Config:           compilerConfig,
+			CFlags:           pkg.CFlags,
+			FileHashes:       make(map[string]string, len(pkg.FileHashes)),
+			Imports:          make(map[string]string, len(pkg.Pkg.Imports())),
+			OptLevel:         optLevel,
+			SizeLevel:        sizeLevel,
+			UndefinedGlobals: undefinedGlobals,
 		}
 		for filePath, hash := range pkg.FileHashes {
 			actionID.FileHashes[filePath] = hex.EncodeToString(hash)
@@ -191,6 +212,25 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 					return errors.New("verification error after compiling package " + pkg.ImportPath)
 				}
 
+				// Erase all globals that are part of the undefinedGlobals list.
+				// This list comes from the -ldflags="-X pkg.foo=val" option.
+				// Instead of setting the value directly in the AST (which would
+				// mean the value, which may be a secret, is stored in the build
+				// cache), the global itself is left external (undefined) and is
+				// only set at the end of the compilation.
+				for _, name := range undefinedGlobals {
+					globalName := pkg.Pkg.Path() + "." + name
+					global := mod.NamedGlobal(globalName)
+					if global.IsNil() {
+						return errors.New("global not found: " + globalName)
+					}
+					name := global.Name()
+					newGlobal := llvm.AddGlobal(mod, global.Type().ElementType(), name+".tmp")
+					global.ReplaceAllUsesWith(newGlobal)
+					global.EraseFromParentAsGlobal()
+					newGlobal.SetName(name)
+				}
+
 				// Try to interpret package initializers at compile time.
 				// It may only be possible to do this partially, in which case
 				// it is completed after all IR files are linked.
@@ -206,6 +246,38 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 					return errors.New("verification error after interpreting " + pkgInit.Name())
 				}
 
+				if sizeLevel >= 2 {
+					// Set the "optsize" attribute to make slightly smaller
+					// binaries at the cost of some performance.
+					kind := llvm.AttributeKindID("optsize")
+					attr := mod.Context().CreateEnumAttribute(kind, 0)
+					for fn := mod.FirstFunction(); !fn.IsNil(); fn = llvm.NextFunction(fn) {
+						fn.AddFunctionAttr(attr)
+					}
+				}
+
+				// Run function passes for each function in the module.
+				// These passes are intended to be run on each function right
+				// after they're created to reduce IR size (and maybe also for
+				// cache locality to improve performance), but for now they're
+				// run here for each function in turn. Maybe this can be
+				// improved in the future.
+				builder := llvm.NewPassManagerBuilder()
+				defer builder.Dispose()
+				builder.SetOptLevel(optLevel)
+				builder.SetSizeLevel(sizeLevel)
+				funcPasses := llvm.NewFunctionPassManagerForModule(mod)
+				defer funcPasses.Dispose()
+				builder.PopulateFunc(funcPasses)
+				funcPasses.InitializeFunc()
+				for fn := mod.FirstFunction(); !fn.IsNil(); fn = llvm.NextFunction(fn) {
+					if fn.IsDeclaration() {
+						continue
+					}
+					funcPasses.RunFunc(fn)
+				}
+				funcPasses.FinalizeFunc()
+
 				// Serialize the LLVM module as a bitcode file.
 				// Write to a temporary path that is renamed to the destination
 				// file to avoid race conditions with other TinyGo invocatiosn
@@ -430,7 +502,7 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 		job := &compileJob{
 			description: "compile extra file " + path,
 			run: func(job *compileJob) error {
-				result, err := compileAndCacheCFile(abspath, dir, config.CFlags(), config)
+				result, err := compileAndCacheCFile(abspath, dir, config.CFlags(), config.Options.PrintCommands)
 				job.result = result
 				return err
 			},
@@ -449,7 +521,7 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 			job := &compileJob{
 				description: "compile CGo file " + abspath,
 				run: func(job *compileJob) error {
-					result, err := compileAndCacheCFile(abspath, dir, pkg.CFlags, config)
+					result, err := compileAndCacheCFile(abspath, dir, pkg.CFlags, config.Options.PrintCommands)
 					job.result = result
 					return err
 				},
@@ -477,8 +549,8 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 				}
 				ldflags = append(ldflags, dependency.result)
 			}
-			if config.Options.PrintCommands {
-				fmt.Printf("%s %s\n", config.Target.Linker, strings.Join(ldflags, " "))
+			if config.Options.PrintCommands != nil {
+				config.Options.PrintCommands(config.Target.Linker, ldflags...)
 			}
 			err = link(config.Target.Linker, ldflags...)
 			if err != nil {
@@ -496,6 +568,8 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 					return err
 				}
 			}
+
+			// Apply ELF patches
 			if config.AutomaticStackSize() {
 				// Modify the .tinygo_stacksizes section that contains a stack size
 				// for each goroutine.
@@ -504,6 +578,13 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 					return fmt.Errorf("could not modify stack sizes: %w", err)
 				}
 			}
+			if config.RP2040BootPatch() {
+				// Patch the second stage bootloader CRC into the .boot2 section
+				err = patchRP2040BootCRC(executable)
+				if err != nil {
+					return fmt.Errorf("could not patch RP2040 second stage boot loader: %w", err)
+				}
+			}
 
 			if config.Options.PrintSizes == "short" || config.Options.PrintSizes == "full" {
 				sizes, err := loadProgramSize(executable)
@@ -569,12 +650,25 @@ func Build(pkgName, outpath string, config *compileopts.Config, action func(Buil
 		if err != nil {
 			return err
 		}
+	case "nrf-dfu":
+		// special format for nrfutil for Nordic chips
+		tmphexpath := filepath.Join(dir, "main.hex")
+		err := objcopy(executable, tmphexpath, "hex")
+		if err != nil {
+			return err
+		}
+		tmppath = filepath.Join(dir, "main"+outext)
+		err = makeDFUFirmwareImage(config.Options, tmphexpath, tmppath)
+		if err != nil {
+			return err
+		}
 	default:
 		return fmt.Errorf("unknown output binary format: %s", outputBinaryFormat)
 	}
 	return action(BuildResult{
-		Binary:  tmppath,
-		MainDir: lprogram.MainPkg().Dir,
+		Binary:     tmppath,
+		MainDir:    lprogram.MainPkg().Dir,
+		ImportPath: lprogram.MainPkg().ImportPath,
 	})
 }
 
@@ -603,6 +697,12 @@ func optimizeProgram(mod llvm.Module, config *compileopts.Config) error {
 		transform.ApplyFunctionSections(mod) // -ffunction-sections
 	}
 
+	// Insert values from -ldflags="-X ..." into the IR.
+	err = setGlobalValues(mod, config.Options.GlobalValues)
+	if err != nil {
+		return err
+	}
+
 	// Browsers cannot handle external functions that have type i64 because it
 	// cannot be represented exactly in JavaScript (JS only has doubles). To
 	// keep functions interoperable, pass int64 types as pointers to
@@ -617,21 +717,8 @@ func optimizeProgram(mod llvm.Module, config *compileopts.Config) error {
 
 	// Optimization levels here are roughly the same as Clang, but probably not
 	// exactly.
-	var errs []error
-	switch config.Options.Opt {
-	case "none", "0":
-		errs = transform.Optimize(mod, config, 0, 0, 0) // -O0
-	case "1":
-		errs = transform.Optimize(mod, config, 1, 0, 0) // -O1
-	case "2":
-		errs = transform.Optimize(mod, config, 2, 0, 225) // -O2
-	case "s":
-		errs = transform.Optimize(mod, config, 2, 1, 225) // -Os
-	case "z":
-		errs = transform.Optimize(mod, config, 2, 2, 5) // -Oz, default
-	default:
-		return errors.New("unknown optimization level: -opt=" + config.Options.Opt)
-	}
+	optLevel, sizeLevel, inlinerThreshold := config.OptLevels()
+	errs := transform.Optimize(mod, config, optLevel, sizeLevel, inlinerThreshold)
 	if len(errs) > 0 {
 		return newMultiError(errs)
 	}
@@ -653,6 +740,71 @@ func optimizeProgram(mod llvm.Module, config *compileopts.Config) error {
 	return nil
 }
 
+// setGlobalValues sets the global values from the -ldflags="-X ..." compiler
+// option in the given module. An error may be returned if the global is not of
+// the expected type.
+func setGlobalValues(mod llvm.Module, globals map[string]map[string]string) error {
+	var pkgPaths []string
+	for pkgPath := range globals {
+		pkgPaths = append(pkgPaths, pkgPath)
+	}
+	sort.Strings(pkgPaths)
+	for _, pkgPath := range pkgPaths {
+		pkg := globals[pkgPath]
+		var names []string
+		for name := range pkg {
+			names = append(names, name)
+		}
+		sort.Strings(names)
+		for _, name := range names {
+			value := pkg[name]
+			globalName := pkgPath + "." + name
+			global := mod.NamedGlobal(globalName)
+			if global.IsNil() || !global.Initializer().IsNil() {
+				// The global either does not exist (optimized away?) or has
+				// some value, in which case it has already been initialized at
+				// package init time.
+				continue
+			}
+
+			// A strin is a {ptr, len} pair. We need these types to build the
+			// initializer.
+			initializerType := global.Type().ElementType()
+			if initializerType.TypeKind() != llvm.StructTypeKind || initializerType.StructName() == "" {
+				return fmt.Errorf("%s: not a string", globalName)
+			}
+			elementTypes := initializerType.StructElementTypes()
+			if len(elementTypes) != 2 {
+				return fmt.Errorf("%s: not a string", globalName)
+			}
+
+			// Create a buffer for the string contents.
+			bufInitializer := mod.Context().ConstString(value, false)
+			buf := llvm.AddGlobal(mod, bufInitializer.Type(), ".string")
+			buf.SetInitializer(bufInitializer)
+			buf.SetAlignment(1)
+			buf.SetUnnamedAddr(true)
+			buf.SetLinkage(llvm.PrivateLinkage)
+
+			// Create the string value, which is a {ptr, len} pair.
+			zero := llvm.ConstInt(mod.Context().Int32Type(), 0, false)
+			ptr := llvm.ConstGEP(buf, []llvm.Value{zero, zero})
+			if ptr.Type() != elementTypes[0] {
+				return fmt.Errorf("%s: not a string", globalName)
+			}
+			length := llvm.ConstInt(elementTypes[1], uint64(len(value)), false)
+			initializer := llvm.ConstNamedStruct(initializerType, []llvm.Value{
+				ptr,
+				length,
+			})
+
+			// Set the initializer. No initializer should be set at this point.
+			global.SetInitializer(initializer)
+		}
+	}
+	return nil
+}
+
 // functionStackSizes keeps stack size information about a single function
 // (usually a goroutine).
 type functionStackSize struct {
@@ -779,30 +931,7 @@ func determineStackSizes(mod llvm.Module, executable string) ([]string, map[stri
 // stack size information. Before this modification, all stack sizes in the
 // section assume the default stack size (which is relatively big).
 func modifyStackSizes(executable string, stackSizeLoads []string, stackSizes map[string]functionStackSize) error {
-	fp, err := os.OpenFile(executable, os.O_RDWR, 0)
-	if err != nil {
-		return err
-	}
-	defer fp.Close()
-
-	elfFile, err := elf.NewFile(fp)
-	if err != nil {
-		return err
-	}
-
-	section := elfFile.Section(".tinygo_stacksizes")
-	if section == nil {
-		return errors.New("could not find .tinygo_stacksizes section")
-	}
-
-	if section.Size != section.FileSize {
-		// Sanity check.
-		return fmt.Errorf("expected .tinygo_stacksizes to have identical size and file size, got %d and %d", section.Size, section.FileSize)
-	}
-
-	// Read all goroutine stack sizes.
-	data := make([]byte, section.Size)
-	_, err = fp.ReadAt(data, int64(section.Offset))
+	data, fileHeader, err := getElfSectionData(executable, ".tinygo_stacksizes")
 	if err != nil {
 		return err
 	}
@@ -831,7 +960,7 @@ func modifyStackSizes(executable string, stackSizeLoads []string, stackSizes map
 			stackSize += 4
 
 			// Add stack size used by interrupts.
-			switch elfFile.Machine {
+			switch fileHeader.Machine {
 			case elf.EM_ARM:
 				// On Cortex-M (assumed here), this stack size is 8 words or 32
 				// bytes. This is only to store the registers that the interrupt
@@ -847,13 +976,7 @@ func modifyStackSizes(executable string, stackSizeLoads []string, stackSizes map
 		}
 	}
 
-	// Write back the modified stack sizes.
-	_, err = fp.WriteAt(data, int64(section.Offset))
-	if err != nil {
-		return err
-	}
-
-	return nil
+	return replaceElfSection(executable, ".tinygo_stacksizes", data)
 }
 
 // printStacks prints the maximum stack depth for functions that are started as
@@ -885,3 +1008,41 @@ func printStacks(calculatedStacks []string, stackSizes map[string]functionStackS
 		}
 	}
 }
+
+// RP2040 second stage bootloader CRC32 calculation
+//
+// Spec: https://datasheets.raspberrypi.org/rp2040/rp2040-datasheet.pdf
+// Section: 2.8.1.3.1. Checksum
+func patchRP2040BootCRC(executable string) error {
+	bytes, _, err := getElfSectionData(executable, ".boot2")
+	if err != nil {
+		return err
+	}
+
+	if len(bytes) != 256 {
+		return fmt.Errorf("rp2040 .boot2 section must be exactly 256 bytes")
+	}
+
+	// From the 'official' RP2040 checksum script:
+	//
+	//  Our bootrom CRC32 is slightly bass-ackward but it's
+	//  best to work around for now (FIXME)
+	//  100% worth it to save two Thumb instructions
+	revBytes := make([]byte, len(bytes))
+	for i := range bytes {
+		revBytes[i] = bits.Reverse8(bytes[i])
+	}
+
+	// crc32.Update does an initial negate and negates the
+	// result, so to meet RP2040 spec, pass 0x0 as initial
+	// hash and negate returned value.
+	//
+	// Note: checksum is over 252 bytes (256 - 4)
+	hash := bits.Reverse32(crc32.Update(0x0, crc32.IEEETable, revBytes[:252]) ^ 0xFFFFFFFF)
+
+	// Write the CRC to the end of the bootloader.
+	binary.LittleEndian.PutUint32(bytes[252:], hash)
+
+	// Update the .boot2 section to included the CRC
+	return replaceElfSection(executable, ".boot2", bytes)
+}

builder/cc.go

@@ -17,7 +17,6 @@ import (
 	"strings"
 	"unicode"
 
-	"github.com/tinygo-org/tinygo/compileopts"
 	"github.com/tinygo-org/tinygo/goenv"
 	"tinygo.org/x/go-llvm"
 )
@@ -57,7 +56,7 @@ import (
 //   depfile but without invalidating its name. For this reason, the depfile is
 //   written on each new compilation (even when it seems unnecessary). However, it
 //   could in rare cases lead to a stale file fetched from the cache.
-func compileAndCacheCFile(abspath, tmpdir string, cflags []string, config *compileopts.Config) (string, error) {
+func compileAndCacheCFile(abspath, tmpdir string, cflags []string, printCommands func(string, ...string)) (string, error) {
 	// Hash input file.
 	fileHash, err := hashFile(abspath)
 	if err != nil {
@@ -68,14 +67,12 @@ func compileAndCacheCFile(abspath, tmpdir string, cflags []string, config *compi
 	buf, err := json.Marshal(struct {
 		Path        string
 		Hash        string
-		Compiler    string
 		Flags       []string
 		LLVMVersion string
 	}{
 		Path:        abspath,
 		Hash:        fileHash,
-		Compiler:    config.Target.Compiler,
-		Flags:       config.CFlags(),
+		Flags:       cflags,
 		LLVMVersion: llvm.Version,
 	})
 	if err != nil {
@@ -124,10 +121,17 @@ func compileAndCacheCFile(abspath, tmpdir string, cflags []string, config *compi
 	flags := append([]string{}, cflags...)                                   // copy cflags
 	flags = append(flags, "-MD", "-MV", "-MTdeps", "-MF", depTmpFile.Name()) // autogenerate dependencies
 	flags = append(flags, "-c", "-o", objTmpFile.Name(), abspath)
-	if config.Options.PrintCommands {
-		fmt.Printf("%s %s\n", config.Target.Compiler, strings.Join(flags, " "))
+	if strings.ToLower(filepath.Ext(abspath)) == ".s" {
+		// If this is an assembly file (.s or .S, lowercase or uppercase), then
+		// we'll need to add -Qunused-arguments because many parameters are
+		// relevant to C, not assembly. And with -Werror, having meaningless
+		// flags (for the assembler) is a compiler error.
+		flags = append(flags, "-Qunused-arguments")
 	}
-	err = runCCompiler(config.Target.Compiler, flags...)
+	if printCommands != nil {
+		printCommands("clang", flags...)
+	}
+	err = runCCompiler(flags...)
 	if err != nil {
 		return "", &commandError{"failed to build", abspath, err}
 	}

builder/config.go

@@ -17,10 +17,18 @@ func NewConfig(options *compileopts.Options) (*compileopts.Config, error) {
 	if err != nil {
 		return nil, err
 	}
+
+	if options.OpenOCDCommands != nil {
+		// Override the OpenOCDCommands from the target spec if specified on
+		// the command-line
+		spec.OpenOCDCommands = options.OpenOCDCommands
+	}
+
 	goroot := goenv.Get("GOROOT")
 	if goroot == "" {
 		return nil, errors.New("cannot locate $GOROOT, please set it manually")
 	}
+
 	major, minor, err := goenv.GetGorootVersion(goroot)
 	if err != nil {
 		return nil, fmt.Errorf("could not read version from GOROOT (%v): %v", goroot, err)
@@ -28,7 +36,9 @@ func NewConfig(options *compileopts.Options) (*compileopts.Config, error) {
 	if major != 1 || minor < 13 || minor > 16 {
 		return nil, fmt.Errorf("requires go version 1.13 through 1.16, got go%d.%d", major, minor)
 	}
+
 	clangHeaderPath := getClangHeaderPath(goenv.Get("TINYGOROOT"))
+
 	return &compileopts.Config{
 		Options:        options,
 		Target:         spec,

builder/elfpatch.go

@@ -0,0 +1,57 @@
+package builder
+
+import (
+	"debug/elf"
+	"fmt"
+	"os"
+)
+
+func getElfSectionData(executable string, sectionName string) ([]byte, elf.FileHeader, error) {
+	elfFile, err := elf.Open(executable)
+	if err != nil {
+		return nil, elf.FileHeader{}, err
+	}
+	defer elfFile.Close()
+
+	section := elfFile.Section(sectionName)
+	if section == nil {
+		return nil, elf.FileHeader{}, fmt.Errorf("could not find %s section", sectionName)
+	}
+
+	data, err := section.Data()
+
+	return data, elfFile.FileHeader, err
+}
+
+func replaceElfSection(executable string, sectionName string, data []byte) error {
+	fp, err := os.OpenFile(executable, os.O_RDWR, 0)
+	if err != nil {
+		return err
+	}
+	defer fp.Close()
+
+	elfFile, err := elf.Open(executable)
+	if err != nil {
+		return err
+	}
+	defer elfFile.Close()
+
+	section := elfFile.Section(sectionName)
+	if section == nil {
+		return fmt.Errorf("could not find %s section", sectionName)
+	}
+
+	// Implicitly check for compressed sections
+	if section.Size != section.FileSize {
+		return fmt.Errorf("expected section %s to have identical size and file size, got %d and %d", sectionName, section.Size, section.FileSize)
+	}
+
+	// Only permit complete replacement of section
+	if section.Size != uint64(len(data)) {
+		return fmt.Errorf("expected section %s to have size %d, was actually %d", sectionName, len(data), section.Size)
+	}
+
+	// Write the replacement section data
+	_, err = fp.WriteAt(data, int64(section.Offset))
+	return err
+}

builder/library.go

@@ -136,7 +136,7 @@ func (l *Library) load(target, cpu, tmpdir string) (job *compileJob, err error)
 				var compileArgs []string
 				compileArgs = append(compileArgs, args...)
 				compileArgs = append(compileArgs, "-o", objpath, srcpath)
-				err := runCCompiler("clang", compileArgs...)
+				err := runCCompiler(compileArgs...)
 				if err != nil {
 					return &commandError{"failed to build", srcpath, err}
 				}

builder/nrfutil.go

@@ -0,0 +1,27 @@
+package builder
+
+import (
+	"fmt"
+	"io/ioutil"
+	"os/exec"
+
+	"github.com/tinygo-org/tinygo/compileopts"
+)
+
+// https://infocenter.nordicsemi.com/index.jsp?topic=%2Fug_nrfutil%2FUG%2Fnrfutil%2Fnrfutil_intro.html
+
+func makeDFUFirmwareImage(options *compileopts.Options, infile, outfile string) error {
+	cmdLine := []string{"nrfutil", "pkg", "generate", "--hw-version", "52", "--sd-req", "0x0", "--debug-mode", "--application", infile, outfile}
+
+	if options.PrintCommands != nil {
+		options.PrintCommands(cmdLine[0], cmdLine[1:]...)
+	}
+
+	cmd := exec.Command(cmdLine[0], cmdLine[1:]...)
+	cmd.Stdout = ioutil.Discard
+	err := cmd.Run()
+	if err != nil {
+		return fmt.Errorf("could not run nrfutil pkg generate: %w", err)
+	}
+	return nil
+}

builder/tools.go

@@ -9,8 +9,8 @@ import (
 )
 
 // runCCompiler invokes a C compiler with the given arguments.
-func runCCompiler(command string, flags ...string) error {
-	if hasBuiltinTools && command == "clang" {
+func runCCompiler(flags ...string) error {
+	if hasBuiltinTools {
 		// Compile this with the internal Clang compiler.
 		headerPath := getClangHeaderPath(goenv.Get("TINYGOROOT"))
 		if headerPath == "" {
@@ -23,17 +23,8 @@ func runCCompiler(command string, flags ...string) error {
 		return cmd.Run()
 	}
 
-	// Running some other compiler. Maybe it has been defined in the
-	// commands map (unlikely).
-	if cmdNames, ok := commands[command]; ok {
-		return execCommand(cmdNames, flags...)
-	}
-
-	// Alternatively, run the compiler directly.
-	cmd := exec.Command(command, flags...)
-	cmd.Stdout = os.Stdout
-	cmd.Stderr = os.Stderr
-	return cmd.Run()
+	// Compile this with an external invocation of the Clang compiler.
+	return execCommand(commands["clang"], flags...)
 }
 
 // link invokes a linker with the given name and flags.

cgo/const.go

@@ -11,105 +11,184 @@ import (
 	"strings"
 )
 
+var (
+	prefixParseFns map[token.Token]func(*tokenizer) (ast.Expr, *scanner.Error)
+	precedences    = map[token.Token]int{
+		token.ADD: precedenceAdd,
+		token.SUB: precedenceAdd,
+		token.MUL: precedenceMul,
+		token.QUO: precedenceMul,
+		token.REM: precedenceMul,
+	}
+)
+
+const (
+	precedenceLowest = iota + 1
+	precedenceAdd
+	precedenceMul
+	precedencePrefix
+)
+
+func init() {
+	// This must be done in an init function to avoid an initialization order
+	// failure.
+	prefixParseFns = map[token.Token]func(*tokenizer) (ast.Expr, *scanner.Error){
+		token.IDENT:  parseIdent,
+		token.INT:    parseBasicLit,
+		token.FLOAT:  parseBasicLit,
+		token.STRING: parseBasicLit,
+		token.CHAR:   parseBasicLit,
+		token.LPAREN: parseParenExpr,
+		token.SUB:    parseUnaryExpr,
+	}
+}
+
 // parseConst parses the given string as a C constant.
 func parseConst(pos token.Pos, fset *token.FileSet, value string) (ast.Expr, *scanner.Error) {
 	t := newTokenizer(pos, fset, value)
-	expr, err := parseConstExpr(t)
-	if t.token != token.EOF {
+	expr, err := parseConstExpr(t, precedenceLowest)
+	t.Next()
+	if t.curToken != token.EOF {
 		return nil, &scanner.Error{
-			Pos: t.fset.Position(t.pos),
-			Msg: "unexpected token " + t.token.String(),
+			Pos: t.fset.Position(t.curPos),
+			Msg: "unexpected token " + t.curToken.String() + ", expected end of expression",
 		}
 	}
 	return expr, err
 }
 
 // parseConstExpr parses a stream of C tokens to a Go expression.
-func parseConstExpr(t *tokenizer) (ast.Expr, *scanner.Error) {
-	switch t.token {
-	case token.LPAREN:
-		lparen := t.pos
-		t.Next()
-		x, err := parseConstExpr(t)
-		if err != nil {
-			return nil, err
-		}
-		if t.token != token.RPAREN {
-			return nil, unexpectedToken(t, token.RPAREN)
-		}
-		expr := &ast.ParenExpr{
-			Lparen: lparen,
-			X:      x,
-			Rparen: t.pos,
-		}
-		t.Next()
-		return expr, nil
-	case token.INT, token.FLOAT, token.STRING, token.CHAR:
-		expr := &ast.BasicLit{
-			ValuePos: t.pos,
-			Kind:     t.token,
-			Value:    t.value,
-		}
-		t.Next()
-		return expr, nil
-	case token.IDENT:
-		expr := &ast.Ident{
-			NamePos: t.pos,
-			Name:    "C." + t.value,
-		}
-		t.Next()
-		return expr, nil
-	case token.EOF:
+func parseConstExpr(t *tokenizer, precedence int) (ast.Expr, *scanner.Error) {
+	if t.curToken == token.EOF {
 		return nil, &scanner.Error{
-			Pos: t.fset.Position(t.pos),
+			Pos: t.fset.Position(t.curPos),
 			Msg: "empty constant",
 		}
-	default:
+	}
+	prefix := prefixParseFns[t.curToken]
+	if prefix == nil {
 		return nil, &scanner.Error{
-			Pos: t.fset.Position(t.pos),
-			Msg: fmt.Sprintf("unexpected token %s", t.token),
+			Pos: t.fset.Position(t.curPos),
+			Msg: fmt.Sprintf("unexpected token %s", t.curToken),
 		}
 	}
+	leftExpr, err := prefix(t)
+
+	for t.peekToken != token.EOF && precedence < precedences[t.peekToken] {
+		switch t.peekToken {
+		case token.ADD, token.SUB, token.MUL, token.QUO, token.REM:
+			t.Next()
+			leftExpr, err = parseBinaryExpr(t, leftExpr)
+		}
+	}
+
+	return leftExpr, err
+}
+
+func parseIdent(t *tokenizer) (ast.Expr, *scanner.Error) {
+	return &ast.Ident{
+		NamePos: t.curPos,
+		Name:    "C." + t.curValue,
+	}, nil
+}
+
+func parseBasicLit(t *tokenizer) (ast.Expr, *scanner.Error) {
+	return &ast.BasicLit{
+		ValuePos: t.curPos,
+		Kind:     t.curToken,
+		Value:    t.curValue,
+	}, nil
+}
+
+func parseParenExpr(t *tokenizer) (ast.Expr, *scanner.Error) {
+	lparen := t.curPos
+	t.Next()
+	x, err := parseConstExpr(t, precedenceLowest)
+	if err != nil {
+		return nil, err
+	}
+	t.Next()
+	if t.curToken != token.RPAREN {
+		return nil, unexpectedToken(t, token.RPAREN)
+	}
+	expr := &ast.ParenExpr{
+		Lparen: lparen,
+		X:      x,
+		Rparen: t.curPos,
+	}
+	return expr, nil
+}
+
+func parseBinaryExpr(t *tokenizer, left ast.Expr) (ast.Expr, *scanner.Error) {
+	expression := &ast.BinaryExpr{
+		X:     left,
+		Op:    t.curToken,
+		OpPos: t.curPos,
+	}
+	precedence := precedences[t.curToken]
+	t.Next()
+	right, err := parseConstExpr(t, precedence)
+	expression.Y = right
+	return expression, err
+}
+
+func parseUnaryExpr(t *tokenizer) (ast.Expr, *scanner.Error) {
+	expression := &ast.UnaryExpr{
+		OpPos: t.curPos,
+		Op:    t.curToken,
+	}
+	t.Next()
+	x, err := parseConstExpr(t, precedencePrefix)
+	expression.X = x
+	return expression, err
 }
 
 // unexpectedToken returns an error of the form "unexpected token FOO, expected
 // BAR".
 func unexpectedToken(t *tokenizer, expected token.Token) *scanner.Error {
 	return &scanner.Error{
-		Pos: t.fset.Position(t.pos),
-		Msg: fmt.Sprintf("unexpected token %s, expected %s", t.token, expected),
+		Pos: t.fset.Position(t.curPos),
+		Msg: fmt.Sprintf("unexpected token %s, expected %s", t.curToken, expected),
 	}
 }
 
 // tokenizer reads C source code and converts it to Go tokens.
 type tokenizer struct {
-	pos   token.Pos
-	fset  *token.FileSet
-	token token.Token
-	value string
-	buf   string
+	curPos, peekPos     token.Pos
+	fset                *token.FileSet
+	curToken, peekToken token.Token
+	curValue, peekValue string
+	buf                 string
 }
 
 // newTokenizer initializes a new tokenizer, positioned at the first token in
 // the string.
 func newTokenizer(start token.Pos, fset *token.FileSet, buf string) *tokenizer {
 	t := &tokenizer{
-		pos:   start,
-		fset:  fset,
-		buf:   buf,
-		token: token.ILLEGAL,
+		peekPos:   start,
+		fset:      fset,
+		buf:       buf,
+		peekToken: token.ILLEGAL,
 	}
-	t.Next() // Parse the first token.
+	// Parse the first two tokens (cur and peek).
+	t.Next()
+	t.Next()
 	return t
 }
 
 // Next consumes the next token in the stream. There is no return value, read
 // the next token from the pos, token and value properties.
 func (t *tokenizer) Next() {
-	t.pos += token.Pos(len(t.value))
+	// The previous peek is now the current token.
+	t.curPos = t.peekPos
+	t.curToken = t.peekToken
+	t.curValue = t.peekValue
+
+	// Parse the next peek token.
+	t.peekPos += token.Pos(len(t.curValue))
 	for {
 		if len(t.buf) == 0 {
-			t.token = token.EOF
+			t.peekToken = token.EOF
 			return
 		}
 		c := t.buf[0]
@@ -118,17 +197,28 @@ func (t *tokenizer) Next() {
 			// Skip whitespace.
 			// Based on this source, not sure whether it represents C whitespace:
 			// https://en.cppreference.com/w/cpp/string/byte/isspace
-			t.pos++
+			t.peekPos++
 			t.buf = t.buf[1:]
-		case c == '(' || c == ')':
+		case c == '(' || c == ')' || c == '+' || c == '-' || c == '*' || c == '/' || c == '%':
 			// Single-character tokens.
+			// TODO: ++ (increment) and -- (decrement) operators.
 			switch c {
 			case '(':
-				t.token = token.LPAREN
+				t.peekToken = token.LPAREN
 			case ')':
-				t.token = token.RPAREN
+				t.peekToken = token.RPAREN
+			case '+':
+				t.peekToken = token.ADD
+			case '-':
+				t.peekToken = token.SUB
+			case '*':
+				t.peekToken = token.MUL
+			case '/':
+				t.peekToken = token.QUO
+			case '%':
+				t.peekToken = token.REM
 			}
-			t.value = t.buf[:1]
+			t.peekValue = t.buf[:1]
 			t.buf = t.buf[1:]
 			return
 		case c >= '0' && c <= '9':
@@ -146,17 +236,17 @@ func (t *tokenizer) Next() {
 					break
 				}
 			}
-			t.value = t.buf[:tokenLen]
+			t.peekValue = t.buf[:tokenLen]
 			t.buf = t.buf[tokenLen:]
 			if hasDot {
 				// Integer constants are more complicated than this but this is
 				// a close approximation.
 				// https://en.cppreference.com/w/cpp/language/integer_literal
-				t.token = token.FLOAT
-				t.value = strings.TrimRight(t.value, "f")
+				t.peekToken = token.FLOAT
+				t.peekValue = strings.TrimRight(t.peekValue, "f")
 			} else {
-				t.token = token.INT
-				t.value = strings.TrimRight(t.value, "uUlL")
+				t.peekToken = token.INT
+				t.peekValue = strings.TrimRight(t.peekValue, "uUlL")
 			}
 			return
 		case c >= 'A' && c <= 'Z' || c >= 'a' && c <= 'z' || c == '_':
@@ -170,9 +260,9 @@ func (t *tokenizer) Next() {
 					break
 				}
 			}
-			t.value = t.buf[:tokenLen]
+			t.peekValue = t.buf[:tokenLen]
 			t.buf = t.buf[tokenLen:]
-			t.token = token.IDENT
+			t.peekToken = token.IDENT
 			return
 		case c == '"':
 			// String constant. Find the first '"' character that is not
@@ -188,8 +278,8 @@ func (t *tokenizer) Next() {
 					escape = c == '\\'
 				}
 			}
-			t.token = token.STRING
-			t.value = t.buf[:tokenLen]
+			t.peekToken = token.STRING
+			t.peekValue = t.buf[:tokenLen]
 			t.buf = t.buf[tokenLen:]
 			return
 		case c == '\'':
@@ -206,12 +296,12 @@ func (t *tokenizer) Next() {
 					escape = c == '\\'
 				}
 			}
-			t.token = token.CHAR
-			t.value = t.buf[:tokenLen]
+			t.peekToken = token.CHAR
+			t.peekValue = t.buf[:tokenLen]
 			t.buf = t.buf[tokenLen:]
 			return
 		default:
-			t.token = token.ILLEGAL
+			t.peekToken = token.ILLEGAL
 			return
 		}
 	}

cgo/const_test.go

@@ -18,7 +18,7 @@ func TestParseConst(t *testing.T) {
 		{`(5)`, `(5)`},
 		{`(((5)))`, `(5)`},
 		{`)`, `error: 1:1: unexpected token )`},
-		{`5)`, `error: 1:2: unexpected token )`},
+		{`5)`, `error: 1:2: unexpected token ), expected end of expression`},
 		{"  \t)", `error: 1:4: unexpected token )`},
 		{`5.8f`, `5.8`},
 		{`foo`, `C.foo`},
@@ -30,7 +30,24 @@ func TestParseConst(t *testing.T) {
 		{`'a'`, `'a'`},
 		{`0b10`, `0b10`},
 		{`0x1234_5678`, `0x1234_5678`},
-		{`5 5`, `error: 1:3: unexpected token INT`}, // test for a bugfix
+		{`5 5`, `error: 1:3: unexpected token INT, expected end of expression`}, // test for a bugfix
+		// Binary operators.
+		{`5+5`, `5 + 5`},
+		{`5-5`, `5 - 5`},
+		{`5*5`, `5 * 5`},
+		{`5/5`, `5 / 5`},
+		{`5%5`, `5 % 5`},
+		{`(5/5)`, `(5 / 5)`},
+		{`1 - 2`, `1 - 2`},
+		{`1 - 2 + 3`, `1 - 2 + 3`},
+		{`1 - 2 * 3`, `1 - 2*3`},
+		{`(1 - 2) * 3`, `(1 - 2) * 3`},
+		{`1 * 2 - 3`, `1*2 - 3`},
+		{`1 * (2 - 3)`, `1 * (2 - 3)`},
+		// Unary operators.
+		{`-5`, `-5`},
+		{`-5-2`, `-5 - 2`},
+		{`5 - - 2`, `5 - -2`},
 	} {
 		fset := token.NewFileSet()
 		startPos := fset.AddFile("", -1, 1000).Pos(0)

cgo/testdata/errors.out.go

@@ -1,7 +1,7 @@
 // CGo errors:
 //     testdata/errors.go:4:2: warning: some warning
 //     testdata/errors.go:11:9: error: unknown type name 'someType'
-//     testdata/errors.go:13:23: unexpected token )
+//     testdata/errors.go:13:23: unexpected token ), expected end of expression
 
 // Type checking errors after CGo processing:
 //     testdata/errors.go:102: cannot use 2 << 10 (untyped int constant 2048) as uint8 value in variable declaration (overflows)

compileopts/config.go

@@ -55,7 +55,7 @@ func (c *Config) GOARCH() string {
 
 // BuildTags returns the complete list of build tags used during this build.
 func (c *Config) BuildTags() []string {
-	tags := append(c.Target.BuildTags, []string{"tinygo", "gc." + c.GC(), "scheduler." + c.Scheduler()}...)
+	tags := append(c.Target.BuildTags, []string{"tinygo", "gc." + c.GC(), "scheduler." + c.Scheduler(), "serial." + c.Serial()}...)
 	for i := 1; i <= c.GoMinorVersion; i++ {
 		tags = append(tags, fmt.Sprintf("go1.%d", i))
 	}
@@ -80,12 +80,7 @@ func (c *Config) GC() string {
 	if c.Target.GC != "" {
 		return c.Target.GC
 	}
-	for _, tag := range c.Target.BuildTags {
-		if tag == "baremetal" || tag == "wasm" {
-			return "conservative"
-		}
-	}
-	return "extalloc"
+	return "conservative"
 }
 
 // NeedsStackObjects returns true if the compiler should insert stack objects
@@ -94,7 +89,7 @@ func (c *Config) NeedsStackObjects() bool {
 	switch c.GC() {
 	case "conservative", "extalloc":
 		for _, tag := range c.BuildTags() {
-			if tag == "wasm" {
+			if tag == "tinygo.wasm" {
 				return true
 			}
 		}
@@ -118,6 +113,39 @@ func (c *Config) Scheduler() string {
 	return "coroutines"
 }
 
+// Serial returns the serial implementation for this build configuration: uart,
+// usb (meaning USB-CDC), or none.
+func (c *Config) Serial() string {
+	if c.Options.Serial != "" {
+		return c.Options.Serial
+	}
+	if c.Target.Serial != "" {
+		return c.Target.Serial
+	}
+	return "none"
+}
+
+// OptLevels returns the optimization level (0-2), size level (0-2), and inliner
+// threshold as used in the LLVM optimization pipeline.
+func (c *Config) OptLevels() (optLevel, sizeLevel int, inlinerThreshold uint) {
+	switch c.Options.Opt {
+	case "none", "0":
+		return 0, 0, 0 // -O0
+	case "1":
+		return 1, 0, 0 // -O1
+	case "2":
+		return 2, 0, 225 // -O2
+	case "s":
+		return 2, 1, 225 // -Os
+	case "z":
+		return 2, 2, 5 // -Oz, default
+	default:
+		// This is not shown to the user: valid choices are already checked as
+		// part of Options.Verify(). It is here as a sanity check.
+		panic("unknown optimization level: -opt=" + c.Options.Opt)
+	}
+}
+
 // FuncImplementation picks an appropriate func value implementation for the
 // target.
 func (c *Config) FuncImplementation() string {
@@ -160,10 +188,19 @@ func (c *Config) AutomaticStackSize() bool {
 	return false
 }
 
+// RP2040BootPatch returns whether the RP2040 boot patch should be applied that
+// calculates and patches in the checksum for the 2nd stage bootloader.
+func (c *Config) RP2040BootPatch() bool {
+	if c.Target.RP2040BootPatch != nil {
+		return *c.Target.RP2040BootPatch
+	}
+	return false
+}
+
 // CFlags returns the flags to pass to the C compiler. This is necessary for CGo
 // preprocessing.
 func (c *Config) CFlags() []string {
-	cflags := append([]string{}, c.Options.CFlags...)
+	var cflags []string
 	for _, flag := range c.Target.CFlags {
 		cflags = append(cflags, strings.ReplaceAll(flag, "{root}", goenv.Get("TINYGOROOT")))
 	}
@@ -184,7 +221,7 @@ func (c *Config) CFlags() []string {
 func (c *Config) LDFlags() []string {
 	root := goenv.Get("TINYGOROOT")
 	// Merge and adjust LDFlags.
-	ldflags := append([]string{}, c.Options.LDFlags...)
+	var ldflags []string
 	for _, flag := range c.Target.LDFlags {
 		ldflags = append(ldflags, strings.ReplaceAll(flag, "{root}", root))
 	}
@@ -238,6 +275,11 @@ func (c *Config) BinaryFormat(ext string) string {
 		// More information:
 		// https://github.com/Microsoft/uf2
 		return "uf2"
+	case ".zip":
+		if c.Target.BinaryFormat != "" {
+			return c.Target.BinaryFormat
+		}
+		return "zip"
 	default:
 		// Use the ELF format for unrecognized file formats.
 		return "elf"
@@ -321,6 +363,10 @@ func (c *Config) WasmAbi() string {
 	return c.Target.WasmAbi
 }
 
+func (c *Config) LLVMFeatures() string {
+	return c.Options.LLVMFeatures
+}
+
 type TestConfig struct {
 	CompileTestBinary bool
 	// TODO: Filter the test functions to run, include verbose flag, etc

compileopts/options.go

@@ -2,37 +2,43 @@ package compileopts
 
 import (
 	"fmt"
+	"regexp"
 	"strings"
 )
 
 var (
 	validGCOptions            = []string{"none", "leaking", "extalloc", "conservative"}
 	validSchedulerOptions     = []string{"none", "tasks", "coroutines"}
+	validSerialOptions        = []string{"none", "uart", "usb"}
 	validPrintSizeOptions     = []string{"none", "short", "full"}
 	validPanicStrategyOptions = []string{"print", "trap"}
+	validOptOptions           = []string{"none", "0", "1", "2", "s", "z"}
 )
 
 // Options contains extra options to give to the compiler. These options are
 // usually passed from the command line.
 type Options struct {
-	Target        string
-	Opt           string
-	GC            string
-	PanicStrategy string
-	Scheduler     string
-	PrintIR       bool
-	DumpSSA       bool
-	VerifyIR      bool
-	PrintCommands bool
-	Debug         bool
-	PrintSizes    string
-	PrintStacks   bool
-	CFlags        []string
-	LDFlags       []string
-	Tags          string
-	WasmAbi       string
-	TestConfig    TestConfig
-	Programmer    string
+	Target          string
+	Opt             string
+	GC              string
+	PanicStrategy   string
+	Scheduler       string
+	Serial          string
+	PrintIR         bool
+	DumpSSA         bool
+	VerifyIR        bool
+	PrintCommands   func(cmd string, args ...string)
+	Debug           bool
+	PrintSizes      string
+	PrintAllocs     *regexp.Regexp // regexp string
+	PrintStacks     bool
+	Tags            string
+	WasmAbi         string
+	GlobalValues    map[string]map[string]string // map[pkgpath]map[varname]value
+	TestConfig      TestConfig
+	Programmer      string
+	OpenOCDCommands []string
+	LLVMFeatures    string
 }
 
 // Verify performs a validation on the given options, raising an error if options are not valid.
@@ -55,6 +61,15 @@ func (o *Options) Verify() error {
 		}
 	}
 
+	if o.Serial != "" {
+		valid := isInArray(validSerialOptions, o.Serial)
+		if !valid {
+			return fmt.Errorf(`invalid serial option '%s': valid values are %s`,
+				o.Serial,
+				strings.Join(validSerialOptions, ", "))
+		}
+	}
+
 	if o.PrintSizes != "" {
 		valid := isInArray(validPrintSizeOptions, o.PrintSizes)
 		if !valid {
@@ -73,6 +88,12 @@ func (o *Options) Verify() error {
 		}
 	}
 
+	if o.Opt != "" {
+		if !isInArray(validOptOptions, o.Opt) {
+			return fmt.Errorf("invalid -opt=%s: valid values are %s", o.Opt, strings.Join(validOptOptions, ", "))
+		}
+	}
+
 	return nil
 }
 

compileopts/target.go

@@ -31,7 +31,7 @@ type TargetSpec struct {
 	BuildTags        []string `json:"build-tags"`
 	GC               string   `json:"gc"`
 	Scheduler        string   `json:"scheduler"`
-	Compiler         string   `json:"compiler"`
+	Serial           string   `json:"serial"` // which serial output to use (uart, usb, none)
 	Linker           string   `json:"linker"`
 	RTLib            string   `json:"rtlib"` // compiler runtime library (libgcc, compiler-rt)
 	Libc             string   `json:"libc"`
@@ -41,10 +41,12 @@ type TargetSpec struct {
 	LDFlags          []string `json:"ldflags"`
 	LinkerScript     string   `json:"linkerscript"`
 	ExtraFiles       []string `json:"extra-files"`
+	RP2040BootPatch  *bool    `json:"rp2040-boot-patch"`        // Patch RP2040 2nd stage bootloader checksum
 	Emulator         []string `json:"emulator" override:"copy"` // inherited Emulator must not be append
 	FlashCommand     string   `json:"flash-command"`
 	GDB              []string `json:"gdb"`
 	PortReset        string   `json:"flash-1200-bps-reset"`
+	SerialPort       []string `json:"serial-port"` // serial port IDs in the form "acm:vid:pid" or "usb:vid:pid"
 	FlashMethod      string   `json:"flash-method"`
 	FlashVolume      string   `json:"msd-volume-name"`
 	FlashFilename    string   `json:"msd-firmware-name"`
@@ -240,23 +242,26 @@ func defaultTarget(goos, goarch, triple string) (*TargetSpec, error) {
 	// No target spec available. Use the default one, useful on most systems
 	// with a regular OS.
 	spec := TargetSpec{
-		Triple:    triple,
-		GOOS:      goos,
-		GOARCH:    goarch,
-		BuildTags: []string{goos, goarch},
-		Compiler:  "clang",
-		Linker:    "cc",
-		CFlags:    []string{"--target=" + triple},
-		GDB:       []string{"gdb"},
-		PortReset: "false",
+		Triple:           triple,
+		GOOS:             goos,
+		GOARCH:           goarch,
+		BuildTags:        []string{goos, goarch},
+		Scheduler:        "tasks",
+		Linker:           "cc",
+		DefaultStackSize: 1024 * 64, // 64kB
+		CFlags:           []string{"--target=" + triple},
+		GDB:              []string{"gdb"},
+		PortReset:        "false",
 	}
 	if goos == "darwin" {
+		spec.CFlags = append(spec.CFlags, "-isysroot", "/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk")
 		spec.LDFlags = append(spec.LDFlags, "-Wl,-dead_strip")
 	} else {
 		spec.LDFlags = append(spec.LDFlags, "-no-pie", "-Wl,--gc-sections") // WARNING: clang < 5.0 requires -nopie
 	}
 	if goarch != "wasm" {
 		spec.ExtraFiles = append(spec.ExtraFiles, "src/runtime/gc_"+goarch+".S")
+		spec.ExtraFiles = append(spec.ExtraFiles, "src/internal/task/task_stack_"+goarch+".S")
 	}
 	if goarch != runtime.GOARCH {
 		// Some educated guesses as to how to invoke helper programs.

compiler/asserts.go

@@ -162,6 +162,10 @@ func (b *builder) createNilCheck(inst ssa.Value, ptr llvm.Value, blockPrefix str
 	case *ssa.Alloc:
 		// An alloc is never nil.
 		return
+	case *ssa.FreeVar:
+		// A free variable is allocated in a parent function and is thus never
+		// nil.
+		return
 	case *ssa.IndexAddr:
 		// This pointer is the result of an index operation into a slice or
 		// array. Such slices/arrays are already bounds checked so the pointer

compiler/compiler.go

@@ -23,7 +23,7 @@ import (
 // Version of the compiler pacakge. Must be incremented each time the compiler
 // package changes in a way that affects the generated LLVM module.
 // This version is independent of the TinyGo version number.
-const Version = 6 // last change: fix issue 1304
+const Version = 12 // last change: implement syscall.rawSyscallNoError
 
 func init() {
 	llvm.InitializeAllTargets()
@@ -59,6 +59,7 @@ type Config struct {
 	DefaultStackSize   uint64
 	NeedsStackObjects  bool
 	Debug              bool // Whether to emit debug information in the LLVM module.
+	LLVMFeatures       string
 }
 
 // compilerContext contains function-independent data that should still be
@@ -185,7 +186,12 @@ func NewTargetMachine(config *Config) (llvm.TargetMachine, error) {
 	if err != nil {
 		return llvm.TargetMachine{}, err
 	}
-	features := strings.Join(config.Features, ",")
+
+	feat := config.Features
+	if len(config.LLVMFeatures) > 0 {
+		feat = append(feat, config.LLVMFeatures)
+	}
+	features := strings.Join(feat, ",")
 
 	var codeModel llvm.CodeModel
 	var relocationModel llvm.RelocMode
@@ -298,7 +304,7 @@ func CompilePackage(moduleName string, pkg *loader.Package, ssaPkg *ssa.Package,
 		c.dibuilder.Finalize()
 	}
 
-	return c.mod, nil
+	return c.mod, c.diagnostics
 }
 
 // getLLVMRuntimeType obtains a named type from the runtime package and returns
@@ -709,11 +715,11 @@ func (c *compilerContext) createPackage(irbuilder llvm.Builder, pkg *ssa.Package
 		member := pkg.Members[name]
 		switch member := member.(type) {
 		case *ssa.Function:
+			// Create the function definition.
+			b := newBuilder(c, irbuilder, member)
 			if member.Blocks == nil {
 				continue // external function
 			}
-			// Create the function definition.
-			b := newBuilder(c, irbuilder, member)
 			b.createFunction()
 		case *ssa.Type:
 			if types.IsInterface(member.Type()) {
@@ -752,10 +758,13 @@ func (c *compilerContext) createPackage(irbuilder llvm.Builder, pkg *ssa.Package
 		case *ssa.Global:
 			// Global variable.
 			info := c.getGlobalInfo(member)
+			global := c.getGlobal(member)
 			if !info.extern {
-				global := c.getGlobal(member)
 				global.SetInitializer(llvm.ConstNull(global.Type().ElementType()))
 				global.SetVisibility(llvm.HiddenVisibility)
+				if info.section != "" {
+					global.SetSection(info.section)
+				}
 			}
 		}
 	}
@@ -781,6 +790,9 @@ func (b *builder) createFunction() {
 		b.llvmFn.SetVisibility(llvm.HiddenVisibility)
 		b.llvmFn.SetUnnamedAddr(true)
 	}
+	if b.info.section != "" {
+		b.llvmFn.SetSection(b.info.section)
+	}
 	if b.info.exported && strings.HasPrefix(b.Triple, "wasm") {
 		// Set the exported name. This is necessary for WebAssembly because
 		// otherwise the function is not exported.
@@ -964,11 +976,61 @@ func (b *builder) createFunction() {
 	}
 }
 
+// posser is an interface that's implemented by both ssa.Value and
+// ssa.Instruction. It is implemented by everything that has a Pos() method,
+// which is all that getPos() needs.
+type posser interface {
+	Pos() token.Pos
+}
+
+// getPos returns position information for a ssa.Value or ssa.Instruction.
+//
+// Not all instructions have position information, especially when they're
+// implicit (such as implicit casts or implicit returns at the end of a
+// function). In these cases, it makes sense to try a bit harder to guess what
+// the position really should be.
+func getPos(val posser) token.Pos {
+	pos := val.Pos()
+	if pos != token.NoPos {
+		// Easy: position is known.
+		return pos
+	}
+
+	// No position information is known.
+	switch val := val.(type) {
+	case *ssa.MakeInterface:
+		return getPos(val.X)
+	case *ssa.MakeClosure:
+		return val.Fn.(*ssa.Function).Pos()
+	case *ssa.Return:
+		syntax := val.Parent().Syntax()
+		if syntax != nil {
+			// non-synthetic
+			return syntax.End()
+		}
+		return token.NoPos
+	case *ssa.FieldAddr:
+		return getPos(val.X)
+	case *ssa.IndexAddr:
+		return getPos(val.X)
+	case *ssa.Slice:
+		return getPos(val.X)
+	case *ssa.Store:
+		return getPos(val.Addr)
+	case *ssa.Extract:
+		return getPos(val.Tuple)
+	default:
+		// This is reachable, for example with *ssa.Const, *ssa.If, and
+		// *ssa.Jump. They might be implemented in some way in the future.
+		return token.NoPos
+	}
+}
+
 // createInstruction builds the LLVM IR equivalent instructions for the
 // particular Go SSA instruction.
 func (b *builder) createInstruction(instr ssa.Instruction) {
 	if b.Debug {
-		pos := b.program.Fset.Position(instr.Pos())
+		pos := b.program.Fset.Position(getPos(instr))
 		b.SetCurrentDebugLocation(uint(pos.Line), uint(pos.Column), b.difunc, llvm.Metadata{})
 	}
 
@@ -994,52 +1056,8 @@ func (b *builder) createInstruction(instr ssa.Instruction) {
 	case *ssa.Defer:
 		b.createDefer(instr)
 	case *ssa.Go:
-		// Get all function parameters to pass to the goroutine.
-		var params []llvm.Value
-		for _, param := range instr.Call.Args {
-			params = append(params, b.getValue(param))
-		}
-
 		// Start a new goroutine.
-		if callee := instr.Call.StaticCallee(); callee != nil {
-			// Static callee is known. This makes it easier to start a new
-			// goroutine.
-			var context llvm.Value
-			switch value := instr.Call.Value.(type) {
-			case *ssa.Function:
-				// Goroutine call is regular function call. No context is necessary.
-				context = llvm.Undef(b.i8ptrType)
-			case *ssa.MakeClosure:
-				// A goroutine call on a func value, but the callee is trivial to find. For
-				// example: immediately applied functions.
-				funcValue := b.getValue(value)
-				context = b.extractFuncContext(funcValue)
-			default:
-				panic("StaticCallee returned an unexpected value")
-			}
-			params = append(params, context) // context parameter
-			b.createGoInstruction(b.getFunction(callee), params, "", callee.Pos())
-		} else if !instr.Call.IsInvoke() {
-			// This is a function pointer.
-			// At the moment, two extra params are passed to the newly started
-			// goroutine:
-			//   * The function context, for closures.
-			//   * The function pointer (for tasks).
-			funcPtr, context := b.decodeFuncValue(b.getValue(instr.Call.Value), instr.Call.Value.Type().Underlying().(*types.Signature))
-			params = append(params, context) // context parameter
-			switch b.Scheduler {
-			case "none", "coroutines":
-				// There are no additional parameters needed for the goroutine start operation.
-			case "tasks":
-				// Add the function pointer as a parameter to start the goroutine.
-				params = append(params, funcPtr)
-			default:
-				panic("unknown scheduler type")
-			}
-			b.createGoInstruction(funcPtr, params, b.fn.RelString(nil), instr.Pos())
-		} else {
-			b.addError(instr.Pos(), "todo: go on interface call")
-		}
+		b.createGo(instr)
 	case *ssa.If:
 		cond := b.getValue(instr.Cond)
 		block := instr.Block()
@@ -1292,6 +1310,8 @@ func (b *builder) createFunctionCall(instr *ssa.CallCommon) (llvm.Value, error)
 			return b.emitCSROperation(instr)
 		case strings.HasPrefix(name, "syscall.Syscall"):
 			return b.createSyscall(instr)
+		case strings.HasPrefix(name, "syscall.rawSyscallNoError"):
+			return b.createRawSyscallNoError(instr)
 		case strings.HasPrefix(name, "runtime/volatile.Load"):
 			return b.createVolatileLoad(instr)
 		case strings.HasPrefix(name, "runtime/volatile.Store"):
@@ -2258,14 +2278,20 @@ func (b *builder) createConst(prefix string, expr *ssa.Const) llvm.Value {
 		} else if typ.Info()&types.IsString != 0 {
 			str := constant.StringVal(expr.Value)
 			strLen := llvm.ConstInt(b.uintptrType, uint64(len(str)), false)
-			objname := prefix + "$string"
-			global := llvm.AddGlobal(b.mod, llvm.ArrayType(b.ctx.Int8Type(), len(str)), objname)
-			global.SetInitializer(b.ctx.ConstString(str, false))
-			global.SetLinkage(llvm.InternalLinkage)
-			global.SetGlobalConstant(true)
-			global.SetUnnamedAddr(true)
-			zero := llvm.ConstInt(b.ctx.Int32Type(), 0, false)
-			strPtr := b.CreateInBoundsGEP(global, []llvm.Value{zero, zero}, "")
+			var strPtr llvm.Value
+			if str != "" {
+				objname := prefix + "$string"
+				global := llvm.AddGlobal(b.mod, llvm.ArrayType(b.ctx.Int8Type(), len(str)), objname)
+				global.SetInitializer(b.ctx.ConstString(str, false))
+				global.SetLinkage(llvm.InternalLinkage)
+				global.SetGlobalConstant(true)
+				global.SetUnnamedAddr(true)
+				global.SetAlignment(1)
+				zero := llvm.ConstInt(b.ctx.Int32Type(), 0, false)
+				strPtr = b.CreateInBoundsGEP(global, []llvm.Value{zero, zero}, "")
+			} else {
+				strPtr = llvm.ConstNull(b.i8ptrType)
+			}
 			strObj := llvm.ConstNamedStruct(b.getLLVMRuntimeType("_string"), []llvm.Value{strPtr, strLen})
 			return strObj
 		} else if typ.Kind() == types.UnsafePointer {

compiler/compiler_test.go

@@ -19,6 +19,8 @@ var flagUpdate = flag.Bool("update", false, "update tests based on test output")
 // Basic tests for the compiler. Build some Go files and compare the output with
 // the expected LLVM IR for regression testing.
 func TestCompiler(t *testing.T) {
+	t.Parallel()
+
 	// Check LLVM version.
 	llvmMajor, err := strconv.Atoi(strings.SplitN(llvm.Version, ".", 2)[0])
 	if err != nil {
@@ -32,42 +34,56 @@ func TestCompiler(t *testing.T) {
 		t.Skip("compiler tests require LLVM 11 or above, got LLVM ", llvm.Version)
 	}
 
-	target, err := compileopts.LoadTarget("i686--linux")
-	if err != nil {
-		t.Fatal("failed to load target:", err)
-	}
-	config := &compileopts.Config{
-		Options: &compileopts.Options{},
-		Target:  target,
-	}
-	compilerConfig := &Config{
-		Triple:             config.Triple(),
-		GOOS:               config.GOOS(),
-		GOARCH:             config.GOARCH(),
-		CodeModel:          config.CodeModel(),
-		RelocationModel:    config.RelocationModel(),
-		Scheduler:          config.Scheduler(),
-		FuncImplementation: config.FuncImplementation(),
-		AutomaticStackSize: config.AutomaticStackSize(),
-	}
-	machine, err := NewTargetMachine(compilerConfig)
-	if err != nil {
-		t.Fatal("failed to create target machine:", err)
+	tests := []struct {
+		file   string
+		target string
+	}{
+		{"basic.go", ""},
+		{"pointer.go", ""},
+		{"slice.go", ""},
+		{"string.go", ""},
+		{"float.go", ""},
+		{"interface.go", ""},
+		{"func.go", ""},
+		{"pragma.go", ""},
+		{"goroutine.go", "wasm"},
+		{"goroutine.go", "cortex-m-qemu"},
 	}
 
-	tests := []string{
-		"basic.go",
-		"pointer.go",
-		"slice.go",
-		"string.go",
-		"float.go",
-		"interface.go",
-	}
+	for _, tc := range tests {
+		name := tc.file
+		targetString := "wasm"
+		if tc.target != "" {
+			targetString = tc.target
+			name = tc.file + "-" + tc.target
+		}
+
+		t.Run(name, func(t *testing.T) {
+			target, err := compileopts.LoadTarget(targetString)
+			if err != nil {
+				t.Fatal("failed to load target:", err)
+			}
+			config := &compileopts.Config{
+				Options: &compileopts.Options{},
+				Target:  target,
+			}
+			compilerConfig := &Config{
+				Triple:             config.Triple(),
+				GOOS:               config.GOOS(),
+				GOARCH:             config.GOARCH(),
+				CodeModel:          config.CodeModel(),
+				RelocationModel:    config.RelocationModel(),
+				Scheduler:          config.Scheduler(),
+				FuncImplementation: config.FuncImplementation(),
+				AutomaticStackSize: config.AutomaticStackSize(),
+			}
+			machine, err := NewTargetMachine(compilerConfig)
+			if err != nil {
+				t.Fatal("failed to create target machine:", err)
+			}
 
-	for _, testCase := range tests {
-		t.Run(testCase, func(t *testing.T) {
 			// Load entire program AST into memory.
-			lprogram, err := loader.Load(config, []string{"./testdata/" + testCase}, config.ClangHeaders, types.Config{
+			lprogram, err := loader.Load(config, []string{"./testdata/" + tc.file}, config.ClangHeaders, types.Config{
 				Sizes: Sizes(machine),
 			})
 			if err != nil {
@@ -75,20 +91,25 @@ func TestCompiler(t *testing.T) {
 			}
 			err = lprogram.Parse()
 			if err != nil {
-				t.Fatalf("could not parse test case %s: %s", testCase, err)
+				t.Fatalf("could not parse test case %s: %s", tc.file, err)
 			}
 
 			// Compile AST to IR.
 			program := lprogram.LoadSSA()
 			pkg := lprogram.MainPkg()
-			mod, errs := CompilePackage(testCase, pkg, program.Package(pkg.Pkg), machine, compilerConfig, false)
+			mod, errs := CompilePackage(tc.file, pkg, program.Package(pkg.Pkg), machine, compilerConfig, false)
 			if errs != nil {
 				for _, err := range errs {
-					t.Log("error:", err)
+					t.Error(err)
 				}
 				return
 			}
 
+			err = llvm.VerifyModule(mod, llvm.PrintMessageAction)
+			if err != nil {
+				t.Error(err)
+			}
+
 			// Optimize IR a little.
 			funcPasses := llvm.NewFunctionPassManagerForModule(mod)
 			defer funcPasses.Dispose()
@@ -99,18 +120,22 @@ func TestCompiler(t *testing.T) {
 			}
 			funcPasses.FinalizeFunc()
 
-			outfile := "./testdata/" + testCase[:len(testCase)-3] + ".ll"
+			outFilePrefix := tc.file[:len(tc.file)-3]
+			if tc.target != "" {
+				outFilePrefix += "-" + tc.target
+			}
+			outPath := "./testdata/" + outFilePrefix + ".ll"
 
 			// Update test if needed. Do not check the result.
 			if *flagUpdate {
-				err := ioutil.WriteFile(outfile, []byte(mod.String()), 0666)
+				err := ioutil.WriteFile(outPath, []byte(mod.String()), 0666)
 				if err != nil {
 					t.Error("failed to write updated output file:", err)
 				}
 				return
 			}
 
-			expected, err := ioutil.ReadFile(outfile)
+			expected, err := ioutil.ReadFile(outPath)
 			if err != nil {
 				t.Fatal("failed to read golden file:", err)
 			}

compiler/func.go

@@ -25,14 +25,13 @@ func (c *compilerContext) createFuncValue(builder llvm.Builder, funcPtr, context
 		// Closure is: {context, function pointer}
 		funcValueScalar = funcPtr
 	case "switch":
-		sigGlobal := c.getTypeCode(sig)
 		funcValueWithSignatureGlobalName := funcPtr.Name() + "$withSignature"
 		funcValueWithSignatureGlobal := c.mod.NamedGlobal(funcValueWithSignatureGlobalName)
 		if funcValueWithSignatureGlobal.IsNil() {
 			funcValueWithSignatureType := c.getLLVMRuntimeType("funcValueWithSignature")
 			funcValueWithSignature := llvm.ConstNamedStruct(funcValueWithSignatureType, []llvm.Value{
 				llvm.ConstPtrToInt(funcPtr, c.uintptrType),
-				sigGlobal,
+				c.getFuncSignatureID(sig),
 			})
 			funcValueWithSignatureGlobal = llvm.AddGlobal(c.mod, funcValueWithSignatureType, funcValueWithSignatureGlobalName)
 			funcValueWithSignatureGlobal.SetInitializer(funcValueWithSignature)
@@ -50,6 +49,19 @@ func (c *compilerContext) createFuncValue(builder llvm.Builder, funcPtr, context
 	return funcValue
 }
 
+// getFuncSignatureID returns a new external global for a given signature. This
+// global reference is not real, it is only used during func lowering to assign
+// signature types to functions and will then be removed.
+func (c *compilerContext) getFuncSignatureID(sig *types.Signature) llvm.Value {
+	sigGlobalName := "reflect/types.funcid:" + getTypeCodeName(sig)
+	sigGlobal := c.mod.NamedGlobal(sigGlobalName)
+	if sigGlobal.IsNil() {
+		sigGlobal = llvm.AddGlobal(c.mod, c.ctx.Int8Type(), sigGlobalName)
+		sigGlobal.SetGlobalConstant(true)
+	}
+	return sigGlobal
+}
+
 // extractFuncScalar returns some scalar that can be used in comparisons. It is
 // a cheap operation.
 func (b *builder) extractFuncScalar(funcValue llvm.Value) llvm.Value {
@@ -71,7 +83,7 @@ func (b *builder) decodeFuncValue(funcValue llvm.Value, sig *types.Signature) (f
 		funcPtr = b.CreateExtractValue(funcValue, 1, "")
 	case "switch":
 		llvmSig := b.getRawFuncType(sig)
-		sigGlobal := b.getTypeCode(sig)
+		sigGlobal := b.getFuncSignatureID(sig)
 		funcPtr = b.createRuntimeCall("getFuncPtr", []llvm.Value{funcValue, sigGlobal}, "")
 		funcPtr = b.CreateIntToPtr(funcPtr, llvmSig, "")
 	default:

compiler/goroutine.go

@@ -5,25 +5,110 @@ package compiler
 
 import (
 	"go/token"
+	"go/types"
 
 	"github.com/tinygo-org/tinygo/compiler/llvmutil"
 	"golang.org/x/tools/go/ssa"
 	"tinygo.org/x/go-llvm"
 )
 
-// createGoInstruction starts a new goroutine with the provided function pointer
-// and parameters.
-// In general, you should pass all regular parameters plus the context parameter.
-// There is one exception: the task-based scheduler needs to have the function
-// pointer passed in as a parameter too in addition to the context.
-//
-// Because a go statement doesn't return anything, return undef.
-func (b *builder) createGoInstruction(funcPtr llvm.Value, params []llvm.Value, prefix string, pos token.Pos) llvm.Value {
+// createGo emits code to start a new goroutine.
+func (b *builder) createGo(instr *ssa.Go) {
+	// Get all function parameters to pass to the goroutine.
+	var params []llvm.Value
+	for _, param := range instr.Call.Args {
+		params = append(params, b.getValue(param))
+	}
+
+	var prefix string
+	var funcPtr llvm.Value
+	hasContext := false
+	if callee := instr.Call.StaticCallee(); callee != nil {
+		// Static callee is known. This makes it easier to start a new
+		// goroutine.
+		var context llvm.Value
+		switch value := instr.Call.Value.(type) {
+		case *ssa.Function:
+			// Goroutine call is regular function call. No context is necessary.
+			if b.Scheduler == "coroutines" {
+				// The context parameter is assumed to be always present in the
+				// coroutines scheduler.
+				context = llvm.Undef(b.i8ptrType)
+			}
+		case *ssa.MakeClosure:
+			// A goroutine call on a func value, but the callee is trivial to find. For
+			// example: immediately applied functions.
+			funcValue := b.getValue(value)
+			context = b.extractFuncContext(funcValue)
+		default:
+			panic("StaticCallee returned an unexpected value")
+		}
+		if !context.IsNil() {
+			params = append(params, context) // context parameter
+			hasContext = true
+		}
+		funcPtr = b.getFunction(callee)
+	} else if builtin, ok := instr.Call.Value.(*ssa.Builtin); ok {
+		// We cheat. None of the builtins do any long or blocking operation, so
+		// we might as well run these builtins right away without the program
+		// noticing the difference.
+		// Possible exceptions:
+		//   - copy: this is a possibly long operation, but not a blocking
+		//     operation. Semantically it makes no difference to run it right
+		//     away (not in a goroutine). However, in practice it makes no sense
+		//     to run copy in a goroutine as there is no way to (safely) know
+		//     when it is finished.
+		//   - panic: the error message would appear in the parent goroutine.
+		//     But because `go panic("err")` would halt the program anyway
+		//     (there is no recover), panicking right away would give the same
+		//     behavior as creating a goroutine, switching the scheduler to that
+		//     goroutine, and panicking there. So this optimization seems
+		//     correct.
+		//   - recover: because it runs in a new goroutine, it is never a
+		//     deferred function. Thus this is a no-op.
+		if builtin.Name() == "recover" {
+			// This is a no-op, even in a deferred function:
+			//   go recover()
+			return
+		}
+		var argTypes []types.Type
+		var argValues []llvm.Value
+		for _, arg := range instr.Call.Args {
+			argTypes = append(argTypes, arg.Type())
+			argValues = append(argValues, b.getValue(arg))
+		}
+		b.createBuiltin(argTypes, argValues, builtin.Name(), instr.Pos())
+		return
+	} else if !instr.Call.IsInvoke() {
+		// This is a function pointer.
+		// At the moment, two extra params are passed to the newly started
+		// goroutine:
+		//   * The function context, for closures.
+		//   * The function pointer (for tasks).
+		var context llvm.Value
+		funcPtr, context = b.decodeFuncValue(b.getValue(instr.Call.Value), instr.Call.Value.Type().Underlying().(*types.Signature))
+		params = append(params, context) // context parameter
+		hasContext = true
+		switch b.Scheduler {
+		case "none", "coroutines":
+			// There are no additional parameters needed for the goroutine start operation.
+		case "tasks":
+			// Add the function pointer as a parameter to start the goroutine.
+			params = append(params, funcPtr)
+		default:
+			panic("unknown scheduler type")
+		}
+		prefix = b.fn.RelString(nil)
+	} else {
+		b.addError(instr.Pos(), "todo: go on interface call")
+		return
+	}
+
 	paramBundle := b.emitPointerPack(params)
 	var callee, stackSize llvm.Value
 	switch b.Scheduler {
 	case "none", "tasks":
-		callee = b.createGoroutineStartWrapper(funcPtr, prefix, pos)
+		callee = b.createGoroutineStartWrapper(funcPtr, prefix, hasContext, instr.Pos())
 		if b.AutomaticStackSize {
 			// The stack size is not known until after linking. Call a dummy
 			// function that will be replaced with a load from a special ELF
@@ -34,6 +119,9 @@ func (b *builder) createGoInstruction(funcPtr llvm.Value, params []llvm.Value, p
 		} else {
 			// The stack size is fixed at compile time. By emitting it here as a
 			// constant, it can be optimized.
+			if b.Scheduler == "tasks" && b.DefaultStackSize == 0 {
+				b.addError(instr.Pos(), "default stack size for goroutines is not set")
+			}
 			stackSize = llvm.ConstInt(b.uintptrType, b.DefaultStackSize, false)
 		}
 	case "coroutines":
@@ -46,7 +134,6 @@ func (b *builder) createGoInstruction(funcPtr llvm.Value, params []llvm.Value, p
 	}
 	start := b.getFunction(b.program.ImportedPackage("internal/task").Members["start"].(*ssa.Function))
 	b.createCall(start, []llvm.Value{callee, paramBundle, stackSize, llvm.Undef(b.i8ptrType), llvm.ConstPointerNull(b.i8ptrType)}, "")
-	return llvm.Undef(funcPtr.Type().ElementType().ReturnType())
 }
 
 // createGoroutineStartWrapper creates a wrapper for the task-based
@@ -67,7 +154,12 @@ func (b *builder) createGoInstruction(funcPtr llvm.Value, params []llvm.Value, p
 // allows a single (pointer) argument to the newly started goroutine. Also, it
 // ignores the return value because newly started goroutines do not have a
 // return value.
-func (c *compilerContext) createGoroutineStartWrapper(fn llvm.Value, prefix string, pos token.Pos) llvm.Value {
+//
+// The hasContext parameter indicates whether the context parameter (the second
+// to last parameter of the function) is used for this wrapper. If hasContext is
+// false, the parameter bundle is assumed to have no context parameter and undef
+// is passed instead.
+func (c *compilerContext) createGoroutineStartWrapper(fn llvm.Value, prefix string, hasContext bool, pos token.Pos) llvm.Value {
 	var wrapper llvm.Value
 
 	builder := c.ctx.NewBuilder()
@@ -114,8 +206,15 @@ func (c *compilerContext) createGoroutineStartWrapper(fn llvm.Value, prefix stri
 
 		// Create the list of params for the call.
 		paramTypes := fn.Type().ElementType().ParamTypes()
-		params := llvmutil.EmitPointerUnpack(builder, c.mod, wrapper.Param(0), paramTypes[:len(paramTypes)-1])
-		params = append(params, llvm.Undef(c.i8ptrType))
+		paramTypes = paramTypes[:len(paramTypes)-1] // strip parentHandle parameter
+		if !hasContext {
+			paramTypes = paramTypes[:len(paramTypes)-1] // strip context parameter
+		}
+		params := llvmutil.EmitPointerUnpack(builder, c.mod, wrapper.Param(0), paramTypes)
+		if !hasContext {
+			params = append(params, llvm.Undef(c.i8ptrType)) // add dummy context parameter
+		}
+		params = append(params, llvm.Undef(c.i8ptrType)) // add dummy parentHandle parameter
 
 		// Create the call.
 		builder.CreateCall(fn, params, "")
@@ -177,8 +276,11 @@ func (c *compilerContext) createGoroutineStartWrapper(fn llvm.Value, prefix stri
 		// Get the function pointer.
 		fnPtr := params[len(params)-1]
 
-		// Ignore the last param, which isn't used anymore.
-		// TODO: avoid this extra "parent handle" parameter in most functions.
+		// The last parameter in the packed object has somewhat of a dual role.
+		// Inside the parameter bundle it's the function pointer, stored right
+		// after the context pointer. But in the IR call instruction, it's the
+		// parentHandle function that's always undef outside of the coroutines
+		// scheduler. Thus, make the parameter undef here.
 		params[len(params)-1] = llvm.Undef(c.i8ptrType)
 
 		// Create the call.

compiler/interface.go

@@ -46,6 +46,7 @@ func (c *compilerContext) getTypeCode(typ types.Type) llvm.Value {
 		var references llvm.Value
 		var length int64
 		var methodSet llvm.Value
+		var ptrTo llvm.Value
 		switch typ := typ.(type) {
 		case *types.Named:
 			references = c.getTypeCode(typ.Underlying())
@@ -69,22 +70,25 @@ func (c *compilerContext) getTypeCode(typ types.Type) llvm.Value {
 		if _, ok := typ.Underlying().(*types.Interface); !ok {
 			methodSet = c.getTypeMethodSet(typ)
 		}
-		if !references.IsNil() || length != 0 || !methodSet.IsNil() {
-			// Set the 'references' field of the runtime.typecodeID struct.
-			globalValue := llvm.ConstNull(global.Type().ElementType())
-			if !references.IsNil() {
-				globalValue = llvm.ConstInsertValue(globalValue, references, []uint32{0})
-			}
-			if length != 0 {
-				lengthValue := llvm.ConstInt(c.uintptrType, uint64(length), false)
-				globalValue = llvm.ConstInsertValue(globalValue, lengthValue, []uint32{1})
-			}
-			if !methodSet.IsNil() {
-				globalValue = llvm.ConstInsertValue(globalValue, methodSet, []uint32{2})
-			}
-			global.SetInitializer(globalValue)
-			global.SetLinkage(llvm.LinkOnceODRLinkage)
+		if _, ok := typ.Underlying().(*types.Pointer); !ok {
+			ptrTo = c.getTypeCode(types.NewPointer(typ))
 		}
+		globalValue := llvm.ConstNull(global.Type().ElementType())
+		if !references.IsNil() {
+			globalValue = llvm.ConstInsertValue(globalValue, references, []uint32{0})
+		}
+		if length != 0 {
+			lengthValue := llvm.ConstInt(c.uintptrType, uint64(length), false)
+			globalValue = llvm.ConstInsertValue(globalValue, lengthValue, []uint32{1})
+		}
+		if !methodSet.IsNil() {
+			globalValue = llvm.ConstInsertValue(globalValue, methodSet, []uint32{2})
+		}
+		if !ptrTo.IsNil() {
+			globalValue = llvm.ConstInsertValue(globalValue, ptrTo, []uint32{3})
+		}
+		global.SetInitializer(globalValue)
+		global.SetLinkage(llvm.LinkOnceODRLinkage)
 		global.SetGlobalConstant(true)
 	}
 	return global
@@ -307,10 +311,21 @@ func (c *compilerContext) getInterfaceMethodSet(typ types.Type) llvm.Value {
 // used during the interface lowering pass.
 func (c *compilerContext) getMethodSignature(method *types.Func) llvm.Value {
 	signature := methodSignature(method)
-	signatureGlobal := c.mod.NamedGlobal("func " + signature)
+	var globalName string
+	if token.IsExported(method.Name()) {
+		globalName = "reflect/methods." + signature
+	} else {
+		globalName = method.Type().(*types.Signature).Recv().Pkg().Path() + ".$methods." + signature
+	}
+	signatureGlobal := c.mod.NamedGlobal(globalName)
 	if signatureGlobal.IsNil() {
-		signatureGlobal = llvm.AddGlobal(c.mod, c.ctx.Int8Type(), "func "+signature)
+		// TODO: put something useful in these globals, such as the method
+		// signature. Useful to one day implement reflect.Value.Method(n).
+		signatureGlobal = llvm.AddGlobal(c.mod, c.ctx.Int8Type(), globalName)
+		signatureGlobal.SetInitializer(llvm.ConstInt(c.ctx.Int8Type(), 0, false))
+		signatureGlobal.SetLinkage(llvm.LinkOnceODRLinkage)
 		signatureGlobal.SetGlobalConstant(true)
+		signatureGlobal.SetAlignment(1)
 	}
 	return signatureGlobal
 }
@@ -341,7 +356,7 @@ func (b *builder) createTypeAssert(expr *ssa.TypeAssert) llvm.Value {
 		commaOk = b.createRuntimeCall("interfaceImplements", []llvm.Value{actualTypeNum, methodSet}, "")
 
 	} else {
-		globalName := "reflect/types.type:" + getTypeCodeName(expr.AssertedType) + "$id"
+		globalName := "reflect/types.typeid:" + getTypeCodeName(expr.AssertedType)
 		assertedTypeCodeGlobal := b.mod.NamedGlobal(globalName)
 		if assertedTypeCodeGlobal.IsNil() {
 			// Create a new typecode global.

compiler/interrupt.go

@@ -41,10 +41,7 @@ func (b *builder) createInterruptGlobal(instr *ssa.CallCommon) (llvm.Value, erro
 	// type are lowered in the interrupt lowering pass.
 	globalType := b.program.ImportedPackage("runtime/interrupt").Type("handle").Type()
 	globalLLVMType := b.getLLVMType(globalType)
-	globalName := "runtime/interrupt.$interrupt" + strconv.FormatInt(id.Int64(), 10)
-	if global := b.mod.NamedGlobal(globalName); !global.IsNil() {
-		return llvm.Value{}, b.makeError(instr.Pos(), "interrupt redeclared in this program")
-	}
+	globalName := b.fn.Package().Pkg.Path() + "$interrupt" + strconv.FormatInt(id.Int64(), 10)
 	global := llvm.AddGlobal(b.mod, globalLLVMType, globalName)
 	global.SetVisibility(llvm.HiddenVisibility)
 	global.SetGlobalConstant(true)

compiler/symbol.go

@@ -24,6 +24,7 @@ type functionInfo struct {
 	module     string     // go:wasm-module
 	importName string     // go:linkname, go:export - The name the developer assigns
 	linkName   string     // go:linkname, go:export - The name that we map for the particular module -> importName
+	section    string     // go:section - object file section name
 	exported   bool       // go:export, CGo
 	nobounds   bool       // go:nobounds
 	variadic   bool       // go:variadic (CGo only)
@@ -139,6 +140,17 @@ func (c *compilerContext) getFunction(fn *ssa.Function) llvm.Value {
 		for _, attrName := range []string{"noalias", "nonnull"} {
 			llvmFn.AddAttributeAtIndex(0, c.ctx.CreateEnumAttribute(llvm.AttributeKindID(attrName), 0))
 		}
+	case "runtime.sliceAppend":
+		// Appending a slice will only read the to-be-appended slice, it won't
+		// be modified.
+		llvmFn.AddAttributeAtIndex(2, c.ctx.CreateEnumAttribute(llvm.AttributeKindID("nocapture"), 0))
+		llvmFn.AddAttributeAtIndex(2, c.ctx.CreateEnumAttribute(llvm.AttributeKindID("readonly"), 0))
+	case "runtime.sliceCopy":
+		// Copying a slice won't capture any of the parameters.
+		llvmFn.AddAttributeAtIndex(1, c.ctx.CreateEnumAttribute(llvm.AttributeKindID("writeonly"), 0))
+		llvmFn.AddAttributeAtIndex(1, c.ctx.CreateEnumAttribute(llvm.AttributeKindID("nocapture"), 0))
+		llvmFn.AddAttributeAtIndex(2, c.ctx.CreateEnumAttribute(llvm.AttributeKindID("readonly"), 0))
+		llvmFn.AddAttributeAtIndex(2, c.ctx.CreateEnumAttribute(llvm.AttributeKindID("nocapture"), 0))
 	case "runtime.trackPointer":
 		// This function is necessary for tracking pointers on the stack in a
 		// portable way (see gc_stack_portable.go). Indicate to the optimizer
@@ -259,6 +271,10 @@ func (info *functionInfo) parsePragmas(f *ssa.Function) {
 				if hasUnsafeImport(f.Pkg.Pkg) {
 					info.linkName = parts[2]
 				}
+			case "//go:section":
+				if len(parts) == 2 && hasUnsafeImport(f.Pkg.Pkg) {
+					info.section = parts[1]
+				}
 			case "//go:nobounds":
 				// Skip bounds checking in this function. Useful for some
 				// runtime functions.
@@ -314,6 +330,7 @@ type globalInfo struct {
 	linkName string // go:extern
 	extern   bool   // go:extern
 	align    int    // go:align
+	section  string // go:section
 }
 
 // loadASTComments loads comments on globals from the AST, for use later in the
@@ -355,16 +372,18 @@ func (c *compilerContext) getGlobal(g *ssa.Global) llvm.Value {
 
 		// Set alignment from the //go:align comment.
 		var alignInBits uint32
-		if info.align < 0 || info.align&(info.align-1) != 0 {
+		alignment := c.targetData.ABITypeAlignment(llvmType)
+		if info.align > alignment {
+			alignment = info.align
+		}
+		if alignment <= 0 || alignment&(alignment-1) != 0 {
 			// Check for power-of-two (or 0).
 			// See: https://stackoverflow.com/a/108360
 			c.addError(g.Pos(), "global variable alignment must be a positive power of two")
 		} else {
 			// Set the alignment only when it is a power of two.
-			alignInBits = uint32(info.align) ^ uint32(info.align-1)
-			if info.align > c.targetData.ABITypeAlignment(llvmType) {
-				llvmGlobal.SetAlignment(info.align)
-			}
+			alignInBits = uint32(alignment) ^ uint32(alignment-1)
+			llvmGlobal.SetAlignment(alignment)
 		}
 
 		if c.Debug && !info.extern {
@@ -425,6 +444,10 @@ func (info *globalInfo) parsePragmas(doc *ast.CommentGroup) {
 			if err == nil {
 				info.align = align
 			}
+		case "//go:section":
+			if len(parts) == 2 {
+				info.section = parts[1]
+			}
 		}
 	}
 }

compiler/syscall.go

@@ -10,11 +10,11 @@ import (
 	"tinygo.org/x/go-llvm"
 )
 
-// createSyscall emits an inline system call instruction, depending on the
-// target OS/arch.
-func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
+// createRawSyscall creates a system call with the provided system call number
+// and returns the result as a single integer (the system call result). The
+// result is not further interpreted.
+func (b *builder) createRawSyscall(call *ssa.CallCommon) (llvm.Value, error) {
 	num := b.getValue(call.Args[0])
-	var syscallResult llvm.Value
 	switch {
 	case b.GOARCH == "amd64":
 		if b.GOOS == "darwin" {
@@ -57,7 +57,7 @@ func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
 		constraints += ",~{rcx},~{r11}"
 		fnType := llvm.FunctionType(b.uintptrType, argTypes, false)
 		target := llvm.InlineAsm(fnType, "syscall", constraints, true, false, llvm.InlineAsmDialectIntel)
-		syscallResult = b.CreateCall(target, args, "")
+		return b.CreateCall(target, args, ""), nil
 	case b.GOARCH == "386" && b.GOOS == "linux":
 		// Sources:
 		//   syscall(2) man page
@@ -83,7 +83,7 @@ func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
 		}
 		fnType := llvm.FunctionType(b.uintptrType, argTypes, false)
 		target := llvm.InlineAsm(fnType, "int 0x80", constraints, true, false, llvm.InlineAsmDialectIntel)
-		syscallResult = b.CreateCall(target, args, "")
+		return b.CreateCall(target, args, ""), nil
 	case b.GOARCH == "arm" && b.GOOS == "linux":
 		// Implement the EABI system call convention for Linux.
 		// Source: syscall(2) man page.
@@ -115,7 +115,7 @@ func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
 		}
 		fnType := llvm.FunctionType(b.uintptrType, argTypes, false)
 		target := llvm.InlineAsm(fnType, "svc #0", constraints, true, false, 0)
-		syscallResult = b.CreateCall(target, args, "")
+		return b.CreateCall(target, args, ""), nil
 	case b.GOARCH == "arm64" && b.GOOS == "linux":
 		// Source: syscall(2) man page.
 		args := []llvm.Value{}
@@ -147,10 +147,19 @@ func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
 		constraints += ",~{x16},~{x17}" // scratch registers
 		fnType := llvm.FunctionType(b.uintptrType, argTypes, false)
 		target := llvm.InlineAsm(fnType, "svc #0", constraints, true, false, 0)
-		syscallResult = b.CreateCall(target, args, "")
+		return b.CreateCall(target, args, ""), nil
 	default:
 		return llvm.Value{}, b.makeError(call.Pos(), "unknown GOOS/GOARCH for syscall: "+b.GOOS+"/"+b.GOARCH)
 	}
+}
+
+// createSyscall emits instructions for the syscall.Syscall* family of
+// functions, depending on the target OS/arch.
+func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
+	syscallResult, err := b.createRawSyscall(call)
+	if err != nil {
+		return syscallResult, err
+	}
 	switch b.GOOS {
 	case "linux", "freebsd":
 		// Return values: r0, r1 uintptr, err Errno
@@ -190,3 +199,16 @@ func (b *builder) createSyscall(call *ssa.CallCommon) (llvm.Value, error) {
 		return llvm.Value{}, b.makeError(call.Pos(), "unknown GOOS/GOARCH for syscall: "+b.GOOS+"/"+b.GOARCH)
 	}
 }
+
+// createRawSyscallNoError emits instructions for the Linux-specific
+// syscall.rawSyscallNoError function.
+func (b *builder) createRawSyscallNoError(call *ssa.CallCommon) (llvm.Value, error) {
+	syscallResult, err := b.createRawSyscall(call)
+	if err != nil {
+		return syscallResult, err
+	}
+	retval := llvm.ConstNull(b.ctx.StructType([]llvm.Type{b.uintptrType, b.uintptrType}, false))
+	retval = b.CreateInsertValue(retval, syscallResult, 0, "")
+	retval = b.CreateInsertValue(retval, llvm.ConstInt(b.uintptrType, 0, false), 1, "")
+	return retval, nil
+}

compiler/testdata/basic.ll

@@ -1,7 +1,7 @@
 ; ModuleID = 'basic.go'
 source_filename = "basic.go"
-target datalayout = "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-f64:32:64-f80:32-n8:16:32-S128"
-target triple = "i686--linux"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
 
 declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
 

compiler/testdata/float.ll

@@ -1,7 +1,7 @@
 ; ModuleID = 'float.go'
 source_filename = "float.go"
-target datalayout = "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-f64:32:64-f80:32-n8:16:32-S128"
-target triple = "i686--linux"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
 
 declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
 

compiler/testdata/func.go

@@ -0,0 +1,12 @@
+package main
+
+func foo(callback func(int)) {
+	callback(3)
+}
+
+func bar() {
+	foo(someFunc)
+}
+
+func someFunc(int) {
+}

compiler/testdata/func.ll

@@ -0,0 +1,47 @@
+; ModuleID = 'func.go'
+source_filename = "func.go"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
+
+%runtime.funcValueWithSignature = type { i32, i8* }
+
+@"reflect/types.funcid:func:{basic:int}{}" = external constant i8
+@"main.someFunc$withSignature" = linkonce_odr constant %runtime.funcValueWithSignature { i32 ptrtoint (void (i32, i8*, i8*)* @main.someFunc to i32), i8* @"reflect/types.funcid:func:{basic:int}{}" }
+
+declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
+
+define hidden void @main.init(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define hidden void @main.foo(i8* %callback.context, i32 %callback.funcptr, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %0 = call i32 @runtime.getFuncPtr(i8* %callback.context, i32 %callback.funcptr, i8* nonnull @"reflect/types.funcid:func:{basic:int}{}", i8* undef, i8* null)
+  %1 = icmp eq i32 %0, 0
+  br i1 %1, label %fpcall.throw, label %fpcall.next
+
+fpcall.throw:                                     ; preds = %entry
+  call void @runtime.nilPanic(i8* undef, i8* null)
+  unreachable
+
+fpcall.next:                                      ; preds = %entry
+  %2 = inttoptr i32 %0 to void (i32, i8*, i8*)*
+  call void %2(i32 3, i8* %callback.context, i8* undef)
+  ret void
+}
+
+declare i32 @runtime.getFuncPtr(i8*, i32, i8* dereferenceable_or_null(1), i8*, i8*)
+
+declare void @runtime.nilPanic(i8*, i8*)
+
+define hidden void @main.bar(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  call void @main.foo(i8* undef, i32 ptrtoint (%runtime.funcValueWithSignature* @"main.someFunc$withSignature" to i32), i8* undef, i8* undef)
+  ret void
+}
+
+define hidden void @main.someFunc(i32 %arg0, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}

compiler/testdata/goroutine-cortex-m-qemu.ll

@@ -0,0 +1,150 @@
+; ModuleID = 'goroutine.go'
+source_filename = "goroutine.go"
+target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
+target triple = "armv7m-none-eabi"
+
+%runtime.channel = type { i32, i32, i8, %runtime.channelBlockedList*, i32, i32, i32, i8* }
+%runtime.channelBlockedList = type { %runtime.channelBlockedList*, %"internal/task.Task"*, %runtime.chanSelectState*, { %runtime.channelBlockedList*, i32, i32 } }
+%"internal/task.Task" = type { %"internal/task.Task"*, i8*, i64, %"internal/task.state" }
+%"internal/task.state" = type { i32, i32* }
+%runtime.chanSelectState = type { %runtime.channel*, i8* }
+
+declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
+
+define hidden void @main.init(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define hidden void @main.regularFunctionGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %stacksize = call i32 @"internal/task.getGoroutineStackSize"(i32 ptrtoint (void (i8*)* @"main.regularFunction$gowrapper" to i32), i8* undef, i8* undef)
+  call void @"internal/task.start"(i32 ptrtoint (void (i8*)* @"main.regularFunction$gowrapper" to i32), i8* nonnull inttoptr (i32 5 to i8*), i32 %stacksize, i8* undef, i8* null)
+  ret void
+}
+
+declare void @main.regularFunction(i32, i8*, i8*)
+
+define linkonce_odr void @"main.regularFunction$gowrapper"(i8* %0) unnamed_addr #0 {
+entry:
+  %unpack.int = ptrtoint i8* %0 to i32
+  call void @main.regularFunction(i32 %unpack.int, i8* undef, i8* undef)
+  ret void
+}
+
+declare i32 @"internal/task.getGoroutineStackSize"(i32, i8*, i8*)
+
+declare void @"internal/task.start"(i32, i8*, i32, i8*, i8*)
+
+define hidden void @main.inlineFunctionGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %stacksize = call i32 @"internal/task.getGoroutineStackSize"(i32 ptrtoint (void (i8*)* @"main.inlineFunctionGoroutine$1$gowrapper" to i32), i8* undef, i8* undef)
+  call void @"internal/task.start"(i32 ptrtoint (void (i8*)* @"main.inlineFunctionGoroutine$1$gowrapper" to i32), i8* nonnull inttoptr (i32 5 to i8*), i32 %stacksize, i8* undef, i8* null)
+  ret void
+}
+
+define hidden void @"main.inlineFunctionGoroutine$1"(i32 %x, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define linkonce_odr void @"main.inlineFunctionGoroutine$1$gowrapper"(i8* %0) unnamed_addr #1 {
+entry:
+  %unpack.int = ptrtoint i8* %0 to i32
+  call void @"main.inlineFunctionGoroutine$1"(i32 %unpack.int, i8* undef, i8* undef)
+  ret void
+}
+
+define hidden void @main.closureFunctionGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %n = call i8* @runtime.alloc(i32 4, i8* undef, i8* null)
+  %0 = bitcast i8* %n to i32*
+  store i32 3, i32* %0, align 4
+  %1 = call i8* @runtime.alloc(i32 8, i8* undef, i8* null)
+  %2 = bitcast i8* %1 to i32*
+  store i32 5, i32* %2, align 4
+  %3 = getelementptr inbounds i8, i8* %1, i32 4
+  %4 = bitcast i8* %3 to i8**
+  store i8* %n, i8** %4, align 4
+  %stacksize = call i32 @"internal/task.getGoroutineStackSize"(i32 ptrtoint (void (i8*)* @"main.closureFunctionGoroutine$1$gowrapper" to i32), i8* undef, i8* undef)
+  call void @"internal/task.start"(i32 ptrtoint (void (i8*)* @"main.closureFunctionGoroutine$1$gowrapper" to i32), i8* nonnull %1, i32 %stacksize, i8* undef, i8* null)
+  %5 = load i32, i32* %0, align 4
+  call void @runtime.printint32(i32 %5, i8* undef, i8* null)
+  ret void
+}
+
+define hidden void @"main.closureFunctionGoroutine$1"(i32 %x, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %unpack.ptr = bitcast i8* %context to i32*
+  store i32 7, i32* %unpack.ptr, align 4
+  ret void
+}
+
+define linkonce_odr void @"main.closureFunctionGoroutine$1$gowrapper"(i8* %0) unnamed_addr #2 {
+entry:
+  %1 = bitcast i8* %0 to i32*
+  %2 = load i32, i32* %1, align 4
+  %3 = getelementptr inbounds i8, i8* %0, i32 4
+  %4 = bitcast i8* %3 to i8**
+  %5 = load i8*, i8** %4, align 4
+  call void @"main.closureFunctionGoroutine$1"(i32 %2, i8* %5, i8* undef)
+  ret void
+}
+
+declare void @runtime.printint32(i32, i8*, i8*)
+
+define hidden void @main.funcGoroutine(i8* %fn.context, void (i32, i8*, i8*)* %fn.funcptr, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %0 = call i8* @runtime.alloc(i32 12, i8* undef, i8* null)
+  %1 = bitcast i8* %0 to i32*
+  store i32 5, i32* %1, align 4
+  %2 = getelementptr inbounds i8, i8* %0, i32 4
+  %3 = bitcast i8* %2 to i8**
+  store i8* %fn.context, i8** %3, align 4
+  %4 = getelementptr inbounds i8, i8* %0, i32 8
+  %5 = bitcast i8* %4 to void (i32, i8*, i8*)**
+  store void (i32, i8*, i8*)* %fn.funcptr, void (i32, i8*, i8*)** %5, align 4
+  %stacksize = call i32 @"internal/task.getGoroutineStackSize"(i32 ptrtoint (void (i8*)* @main.funcGoroutine.gowrapper to i32), i8* undef, i8* undef)
+  call void @"internal/task.start"(i32 ptrtoint (void (i8*)* @main.funcGoroutine.gowrapper to i32), i8* nonnull %0, i32 %stacksize, i8* undef, i8* null)
+  ret void
+}
+
+define linkonce_odr void @main.funcGoroutine.gowrapper(i8* %0) unnamed_addr #3 {
+entry:
+  %1 = bitcast i8* %0 to i32*
+  %2 = load i32, i32* %1, align 4
+  %3 = getelementptr inbounds i8, i8* %0, i32 4
+  %4 = bitcast i8* %3 to i8**
+  %5 = load i8*, i8** %4, align 4
+  %6 = getelementptr inbounds i8, i8* %0, i32 8
+  %7 = bitcast i8* %6 to void (i32, i8*, i8*)**
+  %8 = load void (i32, i8*, i8*)*, void (i32, i8*, i8*)** %7, align 4
+  call void %8(i32 %2, i8* %5, i8* undef)
+  ret void
+}
+
+define hidden void @main.recoverBuiltinGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define hidden void @main.copyBuiltinGoroutine(i8* %dst.data, i32 %dst.len, i32 %dst.cap, i8* %src.data, i32 %src.len, i32 %src.cap, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %copy.n = call i32 @runtime.sliceCopy(i8* %dst.data, i8* %src.data, i32 %dst.len, i32 %src.len, i32 1, i8* undef, i8* null)
+  ret void
+}
+
+declare i32 @runtime.sliceCopy(i8* nocapture writeonly, i8* nocapture readonly, i32, i32, i32, i8*, i8*)
+
+define hidden void @main.closeBuiltinGoroutine(%runtime.channel* dereferenceable_or_null(32) %ch, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  call void @runtime.chanClose(%runtime.channel* %ch, i8* undef, i8* null)
+  ret void
+}
+
+declare void @runtime.chanClose(%runtime.channel* dereferenceable_or_null(32), i8*, i8*)
+
+attributes #0 = { "tinygo-gowrapper"="main.regularFunction" }
+attributes #1 = { "tinygo-gowrapper"="main.inlineFunctionGoroutine$1" }
+attributes #2 = { "tinygo-gowrapper"="main.closureFunctionGoroutine$1" }
+attributes #3 = { "tinygo-gowrapper" }

compiler/testdata/goroutine-wasm.ll

@@ -0,0 +1,106 @@
+; ModuleID = 'goroutine.go'
+source_filename = "goroutine.go"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
+
+%runtime.funcValueWithSignature = type { i32, i8* }
+%runtime.channel = type { i32, i32, i8, %runtime.channelBlockedList*, i32, i32, i32, i8* }
+%runtime.channelBlockedList = type { %runtime.channelBlockedList*, %"internal/task.Task"*, %runtime.chanSelectState*, { %runtime.channelBlockedList*, i32, i32 } }
+%"internal/task.Task" = type { %"internal/task.Task"*, i8*, i64, %"internal/task.state" }
+%"internal/task.state" = type { i8* }
+%runtime.chanSelectState = type { %runtime.channel*, i8* }
+
+@"main.regularFunctionGoroutine$pack" = private unnamed_addr constant { i32, i8* } { i32 5, i8* undef }
+@"main.inlineFunctionGoroutine$pack" = private unnamed_addr constant { i32, i8* } { i32 5, i8* undef }
+@"reflect/types.funcid:func:{basic:int}{}" = external constant i8
+@"main.closureFunctionGoroutine$1$withSignature" = linkonce_odr constant %runtime.funcValueWithSignature { i32 ptrtoint (void (i32, i8*, i8*)* @"main.closureFunctionGoroutine$1" to i32), i8* @"reflect/types.funcid:func:{basic:int}{}" }
+
+declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
+
+define hidden void @main.init(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define hidden void @main.regularFunctionGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  call void @"internal/task.start"(i32 ptrtoint (void (i32, i8*, i8*)* @main.regularFunction to i32), i8* bitcast ({ i32, i8* }* @"main.regularFunctionGoroutine$pack" to i8*), i32 undef, i8* undef, i8* null)
+  ret void
+}
+
+declare void @main.regularFunction(i32, i8*, i8*)
+
+declare void @"internal/task.start"(i32, i8*, i32, i8*, i8*)
+
+define hidden void @main.inlineFunctionGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  call void @"internal/task.start"(i32 ptrtoint (void (i32, i8*, i8*)* @"main.inlineFunctionGoroutine$1" to i32), i8* bitcast ({ i32, i8* }* @"main.inlineFunctionGoroutine$pack" to i8*), i32 undef, i8* undef, i8* null)
+  ret void
+}
+
+define hidden void @"main.inlineFunctionGoroutine$1"(i32 %x, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define hidden void @main.closureFunctionGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %n = call i8* @runtime.alloc(i32 4, i8* undef, i8* null)
+  %0 = bitcast i8* %n to i32*
+  store i32 3, i32* %0, align 4
+  %1 = call i8* @runtime.alloc(i32 8, i8* undef, i8* null)
+  %2 = bitcast i8* %1 to i32*
+  store i32 5, i32* %2, align 4
+  %3 = getelementptr inbounds i8, i8* %1, i32 4
+  %4 = bitcast i8* %3 to i8**
+  store i8* %n, i8** %4, align 4
+  call void @"internal/task.start"(i32 ptrtoint (void (i32, i8*, i8*)* @"main.closureFunctionGoroutine$1" to i32), i8* nonnull %1, i32 undef, i8* undef, i8* null)
+  %5 = load i32, i32* %0, align 4
+  call void @runtime.printint32(i32 %5, i8* undef, i8* null)
+  ret void
+}
+
+define hidden void @"main.closureFunctionGoroutine$1"(i32 %x, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %unpack.ptr = bitcast i8* %context to i32*
+  store i32 7, i32* %unpack.ptr, align 4
+  ret void
+}
+
+declare void @runtime.printint32(i32, i8*, i8*)
+
+define hidden void @main.funcGoroutine(i8* %fn.context, i32 %fn.funcptr, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %0 = call i32 @runtime.getFuncPtr(i8* %fn.context, i32 %fn.funcptr, i8* nonnull @"reflect/types.funcid:func:{basic:int}{}", i8* undef, i8* null)
+  %1 = call i8* @runtime.alloc(i32 8, i8* undef, i8* null)
+  %2 = bitcast i8* %1 to i32*
+  store i32 5, i32* %2, align 4
+  %3 = getelementptr inbounds i8, i8* %1, i32 4
+  %4 = bitcast i8* %3 to i8**
+  store i8* %fn.context, i8** %4, align 4
+  call void @"internal/task.start"(i32 %0, i8* nonnull %1, i32 undef, i8* undef, i8* null)
+  ret void
+}
+
+declare i32 @runtime.getFuncPtr(i8*, i32, i8* dereferenceable_or_null(1), i8*, i8*)
+
+define hidden void @main.recoverBuiltinGoroutine(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define hidden void @main.copyBuiltinGoroutine(i8* %dst.data, i32 %dst.len, i32 %dst.cap, i8* %src.data, i32 %src.len, i32 %src.cap, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  %copy.n = call i32 @runtime.sliceCopy(i8* %dst.data, i8* %src.data, i32 %dst.len, i32 %src.len, i32 1, i8* undef, i8* null)
+  ret void
+}
+
+declare i32 @runtime.sliceCopy(i8* nocapture writeonly, i8* nocapture readonly, i32, i32, i32, i8*, i8*)
+
+define hidden void @main.closeBuiltinGoroutine(%runtime.channel* dereferenceable_or_null(32) %ch, i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  call void @runtime.chanClose(%runtime.channel* %ch, i8* undef, i8* null)
+  ret void
+}
+
+declare void @runtime.chanClose(%runtime.channel* dereferenceable_or_null(32), i8*, i8*)

compiler/testdata/goroutine.go

@@ -0,0 +1,43 @@
+package main
+
+func regularFunctionGoroutine() {
+	go regularFunction(5)
+}
+
+func inlineFunctionGoroutine() {
+	go func(x int) {
+	}(5)
+}
+
+func closureFunctionGoroutine() {
+	n := 3
+	go func(x int) {
+		n = 7
+	}(5)
+	print(n) // note: this is racy (but good enough for this test)
+}
+
+func funcGoroutine(fn func(x int)) {
+	go fn(5)
+}
+
+func recoverBuiltinGoroutine() {
+	// This is a no-op.
+	go recover()
+}
+
+func copyBuiltinGoroutine(dst, src []byte) {
+	// This is not run in a goroutine. While this copy operation can indeed take
+	// some time (if there is a lot of data to copy), there is no race-free way
+	// to make use of the result so it's unlikely applications will make use of
+	// it. And doing it this way should be just within the Go specification.
+	go copy(dst, src)
+}
+
+func closeBuiltinGoroutine(ch chan int) {
+	// This builtin is executed directly, not in a goroutine.
+	// The observed behavior is the same.
+	go close(ch)
+}
+
+func regularFunction(x int)

compiler/testdata/interface.ll

@@ -1,26 +1,27 @@
 ; ModuleID = 'interface.go'
 source_filename = "interface.go"
-target datalayout = "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-f64:32:64-f80:32-n8:16:32-S128"
-target triple = "i686--linux"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
 
-%runtime.typecodeID = type { %runtime.typecodeID*, i32, %runtime.interfaceMethodInfo* }
+%runtime.typecodeID = type { %runtime.typecodeID*, i32, %runtime.interfaceMethodInfo*, %runtime.typecodeID* }
 %runtime.interfaceMethodInfo = type { i8*, i32 }
 %runtime._interface = type { i32, i8* }
 %runtime._string = type { i8*, i32 }
 
-@"reflect/types.type:basic:int" = linkonce_odr constant %runtime.typecodeID zeroinitializer
-@"reflect/types.type:pointer:basic:int" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:basic:int", i32 0, %runtime.interfaceMethodInfo* null }
-@"reflect/types.type:pointer:named:error" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:named:error", i32 0, %runtime.interfaceMethodInfo* null }
-@"reflect/types.type:named:error" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:interface:{Error:func:{}{basic:string}}", i32 0, %runtime.interfaceMethodInfo* null }
-@"reflect/types.type:interface:{Error:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* bitcast ([1 x i8*]* @"reflect/types.interface:interface{Error() string}$interface" to %runtime.typecodeID*), i32 0, %runtime.interfaceMethodInfo* null }
-@"func Error() string" = external constant i8
-@"reflect/types.interface:interface{Error() string}$interface" = linkonce_odr constant [1 x i8*] [i8* @"func Error() string"]
-@"reflect/types.type:pointer:interface:{String:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:interface:{String:func:{}{basic:string}}", i32 0, %runtime.interfaceMethodInfo* null }
-@"reflect/types.type:interface:{String:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* bitcast ([1 x i8*]* @"reflect/types.interface:interface{String() string}$interface" to %runtime.typecodeID*), i32 0, %runtime.interfaceMethodInfo* null }
-@"func String() string" = external constant i8
-@"reflect/types.interface:interface{String() string}$interface" = linkonce_odr constant [1 x i8*] [i8* @"func String() string"]
-@"reflect/types.type:basic:int$id" = external constant i8
-@"error$interface" = linkonce_odr constant [1 x i8*] [i8* @"func Error() string"]
+@"reflect/types.type:basic:int" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* null, i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* @"reflect/types.type:pointer:basic:int" }
+@"reflect/types.type:pointer:basic:int" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:basic:int", i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* null }
+@"reflect/types.type:pointer:named:error" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:named:error", i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* null }
+@"reflect/types.type:named:error" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:interface:{Error:func:{}{basic:string}}", i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* @"reflect/types.type:pointer:named:error" }
+@"reflect/types.type:interface:{Error:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* bitcast ([1 x i8*]* @"reflect/types.interface:interface{Error() string}$interface" to %runtime.typecodeID*), i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* @"reflect/types.type:pointer:interface:{Error:func:{}{basic:string}}" }
+@"reflect/methods.Error() string" = linkonce_odr constant i8 0, align 1
+@"reflect/types.interface:interface{Error() string}$interface" = linkonce_odr constant [1 x i8*] [i8* @"reflect/methods.Error() string"]
+@"reflect/types.type:pointer:interface:{Error:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:interface:{Error:func:{}{basic:string}}", i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* null }
+@"reflect/types.type:pointer:interface:{String:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:interface:{String:func:{}{basic:string}}", i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* null }
+@"reflect/types.type:interface:{String:func:{}{basic:string}}" = linkonce_odr constant %runtime.typecodeID { %runtime.typecodeID* bitcast ([1 x i8*]* @"reflect/types.interface:interface{String() string}$interface" to %runtime.typecodeID*), i32 0, %runtime.interfaceMethodInfo* null, %runtime.typecodeID* @"reflect/types.type:pointer:interface:{String:func:{}{basic:string}}" }
+@"reflect/methods.String() string" = linkonce_odr constant i8 0, align 1
+@"reflect/types.interface:interface{String() string}$interface" = linkonce_odr constant [1 x i8*] [i8* @"reflect/methods.String() string"]
+@"reflect/types.typeid:basic:int" = external constant i8
+@"error$interface" = linkonce_odr constant [1 x i8*] [i8* @"reflect/methods.Error() string"]
 
 declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
 
@@ -51,7 +52,7 @@ entry:
 
 define hidden i1 @main.isInt(i32 %itf.typecode, i8* %itf.value, i8* %context, i8* %parentHandle) unnamed_addr {
 entry:
-  %typecode = call i1 @runtime.typeAssert(i32 %itf.typecode, i8* nonnull @"reflect/types.type:basic:int$id", i8* undef, i8* null)
+  %typecode = call i1 @runtime.typeAssert(i32 %itf.typecode, i8* nonnull @"reflect/types.typeid:basic:int", i8* undef, i8* null)
   br i1 %typecode, label %typeassert.ok, label %typeassert.next
 
 typeassert.ok:                                    ; preds = %entry
@@ -91,7 +92,7 @@ typeassert.next:                                  ; preds = %typeassert.ok, %ent
 
 define hidden %runtime._string @main.callErrorMethod(i32 %itf.typecode, i8* %itf.value, i8* %context, i8* %parentHandle) unnamed_addr {
 entry:
-  %invoke.func = call i32 @runtime.interfaceMethod(i32 %itf.typecode, i8** getelementptr inbounds ([1 x i8*], [1 x i8*]* @"error$interface", i32 0, i32 0), i8* nonnull @"func Error() string", i8* undef, i8* null)
+  %invoke.func = call i32 @runtime.interfaceMethod(i32 %itf.typecode, i8** getelementptr inbounds ([1 x i8*], [1 x i8*]* @"error$interface", i32 0, i32 0), i8* nonnull @"reflect/methods.Error() string", i8* undef, i8* null)
   %invoke.func.cast = inttoptr i32 %invoke.func to %runtime._string (i8*, i8*, i8*)*
   %0 = call %runtime._string %invoke.func.cast(i8* %itf.value, i8* undef, i8* undef)
   ret %runtime._string %0

compiler/testdata/pointer.ll

@@ -1,7 +1,7 @@
 ; ModuleID = 'pointer.go'
 source_filename = "pointer.go"
-target datalayout = "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-f64:32:64-f80:32-n8:16:32-S128"
-target triple = "i686--linux"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
 
 declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
 

compiler/testdata/pragma.go

@@ -0,0 +1,66 @@
+package main
+
+import _ "unsafe"
+
+// Creates an external global with name extern_global.
+//go:extern extern_global
+var externGlobal [0]byte
+
+// Creates a
+//go:align 32
+var alignedGlobal [4]uint32
+
+// Test conflicting pragmas (the last one counts).
+//go:align 64
+//go:align 16
+var alignedGlobal16 [4]uint32
+
+// Test exported functions.
+//export extern_func
+func externFunc() {
+}
+
+// Define a function in a different package using go:linkname.
+//go:linkname withLinkageName1 somepkg.someFunction1
+func withLinkageName1() {
+}
+
+// Import a function from a different package using go:linkname.
+//go:linkname withLinkageName2 somepkg.someFunction2
+func withLinkageName2()
+
+// Function has an 'inline hint', similar to the inline keyword in C.
+//go:inline
+func inlineFunc() {
+}
+
+// Function should never be inlined, equivalent to GCC
+// __attribute__((noinline)).
+//go:noinline
+func noinlineFunc() {
+}
+
+// This function should have the specified section.
+//go:section .special_function_section
+func functionInSection() {
+}
+
+//export exportedFunctionInSection
+//go:section .special_function_section
+func exportedFunctionInSection() {
+}
+
+// This function should not: it's only a declaration and not a definition.
+//go:section .special_function_section
+func undefinedFunctionNotInSection()
+
+//go:section .special_global_section
+var globalInSection uint32
+
+//go:section .special_global_section
+//go:extern undefinedGlobalNotInSection
+var undefinedGlobalNotInSection uint32
+
+//go:align 1024
+//go:section .global_section
+var multipleGlobalPragmas uint32

compiler/testdata/pragma.ll

@@ -0,0 +1,59 @@
+; ModuleID = 'pragma.go'
+source_filename = "pragma.go"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
+
+@extern_global = external global [0 x i8], align 1
+@main.alignedGlobal = hidden global [4 x i32] zeroinitializer, align 32
+@main.alignedGlobal16 = hidden global [4 x i32] zeroinitializer, align 16
+@main.globalInSection = hidden global i32 0, section ".special_global_section", align 4
+@undefinedGlobalNotInSection = external global i32, align 4
+@main.multipleGlobalPragmas = hidden global i32 0, section ".global_section", align 1024
+
+declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
+
+define hidden void @main.init(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+define void @extern_func() #0 {
+entry:
+  ret void
+}
+
+define hidden void @somepkg.someFunction1(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret void
+}
+
+declare void @somepkg.someFunction2(i8*, i8*)
+
+; Function Attrs: inlinehint
+define hidden void @main.inlineFunc(i8* %context, i8* %parentHandle) unnamed_addr #1 {
+entry:
+  ret void
+}
+
+; Function Attrs: noinline
+define hidden void @main.noinlineFunc(i8* %context, i8* %parentHandle) unnamed_addr #2 {
+entry:
+  ret void
+}
+
+define hidden void @main.functionInSection(i8* %context, i8* %parentHandle) unnamed_addr section ".special_function_section" {
+entry:
+  ret void
+}
+
+define void @exportedFunctionInSection() #3 section ".special_function_section" {
+entry:
+  ret void
+}
+
+declare void @main.undefinedFunctionNotInSection(i8*, i8*)
+
+attributes #0 = { "wasm-export-name"="extern_func" }
+attributes #1 = { inlinehint }
+attributes #2 = { noinline }
+attributes #3 = { "wasm-export-name"="exportedFunctionInSection" }

compiler/testdata/slice.ll

@@ -1,7 +1,7 @@
 ; ModuleID = 'slice.go'
 source_filename = "slice.go"
-target datalayout = "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-f64:32:64-f80:32-n8:16:32-S128"
-target triple = "i686--linux"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
 
 declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
 
@@ -60,7 +60,7 @@ entry:
   ret { i32*, i32, i32 } %7
 }
 
-declare { i8*, i32, i32 } @runtime.sliceAppend(i8*, i8*, i32, i32, i32, i32, i8*, i8*)
+declare { i8*, i32, i32 } @runtime.sliceAppend(i8*, i8* nocapture readonly, i32, i32, i32, i32, i8*, i8*)
 
 define hidden { i32*, i32, i32 } @main.sliceAppendSlice(i32* %ints.data, i32 %ints.len, i32 %ints.cap, i32* %added.data, i32 %added.len, i32 %added.cap, i8* %context, i8* %parentHandle) unnamed_addr {
 entry:
@@ -85,4 +85,4 @@ entry:
   ret i32 %copy.n
 }
 
-declare i32 @runtime.sliceCopy(i8*, i8*, i32, i32, i32, i8*, i8*)
+declare i32 @runtime.sliceCopy(i8* nocapture writeonly, i8* nocapture readonly, i32, i32, i32, i8*, i8*)

compiler/testdata/string.go

@@ -1,5 +1,13 @@
 package main
 
+func someString() string {
+	return "foo"
+}
+
+func zeroLengthString() string {
+	return ""
+}
+
 func stringLen(s string) int {
 	return len(s)
 }

compiler/testdata/string.ll

@@ -1,7 +1,11 @@
 ; ModuleID = 'string.go'
 source_filename = "string.go"
-target datalayout = "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-f64:32:64-f80:32-n8:16:32-S128"
-target triple = "i686--linux"
+target datalayout = "e-m:e-p:32:32-i64:64-n32:64-S128"
+target triple = "wasm32--wasi"
+
+%runtime._string = type { i8*, i32 }
+
+@"main.someString$string" = internal unnamed_addr constant [3 x i8] c"foo", align 1
 
 declare noalias nonnull i8* @runtime.alloc(i32, i8*, i8*)
 
@@ -10,6 +14,16 @@ entry:
   ret void
 }
 
+define hidden %runtime._string @main.someString(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret %runtime._string { i8* getelementptr inbounds ([3 x i8], [3 x i8]* @"main.someString$string", i32 0, i32 0), i32 3 }
+}
+
+define hidden %runtime._string @main.zeroLengthString(i8* %context, i8* %parentHandle) unnamed_addr {
+entry:
+  ret %runtime._string zeroinitializer
+}
+
 define hidden i32 @main.stringLen(i8* %s.data, i32 %s.len, i8* %context, i8* %parentHandle) unnamed_addr {
 entry:
   ret i32 %s.len

go.mod

@@ -12,5 +12,5 @@ require (
 	go.bug.st/serial v1.1.2
 	golang.org/x/sys v0.0.0-20210113181707-4bcb84eeeb78
 	golang.org/x/tools v0.0.0-20200216192241-b320d3a0f5a2
-	tinygo.org/x/go-llvm v0.0.0-20210308112806-9ef958b6bed4
+	tinygo.org/x/go-llvm v0.0.0-20210325115028-e7b85195e81c
 )

go.sum

@@ -59,3 +59,5 @@ gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8
 gopkg.in/yaml.v2 v2.2.2/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
 tinygo.org/x/go-llvm v0.0.0-20210308112806-9ef958b6bed4 h1:CMUHxVTb+UuUePuMf8vkWjZ3gTp9BBK91KrgOCwoNHs=
 tinygo.org/x/go-llvm v0.0.0-20210308112806-9ef958b6bed4/go.mod h1:fv1F0BSNpxMfCL0zF3M4OPFbgYHnhtB6ST0HvUtu/LE=
+tinygo.org/x/go-llvm v0.0.0-20210325115028-e7b85195e81c h1:vn9IPshzYmzZis10UEVrsIBRv9FpykADw6M3/tHHROg=
+tinygo.org/x/go-llvm v0.0.0-20210325115028-e7b85195e81c/go.mod h1:fv1F0BSNpxMfCL0zF3M4OPFbgYHnhtB6ST0HvUtu/LE=

goenv/goenv.go

@@ -171,8 +171,9 @@ func getGoroot() string {
 	switch runtime.GOOS {
 	case "linux":
 		candidates = []string{
-			"/usr/local/go", // manually installed
-			"/usr/lib/go",   // from the distribution
+			"/usr/local/go",     // manually installed
+			"/usr/lib/go",       // from the distribution
+			"/snap/go/current/", // installed using snap
 		}
 	case "darwin":
 		candidates = []string{

goenv/version.go

@@ -12,7 +12,7 @@ import (
 
 // Version of TinyGo.
 // Update this value before release of new version of software.
-const Version = "0.18.0-dev"
+const Version = "0.19.0"
 
 // GetGorootVersion returns the major and minor version for a given GOROOT path.
 // If the goroot cannot be determined, (0, 0) is returned.

interp/compiler.go

@@ -25,6 +25,7 @@ type function struct {
 // basicBlock represents a LLVM basic block and contains a slice of
 // instructions. The last instruction must be a terminator instruction.
 type basicBlock struct {
+	phiNodes     []instruction
 	instructions []instruction
 }
 
@@ -135,6 +136,15 @@ func (r *runner) compileFunction(llvmFn llvm.Value) *function {
 				default:
 					panic("unknown number of operands")
 				}
+			case llvm.Switch:
+				// A switch is an array of (value, label) pairs, of which the
+				// first one indicates the to-switch value and the default
+				// label.
+				numOperands := llvmInst.OperandsCount()
+				for i := 0; i < numOperands; i += 2 {
+					inst.operands = append(inst.operands, r.getValue(llvmInst.Operand(i)))
+					inst.operands = append(inst.operands, literalValue{uint32(blockIndices[llvmInst.Operand(i+1)])})
+				}
 			case llvm.PHI:
 				inst.name = llvmInst.Name()
 				incomingCount := inst.llvmInst.IncomingCount()
@@ -337,7 +347,13 @@ func (r *runner) compileFunction(llvmFn llvm.Value) *function {
 				// This error is handled when actually trying to interpret this
 				// instruction (to not trigger on code that won't be executed).
 			}
-			bb.instructions = append(bb.instructions, inst)
+			if inst.opcode == llvm.PHI {
+				// PHI nodes need to be treated specially, see the comment in
+				// interpreter.go for an explanation.
+				bb.phiNodes = append(bb.phiNodes, inst)
+			} else {
+				bb.instructions = append(bb.instructions, inst)
+			}
 		}
 	}
 	return fn

interp/interp.go

@@ -129,7 +129,7 @@ func Run(mod llvm.Module, debug bool) error {
 
 	// Update all global variables in the LLVM module.
 	mem := memoryView{r: r}
-	for _, obj := range r.objects {
+	for i, obj := range r.objects {
 		if obj.llvmGlobal.IsNil() {
 			continue
 		}
@@ -159,6 +159,12 @@ func Run(mod llvm.Module, debug bool) error {
 			name := obj.llvmGlobal.Name()
 			obj.llvmGlobal.EraseFromParentAsGlobal()
 			newGlobal.SetName(name)
+
+			// Update interp-internal references.
+			delete(r.globals, obj.llvmGlobal)
+			obj.llvmGlobal = newGlobal
+			r.globals[newGlobal] = i
+			r.objects[i] = obj
 			continue
 		}
 		if err != nil {

interp/interp_test.go

@@ -13,9 +13,9 @@ import (
 func TestInterp(t *testing.T) {
 	for _, name := range []string{
 		"basic",
+		"phi",
 		"slice-copy",
 		"consteval",
-		"map",
 		"interface",
 	} {
 		name := name // make tc local to this closure

interp/interpreter.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"math"
 	"os"
+	"strconv"
 	"strings"
 	"time"
 
@@ -33,6 +34,50 @@ func (r *runner) run(fn *function, params []value, parentMem *memoryView, indent
 	lastBB := -1 // last basic block is undefined, only defined after a branch
 	var operands []value
 	for instIndex := 0; instIndex < len(bb.instructions); instIndex++ {
+		if instIndex == 0 {
+			// This is the start of a new basic block.
+			// There may be PHI nodes that need to be resolved. Resolve all PHI
+			// nodes before continuing with regular instructions.
+			// PHI nodes need to be treated specially because they can have a
+			// mutual dependency:
+			//   for.loop:
+			//     %a = phi i8 [ 1, %entry ], [ %b, %for.loop ]
+			//     %b = phi i8 [ 3, %entry ], [ %a, %for.loop ]
+			// If these PHI nodes are processed like a regular instruction, %a
+			// and %b are both 3 on the second iteration of the loop because %b
+			// loads the value of %a from the second iteration, while it should
+			// load the value from the previous iteration. The correct behavior
+			// is that these two values swap each others place on each
+			// iteration.
+			var phiValues []value
+			var phiIndices []int
+			for _, inst := range bb.phiNodes {
+				var result value
+				for i := 0; i < len(inst.operands); i += 2 {
+					if int(inst.operands[i].(literalValue).value.(uint32)) == lastBB {
+						incoming := inst.operands[i+1]
+						if local, ok := incoming.(localValue); ok {
+							result = locals[fn.locals[local.value]]
+						} else {
+							result = incoming
+						}
+						break
+					}
+				}
+				if r.debug {
+					fmt.Fprintln(os.Stderr, indent+"phi", inst.operands, "->", result)
+				}
+				if result == nil {
+					panic("could not find PHI input")
+				}
+				phiValues = append(phiValues, result)
+				phiIndices = append(phiIndices, inst.localIndex)
+			}
+			for i, value := range phiValues {
+				locals[phiIndices[i]] = value
+			}
+		}
+
 		inst := bb.instructions[instIndex]
 		operands = operands[:0]
 		isRuntimeInst := false
@@ -103,27 +148,24 @@ func (r *runner) run(fn *function, params []value, parentMem *memoryView, indent
 			default:
 				panic("unknown operands length")
 			}
-			break // continue with next block
-		case llvm.PHI:
-			var result value
-			for i := 0; i < len(inst.operands); i += 2 {
-				if int(inst.operands[i].(literalValue).value.(uint32)) == lastBB {
-					incoming := inst.operands[i+1]
-					if local, ok := incoming.(localValue); ok {
-						result = locals[fn.locals[local.value]]
-					} else {
-						result = incoming
-					}
+		case llvm.Switch:
+			// Switch statement: [value, defaultLabel, case0, label0, case1, label1, ...]
+			value := operands[0].Uint()
+			targetLabel := operands[1].Uint() // default label
+			// Do a lazy switch by iterating over all cases.
+			for i := 2; i < len(operands); i += 2 {
+				if value == operands[i].Uint() {
+					targetLabel = operands[i+1].Uint()
 					break
 				}
 			}
+			lastBB = currentBB
+			currentBB = int(targetLabel)
+			bb = fn.blocks[currentBB]
+			instIndex = -1 // start at 0 the next cycle
 			if r.debug {
-				fmt.Fprintln(os.Stderr, indent+"phi", inst.operands, "->", result)
+				fmt.Fprintln(os.Stderr, indent+"switch", operands, "->", currentBB)
 			}
-			if result == nil {
-				panic("could not find PHI input")
-			}
-			locals[inst.localIndex] = result
 		case llvm.Select:
 			// Select is much like a ternary operator: it picks a result from
 			// the second and third operand based on the boolean first operand.
@@ -155,7 +197,7 @@ func (r *runner) run(fn *function, params []value, parentMem *memoryView, indent
 				// which case this call won't even get to this point but will
 				// already be emitted in initAll.
 				continue
-			case strings.HasPrefix(callFn.name, "runtime.print") || callFn.name == "runtime._panic" || callFn.name == "runtime.hashmapGet":
+			case strings.HasPrefix(callFn.name, "runtime.print") || callFn.name == "runtime._panic" || callFn.name == "runtime.hashmapGet" || callFn.name == "os.runtime_args":
 				// These functions should be run at runtime. Specifically:
 				//   * Print and panic functions are best emitted directly without
 				//     interpreting them, otherwise we get a ton of putchar (etc.)
@@ -163,6 +205,9 @@ func (r *runner) run(fn *function, params []value, parentMem *memoryView, indent
 				//   * runtime.hashmapGet tries to access the map value directly.
 				//     This is not possible as the map value is treated as a special
 				//     kind of object in this package.
+				//   * os.runtime_args reads globals that are initialized outside
+				//     the view of the interp package so it always needs to be run
+				//     at runtime.
 				err := r.runAtRuntime(fn, inst, locals, &mem, indent)
 				if err != nil {
 					return nil, mem, err
@@ -328,7 +373,7 @@ func (r *runner) run(fn *function, params []value, parentMem *memoryView, indent
 					return nil, mem, r.errorAt(inst, err)
 				}
 				actualType := actualTypePtrToInt.Operand(0)
-				if actualType.Name()+"$id" == assertedType.Name() {
+				if strings.TrimPrefix(actualType.Name(), "reflect/types.type:") == strings.TrimPrefix(assertedType.Name(), "reflect/types.typeid:") {
 					locals[inst.localIndex] = literalValue{uint8(1)}
 				} else {
 					locals[inst.localIndex] = literalValue{uint8(0)}
@@ -415,74 +460,6 @@ func (r *runner) run(fn *function, params []value, parentMem *memoryView, indent
 					return nil, mem, r.errorAt(inst, errors.New("could not find method: "+signature.Name()))
 				}
 				locals[inst.localIndex] = r.getValue(method)
-			case callFn.name == "runtime.hashmapMake":
-				// Create a new map.
-				hashmapPointerType := inst.llvmInst.Type()
-				keySize := uint32(operands[1].Uint())
-				valueSize := uint32(operands[2].Uint())
-				m := newMapValue(r, hashmapPointerType, keySize, valueSize)
-				alloc := object{
-					llvmType:   hashmapPointerType,
-					globalName: r.pkgName + "$map",
-					buffer:     m,
-					size:       m.len(r),
-				}
-				index := len(r.objects)
-				r.objects = append(r.objects, alloc)
-
-				// Create a pointer to this map. Maps are reference types, so
-				// are implemented as pointers.
-				ptr := newPointerValue(r, index, 0)
-				if r.debug {
-					fmt.Fprintln(os.Stderr, indent+"runtime.hashmapMake:", keySize, valueSize, "->", ptr)
-				}
-				locals[inst.localIndex] = ptr
-			case callFn.name == "runtime.hashmapBinarySet":
-				// Do a mapassign operation with a binary key (that is, without
-				// a string key).
-				if r.debug {
-					fmt.Fprintln(os.Stderr, indent+"runtime.hashmapBinarySet:", operands[1:])
-				}
-				mapPtr, err := operands[1].asPointer(r)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-				m := mem.getWritable(mapPtr.index()).buffer.(*mapValue)
-				keyPtr, err := operands[2].asPointer(r)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-				valuePtr, err := operands[3].asPointer(r)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-				err = m.putBinary(&mem, keyPtr, valuePtr)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-			case callFn.name == "runtime.hashmapStringSet":
-				// Do a mapassign operation with a string key.
-				if r.debug {
-					fmt.Fprintln(os.Stderr, indent+"runtime.hashmapBinarySet:", operands[1:])
-				}
-				mapPtr, err := operands[1].asPointer(r)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-				m := mem.getWritable(mapPtr.index()).buffer.(*mapValue)
-				stringPtr, err := operands[2].asPointer(r)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-				stringLen := operands[3].Uint()
-				valuePtr, err := operands[4].asPointer(r)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
-				err = m.putString(&mem, stringPtr, stringLen, valuePtr)
-				if err != nil {
-					return nil, mem, r.errorAt(inst, err)
-				}
 			default:
 				if len(callFn.blocks) == 0 {
 					// Call to a function declaration without a definition
@@ -954,16 +931,31 @@ func (r *runner) runAtRuntime(fn *function, inst instruction, locals []value, me
 		result = r.builder.CreateBitCast(operands[0], inst.llvmInst.Type(), inst.name)
 	case llvm.ExtractValue:
 		indices := inst.llvmInst.Indices()
-		if len(indices) != 1 {
-			panic("expected exactly one index")
+		// Note: the Go LLVM API doesn't support multiple indices, so simulate
+		// this operation with some extra extractvalue instructions. Hopefully
+		// this is optimized to a single instruction.
+		agg := operands[0]
+		for i := 0; i < len(indices)-1; i++ {
+			agg = r.builder.CreateExtractValue(agg, int(indices[i]), inst.name+".agg")
 		}
-		result = r.builder.CreateExtractValue(operands[0], int(indices[0]), inst.name)
+		result = r.builder.CreateExtractValue(agg, int(indices[len(indices)-1]), inst.name)
 	case llvm.InsertValue:
 		indices := inst.llvmInst.Indices()
-		if len(indices) != 1 {
-			panic("expected exactly one index")
+		// Similar to extractvalue, we're working around a limitation in the Go
+		// LLVM API here by splitting the insertvalue into multiple instructions
+		// if there is more than one operand.
+		agg := operands[0]
+		aggregates := []llvm.Value{agg}
+		for i := 0; i < len(indices)-1; i++ {
+			agg = r.builder.CreateExtractValue(agg, int(indices[i]), inst.name+".agg"+strconv.Itoa(i))
+			aggregates = append(aggregates, agg)
 		}
-		result = r.builder.CreateInsertValue(operands[0], operands[1], int(indices[0]), inst.name)
+		result = operands[1]
+		for i := len(indices) - 1; i >= 0; i-- {
+			agg := aggregates[i]
+			result = r.builder.CreateInsertValue(agg, result, int(indices[i]), inst.name+".insertvalue"+strconv.Itoa(i))
+		}
+
 	case llvm.Add:
 		result = r.builder.CreateAdd(operands[0], operands[1], inst.name)
 	case llvm.Sub:

interp/memory.go

@@ -178,6 +178,9 @@ func (mv *memoryView) markExternal(llvmValue llvm.Value, mark uint8) {
 		default:
 			panic("interp: unknown constant expression")
 		}
+	} else if !llvmValue.IsAInlineAsm().IsNil() {
+		// Inline assembly can modify globals but only exported globals. Let's
+		// hope the author knows what they're doing.
 	} else {
 		llvmType := llvmValue.Type()
 		switch llvmType.TypeKind() {
@@ -640,288 +643,6 @@ func (v pointerValue) toLLVMValue(llvmType llvm.Type, mem *memoryView) (llvm.Val
 	return gep, nil
 }
 
-// mapValue implements a Go map which is created at compile time and stored as a
-// global variable.
-// The value itself is only used as part of an object (object.buffer). Maps are
-// reference types aka pointers, so it can only be used as a pointerValue, not
-// directly.
-type mapValue struct {
-	r           *runner
-	pkgName     string
-	size        uint32 // byte size of runtime.hashmap
-	hashmap     llvm.Value
-	keyIsString bool
-	keys        []interface{} // either rawValue (for binary key) or mapStringKey (for string key)
-	values      []rawValue
-	keySize     uint32
-	valueSize   uint32
-}
-
-type mapStringKey struct {
-	buf  pointerValue
-	size uint64
-	data []uint64
-}
-
-func newMapValue(r *runner, hashmapPointerType llvm.Type, keySize, valueSize uint32) *mapValue {
-	size := uint32(r.targetData.TypeAllocSize(hashmapPointerType.ElementType()))
-	return &mapValue{
-		r:         r,
-		pkgName:   r.pkgName,
-		size:      size,
-		keySize:   keySize,
-		valueSize: valueSize,
-	}
-}
-
-func (v *mapValue) len(r *runner) uint32 {
-	return v.size
-}
-
-func (v *mapValue) clone() value {
-	// Return a copy of mapValue.
-	clone := *v
-	clone.keys = append([]interface{}{}, clone.keys...)
-	clone.values = append([]rawValue{}, clone.values...)
-	return &clone
-}
-
-func (v *mapValue) asPointer(r *runner) (pointerValue, error) {
-	panic("interp: mapValue.asPointer")
-}
-
-func (v *mapValue) asRawValue(r *runner) rawValue {
-	panic("interp: mapValue.asRawValue")
-}
-
-func (v *mapValue) Uint() uint64 {
-	panic("interp: mapValue.Uint")
-}
-
-func (v *mapValue) Int() int64 {
-	panic("interp: mapValue.Int")
-}
-
-// Temporary struct to collect data before turning this into a hashmap bucket
-// LLVM value.
-type mapBucket struct {
-	m       *mapValue
-	tophash [8]uint8
-	keys    []rawValue // can have up to 8 keys
-	values  []rawValue // can have up to 8 values, len(keys) == len(values)
-}
-
-// create returns a (pointer to a) buffer structurally equivalent to
-// runtime.hashmapBucket.
-func (b *mapBucket) create(ctx llvm.Context, nextBucket llvm.Value, mem *memoryView) llvm.Value {
-	// Create tophash array.
-	int8Type := ctx.Int8Type()
-	tophashValues := make([]llvm.Value, 8)
-	for i := range tophashValues {
-		tophashValues[i] = llvm.ConstInt(int8Type, uint64(b.tophash[i]), false)
-	}
-	tophash := llvm.ConstArray(int8Type, tophashValues)
-
-	// Create next pointer (if not set).
-	if nextBucket.IsNil() {
-		nextBucket = llvm.ConstNull(llvm.PointerType(int8Type, 0))
-	}
-
-	// Create data for keys.
-	var keyValues []llvm.Value
-	for _, key := range b.keys {
-		keyValue, err := key.rawLLVMValue(mem)
-		if err != nil {
-			panic(err)
-		}
-		keyValues = append(keyValues, keyValue)
-	}
-	if len(b.keys) < 8 {
-		keyValues = append(keyValues, llvm.ConstNull(llvm.ArrayType(int8Type, int(b.m.keySize)*(8-len(b.keys)))))
-	}
-	keyValue := ctx.ConstStruct(keyValues, false)
-	if checks && uint32(b.m.r.targetData.TypeAllocSize(keyValue.Type())) != b.m.keySize*8 {
-		panic("key size invalid")
-	}
-
-	// Create data for values.
-	var valueValues []llvm.Value
-	for _, value := range b.values {
-		v, err := value.rawLLVMValue(mem)
-		if err != nil {
-			panic(err)
-		}
-		valueValues = append(valueValues, v)
-	}
-	if len(b.values) < 8 {
-		valueValues = append(valueValues, llvm.ConstNull(llvm.ArrayType(int8Type, int(b.m.valueSize)*(8-len(b.values)))))
-	}
-	valueValue := ctx.ConstStruct(valueValues, false)
-	if checks && uint32(b.m.r.targetData.TypeAllocSize(valueValue.Type())) != b.m.valueSize*8 {
-		panic("value size invalid")
-	}
-
-	// Create the bucket.
-	bucketInitializer := ctx.ConstStruct([]llvm.Value{
-		tophash,
-		nextBucket,
-		keyValue,
-		valueValue,
-	}, false)
-	bucket := llvm.AddGlobal(b.m.r.mod, bucketInitializer.Type(), b.m.pkgName+"$mapbucket")
-	bucket.SetInitializer(bucketInitializer)
-	bucket.SetLinkage(llvm.InternalLinkage)
-	bucket.SetUnnamedAddr(true)
-	return bucket
-}
-
-func (v *mapValue) toLLVMValue(hashmapType llvm.Type, mem *memoryView) (llvm.Value, error) {
-	if !v.hashmap.IsNil() {
-		return v.hashmap, nil
-	}
-
-	// Create a slice of buckets with all the keys and values in the hashmap.
-	var buckets []*mapBucket
-	var bucket *mapBucket
-	for i, key := range v.keys {
-		var data []uint64
-		var keyValue rawValue
-		switch key := key.(type) {
-		case mapStringKey:
-			data = key.data
-			keyValue = newRawValue(v.keySize)
-			// runtime._string is {ptr, length}
-			for i := uint32(0); i < v.keySize/2; i++ {
-				keyValue.buf[i] = key.buf.pointer
-			}
-			copy(keyValue.buf[v.keySize/2:], literalValue{key.size}.asRawValue(v.r).buf)
-		case rawValue:
-			if key.hasPointer() {
-				return llvm.Value{}, errors.New("interp: todo: map key with pointer")
-			}
-			data = key.buf
-			keyValue = key
-		default:
-			return llvm.Value{}, errors.New("interp: unknown map key type")
-		}
-		buf := make([]byte, len(data))
-		for i, p := range data {
-			buf[i] = byte(p)
-		}
-		hash := v.hash(buf)
-
-		if i%8 == 0 {
-			bucket = &mapBucket{m: v}
-			buckets = append(buckets, bucket)
-		}
-		bucket.tophash[i%8] = v.topHash(hash)
-		bucket.keys = append(bucket.keys, keyValue)
-		bucket.values = append(bucket.values, v.values[i])
-	}
-
-	// Convert these buckets into LLVM global variables.
-	ctx := v.r.mod.Context()
-	var nextBucket llvm.Value
-	for i := len(buckets) - 1; i >= 0; i-- {
-		bucket = buckets[i]
-		bucketValue := bucket.create(ctx, nextBucket, mem)
-		nextBucket = bucketValue
-	}
-	firstBucket := nextBucket
-	if firstBucket.IsNil() {
-		firstBucket = llvm.ConstNull(mem.r.i8ptrType)
-	} else {
-		firstBucket = llvm.ConstBitCast(firstBucket, mem.r.i8ptrType)
-	}
-
-	// Create the hashmap itself, pointing to these buckets.
-	hashmapPointerType := llvm.PointerType(hashmapType, 0)
-	hashmap := llvm.ConstNamedStruct(hashmapType, []llvm.Value{
-		llvm.ConstPointerNull(hashmapPointerType), // next
-		firstBucket, // buckets
-		llvm.ConstInt(hashmapType.StructElementTypes()[2], uint64(len(v.keys)), false), // count
-		llvm.ConstInt(ctx.Int8Type(), uint64(v.keySize), false),                        // keySize
-		llvm.ConstInt(ctx.Int8Type(), uint64(v.valueSize), false),                      // valueSize
-		llvm.ConstInt(ctx.Int8Type(), 0, false),                                        // bucketBits
-	})
-
-	v.hashmap = hashmap
-	return v.hashmap, nil
-}
-
-// putString does a map assign operation, assuming that the map is of type
-// map[string]T.
-func (v *mapValue) putString(mem *memoryView, stringBuf pointerValue, stringLen uint64, valuePtr pointerValue) error {
-	if !v.hashmap.IsNil() {
-		return errMapAlreadyCreated
-	}
-
-	value := mem.load(valuePtr, v.valueSize)
-	stringValue := mem.load(stringBuf, uint32(stringLen)).asRawValue(v.r)
-	if stringValue.hasPointer() {
-		panic("interp: string contains pointer")
-	}
-
-	// TODO: avoid duplicate keys
-	v.keys = append(v.keys, mapStringKey{stringBuf, stringLen, stringValue.buf})
-	v.values = append(v.values, value.asRawValue(v.r))
-	v.keyIsString = true
-
-	return nil
-}
-
-// putBinary does a map assign operation for binary data (e.g. [3]int etc). The
-// key must not contain pointer values.
-func (v *mapValue) putBinary(mem *memoryView, keyPtr, valuePtr pointerValue) error {
-	if !v.hashmap.IsNil() {
-		return errMapAlreadyCreated
-	}
-
-	key := mem.load(keyPtr, v.keySize)
-	value := mem.load(valuePtr, v.valueSize)
-
-	// Sanity checks.
-	if v.keySize != key.len(mem.r) || v.valueSize != value.len(mem.r) {
-		// This is a bug (not unhandled input), so panic.
-		panic("interp: key or value size mismatch")
-	}
-	if v.keyIsString {
-		panic("cannot put binary keys in string map")
-	}
-
-	// TODO: avoid duplicate keys
-	v.keys = append(v.keys, key.asRawValue(v.r))
-	v.values = append(v.values, value.asRawValue(v.r))
-
-	return nil
-}
-
-// Get FNV-1a hash of this string.
-//
-// https://en.wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function#FNV-1a_hash
-func (v *mapValue) hash(data []byte) uint32 {
-	var result uint32 = 2166136261 // FNV offset basis
-	for _, c := range data {
-		result ^= uint32(c)
-		result *= 16777619 // FNV prime
-	}
-	return result
-}
-
-// Get the topmost 8 bits of the hash, without using a special value (like 0).
-func (v *mapValue) topHash(hash uint32) uint8 {
-	tophash := uint8(hash >> 24)
-	if tophash < 1 {
-		// 0 means empty slot, so make it bigger.
-		tophash++
-	}
-	return tophash
-}
-
-func (v *mapValue) String() string {
-	return "<map keySize=" + strconv.Itoa(int(v.keySize)) + " valueSize=" + strconv.Itoa(int(v.valueSize)) + ">"
-}
-
 // rawValue is a raw memory buffer that can store either pointers or regular
 // data. This is the fallback data for everything that isn't clearly a
 // literalValue or pointerValue.

interp/testdata/basic.ll

@@ -8,6 +8,7 @@ target triple = "x86_64--linux"
 @main.exportedValue = global [1 x i16*] [i16* @main.exposedValue1]
 @main.exposedValue1 = global i16 0
 @main.exposedValue2 = global i16 0
+@main.insertedValue = global {i8, i32, {float, {i64, i16}}} zeroinitializer
 
 declare void @runtime.printint64(i64) unnamed_addr
 
@@ -65,6 +66,22 @@ entry:
   call void @modifyExternal(i32* bitcast (void ()* @willModifyGlobal to i32*))
   store i16 7, i16* @main.exposedValue2
 
+  ; Test that inline assembly is ignored.
+  call void @modifyExternal(i32* bitcast (void ()* @hasInlineAsm to i32*))
+
+  ; Test switch statement.
+  %switch1 = call i64 @testSwitch(i64 1) ; 1 returns 6
+  %switch2 = call i64 @testSwitch(i64 9) ; 9 returns the default value -1
+  call void @runtime.printint64(i64 %switch1)
+  call void @runtime.printint64(i64 %switch2)
+
+  ; Test extractvalue/insertvalue with multiple operands.
+  %agg = call {i8, i32, {float, {i64, i16}}} @nestedStruct()
+  %elt = extractvalue {i8, i32, {float, {i64, i16}}} %agg, 2, 1, 0
+  call void @runtime.printint64(i64 %elt)
+  %agg2 = insertvalue {i8, i32, {float, {i64, i16}}} %agg, i64 5, 2, 1, 0
+  store {i8, i32, {float, {i64, i16}}} %agg2, {i8, i32, {float, {i64, i16}}}* @main.insertedValue
+
   ret void
 }
 
@@ -87,3 +104,33 @@ entry:
   store i16 8, i16* @main.exposedValue2
   ret void
 }
+
+; Inline assembly should be ignored in the interp package. While it is possible
+; to modify other globals that way, usually that's not the case and there is no
+; real way to check.
+define void @hasInlineAsm() {
+entry:
+  call void asm sideeffect "", ""()
+  ret void
+}
+
+define i64 @testSwitch(i64 %val) {
+entry:
+  ; Test switch statement.
+  switch i64 %val, label %otherwise [ i64 0, label %zero
+                                      i64 1, label %one
+                                      i64 2, label %two ]
+zero:
+  ret i64 5
+
+one:
+  ret i64 6
+
+two:
+  ret i64 7
+
+otherwise:
+  ret i64 -1
+}
+
+declare {i8, i32, {float, {i64, i16}}} @nestedStruct()

interp/testdata/basic.out.ll

@@ -7,6 +7,7 @@ target triple = "x86_64--linux"
 @main.exportedValue = global [1 x i16*] [i16* @main.exposedValue1]
 @main.exposedValue1 = global i16 0
 @main.exposedValue2 = local_unnamed_addr global i16 0
+@main.insertedValue = local_unnamed_addr global { i8, i32, { float, { i64, i16 } } } zeroinitializer
 
 declare void @runtime.printint64(i64) unnamed_addr
 
@@ -25,6 +26,20 @@ entry:
   store i16 5, i16* @main.exposedValue1
   call void @modifyExternal(i32* bitcast (void ()* @willModifyGlobal to i32*))
   store i16 7, i16* @main.exposedValue2
+  call void @modifyExternal(i32* bitcast (void ()* @hasInlineAsm to i32*))
+  call void @runtime.printint64(i64 6)
+  call void @runtime.printint64(i64 -1)
+  %agg = call { i8, i32, { float, { i64, i16 } } } @nestedStruct()
+  %elt.agg = extractvalue { i8, i32, { float, { i64, i16 } } } %agg, 2
+  %elt.agg1 = extractvalue { float, { i64, i16 } } %elt.agg, 1
+  %elt = extractvalue { i64, i16 } %elt.agg1, 0
+  call void @runtime.printint64(i64 %elt)
+  %agg2.agg0 = extractvalue { i8, i32, { float, { i64, i16 } } } %agg, 2
+  %agg2.agg1 = extractvalue { float, { i64, i16 } } %agg2.agg0, 1
+  %agg2.insertvalue2 = insertvalue { i64, i16 } %agg2.agg1, i64 5, 0
+  %agg2.insertvalue1 = insertvalue { float, { i64, i16 } } %agg2.agg0, { i64, i16 } %agg2.insertvalue2, 1
+  %agg2.insertvalue0 = insertvalue { i8, i32, { float, { i64, i16 } } } %agg, { float, { i64, i16 } } %agg2.insertvalue1, 2
+  store { i8, i32, { float, { i64, i16 } } } %agg2.insertvalue0, { i8, i32, { float, { i64, i16 } } }* @main.insertedValue
   ret void
 }
 
@@ -44,3 +59,32 @@ entry:
   store i16 8, i16* @main.exposedValue2
   ret void
 }
+
+define void @hasInlineAsm() {
+entry:
+  call void asm sideeffect "", ""()
+  ret void
+}
+
+define i64 @testSwitch(i64 %val) local_unnamed_addr {
+entry:
+  switch i64 %val, label %otherwise [
+    i64 0, label %zero
+    i64 1, label %one
+    i64 2, label %two
+  ]
+
+zero:                                             ; preds = %entry
+  ret i64 5
+
+one:                                              ; preds = %entry
+  ret i64 6
+
+two:                                              ; preds = %entry
+  ret i64 7
+
+otherwise:                                        ; preds = %entry
+  ret i64 -1
+}
+
+declare { i8, i32, { float, { i64, i16 } } } @nestedStruct() local_unnamed_addr

interp/testdata/interface.ll

@@ -6,7 +6,7 @@ target triple = "x86_64--linux"
 
 @main.v1 = global i1 0
 @"reflect/types.type:named:main.foo" = private constant %runtime.typecodeID { %runtime.typecodeID* @"reflect/types.type:basic:int", i64 0, %runtime.interfaceMethodInfo* null }
-@"reflect/types.type:named:main.foo$id" = external constant i8
+@"reflect/types.typeid:named:main.foo" = external constant i8
 @"reflect/types.type:basic:int" = external constant %runtime.typecodeID
 
 
@@ -21,7 +21,7 @@ entry:
 define internal void @main.init() unnamed_addr {
 entry:
   ; Test type asserts.
-  %typecode = call i1 @runtime.typeAssert(i64 ptrtoint (%runtime.typecodeID* @"reflect/types.type:named:main.foo" to i64), i8* @"reflect/types.type:named:main.foo$id", i8* undef, i8* null)
+  %typecode = call i1 @runtime.typeAssert(i64 ptrtoint (%runtime.typecodeID* @"reflect/types.type:named:main.foo" to i64), i8* @"reflect/types.typeid:named:main.foo", i8* undef, i8* null)
   store i1 %typecode, i1* @main.v1
   ret void
 }

interp/testdata/map.ll

@@ -1,74 +0,0 @@
-target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
-target triple = "armv6m-none-eabi"
-
-%runtime._string = type { i8*, i32 }
-%runtime.hashmap = type { %runtime.hashmap*, i8*, i32, i8, i8, i8 }
-
-@main.m = global %runtime.hashmap* null
-@main.binaryMap = global %runtime.hashmap* null
-@main.stringMap = global %runtime.hashmap* null
-@main.init.string = internal unnamed_addr constant [7 x i8] c"CONNECT"
-
-declare %runtime.hashmap* @runtime.hashmapMake(i8, i8, i32, i8* %context, i8* %parentHandle)
-declare void @runtime.hashmapBinarySet(%runtime.hashmap*, i8*, i8*, i8* %context, i8* %parentHandle)
-declare void @runtime.hashmapStringSet(%runtime.hashmap*, i8*, i32, i8*, i8* %context, i8* %parentHandle)
-declare void @llvm.lifetime.end.p0i8(i64, i8*)
-declare void @llvm.lifetime.start.p0i8(i64, i8*)
-
-define void @runtime.initAll() unnamed_addr {
-entry:
-  call void @main.init(i8* undef, i8* null)
-  ret void
-}
-
-define internal void @main.init(i8* %context, i8* %parentHandle) unnamed_addr {
-entry:
-; Test that hashmap optimizations generally work (even with lifetimes).
-  %hashmap.key = alloca i8
-  %hashmap.value = alloca %runtime._string
-  %0 = call %runtime.hashmap* @runtime.hashmapMake(i8 1, i8 8, i32 1, i8* undef, i8* null)
-  %hashmap.value.bitcast = bitcast %runtime._string* %hashmap.value to i8*
-  call void @llvm.lifetime.start.p0i8(i64 8, i8* %hashmap.value.bitcast)
-  store %runtime._string { i8* getelementptr inbounds ([7 x i8], [7 x i8]* @main.init.string, i32 0, i32 0), i32 7 }, %runtime._string* %hashmap.value
-  call void @llvm.lifetime.start.p0i8(i64 1, i8* %hashmap.key)
-  store i8 1, i8* %hashmap.key
-  call void @runtime.hashmapBinarySet(%runtime.hashmap* %0, i8* %hashmap.key, i8* %hashmap.value.bitcast, i8* undef, i8* null)
-  call void @llvm.lifetime.end.p0i8(i64 1, i8* %hashmap.key)
-  call void @llvm.lifetime.end.p0i8(i64 8, i8* %hashmap.value.bitcast)
-  store %runtime.hashmap* %0, %runtime.hashmap** @main.m
-
-  ; Other tests, that can be done in a separate function.
-  call void @main.testNonConstantBinarySet()
-  call void @main.testNonConstantStringSet()
-  ret void
-}
-
-; Test that a map loaded from a global can still be used for mapassign
-; operations (with binary keys).
-define internal void @main.testNonConstantBinarySet() {
-  %hashmap.key = alloca i8
-  %hashmap.value = alloca i8
-  ; Create hashmap from global.
-  %map.new = call %runtime.hashmap* @runtime.hashmapMake(i8 1, i8 1, i32 1, i8* undef, i8* null)
-  store %runtime.hashmap* %map.new, %runtime.hashmap** @main.binaryMap
-  %map = load %runtime.hashmap*, %runtime.hashmap** @main.binaryMap
-  ; Do the binary set to the newly loaded map.
-  store i8 1, i8* %hashmap.key
-  store i8 2, i8* %hashmap.value
-  call void @runtime.hashmapBinarySet(%runtime.hashmap* %map, i8* %hashmap.key, i8* %hashmap.value, i8* undef, i8* null)
-  ret void
-}
-
-; Test that a map loaded from a global can still be used for mapassign
-; operations (with string keys).
-define internal void @main.testNonConstantStringSet() {
-  %hashmap.value = alloca i8
-  ; Create hashmap from global.
-  %map.new = call %runtime.hashmap* @runtime.hashmapMake(i8 8, i8 1, i32 1, i8* undef, i8* null)
-  store %runtime.hashmap* %map.new, %runtime.hashmap** @main.stringMap
-  %map = load %runtime.hashmap*, %runtime.hashmap** @main.stringMap
-  ; Do the string set to the newly loaded map.
-  store i8 2, i8* %hashmap.value
-  call void @runtime.hashmapStringSet(%runtime.hashmap* %map, i8* getelementptr inbounds ([7 x i8], [7 x i8]* @main.init.string, i32 0, i32 0), i32 7, i8* %hashmap.value, i8* undef, i8* null)
-  ret void
-}

interp/testdata/map.out.ll

@@ -1,20 +0,0 @@
-target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
-target triple = "armv6m-none-eabi"
-
-%runtime.hashmap = type { %runtime.hashmap*, i8*, i32, i8, i8, i8 }
-
-@main.m = local_unnamed_addr global %runtime.hashmap* @"main$map"
-@main.binaryMap = local_unnamed_addr global %runtime.hashmap* @"main$map.1"
-@main.stringMap = local_unnamed_addr global %runtime.hashmap* @"main$map.3"
-@main.init.string = internal unnamed_addr constant [7 x i8] c"CONNECT"
-@"main$map" = internal global %runtime.hashmap { %runtime.hashmap* null, i8* getelementptr inbounds ({ [8 x i8], i8*, { i8, [7 x i8] }, { { [7 x i8]*, [4 x i8] }, [56 x i8] } }, { [8 x i8], i8*, { i8, [7 x i8] }, { { [7 x i8]*, [4 x i8] }, [56 x i8] } }* @"main$mapbucket", i32 0, i32 0, i32 0), i32 1, i8 1, i8 8, i8 0 }
-@"main$mapbucket" = internal unnamed_addr global { [8 x i8], i8*, { i8, [7 x i8] }, { { [7 x i8]*, [4 x i8] }, [56 x i8] } } { [8 x i8] c"\04\00\00\00\00\00\00\00", i8* null, { i8, [7 x i8] } { i8 1, [7 x i8] zeroinitializer }, { { [7 x i8]*, [4 x i8] }, [56 x i8] } { { [7 x i8]*, [4 x i8] } { [7 x i8]* @main.init.string, [4 x i8] c"\07\00\00\00" }, [56 x i8] zeroinitializer } }
-@"main$map.1" = internal global %runtime.hashmap { %runtime.hashmap* null, i8* getelementptr inbounds ({ [8 x i8], i8*, { i8, [7 x i8] }, { i8, [7 x i8] } }, { [8 x i8], i8*, { i8, [7 x i8] }, { i8, [7 x i8] } }* @"main$mapbucket.2", i32 0, i32 0, i32 0), i32 1, i8 1, i8 1, i8 0 }
-@"main$mapbucket.2" = internal unnamed_addr global { [8 x i8], i8*, { i8, [7 x i8] }, { i8, [7 x i8] } } { [8 x i8] c"\04\00\00\00\00\00\00\00", i8* null, { i8, [7 x i8] } { i8 1, [7 x i8] zeroinitializer }, { i8, [7 x i8] } { i8 2, [7 x i8] zeroinitializer } }
-@"main$map.3" = internal global %runtime.hashmap { %runtime.hashmap* null, i8* getelementptr inbounds ({ [8 x i8], i8*, { { [7 x i8]*, [4 x i8] }, [56 x i8] }, { i8, [7 x i8] } }, { [8 x i8], i8*, { { [7 x i8]*, [4 x i8] }, [56 x i8] }, { i8, [7 x i8] } }* @"main$mapbucket.4", i32 0, i32 0, i32 0), i32 1, i8 8, i8 1, i8 0 }
-@"main$mapbucket.4" = internal unnamed_addr global { [8 x i8], i8*, { { [7 x i8]*, [4 x i8] }, [56 x i8] }, { i8, [7 x i8] } } { [8 x i8] c"x\00\00\00\00\00\00\00", i8* null, { { [7 x i8]*, [4 x i8] }, [56 x i8] } { { [7 x i8]*, [4 x i8] } { [7 x i8]* @main.init.string, [4 x i8] c"\07\00\00\00" }, [56 x i8] zeroinitializer }, { i8, [7 x i8] } { i8 2, [7 x i8] zeroinitializer } }
-
-define void @runtime.initAll() unnamed_addr {
-entry:
-  ret void
-}

interp/testdata/phi.ll

@@ -0,0 +1,31 @@
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64--linux"
+
+@main.phiNodesResultA = global i8 0
+@main.phiNodesResultB = global i8 0
+
+define void @runtime.initAll() {
+  call void @main.init()
+  ret void
+}
+
+; PHI nodes always use the value from the previous block, even in a loop. This
+; means that the loop below should swap the values %a and %b on each iteration.
+; Previously there was a bug which resulted in %b getting the value 3 on the
+; second iteration while it should have gotten 1 (from the first iteration of
+; %for.loop).
+define internal void @main.init() {
+entry:
+  br label %for.loop
+
+for.loop:
+  %a = phi i8 [ 1, %entry ], [ %b, %for.loop ]
+  %b = phi i8 [ 3, %entry ], [ %a, %for.loop ]
+  %icmp = icmp eq i8 %a, 3
+  br i1 %icmp, label %for.done, label %for.loop
+
+for.done:
+  store i8 %a, i8* @main.phiNodesResultA
+  store i8 %b, i8* @main.phiNodesResultB
+  ret void
+}

interp/testdata/phi.out.ll

@@ -0,0 +1,9 @@
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64--linux"
+
+@main.phiNodesResultA = local_unnamed_addr global i8 3
+@main.phiNodesResultB = local_unnamed_addr global i8 1
+
+define void @runtime.initAll() local_unnamed_addr {
+  ret void
+}

loader/errors.go

@@ -23,3 +23,13 @@ type Error struct {
 func (e Error) Error() string {
 	return e.Err.Error()
 }
+
+// Error returned when loading a *Program for a test binary but no test files
+// are present.
+type NoTestFilesError struct {
+	ImportPath string
+}
+
+func (e NoTestFilesError) Error() string {
+	return "no test files"
+}

loader/goroot.go

@@ -2,6 +2,14 @@ package loader
 
 // This file constructs a new temporary GOROOT directory by merging both the
 // standard Go GOROOT and the GOROOT from TinyGo using symlinks.
+//
+// The goal is to replace specific packages from Go with a TinyGo version. It's
+// never a partial replacement, either a package is fully replaced or it is not.
+// This is important because if we did allow to merge packages (e.g. by adding
+// files to a package), it would lead to a dependency on implementation details
+// with all the maintenance burden that results in. Only allowing to replace
+// packages as a whole avoids this as packages are already designed to have a
+// public (backwards-compatible) API.
 
 import (
 	"crypto/sha512"
@@ -139,6 +147,7 @@ func mergeDirectory(goroot, tinygoroot, tmpgoroot, importPath string, overrides
 		if err != nil {
 			return err
 		}
+		hasTinyGoFiles := false
 		for _, e := range tinygoEntries {
 			if e.IsDir() {
 				// A directory, so merge this thing.
@@ -154,6 +163,7 @@ func mergeDirectory(goroot, tinygoroot, tmpgoroot, importPath string, overrides
 				if err != nil {
 					return err
 				}
+				hasTinyGoFiles = true
 			}
 		}
 
@@ -164,21 +174,30 @@ func mergeDirectory(goroot, tinygoroot, tmpgoroot, importPath string, overrides
 			return err
 		}
 		for _, e := range gorootEntries {
-			if !e.IsDir() {
-				// Don't merge in files from Go. Otherwise we'd end up with a
-				// weird syscall package with files from both roots.
-				continue
-			}
-			if _, ok := overrides[path.Join(importPath, e.Name())+"/"]; ok {
-				// Already included above, so don't bother trying to create this
-				// symlink.
-				continue
-			}
-			newname := filepath.Join(tmpgoroot, "src", importPath, e.Name())
-			oldname := filepath.Join(goroot, "src", importPath, e.Name())
-			err := symlink(oldname, newname)
-			if err != nil {
-				return err
+			if e.IsDir() {
+				if _, ok := overrides[path.Join(importPath, e.Name())+"/"]; ok {
+					// Already included above, so don't bother trying to create this
+					// symlink.
+					continue
+				}
+				newname := filepath.Join(tmpgoroot, "src", importPath, e.Name())
+				oldname := filepath.Join(goroot, "src", importPath, e.Name())
+				err := symlink(oldname, newname)
+				if err != nil {
+					return err
+				}
+			} else {
+				// Only merge files from Go if TinyGo does not have any files.
+				// Otherwise we'd end up with a weird mix from both Go
+				// implementations.
+				if !hasTinyGoFiles {
+					newname := filepath.Join(tmpgoroot, "src", importPath, e.Name())
+					oldname := filepath.Join(goroot, "src", importPath, e.Name())
+					err := symlink(oldname, newname)
+					if err != nil {
+						return err
+					}
+				}
 			}
 		}
 	}
@@ -201,6 +220,8 @@ func needsSyscallPackage(buildTags []string) bool {
 func pathsToOverride(needsSyscallPackage bool) map[string]bool {
 	paths := map[string]bool{
 		"/":                     true,
+		"crypto/":               true,
+		"crypto/rand/":          false,
 		"device/":               false,
 		"examples/":             false,
 		"internal/":             true,
@@ -208,6 +229,7 @@ func pathsToOverride(needsSyscallPackage bool) map[string]bool {
 		"internal/reflectlite/": false,
 		"internal/task/":        false,
 		"machine/":              false,
+		"net/":                  true,
 		"os/":                   true,
 		"reflect/":              false,
 		"runtime/":              false,

loader/loader.go

@@ -210,6 +210,12 @@ func Load(config *compileopts.Config, inputPkgs []string, clangHeaders string, t
 		p.Packages[pkg.ImportPath] = pkg
 	}
 
+	if config.TestConfig.CompileTestBinary && !strings.HasSuffix(p.sorted[len(p.sorted)-1].ImportPath, ".test") {
+		// Trying to compile a test binary but there are no test files in this
+		// package.
+		return p, NoTestFilesError{p.sorted[len(p.sorted)-1].ImportPath}
+	}
+
 	return p, nil
 }
 
@@ -337,12 +343,11 @@ func (p *Package) Check() error {
 	checker.Importer = p
 
 	packageName := p.ImportPath
-	if p.Name == "main" {
-		// The main package normally has a different import path, such as
-		// "command-line-arguments" or "./testdata/cgo". Therefore, use the name
-		// "main" in such a case: this package isn't imported from anywhere.
-		// This is safe as it isn't possible to import a package with the name
-		// "main".
+	if p == p.program.MainPkg() {
+		if p.Name != "main" {
+			// Sanity check. Should not ever trigger.
+			panic("expected main package to have name 'main'")
+		}
 		packageName = "main"
 	}
 	typesPkg, err := checker.Check(packageName, p.program.fset, p.Files, &p.info)

main.go

@@ -13,11 +13,14 @@ import (
 	"os/exec"
 	"os/signal"
 	"path/filepath"
+	"regexp"
 	"runtime"
+	"strconv"
 	"strings"
 	"sync/atomic"
 	"time"
 
+	"github.com/google/shlex"
 	"github.com/mattn/go-colorable"
 	"github.com/tinygo-org/tinygo/builder"
 	"github.com/tinygo-org/tinygo/compileopts"
@@ -28,6 +31,7 @@ import (
 	"tinygo.org/x/go-llvm"
 
 	"go.bug.st/serial"
+	"go.bug.st/serial/enumerator"
 )
 
 var (
@@ -92,12 +96,31 @@ func copyFile(src, dst string) error {
 
 // executeCommand is a simple wrapper to exec.Cmd
 func executeCommand(options *compileopts.Options, name string, arg ...string) *exec.Cmd {
-	if options.PrintCommands {
-		fmt.Printf("%s %s\n", name, strings.Join(arg, " "))
+	if options.PrintCommands != nil {
+		options.PrintCommands(name, arg...)
 	}
 	return exec.Command(name, arg...)
 }
 
+// printCommand prints a command to stdout while formatting it like a real
+// command (escaping characters etc). The resulting command should be easy to
+// run directly in a shell, although it is not guaranteed to be a safe shell
+// escape. That's not a problem as the primary use case is printing the command,
+// not running it.
+func printCommand(cmd string, args ...string) {
+	command := append([]string{cmd}, args...)
+	for i, arg := range command {
+		// Source: https://www.oreilly.com/library/view/learning-the-bash/1565923472/ch01s09.html
+		const specialChars = "~`#$&*()\\|[]{};'\"<>?! "
+		if strings.ContainsAny(arg, specialChars) {
+			// See: https://stackoverflow.com/questions/15783701/which-characters-need-to-be-escaped-when-using-bash
+			arg = "'" + strings.ReplaceAll(arg, `'`, `'\''`) + "'"
+			command[i] = arg
+		}
+	}
+	fmt.Fprintln(os.Stderr, strings.Join(command, " "))
+}
+
 // Build compiles and links the given package and writes it to outpath.
 func Build(pkgName, outpath string, options *compileopts.Options) error {
 	config, err := builder.NewConfig(options)
@@ -133,15 +156,17 @@ func Build(pkgName, outpath string, options *compileopts.Options) error {
 	})
 }
 
-// Test runs the tests in the given package.
-func Test(pkgName string, options *compileopts.Options, testCompileOnly bool, outpath string) error {
+// Test runs the tests in the given package. Returns whether the test passed and
+// possibly an error if the test failed to run.
+func Test(pkgName string, options *compileopts.Options, testCompileOnly bool, outpath string) (bool, error) {
 	options.TestConfig.CompileTestBinary = true
 	config, err := builder.NewConfig(options)
 	if err != nil {
-		return err
+		return false, err
 	}
 
-	return builder.Build(pkgName, outpath, config, func(result builder.BuildResult) error {
+	passed := true
+	err = builder.Build(pkgName, outpath, config, func(result builder.BuildResult) error {
 		if testCompileOnly || outpath != "" {
 			// Write test binary to the specified file name.
 			if outpath == "" {
@@ -155,48 +180,78 @@ func Test(pkgName string, options *compileopts.Options, testCompileOnly bool, ou
 			// Do not run the test.
 			return nil
 		}
-		if len(config.Target.Emulator) == 0 {
-			// Run directly.
-			cmd := executeCommand(config.Options, result.Binary)
-			cmd.Stdout = os.Stdout
-			cmd.Stderr = os.Stderr
-			cmd.Dir = result.MainDir
-			err := cmd.Run()
-			if err != nil {
-				// Propagate the exit code
-				if err, ok := err.(*exec.ExitError); ok {
-					os.Exit(err.ExitCode())
-				}
-				return &commandError{"failed to run compiled binary", result.Binary, err}
-			}
-			return nil
+
+		// Run the test.
+		start := time.Now()
+		var err error
+		passed, err = runPackageTest(config, result)
+		if err != nil {
+			return err
+		}
+		duration := time.Since(start)
+
+		// Print the result.
+		importPath := strings.TrimSuffix(result.ImportPath, ".test")
+		if passed {
+			fmt.Printf("ok  \t%s\t%.3fs\n", importPath, duration.Seconds())
 		} else {
-			// Run in an emulator.
-			args := append(config.Target.Emulator[1:], result.Binary)
-			cmd := executeCommand(config.Options, config.Target.Emulator[0], args...)
-			buf := &bytes.Buffer{}
-			w := io.MultiWriter(os.Stdout, buf)
-			cmd.Stdout = w
-			cmd.Stderr = os.Stderr
-			err := cmd.Run()
-			if err != nil {
-				if err, ok := err.(*exec.ExitError); !ok || !err.Exited() {
-					// Workaround for QEMU which always exits with an error.
-					return &commandError{"failed to run emulator with", result.Binary, err}
-				}
+			fmt.Printf("FAIL\t%s\t%.3fs\n", importPath, duration.Seconds())
+		}
+		return nil
+	})
+	if err, ok := err.(loader.NoTestFilesError); ok {
+		fmt.Printf("?   \t%s\t[no test files]\n", err.ImportPath)
+		// Pretend the test passed - it at least didn't fail.
+		return true, nil
+	}
+	return passed, err
+}
+
+// runPackageTest runs a test binary that was previously built. The return
+// values are whether the test passed and any errors encountered while trying to
+// run the binary.
+func runPackageTest(config *compileopts.Config, result builder.BuildResult) (bool, error) {
+	if len(config.Target.Emulator) == 0 {
+		// Run directly.
+		cmd := executeCommand(config.Options, result.Binary)
+		cmd.Stdout = os.Stdout
+		cmd.Stderr = os.Stderr
+		cmd.Dir = result.MainDir
+		err := cmd.Run()
+		if err != nil {
+			if _, ok := err.(*exec.ExitError); ok {
+				// Binary exited with a non-zero exit code, which means the test
+				// failed.
+				return false, nil
 			}
-			testOutput := string(buf.Bytes())
-			if testOutput == "PASS\n" || strings.HasSuffix(testOutput, "\nPASS\n") {
-				// Test passed.
-				return nil
-			} else {
-				// Test failed, either by ending with the word "FAIL" or with a
-				// panic of some sort.
-				os.Exit(1)
-				return nil // unreachable
+			return false, &commandError{"failed to run compiled binary", result.Binary, err}
+		}
+		return true, nil
+	} else {
+		// Run in an emulator.
+		args := append(config.Target.Emulator[1:], result.Binary)
+		cmd := executeCommand(config.Options, config.Target.Emulator[0], args...)
+		buf := &bytes.Buffer{}
+		w := io.MultiWriter(os.Stdout, buf)
+		cmd.Stdout = w
+		cmd.Stderr = os.Stderr
+		err := cmd.Run()
+		if err != nil {
+			if err, ok := err.(*exec.ExitError); !ok || !err.Exited() {
+				// Workaround for QEMU which always exits with an error.
+				return false, &commandError{"failed to run emulator with", result.Binary, err}
 			}
 		}
-	})
+		testOutput := string(buf.Bytes())
+		if testOutput == "PASS\n" || strings.HasSuffix(testOutput, "\nPASS\n") {
+			// Test passed.
+			return true, nil
+		} else {
+			// Test failed, either by ending with the word "FAIL" or with a
+			// panic of some sort.
+			return false, nil
+		}
+	}
 }
 
 // Flash builds and flashes the built binary to the given serial port.
@@ -221,6 +276,8 @@ func Flash(pkgName, port string, options *compileopts.Options) error {
 			fileExt = ".bin"
 		case strings.Contains(config.Target.FlashCommand, "{uf2}"):
 			fileExt = ".uf2"
+		case strings.Contains(config.Target.FlashCommand, "{zip}"):
+			fileExt = ".zip"
 		default:
 			return errors.New("invalid target file - did you forget the {hex} token in the 'flash-command' section?")
 		}
@@ -240,15 +297,12 @@ func Flash(pkgName, port string, options *compileopts.Options) error {
 	return builder.Build(pkgName, fileExt, config, func(result builder.BuildResult) error {
 		// do we need port reset to put MCU into bootloader mode?
 		if config.Target.PortReset == "true" && flashMethod != "openocd" {
-			if port == "" {
-				var err error
-				port, err = getDefaultPort()
-				if err != nil {
-					return err
-				}
+			port, err := getDefaultPort(port, config.Target.SerialPort)
+			if err != nil {
+				return err
 			}
 
-			err := touchSerialPortAt1200bps(port)
+			err = touchSerialPortAt1200bps(port)
 			if err != nil {
 				return &commandError{"failed to reset port", result.Binary, err}
 			}
@@ -261,36 +315,36 @@ func Flash(pkgName, port string, options *compileopts.Options) error {
 		case "", "command":
 			// Create the command.
 			flashCmd := config.Target.FlashCommand
-			fileToken := "{" + fileExt[1:] + "}"
-			flashCmd = strings.ReplaceAll(flashCmd, fileToken, result.Binary)
+			flashCmdList, err := shlex.Split(flashCmd)
+			if err != nil {
+				return fmt.Errorf("could not parse flash command %#v: %w", flashCmd, err)
+			}
 
-			if port == "" && strings.Contains(flashCmd, "{port}") {
+			if strings.Contains(flashCmd, "{port}") {
 				var err error
-				port, err = getDefaultPort()
+				port, err = getDefaultPort(port, config.Target.SerialPort)
 				if err != nil {
 					return err
 				}
 			}
 
-			flashCmd = strings.ReplaceAll(flashCmd, "{port}", port)
-
-			// Execute the command.
-			var cmd *exec.Cmd
-			switch runtime.GOOS {
-			case "windows":
-				command := strings.Split(flashCmd, " ")
-				if len(command) < 2 {
-					return errors.New("invalid flash command")
-				}
-				cmd = executeCommand(config.Options, command[0], command[1:]...)
-			default:
-				cmd = executeCommand(config.Options, "/bin/sh", "-c", flashCmd)
+			// Fill in fields in the command template.
+			fileToken := "{" + fileExt[1:] + "}"
+			for i, arg := range flashCmdList {
+				arg = strings.ReplaceAll(arg, fileToken, result.Binary)
+				arg = strings.ReplaceAll(arg, "{port}", port)
+				flashCmdList[i] = arg
 			}
 
+			// Execute the command.
+			if len(flashCmdList) < 2 {
+				return fmt.Errorf("invalid flash command: %#v", flashCmd)
+			}
+			cmd := executeCommand(config.Options, flashCmdList[0], flashCmdList[1:]...)
 			cmd.Stdout = os.Stdout
 			cmd.Stderr = os.Stderr
 			cmd.Dir = goenv.Get("TINYGOROOT")
-			err := cmd.Run()
+			err = cmd.Run()
 			if err != nil {
 				return &commandError{"failed to flash", result.Binary, err}
 			}
@@ -381,7 +435,7 @@ func FlashGDB(pkgName string, ocdOutput bool, options *compileopts.Options) erro
 		case "native":
 			// Run GDB directly.
 		case "openocd":
-			gdbCommands = append(gdbCommands, "target remote :3333", "monitor halt", "load", "monitor reset halt")
+			gdbCommands = append(gdbCommands, "target extended-remote :3333", "monitor halt", "load", "monitor reset halt")
 
 			// We need a separate debugging daemon for on-chip debugging.
 			args, err := config.OpenOCDConfiguration()
@@ -400,7 +454,7 @@ func FlashGDB(pkgName string, ocdOutput bool, options *compileopts.Options) erro
 				daemon.Stderr = w
 			}
 		case "jlink":
-			gdbCommands = append(gdbCommands, "target remote :2331", "load", "monitor reset halt")
+			gdbCommands = append(gdbCommands, "target extended-remote :2331", "load", "monitor reset halt")
 
 			// We need a separate debugging daemon for on-chip debugging.
 			daemon = executeCommand(config.Options, "JLinkGDBServer", "-device", config.Target.JLinkDevice)
@@ -415,7 +469,7 @@ func FlashGDB(pkgName string, ocdOutput bool, options *compileopts.Options) erro
 				daemon.Stderr = w
 			}
 		case "qemu":
-			gdbCommands = append(gdbCommands, "target remote :1234")
+			gdbCommands = append(gdbCommands, "target extended-remote :1234")
 
 			// Run in an emulator.
 			args := append(config.Target.Emulator[1:], result.Binary, "-s", "-S")
@@ -423,7 +477,7 @@ func FlashGDB(pkgName string, ocdOutput bool, options *compileopts.Options) erro
 			daemon.Stdout = os.Stdout
 			daemon.Stderr = os.Stderr
 		case "qemu-user":
-			gdbCommands = append(gdbCommands, "target remote :1234")
+			gdbCommands = append(gdbCommands, "target extended-remote :1234")
 
 			// Run in an emulator.
 			args := append(config.Target.Emulator[1:], "-g", "1234", result.Binary)
@@ -431,7 +485,7 @@ func FlashGDB(pkgName string, ocdOutput bool, options *compileopts.Options) erro
 			daemon.Stdout = os.Stdout
 			daemon.Stderr = os.Stderr
 		case "mgba":
-			gdbCommands = append(gdbCommands, "target remote :2345")
+			gdbCommands = append(gdbCommands, "target extended-remote :2345")
 
 			// Run in an emulator.
 			args := append(config.Target.Emulator[1:], result.Binary, "-g")
@@ -439,7 +493,7 @@ func FlashGDB(pkgName string, ocdOutput bool, options *compileopts.Options) erro
 			daemon.Stdout = os.Stdout
 			daemon.Stderr = os.Stderr
 		case "simavr":
-			gdbCommands = append(gdbCommands, "target remote :1234")
+			gdbCommands = append(gdbCommands, "target extended-remote :1234")
 
 			// Run in an emulator.
 			args := append(config.Target.Emulator[1:], "-g", result.Binary)
@@ -665,41 +719,126 @@ func windowsFindUSBDrive(volume string, options *compileopts.Options) (string, e
 }
 
 // getDefaultPort returns the default serial port depending on the operating system.
-func getDefaultPort() (port string, err error) {
-	var portPath string
+func getDefaultPort(portFlag string, usbInterfaces []string) (port string, err error) {
+	portCandidates := strings.FieldsFunc(portFlag, func(c rune) bool { return c == ',' })
+	if len(portCandidates) == 1 {
+		return portCandidates[0], nil
+	}
+
+	var ports []string
 	switch runtime.GOOS {
-	case "darwin":
-		portPath = "/dev/cu.usb*"
-	case "linux":
-		portPath = "/dev/ttyACM*"
 	case "freebsd":
-		portPath = "/dev/cuaU*"
-	case "windows":
-		ports, err := serial.GetPortsList()
+		ports, err = filepath.Glob("/dev/cuaU*")
+	case "darwin", "linux", "windows":
+		var portsList []*enumerator.PortDetails
+		portsList, err = enumerator.GetDetailedPortsList()
 		if err != nil {
 			return "", err
 		}
 
-		if len(ports) == 0 {
-			return "", errors.New("no serial ports available")
-		} else if len(ports) > 1 {
-			return "", errors.New("multiple serial ports available - use -port flag")
+		var preferredPortIDs [][2]uint16
+		for _, s := range usbInterfaces {
+			parts := strings.Split(s, ":")
+			if len(parts) != 3 || (parts[0] != "acm" && parts[0] == "usb") {
+				// acm and usb are the two types of serial ports recognized
+				// under Linux (ttyACM*, ttyUSB*). Other operating systems don't
+				// generally make this distinction. If this is not one of the
+				// given USB devices, don't try to parse the USB IDs.
+				continue
+			}
+			vid, err := strconv.ParseUint(parts[1], 16, 16)
+			if err != nil {
+				return "", fmt.Errorf("could not parse USB vendor ID %q: %w", parts[1], err)
+			}
+			pid, err := strconv.ParseUint(parts[2], 16, 16)
+			if err != nil {
+				return "", fmt.Errorf("could not parse USB product ID %q: %w", parts[1], err)
+			}
+			preferredPortIDs = append(preferredPortIDs, [2]uint16{uint16(vid), uint16(pid)})
 		}
 
-		return ports[0], nil
+		var primaryPorts []string   // ports picked from preferred USB VID/PID
+		var secondaryPorts []string // other ports (as a fallback)
+		for _, p := range portsList {
+			if !p.IsUSB {
+				continue
+			}
+			if p.VID != "" && p.PID != "" {
+				foundPort := false
+				vid, vidErr := strconv.ParseUint(p.VID, 16, 16)
+				pid, pidErr := strconv.ParseUint(p.PID, 16, 16)
+				if vidErr == nil && pidErr == nil {
+					for _, id := range preferredPortIDs {
+						if uint16(vid) == id[0] && uint16(pid) == id[1] {
+							primaryPorts = append(primaryPorts, p.Name)
+							foundPort = true
+							continue
+						}
+					}
+				}
+				if foundPort {
+					continue
+				}
+			}
+
+			secondaryPorts = append(secondaryPorts, p.Name)
+		}
+		if len(primaryPorts) == 1 {
+			// There is exactly one match in the set of preferred ports. Use
+			// this port, even if there may be others available. This allows
+			// flashing a specific board even if there are multiple available.
+			return primaryPorts[0], nil
+		} else if len(primaryPorts) > 1 {
+			// There are multiple preferred ports, probably because more than
+			// one device of the same type are connected (e.g. two Arduino
+			// Unos).
+			ports = primaryPorts
+		} else {
+			// No preferred ports found. Fall back to other serial ports
+			// available in the system.
+			ports = secondaryPorts
+		}
+
+		if len(ports) == 0 {
+			// fallback
+			switch runtime.GOOS {
+			case "darwin":
+				ports, err = filepath.Glob("/dev/cu.usb*")
+			case "linux":
+				ports, err = filepath.Glob("/dev/ttyACM*")
+			case "windows":
+				ports, err = serial.GetPortsList()
+			}
+		}
 	default:
 		return "", errors.New("unable to search for a default USB device to be flashed on this OS")
 	}
 
-	d, err := filepath.Glob(portPath)
 	if err != nil {
 		return "", err
-	}
-	if d == nil {
+	} else if ports == nil {
 		return "", errors.New("unable to locate a serial port")
+	} else if len(ports) == 0 {
+		return "", errors.New("no serial ports available")
 	}
 
-	return d[0], nil
+	if len(portCandidates) == 0 {
+		if len(ports) == 1 {
+			return ports[0], nil
+		} else {
+			return "", errors.New("multiple serial ports available - use -port flag, available ports are " + strings.Join(ports, ", "))
+		}
+	}
+
+	for _, ps := range portCandidates {
+		for _, p := range ports {
+			if p == ps {
+				return p, nil
+			}
+		}
+	}
+
+	return "", errors.New("port you specified '" + strings.Join(portCandidates, ",") + "' does not exist, available ports are " + strings.Join(ports, ", "))
 }
 
 func usage() {
@@ -812,6 +951,52 @@ func handleCompilerError(err error) {
 	}
 }
 
+// This is a special type for the -X flag to parse the pkgpath.Var=stringVal
+// format. It has to be a special type to allow multiple variables to be defined
+// this way.
+type globalValuesFlag map[string]map[string]string
+
+func (m globalValuesFlag) String() string {
+	return "pkgpath.Var=value"
+}
+
+func (m globalValuesFlag) Set(value string) error {
+	equalsIndex := strings.IndexByte(value, '=')
+	if equalsIndex < 0 {
+		return errors.New("expected format pkgpath.Var=value")
+	}
+	pathAndName := value[:equalsIndex]
+	pointIndex := strings.LastIndexByte(pathAndName, '.')
+	if pointIndex < 0 {
+		return errors.New("expected format pkgpath.Var=value")
+	}
+	path := pathAndName[:pointIndex]
+	name := pathAndName[pointIndex+1:]
+	stringValue := value[equalsIndex+1:]
+	if m[path] == nil {
+		m[path] = make(map[string]string)
+	}
+	m[path][name] = stringValue
+	return nil
+}
+
+// parseGoLinkFlag parses the -ldflags parameter. Its primary purpose right now
+// is the -X flag, for setting the value of global string variables.
+func parseGoLinkFlag(flagsString string) (map[string]map[string]string, error) {
+	set := flag.NewFlagSet("link", flag.ExitOnError)
+	globalVarValues := make(globalValuesFlag)
+	set.Var(globalVarValues, "X", "Set the value of the string variable to the given value.")
+	flags, err := shlex.Split(flagsString)
+	if err != nil {
+		return nil, err
+	}
+	err = set.Parse(flags)
+	if err != nil {
+		return nil, err
+	}
+	return map[string]map[string]string(globalVarValues), nil
+}
+
 func main() {
 	if len(os.Args) < 2 {
 		fmt.Fprintln(os.Stderr, "No command-line arguments supplied.")
@@ -824,6 +1009,7 @@ func main() {
 	gc := flag.String("gc", "", "garbage collector to use (none, leaking, extalloc, conservative)")
 	panicStrategy := flag.String("panic", "print", "panic strategy (print, trap)")
 	scheduler := flag.String("scheduler", "", "which scheduler to use (none, coroutines, tasks)")
+	serial := flag.String("serial", "", "which serial output to use (none, uart, usb)")
 	printIR := flag.Bool("printir", false, "print LLVM IR")
 	dumpSSA := flag.Bool("dumpssa", false, "dump internal Go SSA")
 	verifyIR := flag.Bool("verifyir", false, "run extra verification steps on LLVM IR")
@@ -831,19 +1017,22 @@ func main() {
 	target := flag.String("target", "", "LLVM target | .json file with TargetSpec")
 	printSize := flag.String("size", "", "print sizes (none, short, full)")
 	printStacks := flag.Bool("print-stacks", false, "print stack sizes of goroutines")
+	printAllocsString := flag.String("print-allocs", "", "regular expression of functions for which heap allocations should be printed")
 	printCommands := flag.Bool("x", false, "Print commands")
 	nodebug := flag.Bool("no-debug", false, "disable DWARF debug symbol generation")
+	ocdCommandsString := flag.String("ocd-commands", "", "OpenOCD commands, overriding target spec (can specify multiple separated by commas)")
 	ocdOutput := flag.Bool("ocd-output", false, "print OCD daemon output during debug")
-	port := flag.String("port", "", "flash port")
+	port := flag.String("port", "", "flash port (can specify multiple candidates separated by commas)")
 	programmer := flag.String("programmer", "", "which hardware programmer to use")
-	cFlags := flag.String("cflags", "", "additional cflags for compiler")
-	ldFlags := flag.String("ldflags", "", "additional ldflags for linker")
+	ldflags := flag.String("ldflags", "", "Go link tool compatible ldflags")
 	wasmAbi := flag.String("wasm-abi", "", "WebAssembly ABI conventions: js (no i64 params) or generic")
+	llvmFeatures := flag.String("llvm-features", "", "comma separated LLVM features to enable")
 
-	var flagJSON, flagDeps *bool
+	var flagJSON, flagDeps, flagTest *bool
 	if command == "help" || command == "list" {
 		flagJSON = flag.Bool("json", false, "print data in JSON format")
-		flagDeps = flag.Bool("deps", false, "")
+		flagDeps = flag.Bool("deps", false, "supply -deps flag to go list")
+		flagTest = flag.Bool("test", false, "supply -test flag to go list")
 	}
 	var outpath string
 	if command == "help" || command == "build" || command == "build-library" || command == "test" {
@@ -867,35 +1056,54 @@ func main() {
 	}
 
 	flag.CommandLine.Parse(os.Args[2:])
+	globalVarValues, err := parseGoLinkFlag(*ldflags)
+	if err != nil {
+		fmt.Fprintln(os.Stderr, err)
+		os.Exit(1)
+	}
+
+	var printAllocs *regexp.Regexp
+	if *printAllocsString != "" {
+		printAllocs, err = regexp.Compile(*printAllocsString)
+		if err != nil {
+			fmt.Fprintln(os.Stderr, err)
+			os.Exit(1)
+		}
+	}
+
+	var ocdCommands []string
+	if *ocdCommandsString != "" {
+		ocdCommands = strings.Split(*ocdCommandsString, ",")
+	}
+
 	options := &compileopts.Options{
-		Target:        *target,
-		Opt:           *opt,
-		GC:            *gc,
-		PanicStrategy: *panicStrategy,
-		Scheduler:     *scheduler,
-		PrintIR:       *printIR,
-		DumpSSA:       *dumpSSA,
-		VerifyIR:      *verifyIR,
-		Debug:         !*nodebug,
-		PrintSizes:    *printSize,
-		PrintStacks:   *printStacks,
-		PrintCommands: *printCommands,
-		Tags:          *tags,
-		WasmAbi:       *wasmAbi,
-		Programmer:    *programmer,
+		Target:          *target,
+		Opt:             *opt,
+		GC:              *gc,
+		PanicStrategy:   *panicStrategy,
+		Scheduler:       *scheduler,
+		Serial:          *serial,
+		PrintIR:         *printIR,
+		DumpSSA:         *dumpSSA,
+		VerifyIR:        *verifyIR,
+		Debug:           !*nodebug,
+		PrintSizes:      *printSize,
+		PrintStacks:     *printStacks,
+		PrintAllocs:     printAllocs,
+		Tags:            *tags,
+		GlobalValues:    globalVarValues,
+		WasmAbi:         *wasmAbi,
+		Programmer:      *programmer,
+		OpenOCDCommands: ocdCommands,
+		LLVMFeatures:    *llvmFeatures,
 	}
-
-	if *cFlags != "" {
-		options.CFlags = strings.Split(*cFlags, " ")
-	}
-
-	if *ldFlags != "" {
-		options.LDFlags = strings.Split(*ldFlags, " ")
+	if *printCommands {
+		options.PrintCommands = printCommand
 	}
 
 	os.Setenv("CC", "clang -target="+*target)
 
-	err := options.Verify()
+	err = options.Verify()
 	if err != nil {
 		fmt.Fprintln(os.Stderr, err.Error())
 		usage()
@@ -983,16 +1191,26 @@ func main() {
 		err := Run(pkgName, options)
 		handleCompilerError(err)
 	case "test":
-		pkgName := "."
-		if flag.NArg() == 1 {
-			pkgName = filepath.ToSlash(flag.Arg(0))
-		} else if flag.NArg() > 1 {
-			fmt.Fprintln(os.Stderr, "test only accepts a single positional argument: package name, but multiple were specified")
-			usage()
+		var pkgNames []string
+		for i := 0; i < flag.NArg(); i++ {
+			pkgNames = append(pkgNames, filepath.ToSlash(flag.Arg(i)))
+		}
+		if len(pkgNames) == 0 {
+			pkgNames = []string{"."}
+		}
+		allTestsPassed := true
+		for _, pkgName := range pkgNames {
+			// TODO: parallelize building the test binaries
+			passed, err := Test(pkgName, options, *testCompileOnlyFlag, outpath)
+			handleCompilerError(err)
+			if !passed {
+				allTestsPassed = false
+			}
+		}
+		if !allTestsPassed {
+			fmt.Println("FAIL")
 			os.Exit(1)
 		}
-		err := Test(pkgName, options, *testCompileOnlyFlag, outpath)
-		handleCompilerError(err)
 	case "targets":
 		dir := filepath.Join(goenv.Get("TINYGOROOT"), "targets")
 		entries, err := ioutil.ReadDir(dir)
@@ -1067,6 +1285,9 @@ func main() {
 		if *flagDeps {
 			extraArgs = append(extraArgs, "-deps")
 		}
+		if *flagTest {
+			extraArgs = append(extraArgs, "-test")
+		}
 		cmd, err := loader.List(config, extraArgs, flag.Args())
 		if err != nil {
 			fmt.Fprintln(os.Stderr, "failed to run `go list`:", err)

main_test.go

@@ -13,7 +13,6 @@ import (
 	"os/exec"
 	"path/filepath"
 	"runtime"
-	"sort"
 	"strings"
 	"sync"
 	"testing"
@@ -29,41 +28,43 @@ const TESTDATA = "testdata"
 var testTarget = flag.String("target", "", "override test target")
 
 func TestCompiler(t *testing.T) {
-	matches, err := filepath.Glob(filepath.Join(TESTDATA, "*.go"))
-	if err != nil {
-		t.Fatal("could not read test files:", err)
+	tests := []string{
+		"alias.go",
+		"atomic.go",
+		"binop.go",
+		"calls.go",
+		"cgo/",
+		"channel.go",
+		"float.go",
+		"gc.go",
+		"goroutines.go",
+		"init.go",
+		"init_multi.go",
+		"interface.go",
+		"json.go",
+		"map.go",
+		"math.go",
+		"print.go",
+		"reflect.go",
+		"slice.go",
+		"sort.go",
+		"stdlib.go",
+		"string.go",
+		"structs.go",
+		"zeroalloc.go",
 	}
 
-	dirMatches, err := filepath.Glob(filepath.Join(TESTDATA, "*", "main.go"))
-	if err != nil {
-		t.Fatal("could not read test packages:", err)
-	}
-	if len(matches) == 0 || len(dirMatches) == 0 {
-		t.Fatal("no test files found")
-	}
-	for _, m := range dirMatches {
-		matches = append(matches, filepath.Dir(m)+string(filepath.Separator))
-	}
-
-	sort.Strings(matches)
-
 	if *testTarget != "" {
 		// This makes it possible to run one specific test (instead of all),
 		// which is especially useful to quickly check whether some changes
 		// affect a particular target architecture.
-		runPlatTests(*testTarget, matches, t)
+		runPlatTests(*testTarget, tests, t)
 		return
 	}
 
 	if runtime.GOOS != "windows" {
 		t.Run("Host", func(t *testing.T) {
-			runPlatTests("", matches, t)
-			if runtime.GOOS == "darwin" {
-				runTest("testdata/libc/filesystem.go", "", t,
-					nil, nil)
-				runTest("testdata/libc/env.go", "", t,
-					[]string{"ENV1=VALUE1", "ENV2=VALUE2"}, nil)
-			}
+			runPlatTests("", tests, t)
 		})
 	}
 
@@ -72,26 +73,26 @@ func TestCompiler(t *testing.T) {
 	}
 
 	t.Run("EmulatedCortexM3", func(t *testing.T) {
-		runPlatTests("cortex-m-qemu", matches, t)
+		runPlatTests("cortex-m-qemu", tests, t)
 	})
 
 	if runtime.GOOS == "windows" || runtime.GOOS == "darwin" {
 		// Note: running only on Windows and macOS because Linux (as of 2020)
 		// usually has an outdated QEMU version that doesn't support RISC-V yet.
 		t.Run("EmulatedRISCV", func(t *testing.T) {
-			runPlatTests("riscv-qemu", matches, t)
+			runPlatTests("riscv-qemu", tests, t)
 		})
 	}
 
 	if runtime.GOOS == "linux" {
 		t.Run("X86Linux", func(t *testing.T) {
-			runPlatTests("i386--linux-gnu", matches, t)
+			runPlatTests("i386--linux-gnu", tests, t)
 		})
 		t.Run("ARMLinux", func(t *testing.T) {
-			runPlatTests("arm--linux-gnueabihf", matches, t)
+			runPlatTests("arm--linux-gnueabihf", tests, t)
 		})
 		t.Run("ARM64Linux", func(t *testing.T) {
-			runPlatTests("aarch64--linux-gnu", matches, t)
+			runPlatTests("aarch64--linux-gnu", tests, t)
 		})
 		goVersion, err := goenv.GorootVersionString(goenv.Get("GOROOT"))
 		if err != nil {
@@ -104,27 +105,72 @@ func TestCompiler(t *testing.T) {
 			// below that are also not supported but still seem to pass, so
 			// include them in the tests for now.
 			t.Run("WebAssembly", func(t *testing.T) {
-				runPlatTests("wasm", matches, t)
+				runPlatTests("wasm", tests, t)
 			})
 		}
 
 		t.Run("WASI", func(t *testing.T) {
-			runPlatTests("wasi", matches, t)
-			runTest("testdata/libc/env.go", "wasi", t,
-				[]string{"--env", "ENV1=VALUE1", "--env", "ENV2=VALUE2"}, nil)
-			runTest("testdata/libc/filesystem.go", "wasi", t, nil, []string{"--dir=."})
+			runPlatTests("wasi", tests, t)
 		})
 	}
+
+	// Test a few build options.
+	t.Run("build-options", func(t *testing.T) {
+		if runtime.GOOS == "windows" {
+			// These tests assume a host that is supported by TinyGo.
+			t.Skip("can't test build options on Windows")
+		}
+		t.Parallel()
+
+		// Test with few optimizations enabled (no inlining, etc).
+		t.Run("opt=1", func(t *testing.T) {
+			t.Parallel()
+			runTestWithConfig("stdlib.go", "", t, &compileopts.Options{
+				Opt: "1",
+			}, nil, nil)
+		})
+
+		// Test with only the bare minimum of optimizations enabled.
+		// TODO: fix this for stdlib.go, which currently fails.
+		t.Run("opt=0", func(t *testing.T) {
+			t.Parallel()
+			runTestWithConfig("print.go", "", t, &compileopts.Options{
+				Opt: "0",
+			}, nil, nil)
+		})
+
+		t.Run("ldflags", func(t *testing.T) {
+			t.Parallel()
+			runTestWithConfig("ldflags.go", "", t, &compileopts.Options{
+				Opt: "z",
+				GlobalValues: map[string]map[string]string{
+					"main": {
+						"someGlobal": "foobar",
+					},
+				},
+			}, nil, nil)
+		})
+	})
 }
 
-func runPlatTests(target string, matches []string, t *testing.T) {
+func runPlatTests(target string, tests []string, t *testing.T) {
 	t.Parallel()
 
-	for _, path := range matches {
-		path := path // redefine to avoid race condition
-		t.Run(filepath.Base(path), func(t *testing.T) {
+	for _, name := range tests {
+		name := name // redefine to avoid race condition
+		t.Run(name, func(t *testing.T) {
 			t.Parallel()
-			runTest(path, target, t, nil, nil)
+			runTest(name, target, t, nil, nil)
+		})
+	}
+	if target == "wasi" || target == "" {
+		t.Run("filesystem.go", func(t *testing.T) {
+			t.Parallel()
+			runTest("filesystem.go", target, t, nil, nil)
+		})
+		t.Run("env.go", func(t *testing.T) {
+			t.Parallel()
+			runTest("env.go", target, t, []string{"first", "second"}, []string{"ENV1=VALUE1", "ENV2=VALUE2"})
 		})
 	}
 }
@@ -141,10 +187,28 @@ func runBuild(src, out string, opts *compileopts.Options) error {
 	return Build(src, out, opts)
 }
 
-func runTest(path, target string, t *testing.T, environmentVars []string, additionalArgs []string) {
+func runTest(name, target string, t *testing.T, cmdArgs, environmentVars []string) {
+	options := &compileopts.Options{
+		Target:     target,
+		Opt:        "z",
+		PrintIR:    false,
+		DumpSSA:    false,
+		VerifyIR:   true,
+		Debug:      true,
+		PrintSizes: "",
+		WasmAbi:    "",
+	}
+	runTestWithConfig(name, target, t, options, cmdArgs, environmentVars)
+}
+
+func runTestWithConfig(name, target string, t *testing.T, options *compileopts.Options, cmdArgs, environmentVars []string) {
+	// Get the expected output for this test.
+	// Note: not using filepath.Join as it strips the path separator at the end
+	// of the path.
+	path := TESTDATA + "/" + name
 	// Get the expected output for this test.
 	txtpath := path[:len(path)-3] + ".txt"
-	if path[len(path)-1] == os.PathSeparator {
+	if path[len(path)-1] == '/' {
 		txtpath = path + "out.txt"
 	}
 	expected, err := ioutil.ReadFile(txtpath)
@@ -165,19 +229,8 @@ func runTest(path, target string, t *testing.T, environmentVars []string, additi
 	}()
 
 	// Build the test binary.
-	config := &compileopts.Options{
-		Target:     target,
-		Opt:        "z",
-		PrintIR:    false,
-		DumpSSA:    false,
-		VerifyIR:   true,
-		Debug:      true,
-		PrintSizes: "",
-		WasmAbi:    "",
-	}
-
 	binary := filepath.Join(tmpdir, "test")
-	err = runBuild("./"+path, binary, config)
+	err = runBuild("./"+path, binary, options)
 	if err != nil {
 		printCompilerError(t.Log, err)
 		t.Fail()
@@ -191,6 +244,7 @@ func runTest(path, target string, t *testing.T, environmentVars []string, additi
 	if target == "" {
 		cmd = exec.Command(binary)
 		cmd.Env = append(cmd.Env, environmentVars...)
+		cmd.Args = append(cmd.Args, cmdArgs...)
 	} else {
 		spec, err := compileopts.LoadTarget(target)
 		if err != nil {
@@ -200,13 +254,16 @@ func runTest(path, target string, t *testing.T, environmentVars []string, additi
 			cmd = exec.Command(binary)
 		} else {
 			args := append(spec.Emulator[1:], binary)
-			cmd = exec.Command(spec.Emulator[0], append(args, additionalArgs...)...)
+			cmd = exec.Command(spec.Emulator[0], args...)
 		}
 
 		if len(spec.Emulator) != 0 && spec.Emulator[0] == "wasmtime" {
+			// Allow reading from the current directory.
+			cmd.Args = append(cmd.Args, "--dir=.")
 			for _, v := range environmentVars {
 				cmd.Args = append(cmd.Args, "--env", v)
 			}
+			cmd.Args = append(cmd.Args, cmdArgs...)
 		} else {
 			cmd.Env = append(cmd.Env, environmentVars...)
 		}

src/crypto/rand/rand.go

@@ -0,0 +1,19 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package rand implements a cryptographically secure
+// random number generator.
+package rand
+
+import "io"
+
+// Reader is a global, shared instance of a cryptographically
+// secure random number generator.
+var Reader io.Reader
+
+// Read is a helper function that calls Reader.Read using io.ReadFull.
+// On return, n == len(b) if and only if err == nil.
+func Read(b []byte) (n int, err error) {
+	return io.ReadFull(Reader, b)
+}

src/crypto/rand/rand_getentropy.go

@@ -0,0 +1,38 @@
+// +build darwin freebsd wasi
+
+// This implementation of crypto/rand uses the getentropy system call (available
+// on both MacOS and WASI) to generate random numbers.
+
+package rand
+
+import (
+	"errors"
+	"unsafe"
+)
+
+var errReadFailed = errors.New("rand: could not read random bytes")
+
+func init() {
+	Reader = &reader{}
+}
+
+type reader struct {
+}
+
+func (r *reader) Read(b []byte) (n int, err error) {
+	if len(b) != 0 {
+		if len(b) > 256 {
+			b = b[:256]
+		}
+		result := libc_getentropy(unsafe.Pointer(&b[0]), len(b))
+		if result < 0 {
+			// Maybe we should return a syscall.Errno here?
+			return 0, errReadFailed
+		}
+	}
+	return len(b), nil
+}
+
+// int getentropy(void *buf, size_t buflen);
+//export getentropy
+func libc_getentropy(buf unsafe.Pointer, buflen int) int

src/crypto/rand/rand_urandom.go

@@ -0,0 +1,38 @@
+// +build linux,!baremetal,!wasi
+
+// This implementation of crypto/rand uses the /dev/urandom pseudo-file to
+// generate random numbers.
+// TODO: convert to the getentropy or getrandom libc function on Linux once it
+// is more widely supported.
+
+package rand
+
+import (
+	"syscall"
+)
+
+func init() {
+	Reader = &reader{}
+}
+
+type reader struct {
+	fd int
+}
+
+func (r *reader) Read(b []byte) (n int, err error) {
+	if len(b) == 0 {
+		return
+	}
+
+	// Open /dev/urandom first if needed.
+	if r.fd == 0 {
+		fd, err := syscall.Open("/dev/urandom", syscall.O_RDONLY, 0)
+		if err != nil {
+			return 0, err
+		}
+		r.fd = fd
+	}
+
+	// Read from the file.
+	return syscall.Read(r.fd, b)
+}

src/crypto/rand/util.go

@@ -0,0 +1,143 @@
+// Copyright 2011 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package rand
+
+import (
+	"errors"
+	"io"
+	"math/big"
+)
+
+// smallPrimes is a list of small, prime numbers that allows us to rapidly
+// exclude some fraction of composite candidates when searching for a random
+// prime. This list is truncated at the point where smallPrimesProduct exceeds
+// a uint64. It does not include two because we ensure that the candidates are
+// odd by construction.
+var smallPrimes = []uint8{
+	3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53,
+}
+
+// smallPrimesProduct is the product of the values in smallPrimes and allows us
+// to reduce a candidate prime by this number and then determine whether it's
+// coprime to all the elements of smallPrimes without further big.Int
+// operations.
+var smallPrimesProduct = new(big.Int).SetUint64(16294579238595022365)
+
+// Prime returns a number, p, of the given size, such that p is prime
+// with high probability.
+// Prime will return error for any error returned by rand.Read or if bits < 2.
+func Prime(rand io.Reader, bits int) (p *big.Int, err error) {
+	if bits < 2 {
+		err = errors.New("crypto/rand: prime size must be at least 2-bit")
+		return
+	}
+
+	b := uint(bits % 8)
+	if b == 0 {
+		b = 8
+	}
+
+	bytes := make([]byte, (bits+7)/8)
+	p = new(big.Int)
+
+	bigMod := new(big.Int)
+
+	for {
+		_, err = io.ReadFull(rand, bytes)
+		if err != nil {
+			return nil, err
+		}
+
+		// Clear bits in the first byte to make sure the candidate has a size <= bits.
+		bytes[0] &= uint8(int(1<<b) - 1)
+		// Don't let the value be too small, i.e, set the most significant two bits.
+		// Setting the top two bits, rather than just the top bit,
+		// means that when two of these values are multiplied together,
+		// the result isn't ever one bit short.
+		if b >= 2 {
+			bytes[0] |= 3 << (b - 2)
+		} else {
+			// Here b==1, because b cannot be zero.
+			bytes[0] |= 1
+			if len(bytes) > 1 {
+				bytes[1] |= 0x80
+			}
+		}
+		// Make the value odd since an even number this large certainly isn't prime.
+		bytes[len(bytes)-1] |= 1
+
+		p.SetBytes(bytes)
+
+		// Calculate the value mod the product of smallPrimes. If it's
+		// a multiple of any of these primes we add two until it isn't.
+		// The probability of overflowing is minimal and can be ignored
+		// because we still perform Miller-Rabin tests on the result.
+		bigMod.Mod(p, smallPrimesProduct)
+		mod := bigMod.Uint64()
+
+	NextDelta:
+		for delta := uint64(0); delta < 1<<20; delta += 2 {
+			m := mod + delta
+			for _, prime := range smallPrimes {
+				if m%uint64(prime) == 0 && (bits > 6 || m != uint64(prime)) {
+					continue NextDelta
+				}
+			}
+
+			if delta > 0 {
+				bigMod.SetUint64(delta)
+				p.Add(p, bigMod)
+			}
+			break
+		}
+
+		// There is a tiny possibility that, by adding delta, we caused
+		// the number to be one bit too long. Thus we check BitLen
+		// here.
+		if p.ProbablyPrime(20) && p.BitLen() == bits {
+			return
+		}
+	}
+}
+
+// Int returns a uniform random value in [0, max). It panics if max <= 0.
+func Int(rand io.Reader, max *big.Int) (n *big.Int, err error) {
+	if max.Sign() <= 0 {
+		panic("crypto/rand: argument to Int is <= 0")
+	}
+	n = new(big.Int)
+	n.Sub(max, n.SetUint64(1))
+	// bitLen is the maximum bit length needed to encode a value < max.
+	bitLen := n.BitLen()
+	if bitLen == 0 {
+		// the only valid result is 0
+		return
+	}
+	// k is the maximum byte length needed to encode a value < max.
+	k := (bitLen + 7) / 8
+	// b is the number of bits in the most significant byte of max-1.
+	b := uint(bitLen % 8)
+	if b == 0 {
+		b = 8
+	}
+
+	bytes := make([]byte, k)
+
+	for {
+		_, err = io.ReadFull(rand, bytes)
+		if err != nil {
+			return nil, err
+		}
+
+		// Clear bits in the first byte to increase the probability
+		// that the candidate is < max.
+		bytes[0] &= uint8(int(1<<b) - 1)
+
+		n.SetBytes(bytes)
+		if n.Cmp(max) < 0 {
+			return
+		}
+	}
+}

src/device/arm/scb.go

@@ -26,12 +26,21 @@ type SCB_Type struct {
 	SHPR2 volatile.Register32 // 0xD1C: System Handler Priority Register 2
 	SHPR3 volatile.Register32 // 0xD20: System Handler Priority Register 3
 	// the following are only applicable for Cortex-M3/M33/M4/M7
-	SHCSR volatile.Register32 // 0xD24: System Handler Control and State Register
-	CFSR  volatile.Register32 // 0xD28: Configurable Fault Status Register
-	HFSR  volatile.Register32 // 0xD2C: HardFault Status Register
-	DFSR  volatile.Register32 // 0xD30: Debug Fault Status Register
-	MMFAR volatile.Register32 // 0xD34: MemManage Fault Address Register
-	BFAR  volatile.Register32 // 0xD38: BusFault Address Register
+	SHCSR volatile.Register32    // 0xD24: System Handler Control and State Register
+	CFSR  volatile.Register32    // 0xD28: Configurable Fault Status Register
+	HFSR  volatile.Register32    // 0xD2C: HardFault Status Register
+	DFSR  volatile.Register32    // 0xD30: Debug Fault Status Register
+	MMFAR volatile.Register32    // 0xD34: MemManage Fault Address Register
+	BFAR  volatile.Register32    // 0xD38: BusFault Address Register
+	AFSR  volatile.Register32    // 0xD3C: Auxiliary Fault Status Register
+	PFR   [2]volatile.Register32 // 0xD40: Processor Feature Register
+	DFR   volatile.Register32    // 0xD48: Debug Feature Register
+	ADR   volatile.Register32    // 0xD4C: Auxiliary Feature Register
+	MMFR  [4]volatile.Register32 // 0xD50: Memory Model Feature Register
+	ISAR  [5]volatile.Register32 // 0xD60: Instruction Set Attributes Register
+	_     [5]uint32              // reserved
+	CPACR volatile.Register32    // 0xD88: Coprocessor Access Control Register
+
 }
 
 var SCB = (*SCB_Type)(unsafe.Pointer(uintptr(SCB_BASE)))

src/device/sam/atsame5x-bitfields.go

@@ -0,0 +1,71 @@
+// Hand created file. DO NOT DELETE.
+// atsamd51x bitfield definitions that are not auto-generated by gen-device-svd.go
+
+// +build sam,atsame5x
+
+// These are the supported pchctrl function numberings on the atsamd51x
+// See http://ww1.microchip.com/downloads/en/DeviceDoc/SAM_D5xE5x_Family_Data_Sheet_DS60001507F.pdf
+// table 14-9
+
+package sam
+
+const (
+	PCHCTRL_GCLK_OSCCTRL_DFLL48     = 0 // DFLL48 input clock source
+	PCHCTRL_GCLK_OSCCTRL_FDPLL0     = 1 // Reference clock for FDPLL0
+	PCHCTRL_GCLK_OSCCTRL_FDPLL1     = 2 // Reference clock for FDPLL1
+	PCHCTRL_GCLK_OSCCTRL_FDPLL0_32K = 3 // FDPLL0 = 3 // 32KHz clock for internal lock timer
+	PCHCTRL_GCLK_OSCCTRL_FDPLL1_32K = 3 // FDPLL1 = 3 // 32KHz clock for internal lock timer
+	PCHCTRL_GCLK_SDHC0_SLOW         = 3 // SDHC0 = 3 // Slow
+	PCHCTRL_GCLK_SDHC1_SLOW         = 3 // SDHC1 = 3 // Slow
+	PCHCTRL_GCLK_SERCOMX_SLOW       = 3 // GCLK_SERCOM[0..7]_SLOW = 3
+	PCHCTRL_GCLK_EIC                = 4
+	PCHCTRL_GCLK_FREQM_MSR          = 5 // FREQM Measure
+	PCHCTRL_GCLK_FREQM_REF          = 6 // FREQM Reference
+	PCHCTRL_GCLK_SERCOM0_CORE       = 7 // SERCOM0 Core
+	PCHCTRL_GCLK_SERCOM1_CORE       = 8 // SERCOM1 Core
+	PCHCTRL_GCLK_TC0                = 9
+	PCHCTRL_GCLK_TC1                = 9  // TC0, TC1
+	PCHCTRL_GCLK_USB                = 10 // USB
+	PCHCTRL_GCLK_EVSYS0             = 11
+	PCHCTRL_GCLK_EVSYS1             = 12
+	PCHCTRL_GCLK_EVSYS2             = 13
+	PCHCTRL_GCLK_EVSYS3             = 14
+	PCHCTRL_GCLK_EVSYS4             = 15
+	PCHCTRL_GCLK_EVSYS5             = 16
+	PCHCTRL_GCLK_EVSYS6             = 17
+	PCHCTRL_GCLK_EVSYS7             = 18
+	PCHCTRL_GCLK_EVSYS8             = 19
+	PCHCTRL_GCLK_EVSYS9             = 20
+	PCHCTRL_GCLK_EVSYS10            = 21
+	PCHCTRL_GCLK_EVSYS11            = 22
+	PCHCTRL_GCLK_SERCOM2_CORE       = 23 // SERCOM2 Core
+	PCHCTRL_GCLK_SERCOM3_CORE       = 24 // SERCOM3 Core
+	PCHCTRL_GCLK_TCC0               = 25
+	PCHCTRL_GCLK_TCC1               = 25 // TCC0, TCC1
+	PCHCTRL_GCLK_TC2                = 26
+	PCHCTRL_GCLK_TC3                = 26 // TC2, TC3
+	PCHCTRL_GCLK_CAN0               = 27 // CAN0
+	PCHCTRL_GCLK_CAN1               = 28 // CAN1
+	PCHCTRL_GCLK_TCC2               = 29
+	PCHCTRL_GCLK_TCC3               = 29 // TCC2, TCC3
+	PCHCTRL_GCLK_TC4                = 30
+	PCHCTRL_GCLK_TC5                = 30 // TC4, TC5
+	PCHCTRL_GCLK_PDEC               = 31 // PDEC
+	PCHCTRL_GCLK_AC                 = 32 // AC
+	PCHCTRL_GCLK_CCL                = 33 // CCL
+	PCHCTRL_GCLK_SERCOM4_CORE       = 34 // SERCOM4 Core
+	PCHCTRL_GCLK_SERCOM5_CORE       = 35 // SERCOM5 Core
+	PCHCTRL_GCLK_SERCOM6_CORE       = 36 // SERCOM6 Core
+	PCHCTRL_GCLK_SERCOM7_CORE       = 37 // SERCOM7 Core
+	PCHCTRL_GCLK_TCC4               = 38 // TCC4
+	PCHCTRL_GCLK_TC6                = 39
+	PCHCTRL_GCLK_TC7                = 39 // TC6, TC7
+	PCHCTRL_GCLK_ADC0               = 40 // ADC0
+	PCHCTRL_GCLK_ADC1               = 41 // ADC1
+	PCHCTRL_GCLK_DAC                = 42 // DAC
+	PCHCTRL_GCLK_I2S0               = 43
+	PCHCTRL_GCLK_I2S1               = 44
+	PCHCTRL_GCLK_SDHC0              = 45 // SDHC0
+	PCHCTRL_GCLK_SDHC1              = 46 // SDHC1
+	PCHCTRL_GCLK_CM4_TRACE          = 47 // CM4 Trace
+)

src/examples/can/feather-m4-can.go

@@ -0,0 +1,15 @@
+// +build feather_m4_can
+
+package main
+
+import (
+	"machine"
+)
+
+func init() {
+	// power on the CAN Transceiver
+	// https://learn.adafruit.com/adafruit-feather-m4-can-express/pinouts#can-bus-3078990-8
+	boost_en := machine.BOOST_EN
+	boost_en.Configure(machine.PinConfig{Mode: machine.PinOutput})
+	boost_en.High()
+}

src/examples/can/main.go

@@ -0,0 +1,53 @@
+package main
+
+import (
+	"fmt"
+	"machine"
+	"time"
+)
+
+func main() {
+	can1 := machine.CAN1
+	can1.Configure(machine.CANConfig{
+		TransferRate:   machine.CANTransferRate500kbps,
+		TransferRateFD: machine.CANTransferRate1000kbps,
+		Rx:             machine.CAN1_RX,
+		Tx:             machine.CAN1_TX,
+		Standby:        machine.CAN1_STANDBY,
+	})
+
+	can0 := machine.CAN0
+	can0.Configure(machine.CANConfig{
+		TransferRate:   machine.CANTransferRate500kbps,
+		TransferRateFD: machine.CANTransferRate1000kbps,
+		Rx:             machine.CAN0_RX,
+		Tx:             machine.CAN0_TX,
+		Standby:        machine.NoPin,
+	})
+
+	rxMsg := machine.CANRxBufferElement{}
+
+	for {
+		can1.Tx(0x123, []byte{0x01, 0x23, 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF}, false, false)
+		can1.Tx(0x789, []byte{0x02, 0x24, 0x46, 0x67, 0x89, 0xAB, 0xCD, 0xEF, 0x01, 0x23, 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF}, true, false)
+		time.Sleep(time.Millisecond * 1000)
+
+		sz0 := can0.RxFifoSize()
+		if sz0 > 0 {
+			fmt.Printf("CAN0 %d\r\n", sz0)
+			for i := 0; i < sz0; i++ {
+				can0.RxRaw(&rxMsg)
+				fmt.Printf("-> %08X %X %#v\r\n", rxMsg.ID, rxMsg.DLC, rxMsg.Data())
+			}
+		}
+
+		sz1 := can1.RxFifoSize()
+		if sz1 > 0 {
+			fmt.Printf("CAN1 %d\r\n", sz1)
+			for i := 0; i < sz1; i++ {
+				can1.RxRaw(&rxMsg)
+				fmt.Printf("-> %08X %X %#v\r\n", rxMsg.ID, rxMsg.DLC, rxMsg.Data())
+			}
+		}
+	}
+}

src/examples/caninterrupt/feather-m4-can.go

@@ -0,0 +1,15 @@
+// +build feather_m4_can
+
+package main
+
+import (
+	"machine"
+)
+
+func init() {
+	// power on the CAN Transceiver
+	// https://learn.adafruit.com/adafruit-feather-m4-can-express/pinouts#can-bus-3078990-8
+	boost_en := machine.BOOST_EN
+	boost_en.Configure(machine.PinConfig{Mode: machine.PinOutput})
+	boost_en.High()
+}

src/examples/caninterrupt/main.go

@@ -0,0 +1,75 @@
+package main
+
+import (
+	"device/sam"
+	"fmt"
+	"machine"
+	"time"
+)
+
+type canMsg struct {
+	ch   byte
+	id   uint32
+	dlc  byte
+	data []byte
+}
+
+func main() {
+	ch := make(chan canMsg, 10)
+	go func() {
+		for {
+			select {
+			case m := <-ch:
+				fmt.Printf("%d %03X %X ", m.ch, m.id, m.dlc)
+				for _, d := range m.data {
+					fmt.Printf("%02X ", d)
+				}
+				fmt.Printf("\r\n")
+			}
+
+		}
+	}()
+
+	can1 := machine.CAN1
+	can1.Configure(machine.CANConfig{
+		TransferRate:   machine.CANTransferRate500kbps,
+		TransferRateFD: machine.CANTransferRate1000kbps,
+		Rx:             machine.CAN1_RX,
+		Tx:             machine.CAN1_TX,
+		Standby:        machine.CAN1_STANDBY,
+	})
+	// RF0NE : Rx FIFO 0 New Message Interrupt Enable
+	can1.SetInterrupt(sam.CAN_IE_RF0NE, func(*machine.CAN) {
+		rxMsg := machine.CANRxBufferElement{}
+		can1.RxRaw(&rxMsg)
+		m := canMsg{ch: 1, id: rxMsg.ID, dlc: rxMsg.DLC, data: rxMsg.Data()}
+		select {
+		case ch <- m:
+		}
+	})
+
+	can0 := machine.CAN0
+	can0.Configure(machine.CANConfig{
+		TransferRate:   machine.CANTransferRate500kbps,
+		TransferRateFD: machine.CANTransferRate1000kbps,
+		Rx:             machine.CAN0_RX,
+		Tx:             machine.CAN0_TX,
+		Standby:        machine.NoPin,
+	})
+	// RF0NE : Rx FIFO 0 New Message Interrupt Enable
+	can0.SetInterrupt(sam.CAN_IE_RF0NE, func(*machine.CAN) {
+		rxMsg := machine.CANRxBufferElement{}
+		can0.RxRaw(&rxMsg)
+		m := canMsg{ch: 2, id: rxMsg.ID, dlc: rxMsg.DLC, data: rxMsg.Data()}
+		select {
+		case ch <- m:
+		}
+	})
+
+	for {
+		can0.Tx(0x123, []byte{0x01, 0x23, 0x45, 0x67, 0x89, 0xAB, 0xCD, 0xEF}, false, false)
+		time.Sleep(time.Millisecond * 500)
+		can1.Tx(0x456, []byte{0xAA, 0xBB, 0xCC}, false, false)
+		time.Sleep(time.Millisecond * 1000)
+	}
+}

src/examples/echo/echo.go

@@ -9,7 +9,7 @@ import (
 
 // change these to test a different UART or pins if available
 var (
-	uart = machine.UART0
+	uart = machine.Serial
 	tx   = machine.UART_TX_PIN
 	rx   = machine.UART_RX_PIN
 )

src/examples/memstats/memstats.go

@@ -0,0 +1,32 @@
+package main
+
+import (
+	"math/rand"
+	"runtime"
+	"time"
+)
+
+func main() {
+
+	ms := runtime.MemStats{}
+
+	for {
+		escapesToHeap()
+		runtime.ReadMemStats(&ms)
+		println("Heap before GC. Used: ", ms.HeapInuse, " Free: ", ms.HeapIdle, " Meta: ", ms.GCSys)
+		runtime.GC()
+		runtime.ReadMemStats(&ms)
+		println("Heap after  GC. Used: ", ms.HeapInuse, " Free: ", ms.HeapIdle, " Meta: ", ms.GCSys)
+		time.Sleep(5 * time.Second)
+	}
+
+}
+
+func escapesToHeap() {
+	n := rand.Intn(100)
+	println("Doing ", n, " iterations")
+	for i := 0; i < n; i++ {
+		s := make([]byte, i)
+		_ = append(s, 42)
+	}
+}

src/examples/pininterrupt/stm32.go

@@ -0,0 +1,10 @@
+// +build stm32
+
+package main
+
+import "machine"
+
+const (
+	buttonMode      = machine.PinInputPulldown
+	buttonPinChange = machine.PinRising | machine.PinFalling
+)

src/examples/pwm/arduino-mega1280.go

@@ -0,0 +1,12 @@
+// +build arduino_mega1280
+
+package main
+
+import "machine"
+
+var (
+	// Configuration on an Arduino Uno.
+	pwm  = machine.Timer3
+	pinA = machine.PH3 // pin 6 on the Mega
+	pinB = machine.PH4 // pin 7 on the Mega
+)

src/examples/pwm/arduino.go

@@ -0,0 +1,12 @@
+// +build arduino
+
+package main
+
+import "machine"
+
+var (
+	// Configuration on an Arduino Uno.
+	pwm  = machine.Timer2
+	pinA = machine.PB3 // pin 11 on the Uno
+	pinB = machine.PD3 // pin 3 on the Uno
+)

src/examples/pwm/bluepill.go

@@ -0,0 +1,11 @@
+// +build bluepill
+
+package main
+
+import "machine"
+
+var (
+	pwm  = &machine.TIM2
+	pinA = machine.PA0
+	pinB = machine.PA1
+)

src/examples/pwm/feather-m4.go

@@ -0,0 +1,11 @@
+// +build feather_m4
+
+package main
+
+import "machine"
+
+var (
+	pwm  = machine.TCC0
+	pinA = machine.D12
+	pinB = machine.D13
+)

src/examples/pwm/itsybitsy-m0.go

@@ -0,0 +1,11 @@
+// +build itsybitsy_m0
+
+package main
+
+import "machine"
+
+var (
+	pwm  = machine.TCC0
+	pinA = machine.D3
+	pinB = machine.D4
+)

src/examples/pwm/itsybitsy-m4.go

@@ -0,0 +1,11 @@
+// +build itsybitsy_m4
+
+package main
+
+import "machine"
+
+var (
+	pwm  = machine.TCC0
+	pinA = machine.D12
+	pinB = machine.D13
+)

src/examples/pwm/nucleo-f722ze.go

@@ -0,0 +1,11 @@
+// +build stm32f7
+
+package main
+
+import "machine"
+
+var (
+	pwm  = &machine.TIM1
+	pinA = machine.PA8
+	pinB = machine.PA9
+)

src/examples/pwm/nucleo-l031k6.go

@@ -0,0 +1,11 @@
+// +build stm32l0
+
+package main
+
+import "machine"
+
+var (
+	pwm  = &machine.TIM2
+	pinA = machine.PA0
+	pinB = machine.PB3
+)

src/examples/pwm/nucleo-l432kc.go

@@ -0,0 +1,11 @@
+// +build stm32l4
+
+package main
+
+import "machine"
+
+var (
+	pwm  = &machine.TIM2
+	pinA = machine.PA0
+	pinB = machine.PB3
+)

src/examples/pwm/nucleo-l552ze.go

@@ -0,0 +1,11 @@
+// +build stm32l5
+
+package main
+
+import "machine"
+
+var (
+	pwm  = &machine.TIM1
+	pinA = machine.PA8
+	pinB = machine.PA9
+)

src/examples/pwm/pwm.go

@@ -1,64 +1,74 @@
 package main
 
+// This example demonstrates some features of the PWM support.
+
 import (
 	"machine"
 	"time"
 )
 
-// This example assumes that an RGB LED is connected to pins 3, 5 and 6 on an Arduino.
-// Change the values below to use different pins.
-const (
-	redPin   = machine.D4
-	greenPin = machine.D5
-	bluePin  = machine.D6
-)
-
-// cycleColor is just a placeholder until math/rand or some equivalent is working.
-func cycleColor(color uint8) uint8 {
-	if color < 10 {
-		return color + 1
-	} else if color < 200 {
-		return color + 10
-	} else {
-		return 0
-	}
-}
+const delayBetweenPeriods = time.Second * 5
 
 func main() {
-	machine.InitPWM()
+	// Delay a bit on startup to easily catch the first messages.
+	time.Sleep(time.Second * 2)
 
-	red := machine.PWM{redPin}
-	err := red.Configure()
-	checkError(err, "failed to configure red pin")
+	// Configure the PWM with the given period.
+	err := pwm.Configure(machine.PWMConfig{
+		Period: 16384e3, // 16.384ms
+	})
+	if err != nil {
+		println("failed to configure PWM")
+		return
+	}
 
-	green := machine.PWM{greenPin}
-	err = green.Configure()
-	checkError(err, "failed to configure green pin")
+	// The top value is the highest value that can be passed to PWMChannel.Set.
+	// It is usually an even number.
+	println("top:", pwm.Top())
 
-	blue := machine.PWM{bluePin}
-	err = blue.Configure()
-	checkError(err, "failed to configure blue pin")
+	// Configure the two channels we'll use as outputs.
+	channelA, err := pwm.Channel(pinA)
+	if err != nil {
+		println("failed to configure channel A")
+		return
+	}
+	channelB, err := pwm.Channel(pinB)
+	if err != nil {
+		println("failed to configure channel B")
+		return
+	}
 
-	var rc uint8
-	var gc uint8 = 20
-	var bc uint8 = 30
+	// Invert one of the channels to demonstrate output polarity.
+	pwm.SetInverting(channelB, true)
+
+	// Test out various frequencies below, including some edge cases.
+
+	println("running at 0% duty cycle")
+	pwm.Set(channelA, 0)
+	pwm.Set(channelB, 0)
+	time.Sleep(delayBetweenPeriods)
+
+	println("running at 1")
+	pwm.Set(channelA, 1)
+	pwm.Set(channelB, 1)
+	time.Sleep(delayBetweenPeriods)
+
+	println("running at 25% duty cycle")
+	pwm.Set(channelA, pwm.Top()/4)
+	pwm.Set(channelB, pwm.Top()/4)
+	time.Sleep(delayBetweenPeriods)
+
+	println("running at top-1")
+	pwm.Set(channelA, pwm.Top()-1)
+	pwm.Set(channelB, pwm.Top()-1)
+	time.Sleep(delayBetweenPeriods)
+
+	println("running at 100% duty cycle")
+	pwm.Set(channelA, pwm.Top())
+	pwm.Set(channelB, pwm.Top())
+	time.Sleep(delayBetweenPeriods)
 
 	for {
-		rc = cycleColor(rc)
-		gc = cycleColor(gc)
-		bc = cycleColor(bc)
-
-		red.Set(uint16(rc) << 8)
-		green.Set(uint16(gc) << 8)
-		blue.Set(uint16(bc) << 8)
-
-		time.Sleep(time.Millisecond * 500)
-	}
-}
-
-func checkError(err error, msg string) {
-	if err != nil {
-		print(msg, ": ", err.Error())
-		println()
+		time.Sleep(time.Second)
 	}
 }

src/examples/pwm/stm32f4disco.go

@@ -0,0 +1,13 @@
+// +build stm32f4disco
+
+package main
+
+import "machine"
+
+var (
+	// These pins correspond to LEDs on the discovery
+	// board
+	pwm  = &machine.TIM4
+	pinA = machine.PD12
+	pinB = machine.PD13
+)

src/internal/task/task.go

@@ -13,7 +13,7 @@ type Task struct {
 	Ptr unsafe.Pointer
 
 	// Data is a field which can be used for storing state information.
-	Data uint
+	Data uint64
 
 	// state is the underlying running state of the task.
 	state state

src/internal/task/task_stack_386.S

@@ -0,0 +1,58 @@
+.section .text.tinygo_startTask
+.global  tinygo_startTask
+.type    tinygo_startTask, %function
+tinygo_startTask:
+    .cfi_startproc
+    // Small assembly stub for starting a goroutine. This is already run on the
+    // new stack, with the callee-saved registers already loaded.
+    // Most importantly, EBX contain the pc of the to-be-started function and
+    // ESI contain the only argument it is given. Multiple arguments are packed
+    // into one by storing them in a new allocation.
+
+    // Indicate to the unwinder that there is nothing to unwind, this is the
+    // root frame. It avoids bogus extra frames in GDB.
+    .cfi_undefined eip
+
+    // Set the first argument of the goroutine start wrapper, which contains all
+    // the arguments.
+    pushl %esi
+
+    // Branch to the "goroutine start" function.
+    calll *%ebx
+
+    // Rebalance the stack (to undo the above push).
+    addl $4, %esp
+
+    // After return, exit this goroutine. This is a tail call.
+    jmp tinygo_pause
+    .cfi_endproc
+
+.global tinygo_swapTask
+.type tinygo_swapTask, %function
+tinygo_swapTask:
+    // This function gets the following parameters:
+    movl 4(%esp), %eax // newStack uintptr
+    movl 8(%esp), %ecx // oldStack *uintptr
+    // More information on the calling convention:
+    // https://wiki.osdev.org/System_V_ABI#i386
+
+    // Save all callee-saved registers:
+    pushl %ebp
+    pushl %edi
+    pushl %esi
+    pushl %ebx
+
+    // Save the current stack pointer in oldStack.
+    movl %esp, (%ecx)
+
+    // Switch to the new stack pointer.
+    movl %eax, %esp
+
+    // Load saved register from the new stack.
+    popl %ebx
+    popl %esi
+    popl %edi
+    popl %ebp
+
+    // Return into the new task, as if tinygo_swapTask was a regular call.
+    ret

src/internal/task/task_stack_386.go

@@ -0,0 +1,59 @@
+// +build scheduler.tasks,386
+
+package task
+
+import "unsafe"
+
+var systemStack uintptr
+
+// calleeSavedRegs is the list of registers that must be saved and restored when
+// switching between tasks. Also see task_stack_386.S that relies on the exact
+// layout of this struct.
+type calleeSavedRegs struct {
+	ebx uintptr
+	esi uintptr
+	edi uintptr
+	ebp uintptr
+
+	pc uintptr
+}
+
+// archInit runs architecture-specific setup for the goroutine startup.
+func (s *state) archInit(r *calleeSavedRegs, fn uintptr, args unsafe.Pointer) {
+	// Store the initial sp for the startTask function (implemented in assembly).
+	s.sp = uintptr(unsafe.Pointer(r))
+
+	// Initialize the registers.
+	// These will be popped off of the stack on the first resume of the goroutine.
+
+	// Start the function at tinygo_startTask (defined in
+	// src/internal/task/task_stack_386.S). This assembly code calls a function
+	// (passed in EBX) with a single argument (passed in ESI). After the
+	// function returns, it calls Pause().
+	r.pc = uintptr(unsafe.Pointer(&startTask))
+
+	// Pass the function to call in EBX.
+	// This function is a compiler-generated wrapper which loads arguments out
+	// of a struct pointer. See createGoroutineStartWrapper (defined in
+	// compiler/goroutine.go) for more information.
+	r.ebx = fn
+
+	// Pass the pointer to the arguments struct in ESI.
+	r.esi = uintptr(args)
+}
+
+func (s *state) resume() {
+	swapTask(s.sp, &systemStack)
+}
+
+func (s *state) pause() {
+	newStack := systemStack
+	systemStack = 0
+	swapTask(newStack, &s.sp)
+}
+
+// SystemStack returns the system stack pointer when called from a task stack.
+// When called from the system stack, it returns 0.
+func SystemStack() uintptr {
+	return systemStack
+}

src/internal/task/task_stack_amd64.S

@@ -0,0 +1,74 @@
+#ifdef __MACH__ // Darwin
+.global  _tinygo_startTask
+_tinygo_startTask:
+#else // Linux etc
+.section .text.tinygo_startTask
+.global  tinygo_startTask
+tinygo_startTask:
+#endif
+    .cfi_startproc
+    // Small assembly stub for starting a goroutine. This is already run on the
+    // new stack, with the callee-saved registers already loaded.
+    // Most importantly, r12 contain the pc of the to-be-started function and
+    // r13 contain the only argument it is given. Multiple arguments are packed
+    // into one by storing them in a new allocation.
+
+    // Indicate to the unwinder that there is nothing to unwind, this is the
+    // root frame. It avoids bogus extra frames in GDB like here:
+    //     #10 0x00000000004277b6 in <goroutine wrapper> () at [...]
+    //     #11 0x00000000004278f3 in tinygo_startTask () at [...]
+    //     #12 0x0000000000002030 in ?? ()
+    //     #13 0x0000000000000071 in ?? ()
+    .cfi_undefined rip
+
+    // Set the first argument of the goroutine start wrapper, which contains all
+    // the arguments.
+    movq %r13, %rdi
+
+    // Branch to the "goroutine start" function.
+    callq *%r12
+
+    // After return, exit this goroutine. This is a tail call.
+    #ifdef __MACH__
+    jmp _tinygo_pause
+    #else
+    jmp tinygo_pause
+    #endif
+    .cfi_endproc
+
+#ifdef __MACH__ // Darwin
+.global _tinygo_swapTask
+_tinygo_swapTask:
+#else // Linux etc
+.global tinygo_swapTask
+.section .text.tinygo_swapTask
+tinygo_swapTask:
+#endif
+    // This function gets the following parameters:
+    // %rdi = newStack uintptr
+    // %rsi = oldStack *uintptr
+
+    // Save all callee-saved registers:
+    pushq %r15
+    pushq %r14
+    pushq %r13
+    pushq %r12
+    pushq %rbp
+    pushq %rbx
+
+    // Save the current stack pointer in oldStack.
+    movq %rsp, (%rsi)
+
+    // Switch to the new stack pointer.
+    movq %rdi, %rsp
+
+    // Load saved register from the new stack.
+    popq %rbx
+    popq %rbp
+    popq %r12
+    popq %r13
+    popq %r14
+    popq %r15
+
+    // Return into the new task, as if tinygo_swapTask was a regular call.
+    ret

src/internal/task/task_stack_amd64.go

@@ -0,0 +1,61 @@
+// +build scheduler.tasks,amd64
+
+package task
+
+import "unsafe"
+
+var systemStack uintptr
+
+// calleeSavedRegs is the list of registers that must be saved and restored when
+// switching between tasks. Also see task_stack_amd64.S that relies on the exact
+// layout of this struct.
+type calleeSavedRegs struct {
+	rbx uintptr
+	rbp uintptr
+	r12 uintptr
+	r13 uintptr
+	r14 uintptr
+	r15 uintptr
+
+	pc uintptr
+}
+
+// archInit runs architecture-specific setup for the goroutine startup.
+func (s *state) archInit(r *calleeSavedRegs, fn uintptr, args unsafe.Pointer) {
+	// Store the initial sp for the startTask function (implemented in assembly).
+	s.sp = uintptr(unsafe.Pointer(r))
+
+	// Initialize the registers.
+	// These will be popped off of the stack on the first resume of the goroutine.
+
+	// Start the function at tinygo_startTask (defined in
+	// src/internal/task/task_stack_amd64.S). This assembly code calls a
+	// function (passed in r12) with a single argument (passed in r13). After
+	// the function returns, it calls Pause().
+	r.pc = uintptr(unsafe.Pointer(&startTask))
+
+	// Pass the function to call in r12.
+	// This function is a compiler-generated wrapper which loads arguments out
+	// of a struct pointer. See createGoroutineStartWrapper (defined in
+	// compiler/goroutine.go) for more information.
+	r.r12 = fn
+
+	// Pass the pointer to the arguments struct in r13.
+	r.r13 = uintptr(args)
+}
+
+func (s *state) resume() {
+	swapTask(s.sp, &systemStack)
+}
+
+func (s *state) pause() {
+	newStack := systemStack
+	systemStack = 0
+	swapTask(newStack, &s.sp)
+}
+
+// SystemStack returns the system stack pointer when called from a task stack.
+// When called from the system stack, it returns 0.
+func SystemStack() uintptr {
+	return systemStack
+}

src/internal/task/task_stack_arm.S

@@ -0,0 +1,51 @@
+// Only generate .debug_frame, don't generate .eh_frame.
+.cfi_sections .debug_frame
+
+.section .text.tinygo_startTask
+.global  tinygo_startTask
+.type    tinygo_startTask, %function
+tinygo_startTask:
+    .cfi_startproc
+    // Small assembly stub for starting a goroutine. This is already run on the
+    // new stack, with the callee-saved registers already loaded.
+    // Most importantly, r4 contains the pc of the to-be-started function and r5
+    // contains the only argument it is given. Multiple arguments are packed
+    // into one by storing them in a new allocation.
+
+    // Indicate to the unwinder that there is nothing to unwind, this is the
+    // root frame. It avoids the following (bogus) error message in GDB:
+    //     Backtrace stopped: previous frame identical to this frame (corrupt stack?)
+    .cfi_undefined lr
+
+    // Set the first argument of the goroutine start wrapper, which contains all
+    // the arguments.
+    mov   r0, r5
+
+    // Branch to the "goroutine start" function. By using blx instead of bx,
+    // we'll return here instead of tail calling.
+    blx   r4
+
+    // After return, exit this goroutine. This is a tail call.
+    bl    tinygo_pause
+    .cfi_endproc
+.size tinygo_startTask, .-tinygo_startTask
+
+.global tinygo_swapTask
+.type tinygo_swapTask, %function
+tinygo_swapTask:
+    // This function gets the following parameters:
+    // r0 = newStack uintptr
+    // r1 = oldStack *uintptr
+
+    // Save all callee-saved registers:
+    push {r4-r11, lr}
+
+    // Save the current stack pointer in oldStack.
+    str sp, [r1]
+
+    // Switch to the new stack pointer.
+    mov sp, r0
+
+    // Load state from new task and branch to the previous position in the
+    // program.
+    pop {r4-r11, pc}

src/internal/task/task_stack_arm.go

@@ -0,0 +1,61 @@
+// +build scheduler.tasks,arm,!cortexm,!avr,!xtensa
+
+package task
+
+import "unsafe"
+
+var systemStack uintptr
+
+// calleeSavedRegs is the list of registers that must be saved and restored when
+// switching between tasks. Also see task_stack_arm.S that relies on the exact
+// layout of this struct.
+type calleeSavedRegs struct {
+	r4  uintptr
+	r5  uintptr
+	r6  uintptr
+	r7  uintptr
+	r8  uintptr
+	r9  uintptr
+	r10 uintptr
+	r11 uintptr
+
+	pc uintptr
+}
+
+// archInit runs architecture-specific setup for the goroutine startup.
+func (s *state) archInit(r *calleeSavedRegs, fn uintptr, args unsafe.Pointer) {
+	// Store the initial sp for the startTask function (implemented in assembly).
+	s.sp = uintptr(unsafe.Pointer(r))
+
+	// Initialize the registers.
+	// These will be popped off of the stack on the first resume of the goroutine.
+
+	// Start the function at tinygo_startTask (defined in src/internal/task/task_stack_arm.S).
+	// This assembly code calls a function (passed in r4) with a single argument
+	// (passed in r5). After the function returns, it calls Pause().
+	r.pc = uintptr(unsafe.Pointer(&startTask))
+
+	// Pass the function to call in r4.
+	// This function is a compiler-generated wrapper which loads arguments out of a struct pointer.
+	// See createGoroutineStartWrapper (defined in compiler/goroutine.go) for more information.
+	r.r4 = fn
+
+	// Pass the pointer to the arguments struct in r5.
+	r.r5 = uintptr(args)
+}
+
+func (s *state) resume() {
+	swapTask(s.sp, &systemStack)
+}
+
+func (s *state) pause() {
+	newStack := systemStack
+	systemStack = 0
+	swapTask(newStack, &s.sp)
+}
+
+// SystemStack returns the system stack pointer when called from a task stack.
+// When called from the system stack, it returns 0.
+func SystemStack() uintptr {
+	return systemStack
+}

src/internal/task/task_stack_arm64.S

@@ -0,0 +1,59 @@
+.section .text.tinygo_startTask
+.global  tinygo_startTask
+.type    tinygo_startTask, %function
+tinygo_startTask:
+    .cfi_startproc
+    // Small assembly stub for starting a goroutine. This is already run on the
+    // new stack, with the callee-saved registers already loaded.
+    // Most importantly, x19 contains the pc of the to-be-started function and
+    // x20 contains the only argument it is given. Multiple arguments are packed
+    // into one by storing them in a new allocation.
+
+    // Indicate to the unwinder that there is nothing to unwind, this is the
+    // root frame. It avoids the following (bogus) error message in GDB:
+    //     Backtrace stopped: previous frame identical to this frame (corrupt stack?)
+    .cfi_undefined lr
+
+    // Set the first argument of the goroutine start wrapper, which contains all
+    // the arguments.
+    mov   x0, x20
+
+    // Branch to the "goroutine start" function. By using blx instead of bx,
+    // we'll return here instead of tail calling.
+    blr   x19
+
+    // After return, exit this goroutine. This is a tail call.
+    b     tinygo_pause
+    .cfi_endproc
+.size tinygo_startTask, .-tinygo_startTask
+
+.global tinygo_swapTask
+.type tinygo_swapTask, %function
+tinygo_swapTask:
+    // This function gets the following parameters:
+    // x0 = newStack uintptr
+    // x1 = oldStack *uintptr
+
+    // Save all callee-saved registers:
+    stp     x19, x20, [sp, #-96]!
+    stp     x21, x22, [sp, #16]
+    stp     x23, x24, [sp, #32]
+    stp     x25, x26, [sp, #48]
+    stp     x27, x28, [sp, #64]
+    stp     x29, x30, [sp, #80]
+
+    // Save the current stack pointer in oldStack.
+    mov x8, sp
+    str x8, [x1]
+
+    // Switch to the new stack pointer.
+    mov sp, x0
+
+    // Restore stack state and return.
+    ldp     x29, x30, [sp, #80]
+    ldp     x27, x28, [sp, #64]
+    ldp     x25, x26, [sp, #48]
+    ldp     x23, x24, [sp, #32]
+    ldp     x21, x22, [sp, #16]
+    ldp     x19, x20, [sp], #96
+    ret