Add wasm32-unknown-wasi target to mlton

This requires:
- WASI SDK: https://github.com/WebAssembly/wasi-sdk
- wasmtime: https://wasmtime.dev/

bin/mlton-script

@@ -118,6 +118,9 @@ for arg in "$@"; do
                -target-cc-opt ia64-hpux "-mlp64"                        \
                -target-cc-opt ia64 "-mtune=itanium2"                    \
                -target-cc-opt sparc '-m32 -mcpu=v8 -Wa,-xarch=v8plusa'  \
+               -target-cc-opt wasi '-D_WASI_EMULATED_SIGNAL'            \
+               -target-cc-opt wasi '-D_WASI_EMULATED_PROCESS_CLOCKS'    \
+               -target-cc-opt wasi '-D_WASI_EMULATED_GETPID'            \
                -target-cc-opt x86 '-m32'                                \
                -target-link-opt aix '-maix64'                           \
                -target-link-opt alpha                                   \
@@ -134,6 +137,9 @@ for arg in "$@"; do
                -target-link-opt mingw '-Wl,--enable-stdcall-fixup'      \
                -target-link-opt openbsd '-Wl,--no-execute-only'         \
                -target-link-opt solaris '-lnsl -lsocket -lrt'           \
+               -target-link-opt wasi '-lwasi-emulated-signal'           \
+               -target-link-opt wasi '-lwasi-emulated-process-clocks'   \
+               -target-link-opt wasi '-lwasi-emulated-getpid'           \
                -target-link-opt x86 '-m32'                              \
                -profile-exclude '\$\(SML_LIB\)'
     fi

runtime/Makefile

@@ -20,6 +20,8 @@ endif
 
 EXE :=
 
+WASMTIME :=
+
 MDVARIANTS := OPT DBG
 PIVARIANTS := DPI NPI PIC PIE
 
@@ -110,6 +112,16 @@ ifeq ($(TARGET_OS), cygwin)
 EXE := .exe
 endif
 
+ifeq ($(TARGET_OS), wasi)
+WASMTIME := wasmtime
+XCPPFLAGS += -D_WASI_EMULATED_SIGNAL \
+             -D_WASI_EMULATED_PROCESS_CLOCKS \
+             -D_WASI_EMULATED_GETPID
+XLDFLAGS += -lwasi-emulated-signal \
+            -lwasi-emulated-process-clocks \
+            -lwasi-emulated-getpid
+endif
+
 ifeq ($(TARGET_OS), mingw)
 EXE := .exe
 # GCC doesn't recognize the %I64 format specifier which means %ll on windows
@@ -239,11 +251,11 @@ ml-types.h: gen/ml-types.h
 	$(CP) $< $@
 
 gen/ml-types.h: gen/gen-types$(EXE)
-	./gen/gen-types$(EXE) ml-types.h > gen/ml-types.h
+	$(WASMTIME) ./gen/gen-types$(EXE) ml-types.h > gen/ml-types.h
 gen/c-types.h: gen/gen-types$(EXE)
-	./gen/gen-types$(EXE) c-types.h > gen/c-types.h
+	$(WASMTIME) ./gen/gen-types$(EXE) c-types.h > gen/c-types.h
 gen/c-types.sml: gen/gen-types$(EXE)
-	./gen/gen-types$(EXE) c-types.sml > gen/c-types.sml
+	$(WASMTIME) ./gen/gen-types$(EXE) c-types.sml > gen/c-types.sml
 
 ifneq ($(MAKECMDGOALS),clean)
 -include gen/gen-types.d
@@ -262,7 +274,7 @@ gen/$(1): gen/gen-basis-ffi.sml gen/basis-ffi.def
 	touch gen/$(1)
 else
 gen/$(1): gen/gen-basis-ffi$(EXE) gen/basis-ffi.def
-	./gen/gen-basis-ffi$(EXE) $(1) < gen/basis-ffi.def > gen/$(1)
+	$(WASMTIME) ./gen/gen-basis-ffi$(EXE) $(1) < gen/basis-ffi.def > gen/$(1)
 	$(CAT) gen/gen-basis-ffi.sml gen/basis-ffi.def gen/$(1) | $(SHA1DGST) > gen/$(1).chk
 endif
 endef
@@ -307,7 +319,7 @@ gdtoa/README: gdtoa.tgz gdtoa.may_alias-unions.patch gdtoa.rename-public-fns.pat
 	@touch $@
 
 gdtoa/arith.h: gdtoa/arithchk$(EXE)
-	./gdtoa/arithchk$(EXE) > gdtoa/arith.h
+	$(WASMTIME) ./gdtoa/arithchk$(EXE) > gdtoa/arith.h
 
 gdtoa/arithchk.c: gdtoa/README
 	@touch $@
@@ -319,7 +331,7 @@ ifneq ($(MAKECMDGOALS),clean)
 endif
 
 gdtoa/gd_qnan.h: gdtoa/qnan$(EXE)
-	./gdtoa/qnan$(EXE) > gdtoa/gd_qnan.h
+	$(WASMTIME) ./gdtoa/qnan$(EXE) > gdtoa/gd_qnan.h
 
 gdtoa/qnan.c: gdtoa/README
 	@touch $@
@@ -348,7 +360,7 @@ $(eval $(call A_TEMPLATE,mlton,MLTON))
 ### gen/constants ###
 
 gen/constants: gen/gen-constants$(EXE)
-	./gen/gen-constants > gen/constants
+	$(WASMTIME) ./gen/gen-constants > gen/constants
 
 ifneq ($(MAKECMDGOALS),clean)
 -include gen/gen-constants.d

runtime/cenv.h

@@ -82,6 +82,8 @@ COMPILE_TIME_ASSERT(sizeof_double__is_eight, sizeof(double) == 8);
 #include "platform/openbsd.h"
 #elif (defined (__sun__))
 #include "platform/solaris.h"
+#elif (defined (__wasi__))
+#include "platform/wasi.h"
 #else
 #error unknown platform os
 #endif
@@ -117,6 +119,8 @@ COMPILE_TIME_ASSERT(sizeof_double__is_eight, sizeof(double) == 8);
 #include "platform/x86.h"
 #elif (defined (__loongarch64))
 #include "platform/loongarch64.h"
+#elif (defined (__wasm32))
+#include "platform/wasm32.h"
 #else
 #error unknown platform arch
 #endif

runtime/gc/heap.c

@@ -183,7 +183,9 @@ void shrinkHeap (GC_state s, GC_heap h, size_t keepSize) {
     }
     assert (isAligned (keepWithMapsSize, s->sysvals.pageSize));
     assert (keepWithMapsSize <= h->withMapsSize);
+#if HAS_SHRINK_HEAP
     GC_release (h->start + keepWithMapsSize, h->withMapsSize - keepWithMapsSize);
+#endif
     h->size = keepSize;
     h->withMapsSize = keepWithMapsSize;
   }

runtime/platform.h

@@ -35,6 +35,10 @@
 #error HAS_REMAP not defined
 #endif
 
+#ifndef HAS_SHRINK_HEAP
+#error HAS_SHRINK_HEAP not defined
+#endif
+
 #ifndef HAS_SIGALTSTACK
 #error HAS_SIGALTSTACK not defined
 #else

runtime/platform/aix.h

@@ -34,6 +34,7 @@
 #define HAS_MSG_DONTWAIT FALSE
 #define HAS_PTRACE FALSE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/cygwin.h

@@ -35,6 +35,7 @@
 
 #define HAS_FEROUND FALSE
 #define HAS_REMAP TRUE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK FALSE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN TRUE

runtime/platform/darwin.h

@@ -34,6 +34,7 @@
 #define HAS_FEROUND TRUE
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/freebsd.h

@@ -30,6 +30,7 @@
 #define HAS_FEROUND TRUE
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/hpux.h

@@ -52,6 +52,7 @@
 #define HAS_FEROUND TRUE
 #define HAS_MSG_DONTWAIT FALSE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/hurd.h

@@ -30,6 +30,7 @@
 #define HAS_FEROUND TRUE
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP TRUE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/linux.h

@@ -38,6 +38,7 @@
 #endif
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP TRUE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #if (defined (__hppa__))
 #define NEEDS_SIGALTSTACK_EXEC TRUE

runtime/platform/mingw.h

@@ -33,6 +33,7 @@
 #define HAS_FEROUND FALSE
 #define HAS_MSG_DONTWAIT FALSE
 #define HAS_REMAP TRUE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK FALSE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN TRUE

runtime/platform/netbsd.h

@@ -30,6 +30,7 @@
 #define HAS_FEROUND FALSE
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/openbsd.h

@@ -30,6 +30,7 @@
 #define HAS_FEROUND FALSE
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/solaris.h

@@ -34,6 +34,7 @@
 #define HAS_FEROUND TRUE
 #define HAS_MSG_DONTWAIT TRUE
 #define HAS_REMAP FALSE
+#define HAS_SHRINK_HEAP TRUE
 #define HAS_SIGALTSTACK TRUE
 #define NEEDS_SIGALTSTACK_EXEC FALSE
 #define HAS_SPAWN FALSE

runtime/platform/wasi.c

@@ -0,0 +1,71 @@
+#include "platform.h"
+
+/* WASI only implements a subset of POSIX, and given how it works it doesn't
+ * make too much sense to try too hard to emulate missing functionality.
+ *
+ * Every function needed by the runtime library and not provided by WASI
+ * is declared in wasi.h, but only a tiny number are implemented here.
+ *
+ * This means that any use of missing runtime functions from SML code will
+ * result in linker errors at the end of compilation, rather than as a runtime
+ * error later.
+ *
+ * The functions in this file are the bare minimum to get the GC working,
+ * needed by every program compiled by MLton.
+ */
+
+size_t GC_pageSize (void) {
+        return PAGESIZE;
+}
+
+uintmax_t GC_physMem (void) {
+        /* WASI doesn't provide a way to query actual physical memory, so pick
+         * a reasonable amount (as of Feb 2024).
+         */
+        return 1 << 30;  /* 1 GiB */
+}
+
+void *GC_mmapAnon (__attribute__ ((unused)) void *start, size_t length) {
+        void *mem;
+        int err = posix_memalign (&mem, PAGESIZE, length);
+        if (err) {
+                return (void *) -1;
+        }
+        return memset (mem, 0, length);
+}
+
+void *GC_mmapAnonFlags (void *start, size_t length,
+                        __attribute__ ((unused)) int flags) {
+        return GC_mmapAnon (start, length);
+}
+
+void GC_release (void *base, __attribute__ ((unused)) size_t length) {
+        free (base);
+}
+
+void GC_displayMem (void) {
+        size_t memory_size = (size_t) sbrk(0);
+        size_t pages = memory_size / PAGESIZE;
+        printf ("memory.size: pages=%zu, pagesize=%d, total=%zu\n",
+                pages, PAGESIZE, memory_size);
+}
+
+void GC_diskBack_close (__attribute__ ((unused)) void *data) {
+        die ("Disk-backed heap not supported on WASI");
+}
+
+void GC_diskBack_read (__attribute__ ((unused)) void *data,
+                       __attribute__ ((unused)) pointer buf,
+                       __attribute__ ((unused)) size_t size) {
+        die ("Disk-backed heap not supported on WASI");
+}
+
+void *GC_diskBack_write (__attribute__ ((unused)) pointer buf,
+                         __attribute__ ((unused)) size_t size) {
+        die ("Disk-backed heap not supported on WASI");
+}
+
+int sigemptyset (sigset_t *set) {
+        *set = (sigset_t) 0;
+        return 0;
+}