Apply the sbrk/brk value at compile time (#2366)

We've had an option to set the location of the sbrk ptr, but not the value.
Applying the value as well is necessary for standalone wasm, as otherwise we set
it in JS.

Author: kripken

src/ir/memory-utils.h

@@ -22,17 +22,31 @@
 
 #include "literal.h"
 #include "wasm-binary.h"
+#include "wasm-builder.h"
 #include "wasm.h"
 
 namespace wasm {
 
 namespace MemoryUtils {
-// flattens memory into a single data segment. returns true if successful
-inline bool flatten(Memory& memory) {
+// Flattens memory into a single data segment, or no segment. If there is
+// a segment, it starts at 0.
+// If ensuredSegmentSize is provided, then a segment is always emitted,
+// and of at least that size.
+// Returns true if successful (e.g. relocatable segments cannot be flattened).
+inline bool flatten(Memory& memory,
+                    Index ensuredSegmentSize = 0,
+                    Module* module = nullptr) {
   if (memory.segments.size() == 0) {
+    if (ensuredSegmentSize > 0) {
+      assert(module); // must provide a module if ensuring a size.
+      Builder builder(*module);
+      memory.segments.emplace_back(builder.makeConst(Literal(int32_t(0))));
+      memory.segments[0].data.resize(ensuredSegmentSize);
+    }
     return true;
   }
   std::vector<char> data;
+  data.resize(ensuredSegmentSize);
   for (auto& segment : memory.segments) {
     if (segment.isPassive) {
       return false;

src/passes/PostEmscripten.cpp

@@ -22,6 +22,7 @@
 #include <asmjs/shared-constants.h>
 #include <ir/import-utils.h>
 #include <ir/localize.h>
+#include <ir/memory-utils.h>
 #include <pass.h>
 #include <shared-constants.h>
 #include <wasm-builder.h>
@@ -80,6 +81,27 @@ struct PostEmscripten : public Pass {
         func->body = builder.makeConst(Literal(int32_t(sbrkPtr)));
         func->module = func->base = Name();
       }
+      // Apply the sbrk ptr value, if it was provided. This lets emscripten set
+      // up sbrk entirely in wasm, without depending on the JS side to init
+      // anything; this is necessary for standalone wasm mode, in which we do
+      // not have any JS. Otherwise, the JS would set this value during
+      // startup.
+      auto sbrkValStr =
+        runner->options.getArgumentOrDefault("emscripten-sbrk-val", "");
+      if (sbrkValStr != "") {
+        uint32_t sbrkVal = std::stoi(sbrkValStr);
+        auto end = sbrkPtr + sizeof(sbrkVal);
+        // Flatten memory to make it simple to write to. Later passes can
+        // re-optimize it.
+        MemoryUtils::ensureExists(module->memory);
+        if (!MemoryUtils::flatten(module->memory, end, module)) {
+          Fatal() << "cannot apply sbrk-val since memory is not flattenable\n";
+        }
+        auto& segment = module->memory.segments[0];
+        assert(segment.offset->cast<Const>()->value.geti32() == 0);
+        assert(end <= segment.data.size());
+        memcpy(segment.data.data() + sbrkPtr, &sbrkVal, sizeof(sbrkVal));
+      }
     }
 
     // Optimize calls

test/passes/post-emscripten_pass-arg=emscripten-sbrk-ptr@16_pass-arg=emscripten-sbrk-val@42.txt

@@ -0,0 +1,60 @@
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 1 1)
+ (data (i32.const 0) "\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00*\00\00\00")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)
+(module
+ (memory $0 1 1)
+ (data (i32.const 0) "\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00*\00\00\00")
+)
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 10 10)
+ (data (i32.const 0) "\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00*\00\00\00")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 10 10)
+ (data (i32.const 0) "1234567890123456*\00\00\00")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 10 10)
+ (data (i32.const 0) "1234567890\00\00\00\00\00\00*\00\00\00")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 10 10)
+ (data (i32.const 0) "\00\00\00\00\00\00\00\00\00\00123456*\00\00\001234567890")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 10 10)
+ (data (i32.const 0) "\00\00\00\00\00\00\00\00\00\00123456*\00\00\00")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)
+(module
+ (type $FUNCSIG$i (func (result i32)))
+ (memory $0 10 10)
+ (data (i32.const 0) "\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00*\00\00\00\00\00\00\001234567890")
+ (func $internal (; 0 ;) (type $FUNCSIG$i) (result i32)
+  (i32.const 16)
+ )
+)

test/passes/post-emscripten_pass-arg=emscripten-sbrk-ptr@16_pass-arg=emscripten-sbrk-val@42.wast

@@ -0,0 +1,35 @@
+(module
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+(module
+)
+(module
+  (memory $0 10 10)
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+(module
+  (memory $0 10 10)
+  (data (i32.const 0) "12345678901234567890")
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+(module
+  (memory $0 10 10)
+  (data (i32.const 0) "1234567890")
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+(module
+  (memory $0 10 10)
+  (data (i32.const 10) "12345678901234567890")
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+(module
+  (memory $0 10 10)
+  (data (i32.const 10) "1234567890")
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+(module
+  (memory $0 10 10)
+  (data (i32.const 24) "1234567890")
+  (import "env" "emscripten_get_sbrk_ptr" (func $internal(result i32)))
+)
+