[wasm-ld] Define a `__heap_end` symbol marking the end of allocated memory.

[sunfish]

Summary

Define a __heap_end symbol that marks the end of the memory region
that starts at __heap_base. This will allow malloc implementations to
know how much memory they can use at __heap_base even if someone has
done a memory.grow before they can initialize their state.

[sbc100]

lgtm, with comment.

lld/wasm/Symbols.h

564

Can you maybe group these together under single comment? I started to do that above for stack_pointer/stack_low/__stack_high and I think it makes it easier to read this section of code. I noticed ELF was doing this in its Symbols.h

[sunfish]

• Merge the comments for heap_base and heap_end.

lld/test/wasm/export-all.s

 # CHECK-NEXT:         Kind:            GLOBAL
 # CHECK-NEXT:         Index:           4
 # CHECK-NEXT:       - Name:            __global_base
 # CHECK-NEXT:         Kind:            GLOBAL
 # CHECK-NEXT:         Index:           5
 # CHECK-NEXT:       - Name:            __heap_base
 # CHECK-NEXT:         Kind:            GLOBAL
 # CHECK-NEXT:         Index:           6
-# CHECK-NEXT:       - Name:            __memory_base
+# CHECK-NEXT:       - Name:            __heap_end
 # CHECK-NEXT:         Kind:            GLOBAL
 # CHECK-NEXT:         Index:           7
-# CHECK-NEXT:       - Name:            __table_base
+# CHECK-NEXT:       - Name:            __memory_base
 # CHECK-NEXT:         Kind:            GLOBAL
 # CHECK-NEXT:         Index:           8
+# CHECK-NEXT:       - Name:            __table_base
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         Index:           9

lld/test/wasm/mutable-global-exports.s

 # CHECK-ALL-NEXT:        Kind:            GLOBAL
 # CHECK-ALL-NEXT:        Index:           5
 # CHECK-ALL-NEXT:      - Name:            __global_base
 # CHECK-ALL-NEXT:        Kind:            GLOBAL
 # CHECK-ALL-NEXT:        Index:           6
 # CHECK-ALL-NEXT:      - Name:            __heap_base
 # CHECK-ALL-NEXT:        Kind:            GLOBAL
 # CHECK-ALL-NEXT:        Index:           7
-# CHECK-ALL-NEXT:      - Name:            __memory_base
+# CHECK-ALL-NEXT:      - Name:            __heap_end
 # CHECK-ALL-NEXT:        Kind:            GLOBAL
 # CHECK-ALL-NEXT:        Index:           8
-# CHECK-ALL-NEXT:      - Name:            __table_base
+# CHECK-ALL-NEXT:      - Name:            __memory_base
 # CHECK-ALL-NEXT:        Kind:            GLOBAL
 # CHECK-ALL-NEXT:        Index:           9
+# CHECK-ALL-NEXT:      - Name:            __table_base
+# CHECK-ALL-NEXT:        Kind:            GLOBAL
+# CHECK-ALL-NEXT:        Index:           10
 # CHECK-ALL-NEXT:  - Type:            CODE
 
 # CHECK-ALL:         Name:            target_features
 # CHECK-ALL-NEXT:    Features:
 # CHECK-ALL-NEXT:      - Prefix:          USED
 # CHECK-ALL-NEXT:        Name:            atomics
 # CHECK-ALL-NEXT:      - Prefix:          USED
 # CHECK-ALL-NEXT:        Name:            mutable-globals
 # CHECK-ALL-NEXT: ...
 

lld/wasm/Driver.cpp

static void createOptionalSymbols() {
   if (!config->shared)
     WasmSym::dataEnd = symtab->addOptionalDataSymbol("__data_end");
 
   if (!config->isPic) {
     WasmSym::stackLow = symtab->addOptionalDataSymbol("__stack_low");
     WasmSym::stackHigh = symtab->addOptionalDataSymbol("__stack_high");
     WasmSym::globalBase = symtab->addOptionalDataSymbol("__global_base");
     WasmSym::heapBase = symtab->addOptionalDataSymbol("__heap_base");
+    WasmSym::heapEnd = symtab->addOptionalDataSymbol("__heap_end");
     WasmSym::definedMemoryBase = symtab->addOptionalDataSymbol("__memory_base");
     WasmSym::definedTableBase = symtab->addOptionalDataSymbol("__table_base");
     if (config->is64.value_or(false))
       WasmSym::definedTableBase32 =
           symtab->addOptionalDataSymbol("__table_base32");
   }
 
   // For non-shared memory programs we still need to define __tls_base since we

lld/wasm/Symbols.h

struct WasmSym {
   // __ctors_done
   // Symbol whose value is non-zero if the global ctors have been called.
   static GlobalSymbol *ctorsDone;
 
   // __data_end
   // Symbol marking the end of the data and bss.
   static DefinedData *dataEnd;
 
-  // __heap_base
-  // Symbol marking the end of the data, bss and explicit stack.  Any linear
-  // memory following this address is not used by the linked code and can
-  // therefore be used as a backing store for brk()/malloc() implementations.
+  // __heap_base/__heap_end
+  // Symbols marking the beginning and end of the "heap". It starts at the end
+  // of the data, bss and explicit stack, and extends to the end of the linear
+  // memory allocated by wasm-ld. This region of memory is not used by the
+  // linked code, so it may be used as a backing store for `sbrk` or `malloc`
+  // implementations.
   static DefinedData *heapBase;
+  static DefinedData *heapEnd;
 
   // __wasm_init_memory_flag
   // Symbol whose contents are nonzero iff memory has already been initialized.
   static DefinedData *initMemoryFlag;
 
   // __wasm_init_memory
   // Function that initializes passive data segments during instantiation.
   static DefinedFunction *initMemory;
 
   // __wasm_call_ctors
   // Function that directly calls all ctors in priority order.
   static DefinedFunction *callCtors;
 
   // __wasm_call_dtors

lld/wasm/Symbols.cpp

 DefinedFunction *WasmSym::applyGlobalRelocs;
 DefinedFunction *WasmSym::applyGlobalTLSRelocs;
 DefinedFunction *WasmSym::initTLS;
 DefinedFunction *WasmSym::startFunction;
 DefinedData *WasmSym::dsoHandle;
 DefinedData *WasmSym::dataEnd;
 DefinedData *WasmSym::globalBase;
 DefinedData *WasmSym::heapBase;
+DefinedData *WasmSym::heapEnd;
 DefinedData *WasmSym::initMemoryFlag;
 GlobalSymbol *WasmSym::stackPointer;
 DefinedData *WasmSym::stackLow;
 DefinedData *WasmSym::stackHigh;
 GlobalSymbol *WasmSym::tlsBase;
 GlobalSymbol *WasmSym::tlsSize;
 GlobalSymbol *WasmSym::tlsAlign;
 GlobalSymbol *WasmSym::ctorsDone;

lld/wasm/Writer.cpp

if (config->initialMemory != alignTo(config->initialMemory, WasmPageSize))
       error("initial memory must be " + Twine(WasmPageSize) + "-byte aligned");
     if (memoryPtr > config->initialMemory)
       error("initial memory too small, " + Twine(memoryPtr) + " bytes needed");
     if (config->initialMemory > maxMemorySetting)
       error("initial memory too large, cannot be greater than " +
             Twine(maxMemorySetting));
     memoryPtr = config->initialMemory;
   }
-  out.memorySec->numMemoryPages =
-      alignTo(memoryPtr, WasmPageSize) / WasmPageSize;
+
+  memoryPtr = alignTo(memoryPtr, WasmPageSize);
+
+  out.memorySec->numMemoryPages = memoryPtr / WasmPageSize;
   log("mem: total pages = " + Twine(out.memorySec->numMemoryPages));
 
+  if (WasmSym::heapEnd) {
+    // Set `__heap_end` to follow the end of the statically allocated linear
+    // memory. The fact that this comes last means that a malloc/brk
+    // implementation can grow the heap at runtime.
+    log("mem: heap end    = " + Twine(memoryPtr));
+    WasmSym::heapEnd->setVA(memoryPtr);
+  }
+
   if (config->maxMemory != 0) {
     if (config->maxMemory != alignTo(config->maxMemory, WasmPageSize))
       error("maximum memory must be " + Twine(WasmPageSize) + "-byte aligned");
     if (memoryPtr > config->maxMemory)
       error("maximum memory too small, " + Twine(memoryPtr) + " bytes needed");
     if (config->maxMemory > maxMemorySetting)
       error("maximum memory too large, cannot be greater than " +
             Twine(maxMemorySetting));
   }
 
   // Check max if explicitly supplied or required by shared memory
   if (config->maxMemory != 0 || config->sharedMemory) {
     uint64_t max = config->maxMemory;
     if (max == 0) {
       // If no maxMemory config was supplied but we are building with
       // shared memory, we need to pick a sensible upper limit.
       if (config->isPic)
         max = maxMemorySetting;
       else
-        max = alignTo(memoryPtr, WasmPageSize);
+        max = memoryPtr;
     }
     out.memorySec->maxMemoryPages = max / WasmPageSize;
     log("mem: max pages   = " + Twine(out.memorySec->maxMemoryPages));
   }
 }
 
 void Writer::addSection(OutputSection *sec) {
   if (!sec->isNeeded())