[lld] [lld][WebAssembly] Allow linker-synthetic symbols to be undefine when… (PR #153537)

[YAMAMOTO Takashi via llvm-commits]

https://github.com/yamt updated https://github.com/llvm/llvm-project/pull/153537

>From 51ced5c6a2440533ab8b8fdf039d26eca5ecf2ba Mon Sep 17 00:00:00 2001
From: Sam Clegg <sbc at chromium.org>
Date: Thu, 20 Feb 2025 17:11:20 -0800
Subject: [PATCH] [lld][WebAssembly] Allow linker-synthetic symbols to be
 undefine when building shared libraries

Fixes: #103592

Co-authored-by: YAMAMOTO Takashi <yamamoto at midokura.com>
---
 lld/test/wasm/shared-synthetic-symbols.s | 75 ++++++++++++++++++++++++
 lld/wasm/Driver.cpp                      | 26 +++++---
 2 files changed, 94 insertions(+), 7 deletions(-)
 create mode 100644 lld/test/wasm/shared-synthetic-symbols.s

diff --git a/lld/test/wasm/shared-synthetic-symbols.s b/lld/test/wasm/shared-synthetic-symbols.s
new file mode 100644
index 0000000000000..0c9c68aabc239
--- /dev/null
+++ b/lld/test/wasm/shared-synthetic-symbols.s
@@ -0,0 +1,75 @@
+## Check that synthetic data-layout symbols such as __heap_base and __heap_end
+## can be referenced from shared libraries and pie executables without
+## generating undefined symbols.
+
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t.o %s
+# RUN: wasm-ld --experimental-pic -pie --import-memory -o %t.wasm %t.o
+# RUN: obj2yaml %t.wasm | FileCheck %s
+# RUN: wasm-ld --experimental-pic -shared -o %t.so %t.o
+# RUN: obj2yaml %t.so | FileCheck %s
+
+.globl _start
+
+_start:
+  .functype _start () -> ()
+  i32.const __heap_base at GOT
+  drop
+  i32.const __heap_end at GOT
+  drop
+  i32.const __stack_low at GOT
+  drop
+  i32.const __stack_high at GOT
+  drop
+  i32.const __global_base at GOT
+  drop
+  i32.const __data_end at GOT
+  drop
+  end_function
+
+# CHECK:        - Type:            IMPORT
+# CHECK-NEXT:     Imports:
+# CHECK-NEXT:       - Module:          env
+# CHECK-NEXT:         Field:           memory
+# CHECK-NEXT:         Kind:            MEMORY
+# CHECK-NEXT:         Memory:
+# CHECK-NEXT:           Minimum:         0x0
+# CHECK-NEXT:       - Module:          env
+# CHECK-NEXT:         Field:           __memory_base
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   false
+# CHECK-NEXT:       - Module:          env
+# CHECK-NEXT:         Field:           __table_base
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   false
+# CHECK-NEXT:       - Module:          GOT.mem
+# CHECK-NEXT:         Field:           __heap_base
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   true
+# CHECK-NEXT:       - Module:          GOT.mem
+# CHECK-NEXT:         Field:           __heap_end
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   true
+# CHECK-NEXT:       - Module:          GOT.mem
+# CHECK-NEXT:         Field:           __stack_low
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   true
+# CHECK-NEXT:       - Module:          GOT.mem
+# CHECK-NEXT:         Field:           __stack_high
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   true
+# CHECK-NEXT:       - Module:          GOT.mem
+# CHECK-NEXT:         Field:           __global_base
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   true
+# CHECK-NEXT:       - Module:          GOT.mem
+# CHECK-NEXT:         Field:           __data_end
+# CHECK-NEXT:         Kind:            GLOBAL
+# CHECK-NEXT:         GlobalType:      I32
+# CHECK-NEXT:         GlobalMutable:   true
diff --git a/lld/wasm/Driver.cpp b/lld/wasm/Driver.cpp
index 1c5d21c06f5af..17d06ed5fae6d 100644
--- a/lld/wasm/Driver.cpp
+++ b/lld/wasm/Driver.cpp
@@ -985,15 +985,27 @@ static void createOptionalSymbols() {
 
   ctx.sym.dsoHandle = symtab->addOptionalDataSymbol("__dso_handle");
 
-  if (!ctx.arg.shared)
-    ctx.sym.dataEnd = symtab->addOptionalDataSymbol("__data_end");
+  auto addDataLayoutSymbol = [&](StringRef s) -> DefinedData * {
+    // Data layout symbols are either defined by lld, or (in the case
+    // of PIC code) defined by the dynamic linker / embedder.
+    if (ctx.isPic) {
+      ctx.arg.allowUndefinedSymbols.insert(s);
+      return nullptr;
+    } else {
+      return symtab->addOptionalDataSymbol(s);
+    }
+  };
+
+  ctx.sym.dataEnd = addDataLayoutSymbol("__data_end");
+  ctx.sym.stackLow = addDataLayoutSymbol("__stack_low");
+  ctx.sym.stackHigh = addDataLayoutSymbol("__stack_high");
+  ctx.sym.globalBase = addDataLayoutSymbol("__global_base");
+  ctx.sym.heapBase = addDataLayoutSymbol("__heap_base");
+  ctx.sym.heapEnd = addDataLayoutSymbol("__heap_end");
 
+  // for pic, __memory_base and __table_base are handled in
+  // createSyntheticSymbols.
   if (!ctx.isPic) {
-    ctx.sym.stackLow = symtab->addOptionalDataSymbol("__stack_low");
-    ctx.sym.stackHigh = symtab->addOptionalDataSymbol("__stack_high");
-    ctx.sym.globalBase = symtab->addOptionalDataSymbol("__global_base");
-    ctx.sym.heapBase = symtab->addOptionalDataSymbol("__heap_base");
-    ctx.sym.heapEnd = symtab->addOptionalDataSymbol("__heap_end");
     ctx.sym.definedMemoryBase = symtab->addOptionalDataSymbol("__memory_base");
     ctx.sym.definedTableBase = symtab->addOptionalDataSymbol("__table_base");
   }