[WebAssembly] WASIP3 and component model threading support

clang/include/clang/Options/Options.td

@@ -5763,6 +5763,8 @@ def mfp16 : Flag<["-"], "mfp16">, Group<m_wasm_Features_Group>;
 def mno_fp16 : Flag<["-"], "mno-fp16">, Group<m_wasm_Features_Group>;
 def mgc : Flag<["-"], "mgc">, Group<m_wasm_Features_Group>;
 def mno_gc : Flag<["-"], "mno-gc">, Group<m_wasm_Features_Group>;
+def mlibcall_thread_context : Joined<["-"], "mlibcall-thread-context">, Group<m_wasm_Features_Group>;
+def mno_libcall_thread_context : Joined<["-"], "mno-libcall-thread-context">, Group<m_wasm_Features_Group>;
 def mmultimemory : Flag<["-"], "mmultimemory">, Group<m_wasm_Features_Group>;
 def mno_multimemory : Flag<["-"], "mno-multimemory">, Group<m_wasm_Features_Group>;
 def mmultivalue : Flag<["-"], "mmultivalue">, Group<m_wasm_Features_Group>;

clang/lib/Basic/Targets/WebAssembly.cpp

@@ -123,6 +123,8 @@ void WebAssemblyTargetInfo::getTargetDefines(const LangOptions &Opts,
     Builder.defineMacro("__wasm_tail_call__");
   if (HasWideArithmetic)
     Builder.defineMacro("__wasm_wide_arithmetic__");
+  if (HasLibcallThreadContext)
+    Builder.defineMacro("__wasm_libcall_thread_context__");
   // Note that not all wasm features appear here.   For example,
   // HasCompatctImports
 
@@ -386,6 +388,14 @@ bool WebAssemblyTargetInfo::handleTargetFeatures(
       HasWideArithmetic = false;
       continue;
     }
+    if (Feature == "+libcall-thread-context") {
+      HasLibcallThreadContext = true;
+      continue;
+    }
+    if (Feature == "-libcall-thread-context") {
+      HasLibcallThreadContext = false;
+      continue;
+    }
 
     Diags.Report(diag::err_opt_not_valid_with_opt)
         << Feature << "-target-feature";

clang/lib/Basic/Targets/WebAssembly.h

@@ -68,6 +68,7 @@ class LLVM_LIBRARY_VISIBILITY WebAssemblyTargetInfo : public TargetInfo {
   bool HasExtendedConst = false;
   bool HasFP16 = false;
   bool HasGC = false;
+  bool HasLibcallThreadContext = false;
   bool HasMultiMemory = false;
   bool HasMultivalue = false;
   bool HasMutableGlobals = false;

clang/lib/Driver/ToolChains/WebAssembly.cpp

@@ -88,6 +88,16 @@ static bool WantsPthread(const llvm::Triple &Triple, const ArgList &Args) {
   return WantsPthread;
 }
 
+static bool WantsLibcallThreadContext(const llvm::Triple &Triple,
+                                      const ArgList &Args) {
+  // If the target is WASIP3, then enable the
+  // libcall-thread-context feature by default, unless explicitly
+  // disabled.
+  return Triple.getOS() == llvm::Triple::WASIp3 &&
+         Args.hasFlag(options::OPT_mlibcall_thread_context,
+                      options::OPT_mno_libcall_thread_context, true);
+}
+
 void wasm::Linker::ConstructJob(Compilation &C, const JobAction &JA,
                                 const InputInfo &Output,
                                 const InputInfoList &Inputs,
@@ -169,6 +179,9 @@ void wasm::Linker::ConstructJob(Compilation &C, const JobAction &JA,
 
   AddLinkerInputs(ToolChain, Inputs, Args, CmdArgs, JA);
 
+  if (WantsLibcallThreadContext(ToolChain.getTriple(), Args))
+    CmdArgs.push_back("--libcall-thread-context");
+
   if (WantsPthread(ToolChain.getTriple(), Args))
     CmdArgs.push_back("--shared-memory");
 
@@ -321,6 +334,11 @@ void WebAssembly::addClangTargetOptions(const ArgList &DriverArgs,
                           options::OPT_fno_use_init_array, true))
     CC1Args.push_back("-fno-use-init-array");
 
+  if (WantsLibcallThreadContext(getTriple(), DriverArgs)) {
+    CC1Args.push_back("-target-feature");
+    CC1Args.push_back("+libcall-thread-context");
+  }
+
   // '-pthread' implies atomics, bulk-memory, mutable-globals, and sign-ext
   if (WantsPthread(getTriple(), DriverArgs)) {
     if (DriverArgs.hasFlag(options::OPT_mno_atomics, options::OPT_matomics,

clang/test/Driver/wasm-features.c

@@ -118,3 +118,9 @@
 
 // COMPACT-IMPORTS: "-target-feature" "+compact-imports"
 // NO-COMPACT-IMPORTS: "-target-feature" "-compact-imports"
+
+// RUN: %clang --target=wasm32-unknown-unknown -### %s -mlibcall-thread-context 2>&1 | FileCheck %s -check-prefix=LIBCALL-THREAD-CONTEXT
+// RUN: %clang --target=wasm32-unknown-unknown -### %s -mno-libcall-thread-context 2>&1 | FileCheck %s -check-prefix=NO-LIBCALL-THREAD-CONTEXT
+
+// LIBCALL-THREAD-CONTEXT: "-target-feature" "+libcall-thread-context"
+// NO-LIBCALL-THREAD-CONTEXT: "-target-feature" "-libcall-thread-context"

clang/test/Driver/wasm-toolchain.c

@@ -303,3 +303,15 @@
 // RUN:   | FileCheck -check-prefix=LINK_WALI_BASIC %s
 // LINK_WALI_BASIC: "-cc1" {{.*}} "-o" "[[temp:[^"]*]]"
 // LINK_WALI_BASIC: wasm-ld{{.*}}" "-L/foo/lib/wasm32-linux-muslwali" "crt1.o" "[[temp]]" "-lc" "{{.*[/\\]}}libclang_rt.builtins.a" "-o" "a.out"
+
+// `wasm32-wasip3` passes `+libcall-thread-context` by default.
+
+// RUN: %clang -### --target=wasm32-wasip3 --sysroot=/foo %s 2>&1 \
+// RUN:   | FileCheck -check-prefix=LINK_WASIP3_LIBCALL_THREAD_CONTEXT %s
+// LINK_WASIP3_LIBCALL_THREAD_CONTEXT: "-cc1" {{.*}} "-target-feature" "+libcall-thread-context"
+
+// `wasm32-wasip3` does not pass `+libcall-thread-context` when `-mno-libcall-thread-context` is used.
+
+// RUN: %clang -### --target=wasm32-wasip3 --sysroot=/foo -mno-libcall-thread-context %s 2>&1 \
+// RUN:   | FileCheck -check-prefix=LINK_WASIP3_NO_LIBCALL_THREAD_CONTEXT %s
+// LINK_WASIP3_NO_LIBCALL_THREAD_CONTEXT: "-cc1" {{.*}} "-target-feature" "-libcall-thread-context"

clang/test/Preprocessor/wasm-target-features.c

@@ -251,3 +251,12 @@
 // RUN:   | FileCheck %s -check-prefix=BLEEDING-EDGE-NO-SIMD128
 //
 // BLEEDING-EDGE-NO-SIMD128-NOT: #define __wasm_simd128__ 1{{$}}
+
+// RUN: %clang -E -dM %s -o - 2>&1 \
+// RUN:     -target wasm32-unknown-unknown -mlibcall-thread-context \
+// RUN:   | FileCheck %s -check-prefix=LIBCALL-THREAD-CONTEXT
+// RUN: %clang -E -dM %s -o - 2>&1 \
+// RUN:     -target wasm64-unknown-unknown -mlibcall-thread-context \
+// RUN:   | FileCheck %s -check-prefix=LIBCALL-THREAD-CONTEXT
+
+// LIBCALL-THREAD-CONTEXT: #define __wasm_libcall_thread_context__ 1{{$}}

lld/test/wasm/stack-pointer-abi.s

@@ -0,0 +1,32 @@
+# RUN: split-file %s %t
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t/use.o %t/use.s
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t/disallow.o %t/disallow.s
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t/start.o %t/start.s
+# RUN: wasm-ld --libcall-thread-context -o %t/libcall.wasm %t/use.o %t/start.o
+# RUN: obj2yaml %t/libcall.wasm | FileCheck %s --check-prefix=LIBCALL
+# RUN: wasm-ld -o %t/global.wasm %t/disallow.o %t/start.o
+# RUN: obj2yaml %t/global.wasm | FileCheck %s --check-prefix=GLOBAL
+
+#--- start.s
+  .globl  _start
+_start:
+  .functype _start () -> ()
+  end_function
+
+#--- disallow.s
+.section  .custom_section.target_features,"",@
+  .int8 1
+  .int8 45
+  .int8 22
+  .ascii  "libcall-thread-context"
+
+#--- use.s
+
+.section  .custom_section.target_features,"",@
+  .int8 1
+  .int8 43
+  .int8 22
+  .ascii  "libcall-thread-context"
+
+# LIBCALL: Name: __init_stack_pointer
+# GLOBAL: Name: __stack_pointer

lld/test/wasm/thread-context-abi-mismatch.s

@@ -0,0 +1,48 @@
+# Test that linking object files with mismatched thread context ABIs fails with an error.
+
+# RUN: split-file %s %t
+
+# Test that the presence of an import of __stack_pointer from the env module is treated 
+# as an indication that the global thread context ABI is being used, even if the
+# libcall-thread-context feature is not disallowed.
+
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t/start.o %t/start.s
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t/stack-pointer.o %t/stack-pointer.s
+# RUN: not wasm-ld --libcall-thread-context %t/start.o %t/stack-pointer.o -o %t/fail.wasm 2>&1 | FileCheck %s
+
+# Test that explicitly disallowing the libcall-thread-context feature causes linking to fail 
+# with an error when other files use the feature.
+
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t/disallow.o %t/disallow.s
+# RUN: not wasm-ld --libcall-thread-context %t/start.o %t/disallow.o -o %t/fail.wasm 2>&1 | FileCheck %s
+
+# CHECK: error: --libcall-thread-context is disallowed by {{.*}} because it uses globals for thread context rather than library function calls.
+
+#--- start.s
+.globl _start
+_start:
+  .functype _start () -> ()
+  end_function
+
+.section  .custom_section.target_features,"",@
+  .int8 1
+  .int8 43
+  .int8 22
+  .ascii  "libcall-thread-context"
+
+#--- stack-pointer.s
+.globaltype __stack_pointer, i32
+
+.globl use_stack_pointer
+use_stack_pointer:
+  .functype use_stack_pointer () -> ()
+  global.get __stack_pointer
+  drop
+  end_function
+
+#--- disallow.s
+.section  .custom_section.target_features,"",@
+  .int8 1
+  .int8 45
+  .int8 22
+  .ascii  "libcall-thread-context"

lld/test/wasm/tls-libcall.s

@@ -0,0 +1,74 @@
+# RUN: llvm-mc -filetype=obj -triple=wasm32-unknown-unknown -o %t.o %s
+# RUN: wasm-ld --libcall-thread-context --shared-memory -no-gc-sections -o %t.wasm %t.o
+# RUN: obj2yaml %t.wasm | FileCheck %s
+# RUN: llvm-objdump -d --no-print-imm-hex --no-show-raw-insn %t.wasm | FileCheck %s --check-prefix=DIS
+
+.globl         __wasm_get_tls_base
+__wasm_get_tls_base:
+    .functype   __wasm_get_tls_base () -> (i32)
+    i32.const 0
+    end_function
+
+.globl _start
+_start:
+  .functype _start () -> (i32)
+  call __wasm_get_tls_base
+  i32.const tls1@TLSREL
+  i32.add
+  i32.load 0
+  call __wasm_get_tls_base
+  i32.const tls2@TLSREL
+  i32.add
+  i32.load 0
+  i32.add
+  end_function
+
+.section  .tdata.tls1,"",@
+.globl  tls1
+tls1:
+  .int32  1
+  .size tls1, 4
+
+.section  .tdata.tls2,"",@
+.globl  tls2
+tls2:
+  .int32  2
+  .size tls2, 4
+
+.section  .custom_section.target_features,"",@
+  .int8 3
+  .int8 43
+  .int8 22
+  .ascii  "libcall-thread-context"
+  .int8 43
+  .int8 11
+  .ascii  "bulk-memory"
+  .int8 43
+  .int8 7
+  .ascii  "atomics"
+
+
+# CHECK:      GlobalNames:
+# CHECK-NEXT:      - Index:           0
+# CHECK-NEXT:        Name:            __init_stack_pointer
+# CHECK-NEXT:      - Index:           1
+# CHECK-NEXT:        Name:            __init_tls_base
+# CHECK-NEXT:      - Index:           2
+# CHECK-NEXT:        Name:            __tls_size
+# CHECK-NEXT:      - Index:           3
+# CHECK-NEXT:        Name:            __tls_align
+
+# DIS-LABEL: <__wasm_init_memory>:
+
+# DIS-LABEL: <_start>:
+# DIS-EMPTY:
+# DIS-NEXT:       call    4
+# DIS-NEXT:       i32.const       0
+# DIS-NEXT:       i32.add 
+# DIS-NEXT:       i32.load        0
+# DIS-NEXT:       call    4
+# DIS-NEXT:       i32.const       4
+# DIS-NEXT:       i32.add 
+# DIS-NEXT:       i32.load        0
+# DIS-NEXT:       i32.add 
+# DIS-NEXT:       end

lld/wasm/Config.h

@@ -35,6 +35,7 @@ class Symbol;
 class DefinedData;
 class GlobalSymbol;
 class DefinedFunction;
+class UndefinedFunction;
 class DefinedGlobal;
 class UndefinedGlobal;
 class TableSymbol;
@@ -65,6 +66,7 @@ struct Config {
   bool growableTable;
   bool gcSections;
   llvm::StringSet<> keepSections;
+  bool libcallThreadContext;
   std::optional<std::pair<llvm::StringRef, llvm::StringRef>> memoryImport;
   std::optional<llvm::StringRef> memoryExport;
   bool sharedMemory;
@@ -252,6 +254,14 @@ struct Ctx {
     // Used as an address space for function pointers, with each function that
     // is used as a function pointer being allocated a slot.
     TableSymbol *indirectFunctionTable;
+
+    // __wasm_set_tls_base
+    // Function used to set TLS base in libcall thread context modules.
+    UndefinedFunction *setTLSBase;
+
+    // __wasm_get_tls_base
+    // Function used to get TLS base in libcall thread context modules.
+    UndefinedFunction *getTLSBase;
   };
   WasmSym sym;