Backport #145690, #146068 and #144058

llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp

@@ -2139,16 +2139,20 @@ void AsmPrinter::emitFunctionBody() {
 }
 
 /// Compute the number of Global Variables that uses a Constant.
-static unsigned getNumGlobalVariableUses(const Constant *C) {
-  if (!C)
+static unsigned getNumGlobalVariableUses(const Constant *C,
+                                         bool &HasNonGlobalUsers) {
+  if (!C) {
+    HasNonGlobalUsers = true;
     return 0;
+  }
 
   if (isa<GlobalVariable>(C))
     return 1;
 
   unsigned NumUses = 0;
   for (const auto *CU : C->users())
-    NumUses += getNumGlobalVariableUses(dyn_cast<Constant>(CU));
+    NumUses +=
+        getNumGlobalVariableUses(dyn_cast<Constant>(CU), HasNonGlobalUsers);
 
   return NumUses;
 }
@@ -2159,7 +2163,8 @@ static unsigned getNumGlobalVariableUses(const Constant *C) {
 /// candidates are skipped and are emitted later in case at least one cstexpr
 /// isn't replaced by a PC relative GOT entry access.
 static bool isGOTEquivalentCandidate(const GlobalVariable *GV,
-                                     unsigned &NumGOTEquivUsers) {
+                                     unsigned &NumGOTEquivUsers,
+                                     bool &HasNonGlobalUsers) {
   // Global GOT equivalents are unnamed private globals with a constant
   // pointer initializer to another global symbol. They must point to a
   // GlobalVariable or Function, i.e., as GlobalValue.
@@ -2171,7 +2176,8 @@ static bool isGOTEquivalentCandidate(const GlobalVariable *GV,
   // To be a got equivalent, at least one of its users need to be a constant
   // expression used by another global variable.
   for (const auto *U : GV->users())
-    NumGOTEquivUsers += getNumGlobalVariableUses(dyn_cast<Constant>(U));
+    NumGOTEquivUsers +=
+        getNumGlobalVariableUses(dyn_cast<Constant>(U), HasNonGlobalUsers);
 
   return NumGOTEquivUsers > 0;
 }
@@ -2189,9 +2195,13 @@ void AsmPrinter::computeGlobalGOTEquivs(Module &M) {
 
   for (const auto &G : M.globals()) {
     unsigned NumGOTEquivUsers = 0;
-    if (!isGOTEquivalentCandidate(&G, NumGOTEquivUsers))
+    bool HasNonGlobalUsers = false;
+    if (!isGOTEquivalentCandidate(&G, NumGOTEquivUsers, HasNonGlobalUsers))
       continue;
-
+    // If non-global variables use it, we still need to emit it.
+    // Add 1 here, then emit it in `emitGlobalGOTEquivs`.
+    if (HasNonGlobalUsers)
+      NumGOTEquivUsers += 1;
     const MCSymbol *GOTEquivSym = getSymbol(&G);
     GlobalGOTEquivs[GOTEquivSym] = std::make_pair(&G, NumGOTEquivUsers);
   }

llvm/lib/ObjCopy/MachO/MachOLayoutBuilder.cpp

@@ -116,10 +116,6 @@ uint64_t MachOLayoutBuilder::layoutSegments() {
   const bool IsObjectFile =
       O.Header.FileType == MachO::HeaderFileType::MH_OBJECT;
   uint64_t Offset = IsObjectFile ? (HeaderSize + O.Header.SizeOfCmds) : 0;
-  // If we are emitting an encryptable binary, our load commands must have a
-  // separate (non-encrypted) page to themselves.
-  bool RequiresFirstSectionOutsideFirstPage =
-      O.EncryptionInfoCommandIndex.has_value();
   for (LoadCommand &LC : O.LoadCommands) {
     auto &MLC = LC.MachOLoadCommand;
     StringRef Segname;
@@ -173,10 +169,6 @@ uint64_t MachOLayoutBuilder::layoutSegments() {
         if (!Sec->hasValidOffset()) {
           Sec->Offset = 0;
         } else {
-          if (RequiresFirstSectionOutsideFirstPage) {
-            SectOffset = alignToPowerOf2(SectOffset, PageSize);
-            RequiresFirstSectionOutsideFirstPage = false;
-          }
           Sec->Offset = SegOffset + SectOffset;
           Sec->Size = Sec->Content.size();
           SegFileSize = std::max(SegFileSize, SectOffset + Sec->Size);

llvm/lib/ObjCopy/MachO/MachOObject.cpp

@@ -98,10 +98,6 @@ void Object::updateLoadCommandIndexes() {
     case MachO::LC_DYLD_EXPORTS_TRIE:
       ExportsTrieCommandIndex = Index;
       break;
-    case MachO::LC_ENCRYPTION_INFO:
-    case MachO::LC_ENCRYPTION_INFO_64:
-      EncryptionInfoCommandIndex = Index;
-      break;
     }
   }
 }

llvm/lib/ObjCopy/MachO/MachOObject.h

@@ -341,9 +341,6 @@ struct Object {
   /// The index of the LC_SEGMENT or LC_SEGMENT_64 load command
   /// corresponding to the __TEXT segment.
   std::optional<size_t> TextSegmentCommandIndex;
-  /// The index of the LC_ENCRYPTION_INFO or LC_ENCRYPTION_INFO_64 load command
-  /// if present.
-  std::optional<size_t> EncryptionInfoCommandIndex;
 
   BumpPtrAllocator Alloc;
   StringSaver NewSectionsContents;

llvm/lib/ObjCopy/MachO/MachOReader.cpp

@@ -184,10 +184,6 @@ Error MachOReader::readLoadCommands(Object &O) const {
     case MachO::LC_DYLD_CHAINED_FIXUPS:
       O.ChainedFixupsCommandIndex = O.LoadCommands.size();
       break;
-    case MachO::LC_ENCRYPTION_INFO:
-    case MachO::LC_ENCRYPTION_INFO_64:
-      O.EncryptionInfoCommandIndex = O.LoadCommands.size();
-      break;
     }
 #define HANDLE_LOAD_COMMAND(LCName, LCValue, LCStruct)                         \
   case MachO::LCName:                                                          \

llvm/lib/Transforms/Utils/RelLookupTableConverter.cpp

@@ -67,6 +67,20 @@ static bool shouldConvertToRelLookupTable(Module &M, GlobalVariable &GV) {
   if (!ElemType->isPointerTy() || DL.getPointerTypeSizeInBits(ElemType) != 64)
     return false;
 
+  SmallVector<GlobalVariable *, 4> GVOps;
+  Triple TT(M.getTargetTriple());
+  // FIXME: This should be removed in the future.
+  bool ShouldDropUnnamedAddr =
+      // Drop unnamed_addr to avoid matching pattern in
+      // `handleIndirectSymViaGOTPCRel`, which generates GOTPCREL relocations
+      // not supported by the GNU linker and LLD versions below 18 on aarch64.
+      TT.isAArch64()
+      // Apple's ld64 (and ld-prime on Xcode 15.2) miscompile something on
+      // x86_64-apple-darwin. See
+      // https://github.com/rust-lang/rust/issues/140686 and
+      // https://github.com/rust-lang/rust/issues/141306.
+      || (TT.isX86() && TT.isOSDarwin());
+
   for (const Use &Op : Array->operands()) {
     Constant *ConstOp = cast<Constant>(&Op);
     GlobalValue *GVOp;
@@ -86,8 +100,15 @@ static bool shouldConvertToRelLookupTable(Module &M, GlobalVariable &GV) {
         !GlovalVarOp->isDSOLocal() ||
         !GlovalVarOp->isImplicitDSOLocal())
       return false;
+
+    if (ShouldDropUnnamedAddr)
+      GVOps.push_back(GlovalVarOp);
   }
 
+  if (ShouldDropUnnamedAddr)
+    for (auto *GVOp : GVOps)
+      GVOp->setUnnamedAddr(GlobalValue::UnnamedAddr::None);
+
   return true;
 }
 
@@ -109,24 +130,8 @@ static GlobalVariable *createRelLookupTable(Function &Func,
   uint64_t Idx = 0;
   SmallVector<Constant *, 64> RelLookupTableContents(NumElts);
 
-  Triple TT(M.getTargetTriple());
-  // FIXME: This should be removed in the future.
-  bool ShouldDropUnnamedAddr =
-      // Drop unnamed_addr to avoid matching pattern in
-      // `handleIndirectSymViaGOTPCRel`, which generates GOTPCREL relocations
-      // not supported by the GNU linker and LLD versions below 18 on aarch64.
-      TT.isAArch64()
-      // Apple's ld64 (and ld-prime on Xcode 15.2) miscompile something on
-      // x86_64-apple-darwin. See
-      // https://github.com/rust-lang/rust/issues/140686 and
-      // https://github.com/rust-lang/rust/issues/141306.
-      || (TT.isX86() && TT.isOSDarwin());
-
   for (Use &Operand : LookupTableArr->operands()) {
     Constant *Element = cast<Constant>(Operand);
-    if (ShouldDropUnnamedAddr)
-      if (auto *GlobalElement = dyn_cast<GlobalValue>(Element))
-        GlobalElement->setUnnamedAddr(GlobalValue::UnnamedAddr::None);
     Type *IntPtrTy = M.getDataLayout().getIntPtrType(M.getContext());
     Constant *Base = llvm::ConstantExpr::getPtrToInt(RelLookupTable, IntPtrTy);
     Constant *Target = llvm::ConstantExpr::getPtrToInt(Element, IntPtrTy);

llvm/test/MC/X86/gotpcrel-non-globals.ll

@@ -0,0 +1,36 @@
+; RUN: llc < %s | FileCheck %s
+
+target triple = "x86_64-unknown-linux-gnu"
+
+; Check that we emit the `@bar_*` symbols, and that we don't emit multiple symbols.
+
+; CHECK-LABEL: .Lrel_0:
+; CHECK: .long   foo_0@GOTPCREL+0
+; CHECK-LABEL: .Lrel_1_failed:
+; CHECK: .long   bar_1-foo_0
+; CHECK-LABEL: .Lrel_2:
+; CHECK: .long   foo_2@GOTPCREL+0
+
+; CHECK: bar_0:
+; CHECK: bar_1:
+; CHECK: bar_2_indirect:
+
+@rel_0 = private unnamed_addr constant [1 x i32] [
+  i32 trunc (i64 sub (i64 ptrtoint (ptr @bar_0 to i64), i64 ptrtoint (ptr @rel_0 to i64)) to i32)]
+@rel_1_failed = private unnamed_addr constant [1 x i32] [
+  i32 trunc (i64 sub (i64 ptrtoint (ptr @bar_1 to i64), i64 ptrtoint (ptr @foo_0 to i64)) to i32)]
+@rel_2 = private unnamed_addr constant [1 x i32] [
+  i32 trunc (i64 sub (i64 ptrtoint (ptr @bar_2_indirect to i64), i64 ptrtoint (ptr @rel_2 to i64)) to i32)]
+@bar_0 = internal unnamed_addr constant ptr @foo_0, align 8
+@bar_1 = internal unnamed_addr constant ptr @foo_1, align 8
+@bar_2_indirect = internal unnamed_addr constant ptr @foo_2, align 8
+@foo_0 = external global ptr, align 8
+@foo_1 = external global ptr, align 8
+@foo_2 = external global ptr, align 8
+
+define void @foo(ptr %arg0, ptr %arg1) {
+  store ptr @bar_0, ptr %arg0, align 8
+  store ptr @bar_1, ptr %arg1, align 8
+  store ptr getelementptr (i8, ptr @bar_2_indirect, i32 1), ptr %arg1, align 8
+  ret void
+}

llvm/test/Transforms/RelLookupTableConverter/unnamed_addr.ll

@@ -20,6 +20,14 @@
 @y3 = internal unnamed_addr constant ptr @x0
 @load_relative_2.table = private unnamed_addr constant [4 x ptr] [ptr @y3, ptr @y2, ptr @y1, ptr @y0]
 
+@b0 = private unnamed_addr constant [8 x i8] c"00000000"
+@b1 = private unnamed_addr constant [8 x i8] c"11111111"
+@b2 = private unnamed_addr constant [8 x i8] c"22222222"
+@load_relative_3.table = private unnamed_addr constant [3 x ptr] [
+  ptr getelementptr inbounds (i8, ptr @b0, i64 8),
+  ptr getelementptr inbounds (i8, ptr @b1, i64 8),
+  ptr getelementptr inbounds (i8, ptr @b2, i64 8)]
+
 ;.
 ; x86_64-apple-darwin: @a0 = private constant i32 0
 ; x86_64-apple-darwin: @a1 = private constant i32 1
@@ -34,6 +42,10 @@
 ; x86_64-apple-darwin: @y2 = internal constant ptr @x1
 ; x86_64-apple-darwin: @y3 = internal constant ptr @x0
 ; x86_64-apple-darwin: @load_relative_2.table.rel = private unnamed_addr constant [4 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr @y3 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y2 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y1 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y0 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32)], align 4
+; x86_64-apple-darwin: @b0 = private constant [8 x i8] c"00000000"
+; x86_64-apple-darwin: @b1 = private constant [8 x i8] c"11111111"
+; x86_64-apple-darwin: @b2 = private constant [8 x i8] c"22222222"
+; x86_64-apple-darwin: @load_relative_3.table.rel = private unnamed_addr constant [3 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b0, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b1, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b2, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32)], align 4
 ;.
 ; aarch64: @a0 = private constant i32 0
 ; aarch64: @a1 = private constant i32 1
@@ -48,6 +60,10 @@
 ; aarch64: @y2 = internal constant ptr @x1
 ; aarch64: @y3 = internal constant ptr @x0
 ; aarch64: @load_relative_2.table.rel = private unnamed_addr constant [4 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr @y3 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y2 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y1 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y0 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32)], align 4
+; aarch64: @b0 = private constant [8 x i8] c"00000000"
+; aarch64: @b1 = private constant [8 x i8] c"11111111"
+; aarch64: @b2 = private constant [8 x i8] c"22222222"
+; aarch64: @load_relative_3.table.rel = private unnamed_addr constant [3 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b0, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b1, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b2, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32)], align 4
 ;.
 ; x86_64: @a0 = private unnamed_addr constant i32 0
 ; x86_64: @a1 = private unnamed_addr constant i32 1
@@ -62,6 +78,10 @@
 ; x86_64: @y2 = internal unnamed_addr constant ptr @x1
 ; x86_64: @y3 = internal unnamed_addr constant ptr @x0
 ; x86_64: @load_relative_2.table.rel = private unnamed_addr constant [4 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr @y3 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y2 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y1 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr @y0 to i64), i64 ptrtoint (ptr @load_relative_2.table.rel to i64)) to i32)], align 4
+; x86_64: @b0 = private unnamed_addr constant [8 x i8] c"00000000"
+; x86_64: @b1 = private unnamed_addr constant [8 x i8] c"11111111"
+; x86_64: @b2 = private unnamed_addr constant [8 x i8] c"22222222"
+; x86_64: @load_relative_3.table.rel = private unnamed_addr constant [3 x i32] [i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b0, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b1, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32), i32 trunc (i64 sub (i64 ptrtoint (ptr getelementptr inbounds (i8, ptr @b2, i64 8) to i64), i64 ptrtoint (ptr @load_relative_3.table.rel to i64)) to i32)], align 4
 ;.
 define ptr @load_relative_1(i64 %offset) {
 ; x86_64-apple-darwin-LABEL: define ptr @load_relative_1(
@@ -110,6 +130,31 @@ define ptr @load_relative_2(i64 %offset) {
   %load = load ptr, ptr %gep
   ret ptr %load
 }
+
+define ptr @load_relative_3(i64 %offset) {
+; x86_64-apple-darwin-LABEL: define ptr @load_relative_3(
+; x86_64-apple-darwin-SAME: i64 [[OFFSET:%.*]]) {
+; x86_64-apple-darwin-NEXT:    [[RELTABLE_SHIFT:%.*]] = shl i64 [[OFFSET]], 2
+; x86_64-apple-darwin-NEXT:    [[RELTABLE_INTRINSIC:%.*]] = call ptr @llvm.load.relative.i64(ptr @load_relative_3.table.rel, i64 [[RELTABLE_SHIFT]])
+; x86_64-apple-darwin-NEXT:    ret ptr [[RELTABLE_INTRINSIC]]
+;
+; aarch64-LABEL: define ptr @load_relative_3(
+; aarch64-SAME: i64 [[OFFSET:%.*]]) {
+; aarch64-NEXT:    [[RELTABLE_SHIFT:%.*]] = shl i64 [[OFFSET]], 2
+; aarch64-NEXT:    [[RELTABLE_INTRINSIC:%.*]] = call ptr @llvm.load.relative.i64(ptr @load_relative_3.table.rel, i64 [[RELTABLE_SHIFT]])
+; aarch64-NEXT:    ret ptr [[RELTABLE_INTRINSIC]]
+;
+; x86_64-LABEL: define ptr @load_relative_3(
+; x86_64-SAME: i64 [[OFFSET:%.*]]) {
+; x86_64-NEXT:    [[RELTABLE_SHIFT:%.*]] = shl i64 [[OFFSET]], 2
+; x86_64-NEXT:    [[RELTABLE_INTRINSIC:%.*]] = call ptr @llvm.load.relative.i64(ptr @load_relative_3.table.rel, i64 [[RELTABLE_SHIFT]])
+; x86_64-NEXT:    ret ptr [[RELTABLE_INTRINSIC]]
+;
+  %gep = getelementptr inbounds [3 x ptr], ptr @load_relative_3.table, i64 0, i64 %offset
+  %load = load ptr, ptr %gep
+  ret ptr %load
+}
+
 ;.
 ; x86_64-apple-darwin: attributes #[[ATTR0:[0-9]+]] = { nocallback nofree nosync nounwind willreturn memory(argmem: read) }
 ;.