Merge pull request swiftlang#71182 from aschwaighofer/oslog_string_section

aschwaighofer · web-flow · commit 204b20c4adc0 · 2024-01-26T13:37:10.000-08:00
OSLog: os_log strings should be in the __oslogstring section on machO
diff --git a/SwiftCompilerSources/Sources/SIL/GlobalVariable.swift b/SwiftCompilerSources/Sources/SIL/GlobalVariable.swift
@@ -126,7 +126,7 @@ extension Instruction {
       return false
     case let sli as StringLiteralInst:
       switch sli.encoding {
-      case .Bytes, .UTF8:
+      case .Bytes, .UTF8, .UTF8_OSLOG:
         return true
       case .ObjCSelector:
         // Objective-C selector string literals cannot be used in static
diff --git a/SwiftCompilerSources/Sources/SIL/Instruction.swift b/SwiftCompilerSources/Sources/SIL/Instruction.swift
@@ -750,6 +750,7 @@ final public class StringLiteralInst : SingleValueInstruction {
     case UTF8
     /// UTF-8 encoding of an Objective-C selector.
     case ObjCSelector
+    case UTF8_OSLOG
   }
 
   public var value: StringRef { StringRef(bridged: bridged.StringLiteralInst_getValue()) }
@@ -759,6 +760,7 @@ final public class StringLiteralInst : SingleValueInstruction {
     case 0: return .Bytes
     case 1: return .UTF8
     case 2: return .ObjCSelector
+    case 3: return .UTF8_OSLOG
     default: fatalError("invalid encoding in StringLiteralInst")
     }
   }
diff --git a/include/swift/SIL/SILInstruction.h b/include/swift/SIL/SILInstruction.h
@@ -4328,6 +4328,7 @@ class StringLiteralInst final
     UTF8 = 1,
     /// UTF-8 encoding of an Objective-C selector.
     ObjCSelector = 2,
+    UTF8_OSLOG = 3,
   };
 
 private:
diff --git a/lib/IRGen/GenConstant.cpp b/lib/IRGen/GenConstant.cpp
@@ -106,10 +106,14 @@ llvm::Constant *irgen::emitConstantFP(IRGenModule &IGM, FloatLiteralInst *FLI) {
 
 llvm::Constant *irgen::emitAddrOfConstantString(IRGenModule &IGM,
                                                 StringLiteralInst *SLI) {
-  switch (SLI->getEncoding()) {
+  auto encoding = SLI->getEncoding();
+  bool useOSLogEncoding = encoding == StringLiteralInst::Encoding::UTF8_OSLOG;
+
+  switch (encoding) {
   case StringLiteralInst::Encoding::Bytes:
   case StringLiteralInst::Encoding::UTF8:
-    return IGM.getAddrOfGlobalString(SLI->getValue());
+  case StringLiteralInst::Encoding::UTF8_OSLOG:
+    return IGM.getAddrOfGlobalString(SLI->getValue(), false, useOSLogEncoding);
 
   case StringLiteralInst::Encoding::ObjCSelector:
     llvm_unreachable("cannot get the address of an Objective-C selector");
diff --git a/lib/IRGen/GenDecl.cpp b/lib/IRGen/GenDecl.cpp
@@ -5878,9 +5878,15 @@ Address IRGenFunction::createAlloca(llvm::Type *type,
 /// resolving relative references to coalesceable symbols.
 /// It should be removed when fixed. rdar://problem/22674524
 llvm::Constant *IRGenModule::getAddrOfGlobalString(StringRef data,
-                                               bool willBeRelativelyAddressed) {
+                                               bool willBeRelativelyAddressed,
+                                               bool useOSLogSection) {
+  useOSLogSection = useOSLogSection &&
+    TargetInfo.OutputObjectFormat == llvm::Triple::MachO;
+
   // Check whether this string already exists.
-  auto &entry = GlobalStrings[data];
+  auto &entry = useOSLogSection ? GlobalOSLogStrings[data] :
+    GlobalStrings[data];
+
   if (entry.second) {
     // FIXME: Clear unnamed_addr if the global will be relative referenced
     // to work around an ld64 bug. rdar://problem/22674524
@@ -5900,9 +5906,12 @@ llvm::Constant *IRGenModule::getAddrOfGlobalString(StringRef data,
     name[i] = '_';
     (llvm::Twine(".nul") + llvm::Twine(i)).toVector(name);
   }
-  
+
+  auto sectionName =
+    useOSLogSection ? "__TEXT,__oslogstring,cstring_literals" : "";
+
   entry = createStringConstant(data, willBeRelativelyAddressed,
-                               /*sectionName*/ "", name);
+                               sectionName, name);
   return entry.second;
 }
 
diff --git a/lib/IRGen/IRGenModule.h b/lib/IRGen/IRGenModule.h
@@ -1108,7 +1108,8 @@ class IRGenModule {
       StringRef Str, bool willBeRelativelyAddressed = false,
       StringRef sectionName = "", StringRef name = "");
   llvm::Constant *getAddrOfGlobalString(StringRef utf8,
-                                        bool willBeRelativelyAddressed = false);
+                                        bool willBeRelativelyAddressed = false,
+                                        bool useOSLogSection = false);
   llvm::Constant *getAddrOfGlobalUTF16String(StringRef utf8);
   llvm::Constant *getAddrOfObjCSelectorRef(StringRef selector);
   llvm::Constant *getAddrOfObjCSelectorRef(SILDeclRef method);
@@ -1218,6 +1219,8 @@ class IRGenModule {
   llvm::DenseSet<const clang::Decl *> GlobalClangDecls;
   llvm::StringMap<std::pair<llvm::GlobalVariable*, llvm::Constant*>>
     GlobalStrings;
+  llvm::StringMap<std::pair<llvm::GlobalVariable*, llvm::Constant*>>
+    GlobalOSLogStrings;
   llvm::StringMap<llvm::Constant*> GlobalUTF16Strings;
   llvm::StringMap<std::pair<llvm::GlobalVariable*, llvm::Constant*>>
     StringsForTypeRef;
diff --git a/lib/SIL/IR/SILPrinter.cpp b/lib/SIL/IR/SILPrinter.cpp
@@ -1674,6 +1674,7 @@ class SILPrinter : public SILInstructionVisitor<SILPrinter> {
     switch (kind) {
     case StringLiteralInst::Encoding::Bytes: return "bytes ";
     case StringLiteralInst::Encoding::UTF8: return "utf8 ";
+    case StringLiteralInst::Encoding::UTF8_OSLOG: return "oslog ";
     case StringLiteralInst::Encoding::ObjCSelector: return "objc_selector ";
     }
     llvm_unreachable("bad string literal encoding");
diff --git a/lib/SIL/Parser/ParseSIL.cpp b/lib/SIL/Parser/ParseSIL.cpp
@@ -2729,6 +2729,8 @@ bool SILParser::parseSpecificSILInstruction(SILBuilder &B,
       encoding = StringLiteralInst::Encoding::ObjCSelector;
     } else if (P.Tok.getText() == "bytes") {
       encoding = StringLiteralInst::Encoding::Bytes;
+    } else if (P.Tok.getText() == "oslog") {
+      encoding = StringLiteralInst::Encoding::UTF8_OSLOG;
     } else {
       P.diagnose(P.Tok, diag::sil_string_invalid_encoding, P.Tok.getText());
       return true;
diff --git a/lib/SILGen/SILGenApply.cpp b/lib/SILGen/SILGenApply.cpp
@@ -1852,6 +1852,7 @@ static PreparedArguments emitStringLiteralArgs(SILGenFunction &SGF, SILLocation
     break;
 
   case StringLiteralInst::Encoding::Bytes:
+  case StringLiteralInst::Encoding::UTF8_OSLOG:
   case StringLiteralInst::Encoding::ObjCSelector:
     llvm_unreachable("these cannot be formed here");
   }
diff --git a/lib/SILOptimizer/Mandatory/OSLogOptimization.cpp b/lib/SILOptimizer/Mandatory/OSLogOptimization.cpp
@@ -637,7 +637,7 @@ static SILValue emitCodeForSymbolicValue(SymbolicValue symVal,
 
     StringRef stringVal = symVal.getStringValue();
     StringLiteralInst *stringLitInst = builder.createStringLiteral(
-        loc, stringVal, StringLiteralInst::Encoding::UTF8);
+        loc, stringVal, StringLiteralInst::Encoding::UTF8_OSLOG);
 
     // Create a builtin word for the size of the string
     IntegerLiteralInst *sizeInst = builder.createIntegerLiteral(
@@ -1427,7 +1427,7 @@ suppressGlobalStringTablePointerError(SingleValueInstruction *oslogMessage) {
   for (BuiltinInst *bi : globalStringTablePointerInsts) {
     SILBuilderWithScope builder(bi);
     StringLiteralInst *stringLiteral = builder.createStringLiteral(
-        bi->getLoc(), StringRef(""), StringLiteralInst::Encoding::UTF8);
+        bi->getLoc(), StringRef(""), StringLiteralInst::Encoding::UTF8_OSLOG);
     bi->replaceAllUsesWith(stringLiteral);
     // The builtin instruction is likely dead. But since we are iterating over
     // many instructions, do the cleanup at the end.
diff --git a/lib/SILOptimizer/Utils/SpecializationMangler.cpp b/lib/SILOptimizer/Utils/SpecializationMangler.cpp
@@ -180,6 +180,7 @@ FunctionSignatureSpecializationMangler::mangleConstantProp(SILInstruction *const
     switch (SLI->getEncoding()) {
       case StringLiteralInst::Encoding::Bytes: ArgOpBuffer << 'B'; break;
       case StringLiteralInst::Encoding::UTF8: ArgOpBuffer << 'b'; break;
+      case StringLiteralInst::Encoding::UTF8_OSLOG: ArgOpBuffer << 'o'; break;
       case StringLiteralInst::Encoding::ObjCSelector: ArgOpBuffer << 'c'; break;
     }
     break;
diff --git a/lib/Serialization/DeserializeSIL.cpp b/lib/Serialization/DeserializeSIL.cpp
@@ -60,6 +60,7 @@ fromStableStringEncoding(unsigned value) {
   case SIL_BYTES: return StringLiteralInst::Encoding::Bytes;
   case SIL_UTF8: return StringLiteralInst::Encoding::UTF8;
   case SIL_OBJC_SELECTOR: return StringLiteralInst::Encoding::ObjCSelector;
+  case SIL_UTF8_OSLOG: return StringLiteralInst::Encoding::UTF8_OSLOG;
   default:
     return llvm::None;
   }
diff --git a/lib/Serialization/ModuleFormat.h b/lib/Serialization/ModuleFormat.h
@@ -58,7 +58,7 @@ const uint16_t SWIFTMODULE_VERSION_MAJOR = 0;
 /// describe what change you made. The content of this comment isn't important;
 /// it just ensures a conflict if two people change the module format.
 /// Don't worry about adhering to the 80-column limit for this line.
-const uint16_t SWIFTMODULE_VERSION_MINOR = 842; // function isolation kinds
+const uint16_t SWIFTMODULE_VERSION_MINOR = 843; // add OSLog string encoding
 
 /// A standard hash seed used for all string hashes in a serialized module.
 ///
diff --git a/lib/Serialization/SILFormat.h b/lib/Serialization/SILFormat.h
@@ -33,7 +33,8 @@ using SILValueOwnershipField = BCFixed<2>;
 enum SILStringEncoding : uint8_t {
   SIL_UTF8,
   SIL_OBJC_SELECTOR,
-  SIL_BYTES
+  SIL_BYTES,
+  SIL_UTF8_OSLOG
 };
 
 enum SILLinkageEncoding : uint8_t {
diff --git a/lib/Serialization/SerializeSIL.cpp b/lib/Serialization/SerializeSIL.cpp
@@ -53,6 +53,7 @@ static unsigned toStableStringEncoding(StringLiteralInst::Encoding encoding) {
   case StringLiteralInst::Encoding::Bytes: return SIL_BYTES;
   case StringLiteralInst::Encoding::UTF8: return SIL_UTF8;
   case StringLiteralInst::Encoding::ObjCSelector: return SIL_OBJC_SELECTOR;
+  case StringLiteralInst::Encoding::UTF8_OSLOG: return SIL_UTF8_OSLOG;
   }
   llvm_unreachable("bad string encoding");
 }
diff --git a/test/IRGen/oslog_literals.sil b/test/IRGen/oslog_literals.sil
@@ -0,0 +1,19 @@
+// RUN: %target-swift-frontend -emit-ir %s | %FileCheck %s
+
+// REQUIRES: OS=macosx || OS=ios || OS=watchos
+
+// CHECK: [[U8_0:@.*]] = private unnamed_addr constant [8 x i8] c"help\09me\00", section "__TEXT,__oslogstring,cstring_literals"
+
+sil_stage canonical
+
+import Builtin
+
+sil @oslog_utf8_literal : $@convention(thin) () -> Builtin.RawPointer {
+bb0:
+  %0 = string_literal oslog "help\tme"
+  return %0 : $Builtin.RawPointer
+}
+// CHECK: define swiftcc ptr @oslog_utf8_literal() {{.*}} {
+// CHECK:   ret ptr [[U8_0]]
+
+
diff --git a/test/SIL/Parser/string_literal.sil b/test/SIL/Parser/string_literal.sil
@@ -6,6 +6,8 @@ bb0:
   %1 = string_literal utf8 "\u{0B}"
 // CHECK: string_literal bytes "ABCD"
   %4 = string_literal bytes "ABCD"
+// CHECK: string_literal oslog "os log section"
+  %5 = string_literal oslog "os log section"
   %6 = tuple ()                                   // user: %7
   return %6 : $()
 }
diff --git a/test/SIL/Serialization/literals.sil b/test/SIL/Serialization/literals.sil
@@ -12,6 +12,8 @@ bb0:
   %1 = string_literal utf8 "\u{0B}"
 // CHECK: string_literal bytes "ABCD"
   %4 = string_literal bytes "ABCD"
+// CHECK: string_literal oslog "os log section"
+  %5 = string_literal oslog "os log section"
   %6 = tuple ()
   return %6 : $()
 }
diff --git a/test/SILOptimizer/OSLogMandatoryOptTest.sil b/test/SILOptimizer/OSLogMandatoryOptTest.sil
@@ -38,7 +38,7 @@ sil @useFormatString: $@convention(thin) (@guaranteed String) -> ()
 // CHECK-DAG: [[BORROW]] = begin_borrow [[STRINGCONST:%[0-9]+]]
 // CHECK-DAG: [[STRINGCONST]] = apply [[STRINGINIT:%[0-9]+]]([[LIT:%[0-9]+]], {{%.*}}, {{%.*}}, {{%.*}})
 // CHECK-DAG: [[STRINGINIT]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
-// CHECK-DAG: [[LIT]] = string_literal utf8 "test message: %lld"
+// CHECK-DAG: [[LIT]] = string_literal oslog "test message: %lld"
 // CHECK-DAG: destroy_value [[STRINGCONST]] : $String
 sil [ossa] @testConstantFoldingOfStructExtract : $@convention(thin) () -> () {
 bb0:
@@ -98,7 +98,7 @@ bb0:
   dealloc_stack %11 : $*String
   %15 = tuple ()
   return %15 : $()
-    // CHECK: [[LIT:%[0-9]+]] = string_literal utf8 "some long message: %llx"
+    // CHECK: [[LIT:%[0-9]+]] = string_literal oslog "some long message: %llx"
     // CHECK: [[STRINGINIT:%[0-9]+]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
     // CHECK: [[STRINGCONST:%[0-9]+]] = apply [[STRINGINIT]]([[LIT]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK: [[BORROW:%[0-9]+]] = begin_borrow [[STRINGCONST]]
@@ -136,7 +136,7 @@ bb0:
     // CHECK-DAG: {{%.*}} = apply [[STRINGUSE]]([[STRINGCONST:%[0-9]+]])
     // CHECK-DAG: [[STRINGCONST]] = apply [[STRINGINIT:%[0-9]+]]([[LIT:%[0-9]+]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK-DAG: [[STRINGINIT]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
-    // CHECK-DAG: [[LIT]] = string_literal utf8 "test message: %lld"
+    // CHECK-DAG: [[LIT]] = string_literal oslog "test message: %lld"
     // FIXME: function_ref @extractFormatStringAsOwned will not be removed as of
     // now by OSLogOptimization pass even though it is folded as function calls
     // are not dead-code eliminated.
@@ -177,7 +177,7 @@ bb0:
     // CHECK-DAG {{%.*}} = apply [[STRINGUSE]]([[STRINGCONST:%[0-9]+]])
     // CHECK-DAG [[STRINGCONST]] = apply [[STRINGINIT:%[0-9]+]]([[LIT:%[0-9]+]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK-DAG [[STRINGINIT]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
-    // CHECK-DAG [[LIT]] = string_literal utf8 "test message: %lld"
+    // CHECK-DAG [[LIT]] = string_literal oslog "test message: %lld"
     // CHECK-DAG destroy_value [[STRINGCONST]] : $String
 }
 
@@ -219,7 +219,7 @@ bb0:
     // CHECK-DAG: [[BORROW2]] = begin_borrow [[STRINGCONST:%[0-9]+]]
     // CHECK-DAG: [[STRINGCONST]] = apply [[STRINGINIT:%[0-9]+]]([[LIT:%[0-9]+]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK-DAG: [[STRINGINIT]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
-    // CHECK-DAG: [[LIT]] = string_literal utf8 "test message: %lld"
+    // CHECK-DAG: [[LIT]] = string_literal oslog "test message: %lld"
     // CHECK-DAG: destroy_value [[STRINGCONST]] : $String
 }
 
@@ -271,12 +271,12 @@ bb5:
     // We must have all string literals at the beginning of the borrowed scope,
     // and destroys of the literals at the end of the borrow scope.
 
-    // CHECK: [[LIT:%[0-9]+]] = string_literal utf8 "test message: %lld"
+    // CHECK: [[LIT:%[0-9]+]] = string_literal oslog "test message: %lld"
     // CHECK: [[STRINGINIT:%[0-9]+]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
     // CHECK: [[STRINGCONST:%[0-9]+]] = apply [[STRINGINIT]]([[LIT]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK: [[BORROW:%[0-9]+]] = begin_borrow [[STRINGCONST]]
 
-    // CHECK: [[LIT2:%[0-9]+]] = string_literal utf8 "test message: %lld"
+    // CHECK: [[LIT2:%[0-9]+]] = string_literal oslog "test message: %lld"
     // CHECK: [[STRINGINIT2:%[0-9]+]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
     // CHECK: [[STRINGCONST2:%[0-9]+]] = apply [[STRINGINIT2]]([[LIT2]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK: [[BORROW2:%[0-9]+]] = begin_borrow [[STRINGCONST2]]
@@ -324,7 +324,7 @@ bb0:
     // CHECK-DAG: [[BORROW]] = begin_borrow [[STRINGCONST:%[0-9]+]]
     // CHECK-DAG: [[STRINGCONST]] = apply [[STRINGINIT:%[0-9]+]]([[LIT:%[0-9]+]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK-DAG: [[STRINGINIT]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
-    // CHECK-DAG: [[LIT]] = string_literal utf8 "test message: %lld"
+    // CHECK-DAG: [[LIT]] = string_literal oslog "test message: %lld"
     // CHECK-DAG: destroy_value [[STRINGCONST]] : $String
 }
 
@@ -532,7 +532,7 @@ bb0:
   %8 = tuple ()
   return %8 : $()
     // The first instance of "ab" will be dead code eliminated.
-    // CHECK: [[LIT:%[0-9]+]] = string_literal utf8 "ab"
+    // CHECK: [[LIT:%[0-9]+]] = string_literal oslog "ab"
     // CHECK: [[STRINGINIT:%[0-9]+]] = function_ref @$sSS21_builtinStringLiteral17utf8CodeUnitCount7isASCIISSBp_BwBi1_tcfC
     // CHECK: [[STRINGCONST:%[0-9]+]] = apply [[STRINGINIT]]([[LIT]], {{%.*}}, {{%.*}}, {{%.*}})
     // CHECK: [[NUMELEMS:%[0-9]+]] = integer_literal $Builtin.Word, 1
diff --git a/test/SILOptimizer/OSLogMandatoryOptTest.swift b/test/SILOptimizer/OSLogMandatoryOptTest.swift

Original file line number	Diff line number	Diff line change
`@@ -750,6 +750,7 @@ final public class StringLiteralInst : SingleValueInstruction {`
`750`	`750`	`case UTF8`
`751`	`751`	`/// UTF-8 encoding of an Objective-C selector.`
`752`	`752`	`case ObjCSelector`
	`753`	`+ case UTF8_OSLOG`
`753`	`754`	`}`
`754`	`755`
`755`	`756`	`public var value: StringRef { StringRef(bridged: bridged.StringLiteralInst_getValue()) }`
`@@ -759,6 +760,7 @@ final public class StringLiteralInst : SingleValueInstruction {`
`759`	`760`	`case 0: return .Bytes`
`760`	`761`	`case 1: return .UTF8`
`761`	`762`	`case 2: return .ObjCSelector`
	`763`	`+ case 3: return .UTF8_OSLOG`
`762`	`764`	`default: fatalError("invalid encoding in StringLiteralInst")`
`763`	`765`	`}`
`764`	`766`	`}`
Original file line number	Diff line number	Diff line change
`@@ -1674,6 +1674,7 @@ class SILPrinter : public SILInstructionVisitor<SILPrinter> {`
`1674`	`1674`	`switch (kind) {`
`1675`	`1675`	`case StringLiteralInst::Encoding::Bytes: return "bytes ";`
`1676`	`1676`	`case StringLiteralInst::Encoding::UTF8: return "utf8 ";`
	`1677`	`+ case StringLiteralInst::Encoding::UTF8_OSLOG: return "oslog ";`
`1677`	`1678`	`case StringLiteralInst::Encoding::ObjCSelector: return "objc_selector ";`
`1678`	`1679`	`}`
`1679`	`1680`	`llvm_unreachable("bad string literal encoding");`
Original file line number	Diff line number	Diff line change
`@@ -1852,6 +1852,7 @@ static PreparedArguments emitStringLiteralArgs(SILGenFunction &SGF, SILLocation`
`1852`	`1852`	`break;`
`1853`	`1853`
`1854`	`1854`	`case StringLiteralInst::Encoding::Bytes:`
	`1855`	`+ case StringLiteralInst::Encoding::UTF8_OSLOG:`
`1855`	`1856`	`case StringLiteralInst::Encoding::ObjCSelector:`
`1856`	`1857`	`llvm_unreachable("these cannot be formed here");`
`1857`	`1858`	`}`