Merge pull request #8820 from adrian-prantl/128571037-6.0

Fix the dsymutil heuristic for excluding system interfaces.  (llvm#93745)

Author: adrian-prantl

llvm/include/llvm/DWARFLinker/Utils.h

@@ -37,17 +37,62 @@ inline Error finiteLoop(function_ref<Expected<bool>()> Iteration,
 }
 
 /// Make a best effort to guess the
-/// Xcode.app/Contents/Developer/Toolchains/ path from an SDK path.
-inline SmallString<128> guessToolchainBaseDir(StringRef SysRoot) {
-  SmallString<128> Result;
+/// Xcode.app/Contents/Developer path from an SDK path.
+inline StringRef guessDeveloperDir(StringRef SysRoot) {
   // Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk
-  StringRef Base = sys::path::parent_path(SysRoot);
-  if (sys::path::filename(Base) != "SDKs")
-    return Result;
-  Base = sys::path::parent_path(Base);
-  Result = Base;
-  Result += "/Toolchains";
-  return Result;
+  auto it = sys::path::rbegin(SysRoot);
+  auto end = sys::path::rend(SysRoot);
+  if (it == end || !it->ends_with(".sdk"))
+    return {};
+  ++it;
+  // Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs
+  if (it == end || *it != "SDKs")
+    return {};
+  auto developerEnd = it;
+  ++it;
+  while (it != end) {
+    // Contents/Developer/Platforms/MacOSX.platform/Developer
+    if (*it != "Developer")
+      return {};
+    ++it;
+    if (it == end)
+      return {};
+    if (*it == "Contents")
+      return StringRef(SysRoot.data(),
+                       developerEnd - sys::path::rend(SysRoot) - 1);
+    // Contents/Developer/Platforms/MacOSX.platform
+    if (!it->ends_with(".platform"))
+      return {};
+    ++it;
+    // Contents/Developer/Platforms
+    if (it == end || *it != "Platforms")
+      return {};
+    developerEnd = it;
+    ++it;
+  }
+  return {};
+}
+
+/// Make a best effort to determine whether Path is inside a toolchain.
+inline bool isInToolchainDir(StringRef Path) {
+  // Library/Developer/Toolchains/swift-DEVELOPMENT-SNAPSHOT-2024-05-15-a.xctoolchain/usr/lib/swift/macosx/_StringProcessing.swiftmodule/arm64-apple-macos.private.swiftinterface
+  for (auto it = sys::path::rbegin(Path), end = sys::path::rend(Path);
+       it != end; ++it) {
+    if (it->ends_with(".xctoolchain")) {
+      ++it;
+      if (it == end)
+        return false;
+      if (*it != "Toolchains")
+        return false;
+      ++it;
+      if (it == end)
+        return false;
+      if (*it != "Developer")
+        return false;
+      return true;
+    }
+  }
+  return false;
 }
 
 inline bool isPathAbsoluteOnWindowsOrPosix(const Twine &Path) {

llvm/lib/DWARFLinker/Classic/DWARFLinker.cpp

@@ -200,8 +200,10 @@ static void analyzeImportedModule(
     return;
   // Don't track interfaces that are part of the toolchain.
   // For example: Swift, _Concurrency, ...
-  SmallString<128> Toolchain = guessToolchainBaseDir(SysRoot);
-  if (!Toolchain.empty() && Path.startswith(Toolchain))
+  StringRef DeveloperDir = guessDeveloperDir(SysRoot);
+  if (!DeveloperDir.empty() && Path.starts_with(DeveloperDir))
+    return;
+  if (isInToolchainDir(Path))
     return;
   std::optional<const char *> Name =
       dwarf::toString(DIE.find(dwarf::DW_AT_name));

llvm/lib/DWARFLinker/Parallel/DWARFLinkerCompileUnit.cpp

@@ -266,7 +266,14 @@ void CompileUnit::analyzeImportedModule(const DWARFDebugInfoEntry *DieEntry) {
       dwarf::toStringRef(find(DieEntry, dwarf::DW_AT_LLVM_sysroot));
   if (SysRoot.empty())
     SysRoot = getSysRoot();
-  if (!SysRoot.empty() && Path.startswith(SysRoot))
+  if (!SysRoot.empty() && Path.starts_with(SysRoot))
+    return;
+  // Don't track interfaces that are part of the toolchain.
+  // For example: Swift, _Concurrency, ...
+  StringRef DeveloperDir = guessDeveloperDir(SysRoot);
+  if (!DeveloperDir.empty() && Path.starts_with(DeveloperDir))
+    return;
+  if (isInToolchainDir(Path))
     return;
   if (std::optional<DWARFFormValue> Val = find(DieEntry, dwarf::DW_AT_name)) {
     Expected<const char *> Name = Val->getAsCString();

llvm/unittests/DWARFLinkerParallel/DWARFLinkerTest.cpp

@@ -5,3 +5,31 @@
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
+
+#include "llvm/DWARFLinker/Utils.h"
+#include "gtest/gtest.h"
+
+using namespace llvm;
+using namespace dwarf_linker;
+
+#define DEVELOPER_DIR "/Applications/Xcode.app/Contents/Developer"
+
+namespace {
+
+TEST(DWARFLinker, PathTest) {
+  EXPECT_EQ(guessDeveloperDir("/Foo"), "");
+  EXPECT_EQ(guessDeveloperDir("Foo.sdk"), "");
+  EXPECT_EQ(guessDeveloperDir(
+                DEVELOPER_DIR
+                "/Platforms/MacOSX.platform/Developer/SDKs/MacOSX14.4.sdk"),
+            DEVELOPER_DIR);
+  EXPECT_EQ(guessDeveloperDir(DEVELOPER_DIR "/SDKs/MacOSX.sdk"), DEVELOPER_DIR);
+  EXPECT_TRUE(
+      isInToolchainDir("/Library/Developer/Toolchains/"
+                       "swift-DEVELOPMENT-SNAPSHOT-2024-05-15-a.xctoolchain/"
+                       "usr/lib/swift/macosx/_StringProcessing.swiftmodule/"
+                       "arm64-apple-macos.private.swiftinterface"));
+  EXPECT_FALSE(isInToolchainDir("/Foo/not-an.xctoolchain/Bar/Baz"));
+}
+
+} // anonymous namespace