[DebugInfo] Normalize paths by removing unnecessary dots

[phosek]

Summary

This was suggested in https://reviews.llvm.org/D87657 as a better
alternative which doesn't require having to guess separators.

[dblaikie]

Does this address @rnk's feedback about symlinks? ( https://reviews.llvm.org/D87657#2296028 )

[phosek]

In D95772#2534103, @dblaikie wrote:

Does this address @rnk's feedback about symlinks? ( https://reviews.llvm.org/D87657#2296028 )

It doesn't, I think if we want to handle that case, we'd need to use sys::fs::real_path rather than sys::path::remove_dots. That should produce correct result but it's more expensive.

While working on coverage mapping, I've noticed that we already use sys::path::remove_dots with remove_dots_dots=true to normalize filenames there. Whichever way we decide on, I think it'd be useful to unify the two.

clang/lib/CodeGen/CGDebugInfo.cpp

if (llvm::sys::path::is_absolute(RemappedFile)) {
     File = RemappedFile;
   }
   llvm::DIFile *F = DBuilder.createFile(File, Dir, CSInfo, Source);
   DIFileCache[FileName.data()].reset(F);
   return F;
 }
 
 std::string CGDebugInfo::remapDIPath(StringRef Path) const {
-  if (DebugPrefixMap.empty())
-    return Path.str();
-
   SmallString<256> P = Path;
+  llvm::sys::path::remove_dots(P, /*remove_dot_dot=*/true);
   for (const auto &Entry : DebugPrefixMap)
     if (llvm::sys::path::replace_path_prefix(P, Entry.first, Entry.second))
       break;
   return P.str().str();
 }
 
 unsigned CGDebugInfo::getLineNumber(SourceLocation Loc) {
   if (Loc.isInvalid())

clang/test/CodeGen/debug-info-abspath.c

 // RUN: mkdir -p %t/UNIQUEISH_SENTINEL
 // RUN: cp %s %t/UNIQUEISH_SENTINEL/debug-info-abspath.c
 
 // RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
 // RUN:   %t/UNIQUEISH_SENTINEL/debug-info-abspath.c -emit-llvm -o - \
 // RUN:   | FileCheck %s
 
+// RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
+// RUN:   %t/UNIQUEISH_SENTINEL/../UNIQUEISH_SENTINEL/debug-info-abspath.c \
+// RUN:   -emit-llvm -o - | FileCheck %s
+
 // RUN: cp %s %t.c
 // RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
 // RUN:   %t.c -emit-llvm -o - | FileCheck %s --check-prefix=INTREE
 
 // RUN: cd %t/UNIQUEISH_SENTINEL
 // RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
 // RUN:   debug-info-abspath.c -emit-llvm -o - \
 // RUN:   | FileCheck %s --check-prefix=CURDIR
 // RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
 // RUN:   %s -emit-llvm -o - | FileCheck %s --check-prefix=CURDIR
 
 void foo() {}
 
 // Since %s is an absolute path, directory should be the common
 // prefix, but the directory part should be part of the filename.
 
 // CHECK: = distinct !DISubprogram({{.*}}file: ![[SPFILE:[0-9]+]]
 // CHECK: ![[SPFILE]] = !DIFile(filename: "{{.*}}UNIQUEISH_SENTINEL
 // CHECK-SAME:                  debug-info-abspath.c"
-// CHECK-NOT:                   directory: "{{.*}}UNIQUEISH_SENTINEL
+// CHECK-NOT:                   directory: "{{.*}}UNIQUEISH_SENTINEL)
 
 // INTREE: = distinct !DISubprogram({{.*}}![[SPFILE:[0-9]+]]
 // INTREE: DIFile({{.*}}directory: "{{.+}}CodeGen{{.*}}")
 
 // CURDIR: = distinct !DICompileUnit({{.*}}file: ![[CUFILE:[0-9]+]]
 // CURDIR: ![[CUFILE]] = !DIFile({{.*}}directory: "{{.+}}UNIQUEISH_SENTINEL")
 

clang/test/CodeGen/debug-info-relpath.c

+// RUN: mkdir -p %t/UNIQUEISH_SENTINEL
+// RUN: cp %s %t/UNIQUEISH_SENTINEL/debug-info-abspath.c
+// RUN: cd %t
+
+// RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
+// RUN:   UNIQUEISH_SENTINEL/debug-info-abspath.c -emit-llvm -o - \
+// RUN:   | FileCheck %s
+
+// RUN: %clang_cc1 -debug-info-kind=limited -triple %itanium_abi_triple \
+// RUN:   ./UNIQUEISH_SENTINEL/../UNIQUEISH_SENTINEL/debug-info-abspath.c \
+// RUN:   -emit-llvm -o - | FileCheck %s
+
+void foo() {}
+
+// Since %s is a relative path, directory should be the common
+// prefix, but the directory part should be part of the filename.
+
+// CHECK: = distinct !DISubprogram({{.*}}file: ![[SPFILE:[0-9]+]]
+// CHECK: ![[SPFILE]] = !DIFile(filename: "UNIQUEISH_SENTINEL
+// CHECK-SAME:                  debug-info-abspath.c"
+// CHECK-NOT:                   directory: "{{.*}}UNIQUEISH_SENTINEL")

clang/test/CodeGenCXX/linetable-fnbegin.cpp

 // RUN: %clang_cc1 -emit-llvm -debug-info-kind=limited %s -o - | FileCheck %s
 // Test that the line table info for Foo<T>::bar() is pointing to the
 // right header file.
 // CHECK: define{{.*}}bar
 // CHECK-NOT: define
 // CHECK: ret {{.*}}, !dbg [[DBG:.*]]
-// CHECK: [[HPP:.*]] = !DIFile(filename: "./template.hpp",
+// CHECK: [[HPP:.*]] = !DIFile(filename: "template.hpp",
 // CHECK: [[SP:.*]] = distinct !DISubprogram(name: "bar",
 // CHECK-SAME:                               file: [[HPP]], line: 22
 // CHECK-SAME:                               DISPFlagDefinition
 // We shouldn't need a lexical block for this function.
 // CHECK: [[DBG]] = !DILocation(line: 23, column: 3, scope: [[SP]])
 
 
 # 1 "./template.h" 1

clang/test/Modules/debug-info-moduleimport.m

 // RUN: rm -rf %t
 // RUN: %clang_cc1 -debug-info-kind=limited -fmodules \
 // RUN:    -DGREETING="Hello World" -UNDEBUG \
 // RUN:    -fimplicit-module-maps -fmodules-cache-path=%t %s \
 // RUN:    -I %S/Inputs -isysroot /tmp/.. -I %t -emit-llvm \
 // RUN:    -debugger-tuning=lldb -o - | FileCheck %s
 
 // CHECK: ![[CU:.*]] = distinct !DICompileUnit
-// CHECK-SAME:  sysroot: "/tmp/..")
+// CHECK-SAME:  sysroot: "/")
 @import DebugObjC;
 // CHECK: !DIImportedEntity(tag: DW_TAG_imported_declaration, scope: ![[CU]],
 // CHECK-SAME:              entity: ![[MODULE:.*]], file: ![[F:[0-9]+]],
 // CHECK-SAME:              line: [[@LINE-3]])
 // CHECK: ![[MODULE]] = !DIModule(scope: null, name: "DebugObjC",
 // CHECK-SAME:  configMacros: "\22-DGREETING=Hello World\22 \22-UNDEBUG\22",
 // CHECK-SAME:  includePath: "{{.*}}test{{.*}}Modules{{.*}}Inputs"
 // CHECK: ![[F]] = !DIFile(filename: {{.*}}debug-info-moduleimport.m