Reapply [2]: [VFS] Add support for handling path traversals

This was applied twice r261551 and 263617 and later reverted because:

(1) Windows bot failing on unittests. Change the current behavior to do
not handle path traversals on windows.

(2) Windows bot failed to include llvm/Config/config.h in order to use
HAVE_REALPATH. Use LLVM_ON_UNIX instead, as done in lib/Basic/FileManager.cpp.

Handle ".", ".." and "./" with trailing slashes while collecting files
to be dumped into the vfs overlay directory.

Include the support for symlinks into components. Given the path:

/install-dir/bin/../lib/clang/3.8.0/include/altivec.h, if "bin"
component is a symlink, it's not safe to use `path::remove_dots` here,
and `realpath` is used to get the right answer. Since `realpath`
is expensive, we only do it at collecting time (which only happens
during the crash reproducer) and cache the base directory for fast lookups.

Overall, this makes the input to the VFS YAML file to be canonicalized
to never contain traversal components.

Differential Revision: http://reviews.llvm.org/D17104

rdar://problem/24499339

llvm-svn: 263686

2 files changed, 132 insertions, 0 deletions

diff --git a/clang/test/Modules/crash-vfs-path-symlink-component.m b/clang/test/Modules/crash-vfs-path-symlink-component.m
new file mode 100644
index 00000000000..759bad03cd3
--- /dev/null
+++ b/clang/test/Modules/crash-vfs-path-symlink-component.m
@@ -0,0 +1,72 @@
+// REQUIRES: crash-recovery, shell
+
+// FIXME: This XFAIL is cargo-culted from crash-report.c. Do we need it?
+// XFAIL: mingw32
+
+// Test that clang is capable of collecting the right header files in the
+// crash reproducer if there's a symbolic link component in the path.
+
+// RUN: rm -rf %t
+// RUN: mkdir -p %t/i %t/m %t %t/sysroot
+// RUN: cp -a %S/Inputs/System/usr %t/i/
+// RUN: ln -s include/tcl-private %t/i/usr/x
+
+// RUN: not env FORCE_CLANG_DIAGNOSTICS_CRASH= TMPDIR=%t TEMP=%t TMP=%t \
+// RUN: %clang -fsyntax-only %s -I %/t/i -isysroot %/t/sysroot/ \
+// RUN:     -fmodules -fmodules-cache-path=%t/m/ 2>&1 | FileCheck %s
+
+// RUN: FileCheck --check-prefix=CHECKSRC %s -input-file %t/crash-vfs-*.m
+// RUN: FileCheck --check-prefix=CHECKSH %s -input-file %t/crash-vfs-*.sh
+// RUN: FileCheck --check-prefix=CHECKYAML %s -input-file \
+// RUN: %t/crash-vfs-*.cache/vfs/vfs.yaml
+// RUN: find %t/crash-vfs-*.cache/vfs | \
+// RUN:   grep "usr/include/stdio.h" | count 1
+
+#include "usr/x/../stdio.h"
+
+// CHECK: Preprocessed source(s) and associated run script(s) are located at:
+// CHECK-NEXT: note: diagnostic msg: {{.*}}.m
+// CHECK-NEXT: note: diagnostic msg: {{.*}}.cache
+
+// CHECKSRC: @import cstd.stdio;
+
+// CHECKSH: # Crash reproducer
+// CHECKSH-NEXT: # Driver args: "-fsyntax-only"
+// CHECKSH-NEXT: # Original command: {{.*$}}
+// CHECKSH-NEXT: "-cc1"
+// CHECKSH: "-isysroot" "{{[^"]*}}/sysroot/"
+// CHECKSH-NOT: "-fmodules-cache-path="
+// CHECKSH: "crash-vfs-{{[^ ]*}}.m"
+// CHECKSH: "-ivfsoverlay" "crash-vfs-{{[^ ]*}}.cache/vfs/vfs.yaml"
+
+// CHECKYAML: 'type': 'directory'
+// CHECKYAML: 'name': "{{[^ ]*}}/i/usr/include",
+// CHECKYAML-NEXT: 'contents': [
+// CHECKYAML-NEXT:   {
+// CHECKYAML-NEXT:     'type': 'file',
+// CHECKYAML-NEXT:     'name': "module.map",
+// CHECKYAML-NEXT:     'external-contents': "{{[^ ]*}}.cache/vfs/{{[^ ]*}}/i/usr/include/module.map"
+// CHECKYAML-NEXT:   },
+
+// CHECKYAML: 'type': 'directory'
+// CHECKYAML: 'name': "{{[^ ]*}}/i/usr",
+// CHECKYAML-NEXT: 'contents': [
+// CHECKYAML-NEXT:   {
+// CHECKYAML-NEXT:     'type': 'file',
+// CHECKYAML-NEXT:     'name': "module.map",
+// CHECKYAML-NEXT:     'external-contents': "{{[^ ]*}}.cache/vfs/{{[^ ]*}}/i/usr/include/module.map"
+// CHECKYAML-NEXT:   },
+
+// Test that by using the previous generated YAML file clang is able to find the
+// right files inside the overlay and map the virtual request for a path that
+// previously contained a symlink to work. To make sure of this, wipe out the
+// %/t/i directory containing the symlink component.
+
+// RUN: rm -rf %/t/i
+// RUN: unset FORCE_CLANG_DIAGNOSTICS_CRASH
+// RUN: %clang -E %s -I %/t/i -isysroot %/t/sysroot/ \
+// RUN:     -ivfsoverlay %t/crash-vfs-*.cache/vfs/vfs.yaml -fmodules \
+// RUN:     -fmodules-cache-path=%t/m/ 2>&1 \
+// RUN:     | FileCheck %s --check-prefix=CHECKOVERLAY
+
+// CHECKOVERLAY: @import cstd.stdio; /* clang -E: implicit import for "{{[^ ]*}}.cache/vfs/{{[^ ]*}}/i/usr/include/stdio.h"
diff --git a/clang/test/Modules/crash-vfs-path-traversal.m b/clang/test/Modules/crash-vfs-path-traversal.m
new file mode 100644
index 00000000000..b7610a591e4
--- /dev/null
+++ b/clang/test/Modules/crash-vfs-path-traversal.m
@@ -0,0 +1,60 @@
+// REQUIRES: crash-recovery, shell, non-ms-sdk, non-ps4-sdk
+
+// FIXME: Canonicalizing paths to remove relative traversal components
+// currenty fails a unittest on windows and is disable by default.
+// FIXME: This XFAIL is cargo-culted from crash-report.c. Do we need it?
+// XFAIL: mingw32
+
+// RUN: rm -rf %t
+// RUN: mkdir -p %t/i %t/m %t
+
+// RUN: not env FORCE_CLANG_DIAGNOSTICS_CRASH= TMPDIR=%t TEMP=%t TMP=%t \
+// RUN: %clang -fsyntax-only %s -I %S/Inputs/System -isysroot %/t/i/    \
+// RUN: -fmodules -fmodules-cache-path=%t/m/ 2>&1 | FileCheck %s
+
+// RUN: FileCheck --check-prefix=CHECKSRC %s -input-file %t/crash-vfs-*.m
+// RUN: FileCheck --check-prefix=CHECKSH %s -input-file %t/crash-vfs-*.sh
+// RUN: FileCheck --check-prefix=CHECKYAML %s -input-file \
+// RUN: %t/crash-vfs-*.cache/vfs/vfs.yaml
+// RUN: find %t/crash-vfs-*.cache/vfs | \
+// RUN:   grep "Inputs/System/usr/include/stdio.h" | count 1
+
+#include "usr/././//////include/../include/./././../include/stdio.h"
+
+// CHECK: Preprocessed source(s) and associated run script(s) are located at:
+// CHECK-NEXT: note: diagnostic msg: {{.*}}.m
+// CHECK-NEXT: note: diagnostic msg: {{.*}}.cache
+
+// CHECKSRC: @import cstd.stdio;
+
+// CHECKSH: # Crash reproducer
+// CHECKSH-NEXT: # Driver args: "-fsyntax-only"
+// CHECKSH-NEXT: # Original command: {{.*$}}
+// CHECKSH-NEXT: "-cc1"
+// CHECKSH: "-isysroot" "{{[^"]*}}/i/"
+// CHECKSH-NOT: "-fmodules-cache-path="
+// CHECKSH: "crash-vfs-{{[^ ]*}}.m"
+// CHECKSH: "-ivfsoverlay" "crash-vfs-{{[^ ]*}}.cache/vfs/vfs.yaml"
+
+// CHECKYAML: 'type': 'directory'
+// CHECKYAML: 'name': "{{[^ ]*}}/Inputs/System/usr/include",
+// CHECKYAML-NEXT: 'contents': [
+// CHECKYAML-NEXT:   {
+// CHECKYAML-NEXT:     'type': 'file',
+// CHECKYAML-NEXT:     'name': "module.map",
+// CHECKYAML-NEXT:     'external-contents': "{{[^ ]*}}/Inputs/System/usr/include/module.map"
+// CHECKYAML-NEXT:   },
+
+// Replace the paths in the YAML files with relative ".." traversals
+// and fed into clang to test whether we're correctly representing them
+// in the VFS overlay.
+
+// RUN: sed -e "s@usr/include@usr/include/../include@g" \
+// RUN:     %t/crash-vfs-*.cache/vfs/vfs.yaml > %t/vfs.yaml
+// RUN: unset FORCE_CLANG_DIAGNOSTICS_CRASH
+// RUN: %clang -E %s -I %S/Inputs/System -isysroot %/t/i/ \
+// RUN:     -ivfsoverlay %t/vfs.yaml -fmodules \
+// RUN:     -fmodules-cache-path=%t/m/ 2>&1 \
+// RUN:     | FileCheck %s --check-prefix=CHECKOVERLAY
+
+// CHECKOVERLAY: @import cstd.stdio; /* clang -E: implicit import for "{{[^ ]*}}.cache/vfs/{{[^ ]*}}/usr/include/stdio.h"