New attribute extra_compiler_files (#245)

lalten · siddharthab · web-flow · commit dd351642dcda · 2024-03-13T01:06:59.000-07:00
This helps inject files into the compiler action sandbox. The files can then be used in the compiler command line, e.g. the sanitizer ignore list.

---------

Co-authored-by: Siddhartha Bagaria &lt;starsid@gmail.com&gt;
diff --git a/toolchain/cc_wrapper.sh.tpl b/toolchain/cc_wrapper.sh.tpl
@@ -27,24 +27,55 @@
 
 # shellcheck disable=SC1083
 
-set -eu
+set -euo pipefail
 
 # See note in toolchain/internal/configure.bzl where we define
 # `wrapper_bin_prefix` for why this wrapper is needed.
 
-# Call the C++ compiler.
 if [[ -f %{toolchain_path_prefix}bin/clang ]]; then
-  exec %{toolchain_path_prefix}bin/clang "$@"
+  execroot_path=""
 elif [[ ${BASH_SOURCE[0]} == "/"* ]]; then
   # Some consumers of `CcToolchainConfigInfo` (e.g. `cmake` from rules_foreign_cc)
   # change CWD and call $CC (this script) with its absolute path.
-  # the execroot (i.e. `cmake` from `rules_foreign_cc`) and call CC . For cases like this,
-  # we'll try to find `clang` relative to this script.
-  # This script is at _execroot_/external/_repo_name_/bin/clang_wrapper.sh
-  execroot_path="${BASH_SOURCE[0]%/*/*/*/*}"
-  clang="${execroot_path}/%{toolchain_path_prefix}bin/clang"
-  exec "${clang}" "${@}"
+  # For cases like this, we'll try to find `clang` through an absolute path.
+  # This script is at _execroot_/external/_repo_name_/bin/cc_wrapper.sh
+  execroot_path="${BASH_SOURCE[0]%/*/*/*/*}/"
 else
   echo >&2 "ERROR: could not find clang; PWD=\"${PWD}\"; PATH=\"${PATH}\"."
   exit 5
 fi
+
+function sanitize_option() {
+  local -r opt=$1
+  if [[ ${opt} == */cc_wrapper.sh ]]; then
+    printf "%s" "${execroot_path}%{toolchain_path_prefix}bin/clang"
+  elif [[ ${opt} =~ ^-fsanitize-(ignore|black)list=[^/] ]]; then
+    # shellcheck disable=SC2206
+    parts=(${opt/=/ }) # Split flag name and value into array.
+    printf "%s" "${parts[0]}=${execroot_path}${parts[1]}"
+  else
+    printf "%s" "${opt}"
+  fi
+}
+
+cmd=()
+for ((i = 0; i <= $#; i++)); do
+  if [[ ${!i} == @* ]]; then
+    while IFS= read -r opt; do
+      opt="$(
+        set -e
+        sanitize_option "${opt}"
+      )"
+      cmd+=("${opt}")
+    done <"${!i:1}"
+  else
+    opt="$(
+      set -e
+      sanitize_option "${!i}"
+    )"
+    cmd+=("${opt}")
+  fi
+done
+
+# Call the C++ compiler.
+exec "${cmd[@]}"
diff --git a/toolchain/internal/configure.bzl b/toolchain/internal/configure.bzl
@@ -159,6 +159,7 @@ def llvm_config_impl(rctx):
         coverage_link_flags_dict = rctx.attr.coverage_link_flags,
         unfiltered_compile_flags_dict = rctx.attr.unfiltered_compile_flags,
         llvm_version = llvm_version,
+        extra_compiler_files = rctx.attr.extra_compiler_files,
     )
     host_dl_ext = "dylib" if os == "darwin" else "so"
     host_tools_info = dict([
@@ -412,7 +413,10 @@ filegroup(
         template = template + """
 filegroup(
     name = "compiler-components-{suffix}",
-    srcs = [":sysroot-components-{suffix}"],
+    srcs = [
+        ":sysroot-components-{suffix}",
+        {extra_compiler_files}
+    ],
 )
 
 filegroup(
@@ -445,6 +449,7 @@ filegroup(
         "{llvm_dist_label_prefix}clang",
         "{llvm_dist_label_prefix}include",
         ":sysroot-components-{suffix}",
+        {extra_compiler_files}
     ],
 )
 
@@ -541,6 +546,7 @@ cc_toolchain(
         extra_files_str = extra_files_str,
         host_tools_info = host_tools_info,
         cxx_builtin_include_directories = _list_to_string(cxx_builtin_include_directories),
+        extra_compiler_files = ("\"%s\"," % str(toolchain_info.extra_compiler_files)) if toolchain_info.extra_compiler_files else "",
     )
 
 def _convenience_targets_str(rctx, use_absolute_paths, llvm_dist_rel_path, llvm_dist_label_prefix, host_dl_ext):
diff --git a/toolchain/internal/repo.bzl b/toolchain/internal/repo.bzl
@@ -233,6 +233,12 @@ _compiler_configuration_attrs = {
         mandatory = False,
         doc = ("Override the toolchain's `target_settings` attribute."),
     ),
+    "extra_compiler_files": attr.label(
+        mandatory = False,
+        doc = ("Files to be made available in the sandbox for compile actions. " +
+               "Mostly useful for providing files containing lists of flags, e.g. " +
+               "sanitizer ignorelists."),
+    ),
 }
 
 llvm_config_attrs = dict(common_attrs)
diff --git a/toolchain/osx_cc_wrapper.sh.tpl b/toolchain/osx_cc_wrapper.sh.tpl
@@ -25,9 +25,9 @@
 # See https://blogs.oracle.com/dipol/entry/dynamic_libraries_rpath_and_mac
 # on how to set those paths for Mach-O binaries.
 
-# shellcheck disable=all
+# shellcheck disable=SC1083
 
-set -eu
+set -euo pipefail
 
 INSTALL_NAME_TOOL="/usr/bin/install_name_tool"
 
@@ -39,29 +39,65 @@ OUTPUT=
 function parse_option() {
   local -r opt="$1"
   if [[ ${OUTPUT} == "1" ]]; then
-    OUTPUT=$opt
-  elif [[ $opt =~ ^-l(.*)$ ]]; then
-    LIBS="${BASH_REMATCH[1]} $LIBS"
-  elif [[ $opt =~ ^-L(.*)$ ]]; then
-    LIB_DIRS="${BASH_REMATCH[1]} $LIB_DIRS"
-  elif [[ $opt =~ ^\@loader_path/(.*)$ ]]; then
+    OUTPUT=${opt}
+  elif [[ ${opt} =~ ^-l(.*)$ ]]; then
+    LIBS="${BASH_REMATCH[1]} ${LIBS}"
+  elif [[ ${opt} =~ ^-L(.*)$ ]]; then
+    LIB_DIRS="${BASH_REMATCH[1]} ${LIB_DIRS}"
+  elif [[ ${opt} =~ ^\@loader_path/(.*)$ ]]; then
     RPATHS="${BASH_REMATCH[1]} ${RPATHS}"
-  elif [[ $opt =~ ^-Wl,-rpath,\@loader_path/(.*)$ ]]; then
+  elif [[ ${opt} =~ ^-Wl,-rpath,\@loader_path/(.*)$ ]]; then
     RPATHS="${BASH_REMATCH[1]} ${RPATHS}"
-  elif [[ $opt == "-o" ]]; then
+  elif [[ ${opt} == "-o" ]]; then
     # output is coming
     OUTPUT=1
   fi
 }
 
-# let parse the option list
-for i in "$@"; do
-  if [[ $i == @* ]]; then
+if [[ -f %{toolchain_path_prefix}bin/clang ]]; then
+  execroot_path=""
+elif [[ ${BASH_SOURCE[0]} == "/"* ]]; then
+  # Some consumers of `CcToolchainConfigInfo` (e.g. `cmake` from rules_foreign_cc)
+  # change CWD and call $CC (this script) with its absolute path.
+  # For cases like this, we'll try to find `clang` through an absolute path.
+  # This script is at _execroot_/external/_repo_name_/bin/cc_wrapper.sh
+  execroot_path="${BASH_SOURCE[0]%/*/*/*/*}/"
+else
+  echo >&2 "ERROR: could not find clang; PWD=\"${PWD}\"; PATH=\"${PATH}\"."
+  exit 5
+fi
+
+function sanitize_option() {
+  local -r opt=$1
+  if [[ ${opt} == */cc_wrapper.sh ]]; then
+    printf "%s" "${execroot_path}%{toolchain_path_prefix}bin/clang"
+  elif [[ ${opt} =~ ^-fsanitize-(ignore|black)list=[^/] ]]; then
+    # shellcheck disable=SC2206
+    parts=(${opt/=/ }) # Split flag name and value into array.
+    printf "%s" "${parts[0]}=${execroot_path}${parts[1]}"
+  else
+    printf "%s" "${opt}"
+  fi
+}
+
+cmd=()
+for ((i = 0; i <= $#; i++)); do
+  if [[ ${!i} == @* ]]; then
     while IFS= read -r opt; do
-      parse_option "$opt"
-    done <"${i:1}" || exit 1
+      opt="$(
+        set -e
+        sanitize_option "${opt}"
+      )"
+      parse_option "${opt}"
+      cmd+=("${opt}")
+    done <"${!i:1}"
   else
-    parse_option "$i"
+    opt="$(
+      set -e
+      sanitize_option "${!i}"
+    )"
+    parse_option "${opt}"
+    cmd+=("${opt}")
   fi
 done
 
@@ -86,27 +122,13 @@ if [[ ":${PATH}:" != *":/usr/bin:"* ]]; then
 fi
 
 # Call the C++ compiler.
-if [[ -f %{toolchain_path_prefix}bin/clang ]]; then
-  %{toolchain_path_prefix}bin/clang "$@"
-elif [[ ${BASH_SOURCE[0]} == "/"* ]]; then
-  # Some consumers of `CcToolchainConfigInfo` (e.g. `cmake` from rules_foreign_cc)
-  # change CWD and call $CC (this script) with its absolute path.
-  # the execroot (i.e. `cmake` from `rules_foreign_cc`) and call CC . For cases like this,
-  # we'll try to find `clang` relative to this script.
-  # This script is at _execroot_/external/_repo_name_/bin/cc_wrapper.sh
-  execroot_path="${BASH_SOURCE[0]%/*/*/*/*}"
-  clang="${execroot_path}/%{toolchain_path_prefix}bin/clang"
-  "${clang}" "${@}"
-else
-  echo >&2 "ERROR: could not find clang; PWD=\"$(pwd)\"; PATH=\"${PATH}\"."
-  exit 5
-fi
+"${cmd[@]}"
 
 function get_library_path() {
   for libdir in ${LIB_DIRS}; do
-    if [ -f ${libdir}/lib$1.so ]; then
+    if [[ -f "${libdir}/lib$1".so ]]; then
       echo "${libdir}/lib$1.so"
-    elif [ -f ${libdir}/lib$1.dylib ]; then
+    elif [[ -f "${libdir}"/lib"$1".dylib ]]; then
       echo "${libdir}/lib$1.dylib"
     fi
   done
@@ -116,8 +138,9 @@ function get_library_path() {
 # and multi-level symlinks.
 function get_realpath() {
   local previous="$1"
-  local next=$(readlink "${previous}")
-  while [ -n "${next}" ]; do
+  local next
+  next="$(readlink "${previous}")"
+  while [[ -n ${next} ]]; do
     previous="${next}"
     next=$(readlink "${previous}")
   done
@@ -127,24 +150,25 @@ function get_realpath() {
 # Get the path of a lib inside a tool
 function get_otool_path() {
   # the lib path is the path of the original lib relative to the workspace
-  get_realpath $1 | sed 's|^.*/bazel-out/|bazel-out/|'
+  get_realpath "$1" | sed 's|^.*/bazel-out/|bazel-out/|'
 }
 
 # Do replacements in the output
 for rpath in ${RPATHS}; do
   for lib in ${LIBS}; do
     unset libname
-    if [ -f "$(dirname ${OUTPUT})/${rpath}/lib${lib}.so" ]; then
+    if [[ -f "$(dirname "${OUTPUT}")/${rpath}/lib${lib}.so" ]]; then
       libname="lib${lib}.so"
-    elif [ -f "$(dirname ${OUTPUT})/${rpath}/lib${lib}.dylib" ]; then
+    elif [[ -f "$(dirname "${OUTPUT}")/${rpath}/lib${lib}.dylib" ]]; then
       libname="lib${lib}.dylib"
     fi
     # ${libname-} --> return $libname if defined, or undefined otherwise. This is to make
     # this set -e friendly
     if [[ -n ${libname-} ]]; then
-      libpath=$(get_library_path ${lib})
-      if [ -n "${libpath}" ]; then
-        ${INSTALL_NAME_TOOL} -change $(get_otool_path "${libpath}") \
+      libpath="$(get_library_path "${lib}")"
+      if [[ -n ${libpath} ]]; then
+        otool_path="$(get_otool_path "${libpath}")"
+        "${INSTALL_NAME_TOOL}" -change "${otool_path}" \
           "@loader_path/${rpath}/${libname}" "${OUTPUT}"
       fi
     fi
