1#!/usr/bin/env bash
2#
3# Run as: CLANG=bin/clang ZLIB_SRC=src/zlib \
4#             build_symbolizer.sh runtime_build/lib/clang/4.0.0/lib/linux/
5# zlib can be downloaded from http://www.zlib.net.
6#
7# Script compiles self-contained object file with symbolization code and injects
8# it into the given set of runtime libraries. Script updates only libraries
9# which has unresolved __sanitizer_symbolize_* symbols and matches architecture.
10# Object file is be compiled from LLVM sources with dependencies like libc++ and
11# zlib. Then it internalizes symbols in the file, so that it can be linked
12# into arbitrary programs, avoiding conflicts with the program own symbols and
13# avoiding dependencies on any program symbols. The only acceptable dependencies
14# are libc and __sanitizer::internal_* from sanitizer runtime.
15#
16# Symbols exported by the object file will be used by Sanitizer runtime
17# libraries to symbolize code/data in-process.
18#
19# The script will modify the output directory which is given as the first
20# argument to the script.
21#
22# FIXME: We should really be using a simpler approach to building this object
23# file, and it should be available as a regular cmake rule. Conceptually, we
24# want to be doing "ld -r" followed by "objcopy -G" to create a relocatable
25# object file with only our entry points exposed. However, this does not work at
26# present, see PR30750.
27
28set -x
29set -e
30set -u
31
32SCRIPT_DIR=$(cd "$(dirname "$0")" && pwd)
33SRC_DIR=$(readlink -f $SCRIPT_DIR/..)
34TARGE_DIR=$(readlink -f $1)
35COMPILER_RT_SRC=$(readlink -f ${SCRIPT_DIR}/../../../..)
36LLVM_SRC=${LLVM_SRC:-${COMPILER_RT_SRC}/../llvm}
37LLVM_SRC=$(readlink -f $LLVM_SRC)
38if [[ ! -d "${LLVM_SRC}/../llvm" ]] ; then
39  LLVM_SRC=$(readlink -f ${COMPILER_RT_SRC}/../../../llvm)
40fi
41LIBCXX_SRC=$(readlink -f ${COMPILER_RT_SRC}/../libcxx)
42LIBCXXABI_SRC=$(readlink -f ${COMPILER_RT_SRC}/../libcxxabi)
43
44if [[ ! -d "${LLVM_SRC}/../llvm" ||
45      ! -d "${LIBCXX_SRC}" ||
46      ! -d "${LIBCXXABI_SRC}" ]]; then
47  echo "Missing or incomplete LLVM_SRC"
48  exit 1
49fi
50
51if [[ "$ZLIB_SRC" == ""  ||
52      ! -x "${ZLIB_SRC}/configure" ||
53      ! -f "${ZLIB_SRC}/zlib.h" ]]; then
54  echo "Missing or incomplete ZLIB_SRC"
55  exit 1
56fi
57ZLIB_SRC=$(readlink -f $ZLIB_SRC)
58
59J="${J:-50}"
60
61CLANG="${CLANG:-`which clang`}"
62CLANG_DIR=$(readlink -f $(dirname "$CLANG"))
63
64BUILD_DIR=$(readlink -f ./symbolizer)
65mkdir -p $BUILD_DIR
66cd $BUILD_DIR
67
68CC=$CLANG_DIR/clang
69CXX=$CLANG_DIR/clang++
70TBLGEN=$CLANG_DIR/llvm-tblgen
71OPT=$CLANG_DIR/opt
72AR=$CLANG_DIR/llvm-ar
73LINK=$CLANG_DIR/llvm-link
74TARGET_TRIPLE=$($CC -print-target-triple)
75
76for F in $CC $CXX $TBLGEN $LINK $OPT $AR; do
77  if [[ ! -x "$F" ]]; then
78    echo "Missing $F"
79     exit 1
80  fi
81done
82
83ZLIB_BUILD=${BUILD_DIR}/zlib
84LIBCXX_BUILD=${BUILD_DIR}/libcxx
85LLVM_BUILD=${BUILD_DIR}/llvm
86SYMBOLIZER_BUILD=${BUILD_DIR}/symbolizer
87
88FLAGS=${FLAGS:-}
89FLAGS="$FLAGS -fPIC -flto -Os -g0 -DNDEBUG"
90
91# Build zlib.
92mkdir -p ${ZLIB_BUILD}
93cd ${ZLIB_BUILD}
94cp -r ${ZLIB_SRC}/* .
95CC=$CC CFLAGS="$FLAGS" RANLIB=/bin/true ./configure --static
96make -j${J} libz.a
97
98# Build and install libcxxabi and libcxx.
99if [[ ! -d ${LIBCXX_BUILD} ]]; then
100  mkdir -p ${LIBCXX_BUILD}
101  cd ${LIBCXX_BUILD}
102  LIBCXX_FLAGS="${FLAGS} -Wno-macro-redefined"
103  PROJECTS=
104  if [[ ! -d $LLVM_SRC/projects/libcxxabi ]] ; then
105    PROJECTS="-DLLVM_ENABLE_PROJECTS='libcxx;libcxxabi'"
106  fi
107  cmake -GNinja \
108    ${PROJECTS} \
109    -DCMAKE_BUILD_TYPE=Release \
110    -DCMAKE_C_COMPILER=$CC \
111    -DCMAKE_CXX_COMPILER=$CXX \
112    -DCMAKE_C_FLAGS_RELEASE="${LIBCXX_FLAGS}" \
113    -DCMAKE_CXX_FLAGS_RELEASE="${LIBCXX_FLAGS}" \
114    -DLIBCXXABI_ENABLE_ASSERTIONS=OFF \
115    -DLIBCXXABI_ENABLE_EXCEPTIONS=OFF \
116    -DLIBCXXABI_ENABLE_SHARED=OFF \
117    -DLIBCXX_ENABLE_ASSERTIONS=OFF \
118    -DLIBCXX_ENABLE_EXCEPTIONS=OFF \
119    -DLIBCXX_ENABLE_RTTI=OFF \
120    -DLIBCXX_ENABLE_SHARED=OFF \
121    -DLLVM_DEFAULT_TARGET_TRIPLE="${TARGET_TRIPLE}" \
122  $LLVM_SRC
123fi
124cd ${LIBCXX_BUILD}
125ninja cxx cxxabi
126
127FLAGS="${FLAGS} -fno-rtti -fno-exceptions"
128LLVM_FLAGS="${FLAGS} -nostdinc++ -I${ZLIB_BUILD} -isystem ${LIBCXX_BUILD}/include/${TARGET_TRIPLE}/c++/v1 -isystem ${LIBCXX_BUILD}/include/c++/v1 -Wno-error=global-constructors"
129
130# Build LLVM.
131if [[ ! -d ${LLVM_BUILD} ]]; then
132  mkdir -p ${LLVM_BUILD}
133  cd ${LLVM_BUILD}
134  cmake -GNinja \
135    -DCMAKE_BUILD_TYPE=Release \
136    -DCMAKE_C_COMPILER=$CC \
137    -DCMAKE_CXX_COMPILER=$CXX \
138    -DCMAKE_C_FLAGS_RELEASE="${LLVM_FLAGS}" \
139    -DCMAKE_CXX_FLAGS_RELEASE="${LLVM_FLAGS}" \
140    -DLLVM_TABLEGEN=$TBLGEN \
141    -DLLVM_DEFAULT_TARGET_TRIPLE="${TARGET_TRIPLE}" \
142    -DLLVM_ENABLE_ZLIB=ON \
143    -DLLVM_ENABLE_TERMINFO=OFF \
144    -DLLVM_ENABLE_THREADS=OFF \
145  $LLVM_SRC
146fi
147cd ${LLVM_BUILD}
148ninja LLVMSymbolize LLVMObject LLVMBinaryFormat LLVMDebugInfoDWARF LLVMSupport LLVMDebugInfoPDB LLVMDebuginfod LLVMMC LLVMDemangle LLVMTextAPI
149
150cd ${BUILD_DIR}
151rm -rf ${SYMBOLIZER_BUILD}
152mkdir ${SYMBOLIZER_BUILD}
153cd ${SYMBOLIZER_BUILD}
154
155echo "Compiling..."
156SYMBOLIZER_FLAGS="$LLVM_FLAGS -I${LLVM_SRC}/include -I${LLVM_BUILD}/include -std=c++14"
157$CXX $SYMBOLIZER_FLAGS ${SRC_DIR}/sanitizer_symbolize.cpp ${SRC_DIR}/sanitizer_wrappers.cpp -c
158$AR rc symbolizer.a sanitizer_symbolize.o sanitizer_wrappers.o
159
160SYMBOLIZER_API_LIST=__sanitizer_symbolize_code
161SYMBOLIZER_API_LIST+=,__sanitizer_symbolize_data
162SYMBOLIZER_API_LIST+=,__sanitizer_symbolize_flush
163SYMBOLIZER_API_LIST+=,__sanitizer_symbolize_demangle
164SYMBOLIZER_API_LIST+=,__sanitizer_symbolize_set_demangle
165SYMBOLIZER_API_LIST+=,__sanitizer_symbolize_set_inline_frames
166
167LIBCXX_ARCHIVE_DIR=$(dirname $(find $LIBCXX_BUILD -name libc++.a | head -n1))
168
169# Merge all the object files together and copy the resulting library back.
170$LINK $LIBCXX_ARCHIVE_DIR/libc++.a \
171      $LIBCXX_ARCHIVE_DIR/libc++abi.a \
172      $LLVM_BUILD/lib/libLLVMSymbolize.a \
173      $LLVM_BUILD/lib/libLLVMObject.a \
174      $LLVM_BUILD/lib/libLLVMBinaryFormat.a \
175      $LLVM_BUILD/lib/libLLVMDebugInfoDWARF.a \
176      $LLVM_BUILD/lib/libLLVMSupport.a \
177      $LLVM_BUILD/lib/libLLVMDebugInfoPDB.a \
178      $LLVM_BUILD/lib/libLLVMDebugInfoMSF.a \
179      $LLVM_BUILD/lib/libLLVMDebugInfoCodeView.a \
180      $LLVM_BUILD/lib/libLLVMDebuginfod.a \
181      $LLVM_BUILD/lib/libLLVMDemangle.a \
182      $LLVM_BUILD/lib/libLLVMMC.a \
183      $LLVM_BUILD/lib/libLLVMTextAPI.a \
184      $ZLIB_BUILD/libz.a \
185      symbolizer.a \
186      -ignore-non-bitcode -o all.bc
187
188echo "Optimizing..."
189$OPT -internalize -internalize-public-api-list=${SYMBOLIZER_API_LIST} all.bc -o opt.bc
190$CC $FLAGS -fno-lto -c opt.bc -o symbolizer.o
191
192echo "Checking undefined symbols..."
193nm -f posix -g symbolizer.o | cut -f 1,2 -d \  | LC_COLLATE=C sort -u > undefined.new
194(diff -u $SCRIPT_DIR/global_symbols.txt undefined.new | grep -E "^\+[^+]") && \
195  (echo "Failed: unexpected symbols"; exit 1)
196
197arch() {
198  objdump -f $1 | grep -m1 -Po "(?<=file format ).*$"
199}
200
201SYMBOLIZER_FORMAT=$(arch symbolizer.o)
202echo "Injecting $SYMBOLIZER_FORMAT symbolizer..."
203for A in $TARGE_DIR/libclang_rt.*san*.a; do
204  A_FORMAT=$(arch $A)
205  if [[ "$A_FORMAT" != "$SYMBOLIZER_FORMAT" ]] ; then
206    continue
207  fi
208  (nm -u $A 2>/dev/null | grep -E "__sanitizer_symbolize_code" >/dev/null) || continue
209  echo "$A"
210  $AR rcs $A symbolizer.o
211done
212
213echo "Success!"
214