1#!/bin/bash -eu
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
28SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
29SRC_DIR=$(readlink -f $SCRIPT_DIR/..)
30TARGE_DIR=$(readlink -f $1)
31
32LLVM_SRC="${LLVM_SRC:-$SCRIPT_DIR/../../../../../..}"
33LLVM_SRC=$(readlink -f $LLVM_SRC)
34
35if [[ ! -d "${LLVM_SRC}/projects/libcxxabi" ||
36      ! -d "${LLVM_SRC}/projects/libcxx" ]]; then
37  echo "Missing or incomplete LLVM_SRC"
38  exit 1
39fi
40
41if [[ "$ZLIB_SRC" == ""  ||
42      ! -x "${ZLIB_SRC}/configure" ||
43      ! -f "${ZLIB_SRC}/zlib.h" ]]; then
44  echo "Missing or incomplete ZLIB_SRC"
45  exit 1
46fi
47ZLIB_SRC=$(readlink -f $ZLIB_SRC)
48
49J="${J:-50}"
50
51CLANG="${CLANG:-`which clang`}"
52CLANG_DIR=$(readlink -f $(dirname "$CLANG"))
53
54BUILD_DIR=$(readlink -f ./symbolizer)
55mkdir -p $BUILD_DIR
56cd $BUILD_DIR
57
58CC=$CLANG_DIR/clang
59CXX=$CLANG_DIR/clang++
60TBLGEN=$CLANG_DIR/llvm-tblgen
61OPT=$CLANG_DIR/opt
62export AR=$CLANG_DIR/llvm-ar
63export LINK=$CLANG_DIR/llvm-link
64
65for F in $CC $CXX $TBLGEN $LINK $OPT $AR; do
66  if [[ ! -x "$F" ]]; then
67    echo "Missing $F"
68     exit 1
69  fi
70done
71
72ZLIB_BUILD=${BUILD_DIR}/zlib
73LIBCXX_BUILD=${BUILD_DIR}/libcxx
74LLVM_BUILD=${BUILD_DIR}/llvm
75SYMBOLIZER_BUILD=${BUILD_DIR}/symbolizer
76
77FLAGS=${FLAGS:-}
78FLAGS="$FLAGS -fPIC -flto -Os -g0 -DNDEBUG"
79
80# Build zlib.
81mkdir -p ${ZLIB_BUILD}
82cd ${ZLIB_BUILD}
83cp -r ${ZLIB_SRC}/* .
84CC=$CC CFLAGS="$FLAGS" RANLIB=/bin/true ./configure --static
85make -j${J} libz.a
86
87# Build and install libcxxabi and libcxx.
88if [[ ! -d ${LIBCXX_BUILD} ]]; then
89  mkdir -p ${LIBCXX_BUILD}
90  cd ${LIBCXX_BUILD}
91  LIBCXX_FLAGS="${FLAGS} -Wno-macro-redefined -I${LLVM_SRC}/projects/libcxxabi/include"
92  cmake -GNinja \
93    -DCMAKE_BUILD_TYPE=Release \
94    -DCMAKE_C_COMPILER=$CC \
95    -DCMAKE_CXX_COMPILER=$CXX \
96    -DCMAKE_C_FLAGS_RELEASE="${LIBCXX_FLAGS}" \
97    -DCMAKE_CXX_FLAGS_RELEASE="${LIBCXX_FLAGS}" \
98    -DLIBCXXABI_ENABLE_ASSERTIONS=OFF \
99    -DLIBCXXABI_ENABLE_EXCEPTIONS=OFF \
100    -DLIBCXXABI_ENABLE_SHARED=OFF \
101    -DLIBCXX_ENABLE_ASSERTIONS=OFF \
102    -DLIBCXX_ENABLE_EXCEPTIONS=OFF \
103    -DLIBCXX_ENABLE_RTTI=OFF \
104    -DLIBCXX_ENABLE_SHARED=OFF \
105  $LLVM_SRC
106fi
107cd ${LIBCXX_BUILD}
108ninja cxx cxxabi
109
110FLAGS="${FLAGS} -fno-rtti -fno-exceptions"
111LLVM_FLAGS="${FLAGS} -nostdinc++ -I${ZLIB_BUILD} -I${LIBCXX_BUILD}/include/c++/v1"
112
113# Build LLVM.
114if [[ ! -d ${LLVM_BUILD} ]]; then
115  mkdir -p ${LLVM_BUILD}
116  cd ${LLVM_BUILD}
117  cmake -GNinja \
118    -DCMAKE_BUILD_TYPE=Release \
119    -DCMAKE_C_COMPILER=$CC \
120    -DCMAKE_CXX_COMPILER=$CXX \
121    -DCMAKE_C_FLAGS_RELEASE="${LLVM_FLAGS}" \
122    -DCMAKE_CXX_FLAGS_RELEASE="${LLVM_FLAGS}" \
123    -DLLVM_TABLEGEN=$TBLGEN \
124    -DLLVM_ENABLE_ZLIB=ON \
125    -DLLVM_ENABLE_TERMINFO=OFF \
126    -DLLVM_ENABLE_THREADS=OFF \
127  $LLVM_SRC
128fi
129cd ${LLVM_BUILD}
130ninja LLVMSymbolize LLVMObject LLVMBinaryFormat LLVMDebugInfoDWARF LLVMSupport LLVMDebugInfoPDB LLVMMC LLVMDemangle
131
132cd ${BUILD_DIR}
133rm -rf ${SYMBOLIZER_BUILD}
134mkdir ${SYMBOLIZER_BUILD}
135cd ${SYMBOLIZER_BUILD}
136
137echo "Compiling..."
138SYMBOLIZER_FLAGS="$LLVM_FLAGS -I${LLVM_SRC}/include -I${LLVM_BUILD}/include -std=c++11"
139$CXX $SYMBOLIZER_FLAGS ${SRC_DIR}/sanitizer_symbolize.cc ${SRC_DIR}/sanitizer_wrappers.cc -c
140$AR rc symbolizer.a sanitizer_symbolize.o sanitizer_wrappers.o
141
142SYMBOLIZER_API_LIST=__sanitizer_symbolize_code,__sanitizer_symbolize_data,__sanitizer_symbolize_flush,__sanitizer_symbolize_demangle
143
144# Merge all the object files together and copy the resulting library back.
145$SCRIPT_DIR/ar_to_bc.sh $LIBCXX_BUILD/lib/libc++.a \
146                        $LIBCXX_BUILD/lib/libc++abi.a \
147                        $LLVM_BUILD/lib/libLLVMSymbolize.a \
148                        $LLVM_BUILD/lib/libLLVMObject.a \
149                        $LLVM_BUILD/lib/libLLVMBinaryFormat.a \
150                        $LLVM_BUILD/lib/libLLVMDebugInfoDWARF.a \
151                        $LLVM_BUILD/lib/libLLVMSupport.a \
152                        $LLVM_BUILD/lib/libLLVMDebugInfoPDB.a \
153                        $LLVM_BUILD/lib/libLLVMDemangle.a \
154                        $LLVM_BUILD/lib/libLLVMMC.a \
155                        $ZLIB_BUILD/libz.a \
156                        symbolizer.a \
157                        all.bc
158
159echo "Optimizing..."
160$OPT -internalize -internalize-public-api-list=${SYMBOLIZER_API_LIST} all.bc -o opt.bc
161$CC $FLAGS -fno-lto -c opt.bc -o symbolizer.o
162
163echo "Checking undefined symbols..."
164nm -f posix -g symbolizer.o | cut -f 1,2 -d \  | LC_COLLATE=C sort -u > undefined.new
165(diff -u $SCRIPT_DIR/global_symbols.txt undefined.new | grep -E "^\+[^+]") && \
166  (echo "Failed: unexpected symbols"; exit 1)
167
168arch() {
169  objdump -f $1 | grep -m1 -Po "(?<=file format ).*$"
170}
171
172SYMBOLIZER_FORMAT=$(arch symbolizer.o)
173echo "Injecting $SYMBOLIZER_FORMAT symbolizer..."
174for A in $TARGE_DIR/libclang_rt.*san*.a; do
175  A_FORMAT=$(arch $A)
176  if [[ "$A_FORMAT" != "$SYMBOLIZER_FORMAT" ]] ; then
177    continue
178  fi
179  (nm -u $A 2>/dev/null | grep -E "__sanitizer_symbolize_code" >/dev/null) || continue
180  echo "$A"
181  $AR rcs $A symbolizer.o
182done
183
184echo "Success!"
185