1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
|
{ lib
, stdenv
, llvm_meta
, release_version
, version
, patches ? []
, src ? null
, monorepoSrc ? null
, runCommand
, cmake
, ninja
, python3
, xcbuild
, libllvm
, linuxHeaders
, libxcrypt
# Some platforms have switched to using compiler-rt, but still want a
# libgcc.a for ABI compat purposes. The use case would be old code that
# expects to link `-lgcc` but doesn't care exactly what its contents
# are, so long as it provides some builtins.
, doFakeLibgcc ? stdenv.hostPlatform.isFreeBSD
# In recent releases, the compiler-rt build seems to produce
# many `libclang_rt*` libraries, but not a single unified
# `libcompiler_rt` library, at least under certain configurations. Some
# platforms stil expect this, however, so we symlink one into place.
, forceLinkCompilerRt ? stdenv.hostPlatform.isOpenBSD
}:
let
useLLVM = stdenv.hostPlatform.useLLVM or false;
bareMetal = stdenv.hostPlatform.parsed.kernel.name == "none";
haveLibc = stdenv.cc.libc != null;
isDarwinStatic = stdenv.hostPlatform.isDarwin && stdenv.hostPlatform.isStatic && lib.versionAtLeast release_version "16";
inherit (stdenv.hostPlatform) isMusl isAarch64;
baseName = "compiler-rt";
pname = baseName + lib.optionalString (haveLibc) "-libc";
src' = if monorepoSrc != null then
runCommand "${baseName}-src-${version}" {} ''
mkdir -p "$out"
cp -r ${monorepoSrc}/cmake "$out"
cp -r ${monorepoSrc}/${baseName} "$out"
'' else src;
preConfigure = lib.optionalString (useLLVM && !haveLibc) ''
cmakeFlagsArray+=(-DCMAKE_C_FLAGS="-nodefaultlibs -ffreestanding")
'';
in
stdenv.mkDerivation ({
inherit pname version patches;
src = src';
sourceRoot = if lib.versionOlder release_version "13" then null
else "${src'.name}/${baseName}";
nativeBuildInputs = [ cmake ]
++ (lib.optional (lib.versionAtLeast release_version "15") ninja)
++ [ python3 libllvm.dev ]
++ lib.optional stdenv.isDarwin xcbuild.xcrun;
buildInputs =
lib.optional (stdenv.hostPlatform.isLinux && stdenv.hostPlatform.isRiscV) linuxHeaders;
env.NIX_CFLAGS_COMPILE = toString ([
"-DSCUDO_DEFAULT_OPTIONS=DeleteSizeMismatch=0:DeallocationTypeMismatch=0"
] ++ lib.optionals (!haveLibc) [
# The compiler got stricter about this, and there is a usellvm patch below
# which patches out the assert include causing an implicit definition of
# assert. It would be nicer to understand why compiler-rt thinks it should
# be able to #include <assert.h> in the first place; perhaps it's in the
# wrong, or perhaps there is a way to provide an assert.h.
"-Wno-error=implicit-function-declaration"
]);
cmakeFlags = [
"-DCOMPILER_RT_DEFAULT_TARGET_ONLY=ON"
"-DCMAKE_C_COMPILER_TARGET=${stdenv.hostPlatform.config}"
"-DCMAKE_ASM_COMPILER_TARGET=${stdenv.hostPlatform.config}"
] ++ lib.optionals (haveLibc && stdenv.hostPlatform.libc == "glibc") [
"-DSANITIZER_COMMON_CFLAGS=-I${libxcrypt}/include"
] ++ lib.optionals ((useLLVM || bareMetal || isMusl || isAarch64) && (lib.versions.major release_version == "13")) [
"-DCOMPILER_RT_BUILD_LIBFUZZER=OFF"
] ++ lib.optionals (useLLVM || bareMetal || isMusl || isDarwinStatic) [
"-DCOMPILER_RT_BUILD_SANITIZERS=OFF"
"-DCOMPILER_RT_BUILD_XRAY=OFF"
"-DCOMPILER_RT_BUILD_LIBFUZZER=OFF"
"-DCOMPILER_RT_BUILD_MEMPROF=OFF"
"-DCOMPILER_RT_BUILD_ORC=OFF" # may be possible to build with musl if necessary
] ++ lib.optionals (useLLVM || bareMetal) [
"-DCOMPILER_RT_BUILD_PROFILE=OFF"
] ++ lib.optionals ((useLLVM && !haveLibc) || bareMetal || isDarwinStatic) [
"-DCMAKE_CXX_COMPILER_WORKS=ON"
] ++ lib.optionals ((useLLVM && !haveLibc) || bareMetal) [
"-DCMAKE_C_COMPILER_WORKS=ON"
"-DCOMPILER_RT_BAREMETAL_BUILD=ON"
"-DCMAKE_SIZEOF_VOID_P=${toString (stdenv.hostPlatform.parsed.cpu.bits / 8)}"
] ++ lib.optionals (useLLVM && !haveLibc) [
"-DCMAKE_C_FLAGS=-nodefaultlibs"
] ++ lib.optionals (useLLVM) [
"-DCOMPILER_RT_BUILD_BUILTINS=ON"
#https://stackoverflow.com/questions/53633705/cmake-the-c-compiler-is-not-able-to-compile-a-simple-test-program
"-DCMAKE_TRY_COMPILE_TARGET_TYPE=STATIC_LIBRARY"
] ++ lib.optionals (bareMetal) [
"-DCOMPILER_RT_OS_DIR=baremetal"
] ++ lib.optionals (stdenv.hostPlatform.isDarwin) (lib.optionals (lib.versionAtLeast release_version "16") [
"-DCMAKE_LIPO=${lib.getBin stdenv.cc.bintools.bintools}/bin/${stdenv.cc.targetPrefix}lipo"
] ++ [
"-DDARWIN_macosx_OVERRIDE_SDK_VERSION=ON"
"-DDARWIN_osx_ARCHS=${stdenv.hostPlatform.darwinArch}"
"-DDARWIN_osx_BUILTIN_ARCHS=${stdenv.hostPlatform.darwinArch}"
] ++ lib.optionals (lib.versionAtLeast release_version "15") [
# `COMPILER_RT_DEFAULT_TARGET_ONLY` does not apply to Darwin:
# https://github.com/llvm/llvm-project/blob/27ef42bec80b6c010b7b3729ed0528619521a690/compiler-rt/cmake/base-config-ix.cmake#L153
"-DCOMPILER_RT_ENABLE_IOS=OFF"
]) ++ lib.optionals (lib.versionAtLeast version "19" && stdenv.isDarwin && lib.versionOlder stdenv.hostPlatform.darwinMinVersion "10.13") [
"-DSANITIZER_MIN_OSX_VERSION=10.10"
];
outputs = [ "out" "dev" ];
# TSAN requires XPC on Darwin, which we have no public/free source files for. We can depend on the Apple frameworks
# to get it, but they're unfree. Since LLVM is rather central to the stdenv, we patch out TSAN support so that Hydra
# can build this. If we didn't do it, basically the entire nixpkgs on Darwin would have an unfree dependency and we'd
# get no binary cache for the entire platform. If you really find yourself wanting the TSAN, make this controllable by
# a flag and turn the flag off during the stdenv build.
postPatch = lib.optionalString (!stdenv.isDarwin) ''
substituteInPlace cmake/builtin-config-ix.cmake \
--replace 'set(X86 i386)' 'set(X86 i386 i486 i586 i686)'
'' + lib.optionalString stdenv.isDarwin ''
substituteInPlace cmake/config-ix.cmake \
--replace 'set(COMPILER_RT_HAS_TSAN TRUE)' 'set(COMPILER_RT_HAS_TSAN FALSE)'
'' + lib.optionalString (useLLVM && !haveLibc) ((lib.optionalString (lib.versionAtLeast release_version "18") ''
substituteInPlace lib/builtins/aarch64/sme-libc-routines.c \
--replace "<stdlib.h>" "<stddef.h>"
'') + ''
substituteInPlace lib/builtins/int_util.c \
--replace "#include <stdlib.h>" ""
'' + (if stdenv.hostPlatform.isFreeBSD then
# As per above, but in FreeBSD assert is a macro and simply allowing it to be implicitly declared causes Issues!!!!!
''
substituteInPlace lib/builtins/clear_cache.c lib/builtins/cpu_model${lib.optionalString (lib.versionAtLeast version "18") "/x86"}.c \
--replace "#include <assert.h>" "#define assert(e) ((e)?(void)0:__assert(__FUNCTION__,__FILE__,__LINE__,#e))"
'' else ''
substituteInPlace lib/builtins/clear_cache.c \
--replace "#include <assert.h>" ""
substituteInPlace lib/builtins/cpu_model${lib.optionalString (lib.versionAtLeast version "18") "/x86"}.c \
--replace "#include <assert.h>" ""
''));
# Hack around weird upsream RPATH bug
postInstall = lib.optionalString (stdenv.hostPlatform.isDarwin) ''
ln -s "$out/lib"/*/* "$out/lib"
'' + lib.optionalString (useLLVM && stdenv.hostPlatform.isLinux) ''
ln -s $out/lib/*/clang_rt.crtbegin-*.o $out/lib/crtbegin.o
ln -s $out/lib/*/clang_rt.crtend-*.o $out/lib/crtend.o
# Note the history of crt{begin,end}S in previous versions of llvm in nixpkg:
# The presence of crtbegin_shared has been added and removed; it's possible
# people have added/removed it to get it working on their platforms.
# Try each in turn for now.
ln -s $out/lib/*/clang_rt.crtbegin-*.o $out/lib/crtbeginS.o
ln -s $out/lib/*/clang_rt.crtend-*.o $out/lib/crtendS.o
ln -s $out/lib/*/clang_rt.crtbegin_shared-*.o $out/lib/crtbeginS.o
ln -s $out/lib/*/clang_rt.crtend_shared-*.o $out/lib/crtendS.o
'' + lib.optionalString doFakeLibgcc ''
ln -s $out/lib/*/libclang_rt.builtins-*.a $out/lib/libgcc.a
'' + lib.optionalString forceLinkCompilerRt ''
ln -s $out/lib/*/libclang_rt.builtins-*.a $out/lib/libcompiler_rt.a
'';
meta = llvm_meta // {
homepage = "https://compiler-rt.llvm.org/";
description = "Compiler runtime libraries";
longDescription = ''
The compiler-rt project provides highly tuned implementations of the
low-level code generator support routines like "__fixunsdfdi" and other
calls generated when a target doesn't have a short sequence of native
instructions to implement a core IR operation. It also provides
implementations of run-time libraries for dynamic testing tools such as
AddressSanitizer, ThreadSanitizer, MemorySanitizer, and DataFlowSanitizer.
'';
# "All of the code in the compiler-rt project is dual licensed under the MIT
# license and the UIUC License (a BSD-like license)":
license = with lib.licenses; [ mit ncsa ];
# compiler-rt requires a Clang stdenv on 32-bit RISC-V:
# https://reviews.llvm.org/D43106#1019077
broken = stdenv.hostPlatform.isRiscV32 && !stdenv.cc.isClang;
};
} // (if lib.versionOlder release_version "16" then { inherit preConfigure; } else {}))
|