Split the building of the toolchain into individual shellscripts

This simplifies building the toolchain outside of docker.

build-compiler-rt.sh

@@ -0,0 +1,56 @@
+#!/bin/sh
+
+set -e
+
+if [ $# -lt 2 ]; then
+    echo $0 dest clang-version
+    exit 1
+fi
+PREFIX="$1"
+CLANG_VERSION="$2"
+export PATH=$PREFIX/bin:$PATH
+
+: ${CORES:=4}
+: ${ARCHS:=${TOOLCHAIN_ARCHS-i686 x86_64 armv7 aarch64}}
+
+if [ ! -d compiler-rt ]; then
+    git clone -b master https://github.com/llvm-mirror/compiler-rt.git
+    cd compiler-rt
+    git checkout 1d871d6cd3fed01cd50dd63e743bd2ea6e65eab6
+    cd ..
+fi
+
+# Add a symlink for i386 -> i686; we normally name the toolchain
+# i686-w64-mingw32, but due to the compiler-rt cmake peculiarities, we
+# need to refer to it as i386 at this stage.
+ln -sf i686-w64-mingw32 $PREFIX/i386-w64-mingw32
+
+cd compiler-rt
+for arch in $ARCHS; do
+    buildarchname=$arch
+    libarchname=$arch
+    case $arch in
+    armv7)
+        libarchname=arm
+        ;;
+    i686)
+        buildarchname=i386
+        libarchname=i386
+        ;;
+    esac
+    mkdir -p build-$arch
+    cd build-$arch
+    cmake \
+        -DCMAKE_C_COMPILER=$arch-w64-mingw32-clang \
+        -DCMAKE_SYSTEM_NAME=Windows \
+        -DCMAKE_AR=$PREFIX/bin/llvm-ar \
+        -DCMAKE_RANLIB=$PREFIX/bin/llvm-ranlib \
+        -DCMAKE_C_COMPILER_WORKS=1 \
+        -DCMAKE_C_COMPILER_TARGET=$buildarchname-windows-gnu \
+        -DCOMPILER_RT_DEFAULT_TARGET_ONLY=TRUE \
+        ../lib/builtins
+    make -j$CORES
+    mkdir -p $PREFIX/lib/clang/$CLANG_VERSION/lib/windows
+    cp lib/windows/libclang_rt.builtins-$buildarchname.a $PREFIX/lib/clang/$CLANG_VERSION/lib/windows/libclang_rt.builtins-$libarchname.a
+    cd ..
+done