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
|
#!/usr/bin/env bash
#
# llvm-target generator
#
# Author: Moritz Angermann <moritz.angermann@gmail.com>
#
# This file generates the `llvm-targets` file, which contains the
# data-layout, cpu and attributes for the target. This is done by
# querying `clang` for the data-layout, cpu and attributes based
# on a given target.
#
# To utilize it run it as
#
# $ ./gen-data-layout.sh > llvm-targets
#
# Add missing targets to the list below to have them included in
# llvm-targets file.
# Target sets for which to generate the llvm-targets file
TARGETS=(
# Windows x86
"i386-unknown-windows" "i686-unknown-windows" "x86_64-unknown-windows"
# Linux ARM
"arm-unknown-linux-gnueabihf" "armv6-unknown-linux-gnueabihf" "armv6l-unknown-linux-gnueabihf"
"armv7-unknown-linux-gnueabihf" "armv7a-unknown-linux-gnueabi" "armv7l-unknown-linux-gnueabihf"
"aarch64-unknown-linux-gnu" "aarch64-unknown-linux"
# Linux x86
"i386-unknown-linux-gnu" "i386-unknown-linux" "x86_64-unknown-linux-gnu" "x86_64-unknown-linux"
# Linux Android
"x86_64-unknown-linux-android" "armv7-unknown-linux-androideabi" "aarch64-unknown-linux-android"
# Linux ppc64le
"powerpc64le-unknown-linux"
# FreeBSD amd64
"amd64-portbld-freebsd"
"x86_64-unknown-freebsd" # See #15718
# QNX
"arm-unknown-nto-qnx-eabi"
# macOS
"i386-apple-darwin" "x86_64-apple-darwin"
# iOS
"armv7-apple-ios arm64-apple-ios" "i386-apple-ios x86_64-apple-ios"
# FreeBSD ARM
"aarch64-unknown-freebsd" "armv6-unknown-freebsd-gnueabihf" "armv7-unknown-freebsd-gnueabihf"
)
# given the call to clang -c11 that clang --target -v generates,
# parse the -target-cpu <CPU> and -target-feature <feature> from
# the output.
function get_cpu_and_attr() {
# echo $@
while [ "$#" -gt 0 ]; do
case "$1" in
-target-cpu) CPU=$2; shift 2;;
-target-feature)
# translate clang to opt/llc target features
case "$2" in
# we don't have support in GHC for proper soft-float.
# if we extend the `llvm-target` file to contain two
# additional columns for opt and llc flags, we could
# pass -float-abi=soft; However ghc will use float
# registers unconditionally on arm, and as such true
# soft float with the registerised llvm backend is
# currently not possible.
+soft-float-abi) shift 2;;
*) ATTR+=("$2"); shift 2;;
esac
;;
*) shift 1;;
esac
done
}
# first marker to discriminate the first line being outputted.
FST=1
# a dummy file to use for the clang invocation.
FILE=_____dummy.c
touch $FILE
for target in "${TARGETS[@]}"; do
# find the cpu and attributes emitted by clang for the given $target
CPU=""
ATTR=()
args=$(clang --target=$target -S $FILE -o /dev/null -v 2>&1 |grep $FILE)
get_cpu_and_attr $args
# find the data-layout from the llvm code emitted by clang.
dl=$(clang --target=$target -S $FILE -emit-llvm -o -|grep datalayout |awk -F\ '{ print $4 }')
# GNU and Apple/LLVM can't agree on the aarch64 target.
# aarch64-apple-ios, is understood by autotools but not by LLVM.
# arm64-apple-ios, is understood by LLVM, but not by autotools.
#
# therefore, while we query clang with arm64-apple-ios, we put
# aarch64-apple-ios into the llvm-target list, as that is what
# we have to configure ghc with --target with anyway. Also we
# want to retain the GNU naming for compatibility with libraries
# that use autotools.
if [ "$target" == "arm64-apple-ios" ]; then
target="aarch64-apple-ios"
fi
if [ $FST -eq 1 ]; then
echo "[(\"${target}\", ($dl, \"$CPU\", \"${ATTR[*]}\"))"
FST=0
else
echo ",(\"${target}\", ($dl, \"$CPU\", \"${ATTR[*]}\"))"
fi
done
rm $FILE
echo "]"
|
