It's a floating world!

sorry. japanese only.

CodeSourceryのツールチェインで，Floatオプション毎による生成されるオブジェクトの
アーキテクチャフラグとリンクされるライブラリ(とそのアーキフラグ)について

Floatオプション -mfloat-abi=soft の場合: (HardFaultにならない)
$ readelf -A /tmp/work/template/obj/main.o
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "Cortex-M4"
  Tag_CPU_arch: v7E-M
  Tag_CPU_arch_profile: Microcontroller
  Tag_THUMB_ISA_use: Thumb-2
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align_needed: 8-byte
  Tag_ABI_align_preserved: 8-byte, except leaf SP
  Tag_ABI_enum_size: small
  Tag_ABI_optimization_goals: Aggressive Size
  Tag_CPU_unaligned_access: v6
リンクされるライブラリ:
->c:/program files/codesourcery/bin/../lib/gcc/arm-none-eabi/4.6.1/../../../../arm-none-eabi/lib/thumb2\libc.a(lib_a-memset.o)

Floatオプション -mfloat-abi=softfp -mfpu=fpv4-sp-d16 の場合: (HardFaultにならない)
$ readelf -A /tmp/work/template/obj/main.o
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "Cortex-M4"
  Tag_CPU_arch: v7E-M
  Tag_CPU_arch_profile: Microcontroller
  Tag_THUMB_ISA_use: Thumb-2
  Tag_FP_arch: VFPv4-D16
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align_needed: 8-byte
  Tag_ABI_align_preserved: 8-byte, except leaf SP
  Tag_ABI_enum_size: small
  Tag_ABI_HardFP_use: SP and DP
  Tag_ABI_optimization_goals: Aggressive Size
  Tag_CPU_unaligned_access: v6
  Tag_FP_HP_extension: Allowed
リンクされるライブラリ:
->c:/program files/codesourcery/bin/../lib/gcc/arm-none-eabi/4.6.1/../../../../arm-none-eabi/lib/thumb2\libc.a(lib_a-memset.o)

Floatオプション -mfloat-abi=hard -mfpu=fpv4-sp-d16 の場合: (HardFaultになる)
$ readelf -A /tmp/work/template/obj/main.o
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "Cortex-M4"
  Tag_CPU_arch: v7E-M
  Tag_CPU_arch_profile: Microcontroller
  Tag_THUMB_ISA_use: Thumb-2
  Tag_FP_arch: VFPv4-D16
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align_needed: 8-byte
  Tag_ABI_align_preserved: 8-byte, except leaf SP
  Tag_ABI_enum_size: small
  Tag_ABI_HardFP_use: SP and DP
  Tag_ABI_VFP_args: VFP registers
  Tag_ABI_optimization_goals: Aggressive Size
  Tag_CPU_unaligned_access: v6
  Tag_FP_HP_extension: Allowed
リンクされるライブラリ:
->c:/program files/codesourcery/bin/../lib/gcc/arm-none-eabi/4.6.1/../../../../arm-none-eabi/lib/thumb2\libc.a(lib_a-memset.o)


↑でリンクされるCodeSourceryのライブラリ:
File: /cygdrive/c/Program Files/CodeSourcery/arm-none-eabi/lib/thumb2/libc.a(lib_a-memset.o)
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "7"
  Tag_CPU_arch: v7
  Tag_THUMB_ISA_use: Thumb-2
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align_needed: 8-byte
  Tag_ABI_align_preserved: 8-byte, except leaf SP
  Tag_ABI_enum_size: small
  Tag_ABI_optimization_goals: Aggressive Speed
  Tag_CPU_unaligned_access: v6
  Tag_DIV_use: Not allowed



[オマケ]
SAT(Summon Arm Toolchain)のmultilibパッチを適用してビルドしたツールチェインで
(binutils-2.22,GCC-4.6.2,newlib-1.20.0)生成した場合のアーキテクチャフラグ
Floatオプション -mfloat-abi=hard -mfpu=fpv4-sp-d16 の場合: (HardFaultにならない)
$ readelf -A /tmp/work/template/obj/main.o
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "Cortex-M4"
  Tag_CPU_arch: v7E-M
  Tag_CPU_arch_profile: Microcontroller
  Tag_THUMB_ISA_use: Thumb-2
  Tag_FP_arch: VFPv4-D16
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align_needed: 8-byte
  Tag_ABI_align_preserved: 8-byte, except leaf SP
  Tag_ABI_enum_size: small
  Tag_ABI_HardFP_use: SP and DP
  Tag_ABI_VFP_args: VFP registers
  Tag_ABI_optimization_goals: Aggressive Size
  Tag_FP_HP_extension: Allowed
リンクされるライブラリ:
->/usr/local/lib/gcc/arm-hardfloat-eabi/4.6.2/../../../../arm-hardfloat-eabi/lib/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/libc.a(lib_a-memset.o)

↑のライブラリ:
File: /usr/local/arm-hardfloat-eabi/lib/thumb/cortex-m4/float-abi-hard/fpuv4-sp-d16/libc.a(lib_a-memset.o)
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "Cortex-M4"
  Tag_CPU_arch: v7E-M
  Tag_CPU_arch_profile: Microcontroller
  Tag_THUMB_ISA_use: Thumb-2
  Tag_FP_arch: VFPv4-D16
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align_needed: 8-byte
  Tag_ABI_align_preserved: 8-byte, except leaf SP
  Tag_ABI_enum_size: small
  Tag_ABI_HardFP_use: SP and DP
  Tag_ABI_VFP_args: VFP registers
  Tag_ABI_optimization_goals: Aggressive Speed
  Tag_FP_HP_extension: Allowed

※リンクも実行も可能だけど，FP演算の結果が正しくない...オブジェクト自体は，
vldrやvstr命令は生成されて，Dレジスタ群をビシバシ使っているけど，libgcc含む
ライブラリがS,Dレジスタ群を使っていない...ツールチェインのコンフィグミスかな。



[参考]
-mfloat-abi=hardで実行可能なコードを吐き出せるツールチェインを作ったときのメモ：

required libraries
0. prepare somethings:

fetched & extracted archives.

for dir in gmp* mpfr* mpc*; do mkdir $dir/i686-pc-cygwin-build; done


1. gmp:
cd gmp*/i686-pc-cygwin-build # change current work directory into build directory
CPPFLAGS=-fexceptions \
 ../configure \
 --prefix=/usr/local \
 --with-gnu-ld \
 --disable-shared \
 --enable-cxx
time make CFLAGS="-O2 -fomit-frame-pointer" LDFLAGS="-s" CPPFLAGS="-fexceptions" -j 4
time make CFLAGS="-O2 -fomit-frame-pointer" LDFLAGS="-s" CPPFLAGS="-fexceptions" check
make install


2. mpfr:
cd ../../mpfr*/i686-pc-cygwin-build # change current work directory into build directory
CFLAGS='-I/usr/local/include' \
LDFLAGS='-s -L/usr/local/lib' \
../configure \
 --prefix=/usr/local \
 --with-gnu-ld \
 --with-gmp=/usr/local
time make CFLAGS="-O2 -fomit-frame-pointer" LDFLAGS="-s -L/usr/local/lib" -j 4
time make CFLAGS="-O2 -fomit-frame-pointer" LDFLAGS="-s -L/usr/local/lib" check
make install


3. mpc:
cd ../../mpc*/i686-pc-cygwin-build # change current work directory into build directory
CFLAGS='-I/usr/local/include' \
LDFLAGS='-s -L/usr/local/lib' \
../configure \
 --prefix=/usr/local \
 --with-gnu-ld \
 --disable-shared \
 --with-gmp=/usr/local \
 --with-mpfr=/usr/local
time make CFLAGS="-O2 -fomit-frame-pointer" LDFLAGS="-s -L/usr/local/lib" -j 4
time make CFLAGS="-O2 -fomit-frame-pointer" LDFLAGS="-s -L/usr/local/lib" check
make install


cd ../..


4. prepare somethings for building cross-compiler toolchain:

ln -s cygwin/process.h /usr/include/process.h
↑cygwinのdllを更新したらspawn関係の機能をbinutilsのビルドが要求するので...


for dir in binutils* gcc* newlib*; do mkdir $dir/arm-hardfloat-eabi; done


SAT's patches applied


5. binutils:
cd binutils*/arm-hardloat-eabi-build
CFLAGS="-O2 -fomit-frame-pointer -I/usr/local/include" LDFLAGS="-s -L/usr/local/lib" \
 ../configure \
 --prefix=/usr/local \
 --target=arm-hardfloat-eabi \
 --build=i686-pc-cygwin --host=i686-pc-cygwin \
 --with-gmp=/usr/local --with-mpfr=/usr/local --with-mpc=/usr/local \
 --with-ppl=/usr/local --with-cloog=/usr/local \
 --disable-nls
time make CFLAGS="-O2 -fomit-frame-pointer -I/usr/local/include" LDFLAGS="-s -L/usr/local/lib" -j 4
time make CFLAGS="-O2 -fomit-frame-pointer -I/usr/local/include" LDFLAGS="-s -L/usr/local/lib" check
make install
hash -r
which arm-hardfloat-eabi-ld
`which arm-hardfloat-eabi-ld` --version


6. gcc:
cd ../../gcc*/arm-hardloat-eabi-build
CFLAGS="-O2 -fomit-frame-pointer  -I/usr/local/include" LDFLAGS="-s -L/usr/local/lib" \
 ../configure \
 --prefix=/usr/local \
 --target=arm-hardfloat-eabi \
 --build=i686-pc-cygwin --host=i686-pc-cygwin \
 --enable-languages=c,c++ \
 --enable-lto \
 --enable-multilib \
 --with-gnu-as \
 --with-gnu-ld \
 --with-newlib  --without-headers \
 --disable-newlib-supplied-syscalls \
 --with-libelf=/usr/local \
 --with-gmp=/usr/local \
 --with-mpfr=/usr/local \
 --with-mpc=/usr/local \
 --without-ppl \
 --without-cloog \
 --disable-shared \
 --disable-threads \
 --disable-decimal-float \
 --without-gnattools \
 --disable-boehm-gc \
 --disable-libquadmath \
 --disable-libffi \
 --disable-libjava \
 --disable-libada \
 --disable-libstdc++-v3 \
 --disable-libstdcxx-pch \
 --disable-libgfortran \
 --disable-libgo \
 --disable-libobjc \
 --disable-libgomp \
 --disable-libmudflap \
 --disable-libssp \
 --disable-zlib \
 --disable-nls
time make LANGUAGES="c c++" CFLAGS="-O2 -fomit-frame-pointer -I/usr/local/include" LDFLAGS="-s -L/usr/local/lib" -j 4
time make install
hash -r
which arm-hardfloat-eabi-gcc
`which arm-hardfloat-eabi-gcc` --version
`which arm-hardfloat-eabi-gcc` -print-sysroot
`which arm-hardfloat-eabi-gcc` -print-libgcc-file-name
`which arm-hardfloat-eabi-gcc` -print-multi-lib
`which arm-hardfloat-eabi-gcc` -dumpspecs
`which arm-hardfloat-eabi-gcc` --target-help


7. newlib:
cd ../../newlib*/arm-hardloat-eabi-build
CFLAGS="-g -O2 -fno-unroll-loops -DREENTRANT_SYSCALLS_PROVIDED" \
../configure \
 --prefix=/usr/local \
 --target=arm-hardfloat-eabi \
 --build=i686-pc-cygwin --host=i686-pc-cygwin \
 --enable-newlib-io-long-long \
 --enable-newlib-register-fini \
 --enable-newlib-reent-small \
 --disable-newlib-supplied-syscalls \
 --disable-libgloss \
 --disable-nls
time make CFLAGS="-g -O2 -fno-unroll-loops -DREENTRANT_SYSCALLS_PROVIDED" -j 4
time make CFLAGS="-g -O2 -fno-unroll-loops" check
make install


reference:
where i got archives from
https://github.com/MikeSmith/summon-arm-toolchain/tree/master/patches
	some patches
ftp://gcc.gnu.org/pub/
	binutils-2.22
	gcc-4.6.2
	gdb-7.4
	gmp-5.0.4
	mpfr-3.1.0
ftp://gcc.gnu.org/pub/gcc/infrastructure/
	cloog-ppl-0.15.11
http://sourceware.org/newlib/
	newlib-1.20.0
http://sourceware.org/insight/
	insight-6.8-1a
http://www.multiprecision.org
	mpc-0.8.2
http://bugseng.com/products/ppl
	ppl-0.11.2
http://www.mr511.de/software/
	libelf-0.8.13


[ご注意]
・このページの記載事項については，一切無保証です。
inserted by FC2 system