| Message ID | 20220516210954.1660716-5-ndesaulniers@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | pass -march= only to compiler | expand |
+ Ard (I messed up my command line invocation of git send-email, sorry for more noise) https://lore.kernel.org/llvm/20220516210954.1660716-1-ndesaulniers@google.com/ On Mon, May 16, 2022 at 2:10 PM Nick Desaulniers <ndesaulniers@google.com> wrote: > > When both -march= and -Wa,-march= are specified for assembler or > assembler-with-cpp sources, GCC and Clang will prefer the -Wa,-march= > value but Clang will warn that -march= is unused. > > warning: argument unused during compilation: '-march=armv6k' > [-Wunused-command-line-argument] > > This is the top group of warnings we observe when using clang to > assemble the kernel via `ARCH=arm make LLVM=1`. > > Split the arch-y make variable into two, so that -march= flags only get > passed to the compiler, not the assembler. -D flags are added to > KBUILD_CPPFLAGS which is used for both C and assembler-with-cpp sources. > > Link: https://github.com/ClangBuiltLinux/linux/issues/1315 > Link: https://github.com/ClangBuiltLinux/linux/issues/1587 > Link: https://lore.kernel.org/llvm/628249e8.1c69fb81.d20fd.02ea@mx.google.com/ > Signed-off-by: Nick Desaulniers <ndesaulniers@google.com> > --- > arch/arm/Makefile | 34 ++++++++++++++++++++++++---------- > 1 file changed, 24 insertions(+), 10 deletions(-) > > diff --git a/arch/arm/Makefile b/arch/arm/Makefile > index 1029c2503aef..47a300f6f99c 100644 > --- a/arch/arm/Makefile > +++ b/arch/arm/Makefile > @@ -57,21 +57,34 @@ endif > KBUILD_CFLAGS += $(call cc-option,-fno-ipa-sra) > > # This selects which instruction set is used. > +arch-$(CONFIG_CPU_32v7M) :=-march=armv7-m > +arch-$(CONFIG_CPU_32v7) :=-march=armv7-a > +arch-$(CONFIG_CPU_32v6) :=-march=armv6 > +# Only override the compiler option if ARMv6. The ARMv6K extensions are > +# always available in ARMv7 > +ifeq ($(CONFIG_CPU_32v6),y) > +arch-$(CONFIG_CPU_32v6K) :=-march=armv6k > +endif > +arch-$(CONFIG_CPU_32v5) :=-march=armv5te > +arch-$(CONFIG_CPU_32v4T) :=-march=armv4t > +arch-$(CONFIG_CPU_32v4) :=-march=armv4 > +arch-$(CONFIG_CPU_32v3) :=-march=armv3m > + > # Note that GCC does not numerically define an architecture version > # macro, but instead defines a whole series of macros which makes > # testing for a specific architecture or later rather impossible. > -arch-$(CONFIG_CPU_32v7M) :=-D__LINUX_ARM_ARCH__=7 -march=armv7-m > -arch-$(CONFIG_CPU_32v7) :=-D__LINUX_ARM_ARCH__=7 -march=armv7-a > -arch-$(CONFIG_CPU_32v6) :=-D__LINUX_ARM_ARCH__=6 -march=armv6 > -# Only override the compiler opt:ion if ARMv6. The ARMv6K extensions are > +cpp-$(CONFIG_CPU_32v7M) :=-D__LINUX_ARM_ARCH__=7 > +cpp-$(CONFIG_CPU_32v7) :=-D__LINUX_ARM_ARCH__=7 > +cpp-$(CONFIG_CPU_32v6) :=-D__LINUX_ARM_ARCH__=6 > +# Only override the compiler option if ARMv6. The ARMv6K extensions are > # always available in ARMv7 > ifeq ($(CONFIG_CPU_32v6),y) > -arch-$(CONFIG_CPU_32v6K) :=-D__LINUX_ARM_ARCH__=6 -march=armv6k > +cpp-$(CONFIG_CPU_32v6K) :=-D__LINUX_ARM_ARCH__=6 > endif > -arch-$(CONFIG_CPU_32v5) :=-D__LINUX_ARM_ARCH__=5 -march=armv5te > -arch-$(CONFIG_CPU_32v4T) :=-D__LINUX_ARM_ARCH__=4 -march=armv4t > -arch-$(CONFIG_CPU_32v4) :=-D__LINUX_ARM_ARCH__=4 -march=armv4 > -arch-$(CONFIG_CPU_32v3) :=-D__LINUX_ARM_ARCH__=3 -march=armv3m > +cpp-$(CONFIG_CPU_32v5) :=-D__LINUX_ARM_ARCH__=5 > +cpp-$(CONFIG_CPU_32v4T) :=-D__LINUX_ARM_ARCH__=4 > +cpp-$(CONFIG_CPU_32v4) :=-D__LINUX_ARM_ARCH__=4 > +cpp-$(CONFIG_CPU_32v3) :=-D__LINUX_ARM_ARCH__=3 > > # This selects how we optimise for the processor. > tune-$(CONFIG_CPU_ARM7TDMI) :=-mtune=arm7tdmi > @@ -123,8 +136,9 @@ AFLAGS_ISA :=$(CFLAGS_ISA) > endif > > # Need -Uarm for gcc < 3.x > +KBUILD_CPPFLAGS +=$(cpp-y) > KBUILD_CFLAGS +=$(CFLAGS_ABI) $(CFLAGS_ISA) $(arch-y) $(tune-y) $(call cc-option,-mshort-load-bytes,$(call cc-option,-malignment-traps,)) -msoft-float -Uarm > -KBUILD_AFLAGS +=$(CFLAGS_ABI) $(AFLAGS_ISA) $(arch-y) $(tune-y) -include asm/unified.h -msoft-float > +KBUILD_AFLAGS +=$(CFLAGS_ABI) $(AFLAGS_ISA) -Wa,$(arch-y) $(tune-y) -include asm/unified.h -msoft-float > > CHECKFLAGS += -D__arm__ > > -- > 2.36.0.550.gb090851708-goog >
Hi Nick,
I love your patch! Yet something to improve:
[auto build test ERROR on 0ac824f379fba2c2b17b75fd5ada69cd68c66348]
url: https://github.com/intel-lab-lkp/linux/commits/Nick-Desaulniers/pass-march-only-to-compiler/20220517-051756
base: 0ac824f379fba2c2b17b75fd5ada69cd68c66348
config: arm-randconfig-r026-20220516 (https://download.01.org/0day-ci/archive/20220518/202205180917.RNpZaxIl-lkp@intel.com/config)
compiler: clang version 15.0.0 (https://github.com/llvm/llvm-project 853fa8ee225edf2d0de94b0dcbd31bea916e825e)
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# install arm cross compiling tool for clang build
# apt-get install binutils-arm-linux-gnueabi
# https://github.com/intel-lab-lkp/linux/commit/6da98100eed87e4316be5ec584fe415134f25a3e
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Nick-Desaulniers/pass-march-only-to-compiler/20220517-051756
git checkout 6da98100eed87e4316be5ec584fe415134f25a3e
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=arm SHELL=/bin/bash
If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>
All errors (new ones prefixed by >>):
>> arch/arm/crypto/poly1305-core.S:16:1: error: target does not support ARM mode
.code 32
^
Lore thread, for context: https://lore.kernel.org/llvm/20220516210954.1660716-1-ndesaulniers@google.com/ On Tue, May 17, 2022 at 6:26 PM kernel test robot <lkp@intel.com> wrote: > > Hi Nick, > > I love your patch! Yet something to improve: > > [auto build test ERROR on 0ac824f379fba2c2b17b75fd5ada69cd68c66348] > > url: https://github.com/intel-lab-lkp/linux/commits/Nick-Desaulniers/pass-march-only-to-compiler/20220517-051756 > base: 0ac824f379fba2c2b17b75fd5ada69cd68c66348 > config: arm-randconfig-r026-20220516 (https://download.01.org/0day-ci/archive/20220518/202205180917.RNpZaxIl-lkp@intel.com/config) ^ looks like this is a THUMB2 build. > compiler: clang version 15.0.0 (https://github.com/llvm/llvm-project 853fa8ee225edf2d0de94b0dcbd31bea916e825e) > reproduce (this is a W=1 build): > wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross > chmod +x ~/bin/make.cross > # install arm cross compiling tool for clang build > # apt-get install binutils-arm-linux-gnueabi > # https://github.com/intel-lab-lkp/linux/commit/6da98100eed87e4316be5ec584fe415134f25a3e > git remote add linux-review https://github.com/intel-lab-lkp/linux > git fetch --no-tags linux-review Nick-Desaulniers/pass-march-only-to-compiler/20220517-051756 > git checkout 6da98100eed87e4316be5ec584fe415134f25a3e > # save the config file > mkdir build_dir && cp config build_dir/.config > COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=arm SHELL=/bin/bash > > If you fix the issue, kindly add following tag as appropriate > Reported-by: kernel test robot <lkp@intel.com> > > All errors (new ones prefixed by >>): > > >> arch/arm/crypto/poly1305-core.S:16:1: error: target does not support ARM mode > .code 32 > ^ It looks like arch/arm/crypto/poly1305-core.S has a preprocessor guard for __thumb2__. My change changes the command line from: from: $(CC) ... -march=armv7-m ... to: $(CC ... -Wa,-march=armv7-m ... (where ... contains `-mthumb -Wa,-mthumb` for BOTH). $ arm-linux-gnueabi-gcc -march=armv7-m -mthumb -dM -E - < /dev/null | grep thumb #define __thumb2__ 1 #define __thumb__ 1 $ arm-linux-gnueabi-gcc -Wa,-march=armv7-m -mthumb -dM -E - < /dev/null | grep thumb #define __thumb__ 1 $ clang --target=arm-linux-gnueabi -march=armv7-m -mthumb -dM -E - < /dev/null | grep thumb #define __thumb2__ 1 #define __thumb__ 1 $ clang --target=arm-linux-gnueabi -Wa,-march=armv7-m -mthumb -dM -E - < /dev/null | grep thumb #define __thumb__ 1 $ (so it seems that the preprocessor definition of `__thumb2__` is dependent on `-march=`, not `-Wa,-march=`). David, we might have a very subtle bug in clang: $ clang --target=arm-linux-gnueabi -mthumb -Wa,-mthumb -march=armv7-m -Wa,-march=armv7-m -x assembler-with-cpp -Wunused-command-line-argument - < /dev/null -dM -E <prints a bunch of preprocessor defines, but no instance of -Wunused-command-line-argument> $ clang --target=arm-linux-gnueabi -mthumb -Wa,-mthumb -march=armv7-m -Wa,-march=armv7-m -x assembler-with-cpp -Wunused-command-line-argument - < /dev/null -dM -c clang-15: warning: argument unused during compilation: '-march=armv7-m' [-Wunused-command-line-argument] That seems weird because -c vs -E have different behaviors regarding -Wunused-command-line-argument, and it seems like for `-x assembler-with-cpp` that the -march= flag without -Wa, prefix *is* expected to control the behavior of certain preprocessor defines like `__thumb2__`. Perhaps a more straightforward test case: ```asm @ x.S @ clang --target=arm-linux-gnueabi -mthumb -Wa,-mthumb -march=armv7-m \ @ -Wa,-march=armv7-m -Wunused-command-line-argument x.S -c .syntax unified .text foo: movs r0, #__thumb2__ ``` $ clang --target=arm-linux-gnueabi -mthumb -Wa,-mthumb -march=armv7-m -Wa,-march=armv7-m -Wunused-command-line-argument x.S -c clang-15: warning: argument unused during compilation: '-march=armv7-m' [-Wunused-command-line-argument] $ llvm-objdump -dr x.o | tail -n 2 00000000 <foo>: 0: 01 20 movs r0, #1 Obviously that warning about `-Wunused-command-line-argument` is incorrect; the value of `__thumb2__` comes from the preprocessor and is controlled by `-march=armv7-m`. It can't be simultaneously "unused" and used to define particular preprocessor directives. Using the above command line invocation without `-march=armv7-m` produces the assembler-time failure: /tmp/x-03ab60.s:10:3: error: unsupported relocation on symbol movs r0, #__thumb2__ ^ because __thumb2__ is not defined. (perhaps `#ifundef __thumb2__ #error "oops" #endif` would be simpler yet). > > -- > 0-DAY CI Kernel Test Service > https://01.org/lkp > -- Thanks, ~Nick Desaulniers
On Wed, May 18, 2022 at 03:38:18PM -0700, Nick Desaulniers wrote: > (so it seems that the preprocessor definition of `__thumb2__` is > dependent on `-march=`, not `-Wa,-march=`). Yes, it will be. -Wa,... are only ever passed to the assembler and do not affect the preprocessor nor the compiler with GCC.
diff --git a/arch/arm/Makefile b/arch/arm/Makefile index 1029c2503aef..47a300f6f99c 100644 --- a/arch/arm/Makefile +++ b/arch/arm/Makefile @@ -57,21 +57,34 @@ endif KBUILD_CFLAGS += $(call cc-option,-fno-ipa-sra) # This selects which instruction set is used. +arch-$(CONFIG_CPU_32v7M) :=-march=armv7-m +arch-$(CONFIG_CPU_32v7) :=-march=armv7-a +arch-$(CONFIG_CPU_32v6) :=-march=armv6 +# Only override the compiler option if ARMv6. The ARMv6K extensions are +# always available in ARMv7 +ifeq ($(CONFIG_CPU_32v6),y) +arch-$(CONFIG_CPU_32v6K) :=-march=armv6k +endif +arch-$(CONFIG_CPU_32v5) :=-march=armv5te +arch-$(CONFIG_CPU_32v4T) :=-march=armv4t +arch-$(CONFIG_CPU_32v4) :=-march=armv4 +arch-$(CONFIG_CPU_32v3) :=-march=armv3m + # Note that GCC does not numerically define an architecture version # macro, but instead defines a whole series of macros which makes # testing for a specific architecture or later rather impossible. -arch-$(CONFIG_CPU_32v7M) :=-D__LINUX_ARM_ARCH__=7 -march=armv7-m -arch-$(CONFIG_CPU_32v7) :=-D__LINUX_ARM_ARCH__=7 -march=armv7-a -arch-$(CONFIG_CPU_32v6) :=-D__LINUX_ARM_ARCH__=6 -march=armv6 -# Only override the compiler opt:ion if ARMv6. The ARMv6K extensions are +cpp-$(CONFIG_CPU_32v7M) :=-D__LINUX_ARM_ARCH__=7 +cpp-$(CONFIG_CPU_32v7) :=-D__LINUX_ARM_ARCH__=7 +cpp-$(CONFIG_CPU_32v6) :=-D__LINUX_ARM_ARCH__=6 +# Only override the compiler option if ARMv6. The ARMv6K extensions are # always available in ARMv7 ifeq ($(CONFIG_CPU_32v6),y) -arch-$(CONFIG_CPU_32v6K) :=-D__LINUX_ARM_ARCH__=6 -march=armv6k +cpp-$(CONFIG_CPU_32v6K) :=-D__LINUX_ARM_ARCH__=6 endif -arch-$(CONFIG_CPU_32v5) :=-D__LINUX_ARM_ARCH__=5 -march=armv5te -arch-$(CONFIG_CPU_32v4T) :=-D__LINUX_ARM_ARCH__=4 -march=armv4t -arch-$(CONFIG_CPU_32v4) :=-D__LINUX_ARM_ARCH__=4 -march=armv4 -arch-$(CONFIG_CPU_32v3) :=-D__LINUX_ARM_ARCH__=3 -march=armv3m +cpp-$(CONFIG_CPU_32v5) :=-D__LINUX_ARM_ARCH__=5 +cpp-$(CONFIG_CPU_32v4T) :=-D__LINUX_ARM_ARCH__=4 +cpp-$(CONFIG_CPU_32v4) :=-D__LINUX_ARM_ARCH__=4 +cpp-$(CONFIG_CPU_32v3) :=-D__LINUX_ARM_ARCH__=3 # This selects how we optimise for the processor. tune-$(CONFIG_CPU_ARM7TDMI) :=-mtune=arm7tdmi @@ -123,8 +136,9 @@ AFLAGS_ISA :=$(CFLAGS_ISA) endif # Need -Uarm for gcc < 3.x +KBUILD_CPPFLAGS +=$(cpp-y) KBUILD_CFLAGS +=$(CFLAGS_ABI) $(CFLAGS_ISA) $(arch-y) $(tune-y) $(call cc-option,-mshort-load-bytes,$(call cc-option,-malignment-traps,)) -msoft-float -Uarm -KBUILD_AFLAGS +=$(CFLAGS_ABI) $(AFLAGS_ISA) $(arch-y) $(tune-y) -include asm/unified.h -msoft-float +KBUILD_AFLAGS +=$(CFLAGS_ABI) $(AFLAGS_ISA) -Wa,$(arch-y) $(tune-y) -include asm/unified.h -msoft-float CHECKFLAGS += -D__arm__
When both -march= and -Wa,-march= are specified for assembler or assembler-with-cpp sources, GCC and Clang will prefer the -Wa,-march= value but Clang will warn that -march= is unused. warning: argument unused during compilation: '-march=armv6k' [-Wunused-command-line-argument] This is the top group of warnings we observe when using clang to assemble the kernel via `ARCH=arm make LLVM=1`. Split the arch-y make variable into two, so that -march= flags only get passed to the compiler, not the assembler. -D flags are added to KBUILD_CPPFLAGS which is used for both C and assembler-with-cpp sources. Link: https://github.com/ClangBuiltLinux/linux/issues/1315 Link: https://github.com/ClangBuiltLinux/linux/issues/1587 Link: https://lore.kernel.org/llvm/628249e8.1c69fb81.d20fd.02ea@mx.google.com/ Signed-off-by: Nick Desaulniers <ndesaulniers@google.com> --- arch/arm/Makefile | 34 ++++++++++++++++++++++++---------- 1 file changed, 24 insertions(+), 10 deletions(-)