Re: [PATCH] selftests/mm: Fix cross compilation with LLVM

[Nick Desaulniers <ndesaulniers@google.com>]

On Wed, Jun 14, 2023 at 5:19 PM Mark Brown <broonie@kernel.org> wrote:
>
> Currently the MM selftests attempt to work out the target architecture by
> using CROSS_COMPILE or otherwise querying the host machine, storing the
> target architecture in a variable called MACHINE rather than the usual ARCH
> though as far as I can tell (including for x86_64) the value is the same as
> we would use for architecture.
>
> When cross compiling with LLVM we don't need a CROSS_COMPILE as LLVM can
> support many target architectures in a single build so this logic does not
> work, CROSS_COMPILE is not set and we end up selecting tests for the host
> rather than target architecture. Fix this by using the more standard ARCH
> to describe the architecture, taking it from the environment if specified.
>
> Signed-off-by: Mark Brown <broonie@kernel.org>

Broonie,
Thanks for the patch! What's the best way to test this?

$ ARCH=arm64 make LLVM=1 -j128 -C tools/testing/selfte
sts/mm/

Is what I would have guessed, but I get errors with or without this patch.

Also, fwiw, b4 shows that you don't have a pgp key setup for
broonie@kernel.org; I don't think it matters, but maybe it's a
surprise to you if you thought you had set that up,
$ b4 shazam https://lore.kernel.org/llvm/20230614-kselftest-mm-llvm-v1-1-180523f277d3@kernel.org/
...
  ✗ No key: openpgp/broonie@kernel.org


> ---
>  tools/testing/selftests/mm/Makefile | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
>
> diff --git a/tools/testing/selftests/mm/Makefile b/tools/testing/selftests/mm/Makefile
> index 23af4633f0f4..4f0c50c33ba7 100644
> --- a/tools/testing/selftests/mm/Makefile
> +++ b/tools/testing/selftests/mm/Makefile
> @@ -5,12 +5,15 @@ LOCAL_HDRS += $(selfdir)/mm/local_config.h $(top_srcdir)/mm/gup_test.h
>
>  include local_config.mk
>
> +ifeq ($(ARCH),)
> +
>  ifeq ($(CROSS_COMPILE),)
>  uname_M := $(shell uname -m 2>/dev/null || echo not)
>  else
>  uname_M := $(shell echo $(CROSS_COMPILE) | grep -o '^[a-z0-9]\+')
>  endif
> -MACHINE ?= $(shell echo $(uname_M) | sed -e 's/aarch64.*/arm64/' -e 's/ppc64.*/ppc64/')
> +ARCH ?= $(shell echo $(uname_M) | sed -e 's/aarch64.*/arm64/' -e 's/ppc64.*/ppc64/')
> +endif
>
>  # Without this, failed build products remain, with up-to-date timestamps,
>  # thus tricking Make (and you!) into believing that All Is Well, in subsequent
> @@ -65,7 +68,7 @@ TEST_GEN_PROGS += ksm_tests
>  TEST_GEN_PROGS += ksm_functional_tests
>  TEST_GEN_PROGS += mdwe_test
>
> -ifeq ($(MACHINE),x86_64)
> +ifeq ($(ARCH),x86_64)
>  CAN_BUILD_I386 := $(shell ./../x86/check_cc.sh "$(CC)" ../x86/trivial_32bit_program.c -m32)
>  CAN_BUILD_X86_64 := $(shell ./../x86/check_cc.sh "$(CC)" ../x86/trivial_64bit_program.c)
>  CAN_BUILD_WITH_NOPIE := $(shell ./../x86/check_cc.sh "$(CC)" ../x86/trivial_program.c -no-pie)
> @@ -87,13 +90,13 @@ TEST_GEN_PROGS += $(BINARIES_64)
>  endif
>  else
>
> -ifneq (,$(findstring $(MACHINE),ppc64))
> +ifneq (,$(findstring $(ARCH),ppc64))
>  TEST_GEN_PROGS += protection_keys
>  endif
>
>  endif
>
> -ifneq (,$(filter $(MACHINE),arm64 ia64 mips64 parisc64 ppc64 riscv64 s390x sparc64 x86_64))
> +ifneq (,$(filter $(ARCH),arm64 ia64 mips64 parisc64 ppc64 riscv64 s390x sparc64 x86_64))
>  TEST_GEN_PROGS += va_high_addr_switch
>  TEST_GEN_PROGS += virtual_address_range
>  TEST_GEN_PROGS += write_to_hugetlbfs
> @@ -112,7 +115,7 @@ $(TEST_GEN_PROGS): vm_util.c
>  $(OUTPUT)/uffd-stress: uffd-common.c
>  $(OUTPUT)/uffd-unit-tests: uffd-common.c
>
> -ifeq ($(MACHINE),x86_64)
> +ifeq ($(ARCH),x86_64)
>  BINARIES_32 := $(patsubst %,$(OUTPUT)/%,$(BINARIES_32))
>  BINARIES_64 := $(patsubst %,$(OUTPUT)/%,$(BINARIES_64))
>
>
> ---
> base-commit: 858fd168a95c5b9669aac8db6c14a9aeab446375
> change-id: 20230614-kselftest-mm-llvm-a25a7daffa6f
>
> Best regards,
> --
> Mark Brown <broonie@kernel.org>
>
>


-- 
Thanks,
~Nick Desaulniers