Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20180918023157.23951-1-stefan@agner.ch>
In-Reply-To: <20180918023157.23951-1-stefan@agner.ch>
From:   Nick Desaulniers <ndesaulniers@google.com>
Date:   Tue, 18 Sep 2018 10:16:15 -0700
Message-ID: <CAKwvOd=CRw+5T6Us-Pw+t02PYWFEh=SM49C+28x7_oH+hcPOzQ@mail.gmail.com>
Subject: Re: [PATCH] kbuild: allow to use GCC toolchain not in Clang search path
To:     Stefan Agner <stefan@agner.ch>
Cc:     Masahiro Yamada <yamada.masahiro@socionext.com>,
        Michal Marek <michal.lkml@markovi.net>,
        Arnd Bergmann <arnd@arndb.de>,
        Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Stephen Hines <srhines@google.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Sep 17, 2018 at 7:32 PM Stefan Agner <stefan@agner.ch> wrote:
>
> When using a GCC cross toolchain which is not in a compiled in
> Clang search path, Clang reverts to the system assembler and
> linker. This leads to assembler or linker errors, depending on
> which tool is first used for a given architecture.
>
> It seems that Clang is not searching $PATH for a matching
> assembler or linker.

Yes, Clang uses what's called a sysroot, which is meant to be more
hygenic than $PATH, IIUC.  This is surprising to folks familiar with
$PATH, but really helps include the correct headers and use the
correct toolchain when cross compiling, at the cost of freedom to put
the cross toolchain wherever in the filesystem. (this is my
understanding of sysroot, I just learned about it yesterday, so take
this with a grain of salt).

>
> Make sure that Clang picks up the correct assembler or linker by
> passing the cross compilers bin directory as search path.
>
> This allows to use Clang provided by distributions with GCC
> toolchains not in /usr/bin.

IIUC, its just utilities in binutils we're using (as, bfd.ld,
objcopy), not GCC, but maybe that's just being pedantic.

>
> Link: https://github.com/ClangBuiltLinux/linux/issues/78
> Signed-off-by: Stefan Agner <stefan@agner.ch>
> ---
> Nick, I removed your Reviewed-by since I had to change variable
> assignment slightly...

Yes, thank you, that's the right thing to do.

>
>  Makefile | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/Makefile b/Makefile
> index 4d5c883a98e5..d5de2db4b549 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -495,13 +495,15 @@ endif
>  ifeq ($(cc-name),clang)
>  ifneq ($(CROSS_COMPILE),)
>  CLANG_TARGET   := --target=$(notdir $(CROSS_COMPILE:%-=%))
> -GCC_TOOLCHAIN  := $(realpath $(dir $(shell which $(LD)))/..)
> +GCC_TOOLCHAIN_DIR := $(dir $(shell which $(LD)))
> +CLANG_PREFIX   := --prefix=$(GCC_TOOLCHAIN_DIR)

Yep: https://clang.llvm.org/docs/ClangCommandLineReference.html#cmdoption-clang-b-dir

> +GCC_TOOLCHAIN  := $(realpath $(GCC_TOOLCHAIN_DIR)/..)
>  endif
>  ifneq ($(GCC_TOOLCHAIN),)
>  CLANG_GCC_TC   := --gcc-toolchain=$(GCC_TOOLCHAIN)
>  endif
> -KBUILD_CFLAGS += $(CLANG_TARGET) $(CLANG_GCC_TC)
> -KBUILD_AFLAGS += $(CLANG_TARGET) $(CLANG_GCC_TC)
> +KBUILD_CFLAGS += $(CLANG_TARGET) $(CLANG_GCC_TC) $(CLANG_PREFIX)
> +KBUILD_AFLAGS += $(CLANG_TARGET) $(CLANG_GCC_TC) $(CLANG_PREFIX)
>  KBUILD_CFLAGS += $(call cc-option, -no-integrated-as)
>  KBUILD_AFLAGS += $(call cc-option, -no-integrated-as)
>  endif
> --
> 2.19.0
>

Thanks for this patch Stefan.
Reviewed-and-tested-by: Nick Desaulniers <ndesaulniers@google.com>
-- 
Thanks,
~Nick Desaulniers