Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210708232522.3118208-1-ndesaulniers@google.com>
 <20210708232522.3118208-3-ndesaulniers@google.com> <CAK7LNARye5Opc0AdXpn+DHB7hTaphoRSCUWxJgXu+sjuNjWUCg@mail.gmail.com>
 <CAHk-=wgGxu4_hgzdYpFuKd95SfnkJbPTWAQ9-fMgmMN1Oxs2xQ@mail.gmail.com>
In-Reply-To: <CAHk-=wgGxu4_hgzdYpFuKd95SfnkJbPTWAQ9-fMgmMN1Oxs2xQ@mail.gmail.com>
From:   Arnd Bergmann <arnd@kernel.org>
Date:   Tue, 20 Jul 2021 21:58:32 +0200
Message-ID: <CAK8P3a3=JBQow-Ws6tt81k93aw+OCV5C2CtSWxASkv=iQZPGUw@mail.gmail.com>
Subject: Re: [PATCH v2 2/2] Makefile: infer CROSS_COMPILE from SRCARCH for
 LLVM=1 LLVM_IAS=1
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Masahiro Yamada <masahiroy@kernel.org>,
        Nick Desaulniers <ndesaulniers@google.com>,
        Miguel Ojeda <ojeda@kernel.org>,
        Fangrui Song <maskray@google.com>,
        Michal Marek <michal.lkml@markovi.net>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
        clang-built-linux <clang-built-linux@googlegroups.com>,
        Geert Uytterhoeven <geert@linux-m68k.org>,
        Christoph Hellwig <hch@infradead.org>,
        Nathan Chancellor <nathan@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Jul 20, 2021 at 7:43 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Tue, Jul 20, 2021 at 1:05 AM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> We do most of the other heavy lifting in this area in Kconfig anyway,
> why not add that compiler choice?
>
> Obviously it would be gated by the tests to see which compilers are
> _installed_ (and that they are valid versions), so that it doesn't ask
> stupid things ("do you want gcc or clang" when only one of them is
> installed and/or viable).

I don't see a good way of making Kconfig options both select the
compiler and defining variables based on the compiler, since that
would mean teaching Kconfig about re-evaluating all compiler
dependent settings whenever the first option changes.

I do have another idea that I think would work though.

> Hmm? So then any "LLVM=1" thing would be about the "make config"
> stage, not the actual build stage.
>
> (It has annoyed me for years that if you want to cross-compile, you
> first have to do "make ARCH=xyz config" and then remember to do "make
> ARCH=xyz" for the build too, but I cross-compile so seldom that I've
> never really cared).

The best thing that I have come up with is a pre-configure step, where
an object tree gets seeded with a makefile fragment that gets included
for any 'make' invocation. This would set 'ARCH=', 'CROSS_COMPILE',
'CC=' and possibly any other option that gets passed to 'make' as
a variable and has to exist before calling 'make *config'.

This is probably the easiest part here, and it lets you set up multiple
object directories in which you can then do

make O=obj-x86 defconfig vmlinux modules -skj32
make O=obj-x86-clang
make O=obj-arm64 randconfig
...

without ever having to type the additional CC/CROSS_COMPILE/LLVM
variables again.

One step further is the script that automatically finds a working toolchain
for a given architecture and sets up that object directory accordingly,
or even downloads a working cross-toolchain from kernel.org when
asked to do that.

        Arnd