Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:4 as permitted sender) client-ip=2620:137:e000::3:4;
MIME-Version: 1.0
References: <20231205-libstringheader-v1-1-7f9c573053a7@gmail.com>
 <20231205130449.8e330a26ecbed1f7b5ad5d7a@linux-foundation.org>
 <CAKwvOdn+VTM+LY6ALcgaZTL57JpiKt5rBPMSPNXsgS3MCENhDQ@mail.gmail.com>
 <20231205132452.418722bea8f6878dca88142a@linux-foundation.org>
 <CAKwvOdn=og6h5gVdDCjFDANs3MN-_CD4OZ9oRM=o9YAvoTzkzw@mail.gmail.com> <20231205214359.GF1674809@ZenIV>
In-Reply-To: <20231205214359.GF1674809@ZenIV>
From:   Nick Desaulniers <ndesaulniers@google.com>
Date:   Mon, 11 Dec 2023 12:47:39 -0800
Message-ID: <CAKwvOdkC0oHWKGJ1Z6k_U9cwkjdcf=iweG-G78jXSJgWFk86Jg@mail.gmail.com>
Subject: Re: [PATCH] lib/string: shrink lib/string.i via IWYU
To:     Al Viro <viro@zeniv.linux.org.uk>,
        Greg KH <gregkh@linuxfoundation.org>
Cc:     Andrew Morton <akpm@linux-foundation.org>, tanzirh@google.com,
        Kees Cook <keescook@chromium.org>,
        Andy Shevchenko <andy@kernel.org>,
        linux-hardening@vger.kernel.org, linux-kernel@vger.kernel.org,
        llvm@lists.linux.dev, Christoph Hellwig <hch@infradead.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Dec 5, 2023 at 1:44=E2=80=AFPM Al Viro <viro@zeniv.linux.org.uk> wr=
ote:
>
> On Tue, Dec 05, 2023 at 01:39:47PM -0800, Nick Desaulniers wrote:
>
> > The tooling Tanzir is working on does wrap IWYU, and does support such
> > mapping (of 'low level' to 'high level' headers; more so, if it
> > recommends X you can override to suggest Y instead).
> >
> > arch/nios/ also doesn't provide a bug.h, which this patch is
> > suggesting we include directly.  I guess the same goes for
> > asm/rwonce.h.
>
> See include/asm-generic/Kbuild:
> mandatory-y +=3D bug.h
> ...
> mandatory-y +=3D rwonce.h
>
> IOW, sh will have asm/bug.h and as/rwonce.h copied from asm-generic.
>
> Still, includes of asm/*.h had been a massive headache historically
> and breeding more of those shouldn't be overdone.
>
> More painful problem is arch- and config-dependent stuff, though...

Ah, it looks like include/uapi/asm-generic/Kbuild also makes use of
this pattern using mandatory-y.

So I think we can handle this as a two step translation. We can tell
the tooling to 'nevermind recommending X, always replace
recommendations for X with Y <for raisins>', so:
1. any recommendation to use asm-generic/foo.h should be replaced with
asm/foo.h (always, based on those 2 Kbuild files, could autogenerate)
2. some recommendations to use asm/foo.h should be replaced with
linux/foo.h (not necessarily codified anywhere; depends on if there is
a linux/foo.h, will manually curate this list for now)

Orthogonal but some places in tree can be cleaned up to include
linux/foo.h rather than asm/foo.h.

Does this sound like an improvement to my mental model of the
conventions used for kernel header inclusion within the linux kernel?

Tanzir nearly has the above done (for 1, 2 we will probably need to
iterate on more).  We've also beefed up our local testing to test more
architectures.  I expect Tanzir to send a v2 of this patch (as a
series, with the fix for ARCH=3Dsh) later this week, if the above seems
correct.
--=20
Thanks,
~Nick Desaulniers