Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.35 as permitted sender) client-ip=23.128.96.35;
MIME-Version: 1.0
References: <20231205-libstringheader-v1-1-7f9c573053a7@gmail.com> <20231205213807.GE1674809@ZenIV>
In-Reply-To: <20231205213807.GE1674809@ZenIV>
From:   Nick Desaulniers <ndesaulniers@google.com>
Date:   Tue, 5 Dec 2023 13:51:10 -0800
Message-ID: <CAKwvOd=2VASkaLvjU+7kkbvhu2CimYn5KUGJBDRePyUhtrNK2Q@mail.gmail.com>
Subject: Re: [PATCH] lib/string: shrink lib/string.i via IWYU
To:     Al Viro <viro@zeniv.linux.org.uk>
Cc:     tanzirh@google.com, Kees Cook <keescook@chromium.org>,
        Andy Shevchenko <andy@kernel.org>,
        linux-hardening@vger.kernel.org, linux-kernel@vger.kernel.org,
        Nick DeSaulniers <nnn@google.com>,
        Andrew Morton <akpm@linux-foundation.org>, llvm@lists.linux.dev
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Dec 5, 2023 at 1:38=E2=80=AFPM Al Viro <viro@zeniv.linux.org.uk> wr=
ote:
>
> On Tue, Dec 05, 2023 at 08:58:53PM +0000, tanzirh@google.com wrote:
> > This diff uses an open source tool include-what-you-use (IWYU) to modif=
y
> > the include list changing indirect includes to direct includes.
>
> How does it account for arch- and config-dependent indirect includes?
>
> In particular, on sh this patch breaks, since there word-at-a-time.h does=
 not
> contain an include of kernel.h, even though it uses REPEAT_BYTE.  This is
> a very simple case (they really ought to include kernel.h, same as all ot=
her
> instances of word-at-a-time.h), but I would expect arseloads of more subt=
le
> breakage in anything less trivial.
>
> And I'm not at all sure that there's no config-dependent breakage as well=
 -
> this had been caught by quick make allmodconfig + make lib/string.o on
> a bunch of architectures; the graph of indirects includes (as well as the
> set of symbols needed for given header) is very much config-dependent.

We're sending these to Kees to stage in branch flowing into linux-next
in order for the patches to get soak time in linux-next; it's not
possible to test every possible randconfig, but with enough soak time
and the bots chewing on linux-next, I think we can get to a certain
level of confidence.

We'll ramp up the amount of testing we're doing locally as well. (I
did quite a few randconfigs locally in a loop; didn't test all
architectures) We can probably fetch the kernel.org toolchains for
very extensive testing.
https://mirrors.edge.kernel.org/pub/tools/crosstool/

>
> > IWYU is implemented using the IWYUScripts github repository which is a =
tool that is
> > currently undergoing development. These changes seek to improve build t=
imes.
> >
> > This change to lib/string.c resulted in a preprocessed size of
> > lib/string.i from 26371 lines to 5232 lines (-80%).
>
> It also breeds includes of asm/*.h, by the look of the output, which is
> not a good thing in general ;-/  E.g. #include <asm/uaccess.h> *anywhere*
> outside of linux/uaccess.h is a bad idea.

It's not clear to me when it's ok to #include <asm/*.h>.  Is there a
convention here that I'm missing?

>
> > If there are any problems with the output of the tool please raise an
> > issue on the github.
>
> Sorry, no.  Your tool, your workflow, of course, but I don't do github is=
sues.
> You are welcome to the contents of this reply, but I'm not using browser-=
based
> UI without very strong reasons; this one isn't.

No problem; Tanzir, please drop that part of the commit message in
future patches.
--=20
Thanks,
~Nick Desaulniers