Re: [PATCH RFC kbuild] lib/crypto: blake2s: include as built-in

From: Masahiro Yamada <masahiroy@kernel.org>
To: "Jason A. Donenfeld" <Jason@zx2c4.com>
Cc: Randy Dunlap <rdunlap@infradead.org>,
	Ard Biesheuvel <ardb@kernel.org>,
	Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
	Herbert Xu <herbert@gondor.apana.org.au>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: Re: [PATCH RFC kbuild] lib/crypto: blake2s: include as built-in
Date: Sat, 25 Dec 2021 16:10:52 +0900	[thread overview]
Message-ID: <CAK7LNAQViiExZQTKiRVE_2wiQVjeS+MLCvJO_xn3sJaqcuoPfA@mail.gmail.com> (raw)
In-Reply-To: <CAHmME9oKMm_VZ10yArhstX4_=f6Vj6Po-uii_0t_wjtiG81dvg@mail.gmail.com>

On Thu, Dec 23, 2021 at 12:55 AM Jason A. Donenfeld <Jason@zx2c4.com> wrote:
>
> On Wed, Dec 22, 2021 at 4:49 PM Randy Dunlap <rdunlap@infradead.org> wrote:
> > If lib-y ignores (drops) an object file (usually because it is not used),
> > the usual way to force it to be included is to add it to obj-y instead
> > of lib-y (see many examples in lib/Makefile).
>
> This is not a problem with lib-y. This is a problem with libs-y. Note
> the 's'. The former is working as expected. The latter controls
> whether a directory's lib.a is picked up. For whatever reason, the
> build system isn't respecting a libs-y declaration added to
> lib/Makefile like it respects for one added to arch/*/Makefile.
>
> > However, this may cause problems with weak symbols. I don't recall it being
> > used in that scenario.
>
> The reason we're using lib-y is so that unused code is pruned in the
> case that the weak symbol isn't used. IOW, a usual use for lib-y. And
> it works just fine. As mentioned, the issue is just with libs-y not
> finding that lib.a file.

lib-y does not work like that.

See commit 7273ad2b08f8ac9563579d16a3cf528857b26f49

When CONFIG_MODULES=y (this is true for most of usecases),
there is not much difference between obj-y and lib-y.

So, weak functions will remain in vmlinux even if they are unused.

-- 
Best Regards
Masahiro Yamada