From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:50502) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1esT9U-0001tN-Ot for guix-patches@gnu.org; Sun, 04 Mar 2018 07:54:09 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1esT9O-0005ym-GE for guix-patches@gnu.org; Sun, 04 Mar 2018 07:54:08 -0500 Received: from debbugs.gnu.org ([208.118.235.43]:35632) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1esT9O-0005yP-CW for guix-patches@gnu.org; Sun, 04 Mar 2018 07:54:02 -0500 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1esT9O-0006G3-2W for guix-patches@gnu.org; Sun, 04 Mar 2018 07:54:02 -0500 Subject: [bug#30604] [PATCH v8 2/7] linux-modules: Add install-modules. Resent-Message-ID: Date: Sun, 4 Mar 2018 13:53:27 +0100 From: Danny Milosavljevic Message-ID: <20180304135327.0a103e57@scratchpost.org> In-Reply-To: <87y3j84xtu.fsf@gnu.org> References: <20180302153408.14091-1-dannym@scratchpost.org> <20180303135533.6112-1-dannym@scratchpost.org> <20180303135533.6112-3-dannym@scratchpost.org> <87y3j84xtu.fsf@gnu.org> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; boundary="Sig_/s3HhKKL0oZ8Y2bY0RkrWxz1"; protocol="application/pgp-signature" List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-patches-bounces+kyle=kyleam.com@gnu.org Sender: "Guix-patches" To: Ludovic =?UTF-8?Q?Court=C3=A8s?= Cc: 30604@debbugs.gnu.org --Sig_/s3HhKKL0oZ8Y2bY0RkrWxz1 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Hi Ludo, On Sat, 03 Mar 2018 23:07:57 +0100 ludo@gnu.org (Ludovic Court=C3=A8s) wrote: > We could reuse modules.alias and modules.devname from the =E2=80=98linux-= libre=E2=80=99 > package (right?), but I guess it doesn=E2=80=99t hurt to generate custom = ones. Yes, but it would be somewhat dirty. Here's why: Linux-libre contains a superset of modules and also their modules.alias and modules.devname contain a superset of modules compared to the initrd. That means (1) modules.alias would contain aliases which map to modules which don't exist. Since we don't for-each it, who cares. (2) modules.devname would contain devnames for modules which don't exist. That means the boot code would create files in /dev which, when someone accesses them, would try to load modules which don't exist - and then the access would fail. Technically both are not bad, but (2) is bad "UX" practise nowadays - and not done in the Linux world any more since udev exists. I can still remember the days of 8540 entries static /dev filesystem (as in, really on the hard disk), 7000 of which didn't work (or even valid for this system type :P). If something goes wrong, good luck finding the needle in the haystack (which /dev file you can use to boot). --Sig_/s3HhKKL0oZ8Y2bY0RkrWxz1 Content-Type: application/pgp-signature Content-Description: OpenPGP digital signature -----BEGIN PGP SIGNATURE----- iQEzBAEBCAAdFiEEds7GsXJ0tGXALbPZ5xo1VCwwuqUFAlqb7EcACgkQ5xo1VCww uqXPwAf+P1CjrHIU4Ewpz6ucWfrmOOxqL3oR5zvmf1xtQGitXspjCtfP7OQhEok4 NS43aa4Rcurf32znHZvciYwiTIvkziXGJS3aMh7RbbJc2M8VG13gYTLYr9+YRs28 GSGLIFlHKAjozMeyOHOv9zzpMVgTH0zVE0B1dlV+0ttGUxSXu00v9QX4aoWTh97p p7HNT/yHvdhfYFUJDjiEk63eSvSkLyvWMxqChGU2iTOv3EDCRFREzKMYWHGBfjpu 5wQRUW5uivoX116X0iA2ipwJ6mb8C31yqInEVAFxt6GFzP5+AmwIrdVrGfG4N/we NkQd9yJol2LXAPV4qI8pfU6h7aFkPQ== =P/RW -----END PGP SIGNATURE----- --Sig_/s3HhKKL0oZ8Y2bY0RkrWxz1--