From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751422Ab1IUEvi (ORCPT ); Wed, 21 Sep 2011 00:51:38 -0400 Received: from home.keithp.com ([63.227.221.253]:39706 "EHLO keithp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751050Ab1IUEvi (ORCPT ); Wed, 21 Sep 2011 00:51:38 -0400 From: Keith Packard To: Jesse Barnes Cc: Dave Airlie , intel-gfx@lists.freedesktop.org, linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org Subject: Re: [Intel-gfx] [PATCH 9/9] drm/i915: Disable eDP VDD in a delayed work proc instead of synchronously In-Reply-To: <20110921094759.22423597@jbarnes-x220> References: <1316470923-27832-1-git-send-email-keithp@keithp.com> <1316470923-27832-10-git-send-email-keithp@keithp.com> <20110921094759.22423597@jbarnes-x220> User-Agent: Notmuch/0.6.1-66-ga900dda (http://notmuchmail.org) Emacs/23.3.1 (i486-pc-linux-gnu) Date: Tue, 20 Sep 2011 21:51:33 -0700 Message-ID: MIME-Version: 1.0 Content-Type: multipart/signed; boundary="=-=-="; micalg=pgp-sha1; protocol="application/pgp-signature" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org --=-=-= Content-Transfer-Encoding: quoted-printable On Wed, 21 Sep 2011 09:47:59 +0530, Jesse Barnes = wrote: > I'm worried this makes our PPS even more complex and hard to follow. > I'd rather see VDD AUX applied only when we need it (dpms, mode set and > detect; for hotplug we can assume the panel is alive) and that we > carefully disable it after waiting for a time after a full PPS on > sequence when doing a mode set or dpms on command. You'll see the fairly long list of places where VDD AUX is needed in my other reply; a couple of them could probably be fixed by moving the PPS calls around. > Having all the power stuff at the highest levels would be clearest I > think. Yes, making it cleaner would help a ton. There are some basic problems with the DRM API that make this hard though -- intel_dp_prepare may not ever be followed by a call to intel_dp_commit. That's why I had the VDD AUX stuff get turned off by a delayed work proc instead. Also, leaving VDD AUX high after EDID is fetched means that we can start the mode setting immediately, rather than having to wait for the power-off/power-on delay (which is really long). What we could do is force VDD AUX off after the panel gets turned on; that would ensure that turning the panel off would actually turn the power off, rather than having VDD stay high for some time after that. =2D-=20 keith.packard@intel.com --=-=-= Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iD8DBQFOeW1WQp8BWwlsTdMRApaaAJwKUX3Y9l0PH+WDXmz7A99jXaza7ACgvOZe 6VHL4AbzzphbiQs3X+9ypUw= =M2kl -----END PGP SIGNATURE----- --=-=-=--