From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754709Ab3ACWpo (ORCPT ); Thu, 3 Jan 2013 17:45:44 -0500 Received: from mga02.intel.com ([134.134.136.20]:32678 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754517Ab3ACWpm (ORCPT ); Thu, 3 Jan 2013 17:45:42 -0500 Message-Id: <6c3329$7ukf5e@orsmga002.jf.intel.com> X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.84,405,1355126400"; d="scan'208";a="267009198" From: Chris Wilson Subject: Re: Linux 3.8-rc2 To: Linus Torvalds , Alexey Zaytsev Cc: Linux Kernel Mailing List , Daniel Vetter , Jesse Barnes , Rodrigo Vivi , Peter Ujfalusi , stable In-Reply-To: References: Date: Thu, 03 Jan 2013 22:45:36 +0000 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, 3 Jan 2013 13:32:04 -0800, Linus Torvalds wrote: > On Thu, Jan 3, 2013 at 1:09 PM, Alexey Zaytsev wrote: > > > > i915 got broken on my machine (Lenovo ThinkPad SL500) by this -rc. The screen > > goes blank after loading the module. I've bisected the problem to > > b0a2658acb5bf9ca86b4aab011b7106de3af0add "drm/i915: don't disable > > disconnected outputs". > > Hmm. Daniel - should I just revert it, or do you have a patch to > possibly fix the divide-by-zero error in g4x_compute_wm0() (which may > involve not getting that deep in the call chain to begin with or > whatever)? Maybe that fixes the blank screen too without a revert.. The issue appears to be a disagreement between intel_crtc->active, an attached framebuffer and there not being a valid mode set on the crtc. It looks like we end up in that situation with further confusion over intel_crtc->base.enabled vs intel_crtc->active. I suspect the answer is something like: diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index a9fb046..5e4b818 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -7940,10 +7940,8 @@ bool intel_set_mode(struct drm_crtc *crtc, } } - for_each_intel_crtc_masked(dev, prepare_pipes, intel_crtc) { - if (intel_crtc->base.enabled) - dev_priv->display.crtc_disable(&intel_crtc->base); - } + for_each_intel_crtc_masked(dev, prepare_pipes, intel_crtc) + dev_priv->display.crtc_disable(&intel_crtc->base); /* crtc->mode is already used by the ->mode_set callbacks, hence we need * to set it here already despite that we pass it down the callchain. However, until that confusion is unravelled, reverting b0a2658ac, seems the most sensible approach. -Chris -- Chris Wilson, Intel Open Source Technology Centre