From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754131Ab1AXStJ (ORCPT ); Mon, 24 Jan 2011 13:49:09 -0500 Received: from mailout10.t-online.de ([194.25.134.21]:60930 "EHLO mailout10.t-online.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753562Ab1AXStH (ORCPT ); Mon, 24 Jan 2011 13:49:07 -0500 Message-ID: <4D3DC997.5040605@t-online.de> Date: Mon, 24 Jan 2011 19:48:55 +0100 From: Knut Petersen User-Agent: Mozilla/5.0 (X11; U; Linux i686; de; rv:1.9.1.16) Gecko/20101125 SUSE/3.0.11 Thunderbird/3.0.11 MIME-Version: 1.0 To: linux-kernel@vger.kernel.org CC: Chris Wilson , jesse.barnes@intel.com Subject: [BUG/REGRESSION] DRM / i915 / 2.6.37 and 2.6.38-rc*: DVI output gets disabled/reenabled under load Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit X-ID: T55qMrZvohKz5XVTsWSR8NXocwen65Bb2-9yLI1T0GMIi6WWvxRCmCfVlCZqTQ1wSO X-TOI-MSGID: 17b00744-54b8-43df-8f31-a74a3818cb61 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On an AOpen i915GMm-HFS I see the following problem: The LCD panel connected to DVI-1 gets disabled and then reenabled under high system load (e.g. -j 15 kernel compile) if I am working on the framebuffer console (no problems in X). Kernels affected: 2.6.37, 2.6.38-rc1, 2.6.38-rc2 Kernel _not_ affected: 2.6.31.14, 2.6.36.3 Details: There is a DVI (DVI-1) connector and a VGA (VGA-1) connector present. The kernel also detects a connector VGA-2, but there is no hardware connector for that. I tried to bisect, but that did not provide a reasonable result because of other problems with the code. Further inspection showed: The problem is related to the 10 second connector rescan. Idling at the prompt the status of VGA-2 is an almost stable connector_status_unknown (3). Under load that changes, the connector status is floating between connector_status_unknown and connector_status_disconnected. During the action taken after a status changes DVI-1 temporarily gets disabled. Disabling the connector rescan in the source code fixes the problem, but obviously that is not the best possible fix ;-) Knut