From mboxrd@z Thu Jan  1 00:00:00 1970
From: Kenneth Graunke <kenneth@whitecape.org>
Subject: Re: [PATCH] drm/i915: Use PIPE_CONTROL for flushing on
 gen6+.
Date: Mon, 26 Sep 2011 13:21:49 -0700
Message-ID: <4E80DEDD.5090403@whitecape.org>
References: <1317063563-1526-1-git-send-email-kenneth@whitecape.org>
	<d08817$1l8egs@azsmga001.ch.intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Return-path: <intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org>
Received: from homiemail-a7.g.dreamhost.com (caiajhbdcaid.dreamhost.com
	[208.97.132.83])
	by gabe.freedesktop.org (Postfix) with ESMTP id 558239E8FD
	for <intel-gfx@lists.freedesktop.org>;
	Mon, 26 Sep 2011 13:20:57 -0700 (PDT)
In-Reply-To: <d08817$1l8egs@azsmga001.ch.intel.com>
List-Unsubscribe: <http://lists.freedesktop.org/mailman/options/intel-gfx>,
	<mailto:intel-gfx-request@lists.freedesktop.org?subject=unsubscribe>
List-Archive: <http://lists.freedesktop.org/archives/intel-gfx>
List-Post: <mailto:intel-gfx@lists.freedesktop.org>
List-Help: <mailto:intel-gfx-request@lists.freedesktop.org?subject=help>
List-Subscribe: <http://lists.freedesktop.org/mailman/listinfo/intel-gfx>,
	<mailto:intel-gfx-request@lists.freedesktop.org?subject=subscribe>
Sender: intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org
Errors-To: intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org
To: Chris Wilson <chris@chris-wilson.co.uk>
Cc: intel-gfx@lists.freedesktop.org
List-Id: intel-gfx@lists.freedesktop.org

On 09/26/2011 12:16 PM, Chris Wilson wrote:
> On Mon, 26 Sep 2011 11:59:23 -0700, Kenneth Graunke <kenneth@whitecape.org> wrote:
>> From: Jesse Barnes <jbarnes@virtuousgeek.org>
> 
> From the school of "If ain't broke, don't fix it" there needs to be a real
> explanation of why this change is required here.
> 
> PIPE_CONTROL and its workarounds is a very bitter pill to swallow if
> MI_FLUSH continues to function.
> -Chris

I hear you.

The issue is that (from Eric's reading of the simulator) MI_FLUSH seems
to be equivalent to PIPE_CONTROL with some unknown set of bits
enabled...which means we likely do need workarounds.  It's just not
clear which ones.

Also, according to the BSpec, MI_FLUSH is no longer validated or
guaranteed to work on Ivybridge.  I heard they said that about
Sandybridge as well but later recanted...I don't know if they will this
time, though.

I suspect that it actually is broke, and we do need to fix it.  This
seems like a first step.

--Kenneth