From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:59379 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757161Ab3IDMap (ORCPT ); Wed, 4 Sep 2013 08:30:45 -0400 Date: Wed, 4 Sep 2013 14:32:39 +0200 From: Alexander Gordeev To: Tejun Heo Cc: linux-kernel@vger.kernel.org, x86@kernel.org, linux-pci@vger.kernel.org, linux-ide@vger.kernel.org, Ingo Molnar , Joerg Roedel , Jan Beulich , Bjorn Helgaas Subject: Re: [PATCH 0/4] AHCI: Conserve interrupts with pci_enable_msi_block_part() interface Message-ID: <20130904123238.GC8726@dhcp-26-207.brq.redhat.com> References: <20130903135541.GB10522@htj.dyndns.org> <20130903140954.GC10522@htj.dyndns.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <20130903140954.GC10522@htj.dyndns.org> Sender: linux-pci-owner@vger.kernel.org List-ID: On Tue, Sep 03, 2013 at 10:09:54AM -0400, Tejun Heo wrote: > On Tue, Sep 03, 2013 at 09:55:41AM -0400, Tejun Heo wrote: > > Hmmm.... I've been looking at the code and and a curiosity. Why does > > multiple MSI support implicitly enabled threaded IRQ handling? Why > > are those two linked? Also, do you have any numbers to show that this > > actually is better? Handling the processing off to a thread isn't a > > light operation. > > Also, it probably is a good idea to skip dummy ports when requesting > irqs from ahci_host_activate(), which BTW should probably be renamed > to ahci_host_activate_mmsi(). Hmm.. that is actually a great idea. What I am not sure about whether is a dummy port still can send (spurious?) interrupts? The hardware interrupt handler would have to be reworked then. Seems as a yet another topic to me ;) > -- > tejun -- Regards, Alexander Gordeev agordeev@redhat.com