From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH v4] KVM: Allow host IRQ sharing for assigned PCI 2.3 devices Date: Tue, 06 Mar 2012 17:53:21 +0200 Message-ID: <4F5632F1.6000408@redhat.com> References: <4F4CD47A.3080708@siemens.com> <4F562E69.4040907@redhat.com> <4F563042.1020703@siemens.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Marcelo Tosatti , kvm , Alex Williamson , "Michael S. Tsirkin" To: Jan Kiszka Return-path: Received: from mx1.redhat.com ([209.132.183.28]:50417 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752344Ab2CFPx0 (ORCPT ); Tue, 6 Mar 2012 10:53:26 -0500 In-Reply-To: <4F563042.1020703@siemens.com> Sender: kvm-owner@vger.kernel.org List-ID: On 03/06/2012 05:41 PM, Jan Kiszka wrote: > On 2012-03-06 16:34, Avi Kivity wrote: > > On 02/28/2012 03:19 PM, Jan Kiszka wrote: > >> PCI 2.3 allows to generically disable IRQ sources at device level. This > >> enables us to share legacy IRQs of such devices with other host devices > >> when passing them to a guest. > >> > >> The new IRQ sharing feature introduced here is optional, user space has > >> to request it explicitly. Moreover, user space can inform us about its > >> view of PCI_COMMAND_INTX_DISABLE so that we can avoid unmasking the > >> interrupt and signaling it if the guest masked it via the virtualized > >> PCI config space. > > > > Long delay, sorry. > > > > I'm sure we discussed this before, so a URL would be sufficient: why > > cannot this be transparent to userspace? > > Yes, we did, and you may recall I tried hard to implement it. The > reasons for not following this path were given in one of the previous > postings: > > "To recall the history of it: I tried hard to implement an adaptive > solution that automatically picks the fastest masking technique whenever > possible. However, the changes required to the IRQ core subsystem and > the logic of the device assignment code became so complex and partly > ugly that I gave up on this. It's simply not worth the pain given that > legacy PCI interrupts are rarely raised for performance critical device > at such a high rate (KHz...) that you can measure the difference." Ok. Given the paragraph below I won't try to second-guess you. > > > > As for the actual patch, I am so unfamiliar with the device assignment > > code now that I'll have to rely on Alex's review. > > > > Then let's hope he didn't miss any of my bugs. > > Jan > -- error compiling committee.c: too many arguments to function