From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH v9 1/2] kvm: Use a reserved IRQ source ID for irqfd Date: Wed, 05 Sep 2012 18:35:24 +0300 Message-ID: <5047713C.3080602@redhat.com> References: <20120821195813.GD9027@redhat.com> <1345579579.29292.67.camel@ul30vt.home> <20120821204125.GF9027@redhat.com> <1345583694.29292.91.camel@ul30vt.home> <20120822004138.GM9027@redhat.com> <5047633F.9020405@redhat.com> <20120905145153.GA11048@redhat.com> <504768E2.3020706@redhat.com> <20120905151348.GC11058@redhat.com> <50476E51.5020005@redhat.com> <20120905152832.GE11058@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Alex Williamson , gleb@redhat.com, kvm@vger.kernel.org, linux-kernel@vger.kernel.org To: "Michael S. Tsirkin" Return-path: In-Reply-To: <20120905152832.GE11058@redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 09/05/2012 06:28 PM, Michael S. Tsirkin wrote: >> gsi 3 is configured as edge in the ioapic. It has (unusually) two >> inputs: one driven by userspace, the other by irqfd. >> >> cpu 0 cpu 1 >> ------------------------ ------------------------- >> irqfd: set to 1 >> ioapic: recognize edge >> inject irq >> EOI >> KVM_IRQ_LINE: set to 1 >> ioapic: ignore >> KVM_IRQ_LINE: set to 0 >> irqfd: set to 0 >> >> We had two edges with an EOI between them, but injected just on interrupt. > > I see. Makes sense, ACK this patch. Actually it's wrong. The two sources are not synchronized, so there is no way for them to know the two edges did not coalesce. On real hardware, after all, edge interrupts have a non-zero pulse width, and kvm faithfully emulates this. But this patch makes sense for level irqfd, so we might as well keep it with a different description. -- error compiling committee.c: too many arguments to function