From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [140.186.70.92] (port=45843 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1PBtKv-0006Yo-LZ for qemu-devel@nongnu.org; Fri, 29 Oct 2010 14:05:58 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1PBtKu-00010J-Cw for qemu-devel@nongnu.org; Fri, 29 Oct 2010 14:05:57 -0400 Received: from mx1.redhat.com ([209.132.183.28]:55412) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1PBtKu-00010A-5C for qemu-devel@nongnu.org; Fri, 29 Oct 2010 14:05:56 -0400 From: Alex Williamson Date: Fri, 29 Oct 2010 12:05:52 -0600 Message-ID: <20101029180511.26801.99926.stgit@s20.home> In-Reply-To: <20101029175434.25281.70647.stgit@s20.home> References: <20101029175434.25281.70647.stgit@s20.home> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Subject: [Qemu-devel] [RFC PATCH] kvm: KVM_EOI_EVENTFD support for eoi_client List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: kvm@vger.kernel.org, avi@redhat.com Cc: chrisw@redhat.com, mst@redhat.com, qemu-devel@nongnu.org, alex.williamson@redhat.com, ddutile@redhat.com With the KVM irqchip, we need to get the EOI via an eventfd. This adds support for that, abstracting the details to the caller. The get_fd function allows drivers to make further optimizations in handling the EOI. For instance with VFIO, we can make use of an irqfd-like mechanism to have the VFIO kernel module consume the EOI directly, bypassing qemu userspace. Signed-off-by: Alex Williamson --- hw/ioapic.c | 52 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 52 insertions(+), 0 deletions(-) diff --git a/hw/ioapic.c b/hw/ioapic.c index c43be3a..707f2a2 100644 --- a/hw/ioapic.c +++ b/hw/ioapic.c @@ -72,14 +72,66 @@ static QLIST_HEAD(ioapic_eoi_client_list, ioapic_eoi_client) ioapic_eoi_client_list = QLIST_HEAD_INITIALIZER(ioapic_eoi_client_list); +#ifdef KVM_EOI_EVENTFD +static void ioapic_eoi_callback(void *opaque) +{ + ioapic_eoi_client *client = opaque; + + if (!event_notifier_test_and_clear(&client->notifier)) { + return; + } + + client->eoi(client); +} +#endif + int ioapic_register_eoi_client(ioapic_eoi_client *client) { QLIST_INSERT_HEAD(&ioapic_eoi_client_list, client, list); + +#ifdef KVM_EOI_EVENTFD + if (kvm_enabled() && kvm_irqchip_in_kernel()) { + int ret, fd; + + ret = event_notifier_init(&client->notifier, 0); + if (ret) { + fprintf(stderr, "%s notifier init failed %d\n", __FUNCTION__, ret); + return ret; + } + + fd = event_notifier_get_fd(&client->notifier); + qemu_set_fd_handler(fd, ioapic_eoi_callback, NULL, client); + + ret = kvm_eoi_eventfd(client->irq, fd, KVM_EOI_EVENTFD_FLAG_DEASSERT); + if (ret) { + fprintf(stderr, "%s eoi eventfd failed %d\n", __FUNCTION__, ret); + return ret; + } + client->notifier_enabled = true; + } +#endif return 0; } void ioapic_unregister_eoi_client(ioapic_eoi_client *client) { +#ifdef KVM_EOI_EVENTFD + if (kvm_enabled() && kvm_irqchip_in_kernel()) { + int ret, fd; + + fd = event_notifier_get_fd(&client->notifier); + + ret = kvm_eoi_eventfd(client->irq, fd, KVM_EOI_EVENTFD_FLAG_DEASSIGN); + if (ret) { + fprintf(stderr, "%s eoi eventfd failed %d\n", __FUNCTION__, ret); + } + + qemu_set_fd_handler(fd, NULL, NULL, NULL); + + event_notifier_cleanup(&client->notifier); + client->notifier_enabled = false; + } +#endif QLIST_REMOVE(client, list); }