From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:60118) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TkdxG-0004D4-3x for qemu-devel@nongnu.org; Mon, 17 Dec 2012 11:54:15 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1TkdxE-0007KR-P6 for qemu-devel@nongnu.org; Mon, 17 Dec 2012 11:54:14 -0500 Received: from mx1.redhat.com ([209.132.183.28]:32458) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TkdxE-0007KL-Gq for qemu-devel@nongnu.org; Mon, 17 Dec 2012 11:54:12 -0500 Message-ID: <50CF4E2E.90605@redhat.com> Date: Mon, 17 Dec 2012 17:54:06 +0100 From: Paolo Bonzini MIME-Version: 1.0 References: <1355761490-10073-1-git-send-email-pbonzini@redhat.com> <1355761490-10073-14-git-send-email-pbonzini@redhat.com> <20121217164838.GA29085@redhat.com> In-Reply-To: <20121217164838.GA29085@redhat.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 13/15] virtio-pci: reset all qbuses too when writing to the status field List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Michael S. Tsirkin" Cc: aliguori@us.ibm.com, qemu-devel@nongnu.org Il 17/12/2012 17:48, Michael S. Tsirkin ha scritto: > On Mon, Dec 17, 2012 at 05:24:48PM +0100, Paolo Bonzini wrote: >> virtio-pci devices do not perform a full reset when zero is written >> to the status byte. While PCI-specific status is initialized, the >> reset does not propagate down the qdev bus hierarchy. Because of >> this, a virtio reset does not cancel in-flight I/O for virtio-scsi >> (where the cancellation is handled automatically by the SCSI >> devices underneath virtio-scsi-pci). >> >> Reported-by: Bryan Venteicher >> Cc: Michael S. Tsirkin >> Signed-off-by: Paolo Bonzini > > I would still prefer this logic to reside in virtio.c instead of being > duplicated in each bus. > My idea was to simply call qdev_reset_all on the binding from virtio.c > but other ideas wellcome. I think you're confusing "in the common superclass of all virtio transports" vs "in the common superclass of all virtio devices". virtio.c only implements a common superclass of all virtio devices; in fact, there is no common superclass of all virtio transports, and it is not possible without multiple inheritance or stuff like traits (you're already inheriting from PCIDevice for virtio-*-pci). Such common superclass, if it existed, would abstract stuff like "write zero to the status register" and would call qdev_reset_all. But again, we don't have this concept. Paolo > > >> --- >> hw/virtio-pci.c | 28 ++++++++++------------------ >> 1 file changed, 10 insertions(+), 18 deletions(-) >> >> diff --git a/hw/virtio-pci.c b/hw/virtio-pci.c >> index a7c75fe..2cf5282 100644 >> --- a/hw/virtio-pci.c >> +++ b/hw/virtio-pci.c >> @@ -268,12 +268,10 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val) >> case VIRTIO_PCI_QUEUE_PFN: >> pa = (hwaddr)val << VIRTIO_PCI_QUEUE_ADDR_SHIFT; >> if (pa == 0) { >> - virtio_reset(proxy->vdev); >> - virtio_pci_stop_ioeventfd(proxy); >> - msix_unuse_all_vectors(&proxy->pci_dev); >> - } >> - else >> + qdev_reset_all(&proxy->pci_dev.qdev); >> + } else { >> virtio_queue_set_addr(vdev, vdev->queue_sel, pa); >> + } >> break; >> case VIRTIO_PCI_QUEUE_SEL: >> if (val < VIRTIO_PCI_QUEUE_MAX) >> @@ -285,22 +283,16 @@ static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val) >> } >> break; >> case VIRTIO_PCI_STATUS: >> - if (vdev->status == 0) { >> - virtio_reset(proxy->vdev); >> - } >> - >> - if (!(val & VIRTIO_CONFIG_S_DRIVER_OK)) { >> - virtio_pci_stop_ioeventfd(proxy); >> - } >> - >> virtio_set_status(vdev, val & 0xFF); >> >> - if (val & VIRTIO_CONFIG_S_DRIVER_OK) { >> - virtio_pci_start_ioeventfd(proxy); >> - } >> - >> if (vdev->status == 0) { >> - msix_unuse_all_vectors(&proxy->pci_dev); >> + qdev_reset_all(&proxy->pci_dev.qdev); >> + } else { >> + if (!(val & VIRTIO_CONFIG_S_DRIVER_OK)) { >> + virtio_pci_stop_ioeventfd(proxy); >> + } else { >> + virtio_pci_start_ioeventfd(proxy); >> + } >> } >> >> /* Linux before 2.6.34 sets the device as OK without enabling >> -- >> 1.8.0.2 >>