From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:47372) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Yxiwr-0006gy-C9 for qemu-devel@nongnu.org; Wed, 27 May 2015 17:33:14 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Yxiwn-0002Va-AB for qemu-devel@nongnu.org; Wed, 27 May 2015 17:33:13 -0400 Received: from mx1.redhat.com ([209.132.183.28]:57981) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Yxiwn-0002VW-3r for qemu-devel@nongnu.org; Wed, 27 May 2015 17:33:09 -0400 Message-ID: <1432762387.24271.96.camel@redhat.com> From: Alex Williamson Date: Wed, 27 May 2015 15:33:07 -0600 In-Reply-To: <46619f0d035b8cb160694963cb8d4c249e744ee3.1432694455.git.chen.fan.fnst@cn.fujitsu.com> References: <46619f0d035b8cb160694963cb8d4c249e744ee3.1432694455.git.chen.fan.fnst@cn.fujitsu.com> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [RFC v8.1 10/13] vfio: do hot bus reset when do virtual secondary bus reset List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Chen Fan Cc: izumi.taku@jp.fujitsu.com, qemu-devel@nongnu.org On Wed, 2015-05-27 at 10:46 +0800, Chen Fan wrote: > when do virtual secondary bus reset, the vfio device under > this bus need to do host bus reset to reset the device. > so add this case. > > Signed-off-by: Chen Fan > --- > hw/vfio/pci.c | 75 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 75 insertions(+) > > diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c > index a8c5988..b05ccdf 100644 > --- a/hw/vfio/pci.c > +++ b/hw/vfio/pci.c > @@ -3803,6 +3803,64 @@ out: > g_free(info); > } > > +/* Check when device reset the affected devices is single or multi, > + * because there was differentiate the hot reset of mulitple in-use > + * devices and hot reset of a single in-use device. */ > +static int vfio_pci_affect_devices_is_multi(VFIOPCIDevice *vdev) > +{ > + VFIOGroup *group; > + struct vfio_pci_hot_reset_info *info = NULL; > + struct vfio_pci_dependent_device *devices; > + int ret, i, count = 0; > + > + ret = vfio_get_hot_reset_info(vdev, &info); > + if (ret < 0) { > + goto out; > + } > + > + devices = &info->devices[0]; > + > + /* Verify that we have all the groups required */ > + for (i = 0; i < info->count; i++) { > + PCIHostDeviceAddress host; > + VFIODevice *vbasedev_iter; > + > + host.domain = devices[i].segment; > + host.bus = devices[i].bus; > + host.slot = PCI_SLOT(devices[i].devfn); > + host.function = PCI_FUNC(devices[i].devfn); > + > + /* Skip the current device */ > + if (vfio_pci_host_match(&host, &vdev->host)) { > + continue; > + } > + > + /* Ensure we own the group of the affected device */ > + QLIST_FOREACH(group, &vfio_group_list, next) { > + if (group->groupid == devices[i].group_id) { > + break; > + } > + } > + > + if (!group) { > + ret = -1; > + goto out; > + } > + > + QLIST_FOREACH(vbasedev_iter, &group->device_list, next) { > + if (vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) { > + continue; > + } > + count++; > + } > + } > + > + ret = count; > +out: > + g_free(info); > + return ret; > +} This basically does almost all the work of the hot reset function just to figure out whether it's a single or multi-reset to call the hot reset function with the right parameter. Why not just do a hot reset? > + > static void vfio_pci_machine_done_notify(Notifier *notifier, void *unused) > { > VFIOGroup *group; > @@ -4039,6 +4097,23 @@ static void vfio_pci_reset(DeviceState *dev) > > trace_vfio_pci_reset(vdev->vbasedev.name); > > + if (vdev->needs_bus_reset) { > + vdev->needs_bus_reset = false; A comment would be nice to signal that this secondary test avoids duplicate bus resets. > + if (vdev->vbasedev.needs_reset) { > + int ret; > + > + ret = vfio_pci_affect_devices_is_multi(vdev); > + if (ret < 0) { > + error_report("vfio: Cannot reset device %04x:%02x:%02x.%x, " > + "get hot reset info failed.", vdev->host.domain, > + vdev->host.bus, vdev->host.slot, vdev->host.function); > + } else { > + vfio_pci_hot_reset(vdev, ret ? true : false); > + } > + } > + return; > + } > + > vfio_pci_pre_reset(vdev); > > if (vdev->resetfn && !vdev->resetfn(vdev)) {