hypervisor: add hot plugging support for Qemu Q35

Qemu Q35 machine type does not support to hot plug devices
directly on pcie.0, hence we have to find a way to allow users
hot plug N devices.
The only way to hot plug devices in Qemu Q35 is through PCI brdiges.
Each PCI bridge is able to support until 32 devices, therefore we have a
limitation in the amount of devices we can hot plug.
This patch adds support for hot plugging using PCI bridges in
pc and q35 machine types.

Signed-off-by: Julio Montes <julio.montes@intel.com>

bridge.go

@@ -0,0 +1,104 @@
+//
+// Copyright (c) 2017 Intel Corporation
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//      http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+//
+
+package virtcontainers
+
+import "fmt"
+
+type bridgeType string
+
+const (
+	pciBridge  bridgeType = "pci"
+	pcieBridge            = "pcie"
+)
+
+const pciBridgeMaxCapacity = 30
+
+// Bridge is a bridge where devices can be hot plugged
+type Bridge struct {
+	// Address contains information about devices plugged and its address in the bridge
+	Address map[uint32]string
+
+	// Type is the type of the bridge (pci, pcie, etc)
+	Type bridgeType
+
+	//ID is used to identify the bridge in the hypervisor
+	ID string
+}
+
+// NewBridges creates n new pci(e) bridges depending of the machine type
+func NewBridges(n uint32, machine string) []Bridge {
+	var bridges []Bridge
+	var bt bridgeType
+
+	switch machine {
+	case QemuQ35:
+		// currently only pci bridges are supported
+		// qemu-2.10 will introduce pcie bridges
+		fallthrough
+	case QemuPC:
+		bt = pciBridge
+	default:
+		return nil
+	}
+
+	for i := uint32(0); i < n; i++ {
+		bridges = append(bridges, Bridge{
+			Type:    bt,
+			ID:      fmt.Sprintf("%s-bridge-%d", bt, i),
+			Address: make(map[uint32]string),
+		})
+	}
+
+	return bridges
+}
+
+// addDevice on success adds the device ID to the bridge and return the address
+// where the device was added, otherwise an error is returned
+func (b *Bridge) addDevice(ID string) (uint32, error) {
+	var addr uint32
+
+	// looking for the first available address
+	for i := uint32(1); i <= pciBridgeMaxCapacity; i++ {
+		if _, ok := b.Address[i]; !ok {
+			addr = i
+			break
+		}
+	}
+
+	if addr == 0 {
+		return 0, fmt.Errorf("Unable to hot plug device on bridge: there are not empty slots")
+	}
+
+	// save address and device
+	b.Address[addr] = ID
+	return addr, nil
+}
+
+// removeDevice on success removes the device ID from the bridge and return nil,
+// otherwise an error is returned
+func (b *Bridge) removeDevice(ID string) error {
+	// check if the device was hot plugged in the bridge
+	for addr, devID := range b.Address {
+		if devID == ID {
+			// free address to re-use the same slot with other devices
+			delete(b.Address, addr)
+			return nil
+		}
+	}
+
+	return fmt.Errorf("Unable to hot unplug device %s: not present on bridge", ID)
+}

container.go

@@ -870,7 +870,7 @@ func (c *Container) attachDevices() error {
 
 func (c *Container) detachDevices() error {
 	for _, device := range c.devices {
-		if err := device.detach(c.pod.hypervisor); err != nil {
+		if err := device.detach(c.pod.hypervisor, c); err != nil {
 			return err
 		}
 	}

device.go

@@ -60,7 +60,7 @@ const (
 // Device is the virtcontainers device interface.
 type Device interface {
 	attach(hypervisor, *Container) error
-	detach(hypervisor) error
+	detach(hypervisor, *Container) error
 	deviceType() string
 }
 
@@ -153,7 +153,7 @@ func (device *VFIODevice) attach(h hypervisor, c *Container) error {
 	return nil
 }
 
-func (device *VFIODevice) detach(h hypervisor) error {
+func (device *VFIODevice) detach(h hypervisor, c *Container) error {
 	return nil
 }
 
@@ -232,7 +232,7 @@ func (device *BlockDevice) attach(h hypervisor, c *Container) (err error) {
 
 		device.DeviceInfo.Hotplugged = false
 	} else {
-		if err = h.hotplugAddDevice(drive, blockDev); err != nil {
+		if err := c.pod.hypervisor.hotplugAddDevice(drive, blockDev); err != nil {
 			return err
 		}
 
@@ -243,19 +243,17 @@ func (device *BlockDevice) attach(h hypervisor, c *Container) (err error) {
 	return nil
 }
 
-func (device BlockDevice) detach(h hypervisor) error {
+func (device BlockDevice) detach(h hypervisor, c *Container) error {
 	if device.DeviceInfo.Hotplugged {
 		deviceLogger().WithField("device", device.DeviceInfo.HostPath).Info("Unplugging block device")
 
 		drive := Drive{
 			ID: makeBlockDevIDForHypervisor(device.DeviceInfo.ID),
 		}
 
-		if err := h.hotplugRemoveDevice(drive, blockDev); err != nil {
-			deviceLogger().WithError(err).Error("Failed to unplug block device")
+		if err := c.pod.hypervisor.hotplugRemoveDevice(drive, blockDev); err != nil {
 			return err
 		}
-
 	}
 	return nil
 }
@@ -281,7 +279,7 @@ func (device *GenericDevice) attach(h hypervisor, c *Container) error {
 	return nil
 }
 
-func (device *GenericDevice) detach(h hypervisor) error {
+func (device *GenericDevice) detach(h hypervisor, c *Container) error {
 	return nil
 }
 

filesystem.go

@@ -32,6 +32,8 @@ import (
 // pods and containers.
 type podResource int
 
+type HypervisorState []byte
+
 const (
 	// configFileType represents a configuration file type
 	configFileType podResource = iota
@@ -42,6 +44,9 @@ const (
 	// networkFileType represents a network file type (pod only)
 	networkFileType
 
+	// hypervisorFileType represents a hypervisor file type (pod only)
+	hypervisorFileType
+
 	// processFileType represents a process file type
 	processFileType
 
@@ -75,6 +80,9 @@ const mountsFile = "mounts.json"
 // devicesFile is the file name storing a container's devices.
 const devicesFile = "devices.json"
 
+// hypervisorFile is the file name storing a hypervisor's state.
+const hypervisorFile = "hypervisor.json"
+
 // dirMode is the permission bits used for creating a directory
 const dirMode = os.FileMode(0750)
 
@@ -109,6 +117,10 @@ type resourceStorage interface {
 	fetchPodNetwork(podID string) (NetworkNamespace, error)
 	storePodNetwork(podID string, networkNS NetworkNamespace) error
 
+	// Hypervisor resources
+	fetchHypervisorState(podID string, state interface{}) error
+	storeHypervisorState(podID string, state interface{}) error
+
 	// Container resources
 	storeContainerResource(podID, containerID string, resource podResource, data interface{}) error
 	deleteContainerResources(podID, containerID string, resources []podResource) error
@@ -260,6 +272,21 @@ func (fs *filesystem) fetchFile(file string, data interface{}) error {
 	return nil
 }
 
+func (fs *filesystem) fetchHypervisorFile(file string, state *HypervisorState) error {
+	if file == "" {
+		return errNeedFile
+	}
+
+	fileData, err := ioutil.ReadFile(file)
+	if err != nil {
+		return err
+	}
+
+	*state = fileData
+
+	return nil
+}
+
 // fetchDeviceFile is used for custom unmarshalling of device interface objects.
 func (fs *filesystem) fetchDeviceFile(file string, devices *[]Device) error {
 	if file == "" {
@@ -328,7 +355,7 @@ func (fs *filesystem) fetchDeviceFile(file string, devices *[]Device) error {
 func resourceNeedsContainerID(podSpecific bool, resource podResource) bool {
 
 	switch resource {
-	case lockFileType, networkFileType:
+	case lockFileType, networkFileType, hypervisorFileType:
 		// pod-specific resources
 		return false
 	default:
@@ -351,7 +378,7 @@ func resourceDir(podSpecific bool, podID, containerID string, resource podResour
 	case configFileType:
 		path = configStoragePath
 		break
-	case stateFileType, networkFileType, processFileType, lockFileType, mountsFileType, devicesFileType:
+	case stateFileType, networkFileType, processFileType, lockFileType, mountsFileType, devicesFileType, hypervisorFileType:
 		path = runStoragePath
 		break
 	default:
@@ -387,6 +414,9 @@ func (fs *filesystem) resourceURI(podSpecific bool, podID, containerID string, r
 		filename = stateFile
 	case networkFileType:
 		filename = networkFile
+	case hypervisorFileType:
+		filename = hypervisorFile
+		break
 	case processFileType:
 		filename = processFile
 	case lockFileType:
@@ -477,6 +507,20 @@ func (fs *filesystem) storeNetworkResource(podSpecific bool, podID, containerID
 	return fs.storeFile(networkFile, file)
 }
 
+func (fs *filesystem) storeHypervisorResource(podSpecific bool, podID, containerID string, resource podResource, file interface{}) error {
+	if resource != hypervisorFileType {
+		return errInvalidResource
+	}
+
+	// pod only resource
+	hypervisorFile, _, err := fs.resourceURI(true, podID, containerID, hypervisorFileType)
+	if err != nil {
+		return err
+	}
+
+	return fs.storeFile(hypervisorFile, file)
+}
+
 func (fs *filesystem) storeProcessResource(podSpecific bool, podID, containerID string, resource podResource, file interface{}) error {
 	if resource != processFileType {
 		return errInvalidResource
@@ -599,6 +643,15 @@ func (fs *filesystem) doFetchResource(containerID, path string, resource podReso
 
 		return networkNS, nil
 
+	case hypervisorFileType:
+		var hypervisorSt HypervisorState
+		err = fs.fetchHypervisorFile(path, &hypervisorSt)
+		if err != nil {
+			return nil, err
+		}
+
+		return hypervisorSt, nil
+
 	case processFileType:
 		process := Process{}
 		err = fs.fetchFile(path, &process)
@@ -675,10 +728,31 @@ func (fs *filesystem) fetchPodNetwork(podID string) (NetworkNamespace, error) {
 	return NetworkNamespace{}, fmt.Errorf("Unknown network type")
 }
 
+func (fs *filesystem) fetchHypervisorState(podID string, state interface{}) error {
+	data, err := fs.fetchResource(true, podID, "", hypervisorFileType)
+	if err != nil {
+		return err
+	}
+
+	switch st := data.(type) {
+	case HypervisorState:
+		if err = json.Unmarshal(st, state); err != nil {
+			return err
+		}
+		return nil
+	}
+
+	return fmt.Errorf("Unknown hypervisor type")
+}
+
 func (fs *filesystem) storePodNetwork(podID string, networkNS NetworkNamespace) error {
 	return fs.storePodResource(podID, networkFileType, networkNS)
 }
 
+func (fs *filesystem) storeHypervisorState(podID string, state interface{}) error {
+	return fs.storeHypervisorResource(true, podID, "", hypervisorFileType, state)
+}
+
 func (fs *filesystem) deletePodResources(podID string, resources []podResource) error {
 	if resources == nil {
 		resources = []podResource{configFileType, stateFileType}

hypervisor.go

@@ -161,6 +161,10 @@ type HypervisorConfig struct {
 	// Pod configuration VMConfig.Memory overwrites this.
 	DefaultMemSz uint32
 
+	// DefaultBridges specifies default number of bridges for the VM.
+	// Bridges can be used to hot plug devices
+	DefaultBridges uint32
+
 	// MemPrealloc specifies if the memory should be pre-allocated
 	MemPrealloc bool
 
@@ -335,7 +339,7 @@ func RunningOnVMM(cpuInfoPath string) (bool, error) {
 // hypervisor is the virtcontainers hypervisor interface.
 // The default hypervisor implementation is Qemu.
 type hypervisor interface {
-	init(config HypervisorConfig) error
+	init(pod *Pod) error
 	createPod(podConfig PodConfig) error
 	startPod(startCh, stopCh chan struct{}) error
 	stopPod() error
@@ -346,4 +350,5 @@ type hypervisor interface {
 	hotplugRemoveDevice(devInfo interface{}, devType deviceType) error
 	getPodConsole(podID string) string
 	capabilities() capabilities
+	getState() interface{}
 }

mock_hypervisor.go

@@ -19,8 +19,8 @@ package virtcontainers
 type mockHypervisor struct {
 }
 
-func (m *mockHypervisor) init(config HypervisorConfig) error {
-	valid, err := config.valid()
+func (m *mockHypervisor) init(pod *Pod) error {
+	valid, err := pod.config.HypervisorConfig.valid()
 	if valid == false || err != nil {
 		return err
 	}
@@ -69,3 +69,7 @@ func (m *mockHypervisor) hotplugRemoveDevice(devInfo interface{}, devType device
 func (m *mockHypervisor) getPodConsole(podID string) string {
 	return ""
 }
+
+func (m *mockHypervisor) getState() interface{} {
+	return nil
+}