podman cgroup enhancement

currently, setting any sort of resource limit in a pod does nothing. With the newly refactored creation process in c/common, podman ca now set resources at a pod level
meaning that resource related flags can now be exposed to podman pod create.

cgroupfs and systemd are both supported with varying completion. cgroupfs is a much simpler process and one that is virtually complete for all resource types, the flags now just need to be added. systemd on the other hand
has to be handeled via the dbus api meaning that the limits need to be passed as recognized properties to systemd. The properties added so far are the ones that podman pod create supports as well as `cpuset-mems` as this will
be the next flag I work on.

Signed-off-by: Charlie Doern <cdoern@redhat.com>

libpod/oci_conmon_linux.go

@@ -23,6 +23,9 @@ import (
 	"text/template"
 	"time"
 
+	runcconfig "github.com/opencontainers/runc/libcontainer/configs"
+	"github.com/opencontainers/runc/libcontainer/devices"
+
 	"github.com/containers/common/pkg/cgroups"
 	"github.com/containers/common/pkg/config"
 	conmonConfig "github.com/containers/conmon/runner/config"
@@ -1451,9 +1454,14 @@ func (r *ConmonOCIRuntime) moveConmonToCgroupAndSignal(ctr *Container, cmd *exec
 		// TODO: This should be a switch - we are not guaranteed that
 		// there are only 2 valid cgroup managers
 		cgroupParent := ctr.CgroupParent()
+		cgroupPath := filepath.Join(ctr.config.CgroupParent, "conmon")
+		Resource := ctr.Spec().Linux.Resources
+		cgroupResources, err := GetLimits(Resource)
+		if err != nil {
+			logrus.StandardLogger().Log(logLevel, "Could not get ctr resources")
+		}
 		if ctr.CgroupManager() == config.SystemdCgroupsManager {
 			unitName := createUnitName("libpod-conmon", ctr.ID())
-
 			realCgroupParent := cgroupParent
 			splitParent := strings.Split(cgroupParent, "/")
 			if strings.HasSuffix(cgroupParent, ".slice") && len(splitParent) > 1 {
@@ -1465,8 +1473,7 @@ func (r *ConmonOCIRuntime) moveConmonToCgroupAndSignal(ctr *Container, cmd *exec
 				logrus.StandardLogger().Logf(logLevel, "Failed to add conmon to systemd sandbox cgroup: %v", err)
 			}
 		} else {
-			cgroupPath := filepath.Join(ctr.config.CgroupParent, "conmon")
-			control, err := cgroups.New(cgroupPath, &spec.LinuxResources{})
+			control, err := cgroups.New(cgroupPath, &cgroupResources)
 			if err != nil {
 				logrus.StandardLogger().Logf(logLevel, "Failed to add conmon to cgroupfs sandbox cgroup: %v", err)
 			} else if err := control.AddPid(cmd.Process.Pid); err != nil {
@@ -1748,3 +1755,191 @@ func httpAttachNonTerminalCopy(container *net.UnixConn, http *bufio.ReadWriter,
 		}
 	}
 }
+
+// GetLimits converts spec resource limits to cgroup consumable limits
+func GetLimits(resource *spec.LinuxResources) (runcconfig.Resources, error) {
+	if resource == nil {
+		resource = &spec.LinuxResources{}
+	}
+	final := &runcconfig.Resources{}
+	devs := []*devices.Rule{}
+
+	// Devices
+	for _, entry := range resource.Devices {
+		if entry.Major == nil || entry.Minor == nil {
+			continue
+		}
+		runeType := 'a'
+		switch entry.Type {
+		case "b":
+			runeType = 'b'
+		case "c":
+			runeType = 'c'
+		}
+
+		devs = append(devs, &devices.Rule{
+			Type:        devices.Type(runeType),
+			Major:       *entry.Major,
+			Minor:       *entry.Minor,
+			Permissions: devices.Permissions(entry.Access),
+			Allow:       entry.Allow,
+		})
+	}
+	final.Devices = devs
+
+	// HugepageLimits
+	pageLimits := []*runcconfig.HugepageLimit{}
+	for _, entry := range resource.HugepageLimits {
+		pageLimits = append(pageLimits, &runcconfig.HugepageLimit{
+			Pagesize: entry.Pagesize,
+			Limit:    entry.Limit,
+		})
+	}
+	final.HugetlbLimit = pageLimits
+
+	// Networking
+	netPriorities := []*runcconfig.IfPrioMap{}
+	if resource.Network != nil {
+		for _, entry := range resource.Network.Priorities {
+			netPriorities = append(netPriorities, &runcconfig.IfPrioMap{
+				Interface: entry.Name,
+				Priority:  int64(entry.Priority),
+			})
+		}
+	}
+	final.NetPrioIfpriomap = netPriorities
+	rdma := make(map[string]runcconfig.LinuxRdma)
+	for name, entry := range resource.Rdma {
+		rdma[name] = runcconfig.LinuxRdma{HcaHandles: entry.HcaHandles, HcaObjects: entry.HcaObjects}
+	}
+	final.Rdma = rdma
+
+	// Memory
+	if resource.Memory != nil {
+		if resource.Memory.Limit != nil {
+			final.Memory = *resource.Memory.Limit
+		}
+		if resource.Memory.Reservation != nil {
+			final.MemoryReservation = *resource.Memory.Reservation
+		}
+		if resource.Memory.Swap != nil {
+			final.MemorySwap = *resource.Memory.Swap
+		}
+		if resource.Memory.Swappiness != nil {
+			final.MemorySwappiness = resource.Memory.Swappiness
+		}
+	}
+
+	// CPU
+	if resource.CPU != nil {
+		if resource.CPU.Period != nil {
+			final.CpuPeriod = *resource.CPU.Period
+		}
+		if resource.CPU.Quota != nil {
+			final.CpuQuota = *resource.CPU.Quota
+		}
+		if resource.CPU.RealtimePeriod != nil {
+			final.CpuRtPeriod = *resource.CPU.RealtimePeriod
+		}
+		if resource.CPU.RealtimeRuntime != nil {
+			final.CpuRtRuntime = *resource.CPU.RealtimeRuntime
+		}
+		if resource.CPU.Shares != nil {
+			final.CpuShares = *resource.CPU.Shares
+		}
+		final.CpusetCpus = resource.CPU.Cpus
+		final.CpusetMems = resource.CPU.Mems
+	}
+
+	// BlkIO
+	if resource.BlockIO != nil {
+		if len(resource.BlockIO.ThrottleReadBpsDevice) > 0 {
+			for _, entry := range resource.BlockIO.ThrottleReadBpsDevice {
+				throttle := &runcconfig.ThrottleDevice{}
+				dev := &runcconfig.BlockIODevice{
+					Major: entry.Major,
+					Minor: entry.Minor,
+				}
+				throttle.BlockIODevice = *dev
+				throttle.Rate = entry.Rate
+				final.BlkioThrottleReadBpsDevice = append(final.BlkioThrottleReadBpsDevice, throttle)
+			}
+		}
+		if len(resource.BlockIO.ThrottleWriteBpsDevice) > 0 {
+			for _, entry := range resource.BlockIO.ThrottleWriteBpsDevice {
+				throttle := &runcconfig.ThrottleDevice{}
+				dev := &runcconfig.BlockIODevice{
+					Major: entry.Major,
+					Minor: entry.Minor,
+				}
+				throttle.BlockIODevice = *dev
+				throttle.Rate = entry.Rate
+				final.BlkioThrottleWriteBpsDevice = append(final.BlkioThrottleWriteBpsDevice, throttle)
+			}
+		}
+		if len(resource.BlockIO.ThrottleReadIOPSDevice) > 0 {
+			for _, entry := range resource.BlockIO.ThrottleReadIOPSDevice {
+				throttle := &runcconfig.ThrottleDevice{}
+				dev := &runcconfig.BlockIODevice{
+					Major: entry.Major,
+					Minor: entry.Minor,
+				}
+				throttle.BlockIODevice = *dev
+				throttle.Rate = entry.Rate
+				final.BlkioThrottleReadIOPSDevice = append(final.BlkioThrottleReadIOPSDevice, throttle)
+			}
+		}
+		if len(resource.BlockIO.ThrottleWriteIOPSDevice) > 0 {
+			for _, entry := range resource.BlockIO.ThrottleWriteIOPSDevice {
+				throttle := &runcconfig.ThrottleDevice{}
+				dev := &runcconfig.BlockIODevice{
+					Major: entry.Major,
+					Minor: entry.Minor,
+				}
+				throttle.BlockIODevice = *dev
+				throttle.Rate = entry.Rate
+				final.BlkioThrottleWriteIOPSDevice = append(final.BlkioThrottleWriteIOPSDevice, throttle)
+			}
+		}
+		if resource.BlockIO.LeafWeight != nil {
+			final.BlkioLeafWeight = *resource.BlockIO.LeafWeight
+		}
+		if resource.BlockIO.Weight != nil {
+			final.BlkioWeight = *resource.BlockIO.Weight
+		}
+		if len(resource.BlockIO.WeightDevice) > 0 {
+			for _, entry := range resource.BlockIO.WeightDevice {
+				weight := &runcconfig.WeightDevice{}
+				dev := &runcconfig.BlockIODevice{
+					Major: entry.Major,
+					Minor: entry.Minor,
+				}
+				if entry.Weight != nil {
+					weight.Weight = *entry.Weight
+				}
+				if entry.LeafWeight != nil {
+					weight.LeafWeight = *entry.LeafWeight
+				}
+				weight.BlockIODevice = *dev
+				final.BlkioWeightDevice = append(final.BlkioWeightDevice, weight)
+			}
+		}
+	}
+
+	// Pids
+	if resource.Pids != nil {
+		final.PidsLimit = resource.Pids.Limit
+	}
+
+	// Networking
+	if resource.Network != nil {
+		if resource.Network.ClassID != nil {
+			final.NetClsClassid = *resource.Network.ClassID
+		}
+	}
+
+	// Unified state
+	final.Unified = resource.Unified
+
+	return *final, nil
+}