This repository was archived by the owner on May 12, 2021. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 367
Expand file tree
/
Copy pathcontainer.go
More file actions
1602 lines (1333 loc) · 45.7 KB
/
container.go
File metadata and controls
1602 lines (1333 loc) · 45.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// +build linux
// Copyright (c) 2016 Intel Corporation
// Copyright (c) 2014,2015,2016,2017 Docker, Inc.
// SPDX-License-Identifier: Apache-2.0
//
package virtcontainers
import (
"context"
"encoding/hex"
"fmt"
"io"
"os"
"path/filepath"
"syscall"
"time"
"github.com/containerd/cgroups"
vccgroups "github.com/kata-containers/runtime/virtcontainers/pkg/cgroups"
vcTypes "github.com/kata-containers/runtime/virtcontainers/pkg/types"
"github.com/kata-containers/runtime/virtcontainers/types"
"github.com/kata-containers/runtime/virtcontainers/utils"
specs "github.com/opencontainers/runtime-spec/specs-go"
opentracing "github.com/opentracing/opentracing-go"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
"golang.org/x/sys/unix"
"github.com/kata-containers/runtime/virtcontainers/device/config"
"github.com/kata-containers/runtime/virtcontainers/device/manager"
"github.com/kata-containers/runtime/virtcontainers/pkg/rootless"
"github.com/kata-containers/runtime/virtcontainers/store"
)
// https://github.com/torvalds/linux/blob/master/include/uapi/linux/major.h
// This file has definitions for major device numbers.
var cdromMajors = map[int64]string{
11: "SCSI_CDROM_MAJOR",
15: "CDU31A_CDROM_MAJOR",
16: "GOLDSTAR_CDROM_MAJOR",
17: "OPTICS_CDROM_MAJOR",
18: "SANYO_CDROM_MAJOR",
20: "MITSUMI_X_CDROM_MAJOR",
23: "MITSUMI_CDROM_MAJOR",
24: "CDU535_CDROM_MAJOR",
25: "MATSUSHITA_CDROM_MAJOR",
26: "MATSUSHITA_CDROM2_MAJOR",
27: "MATSUSHITA_CDROM3_MAJOR",
28: "MATSUSHITA_CDROM4_MAJOR",
29: "AZTECH_CDROM_MAJOR",
32: "CM206_CDROM_MAJOR",
}
// https://github.com/torvalds/linux/blob/master/include/uapi/linux/major.h
// #define FLOPPY_MAJOR 2
const floppyMajor = int64(2)
// Process gathers data related to a container process.
type Process struct {
// Token is the process execution context ID. It must be
// unique per sandbox.
// Token is used to manipulate processes for containers
// that have not started yet, and later identify them
// uniquely within a sandbox.
Token string
// Pid is the process ID as seen by the host software
// stack, e.g. CRI-O, containerd. This is typically the
// shim PID.
Pid int
StartTime time.Time
}
// ContainerStatus describes a container status.
type ContainerStatus struct {
ID string
State types.ContainerState
PID int
StartTime time.Time
RootFs string
Spec *specs.Spec
// Annotations allow clients to store arbitrary values,
// for example to add additional status values required
// to support particular specifications.
Annotations map[string]string
}
// ThrottlingData gather the date related to container cpu throttling.
type ThrottlingData struct {
// Number of periods with throttling active
Periods uint64 `json:"periods,omitempty"`
// Number of periods when the container hit its throttling limit.
ThrottledPeriods uint64 `json:"throttled_periods,omitempty"`
// Aggregate time the container was throttled for in nanoseconds.
ThrottledTime uint64 `json:"throttled_time,omitempty"`
}
// CPUUsage denotes the usage of a CPU.
// All CPU stats are aggregate since container inception.
type CPUUsage struct {
// Total CPU time consumed.
// Units: nanoseconds.
TotalUsage uint64 `json:"total_usage,omitempty"`
// Total CPU time consumed per core.
// Units: nanoseconds.
PercpuUsage []uint64 `json:"percpu_usage,omitempty"`
// Time spent by tasks of the cgroup in kernel mode.
// Units: nanoseconds.
UsageInKernelmode uint64 `json:"usage_in_kernelmode"`
// Time spent by tasks of the cgroup in user mode.
// Units: nanoseconds.
UsageInUsermode uint64 `json:"usage_in_usermode"`
}
// CPUStats describes the cpu stats
type CPUStats struct {
CPUUsage CPUUsage `json:"cpu_usage,omitempty"`
ThrottlingData ThrottlingData `json:"throttling_data,omitempty"`
}
// MemoryData gather the data related to memory
type MemoryData struct {
Usage uint64 `json:"usage,omitempty"`
MaxUsage uint64 `json:"max_usage,omitempty"`
Failcnt uint64 `json:"failcnt"`
Limit uint64 `json:"limit"`
}
// MemoryStats describes the memory stats
type MemoryStats struct {
// memory used for cache
Cache uint64 `json:"cache,omitempty"`
// usage of memory
Usage MemoryData `json:"usage,omitempty"`
// usage of memory swap
SwapUsage MemoryData `json:"swap_usage,omitempty"`
// usage of kernel memory
KernelUsage MemoryData `json:"kernel_usage,omitempty"`
// usage of kernel TCP memory
KernelTCPUsage MemoryData `json:"kernel_tcp_usage,omitempty"`
// if true, memory usage is accounted for throughout a hierarchy of cgroups.
UseHierarchy bool `json:"use_hierarchy"`
Stats map[string]uint64 `json:"stats,omitempty"`
}
// PidsStats describes the pids stats
type PidsStats struct {
// number of pids in the cgroup
Current uint64 `json:"current,omitempty"`
// active pids hard limit
Limit uint64 `json:"limit,omitempty"`
}
// BlkioStatEntry gather date related to a block device
type BlkioStatEntry struct {
Major uint64 `json:"major,omitempty"`
Minor uint64 `json:"minor,omitempty"`
Op string `json:"op,omitempty"`
Value uint64 `json:"value,omitempty"`
}
// BlkioStats describes block io stats
type BlkioStats struct {
// number of bytes tranferred to and from the block device
IoServiceBytesRecursive []BlkioStatEntry `json:"io_service_bytes_recursive,omitempty"`
IoServicedRecursive []BlkioStatEntry `json:"io_serviced_recursive,omitempty"`
IoQueuedRecursive []BlkioStatEntry `json:"io_queue_recursive,omitempty"`
IoServiceTimeRecursive []BlkioStatEntry `json:"io_service_time_recursive,omitempty"`
IoWaitTimeRecursive []BlkioStatEntry `json:"io_wait_time_recursive,omitempty"`
IoMergedRecursive []BlkioStatEntry `json:"io_merged_recursive,omitempty"`
IoTimeRecursive []BlkioStatEntry `json:"io_time_recursive,omitempty"`
SectorsRecursive []BlkioStatEntry `json:"sectors_recursive,omitempty"`
}
// HugetlbStats describes hugetable memory stats
type HugetlbStats struct {
// current res_counter usage for hugetlb
Usage uint64 `json:"usage,omitempty"`
// maximum usage ever recorded.
MaxUsage uint64 `json:"max_usage,omitempty"`
// number of times hugetlb usage allocation failure.
Failcnt uint64 `json:"failcnt"`
}
// CgroupStats describes all cgroup subsystem stats
type CgroupStats struct {
CPUStats CPUStats `json:"cpu_stats,omitempty"`
MemoryStats MemoryStats `json:"memory_stats,omitempty"`
PidsStats PidsStats `json:"pids_stats,omitempty"`
BlkioStats BlkioStats `json:"blkio_stats,omitempty"`
// the map is in the format "size of hugepage: stats of the hugepage"
HugetlbStats map[string]HugetlbStats `json:"hugetlb_stats,omitempty"`
}
// NetworkStats describe all network stats.
type NetworkStats struct {
// Name is the name of the network interface.
Name string `json:"name,omitempty"`
RxBytes uint64 `json:"rx_bytes,omitempty"`
RxPackets uint64 `json:"rx_packets,omitempty"`
RxErrors uint64 `json:"rx_errors,omitempty"`
RxDropped uint64 `json:"rx_dropped,omitempty"`
TxBytes uint64 `json:"tx_bytes,omitempty"`
TxPackets uint64 `json:"tx_packets,omitempty"`
TxErrors uint64 `json:"tx_errors,omitempty"`
TxDropped uint64 `json:"tx_dropped,omitempty"`
}
// ContainerStats describes a container stats.
type ContainerStats struct {
CgroupStats *CgroupStats
NetworkStats []*NetworkStats
}
// ContainerResources describes container resources
type ContainerResources struct {
// VCPUs are the number of vCPUs that are being used by the container
VCPUs uint32
// Mem is the memory that is being used by the container
MemByte int64
}
// ContainerConfig describes one container runtime configuration.
type ContainerConfig struct {
ID string
// RootFs is the container workload image on the host.
RootFs RootFs
// ReadOnlyRootfs indicates if the rootfs should be mounted readonly
ReadonlyRootfs bool
// Cmd specifies the command to run on a container
Cmd types.Cmd
// Annotations allow clients to store arbitrary values,
// for example to add additional status values required
// to support particular specifications.
Annotations map[string]string
Mounts []Mount
// Device configuration for devices that must be available within the container.
DeviceInfos []config.DeviceInfo
// Resources container resources
Resources specs.LinuxResources
// Raw OCI specification, it won't be saved to disk.
CustomSpec *specs.Spec `json:"-"`
}
// valid checks that the container configuration is valid.
func (c *ContainerConfig) valid() bool {
if c == nil {
return false
}
if c.ID == "" {
return false
}
return true
}
// SystemMountsInfo describes additional information for system mounts that the agent
// needs to handle
type SystemMountsInfo struct {
// Indicates if /dev has been passed as a bind mount for the host /dev
BindMountDev bool
// Size of /dev/shm assigned on the host.
DevShmSize uint
}
// ContainerDevice describes a device associated with container
type ContainerDevice struct {
// ID is device id referencing the device from sandbox's device manager
ID string
// ContainerPath is device path displayed in container
ContainerPath string
// FileMode permission bits for the device.
FileMode os.FileMode
// UID is user ID in the container namespace
UID uint32
// GID is group ID in the container namespace
GID uint32
}
// RootFs describes the container's rootfs.
type RootFs struct {
// Source specifies the BlockDevice path
Source string
// Target specify where the rootfs is mounted if it has been mounted
Target string
// Type specifies the type of filesystem to mount.
Type string
// Options specifies zero or more fstab style mount options.
Options []string
// Mounted specifies whether the rootfs has be mounted or not
Mounted bool
}
// Container is composed of a set of containers and a runtime environment.
// A Container can be created, deleted, started, stopped, listed, entered, paused and restored.
type Container struct {
id string
sandboxID string
rootFs RootFs
config *ContainerConfig
sandbox *Sandbox
containerPath string
rootfsSuffix string
state types.ContainerState
process Process
mounts []Mount
devices []ContainerDevice
systemMountsInfo SystemMountsInfo
ctx context.Context
store *store.VCStore
}
// ID returns the container identifier string.
func (c *Container) ID() string {
return c.id
}
// Logger returns a logrus logger appropriate for logging Container messages
func (c *Container) Logger() *logrus.Entry {
return virtLog.WithFields(logrus.Fields{
"subsystem": "container",
"sandbox": c.sandboxID,
})
}
func (c *Container) trace(name string) (opentracing.Span, context.Context) {
if c.ctx == nil {
c.Logger().WithField("type", "bug").Error("trace called before context set")
c.ctx = context.Background()
}
span, ctx := opentracing.StartSpanFromContext(c.ctx, name)
span.SetTag("subsystem", "container")
return span, ctx
}
// Sandbox returns the sandbox handler related to this container.
func (c *Container) Sandbox() VCSandbox {
return c.sandbox
}
// Process returns the container process.
func (c *Container) Process() Process {
return c.process
}
// GetToken returns the token related to this container's process.
func (c *Container) GetToken() string {
return c.process.Token
}
// GetPid returns the pid related to this container's process.
func (c *Container) GetPid() int {
return c.process.Pid
}
func (c *Container) setStateFstype(fstype string) error {
c.state.Fstype = fstype
return nil
}
// GetAnnotations returns container's annotations
func (c *Container) GetAnnotations() map[string]string {
return c.config.Annotations
}
// GetPatchedOCISpec returns container's OCI specification
// This OCI specification was patched when the sandbox was created
// by containerCapabilities(), SetEphemeralStorageType() and others
// in order to support:
// * capabilities
// * Ephemeral storage
// * k8s empty dir
// If you need the original (vanilla) OCI spec,
// use compatoci.GetContainerSpec() instead.
func (c *Container) GetPatchedOCISpec() *specs.Spec {
return c.config.CustomSpec
}
// storeContainer stores a container config.
func (c *Container) storeContainer() error {
if err := c.sandbox.Save(); err != nil {
return err
}
return nil
}
// setContainerState sets both the in-memory and on-disk state of the
// container.
func (c *Container) setContainerState(state types.StateString) error {
if state == "" {
return vcTypes.ErrNeedState
}
c.Logger().Debugf("Setting container state from %v to %v", c.state.State, state)
// update in-memory state
c.state.State = state
if useOldStore(c.sandbox.ctx) {
// experimental runtime use "persist.json" which doesn't need "state.json" anymore
// update on-disk state
if err := c.store.Store(store.State, c.state); err != nil {
return err
}
} else {
// flush data to storage
if err := c.sandbox.Save(); err != nil {
return err
}
}
return nil
}
func (c *Container) shareFiles(m Mount, idx int, hostSharedDir, hostMountDir, guestSharedDir string) (string, bool, error) {
randBytes, err := utils.GenerateRandomBytes(8)
if err != nil {
return "", false, err
}
filename := fmt.Sprintf("%s-%s-%s", c.id, hex.EncodeToString(randBytes), filepath.Base(m.Destination))
guestDest := filepath.Join(guestSharedDir, filename)
// copy file to contaier's rootfs if filesystem sharing is not supported, otherwise
// bind mount it in the shared directory.
caps := c.sandbox.hypervisor.capabilities()
if !caps.IsFsSharingSupported() {
c.Logger().Debug("filesystem sharing is not supported, files will be copied")
fileInfo, err := os.Stat(m.Source)
if err != nil {
return "", false, err
}
// Ignore the mount if this is not a regular file (excludes
// directory, socket, device, ...) as it cannot be handled by
// a simple copy. But this should not be treated as an error,
// only as a limitation.
if !fileInfo.Mode().IsRegular() {
c.Logger().WithField("ignored-file", m.Source).Debug("Ignoring non-regular file as FS sharing not supported")
return "", true, nil
}
if err := c.sandbox.agent.copyFile(m.Source, guestDest); err != nil {
return "", false, err
}
} else {
// These mounts are created in the shared dir
mountDest := filepath.Join(hostMountDir, filename)
if err := bindMount(c.ctx, m.Source, mountDest, m.ReadOnly, "private"); err != nil {
return "", false, err
}
// Save HostPath mount value into the mount list of the container.
c.mounts[idx].HostPath = mountDest
// bindmount remount event is not propagated to mount subtrees, so we have to remount the shared dir mountpoint directly.
if m.ReadOnly {
mountDest = filepath.Join(hostSharedDir, filename)
if err := remountRo(c.ctx, mountDest); err != nil {
return "", false, err
}
}
}
return guestDest, false, nil
}
// mountSharedDirMounts handles bind-mounts by bindmounting to the host shared
// directory which is mounted through virtiofs/9pfs in the VM.
// It also updates the container mount list with the HostPath info, and store
// container mounts to the storage. This way, we will have the HostPath info
// available when we will need to unmount those mounts.
func (c *Container) mountSharedDirMounts(hostSharedDir, hostMountDir, guestSharedDir string) (sharedDirMounts map[string]Mount, ignoredMounts map[string]Mount, err error) {
sharedDirMounts = make(map[string]Mount)
ignoredMounts = make(map[string]Mount)
var devicesToDetach []string
defer func() {
if err != nil {
for _, id := range devicesToDetach {
c.sandbox.devManager.DetachDevice(id, c.sandbox)
}
}
}()
for idx, m := range c.mounts {
// Skip mounting certain system paths from the source on the host side
// into the container as it does not make sense to do so.
// Example sources could be /sys/fs/cgroup etc.
if isSystemMount(m.Source) {
continue
}
// Check if mount is a block device file. If it is, the block device will be attached to the host
// instead of passing this as a shared mount:
if len(m.BlockDeviceID) > 0 {
// Attach this block device, all other devices passed in the config have been attached at this point
if err = c.sandbox.devManager.AttachDevice(m.BlockDeviceID, c.sandbox); err != nil {
return nil, nil, err
}
devicesToDetach = append(devicesToDetach, m.BlockDeviceID)
continue
}
// For non-block based mounts, we are only interested in bind mounts
if m.Type != "bind" {
continue
}
// We need to treat /dev/shm as a special case. This is passed as a bind mount in the spec,
// but it does not make sense to pass this as a 9p mount from the host side.
// This needs to be handled purely in the guest, by allocating memory for this inside the VM.
if m.Destination == "/dev/shm" {
continue
}
// Ignore /dev, directories and all other device files. We handle
// only regular files in /dev. It does not make sense to pass the host
// device nodes to the guest.
if isHostDevice(m.Destination) {
continue
}
var ignore bool
var guestDest string
guestDest, ignore, err = c.shareFiles(m, idx, hostSharedDir, hostMountDir, guestSharedDir)
if err != nil {
return nil, nil, err
}
// Expand the list of mounts to ignore.
if ignore {
ignoredMounts[m.Source] = Mount{Source: m.Source}
continue
}
sharedDirMount := Mount{
Source: guestDest,
Destination: m.Destination,
Type: m.Type,
Options: m.Options,
ReadOnly: m.ReadOnly,
}
sharedDirMounts[sharedDirMount.Destination] = sharedDirMount
}
return sharedDirMounts, ignoredMounts, nil
}
func (c *Container) unmountHostMounts() error {
var span opentracing.Span
span, c.ctx = c.trace("unmountHostMounts")
defer span.Finish()
for _, m := range c.mounts {
if m.HostPath != "" {
span, _ := c.trace("unmount")
span.SetTag("host-path", m.HostPath)
if err := syscall.Unmount(m.HostPath, syscall.MNT_DETACH|UmountNoFollow); err != nil {
c.Logger().WithFields(logrus.Fields{
"host-path": m.HostPath,
"error": err,
}).Warn("Could not umount")
return err
}
if m.Type == "bind" {
s, err := os.Stat(m.HostPath)
if err != nil {
return errors.Wrapf(err, "Could not stat host-path %v", m.HostPath)
}
// Remove the empty file or directory
if s.Mode().IsRegular() && s.Size() == 0 {
os.Remove(m.HostPath)
}
if s.Mode().IsDir() {
syscall.Rmdir(m.HostPath)
}
}
span.Finish()
}
}
return nil
}
func filterDevices(c *Container, devices []ContainerDevice) (ret []ContainerDevice) {
for _, dev := range devices {
major, _ := c.sandbox.devManager.GetDeviceByID(dev.ID).GetMajorMinor()
if _, ok := cdromMajors[major]; ok {
c.Logger().WithFields(logrus.Fields{
"device": dev.ContainerPath,
}).Info("Not attach device because it is a CDROM")
continue
}
if major == floppyMajor {
c.Logger().WithFields(logrus.Fields{
"device": dev.ContainerPath,
}).Info("Not attaching device because it is a floppy drive")
continue
}
ret = append(ret, dev)
}
return
}
// Add any mount based block devices to the device manager and save the
// device ID for the particular mount. This'll occur when the mountpoint source
// is a block device.
func (c *Container) createBlockDevices() error {
if !c.checkBlockDeviceSupport() {
c.Logger().Warn("Block device not supported")
return nil
}
// iterate all mounts and create block device if it's block based.
for i, m := range c.mounts {
if len(m.BlockDeviceID) > 0 {
// Non-empty m.BlockDeviceID indicates there's already one device
// associated with the mount,so no need to create a new device for it
// and we only create block device for bind mount
continue
}
if m.Type != "bind" {
// We only handle for bind-mounts
continue
}
var stat unix.Stat_t
if err := unix.Stat(m.Source, &stat); err != nil {
return fmt.Errorf("stat %q failed: %v", m.Source, err)
}
var di *config.DeviceInfo
var err error
// Check if mount is a block device file. If it is, the block device will be attached to the host
// instead of passing this as a shared mount.
if stat.Mode&unix.S_IFBLK == unix.S_IFBLK {
di = &config.DeviceInfo{
HostPath: m.Source,
ContainerPath: m.Destination,
DevType: "b",
Major: int64(unix.Major(stat.Rdev)),
Minor: int64(unix.Minor(stat.Rdev)),
ReadOnly: m.ReadOnly,
}
// check whether source can be used as a pmem device
} else if di, err = config.PmemDeviceInfo(m.Source, m.Destination); err != nil {
c.Logger().WithError(err).
WithField("mount-source", m.Source).
Debug("no loop device")
}
if err == nil && di != nil {
b, err := c.sandbox.devManager.NewDevice(*di)
if err != nil {
// Do not return an error, try to create
// devices for other mounts
c.Logger().WithError(err).WithField("mount-source", m.Source).
Error("device manager failed to create new device")
continue
}
c.mounts[i].BlockDeviceID = b.DeviceID()
}
}
return nil
}
// newContainer creates a Container structure from a sandbox and a container configuration.
func newContainer(sandbox *Sandbox, contConfig *ContainerConfig) (*Container, error) {
span, _ := sandbox.trace("newContainer")
defer span.Finish()
if !contConfig.valid() {
return &Container{}, fmt.Errorf("Invalid container configuration")
}
c := &Container{
id: contConfig.ID,
sandboxID: sandbox.id,
rootFs: contConfig.RootFs,
config: contConfig,
sandbox: sandbox,
containerPath: filepath.Join(sandbox.id, contConfig.ID),
rootfsSuffix: "rootfs",
state: types.ContainerState{},
process: Process{},
mounts: contConfig.Mounts,
ctx: sandbox.ctx,
}
if useOldStore(sandbox.ctx) {
ctrStore, err := store.NewVCContainerStore(sandbox.ctx, c.sandboxID, c.id)
if err != nil {
return nil, err
}
c.store = ctrStore
state, err := c.store.LoadContainerState()
if err == nil {
c.state = state
}
var process Process
if err := c.store.Load(store.Process, &process); err == nil {
c.process = process
}
} else {
// experimental runtime use "persist.json" instead of legacy "state.json" as storage
err := c.Restore()
if err == nil {
//container restored
return c, nil
}
// Unexpected error
if !os.IsNotExist(err) && err != errContainerPersistNotExist {
return nil, err
}
}
// If mounts are block devices, add to devmanager
if err := c.createMounts(); err != nil {
return nil, err
}
// Add container's devices to sandbox's device-manager
if err := c.createDevices(contConfig); err != nil {
return nil, err
}
return c, nil
}
func (c *Container) loadMounts() ([]Mount, error) {
var mounts []Mount
if err := c.store.Load(store.Mounts, &mounts); err != nil {
return []Mount{}, err
}
return mounts, nil
}
func (c *Container) loadDevices() ([]ContainerDevice, error) {
var devices []ContainerDevice
if err := c.store.Load(store.DeviceIDs, &devices); err != nil {
return []ContainerDevice{}, err
}
return devices, nil
}
func (c *Container) createMounts() error {
if useOldStore(c.sandbox.ctx) {
mounts, err := c.loadMounts()
if err == nil {
// restore mounts from disk
c.mounts = mounts
return nil
}
}
// Create block devices for newly created container
return c.createBlockDevices()
}
func (c *Container) createDevices(contConfig *ContainerConfig) error {
// If sandbox supports "newstore", only newly created container can reach this function,
// so we don't call restore when `supportNewStore` is true
if useOldStore(c.sandbox.ctx) {
// Devices will be found in storage after create stage has completed.
// We load devices from storage at all other stages.
storedDevices, err := c.loadDevices()
if err == nil {
c.devices = storedDevices
return nil
}
}
// If devices were not found in storage, create Device implementations
// from the configuration. This should happen at create.
var storedDevices []ContainerDevice
for _, info := range contConfig.DeviceInfos {
dev, err := c.sandbox.devManager.NewDevice(info)
if err != nil {
return err
}
storedDevices = append(storedDevices, ContainerDevice{
ID: dev.DeviceID(),
ContainerPath: info.ContainerPath,
FileMode: info.FileMode,
UID: info.UID,
GID: info.GID,
})
}
c.devices = filterDevices(c, storedDevices)
return nil
}
// rollbackFailingContainerCreation rolls back important steps that might have
// been performed before the container creation failed.
// - Unplug CPU and memory resources from the VM.
// - Unplug devices from the VM.
func (c *Container) rollbackFailingContainerCreation() {
if err := c.detachDevices(); err != nil {
c.Logger().WithError(err).Error("rollback failed detachDevices()")
}
if err := c.removeDrive(); err != nil {
c.Logger().WithError(err).Error("rollback failed removeDrive()")
}
if err := c.unmountHostMounts(); err != nil {
c.Logger().WithError(err).Error("rollback failed unmountHostMounts()")
}
if err := bindUnmountContainerRootfs(c.ctx, getMountPath(c.sandbox.id), c); err != nil {
c.Logger().WithError(err).Error("rollback failed bindUnmountContainerRootfs()")
}
}
func (c *Container) checkBlockDeviceSupport() bool {
if !c.sandbox.config.HypervisorConfig.DisableBlockDeviceUse {
agentCaps := c.sandbox.agent.capabilities()
hypervisorCaps := c.sandbox.hypervisor.capabilities()
if agentCaps.IsBlockDeviceSupported() && hypervisorCaps.IsBlockDeviceHotplugSupported() {
return true
}
}
return false
}
// createContainer creates and start a container inside a Sandbox. It has to be
// called only when a new container, not known by the sandbox, has to be created.
func (c *Container) create() (err error) {
// In case the container creation fails, the following takes care
// of rolling back all the actions previously performed.
defer func() {
if err != nil {
c.Logger().WithError(err).Error("container create failed")
c.rollbackFailingContainerCreation()
}
}()
if c.checkBlockDeviceSupport() {
// If the rootfs is backed by a block device, go ahead and hotplug it to the guest
if err = c.hotplugDrive(); err != nil {
return
}
}
var (
machineType = c.sandbox.config.HypervisorConfig.HypervisorMachineType
normalAttachedDevs []ContainerDevice //for q35: normally attached devices
delayAttachedDevs []ContainerDevice //for q35: delay attached devices, for example, large bar space device
)
// Fix: https://github.com/kata-containers/runtime/issues/2460
if machineType == QemuQ35 {
// add Large Bar space device to delayAttachedDevs
for _, device := range c.devices {
var isLargeBarSpace bool
isLargeBarSpace, err = manager.IsVFIOLargeBarSpaceDevice(device.ContainerPath)
if err != nil {
return
}
if isLargeBarSpace {
delayAttachedDevs = append(delayAttachedDevs, device)
} else {
normalAttachedDevs = append(normalAttachedDevs, device)
}
}
} else {
normalAttachedDevs = c.devices
}
c.Logger().WithFields(logrus.Fields{
"machine_type": machineType,
"devices": normalAttachedDevs,
}).Info("normal attach devices")
if len(normalAttachedDevs) > 0 {
if err = c.attachDevices(normalAttachedDevs); err != nil {
return
}
}
// Deduce additional system mount info that should be handled by the agent
// inside the VM
c.getSystemMountInfo()
process, err := c.sandbox.agent.createContainer(c.sandbox, c)
if err != nil {
return err
}
c.process = *process
// lazy attach device after createContainer for q35
if machineType == QemuQ35 && len(delayAttachedDevs) > 0 {
c.Logger().WithFields(logrus.Fields{
"machine_type": machineType,
"devices": delayAttachedDevs,
}).Info("lazy attach devices")
if err = c.attachDevices(delayAttachedDevs); err != nil {
return
}
}
if !rootless.IsRootless() && !c.sandbox.config.SandboxCgroupOnly {
if err = c.cgroupsCreate(); err != nil {
return
}
}
if err = c.setContainerState(types.StateReady); err != nil {
return
}
return nil
}
func (c *Container) delete() error {
if c.state.State != types.StateReady &&
c.state.State != types.StateStopped {
return fmt.Errorf("Container not ready or stopped, impossible to delete")
}
// Remove the container from sandbox structure
if err := c.sandbox.removeContainer(c.id); err != nil {
return err
}
// If running rootless, there are no cgroups to remove
if !c.sandbox.config.SandboxCgroupOnly || !rootless.IsRootless() {
if err := c.cgroupsDelete(); err != nil {
return err
}
}
return c.sandbox.storeSandbox()
}
// checkSandboxRunning validates the container state.
//
// cmd specifies the operation (or verb) that the retrieval is destined
// for and is only used to make the returned error as descriptive as
// possible.
func (c *Container) checkSandboxRunning(cmd string) error {
if cmd == "" {
return fmt.Errorf("Cmd cannot be empty")
}
if c.sandbox.state.State != types.StateRunning {
return fmt.Errorf("Sandbox not running, impossible to %s the container", cmd)
}
return nil
}
func (c *Container) getSystemMountInfo() {
// check if /dev needs to be bind mounted from host /dev