summaryrefslogtreecommitdiff
path: root/pkg/spec/spec.go
blob: c947467670cb44ddd2426e10ebbfd0e4be31a876 (plain)
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
package createconfig

import (
	"os"
	"path/filepath"
	"strings"

	"github.com/containers/libpod/libpod"
	"github.com/containers/libpod/pkg/cgroups"
	"github.com/containers/libpod/pkg/rootless"
	pmount "github.com/containers/storage/pkg/mount"
	"github.com/docker/docker/oci/caps"
	"github.com/docker/go-units"
	"github.com/opencontainers/runc/libcontainer/user"
	spec "github.com/opencontainers/runtime-spec/specs-go"
	"github.com/opencontainers/runtime-tools/generate"
	"github.com/pkg/errors"
	"github.com/sirupsen/logrus"
)

const cpuPeriod = 100000

func getAvailableGids() (int64, error) {
	idMap, err := user.ParseIDMapFile("/proc/self/gid_map")
	if err != nil {
		return 0, err
	}
	count := int64(0)
	for _, r := range idMap {
		count += r.Count
	}
	return count, nil
}

// CreateConfigToOCISpec parses information needed to create a container into an OCI runtime spec
func (config *CreateConfig) createConfigToOCISpec(runtime *libpod.Runtime, userMounts []spec.Mount) (*spec.Spec, error) {
	cgroupPerm := "ro"
	g, err := generate.New("linux")
	if err != nil {
		return nil, err
	}
	// Remove the default /dev/shm mount to ensure we overwrite it
	g.RemoveMount("/dev/shm")
	g.HostSpecific = true
	addCgroup := true
	canMountSys := true

	isRootless := rootless.IsRootless()
	hasUserns := config.UsernsMode.IsContainer() || config.UsernsMode.IsNS() || len(config.IDMappings.UIDMap) > 0 || len(config.IDMappings.GIDMap) > 0
	inUserNS := isRootless || (hasUserns && !config.UsernsMode.IsHost())

	if inUserNS && config.NetMode.IsHost() {
		canMountSys = false
	}

	if config.Privileged && canMountSys {
		cgroupPerm = "rw"
		g.RemoveMount("/sys")
		sysMnt := spec.Mount{
			Destination: "/sys",
			Type:        "sysfs",
			Source:      "sysfs",
			Options:     []string{"rprivate", "nosuid", "noexec", "nodev", "rw"},
		}
		g.AddMount(sysMnt)
	} else if !canMountSys {
		addCgroup = false
		g.RemoveMount("/sys")
		r := "ro"
		if config.Privileged {
			r = "rw"
		}
		sysMnt := spec.Mount{
			Destination: "/sys",
			Type:        TypeBind,
			Source:      "/sys",
			Options:     []string{"rprivate", "nosuid", "noexec", "nodev", r, "rbind"},
		}
		g.AddMount(sysMnt)
		if !config.Privileged && isRootless {
			g.AddLinuxMaskedPaths("/sys/kernel")
		}
	}
	gid5Available := true
	if isRootless {
		nGids, err := getAvailableGids()
		if err != nil {
			return nil, err
		}
		gid5Available = nGids >= 5
	}
	// When using a different user namespace, check that the GID 5 is mapped inside
	// the container.
	if gid5Available && len(config.IDMappings.GIDMap) > 0 {
		mappingFound := false
		for _, r := range config.IDMappings.GIDMap {
			if r.ContainerID <= 5 && 5 < r.ContainerID+r.Size {
				mappingFound = true
				break
			}
		}
		if !mappingFound {
			gid5Available = false
		}

	}
	if !gid5Available {
		// If we have no GID mappings, the gid=5 default option would fail, so drop it.
		g.RemoveMount("/dev/pts")
		devPts := spec.Mount{
			Destination: "/dev/pts",
			Type:        "devpts",
			Source:      "devpts",
			Options:     []string{"rprivate", "nosuid", "noexec", "newinstance", "ptmxmode=0666", "mode=0620"},
		}
		g.AddMount(devPts)
	}

	if inUserNS && config.IpcMode.IsHost() {
		g.RemoveMount("/dev/mqueue")
		devMqueue := spec.Mount{
			Destination: "/dev/mqueue",
			Type:        TypeBind,
			Source:      "/dev/mqueue",
			Options:     []string{"bind", "nosuid", "noexec", "nodev"},
		}
		g.AddMount(devMqueue)
	}
	if inUserNS && config.PidMode.IsHost() {
		g.RemoveMount("/proc")
		procMount := spec.Mount{
			Destination: "/proc",
			Type:        TypeBind,
			Source:      "/proc",
			Options:     []string{"rbind", "nosuid", "noexec", "nodev"},
		}
		g.AddMount(procMount)
	}

	if addCgroup {
		cgroupMnt := spec.Mount{
			Destination: "/sys/fs/cgroup",
			Type:        "cgroup",
			Source:      "cgroup",
			Options:     []string{"rprivate", "nosuid", "noexec", "nodev", "relatime", cgroupPerm},
		}
		g.AddMount(cgroupMnt)
	}
	g.SetProcessCwd(config.WorkDir)
	g.SetProcessArgs(config.Command)
	g.SetProcessTerminal(config.Tty)

	for key, val := range config.Annotations {
		g.AddAnnotation(key, val)
	}
	g.SetRootReadonly(config.ReadOnlyRootfs)

	if config.HTTPProxy {
		for _, envSpec := range []string{
			"http_proxy",
			"HTTP_PROXY",
			"https_proxy",
			"HTTPS_PROXY",
			"ftp_proxy",
			"FTP_PROXY",
			"no_proxy",
			"NO_PROXY",
		} {
			envVal := os.Getenv(envSpec)
			if envVal != "" {
				g.AddProcessEnv(envSpec, envVal)
			}
		}
	}

	hostname := config.Hostname
	if hostname == "" && (config.NetMode.IsHost() || config.UtsMode.IsHost()) {
		hostname, err = os.Hostname()
		if err != nil {
			return nil, errors.Wrap(err, "unable to retrieve hostname")
		}
	}
	g.RemoveHostname()
	if config.Hostname != "" || !config.UtsMode.IsHost() {
		// Set the hostname in the OCI configuration only
		// if specified by the user or if we are creating
		// a new UTS namespace.
		g.SetHostname(hostname)
	}
	g.AddProcessEnv("HOSTNAME", hostname)

	for sysctlKey, sysctlVal := range config.Sysctl {
		g.AddLinuxSysctl(sysctlKey, sysctlVal)
	}
	g.AddProcessEnv("container", "podman")

	addedResources := false

	// RESOURCES - MEMORY
	if config.Resources.Memory != 0 {
		g.SetLinuxResourcesMemoryLimit(config.Resources.Memory)
		// If a swap limit is not explicitly set, also set a swap limit
		// Default to double the memory limit
		if config.Resources.MemorySwap == 0 {
			g.SetLinuxResourcesMemorySwap(2 * config.Resources.Memory)
		}
		addedResources = true
	}
	if config.Resources.MemoryReservation != 0 {
		g.SetLinuxResourcesMemoryReservation(config.Resources.MemoryReservation)
		addedResources = true
	}
	if config.Resources.MemorySwap != 0 {
		g.SetLinuxResourcesMemorySwap(config.Resources.MemorySwap)
		addedResources = true
	}
	if config.Resources.KernelMemory != 0 {
		g.SetLinuxResourcesMemoryKernel(config.Resources.KernelMemory)
		addedResources = true
	}
	if config.Resources.MemorySwappiness != -1 {
		g.SetLinuxResourcesMemorySwappiness(uint64(config.Resources.MemorySwappiness))
		addedResources = true
	}
	g.SetLinuxResourcesMemoryDisableOOMKiller(config.Resources.DisableOomKiller)
	g.SetProcessOOMScoreAdj(config.Resources.OomScoreAdj)

	// RESOURCES - CPU
	if config.Resources.CPUShares != 0 {
		g.SetLinuxResourcesCPUShares(config.Resources.CPUShares)
		addedResources = true
	}
	if config.Resources.CPUQuota != 0 {
		g.SetLinuxResourcesCPUQuota(config.Resources.CPUQuota)
		addedResources = true
	}
	if config.Resources.CPUPeriod != 0 {
		g.SetLinuxResourcesCPUPeriod(config.Resources.CPUPeriod)
		addedResources = true
	}
	if config.Resources.CPUs != 0 {
		g.SetLinuxResourcesCPUPeriod(cpuPeriod)
		g.SetLinuxResourcesCPUQuota(int64(config.Resources.CPUs * cpuPeriod))
		addedResources = true
	}
	if config.Resources.CPURtRuntime != 0 {
		g.SetLinuxResourcesCPURealtimeRuntime(config.Resources.CPURtRuntime)
		addedResources = true
	}
	if config.Resources.CPURtPeriod != 0 {
		g.SetLinuxResourcesCPURealtimePeriod(config.Resources.CPURtPeriod)
		addedResources = true
	}
	if config.Resources.CPUsetCPUs != "" {
		g.SetLinuxResourcesCPUCpus(config.Resources.CPUsetCPUs)
		addedResources = true
	}
	if config.Resources.CPUsetMems != "" {
		g.SetLinuxResourcesCPUMems(config.Resources.CPUsetMems)
		addedResources = true
	}

	// Devices
	if config.Privileged {
		// If privileged, we need to add all the host devices to the
		// spec.  We do not add the user provided ones because we are
		// already adding them all.
		if err := config.AddPrivilegedDevices(&g); err != nil {
			return nil, err
		}
	} else {
		for _, devicePath := range config.Devices {
			if err := devicesFromPath(&g, devicePath); err != nil {
				return nil, err
			}
		}
	}

	for _, uidmap := range config.IDMappings.UIDMap {
		g.AddLinuxUIDMapping(uint32(uidmap.HostID), uint32(uidmap.ContainerID), uint32(uidmap.Size))
	}
	for _, gidmap := range config.IDMappings.GIDMap {
		g.AddLinuxGIDMapping(uint32(gidmap.HostID), uint32(gidmap.ContainerID), uint32(gidmap.Size))
	}
	// SECURITY OPTS
	g.SetProcessNoNewPrivileges(config.NoNewPrivs)

	if !config.Privileged {
		g.SetProcessApparmorProfile(config.ApparmorProfile)
	}

	blockAccessToKernelFilesystems(config, &g)

	// RESOURCES - PIDS
	if config.Resources.PidsLimit != 0 {
		g.SetLinuxResourcesPidsLimit(config.Resources.PidsLimit)
		addedResources = true
	}

	for name, val := range config.Env {
		g.AddProcessEnv(name, val)
	}

	if err := addRlimits(config, &g); err != nil {
		return nil, err
	}

	if err := addPidNS(config, &g); err != nil {
		return nil, err
	}

	if err := addUserNS(config, &g); err != nil {
		return nil, err
	}

	if err := addNetNS(config, &g); err != nil {
		return nil, err
	}

	if err := addUTSNS(config, &g); err != nil {
		return nil, err
	}

	if err := addIpcNS(config, &g); err != nil {
		return nil, err
	}

	if err := addCgroupNS(config, &g); err != nil {
		return nil, err
	}
	configSpec := g.Config

	// HANDLE CAPABILITIES
	// NOTE: Must happen before SECCOMP
	if !config.Privileged {
		if err := setupCapabilities(config, configSpec); err != nil {
			return nil, err
		}
	} else {
		g.SetupPrivileged(true)
	}

	// HANDLE SECCOMP

	if config.SeccompProfilePath != "unconfined" {
		seccompConfig, err := getSeccompConfig(config, configSpec)
		if err != nil {
			return nil, err
		}
		configSpec.Linux.Seccomp = seccompConfig
	}

	// Clear default Seccomp profile from Generator for privileged containers
	if config.SeccompProfilePath == "unconfined" || config.Privileged {
		configSpec.Linux.Seccomp = nil
	}

	// BIND MOUNTS
	configSpec.Mounts = supercedeUserMounts(userMounts, configSpec.Mounts)
	// Process mounts to ensure correct options
	configSpec.Mounts = initFSMounts(configSpec.Mounts)

	// BLOCK IO
	blkio, err := config.CreateBlockIO()
	if err != nil {
		return nil, errors.Wrapf(err, "error creating block io")
	}
	if blkio != nil {
		configSpec.Linux.Resources.BlockIO = blkio
		addedResources = true
	}

	if rootless.IsRootless() {
		cgroup2, err := cgroups.IsCgroup2UnifiedMode()
		if err != nil {
			return nil, err
		}
		if addedResources && !cgroup2 {
			return nil, errors.New("invalid configuration, cannot set resources with rootless containers not using cgroups v2 unified mode")
		}
		if !cgroup2 {
			// Force the resources block to be empty instead of having default values.
			configSpec.Linux.Resources = &spec.LinuxResources{}
		}
	}

	// Make sure that the bind mounts keep options like nosuid, noexec, nodev.
	mounts, err := pmount.GetMounts()
	if err != nil {
		return nil, err
	}
	for i := range configSpec.Mounts {
		m := &configSpec.Mounts[i]
		isBind := false
		for _, o := range m.Options {
			if o == "bind" || o == "rbind" {
				isBind = true
				break
			}
		}
		if !isBind {
			continue
		}
		mount, err := findMount(m.Source, mounts)
		if err != nil {
			return nil, err
		}
		if mount == nil {
			continue
		}
	next_option:
		for _, o := range strings.Split(mount.Opts, ",") {
			if o == "nosuid" || o == "noexec" || o == "nodev" {
				for _, e := range m.Options {
					if e == o {
						continue next_option
					}
				}
				m.Options = append(m.Options, o)
			}
		}
	}

	// Add annotations
	if configSpec.Annotations == nil {
		configSpec.Annotations = make(map[string]string)
	}

	if config.CidFile != "" {
		configSpec.Annotations[libpod.InspectAnnotationCIDFile] = config.CidFile
	}

	if config.Rm {
		configSpec.Annotations[libpod.InspectAnnotationAutoremove] = libpod.InspectResponseTrue
	} else {
		configSpec.Annotations[libpod.InspectAnnotationAutoremove] = libpod.InspectResponseFalse
	}

	if len(config.VolumesFrom) > 0 {
		configSpec.Annotations[libpod.InspectAnnotationVolumesFrom] = strings.Join(config.VolumesFrom, ",")
	}

	if config.Privileged {
		configSpec.Annotations[libpod.InspectAnnotationPrivileged] = libpod.InspectResponseTrue
	} else {
		configSpec.Annotations[libpod.InspectAnnotationPrivileged] = libpod.InspectResponseFalse
	}

	if config.PublishAll {
		configSpec.Annotations[libpod.InspectAnnotationPublishAll] = libpod.InspectResponseTrue
	} else {
		configSpec.Annotations[libpod.InspectAnnotationPublishAll] = libpod.InspectResponseFalse
	}

	if config.Init {
		configSpec.Annotations[libpod.InspectAnnotationInit] = libpod.InspectResponseTrue
	} else {
		configSpec.Annotations[libpod.InspectAnnotationInit] = libpod.InspectResponseFalse
	}

	for _, opt := range config.SecurityOpts {
		// Split on both : and =
		splitOpt := strings.Split(opt, "=")
		if len(splitOpt) == 1 {
			splitOpt = strings.Split(opt, ":")
		}
		if len(splitOpt) < 2 {
			continue
		}
		switch splitOpt[0] {
		case "label":
			configSpec.Annotations[libpod.InspectAnnotationLabel] = splitOpt[1]
		case "seccomp":
			configSpec.Annotations[libpod.InspectAnnotationSeccomp] = splitOpt[1]
		case "apparmor":
			configSpec.Annotations[libpod.InspectAnnotationApparmor] = splitOpt[1]
		}
	}

	return configSpec, nil
}

func findMount(target string, mounts []*pmount.Info) (*pmount.Info, error) {
	var err error
	target, err = filepath.Abs(target)
	if err != nil {
		return nil, errors.Wrapf(err, "cannot resolve %s", target)
	}
	var bestSoFar *pmount.Info
	for _, i := range mounts {
		if bestSoFar != nil && len(bestSoFar.Mountpoint) > len(i.Mountpoint) {
			// Won't be better than what we have already found
			continue
		}
		if strings.HasPrefix(target, i.Mountpoint) {
			bestSoFar = i
		}
	}
	return bestSoFar, nil
}

func blockAccessToKernelFilesystems(config *CreateConfig, g *generate.Generator) {
	if !config.Privileged {
		for _, mp := range []string{
			"/proc/acpi",
			"/proc/kcore",
			"/proc/keys",
			"/proc/latency_stats",
			"/proc/timer_list",
			"/proc/timer_stats",
			"/proc/sched_debug",
			"/proc/scsi",
			"/sys/firmware",
			"/sys/fs/selinux",
		} {
			g.AddLinuxMaskedPaths(mp)
		}

		if config.PidMode.IsHost() && rootless.IsRootless() {
			return
		}

		for _, rp := range []string{
			"/proc/asound",
			"/proc/bus",
			"/proc/fs",
			"/proc/irq",
			"/proc/sys",
			"/proc/sysrq-trigger",
		} {
			g.AddLinuxReadonlyPaths(rp)
		}
	}
}

func addPidNS(config *CreateConfig, g *generate.Generator) error {
	pidMode := config.PidMode
	if IsNS(string(pidMode)) {
		return g.AddOrReplaceLinuxNamespace(string(spec.PIDNamespace), NS(string(pidMode)))
	}
	if pidMode.IsHost() {
		return g.RemoveLinuxNamespace(string(spec.PIDNamespace))
	}
	if pidMode.IsContainer() {
		logrus.Debug("using container pidmode")
	}
	if IsPod(string(pidMode)) {
		logrus.Debug("using pod pidmode")
	}
	return nil
}

func addUserNS(config *CreateConfig, g *generate.Generator) error {
	if IsNS(string(config.UsernsMode)) {
		if err := g.AddOrReplaceLinuxNamespace(string(spec.UserNamespace), NS(string(config.UsernsMode))); err != nil {
			return err
		}
		// runc complains if no mapping is specified, even if we join another ns.  So provide a dummy mapping
		g.AddLinuxUIDMapping(uint32(0), uint32(0), uint32(1))
		g.AddLinuxGIDMapping(uint32(0), uint32(0), uint32(1))
	}

	if (len(config.IDMappings.UIDMap) > 0 || len(config.IDMappings.GIDMap) > 0) && !config.UsernsMode.IsHost() {
		if err := g.AddOrReplaceLinuxNamespace(string(spec.UserNamespace), ""); err != nil {
			return err
		}
	}
	return nil
}

func addNetNS(config *CreateConfig, g *generate.Generator) error {
	netMode := config.NetMode
	if netMode.IsHost() {
		logrus.Debug("Using host netmode")
		return g.RemoveLinuxNamespace(string(spec.NetworkNamespace))
	} else if netMode.IsNone() {
		logrus.Debug("Using none netmode")
		return nil
	} else if netMode.IsBridge() {
		logrus.Debug("Using bridge netmode")
		return nil
	} else if netMode.IsContainer() {
		logrus.Debug("Using container netmode")
		return nil
	} else if IsNS(string(netMode)) {
		logrus.Debug("Using ns netmode")
		return g.AddOrReplaceLinuxNamespace(string(spec.NetworkNamespace), NS(string(netMode)))
	} else if IsPod(string(netMode)) {
		logrus.Debug("Using pod netmode, unless pod is not sharing")
		return nil
	} else if netMode.IsSlirp4netns() {
		logrus.Debug("Using slirp4netns netmode")
		return nil
	} else if netMode.IsUserDefined() {
		logrus.Debug("Using user defined netmode")
		return nil
	}
	return errors.Errorf("unknown network mode")
}

func addUTSNS(config *CreateConfig, g *generate.Generator) error {
	utsMode := config.UtsMode
	if IsNS(string(utsMode)) {
		return g.AddOrReplaceLinuxNamespace(string(spec.UTSNamespace), NS(string(utsMode)))
	}
	if utsMode.IsHost() {
		return g.RemoveLinuxNamespace(string(spec.UTSNamespace))
	}
	return nil
}

func addIpcNS(config *CreateConfig, g *generate.Generator) error {
	ipcMode := config.IpcMode
	if IsNS(string(ipcMode)) {
		return g.AddOrReplaceLinuxNamespace(string(spec.IPCNamespace), NS(string(ipcMode)))
	}
	if ipcMode.IsHost() {
		return g.RemoveLinuxNamespace(string(spec.IPCNamespace))
	}
	if ipcMode.IsContainer() {
		logrus.Debug("Using container ipcmode")
	}

	return nil
}

func addCgroupNS(config *CreateConfig, g *generate.Generator) error {
	cgroupMode := config.CgroupMode
	if cgroupMode.IsNS() {
		return g.AddOrReplaceLinuxNamespace(string(spec.CgroupNamespace), NS(string(cgroupMode)))
	}
	if cgroupMode.IsHost() {
		return g.RemoveLinuxNamespace(string(spec.CgroupNamespace))
	}
	if cgroupMode.IsPrivate() {
		return g.AddOrReplaceLinuxNamespace(string(spec.CgroupNamespace), "")
	}
	if cgroupMode.IsContainer() {
		logrus.Debug("Using container cgroup mode")
	}
	return nil
}

func addRlimits(config *CreateConfig, g *generate.Generator) error {
	var (
		kernelMax  uint64 = 1048576
		isRootless        = rootless.IsRootless()
		nofileSet         = false
		nprocSet          = false
	)

	for _, u := range config.Resources.Ulimit {
		if u == "host" {
			if len(config.Resources.Ulimit) != 1 {
				return errors.New("ulimit can use host only once")
			}
			g.Config.Process.Rlimits = nil
			break
		}

		ul, err := units.ParseUlimit(u)
		if err != nil {
			return errors.Wrapf(err, "ulimit option %q requires name=SOFT:HARD, failed to be parsed", u)
		}

		if ul.Name == "nofile" {
			nofileSet = true
		} else if ul.Name == "nproc" {
			nprocSet = true
		}

		g.AddProcessRlimits("RLIMIT_"+strings.ToUpper(ul.Name), uint64(ul.Hard), uint64(ul.Soft))
	}

	// If not explicitly overridden by the user, default number of open
	// files and number of processes to the maximum they can be set to
	// (without overriding a sysctl)
	if !nofileSet && !isRootless {
		g.AddProcessRlimits("RLIMIT_NOFILE", kernelMax, kernelMax)
	}
	if !nprocSet && !isRootless {
		g.AddProcessRlimits("RLIMIT_NPROC", kernelMax, kernelMax)
	}

	return nil
}

func setupCapabilities(config *CreateConfig, configSpec *spec.Spec) error {
	useNotRoot := func(user string) bool {
		if user == "" || user == "root" || user == "0" {
			return false
		}
		return true
	}

	var err error
	var caplist []string
	bounding := configSpec.Process.Capabilities.Bounding
	if useNotRoot(config.User) {
		configSpec.Process.Capabilities.Bounding = caplist
	}
	caplist, err = caps.TweakCapabilities(configSpec.Process.Capabilities.Bounding, config.CapAdd, config.CapDrop, nil, false)
	if err != nil {
		return err
	}

	configSpec.Process.Capabilities.Bounding = caplist
	configSpec.Process.Capabilities.Permitted = caplist
	configSpec.Process.Capabilities.Inheritable = caplist
	configSpec.Process.Capabilities.Effective = caplist
	configSpec.Process.Capabilities.Ambient = caplist
	if useNotRoot(config.User) {
		caplist, err = caps.TweakCapabilities(bounding, config.CapAdd, config.CapDrop, nil, false)
		if err != nil {
			return err
		}
	}
	configSpec.Process.Capabilities.Bounding = caplist
	return nil
}