summaryrefslogtreecommitdiffhomepage
path: root/pkg/v2/epoll.go
blob: 76c7b54d695b440d81fc616006220c9b69bd8298 (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
// +build linux

/*
   Copyright The containerd Authors.

   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 v2

import (
	"context"
	"sync"

	"github.com/containerd/cgroups"
	eventstypes "github.com/containerd/containerd/api/events"
	"github.com/containerd/containerd/events"
	"github.com/containerd/containerd/runtime"
	"github.com/sirupsen/logrus"
	"golang.org/x/sys/unix"
)

func newOOMEpoller(publisher events.Publisher) (*epoller, error) {
	fd, err := unix.EpollCreate1(unix.EPOLL_CLOEXEC)
	if err != nil {
		return nil, err
	}
	return &epoller{
		fd:        fd,
		publisher: publisher,
		set:       make(map[uintptr]*item),
	}, nil
}

type epoller struct {
	mu sync.Mutex

	fd        int
	publisher events.Publisher
	set       map[uintptr]*item
}

type item struct {
	id string
	cg cgroups.Cgroup
}

func (e *epoller) Close() error {
	return unix.Close(e.fd)
}

func (e *epoller) run(ctx context.Context) {
	var events [128]unix.EpollEvent
	for {
		select {
		case <-ctx.Done():
			e.Close()
			return
		default:
			n, err := unix.EpollWait(e.fd, events[:], -1)
			if err != nil {
				if err == unix.EINTR {
					continue
				}
				logrus.WithError(err).Error("cgroups: epoll wait")
			}
			for i := 0; i < n; i++ {
				e.process(ctx, uintptr(events[i].Fd))
			}
		}
	}
}

func (e *epoller) add(id string, cg cgroups.Cgroup) error {
	e.mu.Lock()
	defer e.mu.Unlock()
	fd, err := cg.OOMEventFD()
	if err != nil {
		return err
	}
	e.set[fd] = &item{
		id: id,
		cg: cg,
	}
	event := unix.EpollEvent{
		Fd:     int32(fd),
		Events: unix.EPOLLHUP | unix.EPOLLIN | unix.EPOLLERR,
	}
	return unix.EpollCtl(e.fd, unix.EPOLL_CTL_ADD, int(fd), &event)
}

func (e *epoller) process(ctx context.Context, fd uintptr) {
	flush(fd)
	e.mu.Lock()
	i, ok := e.set[fd]
	if !ok {
		e.mu.Unlock()
		return
	}
	e.mu.Unlock()
	if i.cg.State() == cgroups.Deleted {
		e.mu.Lock()
		delete(e.set, fd)
		e.mu.Unlock()
		unix.Close(int(fd))
		return
	}
	if err := e.publisher.Publish(ctx, runtime.TaskOOMEventTopic, &eventstypes.TaskOOM{
		ContainerID: i.id,
	}); err != nil {
		logrus.WithError(err).Error("publish OOM event")
	}
}

func flush(fd uintptr) error {
	var buf [8]byte
	_, err := unix.Read(int(fd), buf[:])
	return err
}