xref: /linux/Documentation/ABI/testing/dev-kmsg (revision cbac924200b838cfb8d8b1415113d788089dc50b)
1What:		/dev/kmsg
2Date:		Mai 2012
3KernelVersion:	3.5
4Contact:	Kay Sievers <kay@vrfy.org>
5Description:	The /dev/kmsg character device node provides userspace access
6		to the kernel's printk buffer.
7
8		Injecting messages:
9
10		Every write() to the opened device node places a log entry in
11		the kernel's printk buffer.
12
13		The logged line can be prefixed with a <N> syslog prefix, which
14		carries the syslog priority and facility. The single decimal
15		prefix number is composed of the 3 lowest bits being the syslog
16		priority and the next 8 bits the syslog facility number.
17
18		If no prefix is given, the priority number is the default kernel
19		log priority and the facility number is set to LOG_USER (1). It
20		is not possible to inject messages from userspace with the
21		facility number LOG_KERN (0), to make sure that the origin of
22		the messages can always be reliably determined.
23
24		Accessing the buffer:
25
26		Every read() from the opened device node receives one record
27		of the kernel's printk buffer.
28
29		The first read() directly following an open() always returns
30		first message in the buffer; there is no kernel-internal
31		persistent state; many readers can concurrently open the device
32		and read from it, without affecting other readers.
33
34		Every read() will receive the next available record. If no more
35		records are available read() will block, or if O_NONBLOCK is
36		used -EAGAIN returned.
37
38		Messages in the record ring buffer get overwritten as whole,
39		there are never partial messages received by read().
40
41		In case messages get overwritten in the circular buffer while
42		the device is kept open, the next read() will return -EPIPE,
43		and the seek position be updated to the next available record.
44		Subsequent reads() will return available records again.
45
46		Unlike the classic syslog() interface, the 64 bit record
47		sequence numbers allow to calculate the amount of lost
48		messages, in case the buffer gets overwritten. And they allow
49		to reconnect to the buffer and reconstruct the read position
50		if needed, without limiting the interface to a single reader.
51
52		The device supports seek with the following parameters:
53
54		SEEK_SET, 0
55		  seek to the first entry in the buffer
56		SEEK_END, 0
57		  seek after the last entry in the buffer
58		SEEK_DATA, 0
59		  seek after the last record available at the time
60		  the last SYSLOG_ACTION_CLEAR was issued.
61
62		Other seek operations or offsets are not supported because of
63		the special behavior this device has. The device allows to read
64		or write only whole variable length messages (records) that are
65		stored in a ring buffer.
66
67		Because of the non-standard behavior also the error values are
68		non-standard. -ESPIPE is returned for non-zero offset. -EINVAL
69		is returned for other operations, e.g. SEEK_CUR. This behavior
70		and values are historical and could not be modified without the
71		risk of breaking userspace.
72
73		The output format consists of a prefix carrying the syslog
74		prefix including priority and facility, the 64 bit message
75		sequence number and the monotonic timestamp in microseconds,
76		and a flag field. All fields are separated by a ','.
77
78		Future extensions might add more comma separated values before
79		the terminating ';'. Unknown fields and values should be
80		gracefully ignored.
81
82		The human readable text string starts directly after the ';'
83		and is terminated by a '\n'. Untrusted values derived from
84		hardware or other facilities are printed, therefore
85		all non-printable characters and '\' itself in the log message
86		are escaped by "\x00" C-style hex encoding.
87
88		A line starting with ' ', is a continuation line, adding
89		key/value pairs to the log message, which provide the machine
90		readable context of the message, for reliable processing in
91		userspace.
92
93		Example::
94
95		  7,160,424069,-;pci_root PNP0A03:00: host bridge window [io  0x0000-0x0cf7] (ignored)
96		   SUBSYSTEM=acpi
97		   DEVICE=+acpi:PNP0A03:00
98		  6,339,5140900,-;NET: Registered protocol family 10
99		  30,340,5690716,-;udevd[80]: starting version 181
100
101		The DEVICE= key uniquely identifies devices the following way:
102
103		  ============  =================
104		  b12:8         block dev_t
105		  c127:3        char dev_t
106		  n8            netdev ifindex
107		  +sound:card0  subsystem:devname
108		  ============  =================
109
110		The flags field carries '-' by default. A 'c' indicates a
111		fragment of a line. Note, that these hints about continuation
112		lines are not necessarily correct, and the stream could be
113		interleaved with unrelated messages, but merging the lines in
114		the output usually produces better human readable results. A
115		similar logic is used internally when messages are printed to
116		the console, /proc/kmsg or the syslog() syscall.
117
118		By default, kernel tries to avoid fragments by concatenating
119		when it can and fragments are rare; however, when extended
120		console support is enabled, the in-kernel concatenation is
121		disabled and /dev/kmsg output will contain more fragments. If
122		the log consumer performs concatenation, the end result
123		should be the same. In the future, the in-kernel concatenation
124		may be removed entirely and /dev/kmsg users are recommended to
125		implement fragment handling.
126
127Users:		dmesg(1), userspace kernel log consumers
128