man-pages/man2/syslog.2

.\" Copyright (C) 1995 Andries Brouwer (aeb@cwi.nl)
.\"
.\" Permission is granted to make and distribute verbatim copies of this
.\" manual provided the copyright notice and this permission notice are
.\" preserved on all copies.
.\"
.\" Permission is granted to copy and distribute modified versions of this
.\" manual under the conditions for verbatim copying, provided that the
.\" entire resulting derived work is distributed under the terms of a
.\" permission notice identical to this one.
.\"
.\" Since the Linux kernel and libraries are constantly changing, this
.\" manual page may be incorrect or out-of-date.  The author(s) assume no
.\" responsibility for errors or omissions, or for damages resulting from
.\" the use of the information contained herein.  The author(s) may not
.\" have taken the same level of care in the production of this manual,
.\" which is licensed free of charge, as they might when working
.\" professionally.
.\"
.\" Formatted or processed versions of this manual, if unaccompanied by
.\" the source, must acknowledge the copyright and authors of this work.
.\"
.\" Written 11 June 1995 by Andries Brouwer <aeb@cwi.nl>
.\" 2008-02-15, Jeremy Kerr <jk@ozlabs.org>
.\"     Add info on command type 10; add details on types 6, 7, 8, & 9.
.\" 2008-02-15, Michael Kerrisk <mtk.manpages@gmail.com>
.\"     Update LOG_BUF_LEN details; update RETURN VALUE section.
.\"     
.TH SYSLOG 2  2008-02-15 "Linux" "Linux Programmer's Manual"
.SH NAME
syslog, klogctl \- read and/or clear kernel message ring buffer;
set console_loglevel
.SH SYNOPSIS
.nf
.BI "int syslog(int " type ", char *" bufp ", int " len );
.B  "                /* No wrapper provided in glibc */"
.sp
/* The glibc interface */
.br
.B "#include <sys/klog.h>"
.sp
.BI "int klogctl(int " type ", char *" bufp ", int " len );
.fi
.SH DESCRIPTION
If you need the libc function
.BR syslog ()
(which talks to
.BR syslogd (8)),
then look at
.BR syslog (3).
The system call of this name is about controlling the kernel
.IR printk ()
buffer, and the glibc version is called
.BR klogctl ().

The \fItype\fP argument determines the action taken by this function.

Quoting from
.IR kernel/printk.c :
.nf
/*
 * Commands to sys_syslog:
 *
 *      0 \-\- Close the log.  Currently a NOP.
 *      1 \-\- Open the log. Currently a NOP.
 *      2 \-\- Read from the log.
 *      3 \-\- Read all messages remaining in the ring buffer.
 *      4 \-\- Read and clear all messages remaining in the ring buffer
 *      5 \-\- Clear ring buffer.
 *      6 \-\- Disable printk to console
 *      7 \-\- Enable printk to console
 *      8 \-\- Set level of messages printed to console
 *      9 \-\- Return number of unread characters in the log buffer
 *     10 \-\- Return size of the log buffer
 */
.fi

Only command types 3 and 10 are allowed to non-root processes.
Type 9 was added in 2.4.10; type 10 in 2.6.6.
.SS The kernel log buffer
The kernel has a cyclic buffer of length
.B LOG_BUF_LEN
in which messages given as arguments to the kernel function
.BR printk ()
are stored (regardless of their loglevel).
In early kernels, 
.B LOG_BUF_LEN
had the value 4096;
from kernel 1.3.54, it was 8192;
from kernel 2.1.113 it was 16384;
since 2.4.23/2.6 the value is a kernel configuration option.
.\" Under "General setup" ==> "Kernel log buffer size"
.\" For 2.6, precisely the option seems to have appeared in 2.5.55.
In recent kernels the size can be queried with command type 10.

The call
.I "syslog(2,buf,len)"
waits until this kernel log buffer is nonempty, and then reads
at most \fIlen\fP bytes into the buffer \fIbuf\fP.
It returns
the number of bytes read.
Bytes read from the log disappear from
the log buffer: the information can only be read once.
This is the function executed by the kernel when a user program
reads
.IR /proc/kmsg .

The call
.I syslog(3,buf,len)
will read the last \fIlen\fP bytes from the log buffer (non-destructively),
but will not read more than was written into the buffer since the
last "clear ring buffer" command (which does not clear the buffer at all).
It returns the number of bytes read.

The call
.I syslog(4,buf,len)
does precisely the same, but also executes the "clear ring buffer" command.

The call
.I syslog(5,dummy,dummy)
executes just the "clear ring buffer" command.
(In each call where
.I buf
or
.I len
is shown as "dummy", the value of the argument is ignored by the call.)

The call
.I syslog(6,dummy,dummy)
sets the console log level to minimum, so that no messages are printed
to the console.

The call
.I syslog(7,dummy,dummy)
sets the console log level to default, so that messages are printed
to the console.

The call
.I syslog(8,dummy,level)
sets the console log level to
.IR level ,
which must be an integer between 1 and 8 (inclusive).
See the
.B loglevel
section for details.

The call
.I syslog(9,dummy,dummy)
returns the number of bytes currently available to be read
on the kernel log buffer.

The call
.I syslog(10,dummy,dummy)
returns the total size of the kernel log buffer.
.SS The loglevel
The kernel routine
.BR printk ()
will only print a message on the
console, if it has a loglevel less than the value of the variable
.IR console_loglevel .
This variable initially has the value
.B DEFAULT_CONSOLE_LOGLEVEL
(7), but is set to 10 if the
kernel command line contains the word "debug", and to 15 in case
of a kernel fault (the 10 and 15 are just silly, and equivalent to 8).
This variable is set (to a value in the range 1-8) by the call
.IR syslog(8,dummy,value) .
The calls
.I syslog(type,dummy,dummy)
with \fItype\fP equal to 6 or 7, set it to 1 (kernel panics only)
or 7 (all except debugging messages), respectively.

Every text line in a message has its own loglevel.
This level is
.I "DEFAULT_MESSAGE_LOGLEVEL \- 1"
(6) unless the line starts with <d>
where \fId\fP is a digit in the range 1-7, in which case the level
is \fId\fP.
The conventional meaning of the loglevel is defined in
.I <linux/kernel.h>
as follows:

.nf
#define KERN_EMERG    "<0>"  /* system is unusable               */
#define KERN_ALERT    "<1>"  /* action must be taken immediately */
#define KERN_CRIT     "<2>"  /* critical conditions              */
#define KERN_ERR      "<3>"  /* error conditions                 */
#define KERN_WARNING  "<4>"  /* warning conditions               */
#define KERN_NOTICE   "<5>"  /* normal but significant condition */
#define KERN_INFO     "<6>"  /* informational                    */
#define KERN_DEBUG    "<7>"  /* debug-level messages             */
.fi
.SH "RETURN VALUE"
For \fItype\fP equal to 2, 3, or 4, a successful call to
.BR syslog ()
returns the number
of bytes read.
For \fItype\fP 9,
.BR syslog ()
returns the number of bytes currently
available to be read on the kernel log buffer.
For \fItype\fP 10,
.BR syslog ()
returns the total size  of  the  kernel log buffer.
For other values of \fItype\fP, 0 is returned on success.

In case of error, \-1 is returned,
and \fIerrno\fP is set to indicate the error.
.SH ERRORS
.TP
.B EINVAL
Bad parameters (e.g.,
bad
.IR type ;
or for
.I type
2, 3, or 4,
.I buf
is NULL,
or
.I len
is less than zero; or for
.I type
8, the
.I level
is outside the range 1 to 8).
.TP
.B EPERM
An attempt was made to change console_loglevel or clear the kernel
message ring buffer by a process without root permissions
(more precisely: without the
.B CAP_SYS_ADMIN
capability).
.TP
.B ERESTARTSYS
System call was interrupted by a signal; nothing was read.
(This can be seen only during a trace.)
.SH "CONFORMING TO"
This system call is Linux-specific and should not be used in programs
intended to be portable.
.SH NOTES
From the very start people noted that it is unfortunate that
a system call and a library routine of the same name are entirely
different animals.
In libc4 and libc5 the number of this call was defined by
.BR SYS_klog .
In glibc 2.0 the syscall is baptized
.BR klogctl ().
.SH "SEE ALSO"
.BR syslog (3)