strlog.7 (2010 09)

ManualsBrandsHP ManualsSoftwareHP-UX Manpages

strlog(7) strlog(7)

User Interface

User processes access the log driver with an

open() call to /dev/strlog. Each open to the device will

obtain a separate stream. After a process opens

/dev/strlog, it indicates whether it is an error logger

or trace logger. It does this by issuing an

I_STR ioctl() system call with the appropriate value in the

ic_cmd ﬁeld of the strioctl structure, and the appropriate data and control information in a

trace_ids structure:

struct trace_ids {

short ti_mid;

short ti_sid;

char ti_level;

short ti_flags;

};

The values for

ic_cmd are:

I_ERRLOG Indicates an error logger. No trace_ids

data is needed.

I_TRCLOG Indicates a trace logger. A data buffer consisting of an array of one or more

trace_ids structures must be included.

If any of the ﬁelds of the

trace_ids structure contain a value of -1, /dev/strlog will accept what-

ever value it receives in that ﬁeld. Otherwise,

strlog only accepts messages only if the values of mid

and sid are the same as their counterparts in the

trace_ids structure, and if the message’s level is

equal to or less than the level value in the trace_ids structure.

Once the logger process has sent the

I_STR ioctl() call, the STREAMS log driver begins to send log

messages matching the restrictions to the logger process. The logger process obtains the log messages via

the getmsg() system call. The control part of the messages passed in this call includes a log_ctl

structure:

struct log_ctl {

short mid;

short sid;

char level;

short flags;

long ltime;

long ttime;

int seq_no;

};

The

log_ctl structure indicates the mid, sid, and level time in ticks since the boot time that the mes-

sage was submitted, the corresponding time in seconds since January 1, 1970, and a sequence number.

The time in seconds since January 1, 1970 is provided so that the date and time of the message can be

easily computed. The time in ticks since boot time is provided so that the relative timing of log messages

can be determined.

A user process, other than an error or trace logger, can send a log message to

strlog. The driver will

accept only the flags and level ﬁelds of the log_ctl structure in the control part of the message,

and a properly formatted data part of the message. The data part of the message is properly formatted if

it contains a null-terminated format string, followed by up to three arguments packed one word each

after the end of the string.

A different series of sequence numbers is provided for error and trace logging streams. These sequence

numbers are intended to help track the delivery of the messages. A gap in a sequence of numbers indi-

cates that the logger process did not successfully deliver them. This can happen if the logger process

stops sending messages for one reason or another (see strace (1M) and strerr (1M) command reference

pages for more information). The data part of messages contains text of the format string (null ter-

minated), followed by up to three arguments.

STREAMS-NetTL Link

Both STREAMS error logging and event tracing messages are mapped to NetTL logging messages, and

are delivered to NetTL. NetTL classiﬁes messages into four log classes: DISASTER, ERROR, WARNING,

and INFORMATIVE. The NetTL log class is determined by the flags according to the following rule:

If (flags & SL_ERROR) NetTL log class

then

if (flags & SL_FATAL) ====> DISASTER

2 Hewlett-Packard Company − 2 − HP-UX 11i Version 3: September 2010