2

main

10

memccpy

19

memmove

23

__milli_rem32U

20

mmap

24

_mmap_sys

21

strcmp

5

strcpy

3

strlen

8

toupper

4

uppercase

 

 

----------------------------------------------------------------------

Diagnostic Messages

----------------------------------------------------------------------

+ Note: Multiple sampling counter variations are not available on HP-UX. Caliper will use the smaller sampling period variation seen (100) for all event counters

----------------------------------------------------------------------

Sampled Call Graph Profile Report Details

This section gives some information about the Sampled Call Graph Profile report. For more information, see “scgprof Measurement Report Description” (p. 211).

Flat Profile Part of the Report

The flat profile part of the report contains two levels of information:

Function Summary: exact counts of CPU metrics summed across an entire run of a measured application

Function Details: sampled IPs that are associated with particular locations in the measured application

You can specify how to arrange the results in the flat profile with the --sort-byoption. The valid sorting parameters are:

call-count

msecs-per-call

samples (default)

seconds

See “Metrics You Can Use for Report Sorting and Cutoffs” (p. 105) for details.

Call Graph Part of the Report

The lines full of dashes divide the call graph into entries, one for each function. Each entry has one or more lines.

In each entry, the primary line is the one that starts with an index number in square brackets. The end of this line says which function the entry is for.

The preceding lines in the entry describe the callers of this function.

The lines following the primary line describe its subroutines (called children in the call graph). The entries are sorted by time spent in the function and its subroutines.

Hot Call Paths Part of the Report

This section reports the most probable hottest call paths. A call path represents a subset of the program's execution. The Hot Call Paths section reports the percentage of a program's CPU time that is attributed to specific call paths. Since scgprof does not store the whole call stack, it only tries to report the most probable hot paths in the program. There could be some false positives and false negatives as well in the report.

You can use the --callpath-cutoffoption to specify cutoff values that limit the hot callpaths reported in this section. See “--callpath-cutoff ” (page 55)“ .

Function Indexes

The Function Indexes, which follows the Call Graph, is an alphabetical list of each function in your program with its index value.

130 Producing a Sampled Call Graph Profile Analysis