![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gprof.1 CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Power 6/32 Unix Tahoe 4.2BSD] / cci / usr / src / man / man1|
Return to gprof.1 CVS log | Up to [Power 6/32 Unix Tahoe 4.2BSD] / cci / usr / src / man / man1 |
1.1 root 1: .TH GPROF 1 "18 January 1983"
2: \" @(#)gprof.1 1.9 (Berkeley) 10/12/82
3: .UC 4
4: .SH NAME
5: gprof \- display call graph profile data
6: .SH SYNOPSIS
7: .B gprof
8: [ options ] [ a.out [ gmon.out ... ] ]
9: .SH DESCRIPTION
10: .I gprof
11: produces an execution profile of C, Pascal, or Fortran77 programs.
12: The effect of called routines is incorporated in the profile of each caller.
13: The profile data is taken from the call graph profile file
14: .RI ( gmon.out
15: default) which is created by programs
16: which are compiled with the
17: .B \-pg
18: option of
19: .IR cc ,
20: .IR pc ,
21: and
22: .IR f77 .
23: That option also links in versions of the library routines
24: which are compiled for profiling.
25: The symbol table in the
26: named object file
27: .RI ( a.out
28: default)
29: is read and correlated with the
30: call graph profile file.
31: If more than one profile file is specified,
32: the
33: .I gprof
34: output shows the sum of the profile information in the given profile files.
35: .PP
36: First, a flat profile is given,
37: similar to that provided by
38: .IR prof (1).
39: This listing gives the total execution times
40: and call counts for each of the functions
41: in the program, sorted by decreasing time.
42: .PP
43: Next, these times are propagated along the edges of the call graph.
44: Cycles are discovered, and calls into a cycle are made to share the time
45: of the cycle.
46: A second listing shows the functions
47: sorted according to the time they represent
48: including the time of their call graph descendents.
49: Below each function entry is shown its (direct) call graph children,
50: and how their times are propagated to this function.
51: A similar display above the function shows how this function's time and the
52: time of its descendents is propagated to its (direct) call graph parents.
53: .PP
54: Cycles are also shown, with an entry for the cycle as a whole and
55: a listing of the members of the cycle and their contributions to the
56: time and call counts of the cycle.
57: .PP
58: The following options are available:
59: .TP
60: .B \-a
61: suppresses the printing of statically declared functions.
62: If this option is given, all relevant information about the static function
63: .RI ( e.g.,
64: time samples, calls to other functions, calls from other functions)
65: belongs to the function loaded just before the static function in the
66: .I a.out
67: file.
68: .TP
69: .B \-b
70: supresses the printing of a description of each field in the profile.
71: .TP
72: .B \-c
73: the static call graph of the program is discovered by a heuristic
74: which examines the text space of the object file.
75: Static-only parents or children are indicated
76: with call counts of 0.
77: .TP
78: .BI \-e " name"
79: suppresses the printing of the graph profile entry for routine
80: .I name
81: and all its descendants
82: (unless they have other ancestors that aren't suppressed).
83: More than one
84: .B \-e
85: option may be given.
86: Only one
87: .I name
88: may be given with each
89: .B \-e
90: option.
91: .TP
92: .BI \-E " name"
93: suppresses the printing of the graph profile entry for routine
94: .I name
95: (and its descendants) as
96: .BR \-e ,
97: above, and also excludes the time spent in
98: .I name
99: (and its descendants) from the total and percentage time computations.
100: (For example,
101: .B \-E
102: .I mcount
103: .B \-E
104: .I mcleanup
105: is the default.)
106: .TP
107: .BI \-f " name"
108: prints the graph profile entry of only the specified routine
109: .I name
110: and its descendants.
111: More than one
112: .B \-f
113: option may be given.
114: Only one
115: .I name
116: may be given with each
117: .B \-f
118: option.
119: .TP
120: .BI \-F " name"
121: prints the graph profile entry of only the routine
122: .I name
123: and its descendants (as
124: .BR \-f,
125: above) and also uses only the times of the printed routines
126: in total time and percentage computations.
127: More than one
128: .B \-F
129: option may be given.
130: Only one
131: .I name
132: may be given with each
133: .B \-F
134: option.
135: The
136: .B \-F
137: option
138: overrides
139: the
140: .B \-E
141: option.
142: .TP
1.1.1.2 ! root 143: .B \-m
! 144: displays the time spent on
! 145: each instruction code for the 10 most
! 146: time-consuming routines.
! 147: The display format includes the address of the instruction
! 148: and time spent in both
! 149: numeric and histogram form.
! 150: .TP
1.1 root 151: .B \-s
152: a profile file
153: .I gmon.sum
154: is produced which represents
155: the sum of the profile information in all the specified profile files.
156: This summary profile file may be given to subsequent
157: executions of gprof (probably also with a
158: .BR \-s )
159: to accumulate profile data across several runs of an
160: .I a.out
161: file.
162: .TP
163: .B \-z
164: displays routines which have zero usage (as indicated by call counts
165: and accumulated time).
166: This is useful in conjunction with the
167: .B \-c
168: option for discovering which routines were never called.
169: .SH FILES
170: .ta 1.5i
171: .I a.out
172: the namelist and text space.
173: .br
174: .I gmon.out
175: dynamic call graph and profile.
176: .br
177: .I gmon.sum
178: summarized dynamic call graph and profile.
179: .SH "SEE ALSO"
180: monitor(3), profil(2), cc(1), prof(1)
181: .br
182: ``gprof: A Call Graph Execution Profiler'', by
183: Graham, S.L., Kessler, P.B., McKusick, M.K.;
184: .IR "Proceedings of the SIGPLAN '82 Symposium on Compiler Construction" ,
185: SIGPLAN Notices, Vol. 17, No. 6, pp. 120-126, June 1982.
186: .SH BUGS
187: Beware of quantization errors.
188: The granularity of the sampling is shown, but remains
189: statistical at best.
190: We assume that the time for each execution of a function
191: can be expressed by the total time for the function divided
192: by the number of times the function is called.
193: Thus the time propagated along the call graph arcs to parents of that
194: function is directly proportional to the number of times that
195: arc is traversed.
196: .PP
197: Parents which are not themselves profiled will have the time of
198: their profiled children propagated to them, but they will appear
199: to be spontaneously invoked in the call graph listing, and will
200: not have their time propagated further.
201: Similarly, signal catchers, even though profiled, will appear
202: to be spontaneous (although for more obscure reasons).
203: Any profiled children of signal catchers should have their times
204: propagated properly, unless the signal catcher was invoked during
205: the execution of the profiling routine, in which case all is lost.
206: .PP
207: The profiled program must call
208: .IR exit (2)
209: or return normally for the profiling information to be saved
210: in the gmon.out file.
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.