* srconv.c (wr_hd): Set afl field to 4 for bfd_arch_sh.
[deliverable/binutils-gdb.git] / gprof / gprof.texi
CommitLineData
be4e1cd5
JO
1\input texinfo @c -*-texinfo-*-
2@setfilename gprof.info
3@settitle GNU gprof
4@setchapternewpage odd
44c8c1d5
DZ
5
6@ifinfo
7@c This is a dir.info fragment to support semi-automated addition of
8@c manuals to an info tree. zoo@cygnus.com is developing this facility.
9@format
10START-INFO-DIR-ENTRY
5ee3dd17 11* gprof: (gprof). Profiling your program's execution
44c8c1d5
DZ
12END-INFO-DIR-ENTRY
13@end format
14@end ifinfo
15
be4e1cd5
JO
16@ifinfo
17This file documents the gprof profiler of the GNU system.
18
19Copyright (C) 1988, 1992 Free Software Foundation, Inc.
20
21Permission is granted to make and distribute verbatim copies of
22this manual provided the copyright notice and this permission notice
23are preserved on all copies.
24
25@ignore
26Permission is granted to process this file through Tex and print the
27results, provided the printed document carries copying permission
28notice identical to this one except for the removal of this paragraph
29(this paragraph not being relevant to the printed manual).
30
31@end ignore
32Permission is granted to copy and distribute modified versions of this
33manual under the conditions for verbatim copying, provided that the entire
34resulting derived work is distributed under the terms of a permission
35notice identical to this one.
36
37Permission is granted to copy and distribute translations of this manual
38into another language, under the above conditions for modified versions.
39@end ifinfo
40
41@finalout
42@smallbook
43
44@titlepage
45@title GNU gprof
46@subtitle The @sc{gnu} Profiler
47@author Jay Fenlason and Richard Stallman
48
49@page
50
51This manual describes the @sc{gnu} profiler, @code{gprof}, and how you
52can use it to determine which parts of a program are taking most of the
53execution time. We assume that you know how to write, compile, and
54execute programs. @sc{gnu} @code{gprof} was written by Jay Fenlason.
55
56This manual was edited January 1993 by Jeffrey Osier.
57
58@vskip 0pt plus 1filll
59Copyright @copyright{} 1988, 1992 Free Software Foundation, Inc.
60
61Permission is granted to make and distribute verbatim copies of
62this manual provided the copyright notice and this permission notice
63are preserved on all copies.
64
65@ignore
66Permission is granted to process this file through TeX and print the
67results, provided the printed document carries copying permission
68notice identical to this one except for the removal of this paragraph
69(this paragraph not being relevant to the printed manual).
70
71@end ignore
72Permission is granted to copy and distribute modified versions of this
73manual under the conditions for verbatim copying, provided that the entire
74resulting derived work is distributed under the terms of a permission
75notice identical to this one.
76
77Permission is granted to copy and distribute translations of this manual
78into another language, under the same conditions as for modified versions.
79
80@end titlepage
81
82@ifinfo
83@node Top
84@top Profiling a Program: Where Does It Spend Its Time?
85
86This manual describes the @sc{gnu} profiler, @code{gprof}, and how you
87can use it to determine which parts of a program are taking most of the
88execution time. We assume that you know how to write, compile, and
89execute programs. @sc{gnu} @code{gprof} was written by Jay Fenlason.
90
a2b34707
JO
91This manual was updated January 1993.
92
be4e1cd5
JO
93@menu
94* Why:: What profiling means, and why it is useful.
95* Compiling:: How to compile your program for profiling.
96* Executing:: How to execute your program to generate the
97 profile data file @file{gmon.out}.
98* Invoking:: How to run @code{gprof}, and how to specify
99 options for it.
100
101* Flat Profile:: The flat profile shows how much time was spent
102 executing directly in each function.
103* Call Graph:: The call graph shows which functions called which
104 others, and how much time each function used
105 when its subroutine calls are included.
106
107* Implementation:: How the profile data is recorded and written.
108* Sampling Error:: Statistical margins of error.
109 How to accumulate data from several runs
110 to make it more accurate.
111
112* Assumptions:: Some of @code{gprof}'s measurements are based
113 on assumptions about your program
114 that could be very wrong.
115
116* Incompatibilities:: (between GNU @code{gprof} and Unix @code{gprof}.)
117@end menu
118@end ifinfo
119
120@node Why
121@chapter Why Profile
122
123Profiling allows you to learn where your program spent its time and which
124functions called which other functions while it was executing. This
125information can show you which pieces of your program are slower than you
126expected, and might be candidates for rewriting to make your program
127execute faster. It can also tell you which functions are being called more
128or less often than you expected. This may help you spot bugs that had
129otherwise been unnoticed.
130
131Since the profiler uses information collected during the actual execution
132of your program, it can be used on programs that are too large or too
133complex to analyze by reading the source. However, how your program is run
134will affect the information that shows up in the profile data. If you
135don't use some feature of your program while it is being profiled, no
136profile information will be generated for that feature.
137
138Profiling has several steps:
139
140@itemize @bullet
141@item
142You must compile and link your program with profiling enabled.
143@xref{Compiling}.
144
145@item
146You must execute your program to generate a profile data file.
147@xref{Executing}.
148
149@item
150You must run @code{gprof} to analyze the profile data.
151@xref{Invoking}.
152@end itemize
153
154The next three chapters explain these steps in greater detail.
155
156The result of the analysis is a file containing two tables, the
157@dfn{flat profile} and the @dfn{call graph} (plus blurbs which briefly
158explain the contents of these tables).
159
160The flat profile shows how much time your program spent in each function,
161and how many times that function was called. If you simply want to know
162which functions burn most of the cycles, it is stated concisely here.
163@xref{Flat Profile}.
164
165The call graph shows, for each function, which functions called it, which
166other functions it called, and how many times. There is also an estimate
167of how much time was spent in the subroutines of each function. This can
168suggest places where you might try to eliminate function calls that use a
169lot of time. @xref{Call Graph}.
170
171@node Compiling
172@chapter Compiling a Program for Profiling
173
174The first step in generating profile information for your program is
175to compile and link it with profiling enabled.
176
177To compile a source file for profiling, specify the @samp{-pg} option when
178you run the compiler. (This is in addition to the options you normally
179use.)
180
181To link the program for profiling, if you use a compiler such as @code{cc}
182to do the linking, simply specify @samp{-pg} in addition to your usual
183options. The same option, @samp{-pg}, alters either compilation or linking
184to do what is necessary for profiling. Here are examples:
185
186@example
187cc -g -c myprog.c utils.c -pg
188cc -o myprog myprog.o utils.o -pg
189@end example
190
191The @samp{-pg} option also works with a command that both compiles and links:
192
193@example
194cc -o myprog myprog.c utils.c -g -pg
195@end example
196
197If you run the linker @code{ld} directly instead of through a compiler
198such as @code{cc}, you must specify the profiling startup file
199@file{/lib/gcrt0.o} as the first input file instead of the usual startup
200file @file{/lib/crt0.o}. In addition, you would probably want to
201specify the profiling C library, @file{/usr/lib/libc_p.a}, by writing
202@samp{-lc_p} instead of the usual @samp{-lc}. This is not absolutely
203necessary, but doing this gives you number-of-calls information for
204standard library functions such as @code{read} and @code{open}. For
205example:
206
207@example
208ld -o myprog /lib/gcrt0.o myprog.o utils.o -lc_p
209@end example
210
211If you compile only some of the modules of the program with @samp{-pg}, you
212can still profile the program, but you won't get complete information about
213the modules that were compiled without @samp{-pg}. The only information
214you get for the functions in those modules is the total time spent in them;
215there is no record of how many times they were called, or from where. This
216will not affect the flat profile (except that the @code{calls} field for
217the functions will be blank), but will greatly reduce the usefulness of the
218call graph.
219
220@node Executing
221@chapter Executing the Program to Generate Profile Data
222
223Once the program is compiled for profiling, you must run it in order to
224generate the information that @code{gprof} needs. Simply run the program
225as usual, using the normal arguments, file names, etc. The program should
226run normally, producing the same output as usual. It will, however, run
227somewhat slower than normal because of the time spent collecting and the
228writing the profile data.
229
230The way you run the program---the arguments and input that you give
231it---may have a dramatic effect on what the profile information shows. The
232profile data will describe the parts of the program that were activated for
233the particular input you use. For example, if the first command you give
234to your program is to quit, the profile data will show the time used in
235initialization and in cleanup, but not much else.
236
237You program will write the profile data into a file called @file{gmon.out}
238just before exiting. If there is already a file called @file{gmon.out},
239its contents are overwritten. There is currently no way to tell the
240program to write the profile data under a different name, but you can rename
241the file afterward if you are concerned that it may be overwritten.
242
243In order to write the @file{gmon.out} file properly, your program must exit
244normally: by returning from @code{main} or by calling @code{exit}. Calling
245the low-level function @code{_exit} does not write the profile data, and
246neither does abnormal termination due to an unhandled signal.
247
248The @file{gmon.out} file is written in the program's @emph{current working
249directory} at the time it exits. This means that if your program calls
250@code{chdir}, the @file{gmon.out} file will be left in the last directory
251your program @code{chdir}'d to. If you don't have permission to write in
252this directory, the file is not written. You may get a confusing error
253message if this happens. (We have not yet replaced the part of Unix
254responsible for this; when we do, we will make the error message
255comprehensible.)
256
257@node Invoking
258@chapter @code{gprof} Command Summary
259
260After you have a profile data file @file{gmon.out}, you can run @code{gprof}
261to interpret the information in it. The @code{gprof} program prints a
262flat profile and a call graph on standard output. Typically you would
263redirect the output of @code{gprof} into a file with @samp{>}.
264
265You run @code{gprof} like this:
266
267@smallexample
268gprof @var{options} [@var{executable-file} [@var{profile-data-files}@dots{}]] [> @var{outfile}]
269@end smallexample
270
271@noindent
272Here square-brackets indicate optional arguments.
273
274If you omit the executable file name, the file @file{a.out} is used. If
275you give no profile data file name, the file @file{gmon.out} is used. If
276any file is not in the proper format, or if the profile data file does not
277appear to belong to the executable file, an error message is printed.
278
279You can give more than one profile data file by entering all their names
280after the executable file name; then the statistics in all the data files
281are summed together.
282
283The following options may be used to selectively include or exclude
284functions in the output:
285
286@table @code
287@item -a
288The @samp{-a} option causes @code{gprof} to suppress the printing of
289statically declared (private) functions. (These are functions whose
290names are not listed as global, and which are not visible outside the
291file/function/block where they were defined.) Time spent in these
292functions, calls to/from them, etc, will all be attributed to the
293function that was loaded directly before it in the executable file.
294@c This is compatible with Unix @code{gprof}, but a bad idea.
295This option affects both the flat profile and the call graph.
296
297@item -e @var{function_name}
298The @samp{-e @var{function}} option tells @code{gprof} to not print
299information about the function @var{function_name} (and its
300children@dots{}) in the call graph. The function will still be listed
301as a child of any functions that call it, but its index number will be
302shown as @samp{[not printed]}. More than one @samp{-e} option may be
303given; only one @var{function_name} may be indicated with each @samp{-e}
304option.
305
306@item -E @var{function_name}
307The @code{-E @var{function}} option works like the @code{-e} option, but
308time spent in the function (and children who were not called from
309anywhere else), will not be used to compute the percentages-of-time for
310the call graph. More than one @samp{-E} option may be given; only one
311@var{function_name} may be indicated with each @samp{-E} option.
312
313@item -f @var{function_name}
314The @samp{-f @var{function}} option causes @code{gprof} to limit the
315call graph to the function @var{function_name} and its children (and
316their children@dots{}). More than one @samp{-f} option may be given;
317only one @var{function_name} may be indicated with each @samp{-f}
318option.
319
320@item -F @var{function_name}
321The @samp{-F @var{function}} option works like the @code{-f} option, but
322only time spent in the function and its children (and their
323children@dots{}) will be used to determine total-time and
324percentages-of-time for the call graph. More than one @samp{-F} option
325may be given; only one @var{function_name} may be indicated with each
326@samp{-F} option. The @samp{-F} option overrides the @samp{-E} option.
327
328@item -k @var{from@dots{}} @var{to@dots{}}
329The @samp{-k} option allows you to delete from the profile any arcs from
330routine @var{from} to routine @var{to}.
331
4fe2350b 332@item -v
4db865d0 333The @samp{-v} flag causes @code{gprof} to print the current version
4fe2350b
DZ
334number, and then exit.
335
be4e1cd5
JO
336@item -z
337If you give the @samp{-z} option, @code{gprof} will mention all
338functions in the flat profile, even those that were never called, and
339that had no time spent in them. This is useful in conjunction with the
340@samp{-c} option for discovering which routines were never called.
341@end table
342
343The order of these options does not matter.
344
345Note that only one function can be specified with each @code{-e},
346@code{-E}, @code{-f} or @code{-F} option. To specify more than one
347function, use multiple options. For example, this command:
348
349@example
350gprof -e boring -f foo -f bar myprogram > gprof.output
351@end example
352
353@noindent
354lists in the call graph all functions that were reached from either
355@code{foo} or @code{bar} and were not reachable from @code{boring}.
356
357There are a few other useful @code{gprof} options:
358
359@table @code
360@item -b
361If the @samp{-b} option is given, @code{gprof} doesn't print the
362verbose blurbs that try to explain the meaning of all of the fields in
363the tables. This is useful if you intend to print out the output, or
364are tired of seeing the blurbs.
365
366@item -c
367The @samp{-c} option causes the static call-graph of the program to be
368discovered by a heuristic which examines the text space of the object
369file. Static-only parents or children are indicated with call counts of
370@samp{0}.
371
372@item -d @var{num}
373The @samp{-d @var{num}} option specifies debugging options.
374@c @xref{debugging}.
375
376@item -s
377The @samp{-s} option causes @code{gprof} to summarize the information
378in the profile data files it read in, and write out a profile data
379file called @file{gmon.sum}, which contains all the information from
380the profile data files that @code{gprof} read in. The file @file{gmon.sum}
381may be one of the specified input files; the effect of this is to
382merge the data in the other input files into @file{gmon.sum}.
383@xref{Sampling Error}.
384
385Eventually you can run @code{gprof} again without @samp{-s} to analyze the
386cumulative data in the file @file{gmon.sum}.
387
388@item -T
389The @samp{-T} option causes @code{gprof} to print its output in
390``traditional'' BSD style.
391@end table
392
393@node Flat Profile
394@chapter How to Understand the Flat Profile
395@cindex flat profile
396
397The @dfn{flat profile} shows the total amount of time your program
398spent executing each function. Unless the @samp{-z} option is given,
399functions with no apparent time spent in them, and no apparent calls
400to them, are not mentioned. Note that if a function was not compiled
401for profiling, and didn't run long enough to show up on the program
402counter histogram, it will be indistinguishable from a function that
403was never called.
404
405This is part of a flat profile for a small program:
406
407@smallexample
408@group
409Flat profile:
410
411Each sample counts as 0.01 seconds.
412 % cumulative self self total
413 time seconds seconds calls ms/call ms/call name
414 33.34 0.02 0.02 7208 0.00 0.00 open
415 16.67 0.03 0.01 244 0.04 0.12 offtime
416 16.67 0.04 0.01 8 1.25 1.25 memccpy
417 16.67 0.05 0.01 7 1.43 1.43 write
418 16.67 0.06 0.01 mcount
419 0.00 0.06 0.00 236 0.00 0.00 tzset
420 0.00 0.06 0.00 192 0.00 0.00 tolower
421 0.00 0.06 0.00 47 0.00 0.00 strlen
422 0.00 0.06 0.00 45 0.00 0.00 strchr
423 0.00 0.06 0.00 1 0.00 50.00 main
424 0.00 0.06 0.00 1 0.00 0.00 memcpy
425 0.00 0.06 0.00 1 0.00 10.11 print
426 0.00 0.06 0.00 1 0.00 0.00 profil
427 0.00 0.06 0.00 1 0.00 50.00 report
428@dots{}
429@end group
430@end smallexample
431
432@noindent
433The functions are sorted by decreasing run-time spent in them. The
434functions @samp{mcount} and @samp{profil} are part of the profiling
435aparatus and appear in every flat profile; their time gives a measure of
436the amount of overhead due to profiling.
437
438The sampling period estimates the margin of error in each of the time
439figures. A time figure that is not much larger than this is not
440reliable. In this example, the @samp{self seconds} field for
441@samp{mcount} might well be @samp{0} or @samp{0.04} in another run.
442@xref{Sampling Error}, for a complete discussion.
443
444Here is what the fields in each line mean:
445
446@table @code
447@item % time
448This is the percentage of the total execution time your program spent
449in this function. These should all add up to 100%.
450
451@item cumulative seconds
452This is the cumulative total number of seconds the computer spent
453executing this functions, plus the time spent in all the functions
454above this one in this table.
455
456@item self seconds
457This is the number of seconds accounted for by this function alone.
458The flat profile listing is sorted first by this number.
459
460@item calls
461This is the total number of times the function was called. If the
462function was never called, or the number of times it was called cannot
463be determined (probably because the function was not compiled with
464profiling enabled), the @dfn{calls} field is blank.
465
466@item self ms/call
467This represents the average number of milliseconds spent in this
468function per call, if this function is profiled. Otherwise, this field
469is blank for this function.
470
471@item total ms/call
472This represents the average number of milliseconds spent in this
473function and its descendants per call, if this function is profiled.
474Otherwise, this field is blank for this function.
475
476@item name
477This is the name of the function. The flat profile is sorted by this
478field alphabetically after the @dfn{self seconds} field is sorted.
479@end table
480
481@node Call Graph
482@chapter How to Read the Call Graph
483@cindex call graph
484
485The @dfn{call graph} shows how much time was spent in each function
486and its children. From this information, you can find functions that,
487while they themselves may not have used much time, called other
488functions that did use unusual amounts of time.
489
490Here is a sample call from a small program. This call came from the
491same @code{gprof} run as the flat profile example in the previous
492chapter.
493
494@smallexample
495@group
496granularity: each sample hit covers 2 byte(s) for 20.00% of 0.05 seconds
497
498index % time self children called name
499 <spontaneous>
500[1] 100.0 0.00 0.05 start [1]
501 0.00 0.05 1/1 main [2]
502 0.00 0.00 1/2 on_exit [28]
503 0.00 0.00 1/1 exit [59]
504-----------------------------------------------
505 0.00 0.05 1/1 start [1]
506[2] 100.0 0.00 0.05 1 main [2]
507 0.00 0.05 1/1 report [3]
508-----------------------------------------------
509 0.00 0.05 1/1 main [2]
510[3] 100.0 0.00 0.05 1 report [3]
511 0.00 0.03 8/8 timelocal [6]
512 0.00 0.01 1/1 print [9]
513 0.00 0.01 9/9 fgets [12]
514 0.00 0.00 12/34 strncmp <cycle 1> [40]
515 0.00 0.00 8/8 lookup [20]
516 0.00 0.00 1/1 fopen [21]
517 0.00 0.00 8/8 chewtime [24]
518 0.00 0.00 8/16 skipspace [44]
519-----------------------------------------------
520[4] 59.8 0.01 0.02 8+472 <cycle 2 as a whole> [4]
521 0.01 0.02 244+260 offtime <cycle 2> [7]
522 0.00 0.00 236+1 tzset <cycle 2> [26]
523-----------------------------------------------
524@end group
525@end smallexample
526
527The lines full of dashes divide this table into @dfn{entries}, one for each
528function. Each entry has one or more lines.
529
530In each entry, the primary line is the one that starts with an index number
531in square brackets. The end of this line says which function the entry is
532for. The preceding lines in the entry describe the callers of this
533function and the following lines describe its subroutines (also called
534@dfn{children} when we speak of the call graph).
535
536The entries are sorted by time spent in the function and its subroutines.
537
538The internal profiling function @code{mcount} (@pxref{Flat Profile})
539is never mentioned in the call graph.
540
541@menu
542* Primary:: Details of the primary line's contents.
543* Callers:: Details of caller-lines' contents.
544* Subroutines:: Details of subroutine-lines' contents.
545* Cycles:: When there are cycles of recursion,
546 such as @code{a} calls @code{b} calls @code{a}@dots{}
547@end menu
548
549@node Primary
550@section The Primary Line
551
552The @dfn{primary line} in a call graph entry is the line that
553describes the function which the entry is about and gives the overall
554statistics for this function.
555
556For reference, we repeat the primary line from the entry for function
557@code{report} in our main example, together with the heading line that
558shows the names of the fields:
559
560@smallexample
561@group
562index % time self children called name
563@dots{}
564[3] 100.0 0.00 0.05 1 report [3]
565@end group
566@end smallexample
567
568Here is what the fields in the primary line mean:
569
570@table @code
571@item index
572Entries are numbered with consecutive integers. Each function
573therefore has an index number, which appears at the beginning of its
574primary line.
575
576Each cross-reference to a function, as a caller or subroutine of
577another, gives its index number as well as its name. The index number
578guides you if you wish to look for the entry for that function.
579
580@item % time
581This is the percentage of the total time that was spent in this
582function, including time spent in subroutines called from this
583function.
584
585The time spent in this function is counted again for the callers of
586this function. Therefore, adding up these percentages is meaningless.
587
588@item self
589This is the total amount of time spent in this function. This
590should be identical to the number printed in the @code{seconds} field
591for this function in the flat profile.
592
593@item children
594This is the total amount of time spent in the subroutine calls made by
595this function. This should be equal to the sum of all the @code{self}
596and @code{children} entries of the children listed directly below this
597function.
598
599@item called
600This is the number of times the function was called.
601
602If the function called itself recursively, there are two numbers,
603separated by a @samp{+}. The first number counts non-recursive calls,
604and the second counts recursive calls.
605
606In the example above, the function @code{report} was called once from
607@code{main}.
608
609@item name
610This is the name of the current function. The index number is
611repeated after it.
612
613If the function is part of a cycle of recursion, the cycle number is
614printed between the function's name and the index number
615(@pxref{Cycles}). For example, if function @code{gnurr} is part of
616cycle number one, and has index number twelve, its primary line would
617be end like this:
618
619@example
620gnurr <cycle 1> [12]
621@end example
622@end table
623
624@node Callers, Subroutines, Primary, Call Graph
625@section Lines for a Function's Callers
626
627A function's entry has a line for each function it was called by.
628These lines' fields correspond to the fields of the primary line, but
629their meanings are different because of the difference in context.
630
631For reference, we repeat two lines from the entry for the function
632@code{report}, the primary line and one caller-line preceding it, together
633with the heading line that shows the names of the fields:
634
635@smallexample
636index % time self children called name
637@dots{}
638 0.00 0.05 1/1 main [2]
639[3] 100.0 0.00 0.05 1 report [3]
640@end smallexample
641
642Here are the meanings of the fields in the caller-line for @code{report}
643called from @code{main}:
644
645@table @code
646@item self
647An estimate of the amount of time spent in @code{report} itself when it was
648called from @code{main}.
649
650@item children
651An estimate of the amount of time spent in subroutines of @code{report}
652when @code{report} was called from @code{main}.
653
654The sum of the @code{self} and @code{children} fields is an estimate
655of the amount of time spent within calls to @code{report} from @code{main}.
656
657@item called
658Two numbers: the number of times @code{report} was called from @code{main},
659followed by the total number of nonrecursive calls to @code{report} from
660all its callers.
661
662@item name and index number
663The name of the caller of @code{report} to which this line applies,
664followed by the caller's index number.
665
666Not all functions have entries in the call graph; some
667options to @code{gprof} request the omission of certain functions.
668When a caller has no entry of its own, it still has caller-lines
669in the entries of the functions it calls.
670
671If the caller is part of a recursion cycle, the cycle number is
672printed between the name and the index number.
673@end table
674
675If the identity of the callers of a function cannot be determined, a
676dummy caller-line is printed which has @samp{<spontaneous>} as the
677``caller's name'' and all other fields blank. This can happen for
678signal handlers.
679@c What if some calls have determinable callers' names but not all?
680@c FIXME - still relevant?
681
682@node Subroutines, Cycles, Callers, Call Graph
683@section Lines for a Function's Subroutines
684
685A function's entry has a line for each of its subroutines---in other
686words, a line for each other function that it called. These lines'
687fields correspond to the fields of the primary line, but their meanings
688are different because of the difference in context.
689
690For reference, we repeat two lines from the entry for the function
691@code{main}, the primary line and a line for a subroutine, together
692with the heading line that shows the names of the fields:
693
694@smallexample
695index % time self children called name
696@dots{}
697[2] 100.0 0.00 0.05 1 main [2]
698 0.00 0.05 1/1 report [3]
699@end smallexample
700
701Here are the meanings of the fields in the subroutine-line for @code{main}
702calling @code{report}:
703
704@table @code
705@item self
706An estimate of the amount of time spent directly within @code{report}
707when @code{report} was called from @code{main}.
708
709@item children
710An estimate of the amount of time spent in subroutines of @code{report}
711when @code{report} was called from @code{main}.
712
713The sum of the @code{self} and @code{children} fields is an estimate
714of the total time spent in calls to @code{report} from @code{main}.
715
716@item called
717Two numbers, the number of calls to @code{report} from @code{main}
718followed by the total number of nonrecursive calls to @code{report}.
719
720@item name
721The name of the subroutine of @code{main} to which this line applies,
722followed by the subroutine's index number.
723
724If the caller is part of a recursion cycle, the cycle number is
725printed between the name and the index number.
726@end table
727
728@node Cycles,, Subroutines, Call Graph
729@section How Mutually Recursive Functions Are Described
730@cindex cycle
731@cindex recursion cycle
732
733The graph may be complicated by the presence of @dfn{cycles of
734recursion} in the call graph. A cycle exists if a function calls
735another function that (directly or indirectly) calls (or appears to
736call) the original function. For example: if @code{a} calls @code{b},
737and @code{b} calls @code{a}, then @code{a} and @code{b} form a cycle.
738
739Whenever there are call-paths both ways between a pair of functions, they
740belong to the same cycle. If @code{a} and @code{b} call each other and
741@code{b} and @code{c} call each other, all three make one cycle. Note that
742even if @code{b} only calls @code{a} if it was not called from @code{a},
743@code{gprof} cannot determine this, so @code{a} and @code{b} are still
744considered a cycle.
745
746The cycles are numbered with consecutive integers. When a function
747belongs to a cycle, each time the function name appears in the call graph
748it is followed by @samp{<cycle @var{number}>}.
749
750The reason cycles matter is that they make the time values in the call
751graph paradoxical. The ``time spent in children'' of @code{a} should
752include the time spent in its subroutine @code{b} and in @code{b}'s
753subroutines---but one of @code{b}'s subroutines is @code{a}! How much of
754@code{a}'s time should be included in the children of @code{a}, when
755@code{a} is indirectly recursive?
756
757The way @code{gprof} resolves this paradox is by creating a single entry
758for the cycle as a whole. The primary line of this entry describes the
759total time spent directly in the functions of the cycle. The
760``subroutines'' of the cycle are the individual functions of the cycle, and
761all other functions that were called directly by them. The ``callers'' of
762the cycle are the functions, outside the cycle, that called functions in
763the cycle.
764
765Here is an example portion of a call graph which shows a cycle containing
766functions @code{a} and @code{b}. The cycle was entered by a call to
767@code{a} from @code{main}; both @code{a} and @code{b} called @code{c}.
768
769@smallexample
770index % time self children called name
771----------------------------------------
772 1.77 0 1/1 main [2]
773[3] 91.71 1.77 0 1+5 <cycle 1 as a whole> [3]
774 1.02 0 3 b <cycle 1> [4]
775 0.75 0 2 a <cycle 1> [5]
776----------------------------------------
777 3 a <cycle 1> [5]
778[4] 52.85 1.02 0 0 b <cycle 1> [4]
779 2 a <cycle 1> [5]
780 0 0 3/6 c [6]
781----------------------------------------
782 1.77 0 1/1 main [2]
783 2 b <cycle 1> [4]
784[5] 38.86 0.75 0 1 a <cycle 1> [5]
785 3 b <cycle 1> [4]
786 0 0 3/6 c [6]
787----------------------------------------
788@end smallexample
789
790@noindent
791(The entire call graph for this program contains in addition an entry for
792@code{main}, which calls @code{a}, and an entry for @code{c}, with callers
793@code{a} and @code{b}.)
794
795@smallexample
796index % time self children called name
797 <spontaneous>
798[1] 100.00 0 1.93 0 start [1]
799 0.16 1.77 1/1 main [2]
800----------------------------------------
801 0.16 1.77 1/1 start [1]
802[2] 100.00 0.16 1.77 1 main [2]
803 1.77 0 1/1 a <cycle 1> [5]
804----------------------------------------
805 1.77 0 1/1 main [2]
806[3] 91.71 1.77 0 1+5 <cycle 1 as a whole> [3]
807 1.02 0 3 b <cycle 1> [4]
808 0.75 0 2 a <cycle 1> [5]
809 0 0 6/6 c [6]
810----------------------------------------
811 3 a <cycle 1> [5]
812[4] 52.85 1.02 0 0 b <cycle 1> [4]
813 2 a <cycle 1> [5]
814 0 0 3/6 c [6]
815----------------------------------------
816 1.77 0 1/1 main [2]
817 2 b <cycle 1> [4]
818[5] 38.86 0.75 0 1 a <cycle 1> [5]
819 3 b <cycle 1> [4]
820 0 0 3/6 c [6]
821----------------------------------------
822 0 0 3/6 b <cycle 1> [4]
823 0 0 3/6 a <cycle 1> [5]
824[6] 0.00 0 0 6 c [6]
825----------------------------------------
826@end smallexample
827
828The @code{self} field of the cycle's primary line is the total time
829spent in all the functions of the cycle. It equals the sum of the
830@code{self} fields for the individual functions in the cycle, found
831in the entry in the subroutine lines for these functions.
832
833The @code{children} fields of the cycle's primary line and subroutine lines
834count only subroutines outside the cycle. Even though @code{a} calls
835@code{b}, the time spent in those calls to @code{b} is not counted in
836@code{a}'s @code{children} time. Thus, we do not encounter the problem of
837what to do when the time in those calls to @code{b} includes indirect
838recursive calls back to @code{a}.
839
840The @code{children} field of a caller-line in the cycle's entry estimates
841the amount of time spent @emph{in the whole cycle}, and its other
842subroutines, on the times when that caller called a function in the cycle.
843
844The @code{calls} field in the primary line for the cycle has two numbers:
845first, the number of times functions in the cycle were called by functions
846outside the cycle; second, the number of times they were called by
847functions in the cycle (including times when a function in the cycle calls
848itself). This is a generalization of the usual split into nonrecursive and
849recursive calls.
850
851The @code{calls} field of a subroutine-line for a cycle member in the
852cycle's entry says how many time that function was called from functions in
853the cycle. The total of all these is the second number in the primary line's
854@code{calls} field.
855
856In the individual entry for a function in a cycle, the other functions in
857the same cycle can appear as subroutines and as callers. These lines show
858how many times each function in the cycle called or was called from each other
859function in the cycle. The @code{self} and @code{children} fields in these
860lines are blank because of the difficulty of defining meanings for them
861when recursion is going on.
862
863@node Implementation, Sampling Error, Call Graph, Top
864@chapter Implementation of Profiling
865
866Profiling works by changing how every function in your program is compiled
867so that when it is called, it will stash away some information about where
868it was called from. From this, the profiler can figure out what function
869called it, and can count how many times it was called. This change is made
870by the compiler when your program is compiled with the @samp{-pg} option.
871
872Profiling also involves watching your program as it runs, and keeping a
873histogram of where the program counter happens to be every now and then.
874Typically the program counter is looked at around 100 times per second of
875run time, but the exact frequency may vary from system to system.
876
877A special startup routine allocates memory for the histogram and sets up
878a clock signal handler to make entries in it. Use of this special
879startup routine is one of the effects of using @samp{gcc @dots{} -pg} to
880link. The startup file also includes an @samp{exit} function which is
881responsible for writing the file @file{gmon.out}.
882
883Number-of-calls information for library routines is collected by using a
884special version of the C library. The programs in it are the same as in
885the usual C library, but they were compiled with @samp{-pg}. If you
886link your program with @samp{gcc @dots{} -pg}, it automatically uses the
887profiling version of the library.
888
889The output from @code{gprof} gives no indication of parts of your program that
890are limited by I/O or swapping bandwidth. This is because samples of the
891program counter are taken at fixed intervals of run time. Therefore, the
892time measurements in @code{gprof} output say nothing about time that your
893program was not running. For example, a part of the program that creates
894so much data that it cannot all fit in physical memory at once may run very
895slowly due to thrashing, but @code{gprof} will say it uses little time. On
896the other hand, sampling by run time has the advantage that the amount of
897load due to other users won't directly affect the output you get.
898
899@node Sampling Error, Assumptions, Implementation, Top
900@chapter Statistical Inaccuracy of @code{gprof} Output
901
902The run-time figures that @code{gprof} gives you are based on a sampling
903process, so they are subject to statistical inaccuracy. If a function runs
904only a small amount of time, so that on the average the sampling process
905ought to catch that function in the act only once, there is a pretty good
906chance it will actually find that function zero times, or twice.
907
908By contrast, the number-of-calls figures are derived by counting, not
909sampling. They are completely accurate and will not vary from run to run
910if your program is deterministic.
911
912The @dfn{sampling period} that is printed at the beginning of the flat
913profile says how often samples are taken. The rule of thumb is that a
914run-time figure is accurate if it is considerably bigger than the sampling
915period.
916
917The actual amount of error is usually more than one sampling period. In
918fact, if a value is @var{n} times the sampling period, the @emph{expected}
919error in it is the square-root of @var{n} sampling periods. If the
920sampling period is 0.01 seconds and @code{foo}'s run-time is 1 second, the
921expected error in @code{foo}'s run-time is 0.1 seconds. It is likely to
922vary this much @emph{on the average} from one profiling run to the next.
923(@emph{Sometimes} it will vary more.)
924
925This does not mean that a small run-time figure is devoid of information.
926If the program's @emph{total} run-time is large, a small run-time for one
927function does tell you that that function used an insignificant fraction of
928the whole program's time. Usually this means it is not worth optimizing.
929
930One way to get more accuracy is to give your program more (but similar)
931input data so it will take longer. Another way is to combine the data from
932several runs, using the @samp{-s} option of @code{gprof}. Here is how:
933
934@enumerate
935@item
936Run your program once.
937
938@item
939Issue the command @samp{mv gmon.out gmon.sum}.
940
941@item
942Run your program again, the same as before.
943
944@item
945Merge the new data in @file{gmon.out} into @file{gmon.sum} with this command:
946
947@example
948gprof -s @var{executable-file} gmon.out gmon.sum
949@end example
950
951@item
952Repeat the last two steps as often as you wish.
953
954@item
955Analyze the cumulative data using this command:
956
957@example
958gprof @var{executable-file} gmon.sum > @var{output-file}
959@end example
960@end enumerate
961
962@node Assumptions, Incompatibilities, Sampling Error, Top
963@chapter Estimating @code{children} Times Uses an Assumption
964
965Some of the figures in the call graph are estimates---for example, the
966@code{children} time values and all the the time figures in caller and
967subroutine lines.
968
969There is no direct information about these measurements in the profile
970data itself. Instead, @code{gprof} estimates them by making an assumption
971about your program that might or might not be true.
972
973The assumption made is that the average time spent in each call to any
974function @code{foo} is not correlated with who called @code{foo}. If
975@code{foo} used 5 seconds in all, and 2/5 of the calls to @code{foo} came
976from @code{a}, then @code{foo} contributes 2 seconds to @code{a}'s
977@code{children} time, by assumption.
978
979This assumption is usually true enough, but for some programs it is far
980from true. Suppose that @code{foo} returns very quickly when its argument
981is zero; suppose that @code{a} always passes zero as an argument, while
982other callers of @code{foo} pass other arguments. In this program, all the
983time spent in @code{foo} is in the calls from callers other than @code{a}.
984But @code{gprof} has no way of knowing this; it will blindly and
985incorrectly charge 2 seconds of time in @code{foo} to the children of
986@code{a}.
987
988@c FIXME - has this been fixed?
989We hope some day to put more complete data into @file{gmon.out}, so that
990this assumption is no longer needed, if we can figure out how. For the
991nonce, the estimated figures are usually more useful than misleading.
992
993@node Incompatibilities, , Assumptions, Top
994@chapter Incompatibilities with Unix @code{gprof}
995
996@sc{gnu} @code{gprof} and Berkeley Unix @code{gprof} use the same data
997file @file{gmon.out}, and provide essentially the same information. But
998there are a few differences.
999
1000@itemize @bullet
1001@item
1002For a recursive function, Unix @code{gprof} lists the function as a
1003parent and as a child, with a @code{calls} field that lists the number
1004of recursive calls. @sc{gnu} @code{gprof} omits these lines and puts
1005the number of recursive calls in the primary line.
1006
1007@item
1008When a function is suppressed from the call graph with @samp{-e}, @sc{gnu}
1009@code{gprof} still lists it as a subroutine of functions that call it.
1010
1011@ignore - it does this now
1012@item
1013The function names printed in @sc{gnu} @code{gprof} output do not include
1014the leading underscores that are added internally to the front of all
1015C identifiers on many operating systems.
1016@end ignore
1017
1018@item
1019The blurbs, field widths, and output formats are different. @sc{gnu}
1020@code{gprof} prints blurbs after the tables, so that you can see the
1021tables without skipping the blurbs.
1022
1023@contents
1024@bye
1025
1026NEEDS AN INDEX
1027
1028Still relevant?
1029 The @file{gmon.out} file is written in the program's @emph{current working
1030 directory} at the time it exits. This means that if your program calls
1031 @code{chdir}, the @file{gmon.out} file will be left in the last directory
1032 your program @code{chdir}'d to. If you don't have permission to write in
1033 this directory, the file is not written. You may get a confusing error
1034 message if this happens. (We have not yet replaced the part of Unix
1035 responsible for this; when we do, we will make the error message
1036 comprehensible.)
1037
1038-k from to...?
1039
1040-d debugging...? should this be documented?
1041
1042-T - "traditional BSD style": How is it different? Should the
1043differences be documented?
1044
1045what is this about? (and to think, I *wrote* it...)
1046 @item -c
1047 The @samp{-c} option causes the static call-graph of the program to be
1048 discovered by a heuristic which examines the text space of the object
1049 file. Static-only parents or children are indicated with call counts of
1050 @samp{0}.
1051
1052example flat file adds up to 100.01%...
1053
1054note: time estimates now only go out to one decimal place (0.0), where
1055they used to extend two (78.67).
This page took 0.144084 seconds and 4 git commands to generate.