1GCOV(1) GNU GCOV(1)
2
3
4
6 gcov - coverage testing tool
7
9 gcov [-v|--version] [-h|--help]
10 [-a|--all-blocks]
11 [-b|--branch-probabilities]
12 [-c|--branch-counts]
13 [-d|--display-progress]
14 [-f|--function-summaries]
15 [-i|--json-format]
16 [-j|--human-readable]
17 [-k|--use-colors]
18 [-l|--long-file-names]
19 [-m|--demangled-names]
20 [-n|--no-output]
21 [-o|--object-directory directory|file]
22 [-p|--preserve-paths]
23 [-q|--use-hotness-colors]
24 [-r|--relative-only]
25 [-s|--source-prefix directory]
26 [-t|--stdout]
27 [-u|--unconditional-branches]
28 [-x|--hash-filenames]
29 files
30
32 gcov is a test coverage program. Use it in concert with GCC to analyze
33 your programs to help create more efficient, faster running code and to
34 discover untested parts of your program. You can use gcov as a
35 profiling tool to help discover where your optimization efforts will
36 best affect your code. You can also use gcov along with the other
37 profiling tool, gprof, to assess which parts of your code use the
38 greatest amount of computing time.
39
40 Profiling tools help you analyze your code's performance. Using a
41 profiler such as gcov or gprof, you can find out some basic performance
42 statistics, such as:
43
44 * how often each line of code executes
45
46 * what lines of code are actually executed
47
48 * how much computing time each section of code uses
49
50 Once you know these things about how your code works when compiled, you
51 can look at each module to see which modules should be optimized. gcov
52 helps you determine where to work on optimization.
53
54 Software developers also use coverage testing in concert with
55 testsuites, to make sure software is actually good enough for a
56 release. Testsuites can verify that a program works as expected; a
57 coverage program tests to see how much of the program is exercised by
58 the testsuite. Developers can then determine what kinds of test cases
59 need to be added to the testsuites to create both better testing and a
60 better final product.
61
62 You should compile your code without optimization if you plan to use
63 gcov because the optimization, by combining some lines of code into one
64 function, may not give you as much information as you need to look for
65 `hot spots' where the code is using a great deal of computer time.
66 Likewise, because gcov accumulates statistics by line (at the lowest
67 resolution), it works best with a programming style that places only
68 one statement on each line. If you use complicated macros that expand
69 to loops or to other control structures, the statistics are less
70 helpful---they only report on the line where the macro call appears.
71 If your complex macros behave like functions, you can replace them with
72 inline functions to solve this problem.
73
74 gcov creates a logfile called sourcefile.gcov which indicates how many
75 times each line of a source file sourcefile.c has executed. You can
76 use these logfiles along with gprof to aid in fine-tuning the
77 performance of your programs. gprof gives timing information you can
78 use along with the information you get from gcov.
79
80 gcov works only on code compiled with GCC. It is not compatible with
81 any other profiling or test coverage mechanism.
82
84 -a
85 --all-blocks
86 Write individual execution counts for every basic block. Normally
87 gcov outputs execution counts only for the main blocks of a line.
88 With this option you can determine if blocks within a single line
89 are not being executed.
90
91 -b
92 --branch-probabilities
93 Write branch frequencies to the output file, and write branch
94 summary info to the standard output. This option allows you to see
95 how often each branch in your program was taken. Unconditional
96 branches will not be shown, unless the -u option is given.
97
98 -c
99 --branch-counts
100 Write branch frequencies as the number of branches taken, rather
101 than the percentage of branches taken.
102
103 -d
104 --display-progress
105 Display the progress on the standard output.
106
107 -f
108 --function-summaries
109 Output summaries for each function in addition to the file level
110 summary.
111
112 -h
113 --help
114 Display help about using gcov (on the standard output), and exit
115 without doing any further processing.
116
117 -i
118 --json-format
119 Output gcov file in an easy-to-parse JSON intermediate format which
120 does not require source code for generation. The JSON file is
121 compressed with gzip compression algorithm and the files have
122 .gcov.json.gz extension.
123
124 Structure of the JSON is following:
125
126 {
127 "current_working_directory": <current_working_directory>,
128 "data_file": <data_file>,
129 "format_version": <format_version>,
130 "gcc_version": <gcc_version>
131 "files": [<file>]
132 }
133
134 Fields of the root element have following semantics:
135
136 * current_working_directory: working directory where a
137 compilation unit was compiled
138
139 * data_file: name of the data file (GCDA)
140
141 * format_version: semantic version of the format
142
143 * gcc_version: version of the GCC compiler
144
145 Each file has the following form:
146
147 {
148 "file": <file_name>,
149 "functions": [<function>],
150 "lines": [<line>]
151 }
152
153 Fields of the file element have following semantics:
154
155 * file_name: name of the source file
156
157 Each function has the following form:
158
159 {
160 "blocks": <blocks>,
161 "blocks_executed": <blocks_executed>,
162 "demangled_name": "<demangled_name>,
163 "end_column": <end_column>,
164 "end_line": <end_line>,
165 "execution_count": <execution_count>,
166 "name": <name>,
167 "start_column": <start_column>
168 "start_line": <start_line>
169 }
170
171 Fields of the function element have following semantics:
172
173 * blocks: number of blocks that are in the function
174
175 * blocks_executed: number of executed blocks of the function
176
177 * demangled_name: demangled name of the function
178
179 * end_column: column in the source file where the function ends
180
181 * end_line: line in the source file where the function ends
182
183 * execution_count: number of executions of the function
184
185 * name: name of the function
186
187 * start_column: column in the source file where the function
188 begins
189
190 * start_line: line in the source file where the function begins
191
192 Note that line numbers and column numbers number from 1. In the
193 current implementation, start_line and start_column do not include
194 any template parameters and the leading return type but that this
195 is likely to be fixed in the future.
196
197 Each line has the following form:
198
199 {
200 "branches": [<branch>],
201 "count": <count>,
202 "line_number": <line_number>,
203 "unexecuted_block": <unexecuted_block>
204 "function_name": <function_name>,
205 }
206
207 Branches are present only with -b option. Fields of the line
208 element have following semantics:
209
210 * count: number of executions of the line
211
212 * line_number: line number
213
214 * unexecuted_block: flag whether the line contains an unexecuted
215 block (not all statements on the line are executed)
216
217 * function_name: a name of a function this line belongs to (for a
218 line with an inlined statements can be not set)
219
220 Each branch has the following form:
221
222 {
223 "count": <count>,
224 "fallthrough": <fallthrough>,
225 "throw": <throw>
226 }
227
228 Fields of the branch element have following semantics:
229
230 * count: number of executions of the branch
231
232 * fallthrough: true when the branch is a fall through branch
233
234 * throw: true when the branch is an exceptional branch
235
236 -j
237 --human-readable
238 Write counts in human readable format (like 24.6k).
239
240 -k
241 --use-colors
242 Use colors for lines of code that have zero coverage. We use red
243 color for non-exceptional lines and cyan for exceptional. Same
244 colors are used for basic blocks with -a option.
245
246 -l
247 --long-file-names
248 Create long file names for included source files. For example, if
249 the header file x.h contains code, and was included in the file
250 a.c, then running gcov on the file a.c will produce an output file
251 called a.c##x.h.gcov instead of x.h.gcov. This can be useful if
252 x.h is included in multiple source files and you want to see the
253 individual contributions. If you use the -p option, both the
254 including and included file names will be complete path names.
255
256 -m
257 --demangled-names
258 Display demangled function names in output. The default is to show
259 mangled function names.
260
261 -n
262 --no-output
263 Do not create the gcov output file.
264
265 -o directory|file
266 --object-directory directory
267 --object-file file
268 Specify either the directory containing the gcov data files, or the
269 object path name. The .gcno, and .gcda data files are searched for
270 using this option. If a directory is specified, the data files are
271 in that directory and named after the input file name, without its
272 extension. If a file is specified here, the data files are named
273 after that file, without its extension.
274
275 -p
276 --preserve-paths
277 Preserve complete path information in the names of generated .gcov
278 files. Without this option, just the filename component is used.
279 With this option, all directories are used, with / characters
280 translated to # characters, . directory components removed and
281 unremoveable .. components renamed to ^. This is useful if
282 sourcefiles are in several different directories.
283
284 -q
285 --use-hotness-colors
286 Emit perf-like colored output for hot lines. Legend of the color
287 scale is printed at the very beginning of the output file.
288
289 -r
290 --relative-only
291 Only output information about source files with a relative pathname
292 (after source prefix elision). Absolute paths are usually system
293 header files and coverage of any inline functions therein is
294 normally uninteresting.
295
296 -s directory
297 --source-prefix directory
298 A prefix for source file names to remove when generating the output
299 coverage files. This option is useful when building in a separate
300 directory, and the pathname to the source directory is not wanted
301 when determining the output file names. Note that this prefix
302 detection is applied before determining whether the source file is
303 absolute.
304
305 -t
306 --stdout
307 Output to standard output instead of output files.
308
309 -u
310 --unconditional-branches
311 When branch probabilities are given, include those of unconditional
312 branches. Unconditional branches are normally not interesting.
313
314 -v
315 --version
316 Display the gcov version number (on the standard output), and exit
317 without doing any further processing.
318
319 -w
320 --verbose
321 Print verbose informations related to basic blocks and arcs.
322
323 -x
324 --hash-filenames
325 When using --preserve-paths, gcov uses the full pathname of the
326 source files to create an output filename. This can lead to long
327 filenames that can overflow filesystem limits. This option creates
328 names of the form source-file##md5.gcov, where the source-file
329 component is the final filename part and the md5 component is
330 calculated from the full mangled name that would have been used
331 otherwise. The option is an alternative to the --preserve-paths on
332 systems which have a filesystem limit.
333
334 gcov should be run with the current directory the same as that when you
335 invoked the compiler. Otherwise it will not be able to locate the
336 source files. gcov produces files called mangledname.gcov in the
337 current directory. These contain the coverage information of the
338 source file they correspond to. One .gcov file is produced for each
339 source (or header) file containing code, which was compiled to produce
340 the data files. The mangledname part of the output file name is
341 usually simply the source file name, but can be something more
342 complicated if the -l or -p options are given. Refer to those options
343 for details.
344
345 If you invoke gcov with multiple input files, the contributions from
346 each input file are summed. Typically you would invoke it with the
347 same list of files as the final link of your executable.
348
349 The .gcov files contain the : separated fields along with program
350 source code. The format is
351
352 <execution_count>:<line_number>:<source line text>
353
354 Additional block information may succeed each line, when requested by
355 command line option. The execution_count is - for lines containing no
356 code. Unexecuted lines are marked ##### or =====, depending on whether
357 they are reachable by non-exceptional paths or only exceptional paths
358 such as C++ exception handlers, respectively. Given the -a option,
359 unexecuted blocks are marked $$$$$ or %%%%%, depending on whether a
360 basic block is reachable via non-exceptional or exceptional paths.
361 Executed basic blocks having a statement with zero execution_count end
362 with * character and are colored with magenta color with the -k option.
363 This functionality is not supported in Ada.
364
365 Note that GCC can completely remove the bodies of functions that are
366 not needed -- for instance if they are inlined everywhere. Such
367 functions are marked with -, which can be confusing. Use the
368 -fkeep-inline-functions and -fkeep-static-functions options to retain
369 these functions and allow gcov to properly show their execution_count.
370
371 Some lines of information at the start have line_number of zero. These
372 preamble lines are of the form
373
374 -:0:<tag>:<value>
375
376 The ordering and number of these preamble lines will be augmented as
377 gcov development progresses --- do not rely on them remaining
378 unchanged. Use tag to locate a particular preamble line.
379
380 The additional block information is of the form
381
382 <tag> <information>
383
384 The information is human readable, but designed to be simple enough for
385 machine parsing too.
386
387 When printing percentages, 0% and 100% are only printed when the values
388 are exactly 0% and 100% respectively. Other values which would
389 conventionally be rounded to 0% or 100% are instead printed as the
390 nearest non-boundary value.
391
392 When using gcov, you must first compile your program with a special GCC
393 option --coverage. This tells the compiler to generate additional
394 information needed by gcov (basically a flow graph of the program) and
395 also includes additional code in the object files for generating the
396 extra profiling information needed by gcov. These additional files are
397 placed in the directory where the object file is located.
398
399 Running the program will cause profile output to be generated. For
400 each source file compiled with -fprofile-arcs, an accompanying .gcda
401 file will be placed in the object file directory.
402
403 Running gcov with your program's source file names as arguments will
404 now produce a listing of the code along with frequency of execution for
405 each line. For example, if your program is called tmp.cpp, this is
406 what you see when you use the basic gcov facility:
407
408 $ g++ --coverage tmp.cpp
409 $ a.out
410 $ gcov tmp.cpp -m
411 File 'tmp.cpp'
412 Lines executed:92.86% of 14
413 Creating 'tmp.cpp.gcov'
414
415 The file tmp.cpp.gcov contains output from gcov. Here is a sample:
416
417 -: 0:Source:tmp.cpp
418 -: 0:Working directory:/home/gcc/testcase
419 -: 0:Graph:tmp.gcno
420 -: 0:Data:tmp.gcda
421 -: 0:Runs:1
422 -: 0:Programs:1
423 -: 1:#include <stdio.h>
424 -: 2:
425 -: 3:template<class T>
426 -: 4:class Foo
427 -: 5:{
428 -: 6: public:
429 1*: 7: Foo(): b (1000) {}
430 ------------------
431 Foo<char>::Foo():
432 #####: 7: Foo(): b (1000) {}
433 ------------------
434 Foo<int>::Foo():
435 1: 7: Foo(): b (1000) {}
436 ------------------
437 2*: 8: void inc () { b++; }
438 ------------------
439 Foo<char>::inc():
440 #####: 8: void inc () { b++; }
441 ------------------
442 Foo<int>::inc():
443 2: 8: void inc () { b++; }
444 ------------------
445 -: 9:
446 -: 10: private:
447 -: 11: int b;
448 -: 12:};
449 -: 13:
450 -: 14:template class Foo<int>;
451 -: 15:template class Foo<char>;
452 -: 16:
453 -: 17:int
454 1: 18:main (void)
455 -: 19:{
456 -: 20: int i, total;
457 1: 21: Foo<int> counter;
458 -: 22:
459 1: 23: counter.inc();
460 1: 24: counter.inc();
461 1: 25: total = 0;
462 -: 26:
463 11: 27: for (i = 0; i < 10; i++)
464 10: 28: total += i;
465 -: 29:
466 1*: 30: int v = total > 100 ? 1 : 2;
467 -: 31:
468 1: 32: if (total != 45)
469 #####: 33: printf ("Failure\n");
470 -: 34: else
471 1: 35: printf ("Success\n");
472 1: 36: return 0;
473 -: 37:}
474
475 Note that line 7 is shown in the report multiple times. First
476 occurrence presents total number of execution of the line and the next
477 two belong to instances of class Foo constructors. As you can also
478 see, line 30 contains some unexecuted basic blocks and thus execution
479 count has asterisk symbol.
480
481 When you use the -a option, you will get individual block counts, and
482 the output looks like this:
483
484 -: 0:Source:tmp.cpp
485 -: 0:Working directory:/home/gcc/testcase
486 -: 0:Graph:tmp.gcno
487 -: 0:Data:tmp.gcda
488 -: 0:Runs:1
489 -: 0:Programs:1
490 -: 1:#include <stdio.h>
491 -: 2:
492 -: 3:template<class T>
493 -: 4:class Foo
494 -: 5:{
495 -: 6: public:
496 1*: 7: Foo(): b (1000) {}
497 ------------------
498 Foo<char>::Foo():
499 #####: 7: Foo(): b (1000) {}
500 ------------------
501 Foo<int>::Foo():
502 1: 7: Foo(): b (1000) {}
503 ------------------
504 2*: 8: void inc () { b++; }
505 ------------------
506 Foo<char>::inc():
507 #####: 8: void inc () { b++; }
508 ------------------
509 Foo<int>::inc():
510 2: 8: void inc () { b++; }
511 ------------------
512 -: 9:
513 -: 10: private:
514 -: 11: int b;
515 -: 12:};
516 -: 13:
517 -: 14:template class Foo<int>;
518 -: 15:template class Foo<char>;
519 -: 16:
520 -: 17:int
521 1: 18:main (void)
522 -: 19:{
523 -: 20: int i, total;
524 1: 21: Foo<int> counter;
525 1: 21-block 0
526 -: 22:
527 1: 23: counter.inc();
528 1: 23-block 0
529 1: 24: counter.inc();
530 1: 24-block 0
531 1: 25: total = 0;
532 -: 26:
533 11: 27: for (i = 0; i < 10; i++)
534 1: 27-block 0
535 11: 27-block 1
536 10: 28: total += i;
537 10: 28-block 0
538 -: 29:
539 1*: 30: int v = total > 100 ? 1 : 2;
540 1: 30-block 0
541 %%%%%: 30-block 1
542 1: 30-block 2
543 -: 31:
544 1: 32: if (total != 45)
545 1: 32-block 0
546 #####: 33: printf ("Failure\n");
547 %%%%%: 33-block 0
548 -: 34: else
549 1: 35: printf ("Success\n");
550 1: 35-block 0
551 1: 36: return 0;
552 1: 36-block 0
553 -: 37:}
554
555 In this mode, each basic block is only shown on one line -- the last
556 line of the block. A multi-line block will only contribute to the
557 execution count of that last line, and other lines will not be shown to
558 contain code, unless previous blocks end on those lines. The total
559 execution count of a line is shown and subsequent lines show the
560 execution counts for individual blocks that end on that line. After
561 each block, the branch and call counts of the block will be shown, if
562 the -b option is given.
563
564 Because of the way GCC instruments calls, a call count can be shown
565 after a line with no individual blocks. As you can see, line 33
566 contains a basic block that was not executed.
567
568 When you use the -b option, your output looks like this:
569
570 -: 0:Source:tmp.cpp
571 -: 0:Working directory:/home/gcc/testcase
572 -: 0:Graph:tmp.gcno
573 -: 0:Data:tmp.gcda
574 -: 0:Runs:1
575 -: 0:Programs:1
576 -: 1:#include <stdio.h>
577 -: 2:
578 -: 3:template<class T>
579 -: 4:class Foo
580 -: 5:{
581 -: 6: public:
582 1*: 7: Foo(): b (1000) {}
583 ------------------
584 Foo<char>::Foo():
585 function Foo<char>::Foo() called 0 returned 0% blocks executed 0%
586 #####: 7: Foo(): b (1000) {}
587 ------------------
588 Foo<int>::Foo():
589 function Foo<int>::Foo() called 1 returned 100% blocks executed 100%
590 1: 7: Foo(): b (1000) {}
591 ------------------
592 2*: 8: void inc () { b++; }
593 ------------------
594 Foo<char>::inc():
595 function Foo<char>::inc() called 0 returned 0% blocks executed 0%
596 #####: 8: void inc () { b++; }
597 ------------------
598 Foo<int>::inc():
599 function Foo<int>::inc() called 2 returned 100% blocks executed 100%
600 2: 8: void inc () { b++; }
601 ------------------
602 -: 9:
603 -: 10: private:
604 -: 11: int b;
605 -: 12:};
606 -: 13:
607 -: 14:template class Foo<int>;
608 -: 15:template class Foo<char>;
609 -: 16:
610 -: 17:int
611 function main called 1 returned 100% blocks executed 81%
612 1: 18:main (void)
613 -: 19:{
614 -: 20: int i, total;
615 1: 21: Foo<int> counter;
616 call 0 returned 100%
617 branch 1 taken 100% (fallthrough)
618 branch 2 taken 0% (throw)
619 -: 22:
620 1: 23: counter.inc();
621 call 0 returned 100%
622 branch 1 taken 100% (fallthrough)
623 branch 2 taken 0% (throw)
624 1: 24: counter.inc();
625 call 0 returned 100%
626 branch 1 taken 100% (fallthrough)
627 branch 2 taken 0% (throw)
628 1: 25: total = 0;
629 -: 26:
630 11: 27: for (i = 0; i < 10; i++)
631 branch 0 taken 91% (fallthrough)
632 branch 1 taken 9%
633 10: 28: total += i;
634 -: 29:
635 1*: 30: int v = total > 100 ? 1 : 2;
636 branch 0 taken 0% (fallthrough)
637 branch 1 taken 100%
638 -: 31:
639 1: 32: if (total != 45)
640 branch 0 taken 0% (fallthrough)
641 branch 1 taken 100%
642 #####: 33: printf ("Failure\n");
643 call 0 never executed
644 branch 1 never executed
645 branch 2 never executed
646 -: 34: else
647 1: 35: printf ("Success\n");
648 call 0 returned 100%
649 branch 1 taken 100% (fallthrough)
650 branch 2 taken 0% (throw)
651 1: 36: return 0;
652 -: 37:}
653
654 For each function, a line is printed showing how many times the
655 function is called, how many times it returns and what percentage of
656 the function's blocks were executed.
657
658 For each basic block, a line is printed after the last line of the
659 basic block describing the branch or call that ends the basic block.
660 There can be multiple branches and calls listed for a single source
661 line if there are multiple basic blocks that end on that line. In this
662 case, the branches and calls are each given a number. There is no
663 simple way to map these branches and calls back to source constructs.
664 In general, though, the lowest numbered branch or call will correspond
665 to the leftmost construct on the source line.
666
667 For a branch, if it was executed at least once, then a percentage
668 indicating the number of times the branch was taken divided by the
669 number of times the branch was executed will be printed. Otherwise,
670 the message "never executed" is printed.
671
672 For a call, if it was executed at least once, then a percentage
673 indicating the number of times the call returned divided by the number
674 of times the call was executed will be printed. This will usually be
675 100%, but may be less for functions that call "exit" or "longjmp", and
676 thus may not return every time they are called.
677
678 The execution counts are cumulative. If the example program were
679 executed again without removing the .gcda file, the count for the
680 number of times each line in the source was executed would be added to
681 the results of the previous run(s). This is potentially useful in
682 several ways. For example, it could be used to accumulate data over a
683 number of program runs as part of a test verification suite, or to
684 provide more accurate long-term information over a large number of
685 program runs.
686
687 The data in the .gcda files is saved immediately before the program
688 exits. For each source file compiled with -fprofile-arcs, the
689 profiling code first attempts to read in an existing .gcda file; if the
690 file doesn't match the executable (differing number of basic block
691 counts) it will ignore the contents of the file. It then adds in the
692 new execution counts and finally writes the data to the file.
693
694 Using gcov with GCC Optimization
695 If you plan to use gcov to help optimize your code, you must first
696 compile your program with a special GCC option --coverage. Aside from
697 that, you can use any other GCC options; but if you want to prove that
698 every single line in your program was executed, you should not compile
699 with optimization at the same time. On some machines the optimizer can
700 eliminate some simple code lines by combining them with other lines.
701 For example, code like this:
702
703 if (a != b)
704 c = 1;
705 else
706 c = 0;
707
708 can be compiled into one instruction on some machines. In this case,
709 there is no way for gcov to calculate separate execution counts for
710 each line because there isn't separate code for each line. Hence the
711 gcov output looks like this if you compiled the program with
712 optimization:
713
714 100: 12:if (a != b)
715 100: 13: c = 1;
716 100: 14:else
717 100: 15: c = 0;
718
719 The output shows that this block of code, combined by optimization,
720 executed 100 times. In one sense this result is correct, because there
721 was only one instruction representing all four of these lines.
722 However, the output does not indicate how many times the result was 0
723 and how many times the result was 1.
724
725 Inlineable functions can create unexpected line counts. Line counts
726 are shown for the source code of the inlineable function, but what is
727 shown depends on where the function is inlined, or if it is not inlined
728 at all.
729
730 If the function is not inlined, the compiler must emit an out of line
731 copy of the function, in any object file that needs it. If fileA.o and
732 fileB.o both contain out of line bodies of a particular inlineable
733 function, they will also both contain coverage counts for that
734 function. When fileA.o and fileB.o are linked together, the linker
735 will, on many systems, select one of those out of line bodies for all
736 calls to that function, and remove or ignore the other. Unfortunately,
737 it will not remove the coverage counters for the unused function body.
738 Hence when instrumented, all but one use of that function will show
739 zero counts.
740
741 If the function is inlined in several places, the block structure in
742 each location might not be the same. For instance, a condition might
743 now be calculable at compile time in some instances. Because the
744 coverage of all the uses of the inline function will be shown for the
745 same source lines, the line counts themselves might seem inconsistent.
746
747 Long-running applications can use the "__gcov_reset" and "__gcov_dump"
748 facilities to restrict profile collection to the program region of
749 interest. Calling "__gcov_reset(void)" will clear all profile counters
750 to zero, and calling "__gcov_dump(void)" will cause the profile
751 information collected at that point to be dumped to .gcda output files.
752 Instrumented applications use a static destructor with priority 99 to
753 invoke the "__gcov_dump" function. Thus "__gcov_dump" is executed after
754 all user defined static destructors, as well as handlers registered
755 with "atexit". If an executable loads a dynamic shared object via
756 dlopen functionality, -Wl,--dynamic-list-data is needed to dump all
757 profile data.
758
759 Profiling run-time library reports various errors related to profile
760 manipulation and profile saving. Errors are printed into standard
761 error output or GCOV_ERROR_FILE file, if environment variable is used.
762 In order to terminate immediately after an errors occurs set
763 GCOV_EXIT_AT_ERROR environment variable. That can help users to find
764 profile clashing which leads to a misleading profile.
765
767 gpl(7), gfdl(7), fsf-funding(7), gcc(1) and the Info entry for gcc.
768
770 Copyright (c) 1996-2020 Free Software Foundation, Inc.
771
772 Permission is granted to copy, distribute and/or modify this document
773 under the terms of the GNU Free Documentation License, Version 1.3 or
774 any later version published by the Free Software Foundation; with the
775 Invariant Sections being "GNU General Public License" and "Funding Free
776 Software", the Front-Cover texts being (a) (see below), and with the
777 Back-Cover Texts being (b) (see below). A copy of the license is
778 included in the gfdl(7) man page.
779
780 (a) The FSF's Front-Cover Text is:
781
782 A GNU Manual
783
784 (b) The FSF's Back-Cover Text is:
785
786 You have freedom to copy and modify this GNU Manual, like GNU
787 software. Copies published by the Free Software Foundation raise
788 funds for GNU development.
789
790
791
792gcc-10.2.1 2020-07-23 GCOV(1)