| Age | Commit message (Collapse) | Author |
|---|
|
Add a line after the code coverage summary printing the filters
that were used to produce the results, if any.
This is important specially when multiple filters could be used
to produce different reports.
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
|
While lcov has already its own report generator, it is interesting
to be able to deal with multiple files exposing each input in
separate.
So, add new command line parameters to allow it to generate html
reports. Also add some command lines to setup html title, add a
css file and include a prolog/epilog at the html body.
The title option can also be useful to rename the titles when
merging multiple info files.
Acked-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
|
It is interesting to have support not only for including, but
also for excluding functions and files. Also, it is trivial to
have support for it.
When either one or both source/function filters are enabled, it
will print at the console the regular expressions that are applied
to functions and sources (if any), e. g.:
$ cat << END >f1
+i915
gem
- display
-selftest
END
$ cat << END >f2
-/selftests
END
$ code_cov_parse_info dg1.info dg2.info --func-filters f1 --source-filters f2 --stat
lines......: 72.1% (176 of 244 lines)
functions..: 77.3% (17 of 22 functions)
branches...: 50.0% (68 of 136 branches)
Filters......: function regex (not match: m`display` m`selftest` and match: m`i915` m`gem`), source regex (match: m`/selftest`) and ignored source files where none of its code ran.
Source files.: 0.99% (7 of 710 total), 77.78% (7 of 9 filtered)
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
|
print_code_coverage function needs to first read the number
of lines and then simplify the $file by removing its prefix.
Without that, it ends trying to access an uninitiated value,
leading to warnings and not doing what's expected.
While here, also check if the ln data is valid, just in case
the .info file might have some FN field(s) missing.
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
|
While on bash scripts we use "||" for error handling, perl
accepts it, but it has a higher priority, thus would require
to use parenthesis before that.
That causes it to print a warning when the open fail while
reading filter files, instead of the original intent of
finishing and reporting an error.
So, let's use "or" instead.
Reviewed-by: Ch Sai Gowtham <sai.gowtham.ch@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
|
The TN field generated by standard lcov is weird: it keeps
repeating the TN field from time to time. At genhtml, it seems
that only the first one is used.
As we're using TN to indicate the test name, preserve all different
test names at the output file.
Also, printing such names doesn't really makes sense when
--print-used and --print-unused command line options are used,
and printing a list of 100+ names won't make much sense.
So, just remove printing the test names.
Reviewed-by: Ch Sai Gowtham <sai.gowtham.ch@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
|
The code coverage info files provide useful information about what
functions were called and how many times. It can also contain data that
it is not useful for the code coverage, like calls to non-DRM related
Kernel APIs.
Add a tool that helps filtering out non-DRM files and prints what
functions were called.
Both the stats, prints and output files are affected by the filters:
$ code_cov_parse_info --stat total.info
lines......: 40.8% (71989 of 176442 lines)
functions..: 50.2% (5975 of 11909 functions)
branches...: 28.7% (35006 of 121893 branches)
Source files: 744
$ code_cov_parse_info total.info --stat --only-drm
lines......: 40.3% (70244 of 174170 lines)
functions..: 50.8% (5835 of 11491 functions)
branches...: 28.4% (33096 of 116707 branches)
Ignored......: non-drm headers.
Source files.: 78.36% (583 of 744 total)
$ code_cov_parse_info total.info --stat --only-drm --ignore-unused -o o.info
lines......: 47.9% (70244 of 146752 lines)
functions..: 50.8% (5835 of 11491 functions)
branches...: 32.8% (33096 of 100900 branches)
Ignored......: non-drm headers and source files where none of its code ran.
Source files.: 58.87% (438 of 744 total), 75.13% (438 of 583 filtered)
$ code_cov_parse_info --stat o.info
lines......: 47.9% (70244 of 146752 lines)
functions..: 59.8% (5835 of 9763 functions)
branches...: 32.8% (33096 of 100900 branches)
Source files: 438
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
|
