Add fault-injection unit tests (coverage part 2/3)

[brawner]

This adds unit tests to rcl_yaml_param_parser to increase coverage to 95.2%. Since the PR is so long, I'm leaving it in draft for now and will plan to refactor it into a couple of parts for easier review.

Depends on #765

Signed-off-by: Stephen Brawner [email protected]

[brawner]

This PR has now been split into several different PRs.

This depends on #771, which adds the basic unit tests.
It is also followed by #772, which adds the mocking unit test.

[Blast545]

LGTM

[brawner]

I checked these tests for memory leaks. I discovered two additional ones when running test_parser. One was easily fixable with #775. The second is allocated in the same spot, but is more perplexing to me.

==2296119== HEAP SUMMARY:
==2296119==     in use at exit: 211 bytes in 18 blocks
==2296119==   total heap usage: 5,540,035 allocs, 5,540,017 frees, 480,244,485 bytes allocated
==2296119== 
==2296119== 171 (96 direct, 75 indirect) bytes in 5 blocks are definitely lost in loss record 3 of 3
==2296119==    at 0x483FFAF: realloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==2296119==    by 0x507B2AA: __default_reallocate (allocator.c:53)
==2296119==    by 0x505F425: add_val_to_string_arr (add_to_arrays.c:107)
==2296119==    by 0x5062155: parse_value (parse.c:363)
==2296119==    by 0x5063C8B: parse_file_events (parse.c:614)
==2296119==    by 0x5065B62: rcl_parse_yaml_file (parser.c:206)
==2296119==    by 0x159ECE: RclYamlParamParser_test_parse_file_with_bad_allocator_Test::TestBody() (test_parser.cpp:344)
==2296119==    by 0x1D0717: void testing::internal::HandleSehExceptionsInMethodIfSupported<testing::Test, void>(testing::Test*, void (testing::Test::*)(), char const*) (gtest.cc:2447)
==2296119==    by 0x1C24B4: void testing::internal::HandleExceptionsInMethodIfSupported<testing::Test, void>(testing::Test*, void (testing::Test::*)(), char const*) (gtest.cc:2483)
==2296119==    by 0x175CB4: testing::Test::Run() (gtest.cc:2522)
==2296119==    by 0x1772AB: testing::TestInfo::Run() (gtest.cc:2703)
==2296119==    by 0x177FE2: testing::TestCase::Run() (gtest.cc:2825)
==2296119== 
==2296119== LEAK SUMMARY:
==2296119==    definitely lost: 96 bytes in 5 blocks
==2296119==    indirectly lost: 75 bytes in 12 blocks
==2296119==      possibly lost: 0 bytes in 0 blocks
==2296119==    still reachable: 40 bytes in 1 blocks
==2296119==         suppressed: 0 bytes in 0 blocks
==2296119== Reachable blocks (those to which a pointer was found) are not shown.
==2296119== To see them, rerun with: --leak-check=full --show-leak-kinds=all

[hidmic]

@brawner could it be a fault injection-induced leak in rcutils_string_array_fini()?

[brawner]

@hidmic Yes I believe that's correct. It might be harder to address since we need those injected faults for coverage.

[hidmic]

LGTM !

[hidmic]

@brawner nit: can be made const?

[brawner]

I think const is already deduced by auto since filenames is const.

[brawner]

Rebased onto master after merging #771

• Linux
• Linux-aarch64
• macOS
• Windows