On 21/08/2025 08:19, Muhammad Usama Anjum wrote:
>>> [...]
>>>
>>> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c
>>> b/tools/testing/selftests/mm/split_huge_page_test.c
>>> index 05de1fc0005b7..a85b2e393e4e8 100644
>>> --- a/tools/testing/selftests/mm/split_huge_page_test.c
>>> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
>>> @@ -296,10 +296,8 @@ void split_file_backed_thp(int order)
>>> ksft_exit_fail_msg("Unable to create a tmpfs for testing\n");
>>>
>>> status = snprintf(testfile, INPUT_MAX, "%s/thp_file", tmpfs_loc);
>>> - if (status >= INPUT_MAX) {
>>> + if (status >= INPUT_MAX)
>>> ksft_exit_fail_msg("Fail to create file-backed THP split
>>> testing file\n");
>>> - goto cleanup;
>> At that point the mount() call has succeeded so I think we do want to
>> keep the goto and just print the error message instead of calling
>> replace ksft_exit_fail_msg().
> The cleanup tag does cleanup and then calls ksft_exit_fail_msg() without
> printing the
> actual reason of failure. So current flow seems good where information about
> the error
> is being printed.
My point is that with this patch, ksft_exit_fail_msg() still causes the
test to exit right away without the cleanup steps (i.e.
unmounting/removing the directory) being executed. Printing a specific
error message is good but we should also run the cleanup steps.
- Kevin
