Re: [PATCH v2 4/4] selftests/mm: Fix test result reporting in gup_longterm

On Tue, Jun 03, 2025 at 02:36:07PM +0200, David Hildenbrand wrote:

On 27.05.25 18:04, Mark Brown wrote:

...

• int result = KSFT_PASS; int ret;
• if (fd < 0) {

• result = KSFT_FAIL;
  

• goto report;
  

• }

Not a fan of that, especially as it suddenly converts ksft_test_result_skip() -- e.g., on the memfd path -- to KSFT_FAIL.

It looks like the memfd path was an outlier here, the others all failed if they couldn't allocate the FD.

Can we just do the log_test_result(KSFT_FAIL/KSFT_SKIP) in the caller?

Nothing stopping that, or doing it for the cases where we want to skip. IIRC this simplified some of the other callers a little.