Re: [PATCH v5 0/5] selftests/mm: separate GUP microbenchmarking from functional testing

[Sarthak Sharma <sarthak.sharma@arm.com>]

Hi Andrew!

On 7/16/26 6:02 PM, Sarthak Sharma wrote:
> gup_test.c currently serves two distinct purposes: microbenchmarking
> (GUP_FAST_BENCHMARK, PIN_FAST_BENCHMARK, PIN_LONGTERM_BENCHMARK) and
> functional correctness testing (GUP_BASIC_TEST, PIN_BASIC_TEST,
> DUMP_USER_PAGES_TEST). Mixing these in a single binary means that
> functional tests cannot be run or reported individually, and
> run_vmtests.sh must invoke the binary multiple times with different flag
> combinations to cover all configurations.
> 
> This series separates the two concerns: tools/mm/gup_bench handles
> benchmarking, while tools/testing/selftests/mm/gup_test handles functional
> testing. To avoid duplicating the HugeTLB and related file helpers, the
> series first prepares the existing helpers for sharing, then moves the
> common code to tools/lib/mm/ for use by both selftests and tools/mm.
> 

I went through Sashiko's review on this patchset [1]. There are two
concerns that it raises:

1. Patch 4: gup_thread() does not print an error when the ioctl() fails.
This is a valid point and it would be helpful to print it, but it is not
a functional bug.

2. Patch 5: This concern is a false positive. ksefltest_harness keeps
fixture teardown disabled while FIXTURE_SETUP() is running and enables
it only when setup succeeds. Therefore, FIXTURE_TEARDOWN() won't run if
setup fails.

I'll wait for people to review this patchset and if nothing else comes
up, I can send a fixup for the first concern. If there are other things
to address, I can include everything in a v6. Please let me know if you
require a different approach.

[1]
https://sashiko.dev/#/patchset/20260716123226.197736-1-sarthak.sharma%40arm.com