Re: [PATCH v3 34/54] selftests/mm: hmm-tests: add setup of HugeTLB pages

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

Hi Mike!

On 4/29/26 2:12 AM, Mike Rapoport wrote:
> From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>
> 
> hmm-tests skips HugeTLB tests if there are no free huge pages prepared by a
> wrapper script.
> 
> Add setup of HugeTLB pages to the test and make sure that the original
> settings are restored on the test exit.
> 
> Since kselftest_harness runs fixture setup and the tests in child
> processes, use HUGETLB_SETUP_DEFAULT_PAGES() that defines a constructor
> that runs in the main process and add verification that there are enough
> free huge pages to the tests that use them.
> 
> Replace exit() calls with _exit() to avoid restoring HugeTLB settings in
> the middle of test and use SIGLILL to kill a child process in
> hmm_cow_in_device test to avoid interference with signal handlers in
> hugepage_restore_settings().
> 
> Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> ---
>  tools/testing/selftests/mm/hmm-tests.c | 22 +++++++++++++---------
>  1 file changed, 13 insertions(+), 9 deletions(-)
> 
> diff --git a/tools/testing/selftests/mm/hmm-tests.c b/tools/testing/selftests/mm/hmm-tests.c
> index 409b11cad4bc..70c7ae989c01 100644
> --- a/tools/testing/selftests/mm/hmm-tests.c
> +++ b/tools/testing/selftests/mm/hmm-tests.c
> @@ -69,6 +69,9 @@ enum {
>  #ifndef FOLL_LONGTERM
>  #define FOLL_LONGTERM   0x100 /* mapping lifetime is indefinite */
>  #endif
> +
> +HUGETLB_SETUP_DEFAULT_PAGES(1)
> +
>  FIXTURE(hmm)
>  {
>  	int		fd;
> @@ -632,7 +635,7 @@ TEST_F(hmm, anon_write_child)
>  			}
>  
>  			close(child_fd);
> -			exit(0);
> +			_exit(0);
>  		}
>  	}
>  }
> @@ -712,7 +715,7 @@ TEST_F(hmm, anon_write_child_shared)
>  		ASSERT_EQ(ptr[i], -i);
>  
>  	close(child_fd);
> -	exit(0);
> +	_exit(0);
>  }
>  
>  /*
> @@ -784,8 +787,9 @@ TEST_F(hmm, anon_write_hugetlbfs)
>  	int *ptr;
>  	int ret;
>  
> -	if (!default_hsize)
> -		SKIP(return, "Huge page size could not be determined");
> +	if (!hugetlb_free_default_pages())
> +		SKIP(return, "Not enough huge pages");
> +	default_hsize = default_huge_page_size();

default_hsize has already been initialized at declaration above, so this
reassignment seems redundant. Shall we drop this?

>  
>  	size = ALIGN(TWOMEG, default_hsize);
>  	npages = size >> self->page_shift;
> @@ -1564,9 +1568,9 @@ TEST_F(hmm, compound)
>  	unsigned long i;
>  
>  	/* Skip test if we can't allocate a hugetlbfs page. */
> -
> -	if (!default_hsize)
> -		SKIP(return, "Huge page size could not be determined");
> +	if (!hugetlb_free_default_pages())
> +		SKIP(return, "Not enough huge pages");
> +	default_hsize = default_huge_page_size();

Same thing here as well.

>  
>  	size = ALIGN(TWOMEG, default_hsize);
>  	npages = size >> self->page_shift;
> @@ -2025,10 +2029,10 @@ TEST_F(hmm, hmm_cow_in_device)
>  		ptr[i] = i;
>  
>  	/* Terminate child and wait */
> -	EXPECT_EQ(0, kill(pid, SIGTERM));
> +	EXPECT_EQ(0, kill(pid, SIGKILL));
>  	EXPECT_EQ(pid, waitpid(pid, &status, 0));
>  	EXPECT_NE(0, WIFSIGNALED(status));
> -	EXPECT_EQ(SIGTERM, WTERMSIG(status));
> +	EXPECT_EQ(SIGKILL, WTERMSIG(status));
>  
>  	/* Take snapshot to CPU pagetables */
>  	ret = hmm_dmirror_cmd(self->fd, HMM_DMIRROR_SNAPSHOT, buffer, npages);