Re: [PATCH 2/2] kselftest: mm: fix intermittent failure khugepaged test

[Yeoreum Yun <yeoreum.yun@arm.com>]

> 
> 
> On 9/15/26 5:21 PM, Yeoreum Yun wrote:
> > There are intermittent failures in collapse_max_ptes_swap() and
> > collapse_max_ptes_shared() when using the khugepaged_context:
> > 
> >    // while running ./khugepaged -s 2
> > 
> >    # Run test: collapse_max_ptes_shared (khugepaged:anon)
> >    # Allocate huge page... OK
> >    # Share huge page over fork()... OK
> >    # Trigger CoW on page 1023 of 2048... OK
> >    # Maybe collapse with max_ptes_shared exceeded.... OK
> >    # Trigger CoW on page 1024 of 2048... Fail
> >    Bail out! Unexpected huge page
> >    # Planned tests != run tests (26 != 23)
> >    # Totals: pass:23 fail:0 xfail:0 xpass:0 skip:0 error:0
> > 
> >    # Run test: collapse_max_ptes_swap (khugepaged:anon)
> >    # Swapout 257 of 2048 pages... OK
> >    # Maybe collapse with max_ptes_swap exceeded.... OK
> >    # Swapout 256 of 2048 pages... OK
> >    Bail out! Unexpected huge page
> >    # Planned tests != run tests (26 != 17)
> >    # Totals: pass:17 fail:0 xfail:0 xpass:0 skip:0 error:0
> > 
> > This happens because khugepaged may collapse the pages before wait_for_scan()
> > is called, causing a sanity check that expects uncollapsed pages to fail.
> > 
> > For example, in collapse_max_ptes_swap(), after faulting the pages back in
> > and paging out up to max_ptes_swap pages, khugepaged may collapse them again
> > before c->collapse() is called.
> > 
> > To prevent this, change the khugepaged setting from ALWAYS to MADVICE for
> > the affected tests, and mark the VMA with MADV_NOHUGEPAGE after it has been
> > collapsed by wait_for_scan(). This prevents khugepaged from collapsing it
> > again before c->collapse() is called.
> > 
> > This failure was observed on NVIDIA Spark with 16KB page.
> > 
> > Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
> > ---
> >   tools/testing/selftests/mm/khugepaged.c | 15 +++++++++++++++
> >   1 file changed, 15 insertions(+)
> > 
> > diff --git a/tools/testing/selftests/mm/khugepaged.c b/tools/testing/selftests/mm/khugepaged.c
> > index c32244b565658..83e9386bbc842 100644
> > --- a/tools/testing/selftests/mm/khugepaged.c
> > +++ b/tools/testing/selftests/mm/khugepaged.c
> > @@ -578,6 +578,8 @@ static bool wait_for_scan(const char *msg, char *p, size_t len,
> >   		usleep(TICK);
> >   	}
> > +	madvise(p, len, MADV_NOHUGEPAGE);
> 
> This looks incorrect to me and would reintroduce the previous problem.
> Please see commit 7962e05a835f ("selftests: khugepaged: fix the shmem
> collapse failure").

But, after the thp_enabled changed the THP_MADVISE and then fault,
there is the chance to be collpased for anon.

I overlook the shmem case. but simple could we do like:

diff --git a/tools/testing/selftests/mm/khugepaged.c b/tools/testing/selftests/mm/khugepaged.c
index c32244b565658..047524f001532 100644
--- a/tools/testing/selftests/mm/khugepaged.c
+++ b/tools/testing/selftests/mm/khugepaged.c
@@ -578,6 +578,9 @@ static bool wait_for_scan(const char *msg, char *p, size_t len,
                usleep(TICK);
        }

+       if (!strncmp(ops->name, "anon", 4))
+               madvise(p, len, MADV_NOHUGEPAGE);
+
        return timeout == -1;
 }

> 
> > +
> >   	return timeout == -1;
> >   }
> > @@ -839,6 +841,7 @@ static void collapse_swapin_single_pte(struct collapse_context *c, struct mem_op
> >   static void collapse_max_ptes_swap(struct collapse_context *c, struct mem_ops *ops)
> >   {
> > +	struct thp_settings settings = *thp_current_settings();
> >   	int max_ptes_swap = thp_read_num("khugepaged/max_ptes_swap");
> >   	void *p;
> > @@ -860,6 +863,9 @@ static void collapse_max_ptes_swap(struct collapse_context *c, struct mem_ops *o
> >   	validate_memory(p, 0, hpage_pmd_size);
> >   	if (c->enforce_pte_scan_limits) {
> > +		settings.hugepages[collapse_order].enabled = THP_MADVISE;
> > +		thp_push_settings(&settings);
> 
> I'm not sure why the collapse_order setting needs to be changed here. In
> your test case, you did not use the '-c' parameter to specify the collapse
> order.

What a stupid of me and I post wrong one... thie should be thp_enabled...
Apologise :(

> 
> > +
> >   		ops->fault(p, 0, hpage_pmd_size);
> >   		ksft_print_msg("Swapout %d of %d pages...", max_ptes_swap,
> >   		       hpage_pmd_nr);
> > @@ -869,12 +875,15 @@ static void collapse_max_ptes_swap(struct collapse_context *c, struct mem_ops *o
> >   			success("OK");
> >   		} else {
> >   			fail("Fail");
> > +			thp_pop_settings();
> >   			goto out;
> >   		}
> >   		c->collapse("Collapse with max_ptes_swap pages swapped out", p,
> >   			    1, ops, true);
> >   		validate_memory(p, 0, hpage_pmd_size);
> > +
> > +		thp_pop_settings();
> >   	}
> >   out:
> >   	ops->cleanup_area(p, hpage_pmd_size);
> > @@ -1075,6 +1084,7 @@ static void collapse_fork_compound(struct collapse_context *c, struct mem_ops *o
> >   static void collapse_max_ptes_shared(struct collapse_context *c, struct mem_ops *ops)
> >   {
> > +	struct thp_settings settings = *thp_current_settings();
> >   	int max_ptes_shared = thp_read_num("khugepaged/max_ptes_shared");
> >   	int wstatus;
> >   	void *p;
> > @@ -1100,6 +1110,9 @@ static void collapse_max_ptes_shared(struct collapse_context *c, struct mem_ops
> >   			    1, ops, !c->enforce_pte_scan_limits);
> >   		if (c->enforce_pte_scan_limits) {
> > +			settings.hugepages[collapse_order].enabled = THP_MADVISE;
> > +			thp_push_settings(&settings);
> 
> Ditto.

Sorry this should be thp_enabled...

> 
> > +
> >   			ksft_print_msg("Trigger CoW on page %d of %d...",
> >   			       hpage_pmd_nr - max_ptes_shared, hpage_pmd_nr);
> >   			ops->fault(p, 0, (hpage_pmd_nr - max_ptes_shared) *
> > @@ -1111,6 +1124,8 @@ static void collapse_max_ptes_shared(struct collapse_context *c, struct mem_ops
> >   			c->collapse("Collapse with max_ptes_shared PTEs shared",
> >   				    p, 1, ops, true);
> > +
> > +			thp_pop_settings();
> >   		}
> >   		validate_memory(p, 0, hpage_pmd_size);
> > 
> 

Thanks!

-- 
Sincerely,
Yeoreum Yun