selftests/mm: fix split_huge_page_test for folio_split() tests.

[PATCH] selftests/mm: fix split_huge_page_test for folio_split() tests.

Posted by Zi Yan 7 months ago

PID_FMT does not have an offset field, so folio_split() tests are not
performed. Add PID_FMT_OFFSET with an offset field and use it to perform
folio_split() tests.

Fixes: 80a5c494c89f ("selftests/mm: add tests for folio_split(), buddy allocator like split")
Signed-off-by: Zi Yan <ziy@nvidia.com>
---
 tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
index aa7400ed0e99..f0d9c035641d 100644
--- a/tools/testing/selftests/mm/split_huge_page_test.c
+++ b/tools/testing/selftests/mm/split_huge_page_test.c
@@ -31,6 +31,7 @@ uint64_t pmd_pagesize;
 #define INPUT_MAX 80
 
 #define PID_FMT "%d,0x%lx,0x%lx,%d"
+#define PID_FMT_OFFSET "%d,0x%lx,0x%lx,%d,%d"
 #define PATH_FMT "%s,0x%lx,0x%lx,%d"
 
 #define PFN_MASK     ((1UL<<55)-1)
@@ -483,7 +484,7 @@ void split_thp_in_pagecache_to_order_at(size_t fd_size, const char *fs_loc,
 		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
 			      (uint64_t)addr + fd_size, order);
 	else
-		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
+		write_debugfs(PID_FMT_OFFSET, getpid(), (uint64_t)addr,
 			      (uint64_t)addr + fd_size, order, offset);
 
 	for (i = 0; i < fd_size; i++)
-- 
2.47.2

Re: [PATCH] selftests/mm: fix split_huge_page_test for folio_split() tests.

Posted by David Hildenbrand 7 months ago

On 09.07.25 03:27, Zi Yan wrote:
> PID_FMT does not have an offset field, so folio_split() tests are not
> performed. Add PID_FMT_OFFSET with an offset field and use it to perform
> folio_split() tests.
> 
> Fixes: 80a5c494c89f ("selftests/mm: add tests for folio_split(), buddy allocator like split")
> Signed-off-by: Zi Yan <ziy@nvidia.com>
> ---
>   tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
> index aa7400ed0e99..f0d9c035641d 100644
> --- a/tools/testing/selftests/mm/split_huge_page_test.c
> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
> @@ -31,6 +31,7 @@ uint64_t pmd_pagesize;
>   #define INPUT_MAX 80
>   
>   #define PID_FMT "%d,0x%lx,0x%lx,%d"
> +#define PID_FMT_OFFSET "%d,0x%lx,0x%lx,%d,%d"
>   #define PATH_FMT "%s,0x%lx,0x%lx,%d"
>   
>   #define PFN_MASK     ((1UL<<55)-1)
> @@ -483,7 +484,7 @@ void split_thp_in_pagecache_to_order_at(size_t fd_size, const char *fs_loc,
>   		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
>   			      (uint64_t)addr + fd_size, order);
>   	else
> -		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
> +		write_debugfs(PID_FMT_OFFSET, getpid(), (uint64_t)addr,
>   			      (uint64_t)addr + fd_size, order, offset);
>   
>   	for (i = 0; i < fd_size; i++)

So I assume the tests still passed. Would there be a way to have made 
them fail? (IOW, detect that the wrong kind-of split was performed)

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb

Re: [PATCH] selftests/mm: fix split_huge_page_test for folio_split() tests.

Posted by Zi Yan 7 months ago

On 11 Jul 2025, at 9:29, David Hildenbrand wrote:

> On 09.07.25 03:27, Zi Yan wrote:
>> PID_FMT does not have an offset field, so folio_split() tests are not
>> performed. Add PID_FMT_OFFSET with an offset field and use it to perform
>> folio_split() tests.
>>
>> Fixes: 80a5c494c89f ("selftests/mm: add tests for folio_split(), buddy allocator like split")
>> Signed-off-by: Zi Yan <ziy@nvidia.com>
>> ---
>>   tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
>> index aa7400ed0e99..f0d9c035641d 100644
>> --- a/tools/testing/selftests/mm/split_huge_page_test.c
>> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
>> @@ -31,6 +31,7 @@ uint64_t pmd_pagesize;
>>   #define INPUT_MAX 80
>>    #define PID_FMT "%d,0x%lx,0x%lx,%d"
>> +#define PID_FMT_OFFSET "%d,0x%lx,0x%lx,%d,%d"
>>   #define PATH_FMT "%s,0x%lx,0x%lx,%d"
>>    #define PFN_MASK     ((1UL<<55)-1)
>> @@ -483,7 +484,7 @@ void split_thp_in_pagecache_to_order_at(size_t fd_size, const char *fs_loc,
>>   		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
>>   			      (uint64_t)addr + fd_size, order);
>>   	else
>> -		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
>> +		write_debugfs(PID_FMT_OFFSET, getpid(), (uint64_t)addr,
>>   			      (uint64_t)addr + fd_size, order, offset);
>>    	for (i = 0; i < fd_size; i++)
>
> So I assume the tests still passed. Would there be a way to have made them fail? (IOW, detect that the wrong kind-of split was performed)

Only if the initial PMD page is not split. To detect whether all after-split
folios are the size we expect, additional code that reads kpageflags
is needed. I will put it in my TODO to make this test more robust.

>
> Acked-by: David Hildenbrand <david@redhat.com>

Thanks.

Best Regards,
Yan, Zi

Re: [PATCH] selftests/mm: fix split_huge_page_test for folio_split() tests.

Posted by Lorenzo Stoakes 7 months ago

On Tue, Jul 08, 2025 at 09:27:59PM -0400, Zi Yan wrote:
> PID_FMT does not have an offset field, so folio_split() tests are not
> performed. Add PID_FMT_OFFSET with an offset field and use it to perform
> folio_split() tests.
>
> Fixes: 80a5c494c89f ("selftests/mm: add tests for folio_split(), buddy allocator like split")
> Signed-off-by: Zi Yan <ziy@nvidia.com>

LGTM, so:

Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>

> ---
>  tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
> index aa7400ed0e99..f0d9c035641d 100644
> --- a/tools/testing/selftests/mm/split_huge_page_test.c
> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
> @@ -31,6 +31,7 @@ uint64_t pmd_pagesize;
>  #define INPUT_MAX 80
>
>  #define PID_FMT "%d,0x%lx,0x%lx,%d"
> +#define PID_FMT_OFFSET "%d,0x%lx,0x%lx,%d,%d"
>  #define PATH_FMT "%s,0x%lx,0x%lx,%d"
>
>  #define PFN_MASK     ((1UL<<55)-1)
> @@ -483,7 +484,7 @@ void split_thp_in_pagecache_to_order_at(size_t fd_size, const char *fs_loc,
>  		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
>  			      (uint64_t)addr + fd_size, order);
>  	else
> -		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
> +		write_debugfs(PID_FMT_OFFSET, getpid(), (uint64_t)addr,
>  			      (uint64_t)addr + fd_size, order, offset);
>
>  	for (i = 0; i < fd_size; i++)
> --
> 2.47.2
>

Re: [PATCH] selftests/mm: fix split_huge_page_test for folio_split() tests.

Posted by Donet Tom 7 months ago

On 7/9/25 6:57 AM, Zi Yan wrote:

Hi Zi Yan

> PID_FMT does not have an offset field, so folio_split() tests are not
> performed. Add PID_FMT_OFFSET with an offset field and use it to perform
> folio_split() tests.
>
> Fixes: 80a5c494c89f ("selftests/mm: add tests for folio_split(), buddy allocator like split")
> Signed-off-by: Zi Yan <ziy@nvidia.com>
> ---
>   tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
> index aa7400ed0e99..f0d9c035641d 100644
> --- a/tools/testing/selftests/mm/split_huge_page_test.c
> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
> @@ -31,6 +31,7 @@ uint64_t pmd_pagesize;
>   #define INPUT_MAX 80
>   
>   #define PID_FMT "%d,0x%lx,0x%lx,%d"
> +#define PID_FMT_OFFSET "%d,0x%lx,0x%lx,%d,%d"
>   #define PATH_FMT "%s,0x%lx,0x%lx,%d"
>   
>   #define PFN_MASK     ((1UL<<55)-1)
> @@ -483,7 +484,7 @@ void split_thp_in_pagecache_to_order_at(size_t fd_size, const char *fs_loc,
>   		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
>   			      (uint64_t)addr + fd_size, order);
>   	else
> -		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
> +		write_debugfs(PID_FMT_OFFSET, getpid(), (uint64_t)addr,
>   			      (uint64_t)addr + fd_size, order, offset);
>   
>   	for (i = 0; i < fd_size; i++)
>

This looks good to me.

I tested it on my system, and the test is passing.

ok 17 Split PMD-mapped pagecache folio to order 0 at in-folio offset 0 
passed
ok 18 Split PMD-mapped pagecache folio to order 0 at in-folio offset 8 
passed
ok 19 Split PMD-mapped pagecache folio to order 0 at in-folio offset 16 
passed
ok 20 Split PMD-mapped pagecache folio to order 0 at in-folio offset 24 
passed
ok 21 Split PMD-mapped pagecache folio to order 1 at in-folio offset 0 
passed
ok 22 Split PMD-mapped pagecache folio to order 1 at in-folio offset 8 
passed
ok 23 Split PMD-mapped pagecache folio to order 1 at in-folio offset 16 
passed
ok 24 Split PMD-mapped pagecache folio to order 1 at in-folio offset 24 
passed
ok 25 Split PMD-mapped pagecache folio to order 2 at in-folio offset 0 
passed
ok 26 Split PMD-mapped pagecache folio to order 2 at in-folio offset 8 
passed
ok 27 Split PMD-mapped pagecache folio to order 2 at in-folio offset 16 
passed
ok 28 Split PMD-mapped pagecache folio to order 2 at in-folio offset 24 
passed
ok 29 Split PMD-mapped pagecache folio to order 3 at in-folio offset 0 
passed
ok 30 Split PMD-mapped pagecache folio to order 3 at in-folio offset 8 
passed
ok 31 Split PMD-mapped pagecache folio to order 3 at in-folio offset 16 
passed
ok 32 Split PMD-mapped pagecache folio to order 3 at in-folio offset 24 
passed
ok 33 Split PMD-mapped pagecache folio to order 4 at in-folio offset 0 
passed
ok 34 Split PMD-mapped pagecache folio to order 4 at in-folio offset 16 
passed


Feel free to add:

Reviewed-by: Donet Tom <donettom@linux.ibm.com>
Tested-by : Donet Tom <donettom@linux.ibm.com>

Re: [PATCH] selftests/mm: fix split_huge_page_test for folio_split() tests.

Posted by Baolin Wang 7 months ago


On 2025/7/9 09:27, Zi Yan wrote:
> PID_FMT does not have an offset field, so folio_split() tests are not
> performed. Add PID_FMT_OFFSET with an offset field and use it to perform
> folio_split() tests.
> 
> Fixes: 80a5c494c89f ("selftests/mm: add tests for folio_split(), buddy allocator like split")
> Signed-off-by: Zi Yan <ziy@nvidia.com>

LGTM.
Tested-by: Baolin Wang <baolin.wang@linux.alibaba.com>

> ---
>   tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
> index aa7400ed0e99..f0d9c035641d 100644
> --- a/tools/testing/selftests/mm/split_huge_page_test.c
> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
> @@ -31,6 +31,7 @@ uint64_t pmd_pagesize;
>   #define INPUT_MAX 80
>   
>   #define PID_FMT "%d,0x%lx,0x%lx,%d"
> +#define PID_FMT_OFFSET "%d,0x%lx,0x%lx,%d,%d"
>   #define PATH_FMT "%s,0x%lx,0x%lx,%d"
>   
>   #define PFN_MASK     ((1UL<<55)-1)
> @@ -483,7 +484,7 @@ void split_thp_in_pagecache_to_order_at(size_t fd_size, const char *fs_loc,
>   		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
>   			      (uint64_t)addr + fd_size, order);
>   	else
> -		write_debugfs(PID_FMT, getpid(), (uint64_t)addr,
> +		write_debugfs(PID_FMT_OFFSET, getpid(), (uint64_t)addr,
>   			      (uint64_t)addr + fd_size, order, offset);
>   
>   	for (i = 0; i < fd_size; i++)