[mm-stable] mm/hugetlb: set head flag before setting compound_order in __prep_compound_gigantic_folio

Commit Message

Sidhartha Kumar Dec. 12, 2022, 10:55 p.m. UTC

folio_set_compound_order() checks if the passed in folio is a large folio.
A large folio is indicated by the PG_head flag. Call __folio_set_head()
before setting the order.

Signed-off-by: Sidhartha Kumar <[e-mail]>
Reported-by: David Hildenbrand <[e-mail]>
---
Hi David, 
I tested this by doing:

echo 10 > /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages
echo 0 > /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages

and checking dmesg. Let me know if there are other ways of testing
you would like to see run. 

Hi Andrew,
This can be folded into d1c6095572d0cf00c0cd30378639ff9387b34edd
mm/hugetlb: convert hugetlb prep functions to folios.

However, there is still ongoing discussion with this patch series,
specifically with patch 1 9fd330582b2f mm: add folio dtor and order setter
functions, so I'm not sure if this series should be a part of the 6.2 merge
window. I am planning to send a v6 within the next few days to implement
the suggested changes.

 mm/hugetlb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

David Hildenbrand Dec. 13, 2022, 8:42 a.m. UTC | #1

On 12.12.22 23:55, Sidhartha Kumar wrote:
> folio_set_compound_order() checks if the passed in folio is a large folio.
> A large folio is indicated by the PG_head flag. Call __folio_set_head()
> before setting the order.
> 
> Signed-off-by: Sidhartha Kumar <[e-mail]>
> Reported-by: David Hildenbrand <[e-mail]>
> ---
> Hi David,
> I tested this by doing:
> 
> echo 10 > /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages
> echo 0 > /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages

Thanks, that should be good enough.

> 
> and checking dmesg. Let me know if there are other ways of testing
> you would like to see run.
> 
> Hi Andrew,
> This can be folded into d1c6095572d0cf00c0cd30378639ff9387b34edd
> mm/hugetlb: convert hugetlb prep functions to folios.
> 
> However, there is still ongoing discussion with this patch series,
> specifically with patch 1 9fd330582b2f mm: add folio dtor and order setter
> functions, so I'm not sure if this series should be a part of the 6.2 merge
> window. I am planning to send a v6 within the next few days to implement
> the suggested changes.

Whatever is in mm-stable is supposed to go upstream as-is and even the
commit ids will remain stable, meaning we an only queue fixes on top.

Therefore, mm-stable now contains:

commit c45bc55a99957b20e4e0333bcd42e12d1833a7f5 (mm/mm-stable)
Author: Sidhartha Kumar <[e-mail]>
Date:   Mon Dec 12 14:55:29 2022 -0800

     mm/hugetlb: set head flag before setting compound_order in __prep_compound_gigantic_folio
     
     folio_set_compound_order() checks if the passed in folio is a large folio.
     A large folio is indicated by the PG_head flag.  Call __folio_set_head()
     before setting the order.
     
     Link: https://lkml.kernel.org/r/[e-mail]
     Fixes: d1c6095572d0 ("mm/hugetlb: convert hugetlb prep functions to folios")
     Signed-off-by: Sidhartha Kumar <[e-mail]>
     Reported-by: David Hildenbrand <[e-mail]>
     Signed-off-by: Andrew Morton <[e-mail]>


So, best send any future changes against mm-stable instead of sending out a v6.

Patch

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 8c6fe2286814..7cdbcc22587b 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -1805,10 +1805,10 @@  static bool __prep_compound_gigantic_folio(struct folio *folio,
 	int nr_pages = 1 << order;
 	struct page *p;
 
-	/* we rely on prep_new_hugetlb_folio to set the destructor */
-	folio_set_compound_order(folio, order);
 	__folio_clear_reserved(folio);
 	__folio_set_head(folio);
+	/* we rely on prep_new_hugetlb_folio to set the destructor */
+	folio_set_compound_order(folio, order);
 	for (i = 0; i < nr_pages; i++) {
 		p = folio_page(folio, i);