diff mbox series

[RFC,v2,03/14] filemap: use mapping_min_order while allocating folios

Message ID 20240213093713.1753368-4-kernel@pankajraghav.com (mailing list archive)
State New
Headers show
Series enable bs > ps in XFS | expand

Commit Message

Pankaj Raghav (Samsung) Feb. 13, 2024, 9:37 a.m. UTC 
From: Pankaj Raghav <p.raghav@samsung.com>

filemap_create_folio() and do_read_cache_folio() were always allocating
folio of order 0. __filemap_get_folio was trying to allocate higher
order folios when fgp_flags had higher order hint set but it will default
to order 0 folio if higher order memory allocation fails.

As we bring the notion of mapping_min_order, make sure these functions
allocate at least folio of mapping_min_order as we need to guarantee it
in the page cache.

Add some additional VM_BUG_ON() in page_cache_delete[batch] and
__filemap_add_folio to catch errors where we delete or add folios that
has order less than min_order.

Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
---
 mm/filemap.c | 25 +++++++++++++++++++++----
 1 file changed, 21 insertions(+), 4 deletions(-)

Comments

Hannes Reinecke Feb. 13, 2024, 2:58 p.m. UTC | #1 
On 2/13/24 10:37, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <p.raghav@samsung.com>
> 
> filemap_create_folio() and do_read_cache_folio() were always allocating
> folio of order 0. __filemap_get_folio was trying to allocate higher
> order folios when fgp_flags had higher order hint set but it will default
> to order 0 folio if higher order memory allocation fails.
> 
> As we bring the notion of mapping_min_order, make sure these functions
> allocate at least folio of mapping_min_order as we need to guarantee it
> in the page cache.
> 
> Add some additional VM_BUG_ON() in page_cache_delete[batch] and
> __filemap_add_folio to catch errors where we delete or add folios that
> has order less than min_order.
> 
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
> ---
>   mm/filemap.c | 25 +++++++++++++++++++++----
>   1 file changed, 21 insertions(+), 4 deletions(-)
> 
Reviewed-by: Hannes Reinecke <hare@suse.de>

Cheers,

Hannes
Darrick J. Wong Feb. 13, 2024, 4:38 p.m. UTC | #2 
On Tue, Feb 13, 2024 at 10:37:02AM +0100, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <p.raghav@samsung.com>
> 
> filemap_create_folio() and do_read_cache_folio() were always allocating
> folio of order 0. __filemap_get_folio was trying to allocate higher
> order folios when fgp_flags had higher order hint set but it will default
> to order 0 folio if higher order memory allocation fails.
> 
> As we bring the notion of mapping_min_order, make sure these functions
> allocate at least folio of mapping_min_order as we need to guarantee it
> in the page cache.
> 
> Add some additional VM_BUG_ON() in page_cache_delete[batch] and
> __filemap_add_folio to catch errors where we delete or add folios that
> has order less than min_order.
> 
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>

Looks good to me,
Acked-by: Darrick J. Wong <djwong@kernel.org>

--D

> ---
>  mm/filemap.c | 25 +++++++++++++++++++++----
>  1 file changed, 21 insertions(+), 4 deletions(-)
> 
> diff --git a/mm/filemap.c b/mm/filemap.c
> index 323a8e169581..7a6e15c47150 100644
> --- a/mm/filemap.c
> +++ b/mm/filemap.c
> @@ -127,6 +127,7 @@
>  static void page_cache_delete(struct address_space *mapping,
>  				   struct folio *folio, void *shadow)
>  {
> +	unsigned int min_order = mapping_min_folio_order(mapping);
>  	XA_STATE(xas, &mapping->i_pages, folio->index);
>  	long nr = 1;
>  
> @@ -135,6 +136,7 @@ static void page_cache_delete(struct address_space *mapping,
>  	xas_set_order(&xas, folio->index, folio_order(folio));
>  	nr = folio_nr_pages(folio);
>  
> +	VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
>  	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
>  
>  	xas_store(&xas, shadow);
> @@ -277,6 +279,7 @@ void filemap_remove_folio(struct folio *folio)
>  static void page_cache_delete_batch(struct address_space *mapping,
>  			     struct folio_batch *fbatch)
>  {
> +	unsigned int min_order = mapping_min_folio_order(mapping);
>  	XA_STATE(xas, &mapping->i_pages, fbatch->folios[0]->index);
>  	long total_pages = 0;
>  	int i = 0;
> @@ -305,6 +308,7 @@ static void page_cache_delete_batch(struct address_space *mapping,
>  
>  		WARN_ON_ONCE(!folio_test_locked(folio));
>  
> +		VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
>  		folio->mapping = NULL;
>  		/* Leave folio->index set: truncation lookup relies on it */
>  
> @@ -846,6 +850,7 @@ noinline int __filemap_add_folio(struct address_space *mapping,
>  	int huge = folio_test_hugetlb(folio);
>  	bool charged = false;
>  	long nr = 1;
> +	unsigned int min_order = mapping_min_folio_order(mapping);
>  
>  	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
>  	VM_BUG_ON_FOLIO(folio_test_swapbacked(folio), folio);
> @@ -896,6 +901,7 @@ noinline int __filemap_add_folio(struct address_space *mapping,
>  			}
>  		}
>  
> +		VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
>  		xas_store(&xas, folio);
>  		if (xas_error(&xas))
>  			goto unlock;
> @@ -1847,6 +1853,10 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
>  		fgf_t fgp_flags, gfp_t gfp)
>  {
>  	struct folio *folio;
> +	unsigned int min_order = mapping_min_folio_order(mapping);
> +	unsigned int min_nrpages = mapping_min_folio_nrpages(mapping);
> +
> +	index = round_down(index, min_nrpages);
>  
>  repeat:
>  	folio = filemap_get_entry(mapping, index);
> @@ -1886,7 +1896,7 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
>  		folio_wait_stable(folio);
>  no_page:
>  	if (!folio && (fgp_flags & FGP_CREAT)) {
> -		unsigned order = FGF_GET_ORDER(fgp_flags);
> +		unsigned int order = max(min_order, FGF_GET_ORDER(fgp_flags));
>  		int err;
>  
>  		if ((fgp_flags & FGP_WRITE) && mapping_can_writeback(mapping))
> @@ -1914,8 +1924,13 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
>  			err = -ENOMEM;
>  			if (order == 1)
>  				order = 0;
> +			if (order < min_order)
> +				order = min_order;
>  			if (order > 0)
>  				alloc_gfp |= __GFP_NORETRY | __GFP_NOWARN;
> +
> +			VM_BUG_ON(index & ((1UL << order) - 1));
> +
>  			folio = filemap_alloc_folio(alloc_gfp, order);
>  			if (!folio)
>  				continue;
> @@ -1929,7 +1944,7 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
>  				break;
>  			folio_put(folio);
>  			folio = NULL;
> -		} while (order-- > 0);
> +		} while (order-- > min_order);
>  
>  		if (err == -EEXIST)
>  			goto repeat;
> @@ -2424,7 +2439,8 @@ static int filemap_create_folio(struct file *file,
>  	struct folio *folio;
>  	int error;
>  
> -	folio = filemap_alloc_folio(mapping_gfp_mask(mapping), 0);
> +	folio = filemap_alloc_folio(mapping_gfp_mask(mapping),
> +				    mapping_min_folio_order(mapping));
>  	if (!folio)
>  		return -ENOMEM;
>  
> @@ -3682,7 +3698,8 @@ static struct folio *do_read_cache_folio(struct address_space *mapping,
>  repeat:
>  	folio = filemap_get_folio(mapping, index);
>  	if (IS_ERR(folio)) {
> -		folio = filemap_alloc_folio(gfp, 0);
> +		folio = filemap_alloc_folio(gfp,
> +					    mapping_min_folio_order(mapping));
>  		if (!folio)
>  			return ERR_PTR(-ENOMEM);
>  		err = filemap_add_folio(mapping, folio, index, gfp);
> -- 
> 2.43.0
> 
>
Dave Chinner Feb. 13, 2024, 10:05 p.m. UTC | #3 
On Tue, Feb 13, 2024 at 10:37:02AM +0100, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <p.raghav@samsung.com>
> 
> filemap_create_folio() and do_read_cache_folio() were always allocating
> folio of order 0. __filemap_get_folio was trying to allocate higher
> order folios when fgp_flags had higher order hint set but it will default
> to order 0 folio if higher order memory allocation fails.
> 
> As we bring the notion of mapping_min_order, make sure these functions
> allocate at least folio of mapping_min_order as we need to guarantee it
> in the page cache.
> 
> Add some additional VM_BUG_ON() in page_cache_delete[batch] and
> __filemap_add_folio to catch errors where we delete or add folios that
> has order less than min_order.
> 
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
> ---
>  mm/filemap.c | 25 +++++++++++++++++++++----
>  1 file changed, 21 insertions(+), 4 deletions(-)
> 
> diff --git a/mm/filemap.c b/mm/filemap.c
> index 323a8e169581..7a6e15c47150 100644
> --- a/mm/filemap.c
> +++ b/mm/filemap.c
> @@ -127,6 +127,7 @@
>  static void page_cache_delete(struct address_space *mapping,
>  				   struct folio *folio, void *shadow)
>  {
> +	unsigned int min_order = mapping_min_folio_order(mapping);
>  	XA_STATE(xas, &mapping->i_pages, folio->index);
>  	long nr = 1;
>  
> @@ -135,6 +136,7 @@ static void page_cache_delete(struct address_space *mapping,
>  	xas_set_order(&xas, folio->index, folio_order(folio));
>  	nr = folio_nr_pages(folio);
>  
> +	VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
>  	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);

If you are only using min_order in the VM_BUG_ON_FOLIO() macro, then
please just do:

	VM_BUG_ON_FOLIO(folio_order(folio) < mapping_min_folio_order(mapping),
			folio);

There is no need to clutter up the function with variables that are
only used in one debug-only check.

> @@ -1847,6 +1853,10 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
>  		fgf_t fgp_flags, gfp_t gfp)
>  {
>  	struct folio *folio;
> +	unsigned int min_order = mapping_min_folio_order(mapping);
> +	unsigned int min_nrpages = mapping_min_folio_nrpages(mapping);
> +
> +	index = round_down(index, min_nrpages);

	index = mapping_align_start_index(mapping, index);

The rest of the function only cares about min_order, not
min_nrpages....

-Dave.
Pankaj Raghav (Samsung) Feb. 14, 2024, 10:13 a.m. UTC | #4 
> > +++ b/mm/filemap.c
> > @@ -127,6 +127,7 @@
> >  static void page_cache_delete(struct address_space *mapping,
> >  				   struct folio *folio, void *shadow)
> >  {
> > +	unsigned int min_order = mapping_min_folio_order(mapping);
> >  	XA_STATE(xas, &mapping->i_pages, folio->index);
> >  	long nr = 1;
> >  
> > @@ -135,6 +136,7 @@ static void page_cache_delete(struct address_space *mapping,
> >  	xas_set_order(&xas, folio->index, folio_order(folio));
> >  	nr = folio_nr_pages(folio);
> >  
> > +	VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
> >  	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
> 
> If you are only using min_order in the VM_BUG_ON_FOLIO() macro, then
> please just do:
> 
> 	VM_BUG_ON_FOLIO(folio_order(folio) < mapping_min_folio_order(mapping),
> 			folio);
> 
> There is no need to clutter up the function with variables that are
> only used in one debug-only check.
> 
Got it. I will fold it in.

> > @@ -1847,6 +1853,10 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
> >  		fgf_t fgp_flags, gfp_t gfp)
> >  {
> >  	struct folio *folio;
> > +	unsigned int min_order = mapping_min_folio_order(mapping);
> > +	unsigned int min_nrpages = mapping_min_folio_nrpages(mapping);
> > +
> > +	index = round_down(index, min_nrpages);
> 
> 	index = mapping_align_start_index(mapping, index);

I will add this helper. Makes the intent more clear. Thanks.

> 
> The rest of the function only cares about min_order, not
> min_nrpages....
> 
> -Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
diff mbox series

Patch

diff --git a/mm/filemap.c b/mm/filemap.c
index 323a8e169581..7a6e15c47150 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -127,6 +127,7 @@ 
 static void page_cache_delete(struct address_space *mapping,
 				   struct folio *folio, void *shadow)
 {
+	unsigned int min_order = mapping_min_folio_order(mapping);
 	XA_STATE(xas, &mapping->i_pages, folio->index);
 	long nr = 1;
 
@@ -135,6 +136,7 @@  static void page_cache_delete(struct address_space *mapping,
 	xas_set_order(&xas, folio->index, folio_order(folio));
 	nr = folio_nr_pages(folio);
 
+	VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
 	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
 
 	xas_store(&xas, shadow);
@@ -277,6 +279,7 @@  void filemap_remove_folio(struct folio *folio)
 static void page_cache_delete_batch(struct address_space *mapping,
 			     struct folio_batch *fbatch)
 {
+	unsigned int min_order = mapping_min_folio_order(mapping);
 	XA_STATE(xas, &mapping->i_pages, fbatch->folios[0]->index);
 	long total_pages = 0;
 	int i = 0;
@@ -305,6 +308,7 @@  static void page_cache_delete_batch(struct address_space *mapping,
 
 		WARN_ON_ONCE(!folio_test_locked(folio));
 
+		VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
 		folio->mapping = NULL;
 		/* Leave folio->index set: truncation lookup relies on it */
 
@@ -846,6 +850,7 @@  noinline int __filemap_add_folio(struct address_space *mapping,
 	int huge = folio_test_hugetlb(folio);
 	bool charged = false;
 	long nr = 1;
+	unsigned int min_order = mapping_min_folio_order(mapping);
 
 	VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
 	VM_BUG_ON_FOLIO(folio_test_swapbacked(folio), folio);
@@ -896,6 +901,7 @@  noinline int __filemap_add_folio(struct address_space *mapping,
 			}
 		}
 
+		VM_BUG_ON_FOLIO(folio_order(folio) < min_order, folio);
 		xas_store(&xas, folio);
 		if (xas_error(&xas))
 			goto unlock;
@@ -1847,6 +1853,10 @@  struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
 		fgf_t fgp_flags, gfp_t gfp)
 {
 	struct folio *folio;
+	unsigned int min_order = mapping_min_folio_order(mapping);
+	unsigned int min_nrpages = mapping_min_folio_nrpages(mapping);
+
+	index = round_down(index, min_nrpages);
 
 repeat:
 	folio = filemap_get_entry(mapping, index);
@@ -1886,7 +1896,7 @@  struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
 		folio_wait_stable(folio);
 no_page:
 	if (!folio && (fgp_flags & FGP_CREAT)) {
-		unsigned order = FGF_GET_ORDER(fgp_flags);
+		unsigned int order = max(min_order, FGF_GET_ORDER(fgp_flags));
 		int err;
 
 		if ((fgp_flags & FGP_WRITE) && mapping_can_writeback(mapping))
@@ -1914,8 +1924,13 @@  struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
 			err = -ENOMEM;
 			if (order == 1)
 				order = 0;
+			if (order < min_order)
+				order = min_order;
 			if (order > 0)
 				alloc_gfp |= __GFP_NORETRY | __GFP_NOWARN;
+
+			VM_BUG_ON(index & ((1UL << order) - 1));
+
 			folio = filemap_alloc_folio(alloc_gfp, order);
 			if (!folio)
 				continue;
@@ -1929,7 +1944,7 @@  struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
 				break;
 			folio_put(folio);
 			folio = NULL;
-		} while (order-- > 0);
+		} while (order-- > min_order);
 
 		if (err == -EEXIST)
 			goto repeat;
@@ -2424,7 +2439,8 @@  static int filemap_create_folio(struct file *file,
 	struct folio *folio;
 	int error;
 
-	folio = filemap_alloc_folio(mapping_gfp_mask(mapping), 0);
+	folio = filemap_alloc_folio(mapping_gfp_mask(mapping),
+				    mapping_min_folio_order(mapping));
 	if (!folio)
 		return -ENOMEM;
 
@@ -3682,7 +3698,8 @@  static struct folio *do_read_cache_folio(struct address_space *mapping,
 repeat:
 	folio = filemap_get_folio(mapping, index);
 	if (IS_ERR(folio)) {
-		folio = filemap_alloc_folio(gfp, 0);
+		folio = filemap_alloc_folio(gfp,
+					    mapping_min_folio_order(mapping));
 		if (!folio)
 			return ERR_PTR(-ENOMEM);
 		err = filemap_add_folio(mapping, folio, index, gfp);