[PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Lorenzo Stoakes (ARM)]

Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
for the former and a tree walk and swap in to swap cache for the latter.

MAP_PRIVATE-file backed mappings straddle the two and currently get treated
as if they were purely file-backed, meaning any swapped out private pages
remain swapped out.

Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
mappings and performing both walks in this case.

Since the logic checks for vma->anon_vma this means un-CoW'd
MAP_PRIVATE-file backed mappings retain only the single file walk.

Reported-by: Mike Kaplinskiy <mike@recall.ai>
Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
---
 mm/madvise.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/mm/madvise.c b/mm/madvise.c
index 73c2901b9adb..d0510dd49dde 100644
--- a/mm/madvise.c
+++ b/mm/madvise.c
@@ -297,10 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
 	loff_t offset;
 
 #ifdef CONFIG_SWAP
-	if (!file) {
+	if (!file || (vma_is_cow_mapping(vma) && vma->anon_vma)) {
 		walk_page_range_vma(vma, start, end, &swapin_walk_ops, vma);
 		lru_add_drain(); /* Push any new pages onto the LRU now */
-		return 0;
+
+		if (!file)
+			return 0;
 	}
 
 	if (shmem_mapping(file->f_mapping)) {

---
base-commit: e3b5239afe1b8f0194db7436b17c33e94c1988c4
change-id: 20260903-madv-will-need-map-private-125e3e2f70c7

Best regards,
-- 
Lorenzo Stoakes (ARM) <ljs@kernel.org>

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Pedro Falcato]

On Fri, Sep 04, 2026 at 04:12:53PM +0100, Lorenzo Stoakes (ARM) wrote:
> Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> for the former and a tree walk and swap in to swap cache for the latter.
> 
> MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> as if they were purely file-backed, meaning any swapped out private pages
> remain swapped out.
> 
> Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> mappings and performing both walks in this case.
> 
> Since the logic checks for vma->anon_vma this means un-CoW'd
> MAP_PRIVATE-file backed mappings retain only the single file walk.
> 
> Reported-by: Mike Kaplinskiy <mike@recall.ai>
> Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> ---
>  mm/madvise.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/madvise.c b/mm/madvise.c
> index 73c2901b9adb..d0510dd49dde 100644
> --- a/mm/madvise.c
> +++ b/mm/madvise.c
> @@ -297,10 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
>  	loff_t offset;
>  
>  #ifdef CONFIG_SWAP
> -	if (!file) {
> +	if (!file || (vma_is_cow_mapping(vma) && vma->anon_vma)) {

Couldn't this all be simplified to

	if (vma->anon_vma) {

? swapin needs anon pages to have been faulted-in. Non-cow mappings won't
have an anon_vma, nor will fully empty anonymous VMAs (and that's fine).
Right?

Regardless, looks cromulent

Reviewed-by: Pedro Falcato <pfalcato@suse.de>

-- 
Pedro

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Lorenzo Stoakes (ARM)]

On Fri, Sep 04, 2026 at 04:24:59PM +0100, Pedro Falcato wrote:
> On Fri, Sep 04, 2026 at 04:12:53PM +0100, Lorenzo Stoakes (ARM) wrote:
> > Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> > entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> > for the former and a tree walk and swap in to swap cache for the latter.
> >
> > MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> > as if they were purely file-backed, meaning any swapped out private pages
> > remain swapped out.
> >
> > Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> > mappings and performing both walks in this case.
> >
> > Since the logic checks for vma->anon_vma this means un-CoW'd
> > MAP_PRIVATE-file backed mappings retain only the single file walk.
> >
> > Reported-by: Mike Kaplinskiy <mike@recall.ai>
> > Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> > Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> > ---
> >  mm/madvise.c | 6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/mm/madvise.c b/mm/madvise.c
> > index 73c2901b9adb..d0510dd49dde 100644
> > --- a/mm/madvise.c
> > +++ b/mm/madvise.c
> > @@ -297,10 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
> >  	loff_t offset;
> >
> >  #ifdef CONFIG_SWAP
> > -	if (!file) {
> > +	if (!file || (vma_is_cow_mapping(vma) && vma->anon_vma)) {
>
> Couldn't this all be simplified to
>
> 	if (vma->anon_vma) {
>
> ? swapin needs anon pages to have been faulted-in. Non-cow mappings won't
> have an anon_vma, nor will fully empty anonymous VMAs (and that's fine).
> Right?

Hmm good point :)

Though the

	if (!file)
		return 0;

Would have to be outside of the block to avoid an anon unfaulted (nop) from
being skipped.

And it's a real improvement to have unfaulted anon skip...

But I find that version is documenting what's going on a lot less.

Since both anon and MAP_PRIVATE file-backed are CoW mappings we could just
reference that.

Andrew - could you swap the patch out in-place with below? Thanks!

Cheers, Lorenzo

----8<----
From a706374a1186b6b4409a0e760e8728bcadba0c96 Mon Sep 17 00:00:00 2001
From: "Lorenzo Stoakes (ARM)" <ljs@kernel.org>
Date: Thu, 3 Sep 2026 20:08:39 +0100
Subject: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on
 MADV_WILLNEED

Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
for the former and a tree walk and swap in to swap cache for the latter.

MAP_PRIVATE-file backed mappings straddle the two and currently get treated
as if they were purely file-backed, meaning any swapped out private pages
remain swapped out.

Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
mappings and performing both walks in this case.

Since the logic checks for vma->anon_vma this means un-CoW'd
MAP_PRIVATE-file backed mappings retain only the single file walk.

Reported-by: Mike Kaplinskiy <mike@recall.ai>
Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
---
 mm/madvise.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/mm/madvise.c b/mm/madvise.c
index 73c2901b9adb..963337f93a7a 100644
--- a/mm/madvise.c
+++ b/mm/madvise.c
@@ -297,11 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
 	loff_t offset;

 #ifdef CONFIG_SWAP
-	if (!file) {
+	if (vma_is_cow_mapping(vma) && vma->anon_vma) {
 		walk_page_range_vma(vma, start, end, &swapin_walk_ops, vma);
 		lru_add_drain(); /* Push any new pages onto the LRU now */
-		return 0;
 	}
+	if (!file)
+		return 0;

 	if (shmem_mapping(file->f_mapping)) {
 		shmem_swapin_range(vma, start, end, file->f_mapping);
--
2.55.0

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Andrew Morton]

On Fri, 4 Sep 2026 17:01:37 +0100 "Lorenzo Stoakes (ARM)" <ljs@kernel.org> wrote:

> Andrew - could you swap the patch out in-place with below? Thanks!
> 

np.

> Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> for the former and a tree walk and swap in to swap cache for the latter.
> 
> MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> as if they were purely file-backed, meaning any swapped out private pages
> remain swapped out.
> 
> Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> mappings and performing both walks in this case.
> 
> Since the logic checks for vma->anon_vma this means un-CoW'd
> MAP_PRIVATE-file backed mappings retain only the single file walk.
> 
> Reported-by: Mike Kaplinskiy <mike@recall.ai>
> Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>

The failure mode looks pretty obscure, so I'm thinking that no backport
is needed.

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[David Hildenbrand (Arm)]

> And it's a real improvement to have unfaulted anon skip...
> 
> But I find that version is documenting what's going on a lot less.
> 
> Since both anon and MAP_PRIVATE file-backed are CoW mappings we could just
> reference that.
> 
> Andrew - could you swap the patch out in-place with below? Thanks!

A proper v2 would have been nice ;)

(and would trigger sashiko and friends properly)

> 
> Cheers, Lorenzo
> 
> ----8<----
> From a706374a1186b6b4409a0e760e8728bcadba0c96 Mon Sep 17 00:00:00 2001
> From: "Lorenzo Stoakes (ARM)" <ljs@kernel.org>
> Date: Thu, 3 Sep 2026 20:08:39 +0100
> Subject: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on
>  MADV_WILLNEED
> 
> Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> for the former and a tree walk and swap in to swap cache for the latter.
> 
> MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> as if they were purely file-backed, meaning any swapped out private pages
> remain swapped out.
> 
> Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> mappings and performing both walks in this case.
> 
> Since the logic checks for vma->anon_vma this means un-CoW'd
> MAP_PRIVATE-file backed mappings retain only the single file walk.
> 
> Reported-by: Mike Kaplinskiy <mike@recall.ai>
> Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> ---
>  mm/madvise.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/madvise.c b/mm/madvise.c
> index 73c2901b9adb..963337f93a7a 100644
> --- a/mm/madvise.c
> +++ b/mm/madvise.c
> @@ -297,11 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
>  	loff_t offset;
> 
>  #ifdef CONFIG_SWAP
> -	if (!file) {
> +	if (vma_is_cow_mapping(vma) && vma->anon_vma) {
>  		walk_page_range_vma(vma, start, end, &swapin_walk_ops, vma);
>  		lru_add_drain(); /* Push any new pages onto the LRU now */
> -		return 0;
>  	}
> +	if (!file)
> +		return 0;
> 
LGTM

Acked-by: David Hildenbrand (Arm) <david@kernel.org>

-- 
Cheers,

David

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Lorenzo Stoakes (ARM)]

On Sat, Sep 05, 2026 at 05:17:57PM -0700, Andrew Morton wrote:
> On Fri, 4 Sep 2026 17:01:37 +0100 "Lorenzo Stoakes (ARM)" <ljs@kernel.org> wrote:
>
> > Andrew - could you swap the patch out in-place with below? Thanks!
> >
>
> np.
>
> > Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> > entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> > for the former and a tree walk and swap in to swap cache for the latter.
> >
> > MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> > as if they were purely file-backed, meaning any swapped out private pages
> > remain swapped out.
> >
> > Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> > mappings and performing both walks in this case.
> >
> > Since the logic checks for vma->anon_vma this means un-CoW'd
> > MAP_PRIVATE-file backed mappings retain only the single file walk.
> >
> > Reported-by: Mike Kaplinskiy <mike@recall.ai>
> > Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> > Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
>
> The failure mode looks pretty obscure, so I'm thinking that no backport
> is needed.

Well it's not even really a failure mode so much as a 'known limitation' :)
So yeah agreed.

--
Cheers, Lorenzo

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Lorenzo Stoakes (ARM)]

On Mon, Sep 07, 2026 at 01:51:20PM +0200, David Hildenbrand (Arm) wrote:
> > And it's a real improvement to have unfaulted anon skip...
> >
> > But I find that version is documenting what's going on a lot less.
> >
> > Since both anon and MAP_PRIVATE file-backed are CoW mappings we could just
> > reference that.
> >
> > Andrew - could you swap the patch out in-place with below? Thanks!
>
> A proper v2 would have been nice ;)
>
> (and would trigger sashiko and friends properly)

Ack, though it's kinda obviously functionally equivalent (or so I felt :P).

I had a local LLM check it FWIW.

>
> >
> > Cheers, Lorenzo
> >
> > ----8<----
> > From a706374a1186b6b4409a0e760e8728bcadba0c96 Mon Sep 17 00:00:00 2001
> > From: "Lorenzo Stoakes (ARM)" <ljs@kernel.org>
> > Date: Thu, 3 Sep 2026 20:08:39 +0100
> > Subject: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on
> >  MADV_WILLNEED
> >
> > Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> > entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> > for the former and a tree walk and swap in to swap cache for the latter.
> >
> > MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> > as if they were purely file-backed, meaning any swapped out private pages
> > remain swapped out.
> >
> > Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> > mappings and performing both walks in this case.
> >
> > Since the logic checks for vma->anon_vma this means un-CoW'd
> > MAP_PRIVATE-file backed mappings retain only the single file walk.
> >
> > Reported-by: Mike Kaplinskiy <mike@recall.ai>
> > Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> > Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> > ---
> >  mm/madvise.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/mm/madvise.c b/mm/madvise.c
> > index 73c2901b9adb..963337f93a7a 100644
> > --- a/mm/madvise.c
> > +++ b/mm/madvise.c
> > @@ -297,11 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
> >  	loff_t offset;
> >
> >  #ifdef CONFIG_SWAP
> > -	if (!file) {
> > +	if (vma_is_cow_mapping(vma) && vma->anon_vma) {
> >  		walk_page_range_vma(vma, start, end, &swapin_walk_ops, vma);
> >  		lru_add_drain(); /* Push any new pages onto the LRU now */
> > -		return 0;
> >  	}
> > +	if (!file)
> > +		return 0;
> >
> LGTM
>
> Acked-by: David Hildenbrand (Arm) <david@kernel.org>

Thanks!

>
> --
> Cheers,
>
> David

--
Cheers, Lorenzo

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Pedro Falcato]

On Fri, Sep 04, 2026 at 05:01:37PM +0100, Lorenzo Stoakes (ARM) wrote:
> On Fri, Sep 04, 2026 at 04:24:59PM +0100, Pedro Falcato wrote:
> > On Fri, Sep 04, 2026 at 04:12:53PM +0100, Lorenzo Stoakes (ARM) wrote:
> > > Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> > > entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> > > for the former and a tree walk and swap in to swap cache for the latter.
> > >
> > > MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> > > as if they were purely file-backed, meaning any swapped out private pages
> > > remain swapped out.
> > >
> > > Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> > > mappings and performing both walks in this case.
> > >
> > > Since the logic checks for vma->anon_vma this means un-CoW'd
> > > MAP_PRIVATE-file backed mappings retain only the single file walk.
> > >
> > > Reported-by: Mike Kaplinskiy <mike@recall.ai>
> > > Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> > > Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> > > ---
> > >  mm/madvise.c | 6 ++++--
> > >  1 file changed, 4 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/mm/madvise.c b/mm/madvise.c
> > > index 73c2901b9adb..d0510dd49dde 100644
> > > --- a/mm/madvise.c
> > > +++ b/mm/madvise.c
> > > @@ -297,10 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
> > >  	loff_t offset;
> > >
> > >  #ifdef CONFIG_SWAP
> > > -	if (!file) {
> > > +	if (!file || (vma_is_cow_mapping(vma) && vma->anon_vma)) {
> >
> > Couldn't this all be simplified to
> >
> > 	if (vma->anon_vma) {
> >
> > ? swapin needs anon pages to have been faulted-in. Non-cow mappings won't
> > have an anon_vma, nor will fully empty anonymous VMAs (and that's fine).
> > Right?
> 
> Hmm good point :)
> 
> Though the
> 
> 	if (!file)
> 		return 0;
> 
> Would have to be outside of the block to avoid an anon unfaulted (nop) from
> being skipped.
> 
> And it's a real improvement to have unfaulted anon skip...
> 
> But I find that version is documenting what's going on a lot less.
> 
> Since both anon and MAP_PRIVATE file-backed are CoW mappings we could just
> reference that.
> 
> Andrew - could you swap the patch out in-place with below? Thanks!
> 
> Cheers, Lorenzo
> 
> ----8<----
> From a706374a1186b6b4409a0e760e8728bcadba0c96 Mon Sep 17 00:00:00 2001
> From: "Lorenzo Stoakes (ARM)" <ljs@kernel.org>
> Date: Thu, 3 Sep 2026 20:08:39 +0100
> Subject: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on
>  MADV_WILLNEED
> 
> Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> for the former and a tree walk and swap in to swap cache for the latter.
> 
> MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> as if they were purely file-backed, meaning any swapped out private pages
> remain swapped out.
> 
> Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> mappings and performing both walks in this case.
> 
> Since the logic checks for vma->anon_vma this means un-CoW'd
> MAP_PRIVATE-file backed mappings retain only the single file walk.
> 
> Reported-by: Mike Kaplinskiy <mike@recall.ai>
> Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>

LGTM, thanks!

Reviewed-by: Pedro Falcato <pfalcato@suse.de>

-- 
Pedro

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Vlastimil Babka (SUSE)]

On 9/4/26 18:01, Lorenzo Stoakes (ARM) wrote:
> On Fri, Sep 04, 2026 at 04:24:59PM +0100, Pedro Falcato wrote:
> Andrew - could you swap the patch out in-place with below? Thanks!
> 
> Cheers, Lorenzo
> 
> ----8<----
> From a706374a1186b6b4409a0e760e8728bcadba0c96 Mon Sep 17 00:00:00 2001
> From: "Lorenzo Stoakes (ARM)" <ljs@kernel.org>
> Date: Thu, 3 Sep 2026 20:08:39 +0100
> Subject: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on
>  MADV_WILLNEED
> 
> Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> for the former and a tree walk and swap in to swap cache for the latter.
> 
> MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> as if they were purely file-backed, meaning any swapped out private pages
> remain swapped out.
> 
> Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> mappings and performing both walks in this case.
> 
> Since the logic checks for vma->anon_vma this means un-CoW'd
> MAP_PRIVATE-file backed mappings retain only the single file walk.
> 
> Reported-by: Mike Kaplinskiy <mike@recall.ai>
> Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>

Acked-by: Vlastimil Babka (SUSE) <vbabka@kernel.org>


> ---
>  mm/madvise.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/madvise.c b/mm/madvise.c
> index 73c2901b9adb..963337f93a7a 100644
> --- a/mm/madvise.c
> +++ b/mm/madvise.c
> @@ -297,11 +297,12 @@ static long madvise_willneed(struct madvise_behavior *madv_behavior)
>  	loff_t offset;
> 
>  #ifdef CONFIG_SWAP
> -	if (!file) {
> +	if (vma_is_cow_mapping(vma) && vma->anon_vma) {
>  		walk_page_range_vma(vma, start, end, &swapin_walk_ops, vma);
>  		lru_add_drain(); /* Push any new pages onto the LRU now */
> -		return 0;
>  	}
> +	if (!file)
> +		return 0;
> 
>  	if (shmem_mapping(file->f_mapping)) {
>  		shmem_swapin_range(vma, start, end, file->f_mapping);
> --
> 2.55.0

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Vlastimil Babka (SUSE)]

On 9/7/26 14:32, Lorenzo Stoakes (ARM) wrote:
> On Sat, Sep 05, 2026 at 05:17:57PM -0700, Andrew Morton wrote:
>> On Fri, 4 Sep 2026 17:01:37 +0100 "Lorenzo Stoakes (ARM)" <ljs@kernel.org> wrote:
>>
>> > Andrew - could you swap the patch out in-place with below? Thanks!
>> >
>>
>> np.
>>
>> > Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
>> > entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
>> > for the former and a tree walk and swap in to swap cache for the latter.
>> >
>> > MAP_PRIVATE-file backed mappings straddle the two and currently get treated
>> > as if they were purely file-backed, meaning any swapped out private pages
>> > remain swapped out.
>> >
>> > Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
>> > mappings and performing both walks in this case.
>> >
>> > Since the logic checks for vma->anon_vma this means un-CoW'd
>> > MAP_PRIVATE-file backed mappings retain only the single file walk.
>> >
>> > Reported-by: Mike Kaplinskiy <mike@recall.ai>
>> > Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
>> > Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
>>
>> The failure mode looks pretty obscure, so I'm thinking that no backport
>> is needed.
> 
> Well it's not even really a failure mode so much as a 'known limitation' :)
> So yeah agreed.

Yeah, as we know madvise() just an advice after all, the kernel is free to
ignore it :)

(MADV_DONTNEED and friends enter the chat)

> --
> Cheers, Lorenzo

Re: [PATCH] mm/madvise: swap in CoW'd MAP_PRIVATE-file mappings on MADV_WILLNEED

[Lorenzo Stoakes (ARM)]

On Mon, Sep 07, 2026 at 05:09:34PM +0200, Vlastimil Babka (SUSE) wrote:
> On 9/7/26 14:32, Lorenzo Stoakes (ARM) wrote:
> > On Sat, Sep 05, 2026 at 05:17:57PM -0700, Andrew Morton wrote:
> >> On Fri, 4 Sep 2026 17:01:37 +0100 "Lorenzo Stoakes (ARM)" <ljs@kernel.org> wrote:
> >>
> >> > Andrew - could you swap the patch out in-place with below? Thanks!
> >> >
> >>
> >> np.
> >>
> >> > Currently MADV_WILLNEED treats file-backed and pure anonymous mappings
> >> > entirely separately - using POSIX_FADV_WILLNEED (equivalent of a readahead)
> >> > for the former and a tree walk and swap in to swap cache for the latter.
> >> >
> >> > MAP_PRIVATE-file backed mappings straddle the two and currently get treated
> >> > as if they were purely file-backed, meaning any swapped out private pages
> >> > remain swapped out.
> >> >
> >> > Resolve the issue by explicitly checking for CoW'd MAP_PRIVATE-file backed
> >> > mappings and performing both walks in this case.
> >> >
> >> > Since the logic checks for vma->anon_vma this means un-CoW'd
> >> > MAP_PRIVATE-file backed mappings retain only the single file walk.
> >> >
> >> > Reported-by: Mike Kaplinskiy <mike@recall.ai>
> >> > Closes: https://lore.kernel.org/all/CABeknB_S2XJSHFgnHdgnN0rjzHhH4oQJs_APq9fvxHztQ_pgiA@mail.gmail.com/
> >> > Signed-off-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
> >>
> >> The failure mode looks pretty obscure, so I'm thinking that no backport
> >> is needed.
> >
> > Well it's not even really a failure mode so much as a 'known limitation' :)
> > So yeah agreed.
>
> Yeah, as we know madvise() just an advice after all, the kernel is free to
> ignore it :)
>
> (MADV_DONTNEED and friends enter the chat)

Haha yeah, another brilliant footgun invented by the proud folks at kernel mm,
catch us next time with some prctl's and BPF APIs! ;)

>
> > --
> > Cheers, Lorenzo
>

--
Cheers, Lorenzo