[PATCH 0/2] kho: fixes for vmalloc restoration

[PATCH 0/2] kho: fixes for vmalloc restoration

[Mike Rapoport]

From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>

Hi,

Pratyush reported off-list that when kho_restore_vmalloc() is used to
restore a vmalloc_huge() allocation it hits VM_BUG_ON() when we reconstruct
the struct pages in kho_restore_pages().

These patches fix the issue.

Mike Rapoport (Microsoft) (2):
  kho: kho_restore_vmalloc: fix initialization of pages array
  kho: fix restoring of contiguous ranges of order-0 pages

 kernel/liveupdate/kexec_handover.c | 22 +++++++++++++---------
 1 file changed, 13 insertions(+), 9 deletions(-)

-- 
2.50.1

[PATCH 1/2] kho: kho_restore_vmalloc: fix initialization of pages array

[Mike Rapoport]

From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>

In case a preserved vmalloc allocation was using huge pages, all pages in
the array of pages added to vm_struct during kho_restore_vmalloc() are
wrongly set to the same page.

Fix the indexing when assigning pages to that array.

Fixes: a667300bd53f ("kho: add support for preserving vmalloc allocations")
Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
---
 kernel/liveupdate/kexec_handover.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/liveupdate/kexec_handover.c b/kernel/liveupdate/kexec_handover.c
index 5809c6fe331c..e64ee87fa62a 100644
--- a/kernel/liveupdate/kexec_handover.c
+++ b/kernel/liveupdate/kexec_handover.c
@@ -1096,7 +1096,7 @@ void *kho_restore_vmalloc(const struct kho_vmalloc *preservation)
 				goto err_free_pages_array;
 
 			for (int j = 0; j < contig_pages; j++)
-				pages[idx++] = page;
+				pages[idx++] = page + j;
 
 			phys += contig_pages * PAGE_SIZE;
 		}
-- 
2.50.1

Re: [PATCH 1/2] kho: kho_restore_vmalloc: fix initialization of pages array

[Pratyush Yadav]

On Tue, Nov 25 2025, Mike Rapoport wrote:

> From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>
>
> In case a preserved vmalloc allocation was using huge pages, all pages in
> the array of pages added to vm_struct during kho_restore_vmalloc() are
> wrongly set to the same page.
>
> Fix the indexing when assigning pages to that array.
>
> Fixes: a667300bd53f ("kho: add support for preserving vmalloc allocations")
> Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>

Reviewed-by: Pratyush Yadav <pratyush@kernel.org>

[...]

-- 
Regards,
Pratyush Yadav

[PATCH 2/2] kho: fix restoring of contiguous ranges of order-0 pages

[Mike Rapoport]

From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>

When contiguous ranges of order-0 pages are restored, kho_restore_page()
calls prep_compound_page() with the first page in the range and order as
parameters and then kho_restore_pages() calls split_page() to make sure all
pages in the range are order-0.

However, since split_page() is not intended to split compound pages and
with VM_DEBUG enabled it will trigger a VM_BUG_ON_PAGE().

Update kho_restore_page() so that it will use prep_compound_page() when it
restores a folio and make sure it properly sets page count for both large
folios and ranges of order-0 pages.

Reported-by: Pratyush Yadav <pratyush@kernel.org>
Fixes: a667300bd53f ("kho: add support for preserving vmalloc allocations")
Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
---
 kernel/liveupdate/kexec_handover.c | 20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

diff --git a/kernel/liveupdate/kexec_handover.c b/kernel/liveupdate/kexec_handover.c
index e64ee87fa62a..61d17ed1f423 100644
--- a/kernel/liveupdate/kexec_handover.c
+++ b/kernel/liveupdate/kexec_handover.c
@@ -219,11 +219,11 @@ static int __kho_preserve_order(struct kho_mem_track *track, unsigned long pfn,
 	return 0;
 }
 
-static struct page *kho_restore_page(phys_addr_t phys)
+static struct page *kho_restore_page(phys_addr_t phys, bool is_folio)
 {
 	struct page *page = pfn_to_online_page(PHYS_PFN(phys));
+	unsigned int nr_pages, ref_cnt;
 	union kho_page_info info;
-	unsigned int nr_pages;
 
 	if (!page)
 		return NULL;
@@ -243,11 +243,16 @@ static struct page *kho_restore_page(phys_addr_t phys)
 	/* Head page gets refcount of 1. */
 	set_page_count(page, 1);
 
-	/* For higher order folios, tail pages get a page count of zero. */
+	/*
+	 * For higher order folios, tail pages get a page count of zero.
+	 * For physically contiguous order-0 pages every pages gets a page
+	 * count of 1
+	 */
+	ref_cnt = is_folio ? 0 : 1;
 	for (unsigned int i = 1; i < nr_pages; i++)
-		set_page_count(page + i, 0);
+		set_page_count(page + i, ref_cnt);
 
-	if (info.order > 0)
+	if (is_folio && info.order)
 		prep_compound_page(page, info.order);
 
 	adjust_managed_page_count(page, nr_pages);
@@ -262,7 +267,7 @@ static struct page *kho_restore_page(phys_addr_t phys)
  */
 struct folio *kho_restore_folio(phys_addr_t phys)
 {
-	struct page *page = kho_restore_page(phys);
+	struct page *page = kho_restore_page(phys, true);
 
 	return page ? page_folio(page) : NULL;
 }
@@ -287,11 +292,10 @@ struct page *kho_restore_pages(phys_addr_t phys, unsigned int nr_pages)
 	while (pfn < end_pfn) {
 		const unsigned int order =
 			min(count_trailing_zeros(pfn), ilog2(end_pfn - pfn));
-		struct page *page = kho_restore_page(PFN_PHYS(pfn));
+		struct page *page = kho_restore_page(PFN_PHYS(pfn), false);
 
 		if (!page)
 			return NULL;
-		split_page(page, order);
 		pfn += 1 << order;
 	}
 
-- 
2.50.1

Re: [PATCH 2/2] kho: fix restoring of contiguous ranges of order-0 pages

[Pratyush Yadav]

On Tue, Nov 25 2025, Mike Rapoport wrote:

> From: "Mike Rapoport (Microsoft)" <rppt@kernel.org>
>
> When contiguous ranges of order-0 pages are restored, kho_restore_page()
> calls prep_compound_page() with the first page in the range and order as
> parameters and then kho_restore_pages() calls split_page() to make sure all
> pages in the range are order-0.
>
> However, since split_page() is not intended to split compound pages and
> with VM_DEBUG enabled it will trigger a VM_BUG_ON_PAGE().
>
> Update kho_restore_page() so that it will use prep_compound_page() when it
> restores a folio and make sure it properly sets page count for both large
> folios and ranges of order-0 pages.
>
> Reported-by: Pratyush Yadav <pratyush@kernel.org>
> Fixes: a667300bd53f ("kho: add support for preserving vmalloc allocations")
> Signed-off-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> ---
>  kernel/liveupdate/kexec_handover.c | 20 ++++++++++++--------
>  1 file changed, 12 insertions(+), 8 deletions(-)
>
> diff --git a/kernel/liveupdate/kexec_handover.c b/kernel/liveupdate/kexec_handover.c
> index e64ee87fa62a..61d17ed1f423 100644
> --- a/kernel/liveupdate/kexec_handover.c
> +++ b/kernel/liveupdate/kexec_handover.c
> @@ -219,11 +219,11 @@ static int __kho_preserve_order(struct kho_mem_track *track, unsigned long pfn,
>  	return 0;
>  }
>  
> -static struct page *kho_restore_page(phys_addr_t phys)
> +static struct page *kho_restore_page(phys_addr_t phys, bool is_folio)
>  {
>  	struct page *page = pfn_to_online_page(PHYS_PFN(phys));
> +	unsigned int nr_pages, ref_cnt;
>  	union kho_page_info info;
> -	unsigned int nr_pages;
>  
>  	if (!page)
>  		return NULL;
> @@ -243,11 +243,16 @@ static struct page *kho_restore_page(phys_addr_t phys)
>  	/* Head page gets refcount of 1. */
>  	set_page_count(page, 1);
>  
> -	/* For higher order folios, tail pages get a page count of zero. */
> +	/*
> +	 * For higher order folios, tail pages get a page count of zero.
> +	 * For physically contiguous order-0 pages every pages gets a page
> +	 * count of 1
> +	 */
> +	ref_cnt = is_folio ? 0 : 1;
>  	for (unsigned int i = 1; i < nr_pages; i++)
> -		set_page_count(page + i, 0);
> +		set_page_count(page + i, ref_cnt);
>  
> -	if (info.order > 0)
> +	if (is_folio && info.order)

This is getting a bit difficult to parse. Let's separate out folio and
page initialization to separate helpers:

	/* Initalize 0-order KHO pages */
	static void kho_init_page(struct page *page, unsigned int nr_pages)
	{
		for (unsigned int i = 0; i < nr_pages; i++)
			set_page_count(page + i, 1);
	}
	
	static void kho_init_folio(struct page *page, unsigned int order)
	{
		unsigned int nr_pages = (1 << order);
	
		/* Head page gets refcount of 1. */
		set_page_count(page, 1);
	
		/* For higher order folios, tail pages get a page count of zero. */
		for (unsigned int i = 1; i < nr_pages; i++)
			set_page_count(page + i, 0);
	
		if (order > 0)
			prep_compound_page(page, order);
	}


>  		prep_compound_page(page, info.order);
>  
>  	adjust_managed_page_count(page, nr_pages);
> @@ -262,7 +267,7 @@ static struct page *kho_restore_page(phys_addr_t phys)
>   */
>  struct folio *kho_restore_folio(phys_addr_t phys)
>  {
> -	struct page *page = kho_restore_page(phys);
> +	struct page *page = kho_restore_page(phys, true);
>  
>  	return page ? page_folio(page) : NULL;
>  }
> @@ -287,11 +292,10 @@ struct page *kho_restore_pages(phys_addr_t phys, unsigned int nr_pages)
>  	while (pfn < end_pfn) {
>  		const unsigned int order =
>  			min(count_trailing_zeros(pfn), ilog2(end_pfn - pfn));
> -		struct page *page = kho_restore_page(PFN_PHYS(pfn));
> +		struct page *page = kho_restore_page(PFN_PHYS(pfn), false);
>  
>  		if (!page)
>  			return NULL;
> -		split_page(page, order);
>  		pfn += 1 << order;
>  	}

-- 
Regards,
Pratyush Yadav

Re: [PATCH 2/2] kho: fix restoring of contiguous ranges of order-0 pages

[Mike Rapoport]

Hi Pratyush,

On Tue, Nov 25, 2025 at 02:45:59PM +0100, Pratyush Yadav wrote:
> On Tue, Nov 25 2025, Mike Rapoport wrote:

...

> > @@ -243,11 +243,16 @@ static struct page *kho_restore_page(phys_addr_t phys)
> >  	/* Head page gets refcount of 1. */
> >  	set_page_count(page, 1);
> >  
> > -	/* For higher order folios, tail pages get a page count of zero. */
> > +	/*
> > +	 * For higher order folios, tail pages get a page count of zero.
> > +	 * For physically contiguous order-0 pages every pages gets a page
> > +	 * count of 1
> > +	 */
> > +	ref_cnt = is_folio ? 0 : 1;
> >  	for (unsigned int i = 1; i < nr_pages; i++)
> > -		set_page_count(page + i, 0);
> > +		set_page_count(page + i, ref_cnt);
> >  
> > -	if (info.order > 0)
> > +	if (is_folio && info.order)
> 
> This is getting a bit difficult to parse. Let's separate out folio and
> page initialization to separate helpers:
 
Sorry, I've missed this earlier and now the patches are in akpm's -stable
branch.
Let's postpone these changes for the next cycle, maybe along with support
for deferred initialization of struct page.

> 	/* Initalize 0-order KHO pages */
> 	static void kho_init_page(struct page *page, unsigned int nr_pages)
> 	{
> 		for (unsigned int i = 0; i < nr_pages; i++)
> 			set_page_count(page + i, 1);
> 	}
> 	
> 	static void kho_init_folio(struct page *page, unsigned int order)
> 	{
> 		unsigned int nr_pages = (1 << order);
> 	
> 		/* Head page gets refcount of 1. */
> 		set_page_count(page, 1);
> 	
> 		/* For higher order folios, tail pages get a page count of zero. */
> 		for (unsigned int i = 1; i < nr_pages; i++)
> 			set_page_count(page + i, 0);
> 	
> 		if (order > 0)
> 			prep_compound_page(page, order);
> 	}

-- 
Sincerely yours,
Mike.

Re: [PATCH 2/2] kho: fix restoring of contiguous ranges of order-0 pages

[Pratyush Yadav]

On Mon, Dec 01 2025, Mike Rapoport wrote:

> Hi Pratyush,
>
> On Tue, Nov 25, 2025 at 02:45:59PM +0100, Pratyush Yadav wrote:
>> On Tue, Nov 25 2025, Mike Rapoport wrote:
>
> ...
>
>> > @@ -243,11 +243,16 @@ static struct page *kho_restore_page(phys_addr_t phys)
>> >  	/* Head page gets refcount of 1. */
>> >  	set_page_count(page, 1);
>> >  
>> > -	/* For higher order folios, tail pages get a page count of zero. */
>> > +	/*
>> > +	 * For higher order folios, tail pages get a page count of zero.
>> > +	 * For physically contiguous order-0 pages every pages gets a page
>> > +	 * count of 1
>> > +	 */
>> > +	ref_cnt = is_folio ? 0 : 1;
>> >  	for (unsigned int i = 1; i < nr_pages; i++)
>> > -		set_page_count(page + i, 0);
>> > +		set_page_count(page + i, ref_cnt);
>> >  
>> > -	if (info.order > 0)
>> > +	if (is_folio && info.order)
>> 
>> This is getting a bit difficult to parse. Let's separate out folio and
>> page initialization to separate helpers:
>  
> Sorry, I've missed this earlier and now the patches are in akpm's -stable
> branch.
> Let's postpone these changes for the next cycle, maybe along with support
> for deferred initialization of struct page.

Sure, no problem.

[...]

-- 
Regards,
Pratyush Yadav