[PATCH 0/2] iommu: debug-pagealloc: Remove pfn_valid() usage

[PATCH 0/2] iommu: debug-pagealloc: Remove pfn_valid() usage

[Mostafa Saleh]

This is a small fix for the new config IOMMU_DEBUG_PAGEALLOC based
on the discussion:
https://lore.kernel.org/linux-iommu/CAFgf54pBAUm3ao-UJksiuGKtvv4wzRyFq_uKwLe0H1ettO4DLQ@mail.gmail.com/

Where it was concluded that pfn_valid() is not enough to validate
physical addresses before access to page_ext.

The first patch introduces a new function in page_ext that takes a
physical address as an argument, and the second patch uses it instead
of calling pfn_valid() and phys_to_page()

Benchmarks with the new implementation can be found in:
https://lore.kernel.org/linux-iommu/20260114164322.787125-1-smostafa@google.com/

This series applies to iommu/core tree.

Mostafa Saleh (2):
  mm/page_ext: Add page_ext_get_phys()
  iommu: debug-pagealloc: Use page_ext_get_phys()

 drivers/iommu/iommu-debug-pagealloc.c | 31 ++++++++++++---------------
 include/linux/page_ext.h              |  6 ++++++
 mm/page_ext.c                         | 23 ++++++++++++++++++++
 3 files changed, 43 insertions(+), 17 deletions(-)

-- 
2.52.0.457.g6b5491de43-goog

[PATCH 1/2] mm/page_ext: Add page_ext_get_phys()

[Mostafa Saleh]

In the next patches, IOMMU code will add data to page_ext.
The IOMMU code would operate on physical addresses which
can be outside of system RAM.

Add a new function page_ext_get_phys() to abstract the logic of
checking the address and returning the page_ext.

Signed-off-by: Mostafa Saleh <smostafa@google.com>
---
 include/linux/page_ext.h |  6 ++++++
 mm/page_ext.c            | 23 +++++++++++++++++++++++
 2 files changed, 29 insertions(+)

diff --git a/include/linux/page_ext.h b/include/linux/page_ext.h
index 76c817162d2f..bd373496e166 100644
--- a/include/linux/page_ext.h
+++ b/include/linux/page_ext.h
@@ -93,6 +93,7 @@ static inline bool page_ext_iter_next_fast_possible(unsigned long next_pfn)
 #endif
 
 extern struct page_ext *page_ext_get(const struct page *page);
+extern struct page_ext *page_ext_get_phys(phys_addr_t phys);
 extern void page_ext_put(struct page_ext *page_ext);
 extern struct page_ext *page_ext_lookup(unsigned long pfn);
 
@@ -215,6 +216,11 @@ static inline struct page_ext *page_ext_get(const struct page *page)
 	return NULL;
 }
 
+static inline struct page_ext *page_ext_get_phys(phys_addr_t phys)
+{
+	return NULL;
+}
+
 static inline void page_ext_put(struct page_ext *page_ext)
 {
 }
diff --git a/mm/page_ext.c b/mm/page_ext.c
index 297e4cd8ce90..5fe65a0ac4f3 100644
--- a/mm/page_ext.c
+++ b/mm/page_ext.c
@@ -538,6 +538,29 @@ struct page_ext *page_ext_get(const struct page *page)
 	return page_ext;
 }
 
+/**
+ * page_ext_get_phys() - Get the page_ext structure for a physical address.
+ * @phys: The physical address to query.
+ *
+ * This function safely gets the `struct page_ext` associated with a given
+ * physical address. It performs validation to ensure the address corresponds
+ * to a valid, online struct page before attempting to access it.
+ * It should return NULL for (MMIO, ZONE_DEVICE, holes, offline memory)
+ *
+ * Return: NULL if no page_ext exists for this physical address.
+ * Context: Any context.  Caller may not sleep until they have called
+ * page_ext_put().
+ */
+struct page_ext *page_ext_get_phys(phys_addr_t phys)
+{
+	struct page *page = pfn_to_online_page(__phys_to_pfn(phys));
+
+	if (!page)
+		return NULL;
+
+	return page_ext_get(page);
+}
+
 /**
  * page_ext_put() - Working with page extended information is done.
  * @page_ext: Page extended information received from page_ext_get().
-- 
2.52.0.457.g6b5491de43-goog

Re: [PATCH 1/2] mm/page_ext: Add page_ext_get_phys()

[Jörg Rödel]

On Mon, Jan 19, 2026 at 02:22:45PM +0000, Mostafa Saleh wrote:
> +static inline struct page_ext *page_ext_get_phys(phys_addr_t phys)

The name is misleading as it indicates that the function returns a physical
address. Maybe name it page_ext_from_phys()?

> +{
> +	return NULL;
> +}
> +
>  static inline void page_ext_put(struct page_ext *page_ext)
>  {
>  }
> diff --git a/mm/page_ext.c b/mm/page_ext.c
> index 297e4cd8ce90..5fe65a0ac4f3 100644
> --- a/mm/page_ext.c
> +++ b/mm/page_ext.c
> @@ -538,6 +538,29 @@ struct page_ext *page_ext_get(const struct page *page)
>  	return page_ext;
>  }
>  
> +/**
> + * page_ext_get_phys() - Get the page_ext structure for a physical address.
> + * @phys: The physical address to query.
> + *
> + * This function safely gets the `struct page_ext` associated with a given
> + * physical address. It performs validation to ensure the address corresponds
> + * to a valid, online struct page before attempting to access it.
> + * It should return NULL for (MMIO, ZONE_DEVICE, holes, offline memory)

It should?

-Joerg

Re: [PATCH 1/2] mm/page_ext: Add page_ext_get_phys()

[Mostafa Saleh]

On Mon, Jan 19, 2026 at 3:49 PM Jörg Rödel <joro@8bytes.org> wrote:
>
> On Mon, Jan 19, 2026 at 02:22:45PM +0000, Mostafa Saleh wrote:
> > +static inline struct page_ext *page_ext_get_phys(phys_addr_t phys)
>
> The name is misleading as it indicates that the function returns a physical
> address. Maybe name it page_ext_from_phys()?

I will update it.

>
> > +{
> > +     return NULL;
> > +}
> > +
> >  static inline void page_ext_put(struct page_ext *page_ext)
> >  {
> >  }
> > diff --git a/mm/page_ext.c b/mm/page_ext.c
> > index 297e4cd8ce90..5fe65a0ac4f3 100644
> > --- a/mm/page_ext.c
> > +++ b/mm/page_ext.c
> > @@ -538,6 +538,29 @@ struct page_ext *page_ext_get(const struct page *page)
> >       return page_ext;
> >  }
> >
> > +/**
> > + * page_ext_get_phys() - Get the page_ext structure for a physical address.
> > + * @phys: The physical address to query.
> > + *
> > + * This function safely gets the `struct page_ext` associated with a given
> > + * physical address. It performs validation to ensure the address corresponds
> > + * to a valid, online struct page before attempting to access it.
> > + * It should return NULL for (MMIO, ZONE_DEVICE, holes, offline memory)
>
> It should?
>

Yes, I can make it more definitive as "it returns"

Thanks,
Mostafa

> -Joerg

[PATCH 2/2] iommu: debug-pagealloc: Use page_ext_get_phys()

[Mostafa Saleh]

Instead of calling pfn_valid() and then getting the page, call
the newly added function page_ext_get_phys(), which would also
check for MMIO and offline memory and return NULL in that case.

Signed-off-by: Mostafa Saleh <smostafa@google.com>
---
 drivers/iommu/iommu-debug-pagealloc.c | 31 ++++++++++++---------------
 1 file changed, 14 insertions(+), 17 deletions(-)

diff --git a/drivers/iommu/iommu-debug-pagealloc.c b/drivers/iommu/iommu-debug-pagealloc.c
index c080a38f45a4..9343af3bdbb6 100644
--- a/drivers/iommu/iommu-debug-pagealloc.c
+++ b/drivers/iommu/iommu-debug-pagealloc.c
@@ -30,14 +30,6 @@ struct page_ext_operations page_iommu_debug_ops = {
 	.need = need_iommu_debug,
 };
 
-static struct page_ext *get_iommu_page_ext(phys_addr_t phys)
-{
-	struct page *page = phys_to_page(phys);
-	struct page_ext *page_ext = page_ext_get(page);
-
-	return page_ext;
-}
-
 static struct iommu_debug_metadata *get_iommu_data(struct page_ext *page_ext)
 {
 	return page_ext_data(page_ext, &page_iommu_debug_ops);
@@ -45,18 +37,26 @@ static struct iommu_debug_metadata *get_iommu_data(struct page_ext *page_ext)
 
 static void iommu_debug_inc_page(phys_addr_t phys)
 {
-	struct page_ext *page_ext = get_iommu_page_ext(phys);
-	struct iommu_debug_metadata *d = get_iommu_data(page_ext);
+	struct page_ext *page_ext = page_ext_get_phys(phys);
+	struct iommu_debug_metadata *d;
+
+	if (!page_ext)
+		return;
 
+	d = get_iommu_data(page_ext);
 	WARN_ON(atomic_inc_return_relaxed(&d->ref) <= 0);
 	page_ext_put(page_ext);
 }
 
 static void iommu_debug_dec_page(phys_addr_t phys)
 {
-	struct page_ext *page_ext = get_iommu_page_ext(phys);
-	struct iommu_debug_metadata *d = get_iommu_data(page_ext);
+	struct page_ext *page_ext = page_ext_get_phys(phys);
+	struct iommu_debug_metadata *d;
+
+	if (!page_ext)
+		return;
 
+	d = get_iommu_data(page_ext);
 	WARN_ON(atomic_dec_return_relaxed(&d->ref) < 0);
 	page_ext_put(page_ext);
 }
@@ -104,11 +104,8 @@ void __iommu_debug_map(struct iommu_domain *domain, phys_addr_t phys, size_t siz
 	if (WARN_ON(!phys || check_add_overflow(phys, size, &end)))
 		return;
 
-	for (off = 0 ; off < size ; off += page_size) {
-		if (!pfn_valid(__phys_to_pfn(phys + off)))
-			continue;
+	for (off = 0 ; off < size ; off += page_size)
 		iommu_debug_inc_page(phys + off);
-	}
 }
 
 static void __iommu_debug_update_iova(struct iommu_domain *domain,
@@ -123,7 +120,7 @@ static void __iommu_debug_update_iova(struct iommu_domain *domain,
 	for (off = 0 ; off < size ; off += page_size) {
 		phys_addr_t phys = iommu_iova_to_phys(domain, iova + off);
 
-		if (!phys || !pfn_valid(__phys_to_pfn(phys)))
+		if (!phys)
 			continue;
 
 		if (inc)
-- 
2.52.0.457.g6b5491de43-goog

Re: [PATCH 0/2] iommu: debug-pagealloc: Remove pfn_valid() usage

[David Hildenbrand (Red Hat)]

On 1/19/26 15:22, Mostafa Saleh wrote:
> This is a small fix for the new config IOMMU_DEBUG_PAGEALLOC based
> on the discussion:
> https://lore.kernel.org/linux-iommu/CAFgf54pBAUm3ao-UJksiuGKtvv4wzRyFq_uKwLe0H1ettO4DLQ@mail.gmail.com/
> 
> Where it was concluded that pfn_valid() is not enough to validate
> physical addresses before access to page_ext.
> 
> The first patch introduces a new function in page_ext that takes a
> physical address as an argument, and the second patch uses it instead
> of calling pfn_valid() and phys_to_page()
> 
> Benchmarks with the new implementation can be found in:
> https://lore.kernel.org/linux-iommu/20260114164322.787125-1-smostafa@google.com/
> 
> This series applies to iommu/core tree.

This is part of v7 [1], right? Can't we just apply v7 instead or are the 
commit IDs already stable?

[1] https://lkml.kernel.org/r/20260114164322.787125-1-smostafa@google.com

-- 
Cheers

David

Re: [PATCH 0/2] iommu: debug-pagealloc: Remove pfn_valid() usage

[Mostafa Saleh]

On Mon, Jan 19, 2026 at 2:27 PM David Hildenbrand (Red Hat)
<david@kernel.org> wrote:
>
> On 1/19/26 15:22, Mostafa Saleh wrote:
> > This is a small fix for the new config IOMMU_DEBUG_PAGEALLOC based
> > on the discussion:
> > https://lore.kernel.org/linux-iommu/CAFgf54pBAUm3ao-UJksiuGKtvv4wzRyFq_uKwLe0H1ettO4DLQ@mail.gmail.com/
> >
> > Where it was concluded that pfn_valid() is not enough to validate
> > physical addresses before access to page_ext.
> >
> > The first patch introduces a new function in page_ext that takes a
> > physical address as an argument, and the second patch uses it instead
> > of calling pfn_valid() and phys_to_page()
> >
> > Benchmarks with the new implementation can be found in:
> > https://lore.kernel.org/linux-iommu/20260114164322.787125-1-smostafa@google.com/
> >
> > This series applies to iommu/core tree.
>
> This is part of v7 [1], right? Can't we just apply v7 instead or are the
> commit IDs already stable?
>

It should be stable according to Will:
https://lore.kernel.org/linux-iommu/20260114164322.787125-1-smostafa@google.com/T/#m59d0b455e3f2160cb6f9980a0ae65bc481c53898

Thanks,
Mostafa

> [1] https://lkml.kernel.org/r/20260114164322.787125-1-smostafa@google.com
>
> --
> Cheers
>
> David

Re: [PATCH 0/2] iommu: debug-pagealloc: Remove pfn_valid() usage

[David Hildenbrand (Red Hat)]

On 1/19/26 16:09, Mostafa Saleh wrote:
> On Mon, Jan 19, 2026 at 2:27 PM David Hildenbrand (Red Hat)
> <david@kernel.org> wrote:
>>
>> On 1/19/26 15:22, Mostafa Saleh wrote:
>>> This is a small fix for the new config IOMMU_DEBUG_PAGEALLOC based
>>> on the discussion:
>>> https://lore.kernel.org/linux-iommu/CAFgf54pBAUm3ao-UJksiuGKtvv4wzRyFq_uKwLe0H1ettO4DLQ@mail.gmail.com/
>>>
>>> Where it was concluded that pfn_valid() is not enough to validate
>>> physical addresses before access to page_ext.
>>>
>>> The first patch introduces a new function in page_ext that takes a
>>> physical address as an argument, and the second patch uses it instead
>>> of calling pfn_valid() and phys_to_page()
>>>
>>> Benchmarks with the new implementation can be found in:
>>> https://lore.kernel.org/linux-iommu/20260114164322.787125-1-smostafa@google.com/
>>>
>>> This series applies to iommu/core tree.
>>
>> This is part of v7 [1], right? Can't we just apply v7 instead or are the
>> commit IDs already stable?
>>
> 
> It should be stable according to Will:
> https://lore.kernel.org/linux-iommu/20260114164322.787125-1-smostafa@google.com/T/#m59d0b455e3f2160cb6f9980a0ae65bc481c53898

Shame, probably applied a bit too fast ...

-- 
Cheers

David