1. Patchew
  2. linux
  3. View series

[PATCH v2] mm/memory-failure: Fix redundant updates for already poisoned pages

Kyle Meyer posted 1 patch 1 month ago 
mm/memory-failure.c | 13 ++++++-------
1 file changed, 6 insertions(+), 7 deletions(-)
[PATCH v2] mm/memory-failure: Fix redundant updates for already poisoned pages
Posted by Kyle Meyer 1 month ago 
Duplicate memory errors can be reported by multiple sources.

Passing an already poisoned page to action_result() causes issues:

* The amount of hardware corrupted memory is incorrectly updated.
* Per NUMA node MF stats are incorrectly updated.
* Redundant "already poisoned" messages are printed.

Avoid those issues by:

* Skipping hardware corrupted memory updates for already poisoned pages.
* Skipping per NUMA node MF stats updates for already poisoned pages.
* Dropping redundant "already poisoned" messages.

Make MF_MSG_ALREADY_POISONED consistent with other action_page_types and
make calls to action_result() consistent for already poisoned
normal pages and huge pages.

Fixes: b8b9488d50b7 ("mm/memory-failure: improve memory failure action_result messages")
Signed-off-by: Kyle Meyer <kyle.meyer@hpe.com>
---

v1 -> v2:
 * Continue passing poisoned pages to action_result() with MF_FAILED but don't
update anything.
 * https://lore.kernel.org/all/20250821164445.14467-1-kyle.meyer@hpe.com

---
 mm/memory-failure.c | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index fc30ca4804bf..10b3c281c2ae 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -956,7 +956,7 @@ static const char * const action_page_types[] = {
 	[MF_MSG_BUDDY]			= "free buddy page",
 	[MF_MSG_DAX]			= "dax page",
 	[MF_MSG_UNSPLIT_THP]		= "unsplit thp",
-	[MF_MSG_ALREADY_POISONED]	= "already poisoned",
+	[MF_MSG_ALREADY_POISONED]	= "already poisoned page",
 	[MF_MSG_UNKNOWN]		= "unknown page",
 };
 
@@ -1349,9 +1349,10 @@ static int action_result(unsigned long pfn, enum mf_action_page_type type,
 {
 	trace_memory_failure_event(pfn, type, result);
 
-	num_poisoned_pages_inc(pfn);
-
-	update_per_node_mf_stats(pfn, result);
+	if (type != MF_MSG_ALREADY_POISONED) {
+		num_poisoned_pages_inc(pfn);
+		update_per_node_mf_stats(pfn, result);
+	}
 
 	pr_err("%#lx: recovery action for %s: %s\n",
 		pfn, action_page_types[type], action_name[result]);
@@ -2094,12 +2095,11 @@ static int try_memory_failure_hugetlb(unsigned long pfn, int flags, int *hugetlb
 		*hugetlb = 0;
 		return 0;
 	} else if (res == -EHWPOISON) {
-		pr_err("%#lx: already hardware poisoned\n", pfn);
 		if (flags & MF_ACTION_REQUIRED) {
 			folio = page_folio(p);
 			res = kill_accessing_process(current, folio_pfn(folio), flags);
-			action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
 		}
+		action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
 		return res;
 	} else if (res == -EBUSY) {
 		if (!(flags & MF_NO_RETRY)) {
@@ -2285,7 +2285,6 @@ int memory_failure(unsigned long pfn, int flags)
 		goto unlock_mutex;
 
 	if (TestSetPageHWPoison(p)) {
-		pr_err("%#lx: already hardware poisoned\n", pfn);
 		res = -EHWPOISON;
 		if (flags & MF_ACTION_REQUIRED)
 			res = kill_accessing_process(current, pfn, flags);
-- 
2.50.1
Re: [PATCH v2] mm/memory-failure: Fix redundant updates for already poisoned pages
Posted by Miaohe Lin 1 month ago 
On 2025/8/29 2:38, Kyle Meyer wrote:
> Duplicate memory errors can be reported by multiple sources.
> 
> Passing an already poisoned page to action_result() causes issues:
> 
> * The amount of hardware corrupted memory is incorrectly updated.
> * Per NUMA node MF stats are incorrectly updated.
> * Redundant "already poisoned" messages are printed.
> 
> Avoid those issues by:
> 
> * Skipping hardware corrupted memory updates for already poisoned pages.
> * Skipping per NUMA node MF stats updates for already poisoned pages.
> * Dropping redundant "already poisoned" messages.
> 
> Make MF_MSG_ALREADY_POISONED consistent with other action_page_types and
> make calls to action_result() consistent for already poisoned
> normal pages and huge pages.
> 
> Fixes: b8b9488d50b7 ("mm/memory-failure: improve memory failure action_result messages")
> Signed-off-by: Kyle Meyer <kyle.meyer@hpe.com>

Thanks for your patch. This patch looks good to me.

Acked-by: Miaohe Lin <linmiaohe@huawei.com>
Thanks.
.

> ---
> 
> v1 -> v2:
>  * Continue passing poisoned pages to action_result() with MF_FAILED but don't
> update anything.
>  * https://lore.kernel.org/all/20250821164445.14467-1-kyle.meyer@hpe.com
> 
> ---
>  mm/memory-failure.c | 13 ++++++-------
>  1 file changed, 6 insertions(+), 7 deletions(-)
> 
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index fc30ca4804bf..10b3c281c2ae 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -956,7 +956,7 @@ static const char * const action_page_types[] = {
>  	[MF_MSG_BUDDY]			= "free buddy page",
>  	[MF_MSG_DAX]			= "dax page",
>  	[MF_MSG_UNSPLIT_THP]		= "unsplit thp",
> -	[MF_MSG_ALREADY_POISONED]	= "already poisoned",
> +	[MF_MSG_ALREADY_POISONED]	= "already poisoned page",
>  	[MF_MSG_UNKNOWN]		= "unknown page",
>  };
>  
> @@ -1349,9 +1349,10 @@ static int action_result(unsigned long pfn, enum mf_action_page_type type,
>  {
>  	trace_memory_failure_event(pfn, type, result);
>  
> -	num_poisoned_pages_inc(pfn);
> -
> -	update_per_node_mf_stats(pfn, result);
> +	if (type != MF_MSG_ALREADY_POISONED) {
> +		num_poisoned_pages_inc(pfn);
> +		update_per_node_mf_stats(pfn, result);
> +	}
>  
>  	pr_err("%#lx: recovery action for %s: %s\n",
>  		pfn, action_page_types[type], action_name[result]);
> @@ -2094,12 +2095,11 @@ static int try_memory_failure_hugetlb(unsigned long pfn, int flags, int *hugetlb
>  		*hugetlb = 0;
>  		return 0;
>  	} else if (res == -EHWPOISON) {
> -		pr_err("%#lx: already hardware poisoned\n", pfn);
>  		if (flags & MF_ACTION_REQUIRED) {
>  			folio = page_folio(p);
>  			res = kill_accessing_process(current, folio_pfn(folio), flags);
> -			action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>  		}
> +		action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>  		return res;
>  	} else if (res == -EBUSY) {
>  		if (!(flags & MF_NO_RETRY)) {
> @@ -2285,7 +2285,6 @@ int memory_failure(unsigned long pfn, int flags)
>  		goto unlock_mutex;
>  
>  	if (TestSetPageHWPoison(p)) {
> -		pr_err("%#lx: already hardware poisoned\n", pfn);
>  		res = -EHWPOISON;
>  		if (flags & MF_ACTION_REQUIRED)
>  			res = kill_accessing_process(current, pfn, flags);
>
Re: [PATCH v2] mm/memory-failure: Fix redundant updates for already poisoned pages
Posted by jane.chu@oracle.com 1 month ago 

On 8/28/2025 11:38 AM, Kyle Meyer wrote:
> Duplicate memory errors can be reported by multiple sources.
> 
> Passing an already poisoned page to action_result() causes issues:
> 
> * The amount of hardware corrupted memory is incorrectly updated.
> * Per NUMA node MF stats are incorrectly updated.
> * Redundant "already poisoned" messages are printed.
> 
> Avoid those issues by:
> 
> * Skipping hardware corrupted memory updates for already poisoned pages.
> * Skipping per NUMA node MF stats updates for already poisoned pages.
> * Dropping redundant "already poisoned" messages.
> 
> Make MF_MSG_ALREADY_POISONED consistent with other action_page_types and
> make calls to action_result() consistent for already poisoned
> normal pages and huge pages.
> 
> Fixes: b8b9488d50b7 ("mm/memory-failure: improve memory failure action_result messages")
> Signed-off-by: Kyle Meyer <kyle.meyer@hpe.com>
> ---
> 
> v1 -> v2:
>   * Continue passing poisoned pages to action_result() with MF_FAILED but don't
> update anything.
>   * https://lore.kernel.org/all/20250821164445.14467-1-kyle.meyer@hpe.com
> 
> ---
>   mm/memory-failure.c | 13 ++++++-------
>   1 file changed, 6 insertions(+), 7 deletions(-)
> 
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index fc30ca4804bf..10b3c281c2ae 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -956,7 +956,7 @@ static const char * const action_page_types[] = {
>   	[MF_MSG_BUDDY]			= "free buddy page",
>   	[MF_MSG_DAX]			= "dax page",
>   	[MF_MSG_UNSPLIT_THP]		= "unsplit thp",
> -	[MF_MSG_ALREADY_POISONED]	= "already poisoned",
> +	[MF_MSG_ALREADY_POISONED]	= "already poisoned page",
>   	[MF_MSG_UNKNOWN]		= "unknown page",
>   };
>   
> @@ -1349,9 +1349,10 @@ static int action_result(unsigned long pfn, enum mf_action_page_type type,
>   {
>   	trace_memory_failure_event(pfn, type, result);
>   
> -	num_poisoned_pages_inc(pfn);
> -
> -	update_per_node_mf_stats(pfn, result);
> +	if (type != MF_MSG_ALREADY_POISONED) {
> +		num_poisoned_pages_inc(pfn);
> +		update_per_node_mf_stats(pfn, result);
> +	}
>   
>   	pr_err("%#lx: recovery action for %s: %s\n",
>   		pfn, action_page_types[type], action_name[result]);
> @@ -2094,12 +2095,11 @@ static int try_memory_failure_hugetlb(unsigned long pfn, int flags, int *hugetlb
>   		*hugetlb = 0;
>   		return 0;
>   	} else if (res == -EHWPOISON) {
> -		pr_err("%#lx: already hardware poisoned\n", pfn);
>   		if (flags & MF_ACTION_REQUIRED) {
>   			folio = page_folio(p);
>   			res = kill_accessing_process(current, folio_pfn(folio), flags);
> -			action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>   		}
> +		action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>   		return res;
>   	} else if (res == -EBUSY) {
>   		if (!(flags & MF_NO_RETRY)) {
> @@ -2285,7 +2285,6 @@ int memory_failure(unsigned long pfn, int flags)
>   		goto unlock_mutex;
>   
>   	if (TestSetPageHWPoison(p)) {
> -		pr_err("%#lx: already hardware poisoned\n", pfn);
>   		res = -EHWPOISON;
>   		if (flags & MF_ACTION_REQUIRED)
>   			res = kill_accessing_process(current, pfn, flags);

Looks good, thanks!

Reviewed-by: Jane Chu <jane.chu@oracle.com>

-jane
Re: [PATCH v2] mm/memory-failure: Fix redundant updates for already poisoned pages
Posted by David Hildenbrand 1 month ago 
On 28.08.25 20:38, Kyle Meyer wrote:
> Duplicate memory errors can be reported by multiple sources.
> 
> Passing an already poisoned page to action_result() causes issues:
> 
> * The amount of hardware corrupted memory is incorrectly updated.
> * Per NUMA node MF stats are incorrectly updated.
> * Redundant "already poisoned" messages are printed.
> 
> Avoid those issues by:
> 
> * Skipping hardware corrupted memory updates for already poisoned pages.
> * Skipping per NUMA node MF stats updates for already poisoned pages.
> * Dropping redundant "already poisoned" messages.
> 
> Make MF_MSG_ALREADY_POISONED consistent with other action_page_types and
> make calls to action_result() consistent for already poisoned
> normal pages and huge pages.
> 
> Fixes: b8b9488d50b7 ("mm/memory-failure: improve memory failure action_result messages")
> Signed-off-by: Kyle Meyer <kyle.meyer@hpe.com>
> ---
> 
> v1 -> v2:
>   * Continue passing poisoned pages to action_result() with MF_FAILED but don't
> update anything.
>   * https://lore.kernel.org/all/20250821164445.14467-1-kyle.meyer@hpe.com
> 
> ---
>   mm/memory-failure.c | 13 ++++++-------
>   1 file changed, 6 insertions(+), 7 deletions(-)
> 
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index fc30ca4804bf..10b3c281c2ae 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -956,7 +956,7 @@ static const char * const action_page_types[] = {
>   	[MF_MSG_BUDDY]			= "free buddy page",
>   	[MF_MSG_DAX]			= "dax page",
>   	[MF_MSG_UNSPLIT_THP]		= "unsplit thp",
> -	[MF_MSG_ALREADY_POISONED]	= "already poisoned",
> +	[MF_MSG_ALREADY_POISONED]	= "already poisoned page",
>   	[MF_MSG_UNKNOWN]		= "unknown page",
>   };
>   
> @@ -1349,9 +1349,10 @@ static int action_result(unsigned long pfn, enum mf_action_page_type type,
>   {
>   	trace_memory_failure_event(pfn, type, result);
>   
> -	num_poisoned_pages_inc(pfn);
> -
> -	update_per_node_mf_stats(pfn, result);
> +	if (type != MF_MSG_ALREADY_POISONED) {
> +		num_poisoned_pages_inc(pfn);
> +		update_per_node_mf_stats(pfn, result);
> +	}
>   
>   	pr_err("%#lx: recovery action for %s: %s\n",
>   		pfn, action_page_types[type], action_name[result]);
> @@ -2094,12 +2095,11 @@ static int try_memory_failure_hugetlb(unsigned long pfn, int flags, int *hugetlb
>   		*hugetlb = 0;
>   		return 0;
>   	} else if (res == -EHWPOISON) {
> -		pr_err("%#lx: already hardware poisoned\n", pfn);
>   		if (flags & MF_ACTION_REQUIRED) {
>   			folio = page_folio(p);
>   			res = kill_accessing_process(current, folio_pfn(folio), flags);
> -			action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>   		}
> +		action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);

Was briefly confused by that change, but action_result() now essentially 
only traces + pr_err() with MF_MSG_ALREADY_POISONED.

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers

David / dhildenb
Re: [PATCH v2] mm/memory-failure: Fix redundant updates for already poisoned pages
Posted by Jiaqi Yan 1 month ago 
On Thu, Aug 28, 2025 at 11:38 AM Kyle Meyer <kyle.meyer@hpe.com> wrote:
>
> Duplicate memory errors can be reported by multiple sources.
>
> Passing an already poisoned page to action_result() causes issues:
>
> * The amount of hardware corrupted memory is incorrectly updated.
> * Per NUMA node MF stats are incorrectly updated.
> * Redundant "already poisoned" messages are printed.
>
> Avoid those issues by:
>
> * Skipping hardware corrupted memory updates for already poisoned pages.
> * Skipping per NUMA node MF stats updates for already poisoned pages.
> * Dropping redundant "already poisoned" messages.
>
> Make MF_MSG_ALREADY_POISONED consistent with other action_page_types and
> make calls to action_result() consistent for already poisoned
> normal pages and huge pages.
>
> Fixes: b8b9488d50b7 ("mm/memory-failure: improve memory failure action_result messages")
> Signed-off-by: Kyle Meyer <kyle.meyer@hpe.com>
> ---
>
> v1 -> v2:
>  * Continue passing poisoned pages to action_result() with MF_FAILED but don't
> update anything.
>  * https://lore.kernel.org/all/20250821164445.14467-1-kyle.meyer@hpe.com
>
> ---
>  mm/memory-failure.c | 13 ++++++-------
>  1 file changed, 6 insertions(+), 7 deletions(-)
>
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index fc30ca4804bf..10b3c281c2ae 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -956,7 +956,7 @@ static const char * const action_page_types[] = {
>         [MF_MSG_BUDDY]                  = "free buddy page",
>         [MF_MSG_DAX]                    = "dax page",
>         [MF_MSG_UNSPLIT_THP]            = "unsplit thp",
> -       [MF_MSG_ALREADY_POISONED]       = "already poisoned",
> +       [MF_MSG_ALREADY_POISONED]       = "already poisoned page",
>         [MF_MSG_UNKNOWN]                = "unknown page",
>  };
>
> @@ -1349,9 +1349,10 @@ static int action_result(unsigned long pfn, enum mf_action_page_type type,
>  {
>         trace_memory_failure_event(pfn, type, result);
>
> -       num_poisoned_pages_inc(pfn);
> -
> -       update_per_node_mf_stats(pfn, result);
> +       if (type != MF_MSG_ALREADY_POISONED) {
> +               num_poisoned_pages_inc(pfn);
> +               update_per_node_mf_stats(pfn, result);
> +       }

Thanks Kyle, the code looks good to me. While you wait for Miaohe's
review/ack...

Reviewed-by: Jiaqi Yan <jiaqiyan@google.com>

>
>         pr_err("%#lx: recovery action for %s: %s\n",
>                 pfn, action_page_types[type], action_name[result]);
> @@ -2094,12 +2095,11 @@ static int try_memory_failure_hugetlb(unsigned long pfn, int flags, int *hugetlb
>                 *hugetlb = 0;
>                 return 0;
>         } else if (res == -EHWPOISON) {
> -               pr_err("%#lx: already hardware poisoned\n", pfn);
>                 if (flags & MF_ACTION_REQUIRED) {
>                         folio = page_folio(p);
>                         res = kill_accessing_process(current, folio_pfn(folio), flags);
> -                       action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>                 }
> +               action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED);
>                 return res;
>         } else if (res == -EBUSY) {
>                 if (!(flags & MF_NO_RETRY)) {
> @@ -2285,7 +2285,6 @@ int memory_failure(unsigned long pfn, int flags)
>                 goto unlock_mutex;
>
>         if (TestSetPageHWPoison(p)) {
> -               pr_err("%#lx: already hardware poisoned\n", pfn);
>                 res = -EHWPOISON;
>                 if (flags & MF_ACTION_REQUIRED)
>                         res = kill_accessing_process(current, pfn, flags);
> --
> 2.50.1
>

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.