[PATCH 0/2] mm: replace custom ratelimiting logic

[PATCH 0/2] mm: replace custom ratelimiting logic

[Pedro Falcato]

The kernel has a perfectly cromulent and mostly-equivalent variant in
lib/ratelimit.c that can be used.

To: Andrew Morton <akpm@linux-foundation.org>
To: David Hildenbrand <david@kernel.org>
Cc: Lorenzo Stoakes <ljs@kernel.org>
Cc: "Liam R. Howlett" <liam@infradead.org>
Cc: Vlastimil Babka <vbabka@kernel.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Brendan Jackman <brendan.jackman@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Zi Yan <ziy@nvidia.com>
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org

Pedro Falcato (2):
  mm: replace custom bad page map ratelimiting logic
  mm/page_alloc: replace custom bad page ratelimiting logic

 mm/memory.c     | 30 +++---------------------------
 mm/page_alloc.c | 28 +++++-----------------------
 2 files changed, 8 insertions(+), 50 deletions(-)

-- 
2.55.0

[PATCH 1/2] mm: replace custom bad page map ratelimiting logic

[Pedro Falcato]

The current logic (allow up to $BURST prints per minute) can be entirely
replaced by the generic version in lib/ratelimit.c, used around the kernel.

Do so. The only functional difference should be that the new logs will
read something like:

KERN_WARNING "print_bad_page_map: %d callbacks suppressed\n", ...

But that should be fine enough.

Signed-off-by: Pedro Falcato <pfalcato@suse.de>
---
 mm/memory.c | 30 +++---------------------------
 1 file changed, 3 insertions(+), 27 deletions(-)

diff --git a/mm/memory.c b/mm/memory.c
index 4134ac607ee0..b4be57b590ce 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -492,32 +492,8 @@ static inline void add_mm_rss_vec(struct mm_struct *mm, int *rss)
 			add_mm_counter(mm, i, rss[i]);
 }
 
-static bool is_bad_page_map_ratelimited(void)
-{
-	static unsigned long resume;
-	static unsigned long nr_shown;
-	static unsigned long nr_unshown;
-
-	/*
-	 * Allow a burst of 60 reports, then keep quiet for that minute;
-	 * or allow a steady drip of one report per second.
-	 */
-	if (nr_shown == 60) {
-		if (time_before(jiffies, resume)) {
-			nr_unshown++;
-			return true;
-		}
-		if (nr_unshown) {
-			pr_alert("BUG: Bad page map: %lu messages suppressed\n",
-				 nr_unshown);
-			nr_unshown = 0;
-		}
-		nr_shown = 0;
-	}
-	if (nr_shown++ == 0)
-		resume = jiffies + 60 * HZ;
-	return false;
-}
+/* Allow a burst of 60 bad page map reports per minute. */
+static DEFINE_RATELIMIT_STATE(bad_page_map_ratelimit, 60 * HZ, 60);
 
 static void ptval_bytes_to_hex_str(char *buf, size_t buf_size, const void *entry, size_t entry_size)
 {
@@ -633,7 +609,7 @@ static void print_bad_page_map(struct vm_area_struct *vma,
 	char entry_str[PTVAL_STR_MAX];
 	pgoff_t index, anon_index;
 
-	if (is_bad_page_map_ratelimited())
+	if (!__ratelimit(&bad_page_map_ratelimit))
 		return;
 
 	mapping = vma->vm_file ? vma->vm_file->f_mapping : NULL;
-- 
2.55.0

[PATCH 2/2] mm/page_alloc: replace custom bad page ratelimiting logic

[Pedro Falcato]

The current logic (allow up to $BURST prints per minute) can be entirely
replaced by the generic version in lib/ratelimit.c, used around the kernel.

Do so. The only functional difference should be that the new logs will
read something like:

KERN_WARNING "bad_page: %d callbacks suppressed\n", ...

But that should be fine enough.

Signed-off-by: Pedro Falcato <pfalcato@suse.de>
---
 mm/page_alloc.c | 28 +++++-----------------------
 1 file changed, 5 insertions(+), 23 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 12fac9084c48..9ccfd87bc273 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -613,31 +613,13 @@ static inline bool __maybe_unused bad_range(struct zone *zone, struct page *page
 }
 #endif
 
+/* Allow a burst of 60 reports per minute */
+static DEFINE_RATELIMIT_STATE(bad_page_ratelimit, 60 * HZ, 60);
+
 static void bad_page(struct page *page, const char *reason)
 {
-	static unsigned long resume;
-	static unsigned long nr_shown;
-	static unsigned long nr_unshown;
-
-	/*
-	 * Allow a burst of 60 reports, then keep quiet for that minute;
-	 * or allow a steady drip of one report per second.
-	 */
-	if (nr_shown == 60) {
-		if (time_before(jiffies, resume)) {
-			nr_unshown++;
-			goto out;
-		}
-		if (nr_unshown) {
-			pr_alert(
-			      "BUG: Bad page state: %lu messages suppressed\n",
-				nr_unshown);
-			nr_unshown = 0;
-		}
-		nr_shown = 0;
-	}
-	if (nr_shown++ == 0)
-		resume = jiffies + 60 * HZ;
+	if (!__ratelimit(&bad_page_ratelimit))
+		goto out;
 
 	pr_alert("BUG: Bad page state in process %s  pfn:%05lx\n",
 		current->comm, page_to_pfn(page));
-- 
2.55.0

Re: [PATCH 1/2] mm: replace custom bad page map ratelimiting logic

[Johannes Weiner]

On Tue, Aug 11, 2026 at 06:21:55PM +0100, Pedro Falcato wrote:
> The current logic (allow up to $BURST prints per minute) can be entirely
> replaced by the generic version in lib/ratelimit.c, used around the kernel.
> 
> Do so. The only functional difference should be that the new logs will
> read something like:
> 
> KERN_WARNING "print_bad_page_map: %d callbacks suppressed\n", ...
> 
> But that should be fine enough.
> 
> Signed-off-by: Pedro Falcato <pfalcato@suse.de>

Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Re: [PATCH 2/2] mm/page_alloc: replace custom bad page ratelimiting logic

[Johannes Weiner]

On Tue, Aug 11, 2026 at 06:21:56PM +0100, Pedro Falcato wrote:
> The current logic (allow up to $BURST prints per minute) can be entirely
> replaced by the generic version in lib/ratelimit.c, used around the kernel.
> 
> Do so. The only functional difference should be that the new logs will
> read something like:
> 
> KERN_WARNING "bad_page: %d callbacks suppressed\n", ...
> 
> But that should be fine enough.
> 
> Signed-off-by: Pedro Falcato <pfalcato@suse.de>

Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Re: [PATCH 1/2] mm: replace custom bad page map ratelimiting logic

[Zi Yan]

On Tue Aug 11, 2026 at 1:21 PM EDT, Pedro Falcato wrote:
> The current logic (allow up to $BURST prints per minute) can be entirely
> replaced by the generic version in lib/ratelimit.c, used around the kernel.
>
> Do so. The only functional difference should be that the new logs will
> read something like:
>
> KERN_WARNING "print_bad_page_map: %d callbacks suppressed\n", ...
>
> But that should be fine enough.
>
> Signed-off-by: Pedro Falcato <pfalcato@suse.de>
> ---
>  mm/memory.c | 30 +++---------------------------
>  1 file changed, 3 insertions(+), 27 deletions(-)
>
Great!

Acked-by: Zi Yan <ziy@nvidia.com>

-- 
Best Regards,
Yan, Zi

Re: [PATCH 2/2] mm/page_alloc: replace custom bad page ratelimiting logic

[Zi Yan]

On Tue Aug 11, 2026 at 1:21 PM EDT, Pedro Falcato wrote:
> The current logic (allow up to $BURST prints per minute) can be entirely
> replaced by the generic version in lib/ratelimit.c, used around the kernel.
>
> Do so. The only functional difference should be that the new logs will
> read something like:
>
> KERN_WARNING "bad_page: %d callbacks suppressed\n", ...
>
> But that should be fine enough.
>
> Signed-off-by: Pedro Falcato <pfalcato@suse.de>
> ---
>  mm/page_alloc.c | 28 +++++-----------------------
>  1 file changed, 5 insertions(+), 23 deletions(-)
>
LGTM.

Acked-by: Zi Yan <ziy@nvidia.com>

-- 
Best Regards,
Yan, Zi