[PATCH] mm: drop stale MAX_ORDER references

[PATCH] mm: drop stale MAX_ORDER references

[Qi Xi]

The treewide rename in commit 5e0a760b4441 ("mm, treewide: rename
MAX_ORDER to MAX_PAGE_ORDER") left a few spots still using the old
name:

  - two comments in include/net/mana/mana.h and mm/page_alloc.c;
  - the gdb helper scripts/gdb/linux/mm.py, where self.MAX_ORDER is
    a local mirror of the kernel's MAX_ORDER define.

Rename the leftover instances to MAX_PAGE_ORDER so the tree is
consistent.

No functional changes.

Signed-off-by: Qi Xi <xiqi2@huawei.com>
---
 include/net/mana/mana.h | 4 ++--
 mm/page_alloc.c         | 2 +-
 scripts/gdb/linux/mm.py | 8 ++++----
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
index 04acb6791dbd..1d5bed71d6a7 100644
--- a/include/net/mana/mana.h
+++ b/include/net/mana/mana.h
@@ -40,8 +40,8 @@ enum TRI_STATE {
 #define COMP_ENTRY_SIZE 64
 
 /* This Max value for RX buffers is derived from __alloc_page()'s max page
- * allocation calculation. It allows maximum 2^(MAX_ORDER -1) pages. RX buffer
- * size beyond this value gets rejected by __alloc_page() call.
+ * allocation calculation. It allows maximum 2^(MAX_PAGE_ORDER -1) pages. RX
+ * buffer size beyond this value gets rejected by __alloc_page() call.
  */
 #define MAX_RX_BUFFERS_PER_QUEUE 8192
 #define DEF_RX_BUFFERS_PER_QUEUE 1024
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index ee902a468c2f..42b5b41432c5 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -7803,7 +7803,7 @@ static bool cond_accept_memory(struct zone *zone, unsigned int order,
 	/*
 	 * Watermarks have not been initialized yet.
 	 *
-	 * Accepting one MAX_ORDER page to ensure progress.
+	 * Accepting one MAX_PAGE_ORDER page to ensure progress.
 	 */
 	if (!wmark)
 		return try_to_accept_memory_one(zone);
diff --git a/scripts/gdb/linux/mm.py b/scripts/gdb/linux/mm.py
index dffadccbb01d..28d33624c38b 100644
--- a/scripts/gdb/linux/mm.py
+++ b/scripts/gdb/linux/mm.py
@@ -56,7 +56,7 @@ class x86_page_ops():
 
         self.MAX_PHYSMEM_BITS = 46
         self.SECTION_SIZE_BITS = 27
-        self.MAX_ORDER = 10
+        self.MAX_PAGE_ORDER = 10
 
         self.SECTIONS_SHIFT = self.MAX_PHYSMEM_BITS - self.SECTION_SIZE_BITS
         self.NR_MEM_SECTIONS = 1 << self.SECTIONS_SHIFT
@@ -233,11 +233,11 @@ class aarch64_page_ops():
         self.SECTIONS_SHIFT = self.MAX_PHYSMEM_BITS - self.SECTION_SIZE_BITS
 
         if str(constants.LX_CONFIG_ARCH_FORCE_MAX_ORDER).isdigit():
-            self.MAX_ORDER = constants.LX_CONFIG_ARCH_FORCE_MAX_ORDER
+            self.MAX_PAGE_ORDER = constants.LX_CONFIG_ARCH_FORCE_MAX_ORDER
         else:
-            self.MAX_ORDER = 10
+            self.MAX_PAGE_ORDER = 10
 
-        self.MAX_ORDER_NR_PAGES = 1 << (self.MAX_ORDER)
+        self.MAX_ORDER_NR_PAGES = 1 << (self.MAX_PAGE_ORDER)
         self.PFN_SECTION_SHIFT = self.SECTION_SIZE_BITS - self.PAGE_SHIFT
         self.NR_MEM_SECTIONS = 1 << self.SECTIONS_SHIFT
         self.PAGES_PER_SECTION = 1 << self.PFN_SECTION_SHIFT
-- 
2.33.0

Re: [PATCH] mm: drop stale MAX_ORDER references

[Zi Yan]

On Tue Aug 18, 2026 at 8:24 AM EDT, Qi Xi wrote:
> The treewide rename in commit 5e0a760b4441 ("mm, treewide: rename
> MAX_ORDER to MAX_PAGE_ORDER") left a few spots still using the old
> name:
>
>   - two comments in include/net/mana/mana.h and mm/page_alloc.c;
>   - the gdb helper scripts/gdb/linux/mm.py, where self.MAX_ORDER is
>     a local mirror of the kernel's MAX_ORDER define.
>
> Rename the leftover instances to MAX_PAGE_ORDER so the tree is
> consistent.
>
> No functional changes.
>
> Signed-off-by: Qi Xi <xiqi2@huawei.com>
> ---
>  include/net/mana/mana.h | 4 ++--
>  mm/page_alloc.c         | 2 +-
>  scripts/gdb/linux/mm.py | 8 ++++----
>  3 files changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> index 04acb6791dbd..1d5bed71d6a7 100644
> --- a/include/net/mana/mana.h
> +++ b/include/net/mana/mana.h
> @@ -40,8 +40,8 @@ enum TRI_STATE {
>  #define COMP_ENTRY_SIZE 64
>  
>  /* This Max value for RX buffers is derived from __alloc_page()'s max page
> - * allocation calculation. It allows maximum 2^(MAX_ORDER -1) pages. RX buffer
> - * size beyond this value gets rejected by __alloc_page() call.
> + * allocation calculation. It allows maximum 2^(MAX_PAGE_ORDER -1) pages. RX

It should be 2^MAX_PAGE_ORDER here, since MAX_PAGE_ORDER is inclusive
now.

> + * buffer size beyond this value gets rejected by __alloc_page() call.
>   */
>  #define MAX_RX_BUFFERS_PER_QUEUE 8192
>  #define DEF_RX_BUFFERS_PER_QUEUE 1024
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index ee902a468c2f..42b5b41432c5 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -7803,7 +7803,7 @@ static bool cond_accept_memory(struct zone *zone, unsigned int order,
>  	/*
>  	 * Watermarks have not been initialized yet.
>  	 *
> -	 * Accepting one MAX_ORDER page to ensure progress.
> +	 * Accepting one MAX_PAGE_ORDER page to ensure progress.
>  	 */
>  	if (!wmark)
>  		return try_to_accept_memory_one(zone);
> diff --git a/scripts/gdb/linux/mm.py b/scripts/gdb/linux/mm.py
> index dffadccbb01d..28d33624c38b 100644
> --- a/scripts/gdb/linux/mm.py
> +++ b/scripts/gdb/linux/mm.py
> @@ -56,7 +56,7 @@ class x86_page_ops():
>  
>          self.MAX_PHYSMEM_BITS = 46
>          self.SECTION_SIZE_BITS = 27
> -        self.MAX_ORDER = 10
> +        self.MAX_PAGE_ORDER = 10
>  
>          self.SECTIONS_SHIFT = self.MAX_PHYSMEM_BITS - self.SECTION_SIZE_BITS
>          self.NR_MEM_SECTIONS = 1 << self.SECTIONS_SHIFT
> @@ -233,11 +233,11 @@ class aarch64_page_ops():
>          self.SECTIONS_SHIFT = self.MAX_PHYSMEM_BITS - self.SECTION_SIZE_BITS
>  
>          if str(constants.LX_CONFIG_ARCH_FORCE_MAX_ORDER).isdigit():
> -            self.MAX_ORDER = constants.LX_CONFIG_ARCH_FORCE_MAX_ORDER
> +            self.MAX_PAGE_ORDER = constants.LX_CONFIG_ARCH_FORCE_MAX_ORDER
>          else:
> -            self.MAX_ORDER = 10
> +            self.MAX_PAGE_ORDER = 10
>  
> -        self.MAX_ORDER_NR_PAGES = 1 << (self.MAX_ORDER)
> +        self.MAX_ORDER_NR_PAGES = 1 << (self.MAX_PAGE_ORDER)
>          self.PFN_SECTION_SHIFT = self.SECTION_SIZE_BITS - self.PAGE_SHIFT
>          self.NR_MEM_SECTIONS = 1 << self.SECTIONS_SHIFT
>          self.PAGES_PER_SECTION = 1 << self.PFN_SECTION_SHIFT

The rest looks good to me.

With the comment in mana.h fixed, feel free to add

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


-- 
Best Regards,
Yan, Zi

Re: [PATCH] mm: drop stale MAX_ORDER references

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Blind text replacement in a comment causes the documented mathematical formula for maximum page allocation size to be halved.
--

--- Patch [1]: [PATCH] mm: drop stale MAX_ORDER references ---
commit dbb29499acbd8adcdd7b1a04573e7581c99841f4
Author: Qi Xi <xiqi2@huawei.com>

mm: drop stale MAX_ORDER references

This commit renames leftover instances of MAX_ORDER to MAX_PAGE_ORDER across
the tree, including comments and gdb helper scripts, to maintain consistency
after the treewide rename.

> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> index 04acb6791dbd9..1d5bed71d6a7e 100644
> --- a/include/net/mana/mana.h
> +++ b/include/net/mana/mana.h
> @@ -40,8 +40,8 @@ enum TRI_STATE {
>  #define COMP_ENTRY_SIZE 64
>  
>  /* This Max value for RX buffers is derived from __alloc_page()'s max page
> - * allocation calculation. It allows maximum 2^(MAX_ORDER -1) pages. RX buffer
> - * size beyond this value gets rejected by __alloc_page() call.
> + * allocation calculation. It allows maximum 2^(MAX_PAGE_ORDER -1) pages. RX
> + * buffer size beyond this value gets rejected by __alloc_page() call.

[Severity: Low]
Does this direct text replacement inadvertently halve the documented
maximum page allocation size?

The old formula used 2^(MAX_ORDER - 1). Because MAX_PAGE_ORDER is an
inclusive limit that is one less than the old exclusive MAX_ORDER, evaluating
2^(MAX_PAGE_ORDER - 1) results in half the actual maximum pages.

Should this formula be updated to just 2^MAX_PAGE_ORDER instead?

>   */
>  #define MAX_RX_BUFFERS_PER_QUEUE 8192
>  #define DEF_RX_BUFFERS_PER_QUEUE 1024

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260818122408.4182417-1-xiqi2@huawei.com?part=1

Re: [PATCH] mm: drop stale MAX_ORDER references

[Kiryl Shutsemau]

On Tue, Aug 18, 2026 at 08:24:08PM +0800, Qi Xi wrote:
> The treewide rename in commit 5e0a760b4441 ("mm, treewide: rename
> MAX_ORDER to MAX_PAGE_ORDER") left a few spots still using the old
> name:
> 
>   - two comments in include/net/mana/mana.h and mm/page_alloc.c;
>   - the gdb helper scripts/gdb/linux/mm.py, where self.MAX_ORDER is
>     a local mirror of the kernel's MAX_ORDER define.
> 
> Rename the leftover instances to MAX_PAGE_ORDER so the tree is
> consistent.
> 
> No functional changes.
> 
> Signed-off-by: Qi Xi <xiqi2@huawei.com>
> ---
>  include/net/mana/mana.h | 4 ++--
>  mm/page_alloc.c         | 2 +-
>  scripts/gdb/linux/mm.py | 8 ++++----
>  3 files changed, 7 insertions(+), 7 deletions(-)
> 
> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> index 04acb6791dbd..1d5bed71d6a7 100644
> --- a/include/net/mana/mana.h
> +++ b/include/net/mana/mana.h
> @@ -40,8 +40,8 @@ enum TRI_STATE {
>  #define COMP_ENTRY_SIZE 64
>  
>  /* This Max value for RX buffers is derived from __alloc_page()'s max page
> - * allocation calculation. It allows maximum 2^(MAX_ORDER -1) pages. RX buffer
> - * size beyond this value gets rejected by __alloc_page() call.
> + * allocation calculation. It allows maximum 2^(MAX_PAGE_ORDER -1) pages. RX

Should it reference MAX_ORDER_NR_PAGES instead?

> + * buffer size beyond this value gets rejected by __alloc_page() call.
>   */
>  #define MAX_RX_BUFFERS_PER_QUEUE 8192
>  #define DEF_RX_BUFFERS_PER_QUEUE 1024

-- 
  Kiryl Shutsemau / Kirill A. Shutemov

Re: [PATCH] mm: drop stale MAX_ORDER references

[Zi Yan]

On 20 Aug 2026, at 9:27, Kiryl Shutsemau wrote:

> On Tue, Aug 18, 2026 at 08:24:08PM +0800, Qi Xi wrote:
>> The treewide rename in commit 5e0a760b4441 ("mm, treewide: rename
>> MAX_ORDER to MAX_PAGE_ORDER") left a few spots still using the old
>> name:
>>
>>   - two comments in include/net/mana/mana.h and mm/page_alloc.c;
>>   - the gdb helper scripts/gdb/linux/mm.py, where self.MAX_ORDER is
>>     a local mirror of the kernel's MAX_ORDER define.
>>
>> Rename the leftover instances to MAX_PAGE_ORDER so the tree is
>> consistent.
>>
>> No functional changes.
>>
>> Signed-off-by: Qi Xi <xiqi2@huawei.com>
>> ---
>>  include/net/mana/mana.h | 4 ++--
>>  mm/page_alloc.c         | 2 +-
>>  scripts/gdb/linux/mm.py | 8 ++++----
>>  3 files changed, 7 insertions(+), 7 deletions(-)
>>
>> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
>> index 04acb6791dbd..1d5bed71d6a7 100644
>> --- a/include/net/mana/mana.h
>> +++ b/include/net/mana/mana.h
>> @@ -40,8 +40,8 @@ enum TRI_STATE {
>>  #define COMP_ENTRY_SIZE 64
>>
>>  /* This Max value for RX buffers is derived from __alloc_page()'s max page
>> - * allocation calculation. It allows maximum 2^(MAX_ORDER -1) pages. RX buffer
>> - * size beyond this value gets rejected by __alloc_page() call.
>> + * allocation calculation. It allows maximum 2^(MAX_PAGE_ORDER -1) pages. RX
>
> Should it reference MAX_ORDER_NR_PAGES instead?

It is fixed in v2[1].

[1] https://lore.kernel.org/all/20260819082052.3338603-1-xiqi2@huawei.com/

>
>> + * buffer size beyond this value gets rejected by __alloc_page() call.
>>   */
>>  #define MAX_RX_BUFFERS_PER_QUEUE 8192
>>  #define DEF_RX_BUFFERS_PER_QUEUE 1024
>
> -- 
>   Kiryl Shutsemau / Kirill A. Shutemov


Best Regards,
Yan, Zi

Re: [PATCH] mm: drop stale MAX_ORDER references

[Kiryl Shutsemau]

On Thu, Aug 20, 2026 at 09:52:03AM -0400, Zi Yan wrote:
> On 20 Aug 2026, at 9:27, Kiryl Shutsemau wrote:
> 
> > On Tue, Aug 18, 2026 at 08:24:08PM +0800, Qi Xi wrote:
> >> The treewide rename in commit 5e0a760b4441 ("mm, treewide: rename
> >> MAX_ORDER to MAX_PAGE_ORDER") left a few spots still using the old
> >> name:
> >>
> >>   - two comments in include/net/mana/mana.h and mm/page_alloc.c;
> >>   - the gdb helper scripts/gdb/linux/mm.py, where self.MAX_ORDER is
> >>     a local mirror of the kernel's MAX_ORDER define.
> >>
> >> Rename the leftover instances to MAX_PAGE_ORDER so the tree is
> >> consistent.
> >>
> >> No functional changes.
> >>
> >> Signed-off-by: Qi Xi <xiqi2@huawei.com>
> >> ---
> >>  include/net/mana/mana.h | 4 ++--
> >>  mm/page_alloc.c         | 2 +-
> >>  scripts/gdb/linux/mm.py | 8 ++++----
> >>  3 files changed, 7 insertions(+), 7 deletions(-)
> >>
> >> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> >> index 04acb6791dbd..1d5bed71d6a7 100644
> >> --- a/include/net/mana/mana.h
> >> +++ b/include/net/mana/mana.h
> >> @@ -40,8 +40,8 @@ enum TRI_STATE {
> >>  #define COMP_ENTRY_SIZE 64
> >>
> >>  /* This Max value for RX buffers is derived from __alloc_page()'s max page
> >> - * allocation calculation. It allows maximum 2^(MAX_ORDER -1) pages. RX buffer
> >> - * size beyond this value gets rejected by __alloc_page() call.
> >> + * allocation calculation. It allows maximum 2^(MAX_PAGE_ORDER -1) pages. RX
> >
> > Should it reference MAX_ORDER_NR_PAGES instead?
> 
> It is fixed in v2[1].

It is not :)

My point is that 2^MAX_PAGE_ORDER is MAX_ORDER_NR_PAGES.

But v2 is good enough.

-- 
  Kiryl Shutsemau / Kirill A. Shutemov