[PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET

[PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET

[Ruihan Li]

Currently, memmap_init initializes pfn_hole with 0 instead of
ARCH_PFN_OFFSET. Then init_unavailable_range will start iterating each
page from the page at address zero to the first available page, but it
won't do anything for pages below ARCH_PFN_OFFSET because pfn_valid
won't pass.

If ARCH_PFN_OFFSET is very large (e.g., something like 2^64-2GiB if the
kernel is used as a library and loaded at a very high address), the
pointless iteration for pages below ARCH_PFN_OFFSET will take a very
long time, and the kernel will look stuck at boot time.

This commit sets the initial value of pfn_hole to ARCH_PFN_OFFSET, which
avoids the problematic and useless iteration mentioned above.

Fixes: 907ec5fca3dc ("mm: zero remaining unavailable struct pages")
Signed-off-by: Ruihan Li <lrh2000@pku.edu.cn>
---
 mm/mm_init.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/mm_init.c b/mm/mm_init.c
index 84f14fa12..b3ae9f797 100644
--- a/mm/mm_init.c
+++ b/mm/mm_init.c
@@ -966,7 +966,7 @@ static void __init memmap_init_zone_range(struct zone *zone,
 static void __init memmap_init(void)
 {
 	unsigned long start_pfn, end_pfn;
-	unsigned long hole_pfn = 0;
+	unsigned long hole_pfn = ARCH_PFN_OFFSET;
 	int i, j, zone_id = 0, nid;
 
 	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
-- 
2.49.0

Re: [PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET

[Mike Rapoport]

On Sat, Apr 19, 2025 at 12:27:27AM +0800, Ruihan Li wrote:
> Currently, memmap_init initializes pfn_hole with 0 instead of
> ARCH_PFN_OFFSET. Then init_unavailable_range will start iterating each
> page from the page at address zero to the first available page, but it
> won't do anything for pages below ARCH_PFN_OFFSET because pfn_valid
> won't pass.
> 
> If ARCH_PFN_OFFSET is very large (e.g., something like 2^64-2GiB if the
> kernel is used as a library and loaded at a very high address), the
> pointless iteration for pages below ARCH_PFN_OFFSET will take a very
> long time, and the kernel will look stuck at boot time.
> 
> This commit sets the initial value of pfn_hole to ARCH_PFN_OFFSET, which
> avoids the problematic and useless iteration mentioned above.
> 
> Fixes: 907ec5fca3dc ("mm: zero remaining unavailable struct pages")

I don't think it should be backported, this does not fix a bug.

> Signed-off-by: Ruihan Li <lrh2000@pku.edu.cn>
> ---
>  mm/mm_init.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/mm_init.c b/mm/mm_init.c
> index 84f14fa12..b3ae9f797 100644
> --- a/mm/mm_init.c
> +++ b/mm/mm_init.c
> @@ -966,7 +966,7 @@ static void __init memmap_init_zone_range(struct zone *zone,
>  static void __init memmap_init(void)
>  {
>  	unsigned long start_pfn, end_pfn;
> -	unsigned long hole_pfn = 0;
> +	unsigned long hole_pfn = ARCH_PFN_OFFSET;

ARCH_PFN_OFFSET is defined only for FLATMEM, this won't even build for
SPARSEMEM.

>  	int i, j, zone_id = 0, nid;
>  
>  	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
> -- 
> 2.49.0
> 
> 

-- 
Sincerely yours,
Mike.

Re: [PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET

[kernel test robot]

Hi Ruihan,

kernel test robot noticed the following build errors:

[auto build test ERROR on akpm-mm/mm-everything]

url:    https://github.com/intel-lab-lkp/linux/commits/Ruihan-Li/mm-mm_init-Don-t-iterate-pages-below-ARCH_PFN_OFFSET/20250419-002910
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/20250418162727.1535335-1-lrh2000%40pku.edu.cn
patch subject: [PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET
config: s390-randconfig-002-20250419 (https://download.01.org/0day-ci/archive/20250419/202504191858.5WVRkO7O-lkp@intel.com/config)
compiler: s390-linux-gcc (GCC) 7.5.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250419/202504191858.5WVRkO7O-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202504191858.5WVRkO7O-lkp@intel.com/

All errors (new ones prefixed by >>):

   mm/mm_init.c: In function 'memmap_init':
>> mm/mm_init.c:975:27: error: 'ARCH_PFN_OFFSET' undeclared (first use in this function); did you mean 'PREEMPT_OFFSET'?
     unsigned long hole_pfn = ARCH_PFN_OFFSET;
                              ^~~~~~~~~~~~~~~
                              PREEMPT_OFFSET
   mm/mm_init.c:975:27: note: each undeclared identifier is reported only once for each function it appears in


vim +975 mm/mm_init.c

   971	
   972	static void __init memmap_init(void)
   973	{
   974		unsigned long start_pfn, end_pfn;
 > 975		unsigned long hole_pfn = ARCH_PFN_OFFSET;
   976		int i, j, zone_id = 0, nid;
   977	
   978		for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
   979			struct pglist_data *node = NODE_DATA(nid);
   980	
   981			for (j = 0; j < MAX_NR_ZONES; j++) {
   982				struct zone *zone = node->node_zones + j;
   983	
   984				if (!populated_zone(zone))
   985					continue;
   986	
   987				memmap_init_zone_range(zone, start_pfn, end_pfn,
   988						       &hole_pfn);
   989				zone_id = j;
   990			}
   991		}
   992	
   993		/*
   994		 * Initialize the memory map for hole in the range [memory_end,
   995		 * section_end] for SPARSEMEM and in the range [memory_end, memmap_end]
   996		 * for FLATMEM.
   997		 * Append the pages in this hole to the highest zone in the last
   998		 * node.
   999		 */
  1000	#ifdef CONFIG_SPARSEMEM
  1001		end_pfn = round_up(end_pfn, PAGES_PER_SECTION);
  1002	#else
  1003		end_pfn = round_up(end_pfn, MAX_ORDER_NR_PAGES);
  1004	#endif
  1005		if (hole_pfn < end_pfn)
  1006			init_unavailable_range(hole_pfn, end_pfn, zone_id, nid);
  1007	}
  1008	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET

[kernel test robot]

Hi Ruihan,

kernel test robot noticed the following build errors:

[auto build test ERROR on akpm-mm/mm-everything]

url:    https://github.com/intel-lab-lkp/linux/commits/Ruihan-Li/mm-mm_init-Don-t-iterate-pages-below-ARCH_PFN_OFFSET/20250419-002910
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/20250418162727.1535335-1-lrh2000%40pku.edu.cn
patch subject: [PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET
config: s390-randconfig-001-20250419 (https://download.01.org/0day-ci/archive/20250419/202504191834.zUYILN1L-lkp@intel.com/config)
compiler: clang version 21.0.0git (https://github.com/llvm/llvm-project f819f46284f2a79790038e1f6649172789734ae8)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250419/202504191834.zUYILN1L-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202504191834.zUYILN1L-lkp@intel.com/

All errors (new ones prefixed by >>):

>> mm/mm_init.c:975:27: error: use of undeclared identifier 'ARCH_PFN_OFFSET'
     975 |         unsigned long hole_pfn = ARCH_PFN_OFFSET;
         |                                  ^
   1 error generated.


vim +/ARCH_PFN_OFFSET +975 mm/mm_init.c

   971	
   972	static void __init memmap_init(void)
   973	{
   974		unsigned long start_pfn, end_pfn;
 > 975		unsigned long hole_pfn = ARCH_PFN_OFFSET;
   976		int i, j, zone_id = 0, nid;
   977	
   978		for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
   979			struct pglist_data *node = NODE_DATA(nid);
   980	
   981			for (j = 0; j < MAX_NR_ZONES; j++) {
   982				struct zone *zone = node->node_zones + j;
   983	
   984				if (!populated_zone(zone))
   985					continue;
   986	
   987				memmap_init_zone_range(zone, start_pfn, end_pfn,
   988						       &hole_pfn);
   989				zone_id = j;
   990			}
   991		}
   992	
   993		/*
   994		 * Initialize the memory map for hole in the range [memory_end,
   995		 * section_end] for SPARSEMEM and in the range [memory_end, memmap_end]
   996		 * for FLATMEM.
   997		 * Append the pages in this hole to the highest zone in the last
   998		 * node.
   999		 */
  1000	#ifdef CONFIG_SPARSEMEM
  1001		end_pfn = round_up(end_pfn, PAGES_PER_SECTION);
  1002	#else
  1003		end_pfn = round_up(end_pfn, MAX_ORDER_NR_PAGES);
  1004	#endif
  1005		if (hole_pfn < end_pfn)
  1006			init_unavailable_range(hole_pfn, end_pfn, zone_id, nid);
  1007	}
  1008	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH] mm/mm_init: Don't iterate pages below ARCH_PFN_OFFSET

[Ruihan Li]

Hi Mike,

Thanks a lot for your review!

On Sat, Apr 19, 2025 at 10:16:34AM +0300, Mike Rapoport wrote:
> On Sat, Apr 19, 2025 at 12:27:27AM +0800, Ruihan Li wrote:
> > Currently, memmap_init initializes pfn_hole with 0 instead of
> > ARCH_PFN_OFFSET. Then init_unavailable_range will start iterating each
> > page from the page at address zero to the first available page, but it
> > won't do anything for pages below ARCH_PFN_OFFSET because pfn_valid
> > won't pass.
> > 
> > If ARCH_PFN_OFFSET is very large (e.g., something like 2^64-2GiB if the
> > kernel is used as a library and loaded at a very high address), the
> > pointless iteration for pages below ARCH_PFN_OFFSET will take a very
> > long time, and the kernel will look stuck at boot time.
> > 
> > This commit sets the initial value of pfn_hole to ARCH_PFN_OFFSET, which
> > avoids the problematic and useless iteration mentioned above.
> > 
> > Fixes: 907ec5fca3dc ("mm: zero remaining unavailable struct pages")
> 
> I don't think it should be backported, this does not fix a bug.

I thought the Fixes tag only indicated that the problem existed since
the specific commit, not the need for a backport (since the backport
should only happen when someone sends or CC's the patch to the
linux-stable mailing list)?

But anyway, I'm going to remove the Fixes tag when I send v2.

> 
> > Signed-off-by: Ruihan Li <lrh2000@pku.edu.cn>
> > ---
> >  mm/mm_init.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/mm/mm_init.c b/mm/mm_init.c
> > index 84f14fa12..b3ae9f797 100644
> > --- a/mm/mm_init.c
> > +++ b/mm/mm_init.c
> > @@ -966,7 +966,7 @@ static void __init memmap_init_zone_range(struct zone *zone,
> >  static void __init memmap_init(void)
> >  {
> >  	unsigned long start_pfn, end_pfn;
> > -	unsigned long hole_pfn = 0;
> > +	unsigned long hole_pfn = ARCH_PFN_OFFSET;
> 
> ARCH_PFN_OFFSET is defined only for FLATMEM, this won't even build for
> SPARSEMEM.

Sorry for my ignorance. I'm not familiar with the code and I'm only
testing the FLATMEM configuration. I'll send a v2 patch to fix this.
Thanks for pointing out my stupid mistake.

> 
> >  	int i, j, zone_id = 0, nid;
> >  
> >  	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
> > -- 
> > 2.49.0
> > 
> > 
> 
> -- 
> Sincerely yours,
> Mike.

Thanks,
Ruihan Li