Skip to contentSkip to conversation

um/mm: get max_low_pfn from memblock

PATCH In review Wei Yang 3 participants 4 messages Jun 14, 2024 – Jun 15, 2024

Subsystems (2)
lore
Share & tools
lore

Wei Yang

author
#0 lore raw

PATCH 1/1 um/mm: get max_low_pfn from memblock

Current calculation of max_low_pfn is introduced in commit af84eab20891 ("[PATCH] uml: fix LVM crash"). It is intended to set max_low_pfn to the same value as max_pfn.

But I am not sure why the max_pfn is set to totalram_pages, which represents the number of usable pages in system instead of an absolute page frame number. (The change history stops there.)

While we can get the maximum page frame number from memblock, this looks more reasonable than setting to totalram_pages.

Also this would help changing totalram_pages accounting, since we plan to move the accounting into __free_pages_core(). With this change, totalram_pages may not represent the total usable pages at this point, since some pages would be deferred initialized.

CC Jason Lunz <[email protected]>
CC Jeff Dike <[email protected]>
Cc Paolo 'Blaisorblade' Giarrusso <[email protected]>
Cc Alasdair G Kergon <[email protected]>
Cc Jens Axboe <[email protected]>
CC Andrew Morton <[email protected]>
CC Mike Rapoport (IBM) <[email protected]>
CC David Hildenbrand <[email protected]>

---
A simple UML bootup test looks good.

---
 arch/um/kernel/mem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/um/kernel/mem.c +1 -1
index ca91accd64fc..ca682879e28f 100644
@@ -73,7 +73,7 @@ void __init mem_init(void)
 
 	/* this will put all low memory onto the freelists */
 	memblock_free_all();
-	max_low_pfn = totalram_pages();
+	max_low_pfn = PFN_DOWN(memblock_end_of_DRAM());
 	max_pfn = max_low_pfn;
 	kmalloc_ok = 1;
 }
Signature · 1 line
2.34.1
Acked-by on v1 1/1
#1 lore raw
Quoted 8 lines · Wei Yang

Current calculation of max_low_pfn is introduced in commit af84eab20891 ("[PATCH] uml: fix LVM crash"). It is intended to set max_low_pfn to the same value as max_pfn.

But I am not sure why the max_pfn is set to totalram_pages, which represents the number of usable pages in system instead of an absolute page frame number. (The change history stops there.)

Quoted 7 lines

While we can get the maximum page frame number from memblock, this looks more reasonable than setting to totalram_pages.

Also this would help changing totalram_pages accounting, since we plan to move the accounting into __free_pages_core(). With this change, totalram_pages may not represent the total usable pages at this point, since some pages would be deferred initialized.

Question is if deferred page init is even a thing on UM. But it certainly looks odd+suspiciously wrong to rely on totalram_pages().

Quoted 22 lines
CC Jason Lunz <[email protected]>
CC Jeff Dike <[email protected]>
Cc Paolo 'Blaisorblade' Giarrusso <[email protected]>
Cc Alasdair G Kergon <[email protected]>
Cc Jens Axboe <[email protected]>
CC Andrew Morton <[email protected]>
CC Mike Rapoport (IBM) <[email protected]>
CC David Hildenbrand <[email protected]>

---
A simple UML bootup test looks good.

---
  arch/um/kernel/mem.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
arch/um/kernel/mem.c +0 -0
index ca91accd64fc..ca682879e28f 100644

arch/um/kernel/mem.c void __init mem_init(void) · line 79

Quoted 8 lines
arch/um/kernel/mem.c +1 -1
@@ -73,7 +73,7 @@ void __init mem_init(void)
 
  	/* this will put all low memory onto the freelists */
  	memblock_free_all();
-	max_low_pfn = totalram_pages();
+	max_low_pfn = PFN_DOWN(memblock_end_of_DRAM());
  	max_pfn = max_low_pfn;
  	kmalloc_ok = 1;
  }

Matches what a bunch of other archs do.

Randomly staring at other users:

arch/loongarch/kernel/numa.c:   max_low_pfn = PHYS_PFN(memblock_end_of_DRAM());
arch/loongarch/kernel/setup.c:  max_low_pfn = PFN_PHYS(memblock_end_of_DRAM());

What? the latter cannot possibly be right, no? Looks odd at least. Only applies to CONFIG_OF_EARLY_FLATTREE. CCing loongarch maintainer.

Signature · 3 lines
Cheers,

David / dhildenb
#2 lore raw
Quoted 12 lines · David Hildenbrand
Quoted 11 lines · Wei Yang

Current calculation of max_low_pfn is introduced in commit af84eab20891 ("[PATCH] uml: fix LVM crash"). It is intended to set max_low_pfn to the same value as max_pfn.

But I am not sure why the max_pfn is set to totalram_pages, which represents the number of usable pages in system instead of an absolute page frame number. (The change history stops there.)

While we can get the maximum page frame number from memblock, this looks more reasonable than setting to totalram_pages.

Quoted 6 lines
Quoted 4 lines

Also this would help changing totalram_pages accounting, since we plan to move the accounting into __free_pages_core(). With this change, totalram_pages may not represent the total usable pages at this point, since some pages would be deferred initialized.

Question is if deferred page init is even a thing on UM. But it certainly looks odd+suspiciously wrong to rely on totalram_pages().

As long as there is no HIGHMEM,

max_pfn = max_low_pfn = PFN_DOWN(memblock_end_of_DRAM())

should be ok.

Quoted 16 lines
Quoted 16 lines
CC Jason Lunz <[email protected]>
CC Jeff Dike <[email protected]>
Cc Paolo 'Blaisorblade' Giarrusso <[email protected]>
Cc Alasdair G Kergon <[email protected]>
Cc Jens Axboe <[email protected]>
CC Andrew Morton <[email protected]>
CC Mike Rapoport (IBM) <[email protected]>
CC David Hildenbrand <[email protected]>

---
A simple UML bootup test looks good.

---
  arch/um/kernel/mem.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 9 lines

arch/um/kernel/mem.c void __init mem_init(void) · line 75

Quoted 9 lines
arch/um/kernel/mem.c +1 -1
index ca91accd64fc..ca682879e28f 100644
@@ -73,7 +73,7 @@ void __init mem_init(void)
  	/* this will put all low memory onto the freelists */
  	memblock_free_all();
-	max_low_pfn = totalram_pages();
+	max_low_pfn = PFN_DOWN(memblock_end_of_DRAM());

This assignment seem redundant as there is already

	max_low_pfn = min_low_pfn + (map_size >> PAGE_SHIFT);

in setup_physmem

Quoted 22 lines

arch/um/kernel/mem.c void __init mem_init(void) · line 79

Quoted 3 lines
arch/um/kernel/mem.c +0 -0
  	max_pfn = max_low_pfn;
  	kmalloc_ok = 1;
  }

Matches what a bunch of other archs do.

Randomly staring at other users:

arch/loongarch/kernel/numa.c:   max_low_pfn = PHYS_PFN(memblock_end_of_DRAM());
arch/loongarch/kernel/setup.c:  max_low_pfn = PFN_PHYS(memblock_end_of_DRAM());

What? the latter cannot possibly be right, no? Looks odd at least. Only applies to CONFIG_OF_EARLY_FLATTREE. CCing loongarch maintainer.

Signature · 3 lines
Cheers,

David / dhildenb
Signature · 2 lines
Sincerely yours,
Mike.
author
#3 lore raw
Quoted 54 lines · Mike Rapoport
Quoted 12 lines · David Hildenbrand
Quoted 11 lines · Wei Yang

Current calculation of max_low_pfn is introduced in commit af84eab20891 ("[PATCH] uml: fix LVM crash"). It is intended to set max_low_pfn to the same value as max_pfn.

But I am not sure why the max_pfn is set to totalram_pages, which represents the number of usable pages in system instead of an absolute page frame number. (The change history stops there.)

While we can get the maximum page frame number from memblock, this looks more reasonable than setting to totalram_pages.

Quoted 6 lines
Quoted 4 lines

Also this would help changing totalram_pages accounting, since we plan to move the accounting into __free_pages_core(). With this change, totalram_pages may not represent the total usable pages at this point, since some pages would be deferred initialized.

Question is if deferred page init is even a thing on UM. But it certainly looks odd+suspiciously wrong to rely on totalram_pages().

As long as there is no HIGHMEM,

max_pfn = max_low_pfn = PFN_DOWN(memblock_end_of_DRAM())

should be ok.

Quoted 16 lines
Quoted 16 lines
CC Jason Lunz <[email protected]>
CC Jeff Dike <[email protected]>
Cc Paolo 'Blaisorblade' Giarrusso <[email protected]>
Cc Alasdair G Kergon <[email protected]>
Cc Jens Axboe <[email protected]>
CC Andrew Morton <[email protected]>
CC Mike Rapoport (IBM) <[email protected]>
CC David Hildenbrand <[email protected]>

---
A simple UML bootup test looks good.

---
  arch/um/kernel/mem.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 9 lines

arch/um/kernel/mem.c void __init mem_init(void) · line 75

Quoted 9 lines
arch/um/kernel/mem.c +1 -1
index ca91accd64fc..ca682879e28f 100644
@@ -73,7 +73,7 @@ void __init mem_init(void)
  	/* this will put all low memory onto the freelists */
  	memblock_free_all();
-	max_low_pfn = totalram_pages();
+	max_low_pfn = PFN_DOWN(memblock_end_of_DRAM());

This assignment seem redundant as there is already

Quoted 4 lines
	max_low_pfn = min_low_pfn + (map_size >> PAGE_SHIFT);

in setup_physmem

Looks right, I added some log and shows they are the same.

Thanks

Quoted 26 lines
Quoted 22 lines

arch/um/kernel/mem.c void __init mem_init(void) · line 79

Quoted 3 lines
arch/um/kernel/mem.c +0 -0
  	max_pfn = max_low_pfn;
  	kmalloc_ok = 1;
  }

Matches what a bunch of other archs do.

Randomly staring at other users:

arch/loongarch/kernel/numa.c:   max_low_pfn = PHYS_PFN(memblock_end_of_DRAM());
arch/loongarch/kernel/setup.c:  max_low_pfn = PFN_PHYS(memblock_end_of_DRAM());

What? the latter cannot possibly be right, no? Looks odd at least. Only applies to CONFIG_OF_EARLY_FLATTREE. CCing loongarch maintainer.

Signature · 3 lines
Cheers,

David / dhildenb
Signature · 2 lines
Sincerely yours,
Mike.
Signature · 2 lines
Wei Yang
Help you, Help me