Skip to contentSkip to conversation

mm: replace is_zero_pfn with is_huge_zero_pmd for thp

PATCH Superseded Yu Zhao 3 participants 5 messages Aug 25, 2019 – Nov 8, 2019

Subsystems (2)
lore
Share & tools
lore

Yu Zhao

author
#0 lore raw

PATCH 1/1 mm: replace is_zero_pfn with is_huge_zero_pmd for thp

For hugely mapped thp, we use is_huge_zero_pmd() to check if it's zero page or not.

We do fill ptes with my_zero_pfn() when we split zero thp pmd, but
this is not what we have in vm_normal_page_pmd().
pmd_trans_huge_lock() makes sure of it.

This is a trivial fix for /proc/pid/numa_maps, and AFAIK nobody complains about it.

---
 mm/memory.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/memory.c +1 -1
index e2bb51b6242e..ea3c74855b23 100644
@@ -654,7 +654,7 @@ struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
 
 	if (pmd_devmap(pmd))
 		return NULL;
-	if (is_zero_pfn(pfn))
+	if (is_huge_zero_pmd(pmd))
 		return NULL;
 	if (unlikely(pfn > highest_memmap_pfn))
 		return NULL;
Signature · 1 line
2.23.0.187.g17f5b7556c-goog
#1 lore raw

Why did you not cc Gerald who wrote the patch? You can't just run get_maintainers.pl and call it good.

Quoted 31 lines · Yu Zhao

For hugely mapped thp, we use is_huge_zero_pmd() to check if it's zero page or not.

We do fill ptes with my_zero_pfn() when we split zero thp pmd, but
this is not what we have in vm_normal_page_pmd().
pmd_trans_huge_lock() makes sure of it.

This is a trivial fix for /proc/pid/numa_maps, and AFAIK nobody complains about it.

---
 mm/memory.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/memory.c +1 -1
index e2bb51b6242e..ea3c74855b23 100644
@@ -654,7 +654,7 @@ struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
 
 	if (pmd_devmap(pmd))
 		return NULL;
-	if (is_zero_pfn(pfn))
+	if (is_huge_zero_pmd(pmd))
 		return NULL;
 	if (unlikely(pfn > highest_memmap_pfn))
 		return NULL;
Signature · 1 line
2.23.0.187.g17f5b7556c-goog
#2 lore raw
Quoted 25 lines · On Mon, 26 Aug 2019 06:18:58 -0700 Matthew Wilcox <[email protected]> wrote:

Why did you not cc Gerald who wrote the patch? You can't just run get_maintainers.pl and call it good.

Quoted 21 lines · Yu Zhao

For hugely mapped thp, we use is_huge_zero_pmd() to check if it's zero page or not.

We do fill ptes with my_zero_pfn() when we split zero thp pmd, but
this is not what we have in vm_normal_page_pmd().
pmd_trans_huge_lock() makes sure of it.

This is a trivial fix for /proc/pid/numa_maps, and AFAIK nobody complains about it.

---
 mm/memory.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/memory.c +0 -0
index e2bb51b6242e..ea3c74855b23 100644
@@ -654,7 +654,7 @@ struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
 
Quoted 10 lines
Quoted 10 lines
mm/memory.c +1 -1
 	if (pmd_devmap(pmd))
 		return NULL;
-	if (is_zero_pfn(pfn))
+	if (is_huge_zero_pmd(pmd))
 		return NULL;
 	if (unlikely(pfn > highest_memmap_pfn))
 		return NULL;
Signature · 1 line
2.23.0.187.g17f5b7556c-goog

Looks good to me. The "_pmd" versions for can_gather_numa_stats() and vm_normal_page() were introduced to avoid using pte_present/dirty() on pmds, which is not affected by this patch.

In fact, for vm_normal_page_pmd() I basically copied most of the code from vm_normal_page(), including the is_zero_pfn(pfn) check, which does look wrong to me now. Using is_huge_zero_pmd() should be correct.

Maybe the description could also mention the symptom of this bug? I would assume that it affects anon/dirty accounting in gather_pte_stats(), for huge mappings, if zero page mappings are not correctly recognized.

Regards,
Gerald

author
#3 lore raw
Quoted 43 lines · Gerald Schaefer
Quoted 25 lines · On Mon, 26 Aug 2019 06:18:58 -0700 Matthew Wilcox <[email protected]> wrote:

Why did you not cc Gerald who wrote the patch? You can't just run get_maintainers.pl and call it good.

Quoted 21 lines · Yu Zhao

For hugely mapped thp, we use is_huge_zero_pmd() to check if it's zero page or not.

We do fill ptes with my_zero_pfn() when we split zero thp pmd, but
this is not what we have in vm_normal_page_pmd().
pmd_trans_huge_lock() makes sure of it.

This is a trivial fix for /proc/pid/numa_maps, and AFAIK nobody complains about it.

---
 mm/memory.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/memory.c +0 -0
index e2bb51b6242e..ea3c74855b23 100644
@@ -654,7 +654,7 @@ struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
 
Quoted 10 lines
Quoted 10 lines
mm/memory.c +1 -1
 	if (pmd_devmap(pmd))
 		return NULL;
-	if (is_zero_pfn(pfn))
+	if (is_huge_zero_pmd(pmd))
 		return NULL;
 	if (unlikely(pfn > highest_memmap_pfn))
 		return NULL;
Signature · 1 line
2.23.0.187.g17f5b7556c-goog

Looks good to me. The "_pmd" versions for can_gather_numa_stats() and vm_normal_page() were introduced to avoid using pte_present/dirty() on pmds, which is not affected by this patch.

Quoted 7 lines

In fact, for vm_normal_page_pmd() I basically copied most of the code from vm_normal_page(), including the is_zero_pfn(pfn) check, which does look wrong to me now. Using is_huge_zero_pmd() should be correct.

Maybe the description could also mention the symptom of this bug? I would assume that it affects anon/dirty accounting in gather_pte_stats(), for huge mappings, if zero page mappings are not correctly recognized.

Hi, sorry for not copying you on the original email. I came across this while I was looking at the code. I'm not aware of any symptom. Thank you.

author
#4 lore raw

PATCH v2 1/1 mm: replace is_zero_pfn with is_huge_zero_pmd for thp

In reply to Yu Zhao

For hugely mapped thp, we use is_huge_zero_pmd() to check if it's zero page or not.

We do fill ptes with my_zero_pfn() when we split zero thp pmd, but this is not what we have in vm_normal_page_pmd() -- pmd_trans_huge_lock() makes sure of it.

This is a trivial fix for /proc/pid/numa_maps, and AFAIK nobody complains about it.

Gerald Schaefer asked:

Quoted 3 lines

Maybe the description could also mention the symptom of this bug? I would assume that it affects anon/dirty accounting in gather_pte_stats(), for huge mappings, if zero page mappings are not correctly recognized.

I came across this while I was looking at the code, so I'm not aware of any symptom.

---
 mm/memory.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/memory.c +1 -1
index b1ca51a079f2..cf209f84ce4a 100644
@@ -654,7 +654,7 @@ struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
 
 	if (pmd_devmap(pmd))
 		return NULL;
-	if (is_zero_pfn(pfn))
+	if (is_huge_zero_pmd(pmd))
 		return NULL;
 	if (unlikely(pfn > highest_memmap_pfn))
 		return NULL;
Signature · 1 line
2.24.0.rc1.363.gb1bccd3e3d-goog