linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] mm/mmap: leave adjust_next as virtual address instead of page frame number
@ 2020-08-28  8:10 Wei Yang
  2020-09-08 15:31 ` Vlastimil Babka
  0 siblings, 1 reply; 3+ messages in thread
From: Wei Yang @ 2020-08-28  8:10 UTC (permalink / raw)
  To: akpm; +Cc: linux-mm, linux-kernel, Wei Yang

Instead of convert adjust_next between virtual address and page frame
number, let's just store the virtual address into adjust_next.

Also, this patch fixes one typo in the comment of
vma_adjust_trans_huge().

Signed-off-by: Wei Yang <richard.weiyang@linux.alibaba.com>
---
 mm/huge_memory.c | 4 ++--
 mm/mmap.c        | 8 ++++----
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 78c84bee7e29..2c633ba14440 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -2300,13 +2300,13 @@ void vma_adjust_trans_huge(struct vm_area_struct *vma,
 
 	/*
 	 * If we're also updating the vma->vm_next->vm_start, if the new
-	 * vm_next->vm_start isn't page aligned and it could previously
+	 * vm_next->vm_start isn't hpage aligned and it could previously
 	 * contain an hugepage: check if we need to split an huge pmd.
 	 */
 	if (adjust_next > 0) {
 		struct vm_area_struct *next = vma->vm_next;
 		unsigned long nstart = next->vm_start;
-		nstart += adjust_next << PAGE_SHIFT;
+		nstart += adjust_next;
 		if (nstart & ~HPAGE_PMD_MASK &&
 		    (nstart & HPAGE_PMD_MASK) >= next->vm_start &&
 		    (nstart & HPAGE_PMD_MASK) + HPAGE_PMD_SIZE <= next->vm_end)
diff --git a/mm/mmap.c b/mm/mmap.c
index 90b1298d4222..e4c9bbfd4103 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -758,7 +758,7 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
 			 * vma expands, overlapping part of the next:
 			 * mprotect case 5 shifting the boundary up.
 			 */
-			adjust_next = (end - next->vm_start) >> PAGE_SHIFT;
+			adjust_next = (end - next->vm_start);
 			exporter = next;
 			importer = vma;
 			VM_WARN_ON(expand != importer);
@@ -768,7 +768,7 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
 			 * split_vma inserting another: so it must be
 			 * mprotect case 4 shifting the boundary down.
 			 */
-			adjust_next = -((vma->vm_end - end) >> PAGE_SHIFT);
+			adjust_next = -(vma->vm_end - end);
 			exporter = vma;
 			importer = next;
 			VM_WARN_ON(expand != importer);
@@ -840,8 +840,8 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
 	}
 	vma->vm_pgoff = pgoff;
 	if (adjust_next) {
-		next->vm_start += adjust_next << PAGE_SHIFT;
-		next->vm_pgoff += adjust_next;
+		next->vm_start += adjust_next;
+		next->vm_pgoff += adjust_next >> PAGE_SHIFT;
 	}
 
 	if (root) {
-- 
2.20.1 (Apple Git-117)



^ permalink raw reply related	[flat|nested] 3+ messages in thread
* Re: [PATCH] mm/mmap: leave adjust_next as virtual address instead of page frame number
  2020-08-28  8:10 [PATCH] mm/mmap: leave adjust_next as virtual address instead of page frame number Wei Yang
@ 2020-09-08 15:31 ` Vlastimil Babka
  2020-09-09  2:28   ` Wei Yang
  0 siblings, 1 reply; 3+ messages in thread
From: Vlastimil Babka @ 2020-09-08 15:31 UTC (permalink / raw)
  To: Wei Yang, akpm; +Cc: linux-mm, linux-kernel

On 8/28/20 10:10 AM, Wei Yang wrote:
> Instead of convert adjust_next between virtual address and page frame
> number, let's just store the virtual address into adjust_next.

IMHO more precisely/less confusing it's "bytes" and "pages" instead of "virtual
address" (which is absolute address, but this variable holds a difference) and
"page frame number" (which is related to absolute physical address, but what we
have is difference in pages in virtual address space).

> Also, this patch fixes one typo in the comment of
> vma_adjust_trans_huge().
> 
> Signed-off-by: Wei Yang <richard.weiyang@linux.alibaba.com>

Other than that, seems like it leads to less shifting, so
Acked-by: Vlastimil Babka <vbabka@suse.cz>

> ---
>  mm/huge_memory.c | 4 ++--
>  mm/mmap.c        | 8 ++++----
>  2 files changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 78c84bee7e29..2c633ba14440 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -2300,13 +2300,13 @@ void vma_adjust_trans_huge(struct vm_area_struct *vma,
>  
>  	/*
>  	 * If we're also updating the vma->vm_next->vm_start, if the new
> -	 * vm_next->vm_start isn't page aligned and it could previously
> +	 * vm_next->vm_start isn't hpage aligned and it could previously
>  	 * contain an hugepage: check if we need to split an huge pmd.
>  	 */
>  	if (adjust_next > 0) {
>  		struct vm_area_struct *next = vma->vm_next;
>  		unsigned long nstart = next->vm_start;
> -		nstart += adjust_next << PAGE_SHIFT;
> +		nstart += adjust_next;
>  		if (nstart & ~HPAGE_PMD_MASK &&
>  		    (nstart & HPAGE_PMD_MASK) >= next->vm_start &&
>  		    (nstart & HPAGE_PMD_MASK) + HPAGE_PMD_SIZE <= next->vm_end)
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 90b1298d4222..e4c9bbfd4103 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -758,7 +758,7 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
>  			 * vma expands, overlapping part of the next:
>  			 * mprotect case 5 shifting the boundary up.
>  			 */
> -			adjust_next = (end - next->vm_start) >> PAGE_SHIFT;
> +			adjust_next = (end - next->vm_start);
>  			exporter = next;
>  			importer = vma;
>  			VM_WARN_ON(expand != importer);
> @@ -768,7 +768,7 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
>  			 * split_vma inserting another: so it must be
>  			 * mprotect case 4 shifting the boundary down.
>  			 */
> -			adjust_next = -((vma->vm_end - end) >> PAGE_SHIFT);
> +			adjust_next = -(vma->vm_end - end);
>  			exporter = vma;
>  			importer = next;
>  			VM_WARN_ON(expand != importer);
> @@ -840,8 +840,8 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
>  	}
>  	vma->vm_pgoff = pgoff;
>  	if (adjust_next) {
> -		next->vm_start += adjust_next << PAGE_SHIFT;
> -		next->vm_pgoff += adjust_next;
> +		next->vm_start += adjust_next;
> +		next->vm_pgoff += adjust_next >> PAGE_SHIFT;
>  	}
>  
>  	if (root) {
> 



^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: [PATCH] mm/mmap: leave adjust_next as virtual address instead of page frame number
  2020-09-08 15:31 ` Vlastimil Babka
@ 2020-09-09  2:28   ` Wei Yang
  0 siblings, 0 replies; 3+ messages in thread
From: Wei Yang @ 2020-09-09  2:28 UTC (permalink / raw)
  To: Vlastimil Babka; +Cc: Wei Yang, akpm, linux-mm, linux-kernel

On Tue, Sep 08, 2020 at 05:31:22PM +0200, Vlastimil Babka wrote:
>On 8/28/20 10:10 AM, Wei Yang wrote:
>> Instead of convert adjust_next between virtual address and page frame
>> number, let's just store the virtual address into adjust_next.
>
>IMHO more precisely/less confusing it's "bytes" and "pages" instead of "virtual
>address" (which is absolute address, but this variable holds a difference) and
>"page frame number" (which is related to absolute physical address, but what we
>have is difference in pages in virtual address space).
>

Thanks for your comment.

To be honest, I am not sure which one is more precise. English is not my
mother tongue. If others think this is better, I am fine to adjust this.

>> Also, this patch fixes one typo in the comment of
>> vma_adjust_trans_huge().
>> 
>> Signed-off-by: Wei Yang <richard.weiyang@linux.alibaba.com>
>
>Other than that, seems like it leads to less shifting, so
>Acked-by: Vlastimil Babka <vbabka@suse.cz>
>
>> ---
>>  mm/huge_memory.c | 4 ++--
>>  mm/mmap.c        | 8 ++++----
>>  2 files changed, 6 insertions(+), 6 deletions(-)
>> 
>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>> index 78c84bee7e29..2c633ba14440 100644
>> --- a/mm/huge_memory.c
>> +++ b/mm/huge_memory.c
>> @@ -2300,13 +2300,13 @@ void vma_adjust_trans_huge(struct vm_area_struct *vma,
>>  
>>  	/*
>>  	 * If we're also updating the vma->vm_next->vm_start, if the new
>> -	 * vm_next->vm_start isn't page aligned and it could previously
>> +	 * vm_next->vm_start isn't hpage aligned and it could previously
>>  	 * contain an hugepage: check if we need to split an huge pmd.
>>  	 */
>>  	if (adjust_next > 0) {
>>  		struct vm_area_struct *next = vma->vm_next;
>>  		unsigned long nstart = next->vm_start;
>> -		nstart += adjust_next << PAGE_SHIFT;
>> +		nstart += adjust_next;
>>  		if (nstart & ~HPAGE_PMD_MASK &&
>>  		    (nstart & HPAGE_PMD_MASK) >= next->vm_start &&
>>  		    (nstart & HPAGE_PMD_MASK) + HPAGE_PMD_SIZE <= next->vm_end)
>> diff --git a/mm/mmap.c b/mm/mmap.c
>> index 90b1298d4222..e4c9bbfd4103 100644
>> --- a/mm/mmap.c
>> +++ b/mm/mmap.c
>> @@ -758,7 +758,7 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
>>  			 * vma expands, overlapping part of the next:
>>  			 * mprotect case 5 shifting the boundary up.
>>  			 */
>> -			adjust_next = (end - next->vm_start) >> PAGE_SHIFT;
>> +			adjust_next = (end - next->vm_start);
>>  			exporter = next;
>>  			importer = vma;
>>  			VM_WARN_ON(expand != importer);
>> @@ -768,7 +768,7 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
>>  			 * split_vma inserting another: so it must be
>>  			 * mprotect case 4 shifting the boundary down.
>>  			 */
>> -			adjust_next = -((vma->vm_end - end) >> PAGE_SHIFT);
>> +			adjust_next = -(vma->vm_end - end);
>>  			exporter = vma;
>>  			importer = next;
>>  			VM_WARN_ON(expand != importer);
>> @@ -840,8 +840,8 @@ int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
>>  	}
>>  	vma->vm_pgoff = pgoff;
>>  	if (adjust_next) {
>> -		next->vm_start += adjust_next << PAGE_SHIFT;
>> -		next->vm_pgoff += adjust_next;
>> +		next->vm_start += adjust_next;
>> +		next->vm_pgoff += adjust_next >> PAGE_SHIFT;
>>  	}
>>  
>>  	if (root) {
>> 

-- 
Wei Yang
Help you, Help me


^ permalink raw reply	[flat|nested] 3+ messages in thread
end of thread, other threads:[~2020-09-09  2:28 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-28  8:10 [PATCH] mm/mmap: leave adjust_next as virtual address instead of page frame number Wei Yang
2020-09-08 15:31 ` Vlastimil Babka
2020-09-09  2:28   ` Wei Yang

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).