shmem: Are we accounting block right?

shmem: Are we accounting block right?

[Hillf Danton]

Hi all

Currently in mainline we do block account if the flags parameter 
carries VM_NORESERVE. 

But blocks should be accounted if reserved, as shown by the
following diff.

Am I missing anything?

thanks
Hillf

--- a/mm/shmem.c	Fri Aug  5 14:01:59 2016
+++ b/mm/shmem.c	Fri Aug  5 14:36:31 2016
@@ -168,7 +168,7 @@ static inline int shmem_reacct_size(unsi
  */
 static inline int shmem_acct_block(unsigned long flags, long pages)
 {
-	if (!(flags & VM_NORESERVE))
+	if (flags & VM_NORESERVE)
 		return 0;
 
 	return security_vm_enough_memory_mm(current->mm,
@@ -177,7 +177,7 @@ static inline int shmem_acct_block(unsig
 
 static inline void shmem_unacct_blocks(unsigned long flags, long pages)
 {
-	if (flags & VM_NORESERVE)
+	if (!(flags & VM_NORESERVE))
 		vm_unacct_memory(pages * VM_ACCT(PAGE_SIZE));
 }
 
--


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: shmem: Are we accounting block right?

[Hugh Dickins]

On Fri, 5 Aug 2016, Hillf Danton wrote:

> Hi all
> 
> Currently in mainline we do block account if the flags parameter 
> carries VM_NORESERVE. 

Yes.  (VM_NORESERVE being set in tmpfs file flags,
but usually not on SysV SHM and mmaps of /dev/zero.)

> 
> But blocks should be accounted if reserved, as shown by the
> following diff.

Blocks should be accounted one by one as they are instantiated
(tmpfs), unless the total size was all reserved upfront (SHM).

> 
> Am I missing anything?

Apparently, but I'm not sure what.  Maybe the comments
above shmem_acct_size() and shmem_acct_block() will help.

Hugh

> 
> thanks
> Hillf
> 
> --- a/mm/shmem.c	Fri Aug  5 14:01:59 2016
> +++ b/mm/shmem.c	Fri Aug  5 14:36:31 2016
> @@ -168,7 +168,7 @@ static inline int shmem_reacct_size(unsi
>   */
>  static inline int shmem_acct_block(unsigned long flags, long pages)
>  {
> -	if (!(flags & VM_NORESERVE))
> +	if (flags & VM_NORESERVE)
>  		return 0;
>  
>  	return security_vm_enough_memory_mm(current->mm,
> @@ -177,7 +177,7 @@ static inline int shmem_acct_block(unsig
>  
>  static inline void shmem_unacct_blocks(unsigned long flags, long pages)
>  {
> -	if (flags & VM_NORESERVE)
> +	if (!(flags & VM_NORESERVE))
>  		vm_unacct_memory(pages * VM_ACCT(PAGE_SIZE));
>  }
>  
> --

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>