ext4 allocator

ext4 allocator

[Alberich de megres]

Hi!!

I'm studyng ext4 code, and i got a doubt on a situation:

Supposing a 4K page size, and a fs with 8K.. when we are filling page info,:

{ 4K page   }{ 4K page   }{ 4K page   }{ 4K page   }
[       group 0 bitmap      ] [      group 0 buddy        ]

Clearly bitmap and buddy info don't fits on a single page, but e4b
structure saves only one page address.

How mb allocator menages this situation??

Thanks!!!

Re: ext4 allocator

[Andreas Dilger]

On 2010-01-25, at 14:14, Alberich de megres wrote:
> I'm studyng ext4 code, and i got a doubt on a situation:
>
> Supposing a 4K page size, and a fs with 8K.. when we are filling  
> page info,:
>
> { 4K page }{ 4K page }{ 4K page }{ 4K page }
> [   group 0 bitmap   ][   group 0 buddy    ]
>
> Clearly bitmap and buddy info don't fits on a single page, but e4b
> structure saves only one page address.
>
> How mb allocator menages this situation??


ext4, and the Linux VM/block layer in general, do not support disk  
block size larger than the page size.  That has been a restriction for  
a long time already, and there have been various patches/proposals to  
fix that, but nothing that ever made it into the mainline kernel.

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.

Re: ext4 allocator

[Alberich de megres]

Wow, how newbie i am...

Thanks man!!


On Mon, Jan 25, 2010 at 10:37 PM, Andreas Dilger <adilger@sun.com> wrote:
> On 2010-01-25, at 14:14, Alberich de megres wrote:
>>
>> I'm studyng ext4 code, and i got a doubt on a situation:
>>
>> Supposing a 4K page size, and a fs with 8K.. when we are filling page
>> info,:
>>
>> { 4K page }{ 4K page }{ 4K page }{ 4K page }
>> [   group 0 bitmap   ][   group 0 buddy    ]
>>
>> Clearly bitmap and buddy info don't fits on a single page, but e4b
>> structure saves only one page address.
>>
>> How mb allocator menages this situation??
>
>
> ext4, and the Linux VM/block layer in general, do not support disk block
> size larger than the page size.  That has been a restriction for a long time
> already, and there have been various patches/proposals to fix that, but
> nothing that ever made it into the mainline kernel.
>
> Cheers, Andreas
> --
> Andreas Dilger
> Sr. Staff Engineer, Lustre Group
> Sun Microsystems of Canada, Inc.
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html