From: Hiroyasu OHYAMA
I'm a student of Japan, and I diddle the Linux for joyment.
Could I ask you a question about block I/O.
The question is that "Does The linux block I/O processing follow a first-in first-out order"
Now, I'm making layer under low level filesystem. And I wonna make it work as Cache of data like page cache.
I could understood if I emit block I/O request with submit_bio(), the kernel delay it, and I could notice the result by .bi_end_io() method that I set.
But I have something on my mind. "Does the kernel assure the order of block I/O ?"
Once I issue a block I/O request to common I/O layer, the I/O processing is delayed. And the actual block I/O processing is executed by kernel thread.
So user who commit block I/O never have to mind the sequence of delayed block I/O requests.
But I want to know that "The block I/O processing executed first is executed earlient, or may not".
If a I/O request issued back is issued earlier than before, I have to protect the target block that is issued block I/O request.
Could you please give me the actuall operation of block I/O component, that is "It may happen, or not".
Thank you in advance.
Hi,
I think you should read: Documentation/block/barrier.txt
Without explicitly asking for barriers, block I/O doesn't guarantee
FIFO ordering (or any other ordering), since it would be too
inefficient on some devices.
Corrado
On Wed, Apr 22, 2009 at 4:50 PM, <[email protected]> wrote:
> From: Hiroyasu OHYAMA
>
> I'm a student of Japan, and I diddle the Linux for joyment.
> Could I ask you a question about block I/O.
>
> The question is that "Does The linux block I/O processing follow a first-in first-out order"
>
> Now, I'm making layer under low level filesystem. And I wonna make it work as Cache of data like page cache.
> I could understood if I emit block I/O request with submit_bio(), the kernel delay it, and I could notice the result by .bi_end_io() method that I set.
>
> But I have something on my mind. "Does the kernel assure the order of block I/O ?"
>
> Once I issue a block I/O request to common I/O layer, the I/O processing is delayed. And the actual block I/O processing is executed by kernel thread.
> So user who commit block I/O never have to mind the sequence of delayed block I/O requests.
> But I want to know that "The block I/O processing executed first is executed earlient, or may not".
>
> If a I/O request issued back is issued earlier than before, I have to protect the target block that is issued block I/O request.
>
> Could you please give me the actuall operation of block I/O component, that is "It may happen, or not".
>
> Thank you in advance.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>