On Tue, Nov 20, 2012 at 8:59 PM, richard -rw- weinberger
<richard.weinber...@gmail.com> wrote:
> On Tue, Nov 20, 2012 at 11:39 PM, Ezequiel Garcia <elezegar...@gmail.com>
> wrote:
>> Block device emulation on top of ubi volumes with read/write support.
>> Block devices get automatically created for each ubi volume present.
>>
>> Each ubiblock is fairly cheap since it's based on workqueues
>> and not on threads.
>>
>> Read/write access is expected to work fairly well because the
>> request queue at block elevator orders block transfers to be space-effective.
>> In other words, it's expected that reads and writes gets ordered
>> to point to the same LEB.
>>
>> To help this and reduce access to the UBI volume, a 1-LEB size
>> write-back cache has been implemented.
>> Every read and every write, goes through this cache and the write is
>> only done when a request arrives to read or write to a different LEB
>> or when the device is released, when the last file handle is closed.
>
> Did you also benchmark your driver with two caches?
> (One for reading and one for writing.)
> By using two caches you can lower the amount of atomic LEB changes.
>
> Maybe it would be also good to ensure that an cache entry becomes not too old.
>
Yes, I thought of this.
For now, I decided to keep the implementation as simple as possible.
Regards,
Ezequiel
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
