On 08/08/2017 04:00 PM, Bart Van Assche wrote:
> On Tue, 2017-08-08 at 15:13 -0600, Jens Axboe wrote:
>> On 08/08/2017 03:05 PM, Shaohua Li wrote:
>>>> I'm curious why null_blk isn't a good fit? You'd just need to add RAM
>>>> storage to it. That would just be a separate option that should be
>>>> set,
>>>> ram_backing=1 or something like that. That would make it less critical
>>>> than using the RAM disk driver as well, since only people that want a
>>>> "real"
>>>> data backing would enable it.
>>>>
>>>> It's not that I'm extremely opposed to adding a(nother) test block
>>>> driver,
>>>> but we at least need some sort of reasoning behind why, which isn't
>>>> just
>>>> "not a good fit".
>>>
>>> Ah, I thought the 'null' of null_blk means we do nothing for the
>>> disks. Of course we can rename it, which means this point less
>>> meaningful. I think the main reason is the interface. We will
>>> configure the disks with different parameters and do power on/off for
>>> each disks (which is the key we can emulate disk cache and power
>>> loss). The module paramter interface of null_blk doesn't work for the
>>> usage. Of course, these issues can be fixed, for example, we can make
>>> null_blk use the configfs interface. If you really prefer a single
>>> driver for all test purpose, I can move the test_blk functionalities
>>> to null_blk.
>>
>> The idea with null_blk is just that it's a test vehicle. As such, it
>> would actually be useful to have a mode where it does store the data in
>> RAM, since that enables you to do other kinds of testing as well. I'd be
>> fine with augmenting it with configfs for certain things.
>
> Hello Jens,
>
> Would you consider it acceptable to make the mode in which null_blk stores
> data the default? I know several people who got confused by null_blk by
> default not retaining data ...
I don't think we should change the default, since that'll then upset
people that currently use it and all of a sudden see a different
performance profile. It's called null_blk, and the device node is
/dev/nullb0. I think either one of those should reasonably set
expectations for the user that it doesn't really store your data at all.
We could add a module info blurb on it not storing data, I see we don't
have that. The initial commit said:
commit f2298c0403b0dfcaef637eba0c02c4a06d7a25ab
Author: Jens Axboe <axboe@xxxxxxxxx>
Date: Fri Oct 25 11:52:25 2013 +0100
null_blk: multi queue aware block test driver
A driver that simply completes IO it receives, it does no
transfers. Written to fascilitate testing of the blk-mq code.
It supports various module options to use either bio queueing,
rq queueing, or mq mode.
We should just add a MODULE_INFO with a short, similarly worded text.
--
Jens Axboe
