Re: [PATCH 1/6] null_blk: allow write zeores on non-membacked

[Chaitanya Kulkarni]

>>> +static bool g_write_zeroes;
>>> +module_param_named(write_zeroes, g_write_zeroes, bool, 0444);
>>> +MODULE_PARM_DESC(write_zeroes, "Support write-zeores operations. Default: false");
>>
>> Why not make this a number of sectors representing the maximum size of a
>> write zero command (blk_queue_max_write_zeroes_sectors()) ? That would
>> allow exercising split write zeros BIOs.
>>
> 
> I kept the implementation identical to the g_discard.
> 
> Perhaps it's time to change it so REQ_OP_DISCARD and
> REQ_OP_WRITE_ZEROES will have same implementation.
> 
> I'll add a discard patch to match your suggested write-zeroes
> behavior.
> 
> -ck
> 

REQ_OP_DISCARD and REQ_OP_WRITE_ZEORES needs to be consistent
when it comes to configuration interface.

I did the change you suggested, it is breaking the backward
compatibility of discard and if we do it only for write-zeroes it
will be inconsistent with the current g_discard behavior.

Let's keep the original behavior and not break the backward
compatibility ?

-ck