Re: [PATCH v4 3/6] xfs: implement write-stream management support

On 8/19/2026 12:54 PM, Christoph Hellwig wrote:
>> +	nr_streams = bdev_max_write_streams(mp->m_ddev_targp->bt_bdev);
> Returning to this.  Given that we are talking about pure software
> write streams, does this make sense?  This limits the use to devices
> that do report hardware write streams despite not using those.  Most
> devices will not report those.
> 

Yeah, I enabled hardware streams first and define software streams as a 
fallback when the block device has none. I can swap the order if that 
reads better.