On 2020/10/06 10:27, Martin K. Petersen wrote: > > Christoph, > >>> max_zone_append_sectors also appears to be gated exclusively by hardware >>> constraints. Is there user-controllable limit in place for append >>> operations? >> >> Zone Append operations can't be split, as they return the first written >> LBA to the caller. If you'd split it you'd now need to return multiple >> start LBA values. > > Yep, this limit would have to be enforced at the top. > > Anyway, not sure how important this is given modern drives. I just know > that for our workloads the soft limit is something we almost always end > up tweaking. We can fix the max_zone_append_sectors limit independently of this btrfs series. Right now, this attribute represents the hardware limit but we can easily rename it to max_hw_zone_append_sectors and define a soft max_zone_append_sectors which would be min(max_sectors, max_hw_zone_append_sectors). That will result in similar soft limit control as for regular writes & max_sectors/max_hw_sectors. -- Damien Le Moal Western Digital Research
