On 9/30/19 2:52 AM, Jackie Liu wrote: > Cool performance improvement, Reviewed-by: Jackie Liu <liuyun01@xxxxxxxxxx> Thanks for the review. > BTW, we always use s->needs_lock to determine if it is in async. Is it > possible to consider replacing it directly with s->in_async? Yeah I think that'd be a good cleanup, would make it clearer without needing comments to say that they are equivalent. -- Jens Axboe
