On 7/19/23 11:01?PM, Christoph Hellwig wrote: >> /* can use bio alloc cache */ >> #define IOCB_ALLOC_CACHE (1 << 21) >> +/* >> + * IOCB_DIO_DEFER can be set by the iocb owner, to indicate that the >> + * iocb completion can be passed back to the owner for execution from a safe >> + * context rather than needing to be punted through a workqueue. If this >> + * flag is set, the completion handling may set iocb->dio_complete to a >> + * handler, which the issuer will then call from task context to complete >> + * the processing of the iocb. iocb->private should then also be set to >> + * the argument being passed to this handler. > > Can you add an explanation when it is safe/destirable to do the deferred > completion? As of the last patch we seem to avoid anything that does > I/O or transaction commits, but we'd still allow blocking operations > like mutexes used in the zonefs completion handler. We need to catch > this so future usuers know what to do. Sure can do - generally it's exactly as you write, anything that does extra IO should still be done in a workqueue to avoid stalling this particular IO completion on further IO. > Similarly on the iomap side I think we need clear documentation for > what context ->end_io can be called in now. Honestly that was needed before as well, not really related to this change (or the next two patches) as they don't change the context of how it is called. -- Jens Axboe
