On Mon, Sep 01 2008, Thomas Petazzoni wrote: > Hi, > > The comment at the top of end_request() says: > > * This is a remnant of how older block drivers handled IO completions. > * Modern drivers typically end IO on the full request in one go, unless > * they have a residual value to account for. For that case this function > * isn't really useful, unless the residual just happens to be the > * full current segment. In other words, don't use this function in new > * code. Either use end_request_completely(), or the > * end_that_request_chunk() (along with end_that_request_last()) for > * partial completions. > > But unless I'm blind, neither end_request_completely() nor > end_request_chunk() exist. > > There are several request completion notification functions > (end_request, end_queued_request, end_dequeued_request, blk_end_request > and __blk_end_request). Which ones should be used in what case ? > > If end_request() is deprecated, I suspect that the goal is to deprecate > all end_*_request() functions as well, is it correct ? If so, is that > correct that one should use blk_end_request() when the queue lock is > not held, and __blk_end_request() if the queue lock is already held ? end_request() is deprecated, and only that one. You can use end_dequeued_request() or end_queued_request() to completely end a dequeued or queued request. Or you can use blk_end_request() to end nr_bytes of IO on that request. Those are the functions that you should use. And yes, you should use __blk_end_request() if you already have the queue lock grabbed. -- Jens Axboe -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
