On Thursday August 3, paul.clements@xxxxxxxxxxxx wrote: > > > The r1_bio->master_bio may already have had end_io() called and been > freed by the time we bio_clone() it. This results in an invalid bio > being sent down to one (or more) of the raid1 components. In my testing, > the original 4K barrier write ended up being a zero length write when it > was retried. Oh bother... > > Solution: > -------- > Can we just bio_alloc instead of cloning? I think we have all the sector > and offset data in the r1_bio. We could then bio_add_page() the pages > into the new bio, possibly? Or could we just use the original bio that > failed, being careful to clean up fields (I think we used to do this but > Neil changed the behavior because of some problems). Yes we did. I don't remember exactly what the problem was, but there are lots of fields in a bio, and the rules about which ones can be modified by drives are not terribly clear. Let's see... The drive can modify: bi_io_vec->bv_offset end_that_request_first bi_io_vec->bv_len end_that_request_first bi_sector raid0/make_request bi_bdev raid0/make_request bi_idx end_that_request_first bi_size But not (I hope) bi_io_vec->bv_page bi_rw bi_vcnt bi_max_vecs bi_end_io bi_private And these can easily be reset or zeroed bi_next bi_phys_segments bi_hw_segments bi_hw_front_size bi_hw_back_size bi_flags ??? sector, bdev, size are all remembered in r1_bio. That leaves bi_idx and an array od len/offset pairs that we need to preserve. So I guess the first step is to change alloc_behind_pages to return a new 'struct bio_vec' array rather than just a list of pages, and we should keep that array attached to the raid1_bio. Sound reasonable? NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
