On Wed, Nov 27, 2024 at 01:01:31PM -0700, Jens Axboe wrote: > On 11/27/24 12:43 PM, Jann Horn wrote: > > On Wed, Nov 27, 2024 at 7:09?PM Jens Axboe <axboe@xxxxxxxxx> wrote: > >> On 11/27/24 9:57 AM, Jann Horn wrote: > >>> Hi! > >>> > >>> In fs/bcachefs/fs-io-direct.c, "struct dio_write" contains a pointer > >>> to an mm_struct. This pointer is grabbed in bch2_direct_write() > >>> (without any kind of refcount increment), and used in > >>> bch2_dio_write_continue() for kthread_use_mm()/kthread_unuse_mm() > >>> which are used to enable userspace memory access from kthread context. > >>> I believe kthread_use_mm()/kthread_unuse_mm() require that the caller > >>> guarantees that the MM hasn't gone through exit_mmap() yet (normally > >>> by holding an mmget() reference). > >>> > >>> If we reach this codepath via io_uring, do we have a guarantee that > >>> the mm_struct that called bch2_direct_write() is still alive and > >>> hasn't yet gone through exit_mmap() when it is accessed from > >>> bch2_dio_write_continue()? > >>> > >>> I don't know the async direct I/O codepath particularly well, so I > >>> cc'ed the uring maintainers, who probably know this better than me. > >> > >> I _think_ this is fine as-is, even if it does look dubious and bcachefs > >> arguably should grab an mm ref for this just for safety to avoid future > >> problems. The reason is that bcachefs doesn't set FMODE_NOWAIT, which > >> means that on the io_uring side it cannot do non-blocking issue of > >> requests. This is slower as it always punts to an io-wq thread, which > >> shares the same mm. Hence if the request is alive, there's always a > >> thread with the same mm alive as well. > >> > >> Now if FMODE_NOWAIT was set, then the original task could exit. I'd need > >> to dig a bit deeper to verify that would always be safe and there's not > >> a of time today with a few days off in the US looming, so I'll defer > >> that to next week. It certainly would be fine with an mm ref grabbed. > > > > Ah, thanks for looking into it! I missed this implication of not > > setting FMODE_NOWAIT. > > > > Anyway, what you said sounds like it would be cleaner for bcachefs to > > grab its own extra reference, maybe by initially grabbing an mm > > reference with mmgrab() in bch2_direct_write(), and then use > > mmget_not_zero() in bch2_dio_write_continue() to ensure the MM is > > stable. > > Yep I think that would definitely make it more sturdy, and also less > headscratchy in terms of being able to verify it's actually safe. > > > What do other file systems do for this? I think they normally grab > > page references so that they don't need the MM anymore when > > asynchronously fulfilling the request, right? Like in > > iomap_dio_bio_iter(), which uses bio_iov_iter_get_pages() to grab > > references to the pages corresponding to the userspace regions in > > dio->submit.iter? > > Not aware of anything else doing it like this, where it's punted to a > kthread and then the mm used from there. The upfront page > getting/mapping is the common approach, like you described. Which does > seem like a much better choice, rather than needing to rely on the mm in > a kworker. More common, but not necessarily better: the "pin everything up front" approach had the disadvantage that - well, you pinned everything up front: if userspace requests a single multi-gigabyte IO, that is likely not what you want. The old dio code didn't pin everything up front for this reason (but IIRC wasn't fully asynchronous, and also had a silly baked in 64 page limit).
