Re: [PATCH] aio: Fix locking in aio_poll()

[Bart Van Assche <bvanassche@acm.org>]

On Tue, 2019-02-05 at 09:12 +0100, Miklos Szeredi wrote:
> Not all variants of the waitqueue interface require irqs to be
> disabled, and fuse has nothing whatsoever to do with irqs, so there's
> no sane reason to disable them.
> 
> Also, AFAICS, the fuse device does not support asynchronous IO.  I
> just don't get what this is about...

Hi Miklos,

Could this be what happens?

aio_poll() calls vfs_poll()
  vfs_poll() calls fuse_dev_poll()
    fuse_dev_poll() calls poll_wait(file, &fiq->waitq, wait)
      poll_wait() calls aio_poll_queue_proc(file, &fiq->waitq, wait)
        aio_poll_queue_proc() stores &fiq->waitq in pt->iocb->poll.head
aio_poll() calls spin_lock_irq(&ctx->ctx_lock)
aio_poll() calls spin_lock(&req->head->lock) (req == &pt->iocb->poll).

I think the lockdep complaint is about the FUSE fiq->waitq lock not being
IRQ-safe and about aio_poll() creating a dependency between an IRQ-safe lock
(ctx->ctx_lock) and a lock that is not IRQ-safe (fiq->waitq).

Thanks,

Bart.