On Sun, Apr 05, 2026 at 09:02:30AM +0800, zhang wrote: > Dear Ming Lei: > > I just looked at your solution, which is to reset immediately after each IO > is claimed. This solution can indeed effectively solve the blocking problem. > However, if there is a task that is repeated and exactly the same, then the > process may run repeatedly many times. Do you think we should add some logic > to handle repeated tasks? >
->cancel_fn() can provide the forward progress, so it shouldn't be one problem. If the userspace wants to repeat this action, let it do it, and kernel won't hang with Uday's fix. > Reviewed-by: zhang, the-essence-of-life. Usually we show the whole email address. Thanks, Ming
