On Mon, Sep 27, 2021 at 12:10:47PM -0400, Hamza Mahfooz wrote:
> Commit 833f4154ed56 ("aio: fold lookup_kiocb() into its sole caller")
> suggests that, the fact that active_reqs is a linked-list means aio_kiocb
> lookups in io_cancel() are inefficient. So, to get faster lookups while
> maintaining similar characteristics elsewhere, turn active_reqs into an
> array and keep track of the free indices of the array (so we know which
> indices are safe to use).
As requested last time: please explain what workload you have where
cancellation performance actually matters.
