On Wed, Jan 12, 2022 at 6:06 AM Peter Zijlstra wrote:
> *urgh*... so you're making the worst case less likely but fundamentally
> you don't change anything.
>
> If one of those low prio threads manages to block while holding
> cp_rwsem your checkpoint thread will still block for a very long time.
On 1/11/22 19:25, Jaegeuk Kim wrote:
On 01/11, Waiman Long wrote:
v5.10 kernel still have reader optimistic spinning enabled in rwsem which
may have worsen the writer wait time. Could you try with a more up-to-date
kernel or backport the relevant rwsem patches into your test kernel to see
how m
On Mon, Jan 10, 2022 at 11:41:23AM -0800, Tim Murray wrote:
> 1. f2fs-ckpt thread is running f2fs_write_checkpoint(), holding the
> cp_rwsem write lock while doing so via f2fs_lock_all() in
> block_operations().
> 2. Random very-low-priority thread A makes some other f2fs call that
> tries to get
