On 5/30/22 00:25, Jens Axboe wrote:
On 5/29/22 10:20 AM, Hao Xu wrote:
From: Hao Xu <howeyxu@xxxxxxxxxxx>
From: Hao Xu <howeyxu@xxxxxxxxxxx>
Use per list lock for cancel_hash, this removes some completion lock
invocation and remove contension between different cancel_hash entries
Interesting, do you have any numbers on this?
Just Theoretically for now, I'll do some tests tomorrow. This is
actually RFC, forgot to change the subject.
Also, I'd make a hash bucket struct:
struct io_hash_bucket {
spinlock_t lock;
struct hlist_head list;
};
rather than two separate structs, that'll have nicer memory locality too
and should further improve it. Could be done as a prep patch with the
old locking in place, making the end patch doing the per-bucket lock
simpler as well.
Sure, if the test number make sense, I'll send v2. I'll test the
hlist_bl list as well(the comment of it says it is much slower than
normal spin_lock, but we may not care the efficiency of poll
cancellation very much?).
Hmm?
