On 2022/10/17 21:39, Christoph Hellwig wrote:
On Thu, Oct 13, 2022 at 05:44:49PM +0800, Chao Leng wrote:+ rcu = kvmalloc(count * sizeof(*rcu), GFP_KERNEL); + if (rcu) { + list_for_each_entry(q, &set->tag_list, tag_set_list) { + if (blk_queue_noquiesced(q)) + continue; + + init_rcu_head(&rcu[i].head); + init_completion(&rcu[i].completion); + call_srcu(q->srcu, &rcu[i].head, wakeme_after_rcu); + i++; + } + + for (i = 0; i < count; i++) { + wait_for_completion(&rcu[i].completion); + destroy_rcu_head(&rcu[i].head); + } + kvfree(rcu); + } else { + list_for_each_entry(q, &set->tag_list, tag_set_list) + synchronize_srcu(q->srcu); + }Having to allocate a struct rcu_synchronize for each of the potentially many queues here is a bit sad. Pull just explained the start_poll_synchronize_rcu interfaces at ALPSS last week, so I wonder if something like that would also be feasible for SRCU, as that would come in really handy here.
Using start_poll_synchronize_rcu will reduce some memory. This is a good idea.
.
