On 5/20/2020 9:19 AM, Devesh Sharma wrote:
+ +static int ib_alloc_cqs(struct ib_device *dev, int nr_cqes, + enum ib_poll_context poll_ctx) +{ + LIST_HEAD(tmp_list); + struct ib_cq *cq; + unsigned long flags; + int nr_cqs, ret, i; + + /* + * Allocated at least as many CQEs as requested, and otherwise + * a reasonable batch size so that we can share CQs between + * multiple users instead of allocating a larger number of CQs. + */ + nr_cqes = min(dev->attrs.max_cqe, max(nr_cqes, IB_MAX_SHARED_CQ_SZ)); + nr_cqs = min_t(int, dev->num_comp_vectors, num_online_cpus());No WARN() or return with failure as pointed by Leon and me. Has anything else changes elsewhere?
Hey Devesh, I am not sure what you are referring to, could you please clarify?
+ for (i = 0; i < nr_cqs; i++) { + cq = ib_alloc_cq(dev, NULL, nr_cqes, i, poll_ctx); + if (IS_ERR(cq)) { + ret = PTR_ERR(cq); + goto out_free_cqs; + } + cq->shared = true; + list_add_tail(&cq->pool_entry, &tmp_list); + } + + spin_lock_irqsave(&dev->cq_pools_lock, flags); + list_splice(&tmp_list, &dev->cq_pools[poll_ctx - 1]); + spin_unlock_irqrestore(&dev->cq_pools_lock, flags); + + return 0; + +out_free_cqs: + list_for_each_entry(cq, &tmp_list, pool_entry) { + cq->shared = false; + ib_free_cq(cq); + } + return ret; +} +
