From patchwork Thu Jun 11 23:08:50 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Marciniszyn, Mike" X-Patchwork-Id: 6592971 Return-Path: X-Original-To: patchwork-linux-rdma@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork1.web.kernel.org (Postfix) with ESMTP id EC4889F326 for ; Thu, 11 Jun 2015 23:08:57 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 5F7BF2052D for ; Thu, 11 Jun 2015 23:08:56 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 871B520555 for ; Thu, 11 Jun 2015 23:08:54 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751731AbbFKXIx (ORCPT ); Thu, 11 Jun 2015 19:08:53 -0400 Received: from mga01.intel.com ([192.55.52.88]:44080 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752695AbbFKXIx (ORCPT ); Thu, 11 Jun 2015 19:08:53 -0400 Received: from orsmga003.jf.intel.com ([10.7.209.27]) by fmsmga101.fm.intel.com with ESMTP; 11 Jun 2015 16:08:52 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.13,598,1427785200"; d="scan'208";a="586351101" Received: from sedona.ch.intel.com ([143.182.228.65]) by orsmga003.jf.intel.com with ESMTP; 11 Jun 2015 16:08:51 -0700 Received: from phlsvsles11.ph.intel.com (phlsvsles11.ph.intel.com [10.228.195.43]) by sedona.ch.intel.com (8.13.6/8.14.3/Standard MailSET/Hub) with ESMTP id t5BN8oVc003393; Thu, 11 Jun 2015 16:08:51 -0700 Received: from phlsvslse11.ph.intel.com (localhost [127.0.0.1]) by phlsvsles11.ph.intel.com with ESMTP id t5BN8oaE016882; Thu, 11 Jun 2015 19:08:50 -0400 Subject: [PATCH 04/41] IB/hfi1: add completion queue processing To: dledford@redhat.com From: Mike Marciniszyn Cc: linux-rdma@vger.kernel.org Date: Thu, 11 Jun 2015 19:08:50 -0400 Message-ID: <20150611230850.16479.68980.stgit@phlsvslse11.ph.intel.com> In-Reply-To: <20150611230710.16479.62955.stgit@phlsvslse11.ph.intel.com> References: <20150611230710.16479.62955.stgit@phlsvslse11.ph.intel.com> User-Agent: StGit/0.16 MIME-Version: 1.0 Sender: linux-rdma-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, T_RP_MATCHES_RCVD, UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP Signed-off-by: Andrew Friedley Signed-off-by: Arthur Kepner Signed-off-by: Brendan Cunningham Signed-off-by: Brian Welty Signed-off-by: Caz Yokoyama Signed-off-by: Dean Luick Signed-off-by: Dennis Dalessandro Signed-off-by: Easwar Hariharan Signed-off-by: Harish Chegondi Signed-off-by: Ira Weiny Signed-off-by: Jim Snow Signed-off-by: John Gregor Signed-off-by: Jubin John Signed-off-by: Kaike Wan Signed-off-by: Kevin Pine Signed-off-by: Kyle Liddell Signed-off-by: Mike Marciniszyn Signed-off-by: Mitko Haralanov Signed-off-by: Ravi Krishnaswamy Signed-off-by: Sadanand Warrier Signed-off-by: Sanath Kumar Signed-off-by: Sudeep Dutt Signed-off-by: Vlad Danushevsky --- drivers/infiniband/hw/hfi1/cq.c | 556 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 556 insertions(+) create mode 100644 drivers/infiniband/hw/hfi1/cq.c -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/drivers/infiniband/hw/hfi1/cq.c b/drivers/infiniband/hw/hfi1/cq.c new file mode 100644 index 0000000..05de35e --- /dev/null +++ b/drivers/infiniband/hw/hfi1/cq.c @@ -0,0 +1,556 @@ +/* + * + * This file is provided under a dual BSD/GPLv2 license. When using or + * redistributing this file, you may do so under either license. + * + * GPL LICENSE SUMMARY + * + * Copyright(c) 2015 Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of version 2 of the GNU General Public License as + * published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * BSD LICENSE + * + * Copyright(c) 2015 Intel Corporation. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * - Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * - Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * - Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + */ + +#include +#include +#include +#include + +#include "verbs.h" +#include "hfi.h" + +/** + * hfi1_cq_enter - add a new entry to the completion queue + * @cq: completion queue + * @entry: work completion entry to add + * @sig: true if @entry is a solicited entry + * + * This may be called with qp->s_lock held. + */ +void hfi1_cq_enter(struct hfi1_cq *cq, struct ib_wc *entry, int solicited) +{ + struct hfi1_cq_wc *wc; + unsigned long flags; + u32 head; + u32 next; + + spin_lock_irqsave(&cq->lock, flags); + + /* + * Note that the head pointer might be writable by user processes. + * Take care to verify it is a sane value. + */ + wc = cq->queue; + head = wc->head; + if (head >= (unsigned) cq->ibcq.cqe) { + head = cq->ibcq.cqe; + next = 0; + } else + next = head + 1; + if (unlikely(next == wc->tail)) { + spin_unlock_irqrestore(&cq->lock, flags); + if (cq->ibcq.event_handler) { + struct ib_event ev; + + ev.device = cq->ibcq.device; + ev.element.cq = &cq->ibcq; + ev.event = IB_EVENT_CQ_ERR; + cq->ibcq.event_handler(&ev, cq->ibcq.cq_context); + } + return; + } + if (cq->ip) { + wc->uqueue[head].wr_id = entry->wr_id; + wc->uqueue[head].status = entry->status; + wc->uqueue[head].opcode = entry->opcode; + wc->uqueue[head].vendor_err = entry->vendor_err; + wc->uqueue[head].byte_len = entry->byte_len; + wc->uqueue[head].ex.imm_data = + (__u32 __force)entry->ex.imm_data; + wc->uqueue[head].qp_num = entry->qp->qp_num; + wc->uqueue[head].src_qp = entry->src_qp; + wc->uqueue[head].wc_flags = entry->wc_flags; + wc->uqueue[head].pkey_index = entry->pkey_index; + wc->uqueue[head].slid = entry->slid; + wc->uqueue[head].sl = entry->sl; + wc->uqueue[head].dlid_path_bits = entry->dlid_path_bits; + wc->uqueue[head].port_num = entry->port_num; + /* Make sure entry is written before the head index. */ + smp_wmb(); + } else + wc->kqueue[head] = *entry; + wc->head = next; + + if (cq->notify == IB_CQ_NEXT_COMP || + (cq->notify == IB_CQ_SOLICITED && + (solicited || entry->status != IB_WC_SUCCESS))) { + struct kthread_worker *worker; + /* + * This will cause send_complete() to be called in + * another thread. + */ + smp_read_barrier_depends(); /* see hfi1_cq_exit */ + worker = cq->dd->worker; + if (likely(worker)) { + cq->notify = IB_CQ_NONE; + cq->triggered++; + queue_kthread_work(worker, &cq->comptask); + } + } + + spin_unlock_irqrestore(&cq->lock, flags); +} + +/** + * hfi1_poll_cq - poll for work completion entries + * @ibcq: the completion queue to poll + * @num_entries: the maximum number of entries to return + * @entry: pointer to array where work completions are placed + * + * Returns the number of completion entries polled. + * + * This may be called from interrupt context. Also called by ib_poll_cq() + * in the generic verbs code. + */ +int hfi1_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry) +{ + struct hfi1_cq *cq = to_icq(ibcq); + struct hfi1_cq_wc *wc; + unsigned long flags; + int npolled; + u32 tail; + + /* The kernel can only poll a kernel completion queue */ + if (cq->ip) { + npolled = -EINVAL; + goto bail; + } + + spin_lock_irqsave(&cq->lock, flags); + + wc = cq->queue; + tail = wc->tail; + if (tail > (u32) cq->ibcq.cqe) + tail = (u32) cq->ibcq.cqe; + for (npolled = 0; npolled < num_entries; ++npolled, ++entry) { + if (tail == wc->head) + break; + /* The kernel doesn't need a RMB since it has the lock. */ + *entry = wc->kqueue[tail]; + if (tail >= cq->ibcq.cqe) + tail = 0; + else + tail++; + } + wc->tail = tail; + + spin_unlock_irqrestore(&cq->lock, flags); + +bail: + return npolled; +} + +static void send_complete(struct kthread_work *work) +{ + struct hfi1_cq *cq = container_of(work, struct hfi1_cq, comptask); + + /* + * The completion handler will most likely rearm the notification + * and poll for all pending entries. If a new completion entry + * is added while we are in this routine, queue_work() + * won't call us again until we return so we check triggered to + * see if we need to call the handler again. + */ + for (;;) { + u8 triggered = cq->triggered; + + /* + * IPoIB connected mode assumes the callback is from a + * soft IRQ. We simulate this by blocking "bottom halves". + * See the implementation for ipoib_cm_handle_tx_wc(), + * netif_tx_lock_bh() and netif_tx_lock(). + */ + local_bh_disable(); + cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context); + local_bh_enable(); + + if (cq->triggered == triggered) + return; + } +} + +/** + * hfi1_create_cq - create a completion queue + * @ibdev: the device this completion queue is attached to + * @entries: the minimum size of the completion queue + * @context: unused by the driver + * @udata: user data for libibverbs.so + * + * Returns a pointer to the completion queue or negative errno values + * for failure. + * + * Called by ib_create_cq() in the generic verbs code. + */ +struct ib_cq *hfi1_create_cq(struct ib_device *ibdev, int entries, + int comp_vector, struct ib_ucontext *context, + struct ib_udata *udata) +{ + struct hfi1_ibdev *dev = to_idev(ibdev); + struct hfi1_cq *cq; + struct hfi1_cq_wc *wc; + struct ib_cq *ret; + u32 sz; + + if (entries < 1 || entries > hfi1_max_cqes) { + ret = ERR_PTR(-EINVAL); + goto done; + } + + /* Allocate the completion queue structure. */ + cq = kmalloc(sizeof(*cq), GFP_KERNEL); + if (!cq) { + ret = ERR_PTR(-ENOMEM); + goto done; + } + + /* + * Allocate the completion queue entries and head/tail pointers. + * This is allocated separately so that it can be resized and + * also mapped into user space. + * We need to use vmalloc() in order to support mmap and large + * numbers of entries. + */ + sz = sizeof(*wc); + if (udata && udata->outlen >= sizeof(__u64)) + sz += sizeof(struct ib_uverbs_wc) * (entries + 1); + else + sz += sizeof(struct ib_wc) * (entries + 1); + wc = vmalloc_user(sz); + if (!wc) { + ret = ERR_PTR(-ENOMEM); + goto bail_cq; + } + + /* + * Return the address of the WC as the offset to mmap. + * See hfi1_mmap() for details. + */ + if (udata && udata->outlen >= sizeof(__u64)) { + int err; + + cq->ip = hfi1_create_mmap_info(dev, sz, context, wc); + if (!cq->ip) { + ret = ERR_PTR(-ENOMEM); + goto bail_wc; + } + + err = ib_copy_to_udata(udata, &cq->ip->offset, + sizeof(cq->ip->offset)); + if (err) { + ret = ERR_PTR(err); + goto bail_ip; + } + } else + cq->ip = NULL; + + spin_lock(&dev->n_cqs_lock); + if (dev->n_cqs_allocated == hfi1_max_cqs) { + spin_unlock(&dev->n_cqs_lock); + ret = ERR_PTR(-ENOMEM); + goto bail_ip; + } + + dev->n_cqs_allocated++; + spin_unlock(&dev->n_cqs_lock); + + if (cq->ip) { + spin_lock_irq(&dev->pending_lock); + list_add(&cq->ip->pending_mmaps, &dev->pending_mmaps); + spin_unlock_irq(&dev->pending_lock); + } + + /* + * ib_create_cq() will initialize cq->ibcq except for cq->ibcq.cqe. + * The number of entries should be >= the number requested or return + * an error. + */ + cq->dd = dd_from_dev(dev); + cq->ibcq.cqe = entries; + cq->notify = IB_CQ_NONE; + cq->triggered = 0; + spin_lock_init(&cq->lock); + init_kthread_work(&cq->comptask, send_complete); + wc->head = 0; + wc->tail = 0; + cq->queue = wc; + + ret = &cq->ibcq; + + goto done; + +bail_ip: + kfree(cq->ip); +bail_wc: + vfree(wc); +bail_cq: + kfree(cq); +done: + return ret; +} + +/** + * hfi1_destroy_cq - destroy a completion queue + * @ibcq: the completion queue to destroy. + * + * Returns 0 for success. + * + * Called by ib_destroy_cq() in the generic verbs code. + */ +int hfi1_destroy_cq(struct ib_cq *ibcq) +{ + struct hfi1_ibdev *dev = to_idev(ibcq->device); + struct hfi1_cq *cq = to_icq(ibcq); + + flush_kthread_work(&cq->comptask); + spin_lock(&dev->n_cqs_lock); + dev->n_cqs_allocated--; + spin_unlock(&dev->n_cqs_lock); + if (cq->ip) + kref_put(&cq->ip->ref, hfi1_release_mmap_info); + else + vfree(cq->queue); + kfree(cq); + + return 0; +} + +/** + * hfi1_req_notify_cq - change the notification type for a completion queue + * @ibcq: the completion queue + * @notify_flags: the type of notification to request + * + * Returns 0 for success. + * + * This may be called from interrupt context. Also called by + * ib_req_notify_cq() in the generic verbs code. + */ +int hfi1_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags notify_flags) +{ + struct hfi1_cq *cq = to_icq(ibcq); + unsigned long flags; + int ret = 0; + + spin_lock_irqsave(&cq->lock, flags); + /* + * Don't change IB_CQ_NEXT_COMP to IB_CQ_SOLICITED but allow + * any other transitions (see C11-31 and C11-32 in ch. 11.4.2.2). + */ + if (cq->notify != IB_CQ_NEXT_COMP) + cq->notify = notify_flags & IB_CQ_SOLICITED_MASK; + + if ((notify_flags & IB_CQ_REPORT_MISSED_EVENTS) && + cq->queue->head != cq->queue->tail) + ret = 1; + + spin_unlock_irqrestore(&cq->lock, flags); + + return ret; +} + +/** + * hfi1_resize_cq - change the size of the CQ + * @ibcq: the completion queue + * + * Returns 0 for success. + */ +int hfi1_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata) +{ + struct hfi1_cq *cq = to_icq(ibcq); + struct hfi1_cq_wc *old_wc; + struct hfi1_cq_wc *wc; + u32 head, tail, n; + int ret; + u32 sz; + + if (cqe < 1 || cqe > hfi1_max_cqes) { + ret = -EINVAL; + goto bail; + } + + /* + * Need to use vmalloc() if we want to support large #s of entries. + */ + sz = sizeof(*wc); + if (udata && udata->outlen >= sizeof(__u64)) + sz += sizeof(struct ib_uverbs_wc) * (cqe + 1); + else + sz += sizeof(struct ib_wc) * (cqe + 1); + wc = vmalloc_user(sz); + if (!wc) { + ret = -ENOMEM; + goto bail; + } + + /* Check that we can write the offset to mmap. */ + if (udata && udata->outlen >= sizeof(__u64)) { + __u64 offset = 0; + + ret = ib_copy_to_udata(udata, &offset, sizeof(offset)); + if (ret) + goto bail_free; + } + + spin_lock_irq(&cq->lock); + /* + * Make sure head and tail are sane since they + * might be user writable. + */ + old_wc = cq->queue; + head = old_wc->head; + if (head > (u32) cq->ibcq.cqe) + head = (u32) cq->ibcq.cqe; + tail = old_wc->tail; + if (tail > (u32) cq->ibcq.cqe) + tail = (u32) cq->ibcq.cqe; + if (head < tail) + n = cq->ibcq.cqe + 1 + head - tail; + else + n = head - tail; + if (unlikely((u32)cqe < n)) { + ret = -EINVAL; + goto bail_unlock; + } + for (n = 0; tail != head; n++) { + if (cq->ip) + wc->uqueue[n] = old_wc->uqueue[tail]; + else + wc->kqueue[n] = old_wc->kqueue[tail]; + if (tail == (u32) cq->ibcq.cqe) + tail = 0; + else + tail++; + } + cq->ibcq.cqe = cqe; + wc->head = n; + wc->tail = 0; + cq->queue = wc; + spin_unlock_irq(&cq->lock); + + vfree(old_wc); + + if (cq->ip) { + struct hfi1_ibdev *dev = to_idev(ibcq->device); + struct hfi1_mmap_info *ip = cq->ip; + + hfi1_update_mmap_info(dev, ip, sz, wc); + + /* + * Return the offset to mmap. + * See hfi1_mmap() for details. + */ + if (udata && udata->outlen >= sizeof(__u64)) { + ret = ib_copy_to_udata(udata, &ip->offset, + sizeof(ip->offset)); + if (ret) + goto bail; + } + + spin_lock_irq(&dev->pending_lock); + if (list_empty(&ip->pending_mmaps)) + list_add(&ip->pending_mmaps, &dev->pending_mmaps); + spin_unlock_irq(&dev->pending_lock); + } + + ret = 0; + goto bail; + +bail_unlock: + spin_unlock_irq(&cq->lock); +bail_free: + vfree(wc); +bail: + return ret; +} + +int hfi1_cq_init(struct hfi1_devdata *dd) +{ + int ret = 0; + int cpu; + struct task_struct *task; + + if (dd->worker) + return 0; + dd->worker = kzalloc(sizeof(*dd->worker), GFP_KERNEL); + if (!dd->worker) + return -ENOMEM; + init_kthread_worker(dd->worker); + task = kthread_create_on_node( + kthread_worker_fn, + dd->worker, + dd->assigned_node_id, + "hfi1_cq%d", dd->unit); + if (IS_ERR(task)) + goto task_fail; + cpu = cpumask_first(cpumask_of_node(dd->assigned_node_id)); + kthread_bind(task, cpu); + wake_up_process(task); +out: + return ret; +task_fail: + ret = PTR_ERR(task); + kfree(dd->worker); + dd->worker = NULL; + goto out; +} + +void hfi1_cq_exit(struct hfi1_devdata *dd) +{ + struct kthread_worker *worker; + + worker = dd->worker; + if (!worker) + return; + /* blocks future queuing from send_complete() */ + dd->worker = NULL; + smp_wmb(); /* See hfi1_cq_enter */ + flush_kthread_worker(worker); + kthread_stop(worker->task); + kfree(worker); +}