From patchwork Mon Sep 12 04:49:23 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Adit Ranadive X-Patchwork-Id: 9325707 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 6DE6A6048B for ; Mon, 12 Sep 2016 04:51:45 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 6286028ACC for ; Mon, 12 Sep 2016 04:51:45 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 56B2128B06; Mon, 12 Sep 2016 04:51:45 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.8 required=2.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,T_DKIM_INVALID autolearn=unavailable version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 7851128AD7 for ; Mon, 12 Sep 2016 04:51:43 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932677AbcILEv3 (ORCPT ); Mon, 12 Sep 2016 00:51:29 -0400 Received: from ex13-edg-ou-001.vmware.com ([208.91.0.189]:38056 "EHLO EX13-EDG-OU-001.vmware.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932585AbcILEup (ORCPT ); Mon, 12 Sep 2016 00:50:45 -0400 Received: from sc9-mailhost2.vmware.com (10.113.161.72) by EX13-EDG-OU-001.vmware.com (10.113.208.155) with Microsoft SMTP Server id 15.0.1156.6; Sun, 11 Sep 2016 21:50:32 -0700 Received: from EX13-CAS-004.vmware.com (smtp-inbound.vmware.com [10.113.191.54]) by sc9-mailhost2.vmware.com (Postfix) with ESMTP id B6180B04DB; Sun, 11 Sep 2016 21:50:43 -0700 (PDT) Received: from EX13-CAS-003.vmware.com (10.113.191.53) by EX13-MBX-027.vmware.com (10.113.191.47) with Microsoft SMTP Server (TLS) id 15.0.1156.6; Sun, 11 Sep 2016 21:50:43 -0700 Received: from NAM03-BY2-obe.outbound.protection.outlook.com (10.113.170.11) by EX13-CAS-003.vmware.com (10.113.191.53) with Microsoft SMTP Server (TLS) id 15.0.1156.6 via Frontend Transport; Sun, 11 Sep 2016 21:50:43 -0700 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=onevmw.onmicrosoft.com; s=selector1-vmware-com; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version; bh=eNErXuSBqOYB7bNWP7/yupqsZ+YEnchWoXBuMCXKE40=; b=h+6aSFSDngxFdz8roI1cA6hWDOFNbhOV7RzD33eSJx6PUK88NNNBVPZsvLgllLNoVBLBwdjdC1GpLPKyioaue+/XL+ppw8GcJXQD5tvlRXdpMaL8lodnLolNL2b6jkbZOvU8r7mbr2O291HZK9JTsKrBlJefsPeiYvO/WSwRs7w= Authentication-Results: redhat.com; dkim=none (message not signed) header.d=none; redhat.com; dmarc=none action=none header.from=vmware.com; Received: from promb-2s-dhcp95-136.eng.vmware.com (208.91.1.34) by BLUPR0501MB836.namprd05.prod.outlook.com (10.141.251.150) with Microsoft SMTP Server (version=TLS1_0, cipher=TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA_P384) id 15.1.619.3; Mon, 12 Sep 2016 04:50:38 +0000 From: Adit Ranadive To: , , CC: Adit Ranadive , , , , , , Subject: [PATCH v4 13/16] IB/pvrdma: Add the main driver module for PVRDMA Date: Sun, 11 Sep 2016 21:49:23 -0700 Message-ID: <1473655766-31628-14-git-send-email-aditr@vmware.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1473655766-31628-1-git-send-email-aditr@vmware.com> References: <1473655766-31628-1-git-send-email-aditr@vmware.com> MIME-Version: 1.0 X-Originating-IP: [208.91.1.34] X-ClientProxiedBy: BY2PR04CA052.namprd04.prod.outlook.com (10.141.249.170) To BLUPR0501MB836.namprd05.prod.outlook.com (10.141.251.150) X-MS-Office365-Filtering-Correlation-Id: b721ecc6-cf61-4b5d-a490-08d3dac8584b X-Microsoft-Exchange-Diagnostics: 1; BLUPR0501MB836; 2:0ufUZ2VUKuW6tvgcOuGXxBe4d7xFlX52HYtx7NEHvF1DZwTbuxrj0W9PX/kUXP0AFpssP3UktyyjNqJoKBv4h2Oh1VS0W7RVWcyrcP0YS0Yu1qUldkHaxDWcGcgVQ6iz/a5Oo4+uUl3jHzj80dJD/zV2/sKsHkv0spceTcI5DVMTWTPsJJ0pqhITJojMtOTJ; 3:ymt6ckYbunVI1SlflbxWIqxO3l7ZWDzX5Iije+upbehEGfB35LDmDDVR8IdXO8suDEXd7OK/bGKBit5sWj9yqkAQdMWSoeHdAnZN8/uAUm64UL+YlOnUWhmiJgwM/K7f X-Microsoft-Antispam: UriScan:;BCL:0;PCL:0;RULEID:;SRVR:BLUPR0501MB836; X-Microsoft-Exchange-Diagnostics: 1; BLUPR0501MB836; 25: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 X-LD-Processed: b39138ca-3cee-4b4a-a4d6-cd83d9dd62f0,ExtAddr X-Microsoft-Exchange-Diagnostics: 1; BLUPR0501MB836; 31:Md2QcmXC9bQHemcEAkPrUwKsfxnpvzLgXHqloiWaUXlrsOlkNXvjZZW2aIoOT4XSVLrZUlcwv87zZXRuM3jSB2u/R/xNiYgVMRSjIVNdQLzX+He+K0SVUQDFp9ivAt/gCqVwpfyLIz4Fz/Sr9wJM3l63a0jJiX4vEJXWsq/CFCnHjpeqzFMAUiBrGeW/UrmXO6l7JAR620fPFEnzZjS/24ItpCmRmJRq4PPUKo28hv0=; 20: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 X-Microsoft-Antispam-PRVS: X-Exchange-Antispam-Report-Test: UriScan:(250305191791016)(61668805478150)(22074186197030)(146099531331640); X-Exchange-Antispam-Report-CFA-Test: BCL:0; PCL:0; RULEID:(6040176)(601004)(2401047)(5005006)(8121501046)(3002001)(10201501046); SRVR:BLUPR0501MB836; BCL:0; PCL:0; RULEID:; SRVR:BLUPR0501MB836; X-Microsoft-Exchange-Diagnostics: 1; BLUPR0501MB836; 4:iZ7jhAlJ4A4Jr9lb2YcU6N6ZwMs/YJ8hPTZlb8XP1uYcWqKVeVJYHPnEOb05yqqxZPqv5KOebRFvAYYjOxHXKS3s3YBix1pZy0xwF93Jf67zFQzVU17L2GTmcp7CW1OsswX6a84eoS9SMzSHfHgCwRJVC0xXztOHbL1y1TlQ67A2sKIACfHeFwdlqMvkt5w0I6O74spTmQ5wRsOfFMUMaT/+oc6xT97bPQcGECyJkRWhzSymiv5gOYOm62K83T+ewm/16JJDEn5VrULaQ4tqzgHJqRnYRPq+WGi2RNQtK37uzITtdcaF+p70DtkJE05fgdtbC7rbGgrvcH14D/ZdTpazXQpOnzsaAAZ0T4eheMQq8Ri5wHqXuaF/dd931C1AaPiMWuAkpluhOH65tLt04u3XFkV7QXcD+D64ID+KXiU6AVs6xkEW1s4GNqAmoL+hBlumqg2NIhYpDVtA578GZKA1uQ2h/1Qi/eNizxV2I2OS7MFvq8y+tYrWyeiu2EZ+rbblrnwlxGoeZaeVxiyNeEyOM+eGlUInjgGjwXM+87k= X-Forefront-PRVS: 006339698F X-Forefront-Antispam-Report: SFV:NSPM; SFS:(10009020)(4630300001)(6009001)(7916002)(66066001)(107886002)(4001430100002)(5003940100001)(36756003)(47776003)(189998001)(4001450100002)(87386001)(86362001)(5890100001)(48376002)(2201001)(50466002)(2906002)(50226002)(8676002)(81166006)(586003)(4326007)(7846002)(5660300001)(305945005)(42186005)(76176999)(7736002)(50986999)(6116002)(3846002)(15975445007)(4001520100001)(5001770100001)(19580405001)(33646002)(92566002)(19580395003)(2950100001)(77096005)(229853001)(7099028)(5001760100003)(2004002); DIR:OUT; SFP:1101; SCL:1; SRVR:BLUPR0501MB836; H:promb-2s-dhcp95-136.eng.vmware.com; FPR:; SPF:None; LANG:en; X-Microsoft-Exchange-Diagnostics: =?us-ascii?Q?1; BLUPR0501MB836; 23:00tj5b4mObMrokxov6jAS71AyUb4zgeLFYCxNW4u?= =?us-ascii?Q?dbu58r6WTym78V0H6cfg9PLblnTAcUggnRCmSMZRoJnp0tWlmiab3xelVPcs?= =?us-ascii?Q?kUT04rz3QPtW+fRFbTcMpuOUny8eAnIeGfEajwqSfcHcIq7XqLrOXEnkjkCS?= =?us-ascii?Q?b8ojpeZjpHowsS0w7fz2DVa7kROdU8LcNTaAQGw4mHEvbjeADIxZeVTld8D/?= =?us-ascii?Q?NpAbiroUAGW0XKtsqKdxdUMJ7mOL0o3gTALRe9LsminAi6/es7cMQA54xuCW?= =?us-ascii?Q?7uHObypPasJkkms1xPg5w4tmouEZ9LB8bSZ/10HO5iH5bpHFzheHi+puVBTC?= =?us-ascii?Q?yPHdR0zNUXpvULXvEm/gwRMjveMnyD9U6E5yYB3KxekOolQvX7mkntDS4e05?= =?us-ascii?Q?ZwVr3+KWrBzLVmlgwK4DZBDAUaFMFDEo8PXdS97JZzrf/+CXkxH7nsbIVlE3?= =?us-ascii?Q?CfKxf6lPzhLZkvej3Y60UZ/9uXsaTFtVd4NV/BEait1VhqwM3w44MBowPTFS?= =?us-ascii?Q?oHp0h6vN/4MZjAiPJ2kV/xfKitEIQcYQBin8JlIU6nbvrUiVU/Bb0HEyno/q?= =?us-ascii?Q?kiAXKKyRVAnQYs8+5Td8WRdiUBrgXrhc8Vh5KmHIXd2dAB1jHHRLNBom8FSQ?= =?us-ascii?Q?XBTz3mUGtP2Aq2Qtl6vNmsaJqIip2BF/W8RZrPxzgQJBBDvG2fkukbQZaDyX?= =?us-ascii?Q?B81ffhCJQxhuq4WuDcuDykNlr0gP+J3819XjdAIUb4yh/y6ESR/v5wwVqL09?= =?us-ascii?Q?Z41vwGmosBKwCuCjvIci2IpE07jMbmYS5pZEtRNScjTUNokedLQFwpGoZJFB?= =?us-ascii?Q?xu9xoRxhv8GJEjzNvUTVAKhsMmoYpYmVm5D6BUKnqmQIVklLYoYY0NAOXl0w?= =?us-ascii?Q?+LTIUPxFdJ7PxeLRo0eOAIG3b0opw3Evax6aCEtov5uWAYwMd9z5Dpfhhu8E?= =?us-ascii?Q?7/Bozzbp8zSvZ13uL35ngL8ZOIjeWm3tA0zVPeeRFxeNEfI6CTgP1fBqaHMy?= =?us-ascii?Q?tU47eyhRB0S+5MuD5W847aCsi37pN4vHHwaCGp6ueoiY7fxMYn/71N6r/oFq?= =?us-ascii?Q?Ibjzxf/0e5Kk5GQ4hSJg9j5YrSdPKFEWsFeMCl5ZeQQOJdOWKF1o1B6+McJH?= =?us-ascii?Q?1euC3mjadJa75SVFm6Y4G4zMcWWd0oV9C/Gpr+3FnDjcVRRNQuppiA=3D=3D?= X-Microsoft-Exchange-Diagnostics: 1; BLUPR0501MB836; 6:m/3tNJx6oXV+c7/N1ciFtt93uDD1ncTFVMZbXYEbhdGuZe2D9samgxP9e8aP+u6rW/jywlt6eSs3MpC0OD+wq6uNdGJDDBK3E6WzFFc6z8DJ0C0SuTa4ylMrDMU3QpU1ZLRJzZj9PKRNK2gaoqBjClp1o9ILBAk+dDvS5QT6QN3MScyJ9eTAFYbU2jsneUnHdEMjUJ4u117qWX1UnUWzkBBHfaRAT4kZEI/kshxVTkkqzii+B0VbMNGBhvgpARhgrIMbgsQIbo1FD/NwbWR58RmOcOcG0GyfkRKvQ55raoY=; 5:DLLScLOtYjzGJj3lJXxVmPlwWJriffslRSAGfNf2N+7k1FKTdiK4qV0kTWFfiyZcWOewuV3aIpATXkSN50l5odAFzPhHrnLiBmjNsmAYv8z88/G1yBBzJLZ8cPTu+RueGqrGnSRuNrCNeuU5Jqba3A==; 24:f+Mfu/B3O91hWMR5KGDrCvgsUeqXZiEiV9RqkB0bHaYPE2/NmxL4MxylTUiFF0gleUZr26FUBsFt+LChvtru20+dW+SqduKrwi9A5lGAhyo=; 7:PLNhOM10W3ErOlT2RIeEwxLceWvkx3FSy/+FjGExz1ew+M+lOoPh6N4G2kbADMQk+JPqUVFH3JgcIHBM8cpSwhVDhL2izxj0UnxELi67PLPuHXFbhzNr3ux8IigMyGe7hWBlZdCKQEyovbQzYPCsuk4mUyjx95BeHABDJ/USFnCgWuegh4XgoifQx+LjYCNehTBmYwQW+7wfIXKqSuAtn9ba01t8DYqONZfIZB+e/Qil7Hj+C+imLmXJz/pLsAGC SpamDiagnosticOutput: 1:99 SpamDiagnosticMetadata: NSPM X-Microsoft-Exchange-Diagnostics: 1; BLUPR0501MB836; 20:+aHkICRQl2KYiO4ZH2cFpD4GByp7MUp84vft3B8/oypoSZ2K7a8I2KJWznrvCN8JlbRomhn0oqLWJaFr8eIRlTAHyl+3p7St5BujuKiOBIrPK3JZ/Jy1GeRNIdIhmkBQu1Mj7pHV2it3J1XyvWWzRz1bQpRTMNJwRgG3TUryqQ8= X-MS-Exchange-CrossTenant-OriginalArrivalTime: 12 Sep 2016 04:50:38.6928 (UTC) X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted X-MS-Exchange-Transport-CrossTenantHeadersStamped: BLUPR0501MB836 Received-SPF: None (EX13-EDG-OU-001.vmware.com: aditr@vmware.com does not designate permitted sender hosts) Sender: linux-rdma-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP This patch adds the support to register a RDMA device with the kernel RDMA stack as well as a kernel module. This also initializes the underlying virtual PCI device. Reviewed-by: Yuval Shaia Reviewed-by: Jorgen Hansen Reviewed-by: George Zhang Reviewed-by: Aditya Sarwade Reviewed-by: Bryan Tan Signed-off-by: Adit Ranadive --- Changes v3->v4: - Fixed some checkpatch warnings. - Added support for new get_dev_fw_str API. - Added event workqueue for netdevice events. - Restructured the pvrdma_pci_remove function a little bit. Changes v2->v3: - Removed boolean in pvrdma_cmd_post. Changes v1->v2: - Addressed 32-bit build errors - Cosmetic change to avoid if else in intr0_handler - Removed unnecessary return assignment. --- drivers/infiniband/hw/pvrdma/pvrdma_main.c | 1222 ++++++++++++++++++++++++++++ 1 file changed, 1222 insertions(+) create mode 100644 drivers/infiniband/hw/pvrdma/pvrdma_main.c diff --git a/drivers/infiniband/hw/pvrdma/pvrdma_main.c b/drivers/infiniband/hw/pvrdma/pvrdma_main.c new file mode 100644 index 0000000..d61f3b2 --- /dev/null +++ b/drivers/infiniband/hw/pvrdma/pvrdma_main.c @@ -0,0 +1,1222 @@ +/* + * Copyright (c) 2012-2016 VMware, Inc. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of EITHER the GNU General Public License + * version 2 as published by the Free Software Foundation or the BSD + * 2-Clause License. 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 version 2 for more details at + * http://www.gnu.org/licenses/old-licenses/gpl-2.0.en.html. + * + * You should have received a copy of the GNU General Public License + * along with this program available in the file COPYING in the main + * directory of this source tree. + * + * The BSD 2-Clause License + * + * 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. + * + * 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 HOLDER 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 +#include +#include +#include +#include + +#include "pvrdma.h" +#include "pvrdma_user.h" + +#define DRV_NAME "pvrdma" +#define DRV_VERSION "1.0" +#define DRV_RELDATE "January 1, 2013" + +static const char pvrdma_version[] = + DRV_NAME ": PVRDMA InfiniBand driver v" + DRV_VERSION " (" DRV_RELDATE ")\n"; + +static DEFINE_MUTEX(pvrdma_device_list_lock); +static LIST_HEAD(pvrdma_device_list); +static struct workqueue_struct *event_wq; + +static int pvrdma_add_gid(struct ib_device *ibdev, + u8 port_num, + unsigned int index, + const union ib_gid *gid, + const struct ib_gid_attr *attr, + void **context); +static int pvrdma_del_gid(struct ib_device *ibdev, + u8 port_num, + unsigned int index, + void **context); + + +static ssize_t show_hca(struct device *device, struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "PVRDMA%s\n", DRV_VERSION); +} + +static ssize_t show_rev(struct device *device, struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "%d\n", PVRDMA_REV_ID); +} + +static ssize_t show_board(struct device *device, struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "%d\n", PVRDMA_BOARD_ID); +} + +static DEVICE_ATTR(hw_rev, S_IRUGO, show_rev, NULL); +static DEVICE_ATTR(hca_type, S_IRUGO, show_hca, NULL); +static DEVICE_ATTR(board_id, S_IRUGO, show_board, NULL); + +static struct device_attribute *pvrdma_class_attributes[] = { + &dev_attr_hw_rev, + &dev_attr_hca_type, + &dev_attr_board_id +}; + +static void pvrdma_get_fw_ver_str(struct ib_device *device, char *str, + size_t str_len) +{ + struct pvrdma_dev *dev = + container_of(device, struct pvrdma_dev, ib_dev); + snprintf(str, str_len, "%d.%d.%d\n", + (int) (dev->dsr->caps.fw_ver >> 32), + (int) (dev->dsr->caps.fw_ver >> 16) & 0xffff, + (int) dev->dsr->caps.fw_ver & 0xffff); +} + +static int pvrdma_init_device(struct pvrdma_dev *dev) +{ + /* Initialize some device related stuff */ + spin_lock_init(&dev->cmd_lock); + sema_init(&dev->cmd_sema, 1); + atomic_set(&dev->num_qps, 0); + atomic_set(&dev->num_cqs, 0); + atomic_set(&dev->num_pds, 0); + atomic_set(&dev->num_ahs, 0); + + return 0; +} + +static int pvrdma_port_immutable(struct ib_device *ibdev, u8 port_num, + struct ib_port_immutable *immutable) +{ + struct ib_port_attr attr; + int err; + + err = pvrdma_query_port(ibdev, port_num, &attr); + if (err) + return err; + + immutable->pkey_tbl_len = attr.pkey_tbl_len; + immutable->gid_tbl_len = attr.gid_tbl_len; + immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE; + immutable->max_mad_size = IB_MGMT_MAD_SIZE; + return 0; +} + +static struct net_device *pvrdma_get_netdev(struct ib_device *ibdev, + u8 port_num) +{ + struct net_device *netdev; + struct pvrdma_dev *dev = to_vdev(ibdev); + + if (port_num != 1) + return NULL; + + rcu_read_lock(); + netdev = dev->netdev; + if (netdev) + dev_hold(netdev); + rcu_read_unlock(); + + return netdev; +} + +static int pvrdma_register_device(struct pvrdma_dev *dev) +{ + int ret = -1; + int i = 0; + + strlcpy(dev->ib_dev.name, "pvrdma%d", IB_DEVICE_NAME_MAX); + dev->ib_dev.node_guid = dev->dsr->caps.node_guid; + dev->sys_image_guid = dev->dsr->caps.sys_image_guid; + dev->flags = 0; + dev->ib_dev.owner = THIS_MODULE; + dev->ib_dev.num_comp_vectors = 1; + dev->ib_dev.dma_device = &dev->pdev->dev; + dev->ib_dev.uverbs_abi_ver = PVRDMA_UVERBS_ABI_VERSION; + dev->ib_dev.uverbs_cmd_mask = + (1ull << IB_USER_VERBS_CMD_GET_CONTEXT) | + (1ull << IB_USER_VERBS_CMD_QUERY_DEVICE) | + (1ull << IB_USER_VERBS_CMD_QUERY_PORT) | + (1ull << IB_USER_VERBS_CMD_ALLOC_PD) | + (1ull << IB_USER_VERBS_CMD_DEALLOC_PD) | + (1ull << IB_USER_VERBS_CMD_REG_MR) | + (1ull << IB_USER_VERBS_CMD_DEREG_MR) | + (1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL) | + (1ull << IB_USER_VERBS_CMD_CREATE_CQ) | + (1ull << IB_USER_VERBS_CMD_POLL_CQ) | + (1ull << IB_USER_VERBS_CMD_REQ_NOTIFY_CQ) | + (1ull << IB_USER_VERBS_CMD_DESTROY_CQ) | + (1ull << IB_USER_VERBS_CMD_CREATE_QP) | + (1ull << IB_USER_VERBS_CMD_MODIFY_QP) | + (1ull << IB_USER_VERBS_CMD_QUERY_QP) | + (1ull << IB_USER_VERBS_CMD_DESTROY_QP) | + (1ull << IB_USER_VERBS_CMD_POST_SEND) | + (1ull << IB_USER_VERBS_CMD_POST_RECV) | + (1ull << IB_USER_VERBS_CMD_CREATE_AH) | + (1ull << IB_USER_VERBS_CMD_DESTROY_AH); + + dev->ib_dev.node_type = RDMA_NODE_IB_CA; + dev->ib_dev.phys_port_cnt = dev->dsr->caps.phys_port_cnt; + + dev->ib_dev.query_device = pvrdma_query_device; + dev->ib_dev.query_port = pvrdma_query_port; + dev->ib_dev.query_gid = pvrdma_query_gid; + dev->ib_dev.query_pkey = pvrdma_query_pkey; + dev->ib_dev.modify_port = pvrdma_modify_port; + dev->ib_dev.alloc_ucontext = pvrdma_alloc_ucontext; + dev->ib_dev.dealloc_ucontext = pvrdma_dealloc_ucontext; + dev->ib_dev.mmap = pvrdma_mmap; + dev->ib_dev.alloc_pd = pvrdma_alloc_pd; + dev->ib_dev.dealloc_pd = pvrdma_dealloc_pd; + dev->ib_dev.create_ah = pvrdma_create_ah; + dev->ib_dev.destroy_ah = pvrdma_destroy_ah; + dev->ib_dev.create_qp = pvrdma_create_qp; + dev->ib_dev.modify_qp = pvrdma_modify_qp; + dev->ib_dev.query_qp = pvrdma_query_qp; + dev->ib_dev.destroy_qp = pvrdma_destroy_qp; + dev->ib_dev.post_send = pvrdma_post_send; + dev->ib_dev.post_recv = pvrdma_post_recv; + dev->ib_dev.create_cq = pvrdma_create_cq; + dev->ib_dev.modify_cq = pvrdma_modify_cq; + dev->ib_dev.resize_cq = pvrdma_resize_cq; + dev->ib_dev.destroy_cq = pvrdma_destroy_cq; + dev->ib_dev.poll_cq = pvrdma_poll_cq; + dev->ib_dev.req_notify_cq = pvrdma_req_notify_cq; + dev->ib_dev.get_dma_mr = pvrdma_get_dma_mr; + dev->ib_dev.reg_user_mr = pvrdma_reg_user_mr; + dev->ib_dev.dereg_mr = pvrdma_dereg_mr; + dev->ib_dev.alloc_mr = pvrdma_alloc_mr; + dev->ib_dev.map_mr_sg = pvrdma_map_mr_sg; + dev->ib_dev.add_gid = pvrdma_add_gid; + dev->ib_dev.del_gid = pvrdma_del_gid; + dev->ib_dev.get_netdev = pvrdma_get_netdev; + dev->ib_dev.get_port_immutable = pvrdma_port_immutable; + dev->ib_dev.get_link_layer = pvrdma_port_link_layer; + dev->ib_dev.get_dev_fw_str = pvrdma_get_fw_ver_str; + + mutex_init(&dev->port_mutex); + spin_lock_init(&dev->desc_lock); + + dev->cq_tbl = kcalloc(dev->dsr->caps.max_cq, sizeof(void *), + GFP_KERNEL); + if (!dev->cq_tbl) + return ret; + spin_lock_init(&dev->cq_tbl_lock); + + dev->qp_tbl = kcalloc(dev->dsr->caps.max_qp, sizeof(void *), + GFP_KERNEL); + if (!dev->qp_tbl) + goto err_cq_free; + spin_lock_init(&dev->qp_tbl_lock); + + ret = ib_register_device(&dev->ib_dev, NULL); + if (ret) + goto err_qp_free; + + for (i = 0; i < ARRAY_SIZE(pvrdma_class_attributes); ++i) { + ret = device_create_file(&dev->ib_dev.dev, + pvrdma_class_attributes[i]); + if (ret) + goto err_class; + } + + dev->ib_active = true; + + return 0; + +err_class: + ib_unregister_device(&dev->ib_dev); +err_qp_free: + kfree(dev->qp_tbl); +err_cq_free: + kfree(dev->cq_tbl); + + return ret; +} + +static irqreturn_t pvrdma_intr0_handler(int irq, void *dev_id) +{ + u32 icr = PVRDMA_INTR_CAUSE_RESPONSE; + struct pvrdma_dev *dev = dev_id; + + dev_dbg(&dev->pdev->dev, "interrupt 0 (response) handler\n"); + + if (dev->intr.type != PVRDMA_INTR_TYPE_MSIX) { + /* Legacy intr */ + icr = pvrdma_read_reg(dev, PVRDMA_REG_ICR); + if (icr == 0) + return IRQ_NONE; + } + + if (icr == PVRDMA_INTR_CAUSE_RESPONSE) + complete(&dev->cmd_done); + + return IRQ_HANDLED; +} + +static void pvrdma_qp_event(struct pvrdma_dev *dev, u32 qpn, int type) +{ + struct pvrdma_qp *qp; + unsigned long flags; + + spin_lock_irqsave(&dev->qp_tbl_lock, flags); + qp = dev->qp_tbl[qpn % dev->dsr->caps.max_qp]; + if (qp) + atomic_inc(&qp->refcnt); + spin_unlock_irqrestore(&dev->qp_tbl_lock, flags); + + if (qp && qp->ibqp.event_handler) { + struct ib_qp *ibqp = &qp->ibqp; + struct ib_event e; + + e.device = ibqp->device; + e.element.qp = ibqp; + e.event = type; /* 1:1 mapping for now. */ + ibqp->event_handler(&e, ibqp->qp_context); + } + if (qp) { + atomic_dec(&qp->refcnt); + if (atomic_read(&qp->refcnt) == 0) + wake_up(&qp->wait); + } +} + +static void pvrdma_cq_event(struct pvrdma_dev *dev, u32 cqn, int type) +{ + struct pvrdma_cq *cq; + unsigned long flags; + + spin_lock_irqsave(&dev->cq_tbl_lock, flags); + cq = dev->cq_tbl[cqn % dev->dsr->caps.max_cq]; + if (cq) + atomic_inc(&cq->refcnt); + spin_unlock_irqrestore(&dev->cq_tbl_lock, flags); + + if (cq && cq->ibcq.event_handler) { + struct ib_cq *ibcq = &cq->ibcq; + struct ib_event e; + + e.device = ibcq->device; + e.element.cq = ibcq; + e.event = type; /* 1:1 mapping for now. */ + ibcq->event_handler(&e, ibcq->cq_context); + } + if (cq) { + atomic_dec(&cq->refcnt); + if (atomic_read(&cq->refcnt) == 0) + wake_up(&cq->wait); + } +} + +static void pvrdma_dispatch_event(struct pvrdma_dev *dev, int port, + enum ib_event_type event) +{ + struct ib_event ib_event; + + memset(&ib_event, 0, sizeof(ib_event)); + ib_event.device = &dev->ib_dev; + ib_event.element.port_num = port; + ib_event.event = event; + ib_dispatch_event(&ib_event); +} + +static void pvrdma_dev_event(struct pvrdma_dev *dev, u8 port, int type) +{ + if (port < 1 || port > dev->dsr->caps.phys_port_cnt) { + dev_warn(&dev->pdev->dev, "event on port %d\n", port); + return; + } + + pvrdma_dispatch_event(dev, port, type); +} + +static inline struct pvrdma_eqe *get_eqe(struct pvrdma_dev *dev, unsigned int i) +{ + return (struct pvrdma_eqe *)pvrdma_page_dir_get_ptr( + &dev->async_pdir, + PAGE_SIZE + + sizeof(struct pvrdma_eqe) * i); +} + +static irqreturn_t pvrdma_intr1_handler(int irq, void *dev_id) +{ + struct pvrdma_dev *dev = dev_id; + struct pvrdma_ring *ring = &dev->async_ring_state->rx; + int ring_slots = (dev->dsr->async_ring_pages.num_pages - 1) * + PAGE_SIZE / sizeof(struct pvrdma_eqe); + unsigned int head; + + dev_dbg(&dev->pdev->dev, "interrupt 1 (async event) handler\n"); + + /* + * Don't process events until the IB device is registered. Otherwise + * we'll try to ib_dispatch_event() on an invalid device. + */ + if (!dev->ib_active) + return IRQ_HANDLED; + + while (pvrdma_idx_ring_has_data(ring, ring_slots, &head) > 0) { + struct pvrdma_eqe *eqe; + + eqe = get_eqe(dev, head); + + switch (eqe->type) { + case PVRDMA_EVENT_QP_FATAL: + case PVRDMA_EVENT_QP_REQ_ERR: + case PVRDMA_EVENT_QP_ACCESS_ERR: + case PVRDMA_EVENT_COMM_EST: + case PVRDMA_EVENT_SQ_DRAINED: + case PVRDMA_EVENT_PATH_MIG: + case PVRDMA_EVENT_PATH_MIG_ERR: + case PVRDMA_EVENT_QP_LAST_WQE_REACHED: + pvrdma_qp_event(dev, eqe->info, eqe->type); + break; + + case PVRDMA_EVENT_CQ_ERR: + pvrdma_cq_event(dev, eqe->info, eqe->type); + break; + + case PVRDMA_EVENT_SRQ_ERR: + case PVRDMA_EVENT_SRQ_LIMIT_REACHED: + break; + + case PVRDMA_EVENT_PORT_ACTIVE: + case PVRDMA_EVENT_PORT_ERR: + case PVRDMA_EVENT_LID_CHANGE: + case PVRDMA_EVENT_PKEY_CHANGE: + case PVRDMA_EVENT_SM_CHANGE: + case PVRDMA_EVENT_CLIENT_REREGISTER: + case PVRDMA_EVENT_GID_CHANGE: + pvrdma_dev_event(dev, eqe->info, eqe->type); + break; + + case PVRDMA_EVENT_DEVICE_FATAL: + pvrdma_dev_event(dev, 1, eqe->type); + break; + + default: + break; + } + + pvrdma_idx_ring_inc(&ring->cons_head, ring_slots); + } + + return IRQ_HANDLED; +} + +static inline struct pvrdma_cqne *get_cqne(struct pvrdma_dev *dev, + unsigned int i) +{ + return (struct pvrdma_cqne *)pvrdma_page_dir_get_ptr( + &dev->cq_pdir, + PAGE_SIZE + + sizeof(struct pvrdma_cqne) * i); +} + +static irqreturn_t pvrdma_intrx_handler(int irq, void *dev_id) +{ + struct pvrdma_dev *dev = dev_id; + struct pvrdma_ring *ring = &dev->cq_ring_state->rx; + int ring_slots = (dev->dsr->cq_ring_pages.num_pages - 1) * PAGE_SIZE / + sizeof(struct pvrdma_cqne); + unsigned int head; + unsigned long flags; + + dev_dbg(&dev->pdev->dev, "interrupt x (completion) handler\n"); + + while (pvrdma_idx_ring_has_data(ring, ring_slots, &head) > 0) { + struct pvrdma_cqne *cqne; + struct pvrdma_cq *cq; + + cqne = get_cqne(dev, head); + spin_lock_irqsave(&dev->cq_tbl_lock, flags); + cq = dev->cq_tbl[cqne->info % dev->dsr->caps.max_cq]; + if (cq) + atomic_inc(&cq->refcnt); + spin_unlock_irqrestore(&dev->cq_tbl_lock, flags); + + if (cq && cq->ibcq.comp_handler) + cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context); + if (cq) { + atomic_dec(&cq->refcnt); + if (atomic_read(&cq->refcnt)) + wake_up(&cq->wait); + } + pvrdma_idx_ring_inc(&ring->cons_head, ring_slots); + } + + return IRQ_HANDLED; +} + +static void pvrdma_disable_msi_all(struct pvrdma_dev *dev) +{ + if (dev->intr.type == PVRDMA_INTR_TYPE_MSIX) + pci_disable_msix(dev->pdev); + else if (dev->intr.type == PVRDMA_INTR_TYPE_MSI) + pci_disable_msi(dev->pdev); +} + +static void pvrdma_free_irq(struct pvrdma_dev *dev) +{ + int i; + + dev_dbg(&dev->pdev->dev, "freeing interrupts\n"); + + if (dev->intr.type == PVRDMA_INTR_TYPE_MSIX) { + for (i = 0; i < dev->intr.size; i++) { + if (dev->intr.enabled[i]) { + free_irq(dev->intr.msix_entry[i].vector, dev); + dev->intr.enabled[i] = 0; + } + } + } else if (dev->intr.type == PVRDMA_INTR_TYPE_INTX || + dev->intr.type == PVRDMA_INTR_TYPE_MSI) { + free_irq(dev->pdev->irq, dev); + } +} + +static void pvrdma_enable_intrs(struct pvrdma_dev *dev) +{ + dev_dbg(&dev->pdev->dev, "enable interrupts\n"); + pvrdma_write_reg(dev, PVRDMA_REG_IMR, 0); +} + +static void pvrdma_disable_intrs(struct pvrdma_dev *dev) +{ + dev_dbg(&dev->pdev->dev, "disable interrupts\n"); + pvrdma_write_reg(dev, PVRDMA_REG_IMR, ~0); +} + +static int pvrdma_enable_msix(struct pci_dev *pdev, struct pvrdma_dev *dev) +{ + int i; + int ret; + + for (i = 0; i < PVRDMA_MAX_INTERRUPTS; i++) { + dev->intr.msix_entry[i].entry = i; + dev->intr.msix_entry[i].vector = i; + + switch (i) { + case 0: + /* CMD ring handler */ + dev->intr.handler[i] = pvrdma_intr0_handler; + break; + case 1: + /* Async event ring handler */ + dev->intr.handler[i] = pvrdma_intr1_handler; + break; + default: + /* Completion queue handler */ + dev->intr.handler[i] = pvrdma_intrx_handler; + break; + } + } + + ret = pci_enable_msix(pdev, dev->intr.msix_entry, + PVRDMA_MAX_INTERRUPTS); + if (!ret) { + dev->intr.type = PVRDMA_INTR_TYPE_MSIX; + dev->intr.size = PVRDMA_MAX_INTERRUPTS; + } else if (ret > 0) { + ret = pci_enable_msix(pdev, dev->intr.msix_entry, ret); + if (!ret) { + dev->intr.type = PVRDMA_INTR_TYPE_MSIX; + dev->intr.size = ret; + } else { + dev->intr.size = 0; + } + } + + dev_dbg(&pdev->dev, "using interrupt type %d, size %d\n", + dev->intr.type, dev->intr.size); + + return ret; +} + +static int pvrdma_alloc_intrs(struct pvrdma_dev *dev) +{ + int ret = 0; + int i; + + if (pci_find_capability(dev->pdev, PCI_CAP_ID_MSIX) && + pvrdma_enable_msix(dev->pdev, dev)) { + /* Try MSI */ + ret = pci_enable_msi(dev->pdev); + if (!ret) { + dev->intr.type = PVRDMA_INTR_TYPE_MSI; + } else { + /* Legacy INTR */ + dev->intr.type = PVRDMA_INTR_TYPE_INTX; + } + } + + /* Request First IRQ */ + switch (dev->intr.type) { + case PVRDMA_INTR_TYPE_INTX: + case PVRDMA_INTR_TYPE_MSI: + ret = request_irq(dev->pdev->irq, pvrdma_intr0_handler, + IRQF_SHARED, DRV_NAME, dev); + if (ret) { + dev_err(&dev->pdev->dev, + "failed to request interrupt\n"); + goto disable_msi; + } + break; + case PVRDMA_INTR_TYPE_MSIX: + ret = request_irq(dev->intr.msix_entry[0].vector, + pvrdma_intr0_handler, 0, DRV_NAME, dev); + if (ret) { + dev_err(&dev->pdev->dev, + "failed to request interrupt 0\n"); + goto disable_msi; + } + dev->intr.enabled[0] = 1; + break; + default: + /* Not reached */ + break; + } + + /* For MSIX: request intr for each vector */ + if (dev->intr.size > 1) { + ret = request_irq(dev->intr.msix_entry[1].vector, + pvrdma_intr1_handler, 0, DRV_NAME, dev); + if (ret) { + dev_err(&dev->pdev->dev, + "failed to request interrupt 1\n"); + goto free_irq; + } + dev->intr.enabled[1] = 1; + + for (i = 2; i < dev->intr.size; i++) { + ret = request_irq(dev->intr.msix_entry[i].vector, + pvrdma_intrx_handler, 0, + DRV_NAME, dev); + if (ret) { + dev_err(&dev->pdev->dev, + "failed to request interrupt %d\n", i); + goto free_irq; + } + dev->intr.enabled[i] = 1; + } + } + + return 0; + +free_irq: + pvrdma_free_irq(dev); +disable_msi: + pvrdma_disable_msi_all(dev); + return ret; +} + +static void pvrdma_free_slots(struct pvrdma_dev *dev) +{ + struct pci_dev *pdev = dev->pdev; + + if (dev->resp_slot) + dma_free_coherent(&pdev->dev, PAGE_SIZE, dev->resp_slot, + dev->dsr->resp_slot_dma); + if (dev->cmd_slot) + dma_free_coherent(&pdev->dev, PAGE_SIZE, dev->cmd_slot, + dev->dsr->cmd_slot_dma); +} + +static int pvrdma_add_gid_at_index(struct pvrdma_dev *dev, + const union ib_gid *gid, + int index) +{ + int ret; + union pvrdma_cmd_req req; + struct pvrdma_cmd_create_bind *cmd_bind = &req.create_bind; + + if (!dev->sgid_tbl) { + dev_warn(&dev->pdev->dev, "sgid table not initialized\n"); + return -EINVAL; + } + + memset(cmd_bind, 0, sizeof(*cmd_bind)); + cmd_bind->hdr.cmd = PVRDMA_CMD_CREATE_BIND; + memcpy(cmd_bind->new_gid, gid->raw, 16); + cmd_bind->mtu = ib_mtu_enum_to_int(IB_MTU_1024); + cmd_bind->vlan = 0xfff; + cmd_bind->index = index; + cmd_bind->gid_type = PVRDMA_GID_TYPE_FLAG_ROCE_V1; + + ret = pvrdma_cmd_post(dev, &req, NULL); + if (ret < 0) { + dev_warn(&dev->pdev->dev, "could not create binding\n"); + return -EINVAL; + } + memcpy(&dev->sgid_tbl[index], gid, sizeof(*gid)); + return 0; +} + +static int pvrdma_add_gid(struct ib_device *ibdev, + u8 port_num, + unsigned int index, + const union ib_gid *gid, + const struct ib_gid_attr *attr, + void **context) +{ + struct pvrdma_dev *dev = to_vdev(ibdev); + + return pvrdma_add_gid_at_index(dev, gid, index); +} + +static int pvrdma_del_gid_at_index(struct pvrdma_dev *dev, int index) +{ + int ret; + union pvrdma_cmd_req req; + struct pvrdma_cmd_destroy_bind *cmd_dest = &req.destroy_bind; + + /* Update sgid table. */ + if (!dev->sgid_tbl) { + dev_warn(&dev->pdev->dev, "sgid table not initialized\n"); + return -EINVAL; + } + + memset(cmd_dest, 0, sizeof(*cmd_dest)); + cmd_dest->hdr.cmd = PVRDMA_CMD_DESTROY_BIND; + memcpy(cmd_dest->dest_gid, &dev->sgid_tbl[index], 16); + cmd_dest->index = index; + + ret = pvrdma_cmd_post(dev, &req, NULL); + if (ret < 0) { + dev_warn(&dev->pdev->dev, + "could not destroy binding\n"); + return ret; + } + memset(&dev->sgid_tbl[index], 0, 16); + return 0; +} + +static int pvrdma_del_gid(struct ib_device *ibdev, + u8 port_num, + unsigned int index, + void **context) +{ + struct pvrdma_dev *dev = to_vdev(ibdev); + + dev_dbg(&dev->pdev->dev, "removing gid at index %u from %s", + index, dev->netdev->name); + + return pvrdma_del_gid_at_index(dev, index); +} + +static void pvrdma_netdevice_event_handle(struct pvrdma_dev *dev, + unsigned long event) +{ + struct net_device *netdev; + + netdev = dev->netdev; + switch (event) { + case NETDEV_REBOOT: + case NETDEV_DOWN: + pvrdma_dispatch_event(dev, 1, IB_EVENT_PORT_ERR); + break; + case NETDEV_UP: + pvrdma_dispatch_event(dev, 1, IB_EVENT_PORT_ACTIVE); + break; + default: + dev_dbg(&dev->pdev->dev, "ignore netdevice event %ld on %s\n", + event, dev->ib_dev.name); + break; + } +} + +static void pvrdma_netdevice_event_work(struct work_struct *work) +{ + struct pvrdma_netdevice_work *netdev_work; + struct pvrdma_dev *dev; + + netdev_work = container_of(work, struct pvrdma_netdevice_work, work); + + mutex_lock(&pvrdma_device_list_lock); + list_for_each_entry(dev, &pvrdma_device_list, device_link) { + if (dev->netdev == netdev_work->event_netdev) { + pvrdma_netdevice_event_handle(dev, netdev_work->event); + break; + } + } + mutex_unlock(&pvrdma_device_list_lock); + + kfree(netdev_work); +} + +static int pvrdma_netdevice_event(struct notifier_block *this, + unsigned long event, void *ptr) +{ + struct net_device *event_netdev = netdev_notifier_info_to_dev(ptr); + struct pvrdma_netdevice_work *netdev_work; + + netdev_work = kmalloc(sizeof(*netdev_work), GFP_ATOMIC); + if (!netdev_work) + return NOTIFY_BAD; + + INIT_WORK(&netdev_work->work, pvrdma_netdevice_event_work); + netdev_work->event_netdev = event_netdev; + netdev_work->event = event; + queue_work(event_wq, &netdev_work->work); + + return NOTIFY_DONE; +} + +static int pvrdma_pci_probe(struct pci_dev *pdev, + const struct pci_device_id *id) +{ + struct pci_dev *pdev_net; + struct pvrdma_dev *dev; + int ret; + unsigned long start; + unsigned long len; + unsigned int version; + dma_addr_t slot_dma = 0; + + dev_dbg(&pdev->dev, "initializing driver %s\n", pci_name(pdev)); + + /* Allocate zero-out device */ + dev = (struct pvrdma_dev *)ib_alloc_device(sizeof(*dev)); + if (!dev) { + dev_err(&pdev->dev, "failed to allocate IB device\n"); + return -ENOMEM; + } + + mutex_lock(&pvrdma_device_list_lock); + list_add(&dev->device_link, &pvrdma_device_list); + mutex_unlock(&pvrdma_device_list_lock); + + ret = pvrdma_init_device(dev); + if (ret) + goto err_free_device; + + dev->pdev = pdev; + pci_set_drvdata(pdev, dev); + + ret = pci_enable_device(pdev); + if (ret) { + dev_err(&pdev->dev, "cannot enable PCI device\n"); + goto err_free_device; + } + + /* + * DEBUG: Checking PCI: 0x20200; _MEM; len=0x2000 + */ + dev_dbg(&pdev->dev, "PCI resource flags BAR0 %#lx\n", + pci_resource_flags(pdev, 0)); + dev_dbg(&pdev->dev, "PCI resource len %#llx\n", + (unsigned long long)pci_resource_len(pdev, 0)); + dev_dbg(&pdev->dev, "PCI resource start %#llx\n", + (unsigned long long)pci_resource_start(pdev, 0)); + dev_dbg(&pdev->dev, "PCI resource flags BAR1 %#lx\n", + pci_resource_flags(pdev, 1)); + dev_dbg(&pdev->dev, "PCI resource len %#llx\n", + (unsigned long long)pci_resource_len(pdev, 1)); + dev_dbg(&pdev->dev, "PCI resource start %#llx\n", + (unsigned long long)pci_resource_start(pdev, 1)); + + if (!(pci_resource_flags(pdev, 0) & IORESOURCE_MEM) || + !(pci_resource_flags(pdev, 1) & IORESOURCE_MEM)) { + dev_err(&pdev->dev, "PCI BAR region not MMIO\n"); + ret = -ENOMEM; + goto err_free_device; + } + + ret = pci_request_regions(pdev, DRV_NAME); + if (ret) { + dev_err(&pdev->dev, "cannot request PCI resources\n"); + goto err_disable_pdev; + } + + /* Enable 64-Bit DMA */ + if (pci_set_dma_mask(pdev, DMA_BIT_MASK(64)) == 0) { + ret = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)); + if (ret != 0) { + dev_err(&pdev->dev, + "pci_set_consistent_dma_mask failed\n"); + goto err_free_resource; + } + } else { + ret = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); + if (ret != 0) { + dev_err(&pdev->dev, + "pci_set_dma_mask failed\n"); + goto err_free_resource; + } + } + + pci_set_master(pdev); + + /* Map register space */ + start = pci_resource_start(dev->pdev, PVRDMA_PCI_RESOURCE_REG); + len = pci_resource_len(dev->pdev, PVRDMA_PCI_RESOURCE_REG); + dev->regs = ioremap(start, len); + if (!dev->regs) { + dev_err(&pdev->dev, "register mapping failed\n"); + ret = -ENOMEM; + goto err_free_resource; + } + + /* Setup per-device UAR. */ + dev->driver_uar.index = 0; + dev->driver_uar.pfn = + pci_resource_start(dev->pdev, PVRDMA_PCI_RESOURCE_UAR) >> + PAGE_SHIFT; + dev->driver_uar.map = + ioremap(dev->driver_uar.pfn << PAGE_SHIFT, PAGE_SIZE); + if (!dev->driver_uar.map) { + dev_err(&pdev->dev, "failed to remap UAR pages\n"); + ret = -ENOMEM; + goto err_unmap_regs; + } + + version = pvrdma_read_reg(dev, PVRDMA_REG_VERSION); + dev_info(&pdev->dev, "device version %d, driver version %d\n", + version, PVRDMA_VERSION); + if (version < PVRDMA_VERSION) { + dev_err(&pdev->dev, "incompatible device version\n"); + goto err_uar_unmap; + } + + dev->dsr = dma_alloc_coherent(&pdev->dev, sizeof(*dev->dsr), + &dev->dsrbase, GFP_KERNEL); + if (!dev->dsr) { + dev_err(&pdev->dev, "failed to allocate shared region\n"); + ret = -ENOMEM; + goto err_uar_unmap; + } + + /* Setup the shared region */ + memset(dev->dsr, 0, sizeof(*dev->dsr)); + dev->dsr->driver_version = PVRDMA_VERSION; + dev->dsr->gos_info.gos_bits = sizeof(void *) == 4 ? + PVRDMA_GOS_BITS_32 : + PVRDMA_GOS_BITS_64; + dev->dsr->gos_info.gos_type = PVRDMA_GOS_TYPE_LINUX; + dev->dsr->gos_info.gos_ver = 1; + dev->dsr->uar_pfn = dev->driver_uar.pfn; + + /* Command slot. */ + dev->cmd_slot = dma_alloc_coherent(&pdev->dev, PAGE_SIZE, + &slot_dma, GFP_KERNEL); + if (!dev->cmd_slot) { + ret = -ENOMEM; + goto err_free_dsr; + } + + dev->dsr->cmd_slot_dma = (u64)slot_dma; + + /* Response slot. */ + dev->resp_slot = dma_alloc_coherent(&pdev->dev, PAGE_SIZE, + &slot_dma, GFP_KERNEL); + if (!dev->resp_slot) { + ret = -ENOMEM; + goto err_free_slots; + } + + dev->dsr->resp_slot_dma = (u64)slot_dma; + + /* Async event ring */ + dev->dsr->async_ring_pages.num_pages = 4; + ret = pvrdma_page_dir_init(dev, &dev->async_pdir, + dev->dsr->async_ring_pages.num_pages, true); + if (ret) + goto err_free_slots; + dev->async_ring_state = dev->async_pdir.pages[0]; + dev->dsr->async_ring_pages.pdir_dma = dev->async_pdir.dir_dma; + + /* CQ notification ring */ + dev->dsr->cq_ring_pages.num_pages = 4; + ret = pvrdma_page_dir_init(dev, &dev->cq_pdir, + dev->dsr->cq_ring_pages.num_pages, true); + if (ret) + goto err_free_async_ring; + dev->cq_ring_state = dev->cq_pdir.pages[0]; + dev->dsr->cq_ring_pages.pdir_dma = dev->cq_pdir.dir_dma; + + /* + * Write the PA of the shared region to the device. The writes must be + * ordered such that the high bits are written last. When the writes + * complete, the device will have filled out the capabilities. + */ + + pvrdma_write_reg(dev, PVRDMA_REG_DSRLOW, (u32)dev->dsrbase); + pvrdma_write_reg(dev, PVRDMA_REG_DSRHIGH, + (u32)((u64)(dev->dsrbase) >> 32)); + + /* Make sure the write is complete before reading status. */ + mb(); + + /* Currently, the driver only supports RoCE mode. */ + if (dev->dsr->caps.mode != PVRDMA_DEVICE_MODE_ROCE) { + dev_err(&pdev->dev, "unsupported transport %d\n", + dev->dsr->caps.mode); + ret = -EINVAL; + goto err_free_cq_ring; + } + + /* Currently, the driver only supports RoCE V1. */ + if (!(dev->dsr->caps.gid_types & PVRDMA_GID_TYPE_FLAG_ROCE_V1)) { + dev_err(&pdev->dev, "driver needs RoCE v1 support\n"); + ret = -EINVAL; + goto err_free_cq_ring; + } + + /* Paired vmxnet3 will have same bus, slot. But func will be 0 */ + pdev_net = pci_get_slot(pdev->bus, PCI_DEVFN(PCI_SLOT(pdev->devfn), 0)); + if (!pdev_net) { + dev_err(&pdev->dev, "failed to find paired net device\n"); + ret = -ENODEV; + goto err_free_cq_ring; + } + + if (pdev_net->vendor != PCI_VENDOR_ID_VMWARE || + pdev_net->device != PCI_DEVICE_ID_VMWARE_VMXNET3) { + dev_err(&pdev->dev, "failed to find paired vmxnet3 device\n"); + pci_dev_put(pdev_net); + ret = -ENODEV; + goto err_free_cq_ring; + } + + dev->netdev = pci_get_drvdata(pdev_net); + pci_dev_put(pdev_net); + if (!dev->netdev) { + dev_err(&pdev->dev, "failed to get vmxnet3 device\n"); + ret = -ENODEV; + goto err_free_cq_ring; + } + + dev_info(&pdev->dev, "paired device to %s\n", dev->netdev->name); + + /* Interrupt setup */ + ret = pvrdma_alloc_intrs(dev); + if (ret) { + dev_err(&pdev->dev, "failed to allocate interrupts\n"); + ret = -ENOMEM; + goto err_netdevice; + } + + /* Allocate UAR table. */ + ret = pvrdma_uar_table_init(dev); + if (ret) { + dev_err(&pdev->dev, "failed to allocate UAR table\n"); + ret = -ENOMEM; + goto err_free_intrs; + } + + /* Allocate GID table */ + dev->sgid_tbl = kcalloc(dev->dsr->caps.gid_tbl_len, + sizeof(union ib_gid), GFP_KERNEL); + if (!dev->sgid_tbl) { + ret = -ENOMEM; + goto err_free_uar_table; + } + dev_dbg(&pdev->dev, "gid table len %d\n", dev->dsr->caps.gid_tbl_len); + + pvrdma_enable_intrs(dev); + + /* Activate pvrdma device */ + pvrdma_write_reg(dev, PVRDMA_REG_CTL, PVRDMA_DEVICE_CTL_ACTIVATE); + + /* Make sure the write is complete before reading status. */ + mb(); + + /* Check if device was successfully activated */ + ret = pvrdma_read_reg(dev, PVRDMA_REG_ERR); + if (ret != 0) { + dev_err(&pdev->dev, "failed to activate device\n"); + ret = -EINVAL; + goto err_disable_intr; + } + + /* Register IB device */ + ret = pvrdma_register_device(dev); + if (ret) { + dev_err(&pdev->dev, "failed to register IB device\n"); + goto err_disable_intr; + } + + dev->nb_netdev.notifier_call = pvrdma_netdevice_event; + ret = register_netdevice_notifier(&dev->nb_netdev); + if (ret) { + dev_err(&pdev->dev, "failed to register netdevice events\n"); + goto err_unreg_ibdev; + } + + dev_info(&pdev->dev, "attached to device\n"); + return 0; + +err_unreg_ibdev: + ib_unregister_device(&dev->ib_dev); +err_disable_intr: + pvrdma_disable_intrs(dev); + kfree(dev->sgid_tbl); +err_free_uar_table: + pvrdma_uar_table_cleanup(dev); +err_free_intrs: + pvrdma_free_irq(dev); + pvrdma_disable_msi_all(dev); +err_netdevice: + unregister_netdevice_notifier(&dev->nb_netdev); +err_free_cq_ring: + pvrdma_page_dir_cleanup(dev, &dev->cq_pdir); +err_free_async_ring: + pvrdma_page_dir_cleanup(dev, &dev->async_pdir); +err_free_slots: + pvrdma_free_slots(dev); +err_free_dsr: + dma_free_coherent(&pdev->dev, sizeof(*dev->dsr), dev->dsr, + dev->dsrbase); +err_uar_unmap: + iounmap(dev->driver_uar.map); +err_unmap_regs: + iounmap(dev->regs); +err_free_resource: + pci_release_regions(pdev); +err_disable_pdev: + pci_disable_device(pdev); + pci_set_drvdata(pdev, NULL); +err_free_device: + mutex_lock(&pvrdma_device_list_lock); + list_del(&dev->device_link); + mutex_unlock(&pvrdma_device_list_lock); + ib_dealloc_device(&dev->ib_dev); + return ret; +} + +static void pvrdma_pci_remove(struct pci_dev *pdev) +{ + struct pvrdma_dev *dev = pci_get_drvdata(pdev); + + if (!dev) + return; + + dev_info(&pdev->dev, "detaching from device\n"); + + unregister_netdevice_notifier(&dev->nb_netdev); + dev->nb_netdev.notifier_call = NULL; + + flush_workqueue(event_wq); + + /* Unregister ib device */ + ib_unregister_device(&dev->ib_dev); + + mutex_lock(&pvrdma_device_list_lock); + list_del(&dev->device_link); + mutex_unlock(&pvrdma_device_list_lock); + + pvrdma_disable_intrs(dev); + pvrdma_free_irq(dev); + pvrdma_disable_msi_all(dev); + + /* Deactivate pvrdma device */ + pvrdma_write_reg(dev, PVRDMA_REG_CTL, PVRDMA_DEVICE_CTL_RESET); + pvrdma_page_dir_cleanup(dev, &dev->cq_pdir); + pvrdma_page_dir_cleanup(dev, &dev->async_pdir); + pvrdma_free_slots(dev); + + iounmap(dev->regs); + kfree(dev->sgid_tbl); + kfree(dev->cq_tbl); + kfree(dev->qp_tbl); + pvrdma_uar_table_cleanup(dev); + iounmap(dev->driver_uar.map); + + ib_dealloc_device(&dev->ib_dev); + + /* Free pci resources */ + pci_release_regions(pdev); + pci_disable_device(pdev); + pci_set_drvdata(pdev, NULL); +} + +static struct pci_device_id pvrdma_pci_table[] = { + { PCI_DEVICE(PCI_VENDOR_ID_VMWARE, PCI_DEVICE_ID_VMWARE_PVRDMA), }, + { 0 }, +}; + +MODULE_DEVICE_TABLE(pci, pvrdma_pci_table); + +static struct pci_driver pvrdma_driver = { + .name = DRV_NAME, + .id_table = pvrdma_pci_table, + .probe = pvrdma_pci_probe, + .remove = pvrdma_pci_remove, +}; + +static int __init pvrdma_init(void) +{ + int err; + + event_wq = alloc_ordered_workqueue("pvrdma_event_wq", WQ_MEM_RECLAIM); + if (!event_wq) + return -ENOMEM; + + err = pci_register_driver(&pvrdma_driver); + if (err) + destroy_workqueue(event_wq); + + return err; +} + +static void __exit pvrdma_cleanup(void) +{ + pci_unregister_driver(&pvrdma_driver); + + destroy_workqueue(event_wq); +} + +module_init(pvrdma_init); +module_exit(pvrdma_cleanup); + +MODULE_AUTHOR("VMware, Inc"); +MODULE_DESCRIPTION("VMware Virtual RDMA/InfiniBand driver"); +MODULE_VERSION(DRV_VERSION); +MODULE_LICENSE("Dual BSD/GPL");