From patchwork Sat Sep 24 23:21:37 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Adit Ranadive X-Patchwork-Id: 9349357 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 DA019601C2 for ; Sat, 24 Sep 2016 23:23:23 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id C8135290C3 for ; Sat, 24 Sep 2016 23:23:23 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id BC079290CE; Sat, 24 Sep 2016 23:23:23 +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=ham 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 B66C9290C3 for ; Sat, 24 Sep 2016 23:23:21 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1033474AbcIXXW7 (ORCPT ); Sat, 24 Sep 2016 19:22:59 -0400 Received: from ex13-edg-ou-002.vmware.com ([208.91.0.190]:3081 "EHLO EX13-EDG-OU-002.vmware.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1030485AbcIXXWg (ORCPT ); Sat, 24 Sep 2016 19:22:36 -0400 Received: from sc9-mailhost3.vmware.com (10.113.161.73) by EX13-EDG-OU-002.vmware.com (10.113.208.156) with Microsoft SMTP Server id 15.0.1156.6; Sat, 24 Sep 2016 16:22:09 -0700 Received: from EX13-CAS-001.vmware.com (smtp-inbound.vmware.com [10.113.191.51]) by sc9-mailhost3.vmware.com (Postfix) with ESMTP id EC5B0404B5; Sat, 24 Sep 2016 16:22:32 -0700 (PDT) Received: from EX13-CAS-004.vmware.com (10.113.191.54) by EX13-MBX-009.vmware.com (10.113.191.29) with Microsoft SMTP Server (TLS) id 15.0.1156.6; Sat, 24 Sep 2016 16:22:32 -0700 Received: from NAM03-CO1-obe.outbound.protection.outlook.com (10.113.170.11) by EX13-CAS-004.vmware.com (10.113.191.54) with Microsoft SMTP Server (TLS) id 15.0.1156.6 via Frontend Transport; Sat, 24 Sep 2016 16:22:32 -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=yBhBpTQLbfCBMngU2+o9IDseVVTKErgSa/CU7OsZAAI=; b=frb59hd64gNvo9h9gLzbCVHdXQJmIKkfAg/Srqwt/W/oLbd6rXko8HWZZnGxO3rI9rhWqe8Ix0FnL9oTptiJ7pWZbxdGELF3llEUHGCO+waE17MVHv+PtE6eklvyBavPVl+euvE94nOGuRdfHHcUbdkiEIa6v2mjttczifpC9e4= Authentication-Results: spf=none (sender IP is ) smtp.mailfrom=aditr@vmware.com; Received: from promb-2s-dhcp95-136.eng.vmware.com (208.91.1.34) by CO2PR0501MB837.namprd05.prod.outlook.com (10.141.244.147) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA384_P384) id 15.1.639.2; Sat, 24 Sep 2016 23:22:28 +0000 From: Adit Ranadive To: , , CC: Adit Ranadive , , , , , , Subject: [PATCH v5 13/16] IB/pvrdma: Add the main driver module for PVRDMA Date: Sat, 24 Sep 2016 16:21:37 -0700 Message-ID: <8db982c37a0a96b37df02826564474a216922f25.1474759181.git.aditr@vmware.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: References: MIME-Version: 1.0 X-Originating-IP: [208.91.1.34] X-ClientProxiedBy: CY1PR18CA0027.namprd18.prod.outlook.com (10.163.31.37) To CO2PR0501MB837.namprd05.prod.outlook.com (10.141.244.147) X-MS-Office365-Filtering-Correlation-Id: 8082a0cd-4dd7-42bf-d743-08d3e4d1a72e X-Microsoft-Exchange-Diagnostics: 1; CO2PR0501MB837; 2:PJzaHZmIVN0UIpJI+PRio0UfA4e2s+nLP3wslAfgaWWQyOVdqWT4ySzzayqN6I18KNTQqbhxn9Jpk+AhL9PmpVdhtYislHk17u/fOBaJsJEixJTZE+0WzA/TDIIsq/4cXf4U89hx0m0lM9MzqCiqwRTUVZ/nniLO37eTnC2aNVZEZEeGc7xv6oM7MOQTcrRX; 3:eQWCiTvIFV9U7L7DYaz5t6RKYV1oCCxxHb7a73Y5fXvFh9WRdAKTdF2kBG+DoM4b6ZO5B7eOQedtI9QqKUXEQB3lF7TZzc+oISM8yBGeBqRvS8S1Oe65w3M1oE3hHKEs; 25:USb0FgmS/FTCy0d2jFCS3WU6UjhiSmMJTgcloN4/aAFUkBmIzY8VtAEsUlroKfa8Jq0H0xLRQx6Zu1l29AUdlb+mm8ritSOczIrrVB8roKweXb6LTHxQFBnJs4kiaYZKvDx2mlyv2wEBA1ufi3fIss2bOYDSZUI+Cd+HxL9xPkrKeaRT6wv0pYPuTXIbREUOG7rXwDc2vASt7zKwposLLjpVA5P8RDrkMUG39w1fp7wgMjgopkp/zumLFqNpNn4wVK7XkaJItXDbBEiB/xS7rbz3su4b9sypKKv0vUhanGVWWEIEK4MW5OWRmF++it8k2LpGvWaxFiCpNWzGS5hyF6CKCvlTXJW80uQn0Yw/sQx+ywbCp56IHTC/Pjs4sN7gMak/v1+lIFH+XgmXjeHAwWFNub2RhWYKRbqicG1ypxA= X-Microsoft-Antispam: UriScan:;BCL:0;PCL:0;RULEID:;SRVR:CO2PR0501MB837; X-LD-Processed: b39138ca-3cee-4b4a-a4d6-cd83d9dd62f0,ExtAddr X-Microsoft-Exchange-Diagnostics: 1; CO2PR0501MB837; 31:ppvPSuI7se9X06aJgAy47CbytmRNEYk+cfR3FoPYIClFNiGZPvDOGao1u8HcaMK6F3HiFEcH3SV1mcKnPW/0W1ZPfLk5rBv/G/DmQVT7NStPkne6bckCvRtCSohTgI8/bJYHOWdGJtdh7jR+XJNGCwahWRnXPLlqxbp15JQLsofJyNloT+2rFbmSyVVlNqnwhLQS9+pNS6Xs2ZYTlj0fEFZUvkRHCT6rxlgEY82RLLg=; 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)(8121501046)(5005006)(3002001)(10201501046); SRVR:CO2PR0501MB837; BCL:0; PCL:0; RULEID:; SRVR:CO2PR0501MB837; X-Microsoft-Exchange-Diagnostics: 1; CO2PR0501MB837; 4:IjuopBS8cFx6nw9XKS7b800eickV5zbTmtGzio8odkyD5jgfseWOATv6Y511oAL7Z9aY+AZRMkO/Pd1vsrEQ+m9p/grS2Yk8ZCb3ocF/xsJ2ZtklOl0qOmPGCa0m4zcscLgPjPycjXU6+tjxrpISNj0n0Go6/ZJTYZfHIUxt2I6LunRh7y91FyuKeUJKIFaUuX9ppxNDDXtuwjbMTIAalrgUGY5GpQxO1Grkl5VnJjoUB5hDV4KnL1A0ZcipAvMZ4FOrUzFyrb0CMngkHzMtQLY9pT8Rh2zciHtLUYWSkPoQntmEde0bx0rtp91QZXqMFJlRv7ajT+wAfdH1ZNetrip04XD+X0WFST8UlABVDKQEUNWvgwIyQobTp/4UdNwSoljZE9oTs66hPGJo1Q/4M8cF/fdV5WtDuX7Td8D3IBPf4fA6lUE2J/i8CgzuZ/tBii4B8dUZL/U3FREPRC0jAu4gR6eQutxJliURY/CqfFFm+JLsvXq3j5EEmwGR3XvyUw37PCD3PJgy46mX4+liv5++Ij/qhzaTNcqJu66iECc= X-Forefront-PRVS: 0075CB064E X-Forefront-Antispam-Report: SFV:NSPM; SFS:(10009020)(4630300001)(6009001)(7916002)(189002)(199003)(107886002)(77096005)(97736004)(5890100001)(4326007)(106356001)(105586002)(15975445007)(76176999)(586003)(6862003)(118296001)(50986999)(2950100002)(33646002)(6116002)(5001770100001)(19580395003)(5660300001)(19580405001)(92566002)(2906002)(3846002)(47776003)(189998001)(50466002)(86362001)(48376002)(2201001)(81156014)(81166006)(101416001)(8676002)(50226002)(5003940100001)(36756003)(7736002)(7846002)(305945005)(229853001)(42186005)(68736007)(66066001)(4001430100002)(7099028)(2004002); DIR:OUT; SFP:1101; SCL:1; SRVR:CO2PR0501MB837; H:promb-2s-dhcp95-136.eng.vmware.com; FPR:; SPF:None; PTR:InfoNoRecords; MX:1; A:1; LANG:en; Received-SPF: None (EX13-EDG-OU-002.vmware.com: aditr@vmware.com does not designate permitted sender hosts) Received-SPF: None (protection.outlook.com: vmware.com does not designate permitted sender hosts) X-Microsoft-Exchange-Diagnostics: =?us-ascii?Q?1; CO2PR0501MB837; 23:w6+Eiw+gUWTtnMbFNBjn0joUDm6S51tI8XQnAKiH?= =?us-ascii?Q?YWWUk9TSBTy84HtvOG+0DooJQ25ueKjHxYaKGkXjxZECe6ORYDZb11D5yVY2?= =?us-ascii?Q?b14nO9s1QfwoNgdMFqcSYAwZyhenS3Pf4XNiGg67av04L7ZxLRrrVGmyyxLS?= =?us-ascii?Q?RK3ck09bwz+5bIFG0irzZYPkTigFPDOlhzYOZeY5jtNOWrLgO/TLVgvom6fG?= =?us-ascii?Q?FXiEtT4LLRfMq+sF5Ptz7ucvVPUUlhmDCgLDLyvt/nYAmaZJVcqKqgHeKNLc?= =?us-ascii?Q?zRDk+GYxoN4HecDWDbWamj3ZPb2hoLTbKXva/OjcDAYzyd98EvaHsnYv5jnR?= =?us-ascii?Q?z0VUgvU3McGf3tRyWNtU0gxaQqtwa6DPHRDDmlPve6FCuG4WPyFu0lWblvem?= =?us-ascii?Q?ta9cQHShWAM5e9qAkN7vs7N0VyShkvVcZsSX5CgMmHoG5ATOiRB+d/WlVSYh?= =?us-ascii?Q?3x6OD6fG2Pwl0w5NhBPyiOSBqKhPOLH9MrIsx7Y6VfW25CLX093hbcbUQUIS?= =?us-ascii?Q?FY0evQ+heup2HleD2QZP6zabOY34Rd7zeaFiObR6LXsuhDOIj18Ad7v3Qhsr?= =?us-ascii?Q?H9ms1bECTfjzwhGgzTm13fHbgiRVWoahYC7aM1l45T004h03Sl4UUCV/P0iF?= =?us-ascii?Q?MWCSa3RCIwQRVT+xloySluEOz0KIEFo6KW3zm+JFdww7MKvf7gt5DOh+7Yoq?= =?us-ascii?Q?+doVyftxr78zZc2yEotARGQH2WZwTKbLoDbO5t53iI0pWsc68Y/F6KCZPi+W?= =?us-ascii?Q?QBdgLXSfKXGMcmWzHkJYd+duYQ2fVwE3Hk51kSfbZHhm5xQl17iw78Cvz+px?= =?us-ascii?Q?Y/I2KEtMAydK3WQ0nmDJcLcYzSl01Kq6733a+9Ih6RGdHWyKQ9uWL9upgnZ0?= =?us-ascii?Q?+s97Ov5k3xHFq6eGdRuMTRcUFIb+cDdqRiVOfM+hMMuZVDlEd6NZl+GwvEi9?= =?us-ascii?Q?Al2Werss7h7fT2Ksz3SNKwNZPMUUpSeaPZU94EdLlCS2Tw5qbuLFcjPtNQjH?= =?us-ascii?Q?sdj6kp/4exz+EvqbDqcKfjjBOSAvHSTdoDGe3mZJ1iixoLUTnxv1eVHoyNAb?= =?us-ascii?Q?1XrXUBjPNc/B5/JBcaAFUftaKAqxkAkp7xTw4PJyjqT6L2dl5PZQQB08y/yq?= =?us-ascii?Q?70h0B2zHvJTtribGCU0vBgltbq0xiAyRB6LSZ3JJjDPADA0FstznOjf01f2Z?= =?us-ascii?Q?+GQuec4RUdpLjmrO5AyA78OTNKPva9xnGFyK985xkokTRW2MmzFoglbuDlyy?= =?us-ascii?Q?q28iFcg69y5gdLbRT00=3D?= X-Microsoft-Exchange-Diagnostics: 1; CO2PR0501MB837; 6:k32av8nPK4HUwYSEmsH7ctoIcgodVZDt14OSEg5qiw4CrpOF9HGQrjXjokb4SfCGsQYeCX0sy1Rcpv7et3v7lColCUCO+2bFocmzpyK2gaEhRFbl2UvbIllb+TZu1P9nM0/lPduawkKly0nBAudOi+srvevpu8NcGoIGKWgU/LAuco5VPoLp2AOpiAKd96BCTEoOJ03fNPJYhk7C28Werb8EYiBa8aQ/eJkotICTnbjN7UAd8msAQ9lLVMjukNkBJLR/jgjazJdr8BckLSzIdMEb3Juv1z+Y/J4/5+889X4=; 5:swr9+1mu3eJBQsBKwrjOaxMKDsoCfM1VrvnG1g5d7+CUJ8GKw+mVf+xqgoVXMtcbupoc9Vc8Hrw1HRPpV5BpICbZFepHztHVSAPPyxMLa0f03Cookzck+972ph+g1JqGoTsUHeumyldE8JIDiYnGtA==; 24:OPqiSCVpX7/0eIykFaLbU9CLjgW5IxULfHHOAblsJPs1PQSRgjWR2/bDL2ET0GUrLpZQyFUWNSuQpUg1PWh2ag94k4JWZPAFiwvMR4f5tOI=; 7:wybk2/cfQcIB2Z1g8HHi+TGpWlfNnk799alOzWjfXdrYwL34jZ33cMUa1FHTUIdOCprWTlgNcPh/e7GnP63FiT4eWPamK5hL6xHPGQsF6OAalG5z86XfR8b37vINhqQLhR9A7ruQkt+p6gCPTtT490ynFqgDPlpQjynuiaGCxpbS49F9A2LATi5y9i5c/Ycj1qu1+Z6dCWInC4g+h6jXCQ88eCBOoi99dBt4bIZCw2tT/p8ftZhmUNOKPuWtA1sEnpwVvOnLqP2yc7zKKfbgttfQnX+BcIcWc4P2R0v44Zd5c+CYzjhicUYWkpgY2c4+ SpamDiagnosticOutput: 1:99 SpamDiagnosticMetadata: NSPM X-Microsoft-Exchange-Diagnostics: 1; CO2PR0501MB837; 20:AgVDROIBm6WGcc6QvlAibsblcjz9utlyBsRh77Z4X3uXx/BScZoO8kLYZ/52cxuBheZUdmMqU0L2uvA6LUKiHEAz7i5hwEfQrChpMHsMR5zM/IKcoIbE0YF+1kc+CN8UHIDXJgeyvUa469uZQk6NImkL1f63iGUPa60XDS6GMWc= X-MS-Exchange-CrossTenant-OriginalArrivalTime: 24 Sep 2016 23:22:28.1214 (UTC) X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted X-MS-Exchange-Transport-CrossTenantHeadersStamped: CO2PR0501MB837 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 v4->v5: - Removed two unnecessary lines. - Updated include for headers in UAPI folder. - Update to pvrdma_cmd_post for add/delete GIDs. - Add error code in dev_warn if pvrdma_cmd_post failed. 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 | 1220 ++++++++++++++++++++++++++++ 1 file changed, 1220 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..94cbbb9 --- /dev/null +++ b/drivers/infiniband/hw/pvrdma/pvrdma_main.c @@ -0,0 +1,1220 @@ +/* + * 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 + +#include "pvrdma.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, 0); + if (ret < 0) { + dev_warn(&dev->pdev->dev, + "could not create binding, error: %d\n", ret); + 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, 0); + if (ret < 0) { + dev_warn(&dev->pdev->dev, + "could not destroy binding, error: %d\n", ret); + 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) +{ + 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");