2 * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
3 * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
5 * This software is available to you under a choice of one of two
6 * licenses. You may choose to be licensed under the terms of the GNU
7 * General Public License (GPL) Version 2, available from the file
8 * COPYING in the main directory of this source tree, or the
9 * OpenIB.org BSD license below:
11 * Redistribution and use in source and binary forms, with or
12 * without modification, are permitted provided that the following
15 * - Redistributions of source code must retain the above
16 * copyright notice, this list of conditions and the following
19 * - Redistributions in binary form must reproduce the above
20 * copyright notice, this list of conditions and the following
21 * disclaimer in the documentation and/or other materials
22 * provided with the distribution.
24 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
25 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
26 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
27 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
28 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
29 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
30 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
34 #include <linux/workqueue.h>
35 #include <linux/module.h>
40 MLX4_CATAS_POLL_INTERVAL = 5 * HZ,
45 int mlx4_internal_err_reset = 1;
46 module_param_named(internal_err_reset, mlx4_internal_err_reset, int, 0644);
47 MODULE_PARM_DESC(internal_err_reset,
48 "Reset device on internal errors if non-zero"
49 " (default 1, in SRIOV mode default is 0)");
51 static int read_vendor_id(struct mlx4_dev *dev)
56 ret = pci_read_config_word(dev->persist->pdev, 0, &vendor_id);
58 mlx4_err(dev, "Failed to read vendor ID, ret=%d\n", ret);
62 if (vendor_id == 0xffff) {
63 mlx4_err(dev, "PCI can't be accessed to read vendor id\n");
70 static int mlx4_reset_master(struct mlx4_dev *dev)
74 if (!pci_channel_offline(dev->persist->pdev)) {
75 err = read_vendor_id(dev);
76 /* If PCI can't be accessed to read vendor ID we assume that its
77 * link was disabled and chip was already reset.
82 err = mlx4_reset(dev);
84 mlx4_err(dev, "Fail to reset HCA\n");
90 void mlx4_enter_error_state(struct mlx4_dev_persistent *persist)
95 if (!mlx4_internal_err_reset)
98 mutex_lock(&persist->device_state_mutex);
99 if (persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
103 mlx4_err(dev, "device is going to be reset\n");
104 err = mlx4_reset_master(dev);
107 dev->persist->state |= MLX4_DEVICE_STATE_INTERNAL_ERROR;
108 mlx4_err(dev, "device was reset successfully\n");
109 mutex_unlock(&persist->device_state_mutex);
111 /* At that step HW was already reset, now notify clients */
112 mlx4_dispatch_event(dev, MLX4_DEV_EVENT_CATASTROPHIC_ERROR, 0);
113 mlx4_cmd_wake_completions(dev);
117 mutex_unlock(&persist->device_state_mutex);
120 static void mlx4_handle_error_state(struct mlx4_dev_persistent *persist)
124 mlx4_enter_error_state(persist);
125 mutex_lock(&persist->interface_state_mutex);
126 if (persist->interface_state & MLX4_INTERFACE_STATE_UP &&
127 !(persist->interface_state & MLX4_INTERFACE_STATE_DELETION)) {
128 err = mlx4_restart_one(persist->pdev);
129 mlx4_info(persist->dev, "mlx4_restart_one was ended, ret=%d\n",
132 mutex_unlock(&persist->interface_state_mutex);
135 static void dump_err_buf(struct mlx4_dev *dev)
137 struct mlx4_priv *priv = mlx4_priv(dev);
141 mlx4_err(dev, "Internal error detected:\n");
142 for (i = 0; i < priv->fw.catas_size; ++i)
143 mlx4_err(dev, " buf[%02x]: %08x\n",
144 i, swab32(readl(priv->catas_err.map + i)));
147 static void poll_catas(unsigned long dev_ptr)
149 struct mlx4_dev *dev = (struct mlx4_dev *) dev_ptr;
150 struct mlx4_priv *priv = mlx4_priv(dev);
152 if (readl(priv->catas_err.map)) {
157 if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
158 mlx4_warn(dev, "Internal error mark was detected on device\n");
162 mod_timer(&priv->catas_err.timer,
163 round_jiffies(jiffies + MLX4_CATAS_POLL_INTERVAL));
167 if (mlx4_internal_err_reset)
168 queue_work(dev->persist->catas_wq, &dev->persist->catas_work);
171 static void catas_reset(struct work_struct *work)
173 struct mlx4_dev_persistent *persist =
174 container_of(work, struct mlx4_dev_persistent,
177 mlx4_handle_error_state(persist);
180 void mlx4_start_catas_poll(struct mlx4_dev *dev)
182 struct mlx4_priv *priv = mlx4_priv(dev);
185 /*If we are in SRIOV the default of the module param must be 0*/
186 if (mlx4_is_mfunc(dev))
187 mlx4_internal_err_reset = 0;
189 INIT_LIST_HEAD(&priv->catas_err.list);
190 init_timer(&priv->catas_err.timer);
191 priv->catas_err.map = NULL;
193 addr = pci_resource_start(dev->persist->pdev, priv->fw.catas_bar) +
194 priv->fw.catas_offset;
196 priv->catas_err.map = ioremap(addr, priv->fw.catas_size * 4);
197 if (!priv->catas_err.map) {
198 mlx4_warn(dev, "Failed to map internal error buffer at 0x%llx\n",
199 (unsigned long long) addr);
203 priv->catas_err.timer.data = (unsigned long) dev;
204 priv->catas_err.timer.function = poll_catas;
205 priv->catas_err.timer.expires =
206 round_jiffies(jiffies + MLX4_CATAS_POLL_INTERVAL);
207 add_timer(&priv->catas_err.timer);
210 void mlx4_stop_catas_poll(struct mlx4_dev *dev)
212 struct mlx4_priv *priv = mlx4_priv(dev);
214 del_timer_sync(&priv->catas_err.timer);
216 if (priv->catas_err.map) {
217 iounmap(priv->catas_err.map);
218 priv->catas_err.map = NULL;
221 if (dev->persist->interface_state & MLX4_INTERFACE_STATE_DELETION)
222 flush_workqueue(dev->persist->catas_wq);
225 int mlx4_catas_init(struct mlx4_dev *dev)
227 INIT_WORK(&dev->persist->catas_work, catas_reset);
228 dev->persist->catas_wq = create_singlethread_workqueue("mlx4_health");
229 if (!dev->persist->catas_wq)
235 void mlx4_catas_end(struct mlx4_dev *dev)
237 if (dev->persist->catas_wq) {
238 destroy_workqueue(dev->persist->catas_wq);
239 dev->persist->catas_wq = NULL;