en_main.c 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398
  1. /*
  2. * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
  3. *
  4. * This software is available to you under a choice of one of two
  5. * licenses. You may choose to be licensed under the terms of the GNU
  6. * General Public License (GPL) Version 2, available from the file
  7. * COPYING in the main directory of this source tree, or the
  8. * OpenIB.org BSD license below:
  9. *
  10. * Redistribution and use in source and binary forms, with or
  11. * without modification, are permitted provided that the following
  12. * conditions are met:
  13. *
  14. * - Redistributions of source code must retain the above
  15. * copyright notice, this list of conditions and the following
  16. * disclaimer.
  17. *
  18. * - Redistributions in binary form must reproduce the above
  19. * copyright notice, this list of conditions and the following
  20. * disclaimer in the documentation and/or other materials
  21. * provided with the distribution.
  22. *
  23. * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  24. * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  25. * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  26. * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
  27. * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
  28. * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
  29. * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  30. * SOFTWARE.
  31. *
  32. */
  33. #include <linux/cpumask.h>
  34. #include <linux/module.h>
  35. #include <linux/delay.h>
  36. #include <linux/netdevice.h>
  37. #include <linux/slab.h>
  38. #include <linux/mlx4/driver.h>
  39. #include <linux/mlx4/device.h>
  40. #include <linux/mlx4/cmd.h>
  41. #include "mlx4_en.h"
  42. MODULE_AUTHOR("Liran Liss, Yevgeny Petrilin");
  43. MODULE_DESCRIPTION("Mellanox ConnectX HCA Ethernet driver");
  44. MODULE_LICENSE("Dual BSD/GPL");
  45. MODULE_VERSION(DRV_VERSION " ("DRV_RELDATE")");
  46. static const char mlx4_en_version[] =
  47. DRV_NAME ": Mellanox ConnectX HCA Ethernet driver v"
  48. DRV_VERSION " (" DRV_RELDATE ")\n";
  49. #define MLX4_EN_PARM_INT(X, def_val, desc) \
  50. static unsigned int X = def_val;\
  51. module_param(X , uint, 0444); \
  52. MODULE_PARM_DESC(X, desc);
  53. /*
  54. * Device scope module parameters
  55. */
  56. /* Enable RSS UDP traffic */
  57. MLX4_EN_PARM_INT(udp_rss, 1,
  58. "Enable RSS for incoming UDP traffic or disabled (0)");
  59. /* Priority pausing */
  60. MLX4_EN_PARM_INT(pfctx, 0, "Priority based Flow Control policy on TX[7:0]."
  61. " Per priority bit mask");
  62. MLX4_EN_PARM_INT(pfcrx, 0, "Priority based Flow Control policy on RX[7:0]."
  63. " Per priority bit mask");
  64. MLX4_EN_PARM_INT(inline_thold, MAX_INLINE,
  65. "Threshold for using inline data (range: 17-104, default: 104)");
  66. #define MAX_PFC_TX 0xff
  67. #define MAX_PFC_RX 0xff
  68. void en_print(const char *level, const struct mlx4_en_priv *priv,
  69. const char *format, ...)
  70. {
  71. va_list args;
  72. struct va_format vaf;
  73. va_start(args, format);
  74. vaf.fmt = format;
  75. vaf.va = &args;
  76. if (priv->registered)
  77. printk("%s%s: %s: %pV",
  78. level, DRV_NAME, priv->dev->name, &vaf);
  79. else
  80. printk("%s%s: %s: Port %d: %pV",
  81. level, DRV_NAME, dev_name(&priv->mdev->pdev->dev),
  82. priv->port, &vaf);
  83. va_end(args);
  84. }
  85. void mlx4_en_update_loopback_state(struct net_device *dev,
  86. netdev_features_t features)
  87. {
  88. struct mlx4_en_priv *priv = netdev_priv(dev);
  89. if (features & NETIF_F_LOOPBACK)
  90. priv->ctrl_flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK);
  91. else
  92. priv->ctrl_flags &= cpu_to_be32(~MLX4_WQE_CTRL_FORCE_LOOPBACK);
  93. priv->flags &= ~(MLX4_EN_FLAG_RX_FILTER_NEEDED|
  94. MLX4_EN_FLAG_ENABLE_HW_LOOPBACK);
  95. /* Drop the packet if SRIOV is not enabled
  96. * and not performing the selftest or flb disabled
  97. */
  98. if (mlx4_is_mfunc(priv->mdev->dev) &&
  99. !(features & NETIF_F_LOOPBACK) && !priv->validate_loopback)
  100. priv->flags |= MLX4_EN_FLAG_RX_FILTER_NEEDED;
  101. /* Set dmac in Tx WQE if we are in SRIOV mode or if loopback selftest
  102. * is requested
  103. */
  104. if (mlx4_is_mfunc(priv->mdev->dev) || priv->validate_loopback)
  105. priv->flags |= MLX4_EN_FLAG_ENABLE_HW_LOOPBACK;
  106. mutex_lock(&priv->mdev->state_lock);
  107. if (priv->mdev->dev->caps.flags2 &
  108. MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB &&
  109. priv->rss_map.indir_qp.qpn) {
  110. int i;
  111. int err = 0;
  112. int loopback = !!(features & NETIF_F_LOOPBACK);
  113. for (i = 0; i < priv->rx_ring_num; i++) {
  114. int ret;
  115. ret = mlx4_en_change_mcast_lb(priv,
  116. &priv->rss_map.qps[i],
  117. loopback);
  118. if (!err)
  119. err = ret;
  120. }
  121. if (err)
  122. mlx4_warn(priv->mdev, "failed to change mcast loopback\n");
  123. }
  124. mutex_unlock(&priv->mdev->state_lock);
  125. }
  126. static int mlx4_en_get_profile(struct mlx4_en_dev *mdev)
  127. {
  128. struct mlx4_en_profile *params = &mdev->profile;
  129. int i;
  130. params->udp_rss = udp_rss;
  131. params->num_tx_rings_p_up = mlx4_low_memory_profile() ?
  132. MLX4_EN_MIN_TX_RING_P_UP :
  133. min_t(int, num_online_cpus(), MLX4_EN_MAX_TX_RING_P_UP);
  134. if (params->udp_rss && !(mdev->dev->caps.flags
  135. & MLX4_DEV_CAP_FLAG_UDP_RSS)) {
  136. mlx4_warn(mdev, "UDP RSS is not supported on this device\n");
  137. params->udp_rss = 0;
  138. }
  139. for (i = 1; i <= MLX4_MAX_PORTS; i++) {
  140. params->prof[i].rx_pause = !(pfcrx || pfctx);
  141. params->prof[i].rx_ppp = pfcrx;
  142. params->prof[i].tx_pause = !(pfcrx || pfctx);
  143. params->prof[i].tx_ppp = pfctx;
  144. params->prof[i].tx_ring_size = MLX4_EN_DEF_TX_RING_SIZE;
  145. params->prof[i].rx_ring_size = MLX4_EN_DEF_RX_RING_SIZE;
  146. params->prof[i].tx_ring_num = params->num_tx_rings_p_up *
  147. MLX4_EN_NUM_UP;
  148. params->prof[i].rss_rings = 0;
  149. params->prof[i].inline_thold = inline_thold;
  150. }
  151. return 0;
  152. }
  153. static void *mlx4_en_get_netdev(struct mlx4_dev *dev, void *ctx, u8 port)
  154. {
  155. struct mlx4_en_dev *endev = ctx;
  156. return endev->pndev[port];
  157. }
  158. static void mlx4_en_event(struct mlx4_dev *dev, void *endev_ptr,
  159. enum mlx4_dev_event event, unsigned long port)
  160. {
  161. struct mlx4_en_dev *mdev = (struct mlx4_en_dev *) endev_ptr;
  162. struct mlx4_en_priv *priv;
  163. switch (event) {
  164. case MLX4_DEV_EVENT_PORT_UP:
  165. case MLX4_DEV_EVENT_PORT_DOWN:
  166. if (!mdev->pndev[port])
  167. return;
  168. priv = netdev_priv(mdev->pndev[port]);
  169. /* To prevent races, we poll the link state in a separate
  170. task rather than changing it here */
  171. priv->link_state = event;
  172. queue_work(mdev->workqueue, &priv->linkstate_task);
  173. break;
  174. case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
  175. mlx4_err(mdev, "Internal error detected, restarting device\n");
  176. break;
  177. case MLX4_DEV_EVENT_SLAVE_INIT:
  178. case MLX4_DEV_EVENT_SLAVE_SHUTDOWN:
  179. break;
  180. default:
  181. if (port < 1 || port > dev->caps.num_ports ||
  182. !mdev->pndev[port])
  183. return;
  184. mlx4_warn(mdev, "Unhandled event %d for port %d\n", event,
  185. (int) port);
  186. }
  187. }
  188. static void mlx4_en_remove(struct mlx4_dev *dev, void *endev_ptr)
  189. {
  190. struct mlx4_en_dev *mdev = endev_ptr;
  191. int i;
  192. mutex_lock(&mdev->state_lock);
  193. mdev->device_up = false;
  194. mutex_unlock(&mdev->state_lock);
  195. mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
  196. if (mdev->pndev[i])
  197. mlx4_en_destroy_netdev(mdev->pndev[i]);
  198. flush_workqueue(mdev->workqueue);
  199. destroy_workqueue(mdev->workqueue);
  200. (void) mlx4_mr_free(dev, &mdev->mr);
  201. iounmap(mdev->uar_map);
  202. mlx4_uar_free(dev, &mdev->priv_uar);
  203. mlx4_pd_free(dev, mdev->priv_pdn);
  204. if (mdev->nb.notifier_call)
  205. unregister_netdevice_notifier(&mdev->nb);
  206. kfree(mdev);
  207. }
  208. static void mlx4_en_activate(struct mlx4_dev *dev, void *ctx)
  209. {
  210. int i;
  211. struct mlx4_en_dev *mdev = ctx;
  212. /* Create a netdev for each port */
  213. mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) {
  214. mlx4_info(mdev, "Activating port:%d\n", i);
  215. if (mlx4_en_init_netdev(mdev, i, &mdev->profile.prof[i]))
  216. mdev->pndev[i] = NULL;
  217. }
  218. /* register notifier */
  219. mdev->nb.notifier_call = mlx4_en_netdev_event;
  220. if (register_netdevice_notifier(&mdev->nb)) {
  221. mdev->nb.notifier_call = NULL;
  222. mlx4_err(mdev, "Failed to create notifier\n");
  223. }
  224. }
  225. static void *mlx4_en_add(struct mlx4_dev *dev)
  226. {
  227. struct mlx4_en_dev *mdev;
  228. int i;
  229. printk_once(KERN_INFO "%s", mlx4_en_version);
  230. mdev = kzalloc(sizeof(*mdev), GFP_KERNEL);
  231. if (!mdev)
  232. goto err_free_res;
  233. if (mlx4_pd_alloc(dev, &mdev->priv_pdn))
  234. goto err_free_dev;
  235. if (mlx4_uar_alloc(dev, &mdev->priv_uar))
  236. goto err_pd;
  237. mdev->uar_map = ioremap((phys_addr_t) mdev->priv_uar.pfn << PAGE_SHIFT,
  238. PAGE_SIZE);
  239. if (!mdev->uar_map)
  240. goto err_uar;
  241. spin_lock_init(&mdev->uar_lock);
  242. mdev->dev = dev;
  243. mdev->dma_device = &dev->persist->pdev->dev;
  244. mdev->pdev = dev->persist->pdev;
  245. mdev->device_up = false;
  246. mdev->LSO_support = !!(dev->caps.flags & (1 << 15));
  247. if (!mdev->LSO_support)
  248. mlx4_warn(mdev, "LSO not supported, please upgrade to later FW version to enable LSO\n");
  249. if (mlx4_mr_alloc(mdev->dev, mdev->priv_pdn, 0, ~0ull,
  250. MLX4_PERM_LOCAL_WRITE | MLX4_PERM_LOCAL_READ,
  251. 0, 0, &mdev->mr)) {
  252. mlx4_err(mdev, "Failed allocating memory region\n");
  253. goto err_map;
  254. }
  255. if (mlx4_mr_enable(mdev->dev, &mdev->mr)) {
  256. mlx4_err(mdev, "Failed enabling memory region\n");
  257. goto err_mr;
  258. }
  259. /* Build device profile according to supplied module parameters */
  260. if (mlx4_en_get_profile(mdev)) {
  261. mlx4_err(mdev, "Bad module parameters, aborting\n");
  262. goto err_mr;
  263. }
  264. /* Configure which ports to start according to module parameters */
  265. mdev->port_cnt = 0;
  266. mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
  267. mdev->port_cnt++;
  268. /* Set default number of RX rings*/
  269. mlx4_en_set_num_rx_rings(mdev);
  270. /* Create our own workqueue for reset/multicast tasks
  271. * Note: we cannot use the shared workqueue because of deadlocks caused
  272. * by the rtnl lock */
  273. mdev->workqueue = create_singlethread_workqueue("mlx4_en");
  274. if (!mdev->workqueue)
  275. goto err_mr;
  276. /* At this stage all non-port specific tasks are complete:
  277. * mark the card state as up */
  278. mutex_init(&mdev->state_lock);
  279. mdev->device_up = true;
  280. return mdev;
  281. err_mr:
  282. (void) mlx4_mr_free(dev, &mdev->mr);
  283. err_map:
  284. if (mdev->uar_map)
  285. iounmap(mdev->uar_map);
  286. err_uar:
  287. mlx4_uar_free(dev, &mdev->priv_uar);
  288. err_pd:
  289. mlx4_pd_free(dev, mdev->priv_pdn);
  290. err_free_dev:
  291. kfree(mdev);
  292. err_free_res:
  293. return NULL;
  294. }
  295. static struct mlx4_interface mlx4_en_interface = {
  296. .add = mlx4_en_add,
  297. .remove = mlx4_en_remove,
  298. .event = mlx4_en_event,
  299. .get_dev = mlx4_en_get_netdev,
  300. .protocol = MLX4_PROT_ETH,
  301. .activate = mlx4_en_activate,
  302. };
  303. static void mlx4_en_verify_params(void)
  304. {
  305. if (pfctx > MAX_PFC_TX) {
  306. pr_warn("mlx4_en: WARNING: illegal module parameter pfctx 0x%x - should be in range 0-0x%x, will be changed to default (0)\n",
  307. pfctx, MAX_PFC_TX);
  308. pfctx = 0;
  309. }
  310. if (pfcrx > MAX_PFC_RX) {
  311. pr_warn("mlx4_en: WARNING: illegal module parameter pfcrx 0x%x - should be in range 0-0x%x, will be changed to default (0)\n",
  312. pfcrx, MAX_PFC_RX);
  313. pfcrx = 0;
  314. }
  315. if (inline_thold < MIN_PKT_LEN || inline_thold > MAX_INLINE) {
  316. pr_warn("mlx4_en: WARNING: illegal module parameter inline_thold %d - should be in range %d-%d, will be changed to default (%d)\n",
  317. inline_thold, MIN_PKT_LEN, MAX_INLINE, MAX_INLINE);
  318. inline_thold = MAX_INLINE;
  319. }
  320. }
  321. static int __init mlx4_en_init(void)
  322. {
  323. mlx4_en_verify_params();
  324. mlx4_en_init_ptys2ethtool_map();
  325. return mlx4_register_interface(&mlx4_en_interface);
  326. }
  327. static void __exit mlx4_en_cleanup(void)
  328. {
  329. mlx4_unregister_interface(&mlx4_en_interface);
  330. }
  331. module_init(mlx4_en_init);
  332. module_exit(mlx4_en_cleanup);