be_main.c 160.1 KB
Newer Older
S
Sathya Perla 已提交
1
/*
2
 * Copyright (C) 2005 - 2016 Broadcom
S
Sathya Perla 已提交
3 4 5 6 7 8 9 10
 * All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License version 2
 * as published by the Free Software Foundation.  The full GNU General
 * Public License is included in this distribution in the file called COPYING.
 *
 * Contact Information:
11
 * linux-drivers@emulex.com
S
Sathya Perla 已提交
12
 *
13 14 15
 * Emulex
 * 3333 Susan Street
 * Costa Mesa, CA 92626
S
Sathya Perla 已提交
16 17
 */

18
#include <linux/prefetch.h>
19
#include <linux/module.h>
S
Sathya Perla 已提交
20
#include "be.h"
21
#include "be_cmds.h"
22
#include <asm/div64.h>
S
Sathya Perla 已提交
23
#include <linux/aer.h>
24
#include <linux/if_bridge.h>
25
#include <net/busy_poll.h>
26
#include <net/vxlan.h>
S
Sathya Perla 已提交
27 28 29

MODULE_VERSION(DRV_VER);
MODULE_DESCRIPTION(DRV_DESC " " DRV_VER);
30
MODULE_AUTHOR("Emulex Corporation");
S
Sathya Perla 已提交
31 32
MODULE_LICENSE("GPL");

33 34 35
/* num_vfs module param is obsolete.
 * Use sysfs method to enable/disable VFs.
 */
36 37 38
static unsigned int num_vfs;
module_param(num_vfs, uint, S_IRUGO);
MODULE_PARM_DESC(num_vfs, "Number of PCI VFs to initialize");
S
Sathya Perla 已提交
39

40 41 42 43
static ushort rx_frag_size = 2048;
module_param(rx_frag_size, ushort, S_IRUGO);
MODULE_PARM_DESC(rx_frag_size, "Size of a fragment that holds rcvd data.");

44 45 46 47 48
/* Per-module error detection/recovery workq shared across all functions.
 * Each function schedules its own work request on this shared workq.
 */
struct workqueue_struct *be_err_recovery_workq;

49
static const struct pci_device_id be_dev_ids[] = {
50
	{ PCI_DEVICE(BE_VENDOR_ID, BE_DEVICE_ID1) },
51
	{ PCI_DEVICE(BE_VENDOR_ID, BE_DEVICE_ID2) },
52 53
	{ PCI_DEVICE(BE_VENDOR_ID, OC_DEVICE_ID1) },
	{ PCI_DEVICE(BE_VENDOR_ID, OC_DEVICE_ID2) },
54
	{ PCI_DEVICE(EMULEX_VENDOR_ID, OC_DEVICE_ID3)},
55
	{ PCI_DEVICE(EMULEX_VENDOR_ID, OC_DEVICE_ID4)},
56
	{ PCI_DEVICE(EMULEX_VENDOR_ID, OC_DEVICE_ID5)},
57
	{ PCI_DEVICE(EMULEX_VENDOR_ID, OC_DEVICE_ID6)},
S
Sathya Perla 已提交
58 59 60
	{ 0 }
};
MODULE_DEVICE_TABLE(pci, be_dev_ids);
61 62 63 64

/* Workqueue used by all functions for defering cmd calls to the adapter */
struct workqueue_struct *be_wq;

65
/* UE Status Low CSR */
66
static const char * const ue_status_low_desc[] = {
67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
	"CEV",
	"CTX",
	"DBUF",
	"ERX",
	"Host",
	"MPU",
	"NDMA",
	"PTC ",
	"RDMA ",
	"RXF ",
	"RXIPS ",
	"RXULP0 ",
	"RXULP1 ",
	"RXULP2 ",
	"TIM ",
	"TPOST ",
	"TPRE ",
	"TXIPS ",
	"TXULP0 ",
	"TXULP1 ",
	"UC ",
	"WDMA ",
	"TXULP2 ",
	"HOST1 ",
	"P0_OB_LINK ",
	"P1_OB_LINK ",
	"HOST_GPIO ",
	"MBOX ",
95 96 97 98
	"ERX2 ",
	"SPARE ",
	"JTAG ",
	"MPU_INTPEND "
99
};
100

101
/* UE Status High CSR */
102
static const char * const ue_status_hi_desc[] = {
103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
	"LPCMEMHOST",
	"MGMT_MAC",
	"PCS0ONLINE",
	"MPU_IRAM",
	"PCS1ONLINE",
	"PCTL0",
	"PCTL1",
	"PMEM",
	"RR",
	"TXPB",
	"RXPP",
	"XAUI",
	"TXP",
	"ARM",
	"IPC",
	"HOST2",
	"HOST3",
	"HOST4",
	"HOST5",
	"HOST6",
	"HOST7",
124 125
	"ECRC",
	"Poison TLP",
126
	"NETC",
127 128 129 130 131 132 133
	"PERIPH",
	"LLTXULP",
	"D2P",
	"RCON",
	"LDMA",
	"LLTXP",
	"LLTXPB",
134 135
	"Unknown"
};
S
Sathya Perla 已提交
136

137 138 139 140 141
#define BE_VF_IF_EN_FLAGS	(BE_IF_FLAGS_UNTAGGED | \
				 BE_IF_FLAGS_BROADCAST | \
				 BE_IF_FLAGS_MULTICAST | \
				 BE_IF_FLAGS_PASS_L3L4_ERRORS)

S
Sathya Perla 已提交
142 143 144
static void be_queue_free(struct be_adapter *adapter, struct be_queue_info *q)
{
	struct be_dma_mem *mem = &q->dma_mem;
145

146
	if (mem->va) {
I
Ivan Vecera 已提交
147 148
		dma_free_coherent(&adapter->pdev->dev, mem->size, mem->va,
				  mem->dma);
149 150
		mem->va = NULL;
	}
S
Sathya Perla 已提交
151 152 153
}

static int be_queue_alloc(struct be_adapter *adapter, struct be_queue_info *q,
154
			  u16 len, u16 entry_size)
S
Sathya Perla 已提交
155 156 157 158 159 160 161
{
	struct be_dma_mem *mem = &q->dma_mem;

	memset(q, 0, sizeof(*q));
	q->len = len;
	q->entry_size = entry_size;
	mem->size = len * entry_size;
162 163
	mem->va = dma_zalloc_coherent(&adapter->pdev->dev, mem->size, &mem->dma,
				      GFP_KERNEL);
S
Sathya Perla 已提交
164
	if (!mem->va)
S
Sathya Perla 已提交
165
		return -ENOMEM;
S
Sathya Perla 已提交
166 167 168
	return 0;
}

169
static void be_reg_intr_set(struct be_adapter *adapter, bool enable)
S
Sathya Perla 已提交
170
{
171
	u32 reg, enabled;
172

173
	pci_read_config_dword(adapter->pdev, PCICFG_MEMBAR_CTRL_INT_CTRL_OFFSET,
174
			      &reg);
175 176
	enabled = reg & MEMBAR_CTRL_INT_CTRL_HOSTINTR_MASK;

177
	if (!enabled && enable)
S
Sathya Perla 已提交
178
		reg |= MEMBAR_CTRL_INT_CTRL_HOSTINTR_MASK;
179
	else if (enabled && !enable)
S
Sathya Perla 已提交
180
		reg &= ~MEMBAR_CTRL_INT_CTRL_HOSTINTR_MASK;
181
	else
S
Sathya Perla 已提交
182
		return;
183

184
	pci_write_config_dword(adapter->pdev,
185
			       PCICFG_MEMBAR_CTRL_INT_CTRL_OFFSET, reg);
S
Sathya Perla 已提交
186 187
}

188 189 190 191 192 193 194 195
static void be_intr_set(struct be_adapter *adapter, bool enable)
{
	int status = 0;

	/* On lancer interrupts can't be controlled via this register */
	if (lancer_chip(adapter))
		return;

196
	if (be_check_error(adapter, BE_ERROR_EEH))
197 198 199 200 201 202 203
		return;

	status = be_cmd_intr_set(adapter, enable);
	if (status)
		be_reg_intr_set(adapter, enable);
}

204
static void be_rxq_notify(struct be_adapter *adapter, u16 qid, u16 posted)
S
Sathya Perla 已提交
205 206
{
	u32 val = 0;
207

208 209 210
	if (be_check_error(adapter, BE_ERROR_HW))
		return;

S
Sathya Perla 已提交
211 212
	val |= qid & DB_RQ_RING_ID_MASK;
	val |= posted << DB_RQ_NUM_POSTED_SHIFT;
213 214

	wmb();
215
	iowrite32(val, adapter->db + DB_RQ_OFFSET);
S
Sathya Perla 已提交
216 217
}

V
Vasundhara Volam 已提交
218 219
static void be_txq_notify(struct be_adapter *adapter, struct be_tx_obj *txo,
			  u16 posted)
S
Sathya Perla 已提交
220 221
{
	u32 val = 0;
222

223 224 225
	if (be_check_error(adapter, BE_ERROR_HW))
		return;

V
Vasundhara Volam 已提交
226
	val |= txo->q.id & DB_TXULP_RING_ID_MASK;
S
Sathya Perla 已提交
227
	val |= (posted & DB_TXULP_NUM_POSTED_MASK) << DB_TXULP_NUM_POSTED_SHIFT;
228 229

	wmb();
V
Vasundhara Volam 已提交
230
	iowrite32(val, adapter->db + txo->db_offset);
S
Sathya Perla 已提交
231 232
}

233
static void be_eq_notify(struct be_adapter *adapter, u16 qid,
234 235
			 bool arm, bool clear_int, u16 num_popped,
			 u32 eq_delay_mult_enc)
S
Sathya Perla 已提交
236 237
{
	u32 val = 0;
238

S
Sathya Perla 已提交
239
	val |= qid & DB_EQ_RING_ID_MASK;
240
	val |= ((qid & DB_EQ_RING_ID_EXT_MASK) << DB_EQ_RING_ID_EXT_MASK_SHIFT);
241

242
	if (be_check_error(adapter, BE_ERROR_HW))
243 244
		return;

S
Sathya Perla 已提交
245 246 247 248 249 250
	if (arm)
		val |= 1 << DB_EQ_REARM_SHIFT;
	if (clear_int)
		val |= 1 << DB_EQ_CLR_SHIFT;
	val |= 1 << DB_EQ_EVNT_SHIFT;
	val |= num_popped << DB_EQ_NUM_POPPED_SHIFT;
251
	val |= eq_delay_mult_enc << DB_EQ_R2I_DLY_SHIFT;
252
	iowrite32(val, adapter->db + DB_EQ_OFFSET);
S
Sathya Perla 已提交
253 254
}

255
void be_cq_notify(struct be_adapter *adapter, u16 qid, bool arm, u16 num_popped)
S
Sathya Perla 已提交
256 257
{
	u32 val = 0;
258

S
Sathya Perla 已提交
259
	val |= qid & DB_CQ_RING_ID_MASK;
260 261
	val |= ((qid & DB_CQ_RING_ID_EXT_MASK) <<
			DB_CQ_RING_ID_EXT_MASK_SHIFT);
262

263
	if (be_check_error(adapter, BE_ERROR_HW))
264 265
		return;

S
Sathya Perla 已提交
266 267 268
	if (arm)
		val |= 1 << DB_CQ_REARM_SHIFT;
	val |= num_popped << DB_CQ_NUM_POPPED_SHIFT;
269
	iowrite32(val, adapter->db + DB_CQ_OFFSET);
S
Sathya Perla 已提交
270 271
}

272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303
static int be_dev_mac_add(struct be_adapter *adapter, u8 *mac)
{
	int i;

	/* Check if mac has already been added as part of uc-list */
	for (i = 0; i < adapter->uc_macs; i++) {
		if (ether_addr_equal((u8 *)&adapter->uc_list[i * ETH_ALEN],
				     mac)) {
			/* mac already added, skip addition */
			adapter->pmac_id[0] = adapter->pmac_id[i + 1];
			return 0;
		}
	}

	return be_cmd_pmac_add(adapter, mac, adapter->if_handle,
			       &adapter->pmac_id[0], 0);
}

static void be_dev_mac_del(struct be_adapter *adapter, int pmac_id)
{
	int i;

	/* Skip deletion if the programmed mac is
	 * being used in uc-list
	 */
	for (i = 0; i < adapter->uc_macs; i++) {
		if (adapter->pmac_id[i + 1] == pmac_id)
			return;
	}
	be_cmd_pmac_del(adapter, adapter->if_handle, pmac_id, 0);
}

S
Sathya Perla 已提交
304 305 306
static int be_mac_addr_set(struct net_device *netdev, void *p)
{
	struct be_adapter *adapter = netdev_priv(netdev);
307
	struct device *dev = &adapter->pdev->dev;
S
Sathya Perla 已提交
308
	struct sockaddr *addr = p;
309 310
	int status;
	u8 mac[ETH_ALEN];
311
	u32 old_pmac_id = adapter->pmac_id[0];
S
Sathya Perla 已提交
312

313 314 315
	if (!is_valid_ether_addr(addr->sa_data))
		return -EADDRNOTAVAIL;

316 317 318
	/* Proceed further only if, User provided MAC is different
	 * from active MAC
	 */
319
	if (ether_addr_equal(addr->sa_data, adapter->dev_mac))
320 321
		return 0;

322 323 324 325
	/* if device is not running, copy MAC to netdev->dev_addr */
	if (!netif_running(netdev))
		goto done;

326 327 328 329 330
	/* The PMAC_ADD cmd may fail if the VF doesn't have FILTMGMT
	 * privilege or if PF did not provision the new MAC address.
	 * On BE3, this cmd will always fail if the VF doesn't have the
	 * FILTMGMT privilege. This failure is OK, only if the PF programmed
	 * the MAC for the VF.
331
	 */
332 333
	mutex_lock(&adapter->rx_filter_lock);
	status = be_dev_mac_add(adapter, (u8 *)addr->sa_data);
334 335 336 337 338 339
	if (!status) {

		/* Delete the old programmed MAC. This call may fail if the
		 * old MAC was already deleted by the PF driver.
		 */
		if (adapter->pmac_id[0] != old_pmac_id)
340
			be_dev_mac_del(adapter, old_pmac_id);
341 342
	}

343
	mutex_unlock(&adapter->rx_filter_lock);
344 345
	/* Decide if the new MAC is successfully activated only after
	 * querying the FW
346
	 */
347
	status = be_cmd_get_active_mac(adapter, adapter->pmac_id[0], mac,
348
				       adapter->if_handle, true, 0);
349
	if (status)
350
		goto err;
S
Sathya Perla 已提交
351

352 353 354
	/* The MAC change did not happen, either due to lack of privilege
	 * or PF didn't pre-provision.
	 */
355
	if (!ether_addr_equal(addr->sa_data, mac)) {
356 357 358
		status = -EPERM;
		goto err;
	}
359
done:
360
	ether_addr_copy(adapter->dev_mac, addr->sa_data);
361 362
	ether_addr_copy(netdev->dev_addr, addr->sa_data);
	dev_info(dev, "MAC address changed to %pM\n", addr->sa_data);
363 364
	return 0;
err:
365
	dev_warn(dev, "MAC address change to %pM failed\n", addr->sa_data);
S
Sathya Perla 已提交
366 367 368
	return status;
}

369 370 371 372 373 374 375
/* BE2 supports only v0 cmd */
static void *hw_stats_from_cmd(struct be_adapter *adapter)
{
	if (BE2_chip(adapter)) {
		struct be_cmd_resp_get_stats_v0 *cmd = adapter->stats_cmd.va;

		return &cmd->hw_stats;
376
	} else if (BE3_chip(adapter)) {
377 378
		struct be_cmd_resp_get_stats_v1 *cmd = adapter->stats_cmd.va;

379 380 381 382
		return &cmd->hw_stats;
	} else {
		struct be_cmd_resp_get_stats_v2 *cmd = adapter->stats_cmd.va;

383 384 385 386 387 388 389 390 391 392 393
		return &cmd->hw_stats;
	}
}

/* BE2 supports only v0 cmd */
static void *be_erx_stats_from_cmd(struct be_adapter *adapter)
{
	if (BE2_chip(adapter)) {
		struct be_hw_stats_v0 *hw_stats = hw_stats_from_cmd(adapter);

		return &hw_stats->erx;
394
	} else if (BE3_chip(adapter)) {
395 396
		struct be_hw_stats_v1 *hw_stats = hw_stats_from_cmd(adapter);

397 398 399 400
		return &hw_stats->erx;
	} else {
		struct be_hw_stats_v2 *hw_stats = hw_stats_from_cmd(adapter);

401 402 403 404 405
		return &hw_stats->erx;
	}
}

static void populate_be_v0_stats(struct be_adapter *adapter)
406
{
407 408 409
	struct be_hw_stats_v0 *hw_stats = hw_stats_from_cmd(adapter);
	struct be_pmem_stats *pmem_sts = &hw_stats->pmem;
	struct be_rxf_stats_v0 *rxf_stats = &hw_stats->rxf;
410
	struct be_port_rxf_stats_v0 *port_stats =
411 412
					&rxf_stats->port[adapter->port_num];
	struct be_drv_stats *drvs = &adapter->drv_stats;
413

414
	be_dws_le_to_cpu(hw_stats, sizeof(*hw_stats));
415 416 417 418 419 420 421 422 423 424 425 426 427 428
	drvs->rx_pause_frames = port_stats->rx_pause_frames;
	drvs->rx_crc_errors = port_stats->rx_crc_errors;
	drvs->rx_control_frames = port_stats->rx_control_frames;
	drvs->rx_in_range_errors = port_stats->rx_in_range_errors;
	drvs->rx_frame_too_long = port_stats->rx_frame_too_long;
	drvs->rx_dropped_runt = port_stats->rx_dropped_runt;
	drvs->rx_ip_checksum_errs = port_stats->rx_ip_checksum_errs;
	drvs->rx_tcp_checksum_errs = port_stats->rx_tcp_checksum_errs;
	drvs->rx_udp_checksum_errs = port_stats->rx_udp_checksum_errs;
	drvs->rxpp_fifo_overflow_drop = port_stats->rx_fifo_overflow;
	drvs->rx_dropped_tcp_length = port_stats->rx_dropped_tcp_length;
	drvs->rx_dropped_too_small = port_stats->rx_dropped_too_small;
	drvs->rx_dropped_too_short = port_stats->rx_dropped_too_short;
	drvs->rx_out_range_errors = port_stats->rx_out_range_errors;
429
	drvs->rx_input_fifo_overflow_drop = port_stats->rx_input_fifo_overflow;
430 431
	drvs->rx_dropped_header_too_small =
		port_stats->rx_dropped_header_too_small;
432 433 434
	drvs->rx_address_filtered =
					port_stats->rx_address_filtered +
					port_stats->rx_vlan_filtered;
435 436 437 438 439 440 441
	drvs->rx_alignment_symbol_errors =
		port_stats->rx_alignment_symbol_errors;

	drvs->tx_pauseframes = port_stats->tx_pauseframes;
	drvs->tx_controlframes = port_stats->tx_controlframes;

	if (adapter->port_num)
442
		drvs->jabber_events = rxf_stats->port1_jabber_events;
443
	else
444
		drvs->jabber_events = rxf_stats->port0_jabber_events;
445 446 447 448
	drvs->rx_drops_no_pbuf = rxf_stats->rx_drops_no_pbuf;
	drvs->rx_drops_no_erx_descr = rxf_stats->rx_drops_no_erx_descr;
	drvs->forwarded_packets = rxf_stats->forwarded_packets;
	drvs->rx_drops_mtu = rxf_stats->rx_drops_mtu;
449 450
	drvs->rx_drops_no_tpre_descr = rxf_stats->rx_drops_no_tpre_descr;
	drvs->rx_drops_too_many_frags = rxf_stats->rx_drops_too_many_frags;
451 452 453
	adapter->drv_stats.eth_red_drops = pmem_sts->eth_red_drops;
}

454
static void populate_be_v1_stats(struct be_adapter *adapter)
455
{
456 457 458
	struct be_hw_stats_v1 *hw_stats = hw_stats_from_cmd(adapter);
	struct be_pmem_stats *pmem_sts = &hw_stats->pmem;
	struct be_rxf_stats_v1 *rxf_stats = &hw_stats->rxf;
459
	struct be_port_rxf_stats_v1 *port_stats =
460 461
					&rxf_stats->port[adapter->port_num];
	struct be_drv_stats *drvs = &adapter->drv_stats;
462

463
	be_dws_le_to_cpu(hw_stats, sizeof(*hw_stats));
464 465
	drvs->pmem_fifo_overflow_drop = port_stats->pmem_fifo_overflow_drop;
	drvs->rx_priority_pause_frames = port_stats->rx_priority_pause_frames;
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482
	drvs->rx_pause_frames = port_stats->rx_pause_frames;
	drvs->rx_crc_errors = port_stats->rx_crc_errors;
	drvs->rx_control_frames = port_stats->rx_control_frames;
	drvs->rx_in_range_errors = port_stats->rx_in_range_errors;
	drvs->rx_frame_too_long = port_stats->rx_frame_too_long;
	drvs->rx_dropped_runt = port_stats->rx_dropped_runt;
	drvs->rx_ip_checksum_errs = port_stats->rx_ip_checksum_errs;
	drvs->rx_tcp_checksum_errs = port_stats->rx_tcp_checksum_errs;
	drvs->rx_udp_checksum_errs = port_stats->rx_udp_checksum_errs;
	drvs->rx_dropped_tcp_length = port_stats->rx_dropped_tcp_length;
	drvs->rx_dropped_too_small = port_stats->rx_dropped_too_small;
	drvs->rx_dropped_too_short = port_stats->rx_dropped_too_short;
	drvs->rx_out_range_errors = port_stats->rx_out_range_errors;
	drvs->rx_dropped_header_too_small =
		port_stats->rx_dropped_header_too_small;
	drvs->rx_input_fifo_overflow_drop =
		port_stats->rx_input_fifo_overflow_drop;
483
	drvs->rx_address_filtered = port_stats->rx_address_filtered;
484 485
	drvs->rx_alignment_symbol_errors =
		port_stats->rx_alignment_symbol_errors;
486
	drvs->rxpp_fifo_overflow_drop = port_stats->rxpp_fifo_overflow_drop;
487 488
	drvs->tx_pauseframes = port_stats->tx_pauseframes;
	drvs->tx_controlframes = port_stats->tx_controlframes;
489
	drvs->tx_priority_pauseframes = port_stats->tx_priority_pauseframes;
490 491 492 493 494
	drvs->jabber_events = port_stats->jabber_events;
	drvs->rx_drops_no_pbuf = rxf_stats->rx_drops_no_pbuf;
	drvs->rx_drops_no_erx_descr = rxf_stats->rx_drops_no_erx_descr;
	drvs->forwarded_packets = rxf_stats->forwarded_packets;
	drvs->rx_drops_mtu = rxf_stats->rx_drops_mtu;
495 496
	drvs->rx_drops_no_tpre_descr = rxf_stats->rx_drops_no_tpre_descr;
	drvs->rx_drops_too_many_frags = rxf_stats->rx_drops_too_many_frags;
497 498 499
	adapter->drv_stats.eth_red_drops = pmem_sts->eth_red_drops;
}

500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
static void populate_be_v2_stats(struct be_adapter *adapter)
{
	struct be_hw_stats_v2 *hw_stats = hw_stats_from_cmd(adapter);
	struct be_pmem_stats *pmem_sts = &hw_stats->pmem;
	struct be_rxf_stats_v2 *rxf_stats = &hw_stats->rxf;
	struct be_port_rxf_stats_v2 *port_stats =
					&rxf_stats->port[adapter->port_num];
	struct be_drv_stats *drvs = &adapter->drv_stats;

	be_dws_le_to_cpu(hw_stats, sizeof(*hw_stats));
	drvs->pmem_fifo_overflow_drop = port_stats->pmem_fifo_overflow_drop;
	drvs->rx_priority_pause_frames = port_stats->rx_priority_pause_frames;
	drvs->rx_pause_frames = port_stats->rx_pause_frames;
	drvs->rx_crc_errors = port_stats->rx_crc_errors;
	drvs->rx_control_frames = port_stats->rx_control_frames;
	drvs->rx_in_range_errors = port_stats->rx_in_range_errors;
	drvs->rx_frame_too_long = port_stats->rx_frame_too_long;
	drvs->rx_dropped_runt = port_stats->rx_dropped_runt;
	drvs->rx_ip_checksum_errs = port_stats->rx_ip_checksum_errs;
	drvs->rx_tcp_checksum_errs = port_stats->rx_tcp_checksum_errs;
	drvs->rx_udp_checksum_errs = port_stats->rx_udp_checksum_errs;
	drvs->rx_dropped_tcp_length = port_stats->rx_dropped_tcp_length;
	drvs->rx_dropped_too_small = port_stats->rx_dropped_too_small;
	drvs->rx_dropped_too_short = port_stats->rx_dropped_too_short;
	drvs->rx_out_range_errors = port_stats->rx_out_range_errors;
	drvs->rx_dropped_header_too_small =
		port_stats->rx_dropped_header_too_small;
	drvs->rx_input_fifo_overflow_drop =
		port_stats->rx_input_fifo_overflow_drop;
	drvs->rx_address_filtered = port_stats->rx_address_filtered;
	drvs->rx_alignment_symbol_errors =
		port_stats->rx_alignment_symbol_errors;
	drvs->rxpp_fifo_overflow_drop = port_stats->rxpp_fifo_overflow_drop;
	drvs->tx_pauseframes = port_stats->tx_pauseframes;
	drvs->tx_controlframes = port_stats->tx_controlframes;
	drvs->tx_priority_pauseframes = port_stats->tx_priority_pauseframes;
	drvs->jabber_events = port_stats->jabber_events;
	drvs->rx_drops_no_pbuf = rxf_stats->rx_drops_no_pbuf;
	drvs->rx_drops_no_erx_descr = rxf_stats->rx_drops_no_erx_descr;
	drvs->forwarded_packets = rxf_stats->forwarded_packets;
	drvs->rx_drops_mtu = rxf_stats->rx_drops_mtu;
	drvs->rx_drops_no_tpre_descr = rxf_stats->rx_drops_no_tpre_descr;
	drvs->rx_drops_too_many_frags = rxf_stats->rx_drops_too_many_frags;
	adapter->drv_stats.eth_red_drops = pmem_sts->eth_red_drops;
544
	if (be_roce_supported(adapter)) {
545 546 547 548 549 550 551
		drvs->rx_roce_bytes_lsd = port_stats->roce_bytes_received_lsd;
		drvs->rx_roce_bytes_msd = port_stats->roce_bytes_received_msd;
		drvs->rx_roce_frames = port_stats->roce_frames_received;
		drvs->roce_drops_crc = port_stats->roce_drops_crc;
		drvs->roce_drops_payload_len =
			port_stats->roce_drops_payload_len;
	}
552 553
}

S
Selvin Xavier 已提交
554 555 556
static void populate_lancer_stats(struct be_adapter *adapter)
{
	struct be_drv_stats *drvs = &adapter->drv_stats;
557
	struct lancer_pport_stats *pport_stats = pport_stats_from_cmd(adapter);
558 559 560 561 562

	be_dws_le_to_cpu(pport_stats, sizeof(*pport_stats));
	drvs->rx_pause_frames = pport_stats->rx_pause_frames_lo;
	drvs->rx_crc_errors = pport_stats->rx_crc_errors_lo;
	drvs->rx_control_frames = pport_stats->rx_control_frames_lo;
S
Selvin Xavier 已提交
563
	drvs->rx_in_range_errors = pport_stats->rx_in_range_errors;
564
	drvs->rx_frame_too_long = pport_stats->rx_frames_too_long_lo;
S
Selvin Xavier 已提交
565 566 567 568 569 570 571 572 573 574 575 576
	drvs->rx_dropped_runt = pport_stats->rx_dropped_runt;
	drvs->rx_ip_checksum_errs = pport_stats->rx_ip_checksum_errors;
	drvs->rx_tcp_checksum_errs = pport_stats->rx_tcp_checksum_errors;
	drvs->rx_udp_checksum_errs = pport_stats->rx_udp_checksum_errors;
	drvs->rx_dropped_tcp_length =
				pport_stats->rx_dropped_invalid_tcp_length;
	drvs->rx_dropped_too_small = pport_stats->rx_dropped_too_small;
	drvs->rx_dropped_too_short = pport_stats->rx_dropped_too_short;
	drvs->rx_out_range_errors = pport_stats->rx_out_of_range_errors;
	drvs->rx_dropped_header_too_small =
				pport_stats->rx_dropped_header_too_small;
	drvs->rx_input_fifo_overflow_drop = pport_stats->rx_fifo_overflow;
577 578 579
	drvs->rx_address_filtered =
					pport_stats->rx_address_filtered +
					pport_stats->rx_vlan_filtered;
580
	drvs->rx_alignment_symbol_errors = pport_stats->rx_symbol_errors_lo;
S
Selvin Xavier 已提交
581
	drvs->rxpp_fifo_overflow_drop = pport_stats->rx_fifo_overflow;
582 583
	drvs->tx_pauseframes = pport_stats->tx_pause_frames_lo;
	drvs->tx_controlframes = pport_stats->tx_control_frames_lo;
S
Selvin Xavier 已提交
584
	drvs->jabber_events = pport_stats->rx_jabbers;
585 586
	drvs->forwarded_packets = pport_stats->num_forwards_lo;
	drvs->rx_drops_mtu = pport_stats->rx_drops_mtu_lo;
S
Selvin Xavier 已提交
587
	drvs->rx_drops_too_many_frags =
588
				pport_stats->rx_drops_too_many_frags_lo;
S
Selvin Xavier 已提交
589
}
590

591 592 593 594 595 596 597 598 599 600 601 602
static void accumulate_16bit_val(u32 *acc, u16 val)
{
#define lo(x)			(x & 0xFFFF)
#define hi(x)			(x & 0xFFFF0000)
	bool wrapped = val < lo(*acc);
	u32 newacc = hi(*acc) + val;

	if (wrapped)
		newacc += 65536;
	ACCESS_ONCE(*acc) = newacc;
}

J
Jingoo Han 已提交
603
static void populate_erx_stats(struct be_adapter *adapter,
604
			       struct be_rx_obj *rxo, u32 erx_stat)
605 606 607 608 609 610 611 612 613 614 615
{
	if (!BEx_chip(adapter))
		rx_stats(rxo)->rx_drops_no_frags = erx_stat;
	else
		/* below erx HW counter can actually wrap around after
		 * 65535. Driver accumulates a 32-bit value
		 */
		accumulate_16bit_val(&rx_stats(rxo)->rx_drops_no_frags,
				     (u16)erx_stat);
}

616 617
void be_parse_stats(struct be_adapter *adapter)
{
618
	struct be_erx_stats_v2 *erx = be_erx_stats_from_cmd(adapter);
619 620
	struct be_rx_obj *rxo;
	int i;
621
	u32 erx_stat;
622

623 624
	if (lancer_chip(adapter)) {
		populate_lancer_stats(adapter);
S
Selvin Xavier 已提交
625
	} else {
626 627
		if (BE2_chip(adapter))
			populate_be_v0_stats(adapter);
628 629
		else if (BE3_chip(adapter))
			/* for BE3 */
630
			populate_be_v1_stats(adapter);
631 632
		else
			populate_be_v2_stats(adapter);
633

634
		/* erx_v2 is longer than v0, v1. use v2 for v0, v1 access */
635
		for_all_rx_queues(adapter, rxo, i) {
636 637
			erx_stat = erx->rx_drops_no_fragments[rxo->q.id];
			populate_erx_stats(adapter, rxo, erx_stat);
638
		}
639
	}
640 641
}

642
static struct rtnl_link_stats64 *be_get_stats64(struct net_device *netdev,
643
						struct rtnl_link_stats64 *stats)
S
Sathya Perla 已提交
644
{
645
	struct be_adapter *adapter = netdev_priv(netdev);
646
	struct be_drv_stats *drvs = &adapter->drv_stats;
647
	struct be_rx_obj *rxo;
648
	struct be_tx_obj *txo;
649 650
	u64 pkts, bytes;
	unsigned int start;
651
	int i;
S
Sathya Perla 已提交
652

653
	for_all_rx_queues(adapter, rxo, i) {
654
		const struct be_rx_stats *rx_stats = rx_stats(rxo);
655

656
		do {
657
			start = u64_stats_fetch_begin_irq(&rx_stats->sync);
658 659
			pkts = rx_stats(rxo)->rx_pkts;
			bytes = rx_stats(rxo)->rx_bytes;
660
		} while (u64_stats_fetch_retry_irq(&rx_stats->sync, start));
661 662 663 664 665
		stats->rx_packets += pkts;
		stats->rx_bytes += bytes;
		stats->multicast += rx_stats(rxo)->rx_mcast_pkts;
		stats->rx_dropped += rx_stats(rxo)->rx_drops_no_skbs +
					rx_stats(rxo)->rx_drops_no_frags;
666 667
	}

668
	for_all_tx_queues(adapter, txo, i) {
669
		const struct be_tx_stats *tx_stats = tx_stats(txo);
670

671
		do {
672
			start = u64_stats_fetch_begin_irq(&tx_stats->sync);
673 674
			pkts = tx_stats(txo)->tx_pkts;
			bytes = tx_stats(txo)->tx_bytes;
675
		} while (u64_stats_fetch_retry_irq(&tx_stats->sync, start));
676 677
		stats->tx_packets += pkts;
		stats->tx_bytes += bytes;
678
	}
S
Sathya Perla 已提交
679 680

	/* bad pkts received */
681
	stats->rx_errors = drvs->rx_crc_errors +
682 683 684 685 686 687 688 689
		drvs->rx_alignment_symbol_errors +
		drvs->rx_in_range_errors +
		drvs->rx_out_range_errors +
		drvs->rx_frame_too_long +
		drvs->rx_dropped_too_small +
		drvs->rx_dropped_too_short +
		drvs->rx_dropped_header_too_small +
		drvs->rx_dropped_tcp_length +
690
		drvs->rx_dropped_runt;
691

S
Sathya Perla 已提交
692
	/* detailed rx errors */
693
	stats->rx_length_errors = drvs->rx_in_range_errors +
694 695
		drvs->rx_out_range_errors +
		drvs->rx_frame_too_long;
696

697
	stats->rx_crc_errors = drvs->rx_crc_errors;
S
Sathya Perla 已提交
698 699

	/* frame alignment errors */
700
	stats->rx_frame_errors = drvs->rx_alignment_symbol_errors;
701

S
Sathya Perla 已提交
702 703
	/* receiver fifo overrun */
	/* drops_no_pbuf is no per i/f, it's per BE card */
704
	stats->rx_fifo_errors = drvs->rxpp_fifo_overflow_drop +
705 706
				drvs->rx_input_fifo_overflow_drop +
				drvs->rx_drops_no_pbuf;
707
	return stats;
S
Sathya Perla 已提交
708 709
}

710
void be_link_status_update(struct be_adapter *adapter, u8 link_status)
S
Sathya Perla 已提交
711 712 713
{
	struct net_device *netdev = adapter->netdev;

714
	if (!(adapter->flags & BE_FLAGS_LINK_STATUS_INIT)) {
715
		netif_carrier_off(netdev);
716
		adapter->flags |= BE_FLAGS_LINK_STATUS_INIT;
S
Sathya Perla 已提交
717
	}
718

719
	if (link_status)
720 721 722
		netif_carrier_on(netdev);
	else
		netif_carrier_off(netdev);
I
Ivan Vecera 已提交
723 724

	netdev_info(netdev, "Link is %s\n", link_status ? "Up" : "Down");
S
Sathya Perla 已提交
725 726
}

727
static void be_tx_stats_update(struct be_tx_obj *txo, struct sk_buff *skb)
S
Sathya Perla 已提交
728
{
729
	struct be_tx_stats *stats = tx_stats(txo);
730
	u64 tx_pkts = skb_shinfo(skb)->gso_segs ? : 1;
731

732
	u64_stats_update_begin(&stats->sync);
733
	stats->tx_reqs++;
734
	stats->tx_bytes += skb->len;
735 736 737
	stats->tx_pkts += tx_pkts;
	if (skb->encapsulation && skb->ip_summed == CHECKSUM_PARTIAL)
		stats->tx_vxlan_offload_pkts += tx_pkts;
738
	u64_stats_update_end(&stats->sync);
S
Sathya Perla 已提交
739 740
}

741 742
/* Returns number of WRBs needed for the skb */
static u32 skb_wrb_cnt(struct sk_buff *skb)
S
Sathya Perla 已提交
743
{
744 745
	/* +1 for the header wrb */
	return 1 + (skb_headlen(skb) ? 1 : 0) + skb_shinfo(skb)->nr_frags;
S
Sathya Perla 已提交
746 747 748 749
}

static inline void wrb_fill(struct be_eth_wrb *wrb, u64 addr, int len)
{
750 751 752 753 754 755 756 757 758 759 760 761 762 763
	wrb->frag_pa_hi = cpu_to_le32(upper_32_bits(addr));
	wrb->frag_pa_lo = cpu_to_le32(lower_32_bits(addr));
	wrb->frag_len = cpu_to_le32(len & ETH_WRB_FRAG_LEN_MASK);
	wrb->rsvd0 = 0;
}

/* A dummy wrb is just all zeros. Using a separate routine for dummy-wrb
 * to avoid the swap and shift/mask operations in wrb_fill().
 */
static inline void wrb_fill_dummy(struct be_eth_wrb *wrb)
{
	wrb->frag_pa_hi = 0;
	wrb->frag_pa_lo = 0;
	wrb->frag_len = 0;
764
	wrb->rsvd0 = 0;
S
Sathya Perla 已提交
765 766
}

767
static inline u16 be_get_tx_vlan_tag(struct be_adapter *adapter,
768
				     struct sk_buff *skb)
769 770 771 772
{
	u8 vlan_prio;
	u16 vlan_tag;

773
	vlan_tag = skb_vlan_tag_get(skb);
774 775 776 777
	vlan_prio = (vlan_tag & VLAN_PRIO_MASK) >> VLAN_PRIO_SHIFT;
	/* If vlan priority provided by OS is NOT in available bmap */
	if (!(adapter->vlan_prio_bmap & (1 << vlan_prio)))
		vlan_tag = (vlan_tag & ~VLAN_PRIO_MASK) |
778
				adapter->recommended_prio_bits;
779 780 781 782

	return vlan_tag;
}

783 784 785 786 787 788 789 790 791 792 793 794 795
/* Used only for IP tunnel packets */
static u16 skb_inner_ip_proto(struct sk_buff *skb)
{
	return (inner_ip_hdr(skb)->version == 4) ?
		inner_ip_hdr(skb)->protocol : inner_ipv6_hdr(skb)->nexthdr;
}

static u16 skb_ip_proto(struct sk_buff *skb)
{
	return (ip_hdr(skb)->version == 4) ?
		ip_hdr(skb)->protocol : ipv6_hdr(skb)->nexthdr;
}

796 797 798 799 800 801 802 803 804 805 806 807 808 809 810
static inline bool be_is_txq_full(struct be_tx_obj *txo)
{
	return atomic_read(&txo->q.used) + BE_MAX_TX_FRAG_COUNT >= txo->q.len;
}

static inline bool be_can_txq_wake(struct be_tx_obj *txo)
{
	return atomic_read(&txo->q.used) < txo->q.len / 2;
}

static inline bool be_is_tx_compl_pending(struct be_tx_obj *txo)
{
	return atomic_read(&txo->q.used) > txo->pend_wrb_cnt;
}

811 812 813
static void be_get_wrb_params_from_skb(struct be_adapter *adapter,
				       struct sk_buff *skb,
				       struct be_wrb_params *wrb_params)
S
Sathya Perla 已提交
814
{
815
	u16 proto;
S
Sathya Perla 已提交
816

A
Ajit Khaparde 已提交
817
	if (skb_is_gso(skb)) {
818 819
		BE_WRB_F_SET(wrb_params->features, LSO, 1);
		wrb_params->lso_mss = skb_shinfo(skb)->gso_size;
820
		if (skb_is_gso_v6(skb) && !lancer_chip(adapter))
821
			BE_WRB_F_SET(wrb_params->features, LSO6, 1);
S
Sathya Perla 已提交
822
	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
823
		if (skb->encapsulation) {
824
			BE_WRB_F_SET(wrb_params->features, IPCS, 1);
825 826 827 828 829
			proto = skb_inner_ip_proto(skb);
		} else {
			proto = skb_ip_proto(skb);
		}
		if (proto == IPPROTO_TCP)
830
			BE_WRB_F_SET(wrb_params->features, TCPCS, 1);
831
		else if (proto == IPPROTO_UDP)
832
			BE_WRB_F_SET(wrb_params->features, UDPCS, 1);
S
Sathya Perla 已提交
833 834
	}

835
	if (skb_vlan_tag_present(skb)) {
836 837
		BE_WRB_F_SET(wrb_params->features, VLAN, 1);
		wrb_params->vlan_tag = be_get_tx_vlan_tag(adapter, skb);
S
Sathya Perla 已提交
838 839
	}

840 841
	BE_WRB_F_SET(wrb_params->features, CRC, 1);
}
842

843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866
static void wrb_fill_hdr(struct be_adapter *adapter,
			 struct be_eth_hdr_wrb *hdr,
			 struct be_wrb_params *wrb_params,
			 struct sk_buff *skb)
{
	memset(hdr, 0, sizeof(*hdr));

	SET_TX_WRB_HDR_BITS(crc, hdr,
			    BE_WRB_F_GET(wrb_params->features, CRC));
	SET_TX_WRB_HDR_BITS(ipcs, hdr,
			    BE_WRB_F_GET(wrb_params->features, IPCS));
	SET_TX_WRB_HDR_BITS(tcpcs, hdr,
			    BE_WRB_F_GET(wrb_params->features, TCPCS));
	SET_TX_WRB_HDR_BITS(udpcs, hdr,
			    BE_WRB_F_GET(wrb_params->features, UDPCS));

	SET_TX_WRB_HDR_BITS(lso, hdr,
			    BE_WRB_F_GET(wrb_params->features, LSO));
	SET_TX_WRB_HDR_BITS(lso6, hdr,
			    BE_WRB_F_GET(wrb_params->features, LSO6));
	SET_TX_WRB_HDR_BITS(lso_mss, hdr, wrb_params->lso_mss);

	/* Hack to skip HW VLAN tagging needs evt = 1, compl = 0. When this
	 * hack is not needed, the evt bit is set while ringing DB.
867
	 */
868 869 870 871 872 873 874 875
	SET_TX_WRB_HDR_BITS(event, hdr,
			    BE_WRB_F_GET(wrb_params->features, VLAN_SKIP_HW));
	SET_TX_WRB_HDR_BITS(vlan, hdr,
			    BE_WRB_F_GET(wrb_params->features, VLAN));
	SET_TX_WRB_HDR_BITS(vlan_tag, hdr, wrb_params->vlan_tag);

	SET_TX_WRB_HDR_BITS(num_wrb, hdr, skb_wrb_cnt(skb));
	SET_TX_WRB_HDR_BITS(len, hdr, skb->len);
V
Venkata Duvvuru 已提交
876 877
	SET_TX_WRB_HDR_BITS(mgmt, hdr,
			    BE_WRB_F_GET(wrb_params->features, OS2BMC));
S
Sathya Perla 已提交
878 879
}

I
Ivan Vecera 已提交
880
static void unmap_tx_frag(struct device *dev, struct be_eth_wrb *wrb,
881
			  bool unmap_single)
882 883
{
	dma_addr_t dma;
884
	u32 frag_len = le32_to_cpu(wrb->frag_len);
885 886


887 888 889
	dma = (u64)le32_to_cpu(wrb->frag_pa_hi) << 32 |
		(u64)le32_to_cpu(wrb->frag_pa_lo);
	if (frag_len) {
890
		if (unmap_single)
891
			dma_unmap_single(dev, dma, frag_len, DMA_TO_DEVICE);
892
		else
893
			dma_unmap_page(dev, dma, frag_len, DMA_TO_DEVICE);
894 895
	}
}
S
Sathya Perla 已提交
896

897
/* Grab a WRB header for xmit */
898
static u32 be_tx_get_wrb_hdr(struct be_tx_obj *txo)
899
{
900
	u32 head = txo->q.head;
901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943

	queue_head_inc(&txo->q);
	return head;
}

/* Set up the WRB header for xmit */
static void be_tx_setup_wrb_hdr(struct be_adapter *adapter,
				struct be_tx_obj *txo,
				struct be_wrb_params *wrb_params,
				struct sk_buff *skb, u16 head)
{
	u32 num_frags = skb_wrb_cnt(skb);
	struct be_queue_info *txq = &txo->q;
	struct be_eth_hdr_wrb *hdr = queue_index_node(txq, head);

	wrb_fill_hdr(adapter, hdr, wrb_params, skb);
	be_dws_cpu_to_le(hdr, sizeof(*hdr));

	BUG_ON(txo->sent_skb_list[head]);
	txo->sent_skb_list[head] = skb;
	txo->last_req_hdr = head;
	atomic_add(num_frags, &txq->used);
	txo->last_req_wrb_cnt = num_frags;
	txo->pend_wrb_cnt += num_frags;
}

/* Setup a WRB fragment (buffer descriptor) for xmit */
static void be_tx_setup_wrb_frag(struct be_tx_obj *txo, dma_addr_t busaddr,
				 int len)
{
	struct be_eth_wrb *wrb;
	struct be_queue_info *txq = &txo->q;

	wrb = queue_head_node(txq);
	wrb_fill(wrb, busaddr, len);
	queue_head_inc(txq);
}

/* Bring the queue back to the state it was in before be_xmit_enqueue() routine
 * was invoked. The producer index is restored to the previous packet and the
 * WRBs of the current packet are unmapped. Invoked to handle tx setup errors.
 */
static void be_xmit_restore(struct be_adapter *adapter,
944
			    struct be_tx_obj *txo, u32 head, bool map_single,
945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
			    u32 copied)
{
	struct device *dev;
	struct be_eth_wrb *wrb;
	struct be_queue_info *txq = &txo->q;

	dev = &adapter->pdev->dev;
	txq->head = head;

	/* skip the first wrb (hdr); it's not mapped */
	queue_head_inc(txq);
	while (copied) {
		wrb = queue_head_node(txq);
		unmap_tx_frag(dev, wrb, map_single);
		map_single = false;
		copied -= le32_to_cpu(wrb->frag_len);
		queue_head_inc(txq);
	}

	txq->head = head;
}

/* Enqueue the given packet for transmit. This routine allocates WRBs for the
 * packet, dma maps the packet buffers and sets up the WRBs. Returns the number
 * of WRBs used up by the packet.
 */
971
static u32 be_xmit_enqueue(struct be_adapter *adapter, struct be_tx_obj *txo,
972 973
			   struct sk_buff *skb,
			   struct be_wrb_params *wrb_params)
S
Sathya Perla 已提交
974
{
975
	u32 i, copied = 0, wrb_cnt = skb_wrb_cnt(skb);
I
Ivan Vecera 已提交
976
	struct device *dev = &adapter->pdev->dev;
977
	struct be_queue_info *txq = &txo->q;
978
	bool map_single = false;
979
	u32 head = txq->head;
980 981
	dma_addr_t busaddr;
	int len;
S
Sathya Perla 已提交
982

983
	head = be_tx_get_wrb_hdr(txo);
S
Sathya Perla 已提交
984

985
	if (skb->len > skb->data_len) {
986
		len = skb_headlen(skb);
987

I
Ivan Vecera 已提交
988 989
		busaddr = dma_map_single(dev, skb->data, len, DMA_TO_DEVICE);
		if (dma_mapping_error(dev, busaddr))
990 991
			goto dma_err;
		map_single = true;
992
		be_tx_setup_wrb_frag(txo, busaddr, len);
993 994
		copied += len;
	}
S
Sathya Perla 已提交
995

996
	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
997
		const struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i];
998
		len = skb_frag_size(frag);
999

1000
		busaddr = skb_frag_dma_map(dev, frag, 0, len, DMA_TO_DEVICE);
I
Ivan Vecera 已提交
1001
		if (dma_mapping_error(dev, busaddr))
1002
			goto dma_err;
1003 1004
		be_tx_setup_wrb_frag(txo, busaddr, len);
		copied += len;
S
Sathya Perla 已提交
1005 1006
	}

1007
	be_tx_setup_wrb_hdr(adapter, txo, wrb_params, skb, head);
S
Sathya Perla 已提交
1008

1009 1010
	be_tx_stats_update(txo, skb);
	return wrb_cnt;
S
Sathya Perla 已提交
1011

1012
dma_err:
1013 1014
	adapter->drv_stats.dma_map_errors++;
	be_xmit_restore(adapter, txo, head, map_single, copied);
1015
	return 0;
S
Sathya Perla 已提交
1016 1017
}

1018 1019 1020 1021 1022
static inline int qnq_async_evt_rcvd(struct be_adapter *adapter)
{
	return adapter->flags & BE_FLAGS_QNQ_ASYNC_EVT_RCVD;
}

1023
static struct sk_buff *be_insert_vlan_in_pkt(struct be_adapter *adapter,
1024
					     struct sk_buff *skb,
1025 1026
					     struct be_wrb_params
					     *wrb_params)
1027 1028 1029 1030 1031 1032 1033
{
	u16 vlan_tag = 0;

	skb = skb_share_check(skb, GFP_ATOMIC);
	if (unlikely(!skb))
		return skb;

1034
	if (skb_vlan_tag_present(skb))
1035
		vlan_tag = be_get_tx_vlan_tag(adapter, skb);
1036 1037 1038 1039 1040 1041 1042

	if (qnq_async_evt_rcvd(adapter) && adapter->pvid) {
		if (!vlan_tag)
			vlan_tag = adapter->pvid;
		/* f/w workaround to set skip_hw_vlan = 1, informs the F/W to
		 * skip VLAN insertion
		 */
1043
		BE_WRB_F_SET(wrb_params->features, VLAN_SKIP_HW, 1);
1044
	}
1045 1046

	if (vlan_tag) {
1047 1048
		skb = vlan_insert_tag_set_proto(skb, htons(ETH_P_8021Q),
						vlan_tag);
1049 1050 1051 1052 1053 1054 1055 1056
		if (unlikely(!skb))
			return skb;
		skb->vlan_tci = 0;
	}

	/* Insert the outer VLAN, if any */
	if (adapter->qnq_vid) {
		vlan_tag = adapter->qnq_vid;
1057 1058
		skb = vlan_insert_tag_set_proto(skb, htons(ETH_P_8021Q),
						vlan_tag);
1059 1060
		if (unlikely(!skb))
			return skb;
1061
		BE_WRB_F_SET(wrb_params->features, VLAN_SKIP_HW, 1);
1062 1063
	}

1064 1065 1066
	return skb;
}

1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
static bool be_ipv6_exthdr_check(struct sk_buff *skb)
{
	struct ethhdr *eh = (struct ethhdr *)skb->data;
	u16 offset = ETH_HLEN;

	if (eh->h_proto == htons(ETH_P_IPV6)) {
		struct ipv6hdr *ip6h = (struct ipv6hdr *)(skb->data + offset);

		offset += sizeof(struct ipv6hdr);
		if (ip6h->nexthdr != NEXTHDR_TCP &&
		    ip6h->nexthdr != NEXTHDR_UDP) {
			struct ipv6_opt_hdr *ehdr =
K
Kalesh AP 已提交
1079
				(struct ipv6_opt_hdr *)(skb->data + offset);
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090

			/* offending pkt: 2nd byte following IPv6 hdr is 0xff */
			if (ehdr->hdrlen == 0xff)
				return true;
		}
	}
	return false;
}

static int be_vlan_tag_tx_chk(struct be_adapter *adapter, struct sk_buff *skb)
{
1091
	return skb_vlan_tag_present(skb) || adapter->pvid || adapter->qnq_vid;
1092 1093
}

1094
static int be_ipv6_tx_stall_chk(struct be_adapter *adapter, struct sk_buff *skb)
1095
{
1096
	return BE3_chip(adapter) && be_ipv6_exthdr_check(skb);
1097 1098
}

1099 1100
static struct sk_buff *be_lancer_xmit_workarounds(struct be_adapter *adapter,
						  struct sk_buff *skb,
1101 1102
						  struct be_wrb_params
						  *wrb_params)
S
Sathya Perla 已提交
1103
{
1104
	struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data;
1105 1106
	unsigned int eth_hdr_len;
	struct iphdr *ip;
1107

1108 1109
	/* For padded packets, BE HW modifies tot_len field in IP header
	 * incorrecly when VLAN tag is inserted by HW.
1110
	 * For padded packets, Lancer computes incorrect checksum.
1111
	 */
1112 1113
	eth_hdr_len = ntohs(skb->protocol) == ETH_P_8021Q ?
						VLAN_ETH_HLEN : ETH_HLEN;
1114
	if (skb->len <= 60 &&
1115
	    (lancer_chip(adapter) || skb_vlan_tag_present(skb)) &&
1116
	    is_ipv4_pkt(skb)) {
1117 1118 1119
		ip = (struct iphdr *)ip_hdr(skb);
		pskb_trim(skb, eth_hdr_len + ntohs(ip->tot_len));
	}
1120

1121
	/* If vlan tag is already inlined in the packet, skip HW VLAN
1122
	 * tagging in pvid-tagging mode
1123
	 */
1124
	if (be_pvid_tagging_enabled(adapter) &&
1125
	    veh->h_vlan_proto == htons(ETH_P_8021Q))
1126
		BE_WRB_F_SET(wrb_params->features, VLAN_SKIP_HW, 1);
1127

1128 1129 1130 1131 1132
	/* HW has a bug wherein it will calculate CSUM for VLAN
	 * pkts even though it is disabled.
	 * Manually insert VLAN in pkt.
	 */
	if (skb->ip_summed != CHECKSUM_PARTIAL &&
1133
	    skb_vlan_tag_present(skb)) {
1134
		skb = be_insert_vlan_in_pkt(adapter, skb, wrb_params);
1135
		if (unlikely(!skb))
1136
			goto err;
1137 1138 1139 1140 1141 1142 1143
	}

	/* HW may lockup when VLAN HW tagging is requested on
	 * certain ipv6 packets. Drop such pkts if the HW workaround to
	 * skip HW tagging is not enabled by FW.
	 */
	if (unlikely(be_ipv6_tx_stall_chk(adapter, skb) &&
K
Kalesh AP 已提交
1144 1145
		     (adapter->pvid || adapter->qnq_vid) &&
		     !qnq_async_evt_rcvd(adapter)))
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
		goto tx_drop;

	/* Manual VLAN tag insertion to prevent:
	 * ASIC lockup when the ASIC inserts VLAN tag into
	 * certain ipv6 packets. Insert VLAN tags in driver,
	 * and set event, completion, vlan bits accordingly
	 * in the Tx WRB.
	 */
	if (be_ipv6_tx_stall_chk(adapter, skb) &&
	    be_vlan_tag_tx_chk(adapter, skb)) {
1156
		skb = be_insert_vlan_in_pkt(adapter, skb, wrb_params);
1157
		if (unlikely(!skb))
1158
			goto err;
1159 1160
	}

1161 1162 1163
	return skb;
tx_drop:
	dev_kfree_skb_any(skb);
1164
err:
1165 1166 1167
	return NULL;
}

1168 1169
static struct sk_buff *be_xmit_workarounds(struct be_adapter *adapter,
					   struct sk_buff *skb,
1170
					   struct be_wrb_params *wrb_params)
1171
{
1172 1173
	int err;

1174 1175 1176 1177
	/* Lancer, SH and BE3 in SRIOV mode have a bug wherein
	 * packets that are 32b or less may cause a transmit stall
	 * on that port. The workaround is to pad such packets
	 * (len <= 32 bytes) to a minimum length of 36b.
1178
	 */
1179
	if (skb->len <= 32) {
1180
		if (skb_put_padto(skb, 36))
1181 1182 1183 1184
			return NULL;
	}

	if (BEx_chip(adapter) || lancer_chip(adapter)) {
1185
		skb = be_lancer_xmit_workarounds(adapter, skb, wrb_params);
1186 1187 1188 1189
		if (!skb)
			return NULL;
	}

1190 1191 1192 1193 1194 1195 1196
	/* The stack can send us skbs with length greater than
	 * what the HW can handle. Trim the extra bytes.
	 */
	WARN_ON_ONCE(skb->len > BE_MAX_GSO_SIZE);
	err = pskb_trim(skb, BE_MAX_GSO_SIZE);
	WARN_ON(err);

1197 1198 1199
	return skb;
}

1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210
static void be_xmit_flush(struct be_adapter *adapter, struct be_tx_obj *txo)
{
	struct be_queue_info *txq = &txo->q;
	struct be_eth_hdr_wrb *hdr = queue_index_node(txq, txo->last_req_hdr);

	/* Mark the last request eventable if it hasn't been marked already */
	if (!(hdr->dw[2] & cpu_to_le32(TX_HDR_WRB_EVT)))
		hdr->dw[2] |= cpu_to_le32(TX_HDR_WRB_EVT | TX_HDR_WRB_COMPL);

	/* compose a dummy wrb if there are odd set of wrbs to notify */
	if (!lancer_chip(adapter) && (txo->pend_wrb_cnt & 1)) {
1211
		wrb_fill_dummy(queue_head_node(txq));
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223
		queue_head_inc(txq);
		atomic_inc(&txq->used);
		txo->pend_wrb_cnt++;
		hdr->dw[2] &= ~cpu_to_le32(TX_HDR_WRB_NUM_MASK <<
					   TX_HDR_WRB_NUM_SHIFT);
		hdr->dw[2] |= cpu_to_le32((txo->last_req_wrb_cnt + 1) <<
					  TX_HDR_WRB_NUM_SHIFT);
	}
	be_txq_notify(adapter, txo, txo->pend_wrb_cnt);
	txo->pend_wrb_cnt = 0;
}

V
Venkata Duvvuru 已提交
1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
/* OS2BMC related */

#define DHCP_CLIENT_PORT	68
#define DHCP_SERVER_PORT	67
#define NET_BIOS_PORT1		137
#define NET_BIOS_PORT2		138
#define DHCPV6_RAS_PORT		547

#define is_mc_allowed_on_bmc(adapter, eh)	\
	(!is_multicast_filt_enabled(adapter) &&	\
	 is_multicast_ether_addr(eh->h_dest) &&	\
	 !is_broadcast_ether_addr(eh->h_dest))

#define is_bc_allowed_on_bmc(adapter, eh)	\
	(!is_broadcast_filt_enabled(adapter) &&	\
	 is_broadcast_ether_addr(eh->h_dest))

#define is_arp_allowed_on_bmc(adapter, skb)	\
	(is_arp(skb) && is_arp_filt_enabled(adapter))

#define is_broadcast_packet(eh, adapter)	\
		(is_multicast_ether_addr(eh->h_dest) && \
		!compare_ether_addr(eh->h_dest, adapter->netdev->broadcast))

#define is_arp(skb)	(skb->protocol == htons(ETH_P_ARP))

#define is_arp_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & (BMC_FILT_BROADCAST_ARP))

#define is_dhcp_client_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_BROADCAST_DHCP_CLIENT)

#define is_dhcp_srvr_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_BROADCAST_DHCP_SERVER)

#define is_nbios_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_BROADCAST_NET_BIOS)

#define is_ipv6_na_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask &	\
			BMC_FILT_MULTICAST_IPV6_NEIGH_ADVER)

#define is_ipv6_ra_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_MULTICAST_IPV6_RA)

#define is_ipv6_ras_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_MULTICAST_IPV6_RAS)

#define is_broadcast_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_BROADCAST)

#define is_multicast_filt_enabled(adapter)	\
		(adapter->bmc_filt_mask & BMC_FILT_MULTICAST)

static bool be_send_pkt_to_bmc(struct be_adapter *adapter,
			       struct sk_buff **skb)
{
	struct ethhdr *eh = (struct ethhdr *)(*skb)->data;
	bool os2bmc = false;

	if (!be_is_os2bmc_enabled(adapter))
		goto done;

	if (!is_multicast_ether_addr(eh->h_dest))
		goto done;

	if (is_mc_allowed_on_bmc(adapter, eh) ||
	    is_bc_allowed_on_bmc(adapter, eh) ||
	    is_arp_allowed_on_bmc(adapter, (*skb))) {
		os2bmc = true;
		goto done;
	}

	if ((*skb)->protocol == htons(ETH_P_IPV6)) {
		struct ipv6hdr *hdr = ipv6_hdr((*skb));
		u8 nexthdr = hdr->nexthdr;

		if (nexthdr == IPPROTO_ICMPV6) {
			struct icmp6hdr *icmp6 = icmp6_hdr((*skb));

			switch (icmp6->icmp6_type) {
			case NDISC_ROUTER_ADVERTISEMENT:
				os2bmc = is_ipv6_ra_filt_enabled(adapter);
				goto done;
			case NDISC_NEIGHBOUR_ADVERTISEMENT:
				os2bmc = is_ipv6_na_filt_enabled(adapter);
				goto done;
			default:
				break;
			}
		}
	}

	if (is_udp_pkt((*skb))) {
		struct udphdr *udp = udp_hdr((*skb));

1320
		switch (ntohs(udp->dest)) {
V
Venkata Duvvuru 已提交
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347
		case DHCP_CLIENT_PORT:
			os2bmc = is_dhcp_client_filt_enabled(adapter);
			goto done;
		case DHCP_SERVER_PORT:
			os2bmc = is_dhcp_srvr_filt_enabled(adapter);
			goto done;
		case NET_BIOS_PORT1:
		case NET_BIOS_PORT2:
			os2bmc = is_nbios_filt_enabled(adapter);
			goto done;
		case DHCPV6_RAS_PORT:
			os2bmc = is_ipv6_ras_filt_enabled(adapter);
			goto done;
		default:
			break;
		}
	}
done:
	/* For packets over a vlan, which are destined
	 * to BMC, asic expects the vlan to be inline in the packet.
	 */
	if (os2bmc)
		*skb = be_insert_vlan_in_pkt(adapter, *skb, NULL);

	return os2bmc;
}

1348 1349 1350
static netdev_tx_t be_xmit(struct sk_buff *skb, struct net_device *netdev)
{
	struct be_adapter *adapter = netdev_priv(netdev);
1351 1352
	u16 q_idx = skb_get_queue_mapping(skb);
	struct be_tx_obj *txo = &adapter->tx_obj[q_idx];
1353 1354
	struct be_wrb_params wrb_params = { 0 };
	bool flush = !skb->xmit_more;
1355
	u16 wrb_cnt;
1356

1357
	skb = be_xmit_workarounds(adapter, skb, &wrb_params);
1358 1359
	if (unlikely(!skb))
		goto drop;
S
Sathya Perla 已提交
1360

1361 1362 1363
	be_get_wrb_params_from_skb(adapter, skb, &wrb_params);

	wrb_cnt = be_xmit_enqueue(adapter, txo, skb, &wrb_params);
1364 1365 1366 1367
	if (unlikely(!wrb_cnt)) {
		dev_kfree_skb_any(skb);
		goto drop;
	}
E
Eric Dumazet 已提交
1368

V
Venkata Duvvuru 已提交
1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
	/* if os2bmc is enabled and if the pkt is destined to bmc,
	 * enqueue the pkt a 2nd time with mgmt bit set.
	 */
	if (be_send_pkt_to_bmc(adapter, &skb)) {
		BE_WRB_F_SET(wrb_params.features, OS2BMC, 1);
		wrb_cnt = be_xmit_enqueue(adapter, txo, skb, &wrb_params);
		if (unlikely(!wrb_cnt))
			goto drop;
		else
			skb_get(skb);
	}

1381
	if (be_is_txq_full(txo)) {
1382 1383 1384
		netif_stop_subqueue(netdev, q_idx);
		tx_stats(txo)->tx_stops++;
	}
1385

1386 1387
	if (flush || __netif_subqueue_stopped(netdev, q_idx))
		be_xmit_flush(adapter, txo);
S
Sathya Perla 已提交
1388

1389 1390 1391 1392 1393 1394
	return NETDEV_TX_OK;
drop:
	tx_stats(txo)->tx_drv_drops++;
	/* Flush the already enqueued tx requests */
	if (flush && txo->pend_wrb_cnt)
		be_xmit_flush(adapter, txo);
S
Sathya Perla 已提交
1395 1396 1397 1398 1399 1400 1401

	return NETDEV_TX_OK;
}

static int be_change_mtu(struct net_device *netdev, int new_mtu)
{
	struct be_adapter *adapter = netdev_priv(netdev);
K
Kalesh AP 已提交
1402 1403 1404 1405 1406
	struct device *dev = &adapter->pdev->dev;

	if (new_mtu < BE_MIN_MTU || new_mtu > BE_MAX_MTU) {
		dev_info(dev, "MTU must be between %d and %d bytes\n",
			 BE_MIN_MTU, BE_MAX_MTU);
S
Sathya Perla 已提交
1407 1408
		return -EINVAL;
	}
K
Kalesh AP 已提交
1409 1410

	dev_info(dev, "MTU changed from %d to %d bytes\n",
1411
		 netdev->mtu, new_mtu);
S
Sathya Perla 已提交
1412 1413 1414 1415
	netdev->mtu = new_mtu;
	return 0;
}

1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452
static inline bool be_in_all_promisc(struct be_adapter *adapter)
{
	return (adapter->if_flags & BE_IF_FLAGS_ALL_PROMISCUOUS) ==
			BE_IF_FLAGS_ALL_PROMISCUOUS;
}

static int be_set_vlan_promisc(struct be_adapter *adapter)
{
	struct device *dev = &adapter->pdev->dev;
	int status;

	if (adapter->if_flags & BE_IF_FLAGS_VLAN_PROMISCUOUS)
		return 0;

	status = be_cmd_rx_filter(adapter, BE_IF_FLAGS_VLAN_PROMISCUOUS, ON);
	if (!status) {
		dev_info(dev, "Enabled VLAN promiscuous mode\n");
		adapter->if_flags |= BE_IF_FLAGS_VLAN_PROMISCUOUS;
	} else {
		dev_err(dev, "Failed to enable VLAN promiscuous mode\n");
	}
	return status;
}

static int be_clear_vlan_promisc(struct be_adapter *adapter)
{
	struct device *dev = &adapter->pdev->dev;
	int status;

	status = be_cmd_rx_filter(adapter, BE_IF_FLAGS_VLAN_PROMISCUOUS, OFF);
	if (!status) {
		dev_info(dev, "Disabling VLAN promiscuous mode\n");
		adapter->if_flags &= ~BE_IF_FLAGS_VLAN_PROMISCUOUS;
	}
	return status;
}

S
Sathya Perla 已提交
1453
/*
1454 1455
 * A max of 64 (BE_NUM_VLANS_SUPPORTED) vlans can be configured in BE.
 * If the user configures more, place BE in vlan promiscuous mode.
S
Sathya Perla 已提交
1456
 */
S
Sathya Perla 已提交
1457
static int be_vid_config(struct be_adapter *adapter)
S
Sathya Perla 已提交
1458
{
V
Vasundhara Volam 已提交
1459
	struct device *dev = &adapter->pdev->dev;
S
Sathya Perla 已提交
1460
	u16 vids[BE_NUM_VLANS_SUPPORTED];
1461
	u16 num = 0, i = 0;
1462
	int status = 0;
1463

1464 1465
	/* No need to change the VLAN state if the I/F is in promiscuous */
	if (adapter->netdev->flags & IFF_PROMISC)
1466 1467
		return 0;

1468
	if (adapter->vlans_added > be_max_vlans(adapter))
1469
		return be_set_vlan_promisc(adapter);
1470

1471 1472 1473 1474 1475
	if (adapter->if_flags & BE_IF_FLAGS_VLAN_PROMISCUOUS) {
		status = be_clear_vlan_promisc(adapter);
		if (status)
			return status;
	}
1476
	/* Construct VLAN Table to give to HW */
1477 1478
	for_each_set_bit(i, adapter->vids, VLAN_N_VID)
		vids[num++] = cpu_to_le16(i);
1479

1480
	status = be_cmd_vlan_config(adapter, adapter->if_handle, vids, num, 0);
1481
	if (status) {
1482
		dev_err(dev, "Setting HW VLAN filtering failed\n");
1483
		/* Set to VLAN promisc mode as setting VLAN filter failed */
1484 1485
		if (addl_status(status) == MCC_ADDL_STATUS_INSUFFICIENT_VLANS ||
		    addl_status(status) ==
1486
				MCC_ADDL_STATUS_INSUFFICIENT_RESOURCES)
1487
			return be_set_vlan_promisc(adapter);
S
Sathya Perla 已提交
1488
	}
1489
	return status;
S
Sathya Perla 已提交
1490 1491
}

1492
static int be_vlan_add_vid(struct net_device *netdev, __be16 proto, u16 vid)
S
Sathya Perla 已提交
1493 1494
{
	struct be_adapter *adapter = netdev_priv(netdev);
A
Ajit Khaparde 已提交
1495
	int status = 0;
S
Sathya Perla 已提交
1496

1497 1498
	mutex_lock(&adapter->rx_filter_lock);

1499 1500
	/* Packets with VID 0 are always received by Lancer by default */
	if (lancer_chip(adapter) && vid == 0)
1501
		goto done;
1502

1503
	if (test_bit(vid, adapter->vids))
1504
		goto done;
1505

1506
	set_bit(vid, adapter->vids);
1507
	adapter->vlans_added++;
1508

1509 1510 1511 1512
	status = be_vid_config(adapter);
done:
	mutex_unlock(&adapter->rx_filter_lock);
	return status;
S
Sathya Perla 已提交
1513 1514
}

1515
static int be_vlan_rem_vid(struct net_device *netdev, __be16 proto, u16 vid)
S
Sathya Perla 已提交
1516 1517
{
	struct be_adapter *adapter = netdev_priv(netdev);
1518 1519 1520
	int status = 0;

	mutex_lock(&adapter->rx_filter_lock);
S
Sathya Perla 已提交
1521

1522 1523
	/* Packets with VID 0 are always received by Lancer by default */
	if (lancer_chip(adapter) && vid == 0)
1524
		goto done;
1525

1526
	if (!test_bit(vid, adapter->vids))
1527
		goto done;
1528

1529
	clear_bit(vid, adapter->vids);
1530 1531
	adapter->vlans_added--;

1532 1533 1534 1535
	status = be_vid_config(adapter);
done:
	mutex_unlock(&adapter->rx_filter_lock);
	return status;
S
Sathya Perla 已提交
1536 1537
}

1538 1539 1540 1541 1542 1543 1544
static void be_set_all_promisc(struct be_adapter *adapter)
{
	be_cmd_rx_filter(adapter, BE_IF_FLAGS_ALL_PROMISCUOUS, ON);
	adapter->if_flags |= BE_IF_FLAGS_ALL_PROMISCUOUS;
}

static void be_set_mc_promisc(struct be_adapter *adapter)
S
Sathya Perla 已提交
1545
{
1546
	int status;
S
Sathya Perla 已提交
1547

1548 1549
	if (adapter->if_flags & BE_IF_FLAGS_MCAST_PROMISCUOUS)
		return;
S
Sathya Perla 已提交
1550

1551 1552 1553 1554 1555
	status = be_cmd_rx_filter(adapter, BE_IF_FLAGS_MCAST_PROMISCUOUS, ON);
	if (!status)
		adapter->if_flags |= BE_IF_FLAGS_MCAST_PROMISCUOUS;
}

1556
static void be_set_uc_promisc(struct be_adapter *adapter)
1557 1558 1559
{
	int status;

1560 1561 1562 1563
	if (adapter->if_flags & BE_IF_FLAGS_PROMISCUOUS)
		return;

	status = be_cmd_rx_filter(adapter, BE_IF_FLAGS_PROMISCUOUS, ON);
1564
	if (!status)
1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605
		adapter->if_flags |= BE_IF_FLAGS_PROMISCUOUS;
}

static void be_clear_uc_promisc(struct be_adapter *adapter)
{
	int status;

	if (!(adapter->if_flags & BE_IF_FLAGS_PROMISCUOUS))
		return;

	status = be_cmd_rx_filter(adapter, BE_IF_FLAGS_PROMISCUOUS, OFF);
	if (!status)
		adapter->if_flags &= ~BE_IF_FLAGS_PROMISCUOUS;
}

/* The below 2 functions are the callback args for __dev_mc_sync/dev_uc_sync().
 * We use a single callback function for both sync and unsync. We really don't
 * add/remove addresses through this callback. But, we use it to detect changes
 * to the uc/mc lists. The entire uc/mc list is programmed in be_set_rx_mode().
 */
static int be_uc_list_update(struct net_device *netdev,
			     const unsigned char *addr)
{
	struct be_adapter *adapter = netdev_priv(netdev);

	adapter->update_uc_list = true;
	return 0;
}

static int be_mc_list_update(struct net_device *netdev,
			     const unsigned char *addr)
{
	struct be_adapter *adapter = netdev_priv(netdev);

	adapter->update_mc_list = true;
	return 0;
}

static void be_set_mc_list(struct be_adapter *adapter)
{
	struct net_device *netdev = adapter->netdev;
1606
	struct netdev_hw_addr *ha;
1607 1608 1609
	bool mc_promisc = false;
	int status;

1610
	netif_addr_lock_bh(netdev);
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628
	__dev_mc_sync(netdev, be_mc_list_update, be_mc_list_update);

	if (netdev->flags & IFF_PROMISC) {
		adapter->update_mc_list = false;
	} else if (netdev->flags & IFF_ALLMULTI ||
		   netdev_mc_count(netdev) > be_max_mc(adapter)) {
		/* Enable multicast promisc if num configured exceeds
		 * what we support
		 */
		mc_promisc = true;
		adapter->update_mc_list = false;
	} else if (adapter->if_flags & BE_IF_FLAGS_MCAST_PROMISCUOUS) {
		/* Update mc-list unconditionally if the iface was previously
		 * in mc-promisc mode and now is out of that mode.
		 */
		adapter->update_mc_list = true;
	}

1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
	if (adapter->update_mc_list) {
		int i = 0;

		/* cache the mc-list in adapter */
		netdev_for_each_mc_addr(ha, netdev) {
			ether_addr_copy(adapter->mc_list[i].mac, ha->addr);
			i++;
		}
		adapter->mc_count = netdev_mc_count(netdev);
	}
	netif_addr_unlock_bh(netdev);

1641
	if (mc_promisc) {
1642
		be_set_mc_promisc(adapter);
1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659
	} else if (adapter->update_mc_list) {
		status = be_cmd_rx_filter(adapter, BE_IF_FLAGS_MULTICAST, ON);
		if (!status)
			adapter->if_flags &= ~BE_IF_FLAGS_MCAST_PROMISCUOUS;
		else
			be_set_mc_promisc(adapter);

		adapter->update_mc_list = false;
	}
}

static void be_clear_mc_list(struct be_adapter *adapter)
{
	struct net_device *netdev = adapter->netdev;

	__dev_mc_unsync(netdev, NULL);
	be_cmd_rx_filter(adapter, BE_IF_FLAGS_MULTICAST, OFF);
1660
	adapter->mc_count = 0;
1661 1662
}

1663 1664 1665
static int be_uc_mac_add(struct be_adapter *adapter, int uc_idx)
{
	if (ether_addr_equal((u8 *)&adapter->uc_list[uc_idx * ETH_ALEN],
1666
			     adapter->dev_mac)) {
1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684
		adapter->pmac_id[uc_idx + 1] = adapter->pmac_id[0];
		return 0;
	}

	return be_cmd_pmac_add(adapter,
			       (u8 *)&adapter->uc_list[uc_idx * ETH_ALEN],
			       adapter->if_handle,
			       &adapter->pmac_id[uc_idx + 1], 0);
}

static void be_uc_mac_del(struct be_adapter *adapter, int pmac_id)
{
	if (pmac_id == adapter->pmac_id[0])
		return;

	be_cmd_pmac_del(adapter, adapter->if_handle, pmac_id, 0);
}

1685 1686
static void be_set_uc_list(struct be_adapter *adapter)
{
1687
	struct net_device *netdev = adapter->netdev;
1688
	struct netdev_hw_addr *ha;
1689
	bool uc_promisc = false;
1690
	int curr_uc_macs = 0, i;
1691

1692
	netif_addr_lock_bh(netdev);
1693
	__dev_uc_sync(netdev, be_uc_list_update, be_uc_list_update);
1694

1695 1696 1697 1698 1699 1700 1701 1702 1703 1704
	if (netdev->flags & IFF_PROMISC) {
		adapter->update_uc_list = false;
	} else if (netdev_uc_count(netdev) > (be_max_uc(adapter) - 1)) {
		uc_promisc = true;
		adapter->update_uc_list = false;
	}  else if (adapter->if_flags & BE_IF_FLAGS_PROMISCUOUS) {
		/* Update uc-list unconditionally if the iface was previously
		 * in uc-promisc mode and now is out of that mode.
		 */
		adapter->update_uc_list = true;
S
Sathya Perla 已提交
1705 1706
	}

1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
	if (adapter->update_uc_list) {
		i = 1; /* First slot is claimed by the Primary MAC */

		/* cache the uc-list in adapter array */
		netdev_for_each_uc_addr(ha, netdev) {
			ether_addr_copy(adapter->uc_list[i].mac, ha->addr);
			i++;
		}
		curr_uc_macs = netdev_uc_count(netdev);
	}
	netif_addr_unlock_bh(netdev);

1719 1720 1721 1722 1723
	if (uc_promisc) {
		be_set_uc_promisc(adapter);
	} else if (adapter->update_uc_list) {
		be_clear_uc_promisc(adapter);

1724
		for (i = 0; i < adapter->uc_macs; i++)
1725
			be_uc_mac_del(adapter, adapter->pmac_id[i + 1]);
1726

1727
		for (i = 0; i < curr_uc_macs; i++)
1728
			be_uc_mac_add(adapter, i);
1729
		adapter->uc_macs = curr_uc_macs;
1730
		adapter->update_uc_list = false;
1731 1732
	}
}
S
Sathya Perla 已提交
1733

1734 1735
static void be_clear_uc_list(struct be_adapter *adapter)
{
1736
	struct net_device *netdev = adapter->netdev;
1737
	int i;
1738

1739
	__dev_uc_unsync(netdev, NULL);
1740
	for (i = 0; i < adapter->uc_macs; i++)
1741 1742
		be_uc_mac_del(adapter, adapter->pmac_id[i + 1]);

1743 1744
	adapter->uc_macs = 0;
}
1745

1746
static void __be_set_rx_mode(struct be_adapter *adapter)
1747
{
1748 1749 1750
	struct net_device *netdev = adapter->netdev;

	mutex_lock(&adapter->rx_filter_lock);
1751

1752
	if (netdev->flags & IFF_PROMISC) {
1753 1754 1755 1756 1757 1758 1759 1760
		if (!be_in_all_promisc(adapter))
			be_set_all_promisc(adapter);
	} else if (be_in_all_promisc(adapter)) {
		/* We need to re-program the vlan-list or clear
		 * vlan-promisc mode (if needed) when the interface
		 * comes out of promisc mode.
		 */
		be_vid_config(adapter);
1761
	}
1762

1763
	be_set_uc_list(adapter);
1764
	be_set_mc_list(adapter);
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775

	mutex_unlock(&adapter->rx_filter_lock);
}

static void be_work_set_rx_mode(struct work_struct *work)
{
	struct be_cmd_work *cmd_work =
				container_of(work, struct be_cmd_work, work);

	__be_set_rx_mode(cmd_work->adapter);
	kfree(cmd_work);
S
Sathya Perla 已提交
1776 1777
}

1778 1779 1780
static int be_set_vf_mac(struct net_device *netdev, int vf, u8 *mac)
{
	struct be_adapter *adapter = netdev_priv(netdev);
1781
	struct be_vf_cfg *vf_cfg = &adapter->vf_cfg[vf];
1782 1783
	int status;

1784
	if (!sriov_enabled(adapter))
1785 1786
		return -EPERM;

1787
	if (!is_valid_ether_addr(mac) || vf >= adapter->num_vfs)
1788 1789
		return -EINVAL;

1790 1791 1792 1793 1794 1795
	/* Proceed further only if user provided MAC is different
	 * from active MAC
	 */
	if (ether_addr_equal(mac, vf_cfg->mac_addr))
		return 0;

1796 1797 1798
	if (BEx_chip(adapter)) {
		be_cmd_pmac_del(adapter, vf_cfg->if_handle, vf_cfg->pmac_id,
				vf + 1);
1799

1800 1801
		status = be_cmd_pmac_add(adapter, mac, vf_cfg->if_handle,
					 &vf_cfg->pmac_id, vf + 1);
1802 1803 1804
	} else {
		status = be_cmd_set_mac(adapter, mac, vf_cfg->if_handle,
					vf + 1);
1805 1806
	}

1807 1808 1809 1810 1811
	if (status) {
		dev_err(&adapter->pdev->dev, "MAC %pM set on VF %d Failed: %#x",
			mac, vf, status);
		return be_cmd_status(status);
	}
1812

1813 1814 1815
	ether_addr_copy(vf_cfg->mac_addr, mac);

	return 0;
1816 1817
}

1818
static int be_get_vf_config(struct net_device *netdev, int vf,
1819
			    struct ifla_vf_info *vi)
1820 1821
{
	struct be_adapter *adapter = netdev_priv(netdev);
1822
	struct be_vf_cfg *vf_cfg = &adapter->vf_cfg[vf];
1823

1824
	if (!sriov_enabled(adapter))
1825 1826
		return -EPERM;

1827
	if (vf >= adapter->num_vfs)
1828 1829 1830
		return -EINVAL;

	vi->vf = vf;
1831 1832
	vi->max_tx_rate = vf_cfg->tx_rate;
	vi->min_tx_rate = 0;
1833 1834
	vi->vlan = vf_cfg->vlan_tag & VLAN_VID_MASK;
	vi->qos = vf_cfg->vlan_tag >> VLAN_PRIO_SHIFT;
1835
	memcpy(&vi->mac, vf_cfg->mac_addr, ETH_ALEN);
1836
	vi->linkstate = adapter->vf_cfg[vf].plink_tracking;
1837
	vi->spoofchk = adapter->vf_cfg[vf].spoofchk;
1838 1839 1840 1841

	return 0;
}

1842 1843 1844 1845 1846 1847 1848 1849
static int be_set_vf_tvt(struct be_adapter *adapter, int vf, u16 vlan)
{
	struct be_vf_cfg *vf_cfg = &adapter->vf_cfg[vf];
	u16 vids[BE_NUM_VLANS_SUPPORTED];
	int vf_if_id = vf_cfg->if_handle;
	int status;

	/* Enable Transparent VLAN Tagging */
1850
	status = be_cmd_set_hsw_config(adapter, vlan, vf + 1, vf_if_id, 0, 0);
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878
	if (status)
		return status;

	/* Clear pre-programmed VLAN filters on VF if any, if TVT is enabled */
	vids[0] = 0;
	status = be_cmd_vlan_config(adapter, vf_if_id, vids, 1, vf + 1);
	if (!status)
		dev_info(&adapter->pdev->dev,
			 "Cleared guest VLANs on VF%d", vf);

	/* After TVT is enabled, disallow VFs to program VLAN filters */
	if (vf_cfg->privileges & BE_PRIV_FILTMGMT) {
		status = be_cmd_set_fn_privileges(adapter, vf_cfg->privileges &
						  ~BE_PRIV_FILTMGMT, vf + 1);
		if (!status)
			vf_cfg->privileges &= ~BE_PRIV_FILTMGMT;
	}
	return 0;
}

static int be_clear_vf_tvt(struct be_adapter *adapter, int vf)
{
	struct be_vf_cfg *vf_cfg = &adapter->vf_cfg[vf];
	struct device *dev = &adapter->pdev->dev;
	int status;

	/* Reset Transparent VLAN Tagging. */
	status = be_cmd_set_hsw_config(adapter, BE_RESET_VLAN_TAG_ID, vf + 1,
1879
				       vf_cfg->if_handle, 0, 0);
1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897
	if (status)
		return status;

	/* Allow VFs to program VLAN filtering */
	if (!(vf_cfg->privileges & BE_PRIV_FILTMGMT)) {
		status = be_cmd_set_fn_privileges(adapter, vf_cfg->privileges |
						  BE_PRIV_FILTMGMT, vf + 1);
		if (!status) {
			vf_cfg->privileges |= BE_PRIV_FILTMGMT;
			dev_info(dev, "VF%d: FILTMGMT priv enabled", vf);
		}
	}

	dev_info(dev,
		 "Disable/re-enable i/f in VM to clear Transparent VLAN tag");
	return 0;
}

1898 1899
static int be_set_vf_vlan(struct net_device *netdev, int vf, u16 vlan, u8 qos,
			  __be16 vlan_proto)
1900 1901
{
	struct be_adapter *adapter = netdev_priv(netdev);
1902
	struct be_vf_cfg *vf_cfg = &adapter->vf_cfg[vf];
1903
	int status;
1904

1905
	if (!sriov_enabled(adapter))
1906 1907
		return -EPERM;

1908
	if (vf >= adapter->num_vfs || vlan > 4095 || qos > 7)
1909 1910
		return -EINVAL;

1911 1912 1913
	if (vlan_proto != htons(ETH_P_8021Q))
		return -EPROTONOSUPPORT;

1914 1915
	if (vlan || qos) {
		vlan |= qos << VLAN_PRIO_SHIFT;
1916
		status = be_set_vf_tvt(adapter, vf, vlan);
1917
	} else {
1918
		status = be_clear_vf_tvt(adapter, vf);
1919 1920
	}

1921 1922
	if (status) {
		dev_err(&adapter->pdev->dev,
1923 1924
			"VLAN %d config on VF %d failed : %#x\n", vlan, vf,
			status);
1925 1926 1927 1928 1929
		return be_cmd_status(status);
	}

	vf_cfg->vlan_tag = vlan;
	return 0;
1930 1931
}

1932 1933
static int be_set_vf_tx_rate(struct net_device *netdev, int vf,
			     int min_tx_rate, int max_tx_rate)
1934 1935
{
	struct be_adapter *adapter = netdev_priv(netdev);
1936 1937 1938 1939
	struct device *dev = &adapter->pdev->dev;
	int percent_rate, status = 0;
	u16 link_speed = 0;
	u8 link_status;
1940

1941
	if (!sriov_enabled(adapter))
1942 1943
		return -EPERM;

1944
	if (vf >= adapter->num_vfs)
1945 1946
		return -EINVAL;

1947 1948 1949
	if (min_tx_rate)
		return -EINVAL;

1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
	if (!max_tx_rate)
		goto config_qos;

	status = be_cmd_link_status_query(adapter, &link_speed,
					  &link_status, 0);
	if (status)
		goto err;

	if (!link_status) {
		dev_err(dev, "TX-rate setting not allowed when link is down\n");
1960
		status = -ENETDOWN;
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977
		goto err;
	}

	if (max_tx_rate < 100 || max_tx_rate > link_speed) {
		dev_err(dev, "TX-rate must be between 100 and %d Mbps\n",
			link_speed);
		status = -EINVAL;
		goto err;
	}

	/* On Skyhawk the QOS setting must be done only as a % value */
	percent_rate = link_speed / 100;
	if (skyhawk_chip(adapter) && (max_tx_rate % percent_rate)) {
		dev_err(dev, "TX-rate must be a multiple of %d Mbps\n",
			percent_rate);
		status = -EINVAL;
		goto err;
1978
	}
1979

1980 1981
config_qos:
	status = be_cmd_config_qos(adapter, max_tx_rate, link_speed, vf + 1);
1982
	if (status)
1983 1984 1985 1986 1987 1988 1989 1990
		goto err;

	adapter->vf_cfg[vf].tx_rate = max_tx_rate;
	return 0;

err:
	dev_err(dev, "TX-rate setting of %dMbps on VF%d failed\n",
		max_tx_rate, vf);
1991
	return be_cmd_status(status);
1992
}
1993

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
static int be_set_vf_link_state(struct net_device *netdev, int vf,
				int link_state)
{
	struct be_adapter *adapter = netdev_priv(netdev);
	int status;

	if (!sriov_enabled(adapter))
		return -EPERM;

	if (vf >= adapter->num_vfs)
		return -EINVAL;

	status = be_cmd_set_logical_link_config(adapter, link_state, vf+1);
2007 2008 2009 2010 2011
	if (status) {
		dev_err(&adapter->pdev->dev,
			"Link state change on VF %d failed: %#x\n", vf, status);
		return be_cmd_status(status);
	}
2012

2013 2014 2015
	adapter->vf_cfg[vf].plink_tracking = link_state;

	return 0;
2016
}
2017

2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
static int be_set_vf_spoofchk(struct net_device *netdev, int vf, bool enable)
{
	struct be_adapter *adapter = netdev_priv(netdev);
	struct be_vf_cfg *vf_cfg = &adapter->vf_cfg[vf];
	u8 spoofchk;
	int status;

	if (!sriov_enabled(adapter))
		return -EPERM;

	if (vf >= adapter->num_vfs)
		return -EINVAL;

	if (BEx_chip(adapter))
		return -EOPNOTSUPP;

	if (enable == vf_cfg->spoofchk)
		return 0;

	spoofchk = enable ? ENABLE_MAC_SPOOFCHK : DISABLE_MAC_SPOOFCHK;

	status = be_cmd_set_hsw_config(adapter, 0, vf + 1, vf_cfg->if_handle,
				       0, spoofchk);
	if (status) {
		dev_err(&adapter->pdev->dev,
			"Spoofchk change on VF %d failed: %#x\n", vf, status);
		return be_cmd_status(status);
	}

	vf_cfg->spoofchk = enable;
	return 0;
}

2051 2052
static void be_aic_update(struct be_aic_obj *aic, u64 rx_pkts, u64 tx_pkts,
			  ulong now)
S
Sathya Perla 已提交
2053
{
2054 2055 2056 2057
	aic->rx_pkts_prev = rx_pkts;
	aic->tx_reqs_prev = tx_pkts;
	aic->jiffies = now;
}
2058

2059
static int be_get_new_eqd(struct be_eq_obj *eqo)
2060
{
2061 2062
	struct be_adapter *adapter = eqo->adapter;
	int eqd, start;
2063 2064 2065
	struct be_aic_obj *aic;
	struct be_rx_obj *rxo;
	struct be_tx_obj *txo;
2066
	u64 rx_pkts = 0, tx_pkts = 0;
2067 2068
	ulong now;
	u32 pps, delta;
2069
	int i;
S
Sathya Perla 已提交
2070

2071 2072 2073 2074 2075 2076 2077
	aic = &adapter->aic_obj[eqo->idx];
	if (!aic->enable) {
		if (aic->jiffies)
			aic->jiffies = 0;
		eqd = aic->et_eqd;
		return eqd;
	}
S
Sathya Perla 已提交
2078

2079
	for_all_rx_queues_on_eq(adapter, eqo, rxo, i) {
2080
		do {
2081
			start = u64_stats_fetch_begin_irq(&rxo->stats.sync);
2082
			rx_pkts += rxo->stats.rx_pkts;
2083
		} while (u64_stats_fetch_retry_irq(&rxo->stats.sync, start));
2084
	}
S
Sathya Perla 已提交
2085

2086
	for_all_tx_queues_on_eq(adapter, eqo, txo, i) {
2087
		do {
2088
			start = u64_stats_fetch_begin_irq(&txo->stats.sync);
2089
			tx_pkts += txo->stats.tx_reqs;
2090
		} while (u64_stats_fetch_retry_irq(&txo->stats.sync, start));
2091
	}
S
Sathya Perla 已提交
2092

2093 2094 2095 2096 2097 2098 2099 2100
	/* Skip, if wrapped around or first calculation */
	now = jiffies;
	if (!aic->jiffies || time_before(now, aic->jiffies) ||
	    rx_pkts < aic->rx_pkts_prev ||
	    tx_pkts < aic->tx_reqs_prev) {
		be_aic_update(aic, rx_pkts, tx_pkts, now);
		return aic->prev_eqd;
	}
2101

2102 2103 2104
	delta = jiffies_to_msecs(now - aic->jiffies);
	if (delta == 0)
		return aic->prev_eqd;
S
Sathya Perla 已提交
2105

2106 2107 2108
	pps = (((u32)(rx_pkts - aic->rx_pkts_prev) * 1000) / delta) +
		(((u32)(tx_pkts - aic->tx_reqs_prev) * 1000) / delta);
	eqd = (pps / 15000) << 2;
2109

2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
	if (eqd < 8)
		eqd = 0;
	eqd = min_t(u32, eqd, aic->max_eqd);
	eqd = max_t(u32, eqd, aic->min_eqd);

	be_aic_update(aic, rx_pkts, tx_pkts, now);

	return eqd;
}

/* For Skyhawk-R only */
static u32 be_get_eq_delay_mult_enc(struct be_eq_obj *eqo)
{
	struct be_adapter *adapter = eqo->adapter;
	struct be_aic_obj *aic = &adapter->aic_obj[eqo->idx];
	ulong now = jiffies;
	int eqd;
	u32 mult_enc;

	if (!aic->enable)
		return 0;

2132
	if (jiffies_to_msecs(now - aic->jiffies) < 1)
2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
		eqd = aic->prev_eqd;
	else
		eqd = be_get_new_eqd(eqo);

	if (eqd > 100)
		mult_enc = R2I_DLY_ENC_1;
	else if (eqd > 60)
		mult_enc = R2I_DLY_ENC_2;
	else if (eqd > 20)
		mult_enc = R2I_DLY_ENC_3;
	else
		mult_enc = R2I_DLY_ENC_0;

	aic->prev_eqd = eqd;

	return mult_enc;
}

void be_eqd_update(struct be_adapter *adapter, bool force_update)
{
	struct be_set_eqd set_eqd[MAX_EVT_QS];
	struct be_aic_obj *aic;
	struct be_eq_obj *eqo;
	int i, num = 0, eqd;

	for_all_evt_queues(adapter, eqo, i) {
		aic = &adapter->aic_obj[eqo->idx];
		eqd = be_get_new_eqd(eqo);
		if (force_update || eqd != aic->prev_eqd) {
2162 2163 2164 2165 2166
			set_eqd[num].delay_multiplier = (eqd * 65)/100;
			set_eqd[num].eq_id = eqo->q.id;
			aic->prev_eqd = eqd;
			num++;
		}
2167
	}
2168 2169 2170

	if (num)
		be_cmd_modify_eqd(adapter, set_eqd, num);
S
Sathya Perla 已提交
2171 2172
}

2173
static void be_rx_stats_update(struct be_rx_obj *rxo,
2174
			       struct be_rx_compl_info *rxcp)
2175
{
2176
	struct be_rx_stats *stats = rx_stats(rxo);
2177

2178
	u64_stats_update_begin(&stats->sync);
2179
	stats->rx_compl++;
2180
	stats->rx_bytes += rxcp->pkt_size;
2181
	stats->rx_pkts++;
2182 2183
	if (rxcp->tunneled)
		stats->rx_vxlan_offload_pkts++;
2184
	if (rxcp->pkt_type == BE_MULTICAST_PACKET)
2185
		stats->rx_mcast_pkts++;
2186
	if (rxcp->err)
2187
		stats->rx_compl_err++;
2188
	u64_stats_update_end(&stats->sync);
2189 2190
}

2191
static inline bool csum_passed(struct be_rx_compl_info *rxcp)
2192
{
2193
	/* L4 checksum is not reliable for non TCP/UDP packets.
2194 2195
	 * Also ignore ipcksm for ipv6 pkts
	 */
2196
	return (rxcp->tcpf || rxcp->udpf) && rxcp->l4_csum &&
2197
		(rxcp->ip_csum || rxcp->ipv6) && !rxcp->err;
2198 2199
}

2200
static struct be_rx_page_info *get_rx_page_info(struct be_rx_obj *rxo)
S
Sathya Perla 已提交
2201
{
S
Sathya Perla 已提交
2202
	struct be_adapter *adapter = rxo->adapter;
S
Sathya Perla 已提交
2203
	struct be_rx_page_info *rx_page_info;
2204
	struct be_queue_info *rxq = &rxo->q;
2205
	u32 frag_idx = rxq->tail;
S
Sathya Perla 已提交
2206

2207
	rx_page_info = &rxo->page_info_tbl[frag_idx];
S
Sathya Perla 已提交
2208 2209
	BUG_ON(!rx_page_info->page);

2210
	if (rx_page_info->last_frag) {
I
Ivan Vecera 已提交
2211 2212 2213
		dma_unmap_page(&adapter->pdev->dev,
			       dma_unmap_addr(rx_page_info, bus),
			       adapter->big_page_size, DMA_FROM_DEVICE);
2214 2215 2216 2217 2218
		rx_page_info->last_frag = false;
	} else {
		dma_sync_single_for_cpu(&adapter->pdev->dev,
					dma_unmap_addr(rx_page_info, bus),
					rx_frag_size, DMA_FROM_DEVICE);
A
Ajit Khaparde 已提交
2219
	}
S
Sathya Perla 已提交
2220

2221
	queue_tail_inc(rxq);
S
Sathya Perla 已提交
2222 2223 2224 2225 2226
	atomic_dec(&rxq->used);
	return rx_page_info;
}

/* Throwaway the data in the Rx completion */
S
Sathya Perla 已提交
2227 2228
static void be_rx_compl_discard(struct be_rx_obj *rxo,
				struct be_rx_compl_info *rxcp)
S
Sathya Perla 已提交
2229 2230
{
	struct be_rx_page_info *page_info;
2231
	u16 i, num_rcvd = rxcp->num_rcvd;
S
Sathya Perla 已提交
2232

2233
	for (i = 0; i < num_rcvd; i++) {
2234
		page_info = get_rx_page_info(rxo);
2235 2236
		put_page(page_info->page);
		memset(page_info, 0, sizeof(*page_info));
S
Sathya Perla 已提交
2237 2238 2239 2240 2241 2242 2243
	}
}

/*
 * skb_fill_rx_data forms a complete skb for an ether frame
 * indicated by rxcp.
 */
S
Sathya Perla 已提交
2244 2245
static void skb_fill_rx_data(struct be_rx_obj *rxo, struct sk_buff *skb,
			     struct be_rx_compl_info *rxcp)
S
Sathya Perla 已提交
2246 2247
{
	struct be_rx_page_info *page_info;
2248 2249
	u16 i, j;
	u16 hdr_len, curr_frag_len, remaining;
S
Sathya Perla 已提交
2250 2251
	u8 *start;

2252
	page_info = get_rx_page_info(rxo);
S
Sathya Perla 已提交
2253 2254 2255 2256
	start = page_address(page_info->page) + page_info->page_offset;
	prefetch(start);

	/* Copy data in the first descriptor of this completion */
2257
	curr_frag_len = min(rxcp->pkt_size, rx_frag_size);
S
Sathya Perla 已提交
2258 2259 2260

	skb->len = curr_frag_len;
	if (curr_frag_len <= BE_HDR_LEN) { /* tiny packet */
2261
		memcpy(skb->data, start, curr_frag_len);
S
Sathya Perla 已提交
2262 2263 2264 2265 2266
		/* Complete packet has now been moved to data */
		put_page(page_info->page);
		skb->data_len = 0;
		skb->tail += curr_frag_len;
	} else {
2267 2268
		hdr_len = ETH_HLEN;
		memcpy(skb->data, start, hdr_len);
S
Sathya Perla 已提交
2269
		skb_shinfo(skb)->nr_frags = 1;
2270
		skb_frag_set_page(skb, 0, page_info->page);
S
Sathya Perla 已提交
2271 2272
		skb_shinfo(skb)->frags[0].page_offset =
					page_info->page_offset + hdr_len;
2273 2274
		skb_frag_size_set(&skb_shinfo(skb)->frags[0],
				  curr_frag_len - hdr_len);
S
Sathya Perla 已提交
2275
		skb->data_len = curr_frag_len - hdr_len;
E
Eric Dumazet 已提交
2276
		skb->truesize += rx_frag_size;
S
Sathya Perla 已提交
2277 2278
		skb->tail += hdr_len;
	}
A
Ajit Khaparde 已提交
2279
	page_info->page = NULL;
S
Sathya Perla 已提交
2280

2281 2282 2283
	if (rxcp->pkt_size <= rx_frag_size) {
		BUG_ON(rxcp->num_rcvd != 1);
		return;
S
Sathya Perla 已提交
2284 2285 2286
	}

	/* More frags present for this completion */
2287 2288
	remaining = rxcp->pkt_size - curr_frag_len;
	for (i = 1, j = 0; i < rxcp->num_rcvd; i++) {
2289
		page_info = get_rx_page_info(rxo);
2290
		curr_frag_len = min(remaining, rx_frag_size);
S
Sathya Perla 已提交
2291

2292 2293 2294 2295
		/* Coalesce all frags from the same physical page in one slot */
		if (page_info->page_offset == 0) {
			/* Fresh page */
			j++;
2296
			skb_frag_set_page(skb, j, page_info->page);
2297 2298
			skb_shinfo(skb)->frags[j].page_offset =
							page_info->page_offset;
E
Eric Dumazet 已提交
2299
			skb_frag_size_set(&skb_shinfo(skb)->frags[j], 0);
2300 2301 2302 2303 2304
			skb_shinfo(skb)->nr_frags++;
		} else {
			put_page(page_info->page);
		}

E
Eric Dumazet 已提交
2305
		skb_frag_size_add(&skb_shinfo(skb)->frags[j], curr_frag_len);
S
Sathya Perla 已提交
2306 2307
		skb->len += curr_frag_len;
		skb->data_len += curr_frag_len;
E
Eric Dumazet 已提交
2308
		skb->truesize += rx_frag_size;
2309
		remaining -= curr_frag_len;
A
Ajit Khaparde 已提交
2310
		page_info->page = NULL;
S
Sathya Perla 已提交
2311
	}
2312
	BUG_ON(j > MAX_SKB_FRAGS);
S
Sathya Perla 已提交
2313 2314
}

2315
/* Process the RX completion indicated by rxcp when GRO is disabled */
2316
static void be_rx_compl_process(struct be_rx_obj *rxo, struct napi_struct *napi,
S
Sathya Perla 已提交
2317
				struct be_rx_compl_info *rxcp)
S
Sathya Perla 已提交
2318
{
S
Sathya Perla 已提交
2319
	struct be_adapter *adapter = rxo->adapter;
2320
	struct net_device *netdev = adapter->netdev;
S
Sathya Perla 已提交
2321
	struct sk_buff *skb;
2322

2323
	skb = netdev_alloc_skb_ip_align(netdev, BE_RX_SKB_ALLOC_SIZE);
2324
	if (unlikely(!skb)) {
2325
		rx_stats(rxo)->rx_drops_no_skbs++;
S
Sathya Perla 已提交
2326
		be_rx_compl_discard(rxo, rxcp);
S
Sathya Perla 已提交
2327 2328 2329
		return;
	}

S
Sathya Perla 已提交
2330
	skb_fill_rx_data(rxo, skb, rxcp);
S
Sathya Perla 已提交
2331

2332
	if (likely((netdev->features & NETIF_F_RXCSUM) && csum_passed(rxcp)))
2333
		skb->ip_summed = CHECKSUM_UNNECESSARY;
S
Somnath Kotur 已提交
2334 2335
	else
		skb_checksum_none_assert(skb);
S
Sathya Perla 已提交
2336

2337
	skb->protocol = eth_type_trans(skb, netdev);
2338
	skb_record_rx_queue(skb, rxo - &adapter->rx_obj[0]);
S
Sathya Perla 已提交
2339
	if (netdev->features & NETIF_F_RXHASH)
T
Tom Herbert 已提交
2340
		skb_set_hash(skb, rxcp->rss_hash, PKT_HASH_TYPE_L3);
2341

2342
	skb->csum_level = rxcp->tunneled;
2343
	skb_mark_napi_id(skb, napi);
S
Sathya Perla 已提交
2344

2345
	if (rxcp->vlanf)
2346
		__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), rxcp->vlan_tag);
A
Ajit Khaparde 已提交
2347 2348

	netif_receive_skb(skb);
S
Sathya Perla 已提交
2349 2350
}

2351
/* Process the RX completion indicated by rxcp when GRO is enabled */
J
Jingoo Han 已提交
2352 2353 2354
static void be_rx_compl_process_gro(struct be_rx_obj *rxo,
				    struct napi_struct *napi,
				    struct be_rx_compl_info *rxcp)
S
Sathya Perla 已提交
2355
{
S
Sathya Perla 已提交
2356
	struct be_adapter *adapter = rxo->adapter;
S
Sathya Perla 已提交
2357
	struct be_rx_page_info *page_info;
2358
	struct sk_buff *skb = NULL;
2359 2360
	u16 remaining, curr_frag_len;
	u16 i, j;
2361

S
Sathya Perla 已提交
2362
	skb = napi_get_frags(napi);
2363
	if (!skb) {
S
Sathya Perla 已提交
2364
		be_rx_compl_discard(rxo, rxcp);
2365 2366 2367
		return;
	}

2368 2369
	remaining = rxcp->pkt_size;
	for (i = 0, j = -1; i < rxcp->num_rcvd; i++) {
2370
		page_info = get_rx_page_info(rxo);
S
Sathya Perla 已提交
2371 2372 2373

		curr_frag_len = min(remaining, rx_frag_size);

2374 2375 2376 2377
		/* Coalesce all frags from the same physical page in one slot */
		if (i == 0 || page_info->page_offset == 0) {
			/* First frag or Fresh page */
			j++;
2378
			skb_frag_set_page(skb, j, page_info->page);
2379 2380
			skb_shinfo(skb)->frags[j].page_offset =
							page_info->page_offset;
E
Eric Dumazet 已提交
2381
			skb_frag_size_set(&skb_shinfo(skb)->frags[j], 0);
2382 2383 2384
		} else {
			put_page(page_info->page);
		}
E
Eric Dumazet 已提交
2385
		skb_frag_size_add(&skb_shinfo(skb)->frags[j], curr_frag_len);
E
Eric Dumazet 已提交
2386
		skb->truesize += rx_frag_size;
2387
		remaining -= curr_frag_len;
S
Sathya Perla 已提交
2388 2389
		memset(page_info, 0, sizeof(*page_info));
	}
2390
	BUG_ON(j > MAX_SKB_FRAGS);
S
Sathya Perla 已提交
2391

2392
	skb_shinfo(skb)->nr_frags = j + 1;
2393 2394
	skb->len = rxcp->pkt_size;
	skb->data_len = rxcp->pkt_size;
2395
	skb->ip_summed = CHECKSUM_UNNECESSARY;
2396
	skb_record_rx_queue(skb, rxo - &adapter->rx_obj[0]);
A
Ajit Khaparde 已提交
2397
	if (adapter->netdev->features & NETIF_F_RXHASH)
T
Tom Herbert 已提交
2398
		skb_set_hash(skb, rxcp->rss_hash, PKT_HASH_TYPE_L3);
2399

2400
	skb->csum_level = rxcp->tunneled;
2401

2402
	if (rxcp->vlanf)
2403
		__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), rxcp->vlan_tag);
A
Ajit Khaparde 已提交
2404

S
Sathya Perla 已提交
2405
	napi_gro_frags(napi);
2406 2407
}

S
Sathya Perla 已提交
2408 2409
static void be_parse_rx_compl_v1(struct be_eth_rx_compl *compl,
				 struct be_rx_compl_info *rxcp)
2410
{
2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421
	rxcp->pkt_size = GET_RX_COMPL_V1_BITS(pktsize, compl);
	rxcp->vlanf = GET_RX_COMPL_V1_BITS(vtp, compl);
	rxcp->err = GET_RX_COMPL_V1_BITS(err, compl);
	rxcp->tcpf = GET_RX_COMPL_V1_BITS(tcpf, compl);
	rxcp->udpf = GET_RX_COMPL_V1_BITS(udpf, compl);
	rxcp->ip_csum = GET_RX_COMPL_V1_BITS(ipcksm, compl);
	rxcp->l4_csum = GET_RX_COMPL_V1_BITS(l4_cksm, compl);
	rxcp->ipv6 = GET_RX_COMPL_V1_BITS(ip_version, compl);
	rxcp->num_rcvd = GET_RX_COMPL_V1_BITS(numfrags, compl);
	rxcp->pkt_type = GET_RX_COMPL_V1_BITS(cast_enc, compl);
	rxcp->rss_hash = GET_RX_COMPL_V1_BITS(rsshash, compl);
2422
	if (rxcp->vlanf) {
2423 2424
		rxcp->qnq = GET_RX_COMPL_V1_BITS(qnq, compl);
		rxcp->vlan_tag = GET_RX_COMPL_V1_BITS(vlan_tag, compl);
2425
	}
2426
	rxcp->port = GET_RX_COMPL_V1_BITS(port, compl);
2427
	rxcp->tunneled =
2428
		GET_RX_COMPL_V1_BITS(tunneled, compl);
2429 2430
}

S
Sathya Perla 已提交
2431 2432
static void be_parse_rx_compl_v0(struct be_eth_rx_compl *compl,
				 struct be_rx_compl_info *rxcp)
2433
{
2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444
	rxcp->pkt_size = GET_RX_COMPL_V0_BITS(pktsize, compl);
	rxcp->vlanf = GET_RX_COMPL_V0_BITS(vtp, compl);
	rxcp->err = GET_RX_COMPL_V0_BITS(err, compl);
	rxcp->tcpf = GET_RX_COMPL_V0_BITS(tcpf, compl);
	rxcp->udpf = GET_RX_COMPL_V0_BITS(udpf, compl);
	rxcp->ip_csum = GET_RX_COMPL_V0_BITS(ipcksm, compl);
	rxcp->l4_csum = GET_RX_COMPL_V0_BITS(l4_cksm, compl);
	rxcp->ipv6 = GET_RX_COMPL_V0_BITS(ip_version, compl);
	rxcp->num_rcvd = GET_RX_COMPL_V0_BITS(numfrags, compl);
	rxcp->pkt_type = GET_RX_COMPL_V0_BITS(cast_enc, compl);
	rxcp->rss_hash = GET_RX_COMPL_V0_BITS(rsshash, compl);
2445
	if (rxcp->vlanf) {
2446 2447
		rxcp->qnq = GET_RX_COMPL_V0_BITS(qnq, compl);
		rxcp->vlan_tag = GET_RX_COMPL_V0_BITS(vlan_tag, compl);
2448
	}
2449 2450
	rxcp->port = GET_RX_COMPL_V0_BITS(port, compl);
	rxcp->ip_frag = GET_RX_COMPL_V0_BITS(ip_frag, compl);
2451 2452 2453 2454 2455 2456 2457
}

static struct be_rx_compl_info *be_rx_compl_get(struct be_rx_obj *rxo)
{
	struct be_eth_rx_compl *compl = queue_tail_node(&rxo->cq);
	struct be_rx_compl_info *rxcp = &rxo->rxcp;
	struct be_adapter *adapter = rxo->adapter;
S
Sathya Perla 已提交
2458

2459 2460 2461 2462
	/* For checking the valid bit it is Ok to use either definition as the
	 * valid bit is at the same position in both v0 and v1 Rx compl */
	if (compl->dw[offsetof(struct amap_eth_rx_compl_v1, valid) / 32] == 0)
		return NULL;
S
Sathya Perla 已提交
2463

2464 2465
	rmb();
	be_dws_le_to_cpu(compl, sizeof(*compl));
S
Sathya Perla 已提交
2466

2467
	if (adapter->be3_native)
S
Sathya Perla 已提交
2468
		be_parse_rx_compl_v1(compl, rxcp);
2469
	else
S
Sathya Perla 已提交
2470
		be_parse_rx_compl_v0(compl, rxcp);
S
Sathya Perla 已提交
2471

2472 2473 2474
	if (rxcp->ip_frag)
		rxcp->l4_csum = 0;

2475
	if (rxcp->vlanf) {
2476 2477 2478 2479 2480
		/* In QNQ modes, if qnq bit is not set, then the packet was
		 * tagged only with the transparent outer vlan-tag and must
		 * not be treated as a vlan packet by host
		 */
		if (be_is_qnq_mode(adapter) && !rxcp->qnq)
2481
			rxcp->vlanf = 0;
S
Sathya Perla 已提交
2482

2483
		if (!lancer_chip(adapter))
2484
			rxcp->vlan_tag = swab16(rxcp->vlan_tag);
S
Sathya Perla 已提交
2485

2486
		if (adapter->pvid == (rxcp->vlan_tag & VLAN_VID_MASK) &&
2487
		    !test_bit(rxcp->vlan_tag, adapter->vids))
2488 2489
			rxcp->vlanf = 0;
	}
2490 2491 2492

	/* As the compl has been parsed, reset it; we wont touch it again */
	compl->dw[offsetof(struct amap_eth_rx_compl_v1, valid) / 32] = 0;
S
Sathya Perla 已提交
2493

2494
	queue_tail_inc(&rxo->cq);
S
Sathya Perla 已提交
2495 2496 2497
	return rxcp;
}

2498
static inline struct page *be_alloc_pages(u32 size, gfp_t gfp)
S
Sathya Perla 已提交
2499 2500
{
	u32 order = get_order(size);
2501

S
Sathya Perla 已提交
2502
	if (order > 0)
2503 2504
		gfp |= __GFP_COMP;
	return  alloc_pages(gfp, order);
S
Sathya Perla 已提交
2505 2506 2507 2508 2509 2510
}

/*
 * Allocate a page, split it to fragments of size rx_frag_size and post as
 * receive buffers to BE
 */
2511
static void be_post_rx_frags(struct be_rx_obj *rxo, gfp_t gfp, u32 frags_needed)
S
Sathya Perla 已提交
2512
{
2513
	struct be_adapter *adapter = rxo->adapter;
2514
	struct be_rx_page_info *page_info = NULL, *prev_page_info = NULL;
2515
	struct be_queue_info *rxq = &rxo->q;
S
Sathya Perla 已提交
2516
	struct page *pagep = NULL;
2517
	struct device *dev = &adapter->pdev->dev;
S
Sathya Perla 已提交
2518 2519
	struct be_eth_rx_d *rxd;
	u64 page_dmaaddr = 0, frag_dmaaddr;
2520
	u32 posted, page_offset = 0, notify = 0;
S
Sathya Perla 已提交
2521

2522
	page_info = &rxo->page_info_tbl[rxq->head];
2523
	for (posted = 0; posted < frags_needed && !page_info->page; posted++) {
S
Sathya Perla 已提交
2524
		if (!pagep) {
2525
			pagep = be_alloc_pages(adapter->big_page_size, gfp);
S
Sathya Perla 已提交
2526
			if (unlikely(!pagep)) {
2527
				rx_stats(rxo)->rx_post_fail++;
S
Sathya Perla 已提交
2528 2529
				break;
			}
2530 2531
			page_dmaaddr = dma_map_page(dev, pagep, 0,
						    adapter->big_page_size,
I
Ivan Vecera 已提交
2532
						    DMA_FROM_DEVICE);
2533 2534 2535
			if (dma_mapping_error(dev, page_dmaaddr)) {
				put_page(pagep);
				pagep = NULL;
2536
				adapter->drv_stats.dma_map_errors++;
2537 2538
				break;
			}
2539
			page_offset = 0;
S
Sathya Perla 已提交
2540 2541
		} else {
			get_page(pagep);
2542
			page_offset += rx_frag_size;
S
Sathya Perla 已提交
2543
		}
2544
		page_info->page_offset = page_offset;
S
Sathya Perla 已提交
2545 2546 2547
		page_info->page = pagep;

		rxd = queue_head_node(rxq);
2548
		frag_dmaaddr = page_dmaaddr + page_info->page_offset;
S
Sathya Perla 已提交
2549 2550 2551 2552 2553 2554 2555
		rxd->fragpa_lo = cpu_to_le32(frag_dmaaddr & 0xFFFFFFFF);
		rxd->fragpa_hi = cpu_to_le32(upper_32_bits(frag_dmaaddr));

		/* Any space left in the current big page for another frag? */
		if ((page_offset + rx_frag_size + rx_frag_size) >
					adapter->big_page_size) {
			pagep = NULL;
2556 2557 2558 2559
			page_info->last_frag = true;
			dma_unmap_addr_set(page_info, bus, page_dmaaddr);
		} else {
			dma_unmap_addr_set(page_info, bus, frag_dmaaddr);
S
Sathya Perla 已提交
2560
		}
2561 2562 2563

		prev_page_info = page_info;
		queue_head_inc(rxq);
S
Sathya Perla 已提交
2564
		page_info = &rxo->page_info_tbl[rxq->head];
S
Sathya Perla 已提交
2565
	}
2566 2567 2568 2569 2570 2571 2572 2573

	/* Mark the last frag of a page when we break out of the above loop
	 * with no more slots available in the RXQ
	 */
	if (pagep) {
		prev_page_info->last_frag = true;
		dma_unmap_addr_set(prev_page_info, bus, page_dmaaddr);
	}
S
Sathya Perla 已提交
2574 2575 2576

	if (posted) {
		atomic_add(posted, &rxq->used);
2577 2578
		if (rxo->rx_post_starved)
			rxo->rx_post_starved = false;
2579
		do {
2580
			notify = min(MAX_NUM_POST_ERX_DB, posted);
2581 2582 2583
			be_rxq_notify(adapter, rxq->id, notify);
			posted -= notify;
		} while (posted);
2584 2585
	} else if (atomic_read(&rxq->used) == 0) {
		/* Let be_worker replenish when memory is available */
2586
		rxo->rx_post_starved = true;
S
Sathya Perla 已提交
2587 2588 2589
	}
}

2590
static struct be_tx_compl_info *be_tx_compl_get(struct be_tx_obj *txo)
S
Sathya Perla 已提交
2591
{
2592 2593 2594
	struct be_queue_info *tx_cq = &txo->cq;
	struct be_tx_compl_info *txcp = &txo->txcp;
	struct be_eth_tx_compl *compl = queue_tail_node(tx_cq);
S
Sathya Perla 已提交
2595

2596
	if (compl->dw[offsetof(struct amap_eth_tx_compl, valid) / 32] == 0)
S
Sathya Perla 已提交
2597 2598
		return NULL;

2599
	/* Ensure load ordering of valid bit dword and other dwords below */
2600
	rmb();
2601
	be_dws_le_to_cpu(compl, sizeof(*compl));
S
Sathya Perla 已提交
2602

2603 2604
	txcp->status = GET_TX_COMPL_BITS(status, compl);
	txcp->end_index = GET_TX_COMPL_BITS(wrb_index, compl);
S
Sathya Perla 已提交
2605

2606
	compl->dw[offsetof(struct amap_eth_tx_compl, valid) / 32] = 0;
S
Sathya Perla 已提交
2607 2608 2609 2610
	queue_tail_inc(tx_cq);
	return txcp;
}

2611
static u16 be_tx_compl_process(struct be_adapter *adapter,
2612
			       struct be_tx_obj *txo, u16 last_index)
S
Sathya Perla 已提交
2613
{
2614
	struct sk_buff **sent_skbs = txo->sent_skb_list;
2615
	struct be_queue_info *txq = &txo->q;
2616 2617
	struct sk_buff *skb = NULL;
	bool unmap_skb_hdr = false;
2618
	struct be_eth_wrb *wrb;
2619 2620
	u16 num_wrbs = 0;
	u32 frag_index;
S
Sathya Perla 已提交
2621

2622
	do {
2623 2624 2625 2626 2627 2628 2629 2630 2631 2632
		if (sent_skbs[txq->tail]) {
			/* Free skb from prev req */
			if (skb)
				dev_consume_skb_any(skb);
			skb = sent_skbs[txq->tail];
			sent_skbs[txq->tail] = NULL;
			queue_tail_inc(txq);  /* skip hdr wrb */
			num_wrbs++;
			unmap_skb_hdr = true;
		}
2633
		wrb = queue_tail_node(txq);
2634
		frag_index = txq->tail;
I
Ivan Vecera 已提交
2635
		unmap_tx_frag(&adapter->pdev->dev, wrb,
2636
			      (unmap_skb_hdr && skb_headlen(skb)));
2637
		unmap_skb_hdr = false;
S
Sathya Perla 已提交
2638
		queue_tail_inc(txq);
2639 2640 2641
		num_wrbs++;
	} while (frag_index != last_index);
	dev_consume_skb_any(skb);
S
Sathya Perla 已提交
2642

2643
	return num_wrbs;
S
Sathya Perla 已提交
2644 2645
}

S
Sathya Perla 已提交
2646 2647
/* Return the number of events in the event queue */
static inline int events_get(struct be_eq_obj *eqo)
2648
{
S
Sathya Perla 已提交
2649 2650
	struct be_eq_entry *eqe;
	int num = 0;
2651

S
Sathya Perla 已提交
2652 2653 2654 2655
	do {
		eqe = queue_tail_node(&eqo->q);
		if (eqe->evt == 0)
			break;
2656

S
Sathya Perla 已提交
2657 2658 2659 2660 2661 2662 2663
		rmb();
		eqe->evt = 0;
		num++;
		queue_tail_inc(&eqo->q);
	} while (true);

	return num;
2664 2665
}

S
Sathya Perla 已提交
2666 2667
/* Leaves the EQ is disarmed state */
static void be_eq_clean(struct be_eq_obj *eqo)
2668
{
S
Sathya Perla 已提交
2669
	int num = events_get(eqo);
2670

2671
	be_eq_notify(eqo->adapter, eqo->q.id, false, true, num, 0);
2672 2673
}

2674 2675
/* Free posted rx buffers that were not used */
static void be_rxq_clean(struct be_rx_obj *rxo)
S
Sathya Perla 已提交
2676
{
2677
	struct be_queue_info *rxq = &rxo->q;
2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691
	struct be_rx_page_info *page_info;

	while (atomic_read(&rxq->used) > 0) {
		page_info = get_rx_page_info(rxo);
		put_page(page_info->page);
		memset(page_info, 0, sizeof(*page_info));
	}
	BUG_ON(atomic_read(&rxq->used));
	rxq->tail = 0;
	rxq->head = 0;
}

static void be_rx_cq_clean(struct be_rx_obj *rxo)
{
2692
	struct be_queue_info *rx_cq = &rxo->cq;
2693
	struct be_rx_compl_info *rxcp;
2694 2695
	struct be_adapter *adapter = rxo->adapter;
	int flush_wait = 0;
S
Sathya Perla 已提交
2696

2697 2698 2699 2700 2701 2702 2703 2704
	/* Consume pending rx completions.
	 * Wait for the flush completion (identified by zero num_rcvd)
	 * to arrive. Notify CQ even when there are no more CQ entries
	 * for HW to flush partially coalesced CQ entries.
	 * In Lancer, there is no need to wait for flush compl.
	 */
	for (;;) {
		rxcp = be_rx_compl_get(rxo);
K
Kalesh AP 已提交
2705
		if (!rxcp) {
2706 2707 2708
			if (lancer_chip(adapter))
				break;

2709 2710 2711
			if (flush_wait++ > 50 ||
			    be_check_error(adapter,
					   BE_ERROR_HW)) {
2712 2713 2714 2715 2716 2717 2718 2719
				dev_warn(&adapter->pdev->dev,
					 "did not receive flush compl\n");
				break;
			}
			be_cq_notify(adapter, rx_cq->id, true, 0);
			mdelay(1);
		} else {
			be_rx_compl_discard(rxo, rxcp);
2720
			be_cq_notify(adapter, rx_cq->id, false, 1);
2721 2722 2723
			if (rxcp->num_rcvd == 0)
				break;
		}
S
Sathya Perla 已提交
2724 2725
	}

2726 2727
	/* After cleanup, leave the CQ in unarmed state */
	be_cq_notify(adapter, rx_cq->id, false, 0);
S
Sathya Perla 已提交
2728 2729
}

S
Sathya Perla 已提交
2730
static void be_tx_compl_clean(struct be_adapter *adapter)
S
Sathya Perla 已提交
2731
{
2732
	struct device *dev = &adapter->pdev->dev;
2733
	u16 cmpl = 0, timeo = 0, num_wrbs = 0;
2734
	struct be_tx_compl_info *txcp;
S
Sathya Perla 已提交
2735
	struct be_queue_info *txq;
2736
	u32 end_idx, notified_idx;
2737
	struct be_tx_obj *txo;
S
Sathya Perla 已提交
2738
	int i, pending_txqs;
2739

2740
	/* Stop polling for compls when HW has been silent for 10ms */
2741
	do {
S
Sathya Perla 已提交
2742 2743 2744
		pending_txqs = adapter->num_tx_qs;

		for_all_tx_queues(adapter, txo, i) {
2745 2746
			cmpl = 0;
			num_wrbs = 0;
S
Sathya Perla 已提交
2747
			txq = &txo->q;
2748 2749 2750 2751
			while ((txcp = be_tx_compl_get(txo))) {
				num_wrbs +=
					be_tx_compl_process(adapter, txo,
							    txcp->end_index);
S
Sathya Perla 已提交
2752 2753 2754 2755 2756
				cmpl++;
			}
			if (cmpl) {
				be_cq_notify(adapter, txo->cq.id, false, cmpl);
				atomic_sub(num_wrbs, &txq->used);
2757
				timeo = 0;
S
Sathya Perla 已提交
2758
			}
2759
			if (!be_is_tx_compl_pending(txo))
S
Sathya Perla 已提交
2760
				pending_txqs--;
2761 2762
		}

2763 2764
		if (pending_txqs == 0 || ++timeo > 10 ||
		    be_check_error(adapter, BE_ERROR_HW))
2765 2766 2767 2768 2769
			break;

		mdelay(1);
	} while (true);

2770
	/* Free enqueued TX that was never notified to HW */
S
Sathya Perla 已提交
2771 2772 2773
	for_all_tx_queues(adapter, txo, i) {
		txq = &txo->q;

2774 2775 2776 2777
		if (atomic_read(&txq->used)) {
			dev_info(dev, "txq%d: cleaning %d pending tx-wrbs\n",
				 i, atomic_read(&txq->used));
			notified_idx = txq->tail;
S
Sathya Perla 已提交
2778
			end_idx = txq->tail;
2779 2780 2781 2782 2783
			index_adv(&end_idx, atomic_read(&txq->used) - 1,
				  txq->len);
			/* Use the tx-compl process logic to handle requests
			 * that were not sent to the HW.
			 */
S
Sathya Perla 已提交
2784 2785
			num_wrbs = be_tx_compl_process(adapter, txo, end_idx);
			atomic_sub(num_wrbs, &txq->used);
2786 2787 2788 2789 2790 2791 2792
			BUG_ON(atomic_read(&txq->used));
			txo->pend_wrb_cnt = 0;
			/* Since hw was never notified of these requests,
			 * reset TXQ indices
			 */
			txq->head = notified_idx;
			txq->tail = notified_idx;
S
Sathya Perla 已提交
2793
		}
2794
	}
S
Sathya Perla 已提交
2795 2796
}

S
Sathya Perla 已提交
2797 2798 2799 2800 2801 2802
static void be_evt_queues_destroy(struct be_adapter *adapter)
{
	struct be_eq_obj *eqo;
	int i;

	for_all_evt_queues(adapter, eqo, i) {
2803 2804
		if (eqo->q.created) {
			be_eq_clean(eqo);
S
Sathya Perla 已提交
2805
			be_cmd_q_destroy(adapter, &eqo->q, QTYPE_EQ);
2806
			napi_hash_del(&eqo->napi);
2807
			netif_napi_del(&eqo->napi);
2808
			free_cpumask_var(eqo->affinity_mask);
2809
		}
S
Sathya Perla 已提交
2810 2811 2812 2813 2814 2815 2816 2817
		be_queue_free(adapter, &eqo->q);
	}
}

static int be_evt_queues_create(struct be_adapter *adapter)
{
	struct be_queue_info *eq;
	struct be_eq_obj *eqo;
2818
	struct be_aic_obj *aic;
S
Sathya Perla 已提交
2819 2820
	int i, rc;

2821
	/* need enough EQs to service both RX and TX queues */
2822
	adapter->num_evt_qs = min_t(u16, num_irqs(adapter),
2823 2824
				    max(adapter->cfg_num_rx_irqs,
					adapter->cfg_num_tx_irqs));
S
Sathya Perla 已提交
2825 2826

	for_all_evt_queues(adapter, eqo, i) {
2827
		int numa_node = dev_to_node(&adapter->pdev->dev);
2828

2829
		aic = &adapter->aic_obj[i];
S
Sathya Perla 已提交
2830 2831
		eqo->adapter = adapter;
		eqo->idx = i;
2832 2833
		aic->max_eqd = BE_MAX_EQD;
		aic->enable = true;
S
Sathya Perla 已提交
2834 2835 2836

		eq = &eqo->q;
		rc = be_queue_alloc(adapter, eq, EVNT_Q_LEN,
2837
				    sizeof(struct be_eq_entry));
S
Sathya Perla 已提交
2838 2839 2840
		if (rc)
			return rc;

S
Sathya Perla 已提交
2841
		rc = be_cmd_eq_create(adapter, eqo);
S
Sathya Perla 已提交
2842 2843
		if (rc)
			return rc;
2844 2845 2846 2847 2848 2849 2850

		if (!zalloc_cpumask_var(&eqo->affinity_mask, GFP_KERNEL))
			return -ENOMEM;
		cpumask_set_cpu(cpumask_local_spread(i, numa_node),
				eqo->affinity_mask);
		netif_napi_add(adapter->netdev, &eqo->napi, be_poll,
			       BE_NAPI_WEIGHT);
S
Sathya Perla 已提交
2851
	}
2852
	return 0;
S
Sathya Perla 已提交
2853 2854
}

2855 2856 2857 2858
static void be_mcc_queues_destroy(struct be_adapter *adapter)
{
	struct be_queue_info *q;

2859
	q = &adapter->mcc_obj.q;
2860
	if (q->created)
2861
		be_cmd_q_destroy(adapter, q, QTYPE_MCCQ);
2862 2863
	be_queue_free(adapter, q);

2864
	q = &adapter->mcc_obj.cq;
2865
	if (q->created)
2866
		be_cmd_q_destroy(adapter, q, QTYPE_CQ);
2867 2868 2869 2870 2871 2872 2873 2874
	be_queue_free(adapter, q);
}

/* Must be called only after TX qs are created as MCC shares TX EQ */
static int be_mcc_queues_create(struct be_adapter *adapter)
{
	struct be_queue_info *q, *cq;

2875
	cq = &adapter->mcc_obj.cq;
2876
	if (be_queue_alloc(adapter, cq, MCC_CQ_LEN,
2877
			   sizeof(struct be_mcc_compl)))
2878 2879
		goto err;

S
Sathya Perla 已提交
2880 2881
	/* Use the default EQ for MCC completions */
	if (be_cmd_cq_create(adapter, cq, &mcc_eqo(adapter)->q, true, 0))
2882 2883
		goto mcc_cq_free;

2884
	q = &adapter->mcc_obj.q;
2885 2886 2887
	if (be_queue_alloc(adapter, q, MCC_Q_LEN, sizeof(struct be_mcc_wrb)))
		goto mcc_cq_destroy;

2888
	if (be_cmd_mccq_create(adapter, q, cq))
2889 2890 2891 2892 2893 2894 2895
		goto mcc_q_free;

	return 0;

mcc_q_free:
	be_queue_free(adapter, q);
mcc_cq_destroy:
2896
	be_cmd_q_destroy(adapter, cq, QTYPE_CQ);
2897 2898 2899 2900 2901 2902
mcc_cq_free:
	be_queue_free(adapter, cq);
err:
	return -1;
}

S
Sathya Perla 已提交
2903 2904 2905
static void be_tx_queues_destroy(struct be_adapter *adapter)
{
	struct be_queue_info *q;
2906 2907
	struct be_tx_obj *txo;
	u8 i;
S
Sathya Perla 已提交
2908

2909 2910 2911 2912 2913
	for_all_tx_queues(adapter, txo, i) {
		q = &txo->q;
		if (q->created)
			be_cmd_q_destroy(adapter, q, QTYPE_TXQ);
		be_queue_free(adapter, q);
S
Sathya Perla 已提交
2914

2915 2916 2917 2918 2919
		q = &txo->cq;
		if (q->created)
			be_cmd_q_destroy(adapter, q, QTYPE_CQ);
		be_queue_free(adapter, q);
	}
S
Sathya Perla 已提交
2920 2921
}

2922
static int be_tx_qs_create(struct be_adapter *adapter)
S
Sathya Perla 已提交
2923
{
S
Sathya Perla 已提交
2924
	struct be_queue_info *cq;
2925
	struct be_tx_obj *txo;
S
Sathya Perla 已提交
2926
	struct be_eq_obj *eqo;
2927
	int status, i;
S
Sathya Perla 已提交
2928

2929
	adapter->num_tx_qs = min(adapter->num_evt_qs, adapter->cfg_num_tx_irqs);
2930

S
Sathya Perla 已提交
2931 2932 2933 2934 2935 2936
	for_all_tx_queues(adapter, txo, i) {
		cq = &txo->cq;
		status = be_queue_alloc(adapter, cq, TX_CQ_LEN,
					sizeof(struct be_eth_tx_compl));
		if (status)
			return status;
2937

2938 2939 2940
		u64_stats_init(&txo->stats.sync);
		u64_stats_init(&txo->stats.sync_compl);

S
Sathya Perla 已提交
2941 2942 2943
		/* If num_evt_qs is less than num_tx_qs, then more than
		 * one txq share an eq
		 */
S
Sathya Perla 已提交
2944 2945
		eqo = &adapter->eq_obj[i % adapter->num_evt_qs];
		status = be_cmd_cq_create(adapter, cq, &eqo->q, false, 3);
S
Sathya Perla 已提交
2946 2947
		if (status)
			return status;
S
Sathya Perla 已提交
2948

S
Sathya Perla 已提交
2949 2950 2951 2952
		status = be_queue_alloc(adapter, &txo->q, TX_Q_LEN,
					sizeof(struct be_eth_wrb));
		if (status)
			return status;
S
Sathya Perla 已提交
2953

V
Vasundhara Volam 已提交
2954
		status = be_cmd_txq_create(adapter, txo);
S
Sathya Perla 已提交
2955 2956
		if (status)
			return status;
S
Sathya Perla 已提交
2957 2958 2959

		netif_set_xps_queue(adapter->netdev, eqo->affinity_mask,
				    eqo->idx);
2960
	}
S
Sathya Perla 已提交
2961

S
Sathya Perla 已提交
2962 2963
	dev_info(&adapter->pdev->dev, "created %d TX queue(s)\n",
		 adapter->num_tx_qs);
S
Sathya Perla 已提交
2964
	return 0;
S
Sathya Perla 已提交
2965 2966
}

S
Sathya Perla 已提交
2967
static void be_rx_cqs_destroy(struct be_adapter *adapter)
S
Sathya Perla 已提交
2968 2969
{
	struct be_queue_info *q;
2970 2971 2972 2973 2974 2975 2976 2977
	struct be_rx_obj *rxo;
	int i;

	for_all_rx_queues(adapter, rxo, i) {
		q = &rxo->cq;
		if (q->created)
			be_cmd_q_destroy(adapter, q, QTYPE_CQ);
		be_queue_free(adapter, q);
2978 2979 2980
	}
}

S
Sathya Perla 已提交
2981
static int be_rx_cqs_create(struct be_adapter *adapter)
S
Sathya Perla 已提交
2982
{
S
Sathya Perla 已提交
2983
	struct be_queue_info *eq, *cq;
2984 2985
	struct be_rx_obj *rxo;
	int rc, i;
S
Sathya Perla 已提交
2986

2987 2988
	adapter->num_rss_qs =
			min(adapter->num_evt_qs, adapter->cfg_num_rx_irqs);
2989

2990
	/* We'll use RSS only if atleast 2 RSS rings are supported. */
2991
	if (adapter->num_rss_qs < 2)
2992 2993 2994 2995 2996 2997
		adapter->num_rss_qs = 0;

	adapter->num_rx_qs = adapter->num_rss_qs + adapter->need_def_rxq;

	/* When the interface is not capable of RSS rings (and there is no
	 * need to create a default RXQ) we'll still need one RXQ
S
Sathya Perla 已提交
2998
	 */
2999 3000
	if (adapter->num_rx_qs == 0)
		adapter->num_rx_qs = 1;
3001

S
Sathya Perla 已提交
3002
	adapter->big_page_size = (1 << get_order(rx_frag_size)) * PAGE_SIZE;
3003 3004 3005 3006
	for_all_rx_queues(adapter, rxo, i) {
		rxo->adapter = adapter;
		cq = &rxo->cq;
		rc = be_queue_alloc(adapter, cq, RX_CQ_LEN,
3007
				    sizeof(struct be_eth_rx_compl));
3008
		if (rc)
S
Sathya Perla 已提交
3009
			return rc;
3010

3011
		u64_stats_init(&rxo->stats.sync);
S
Sathya Perla 已提交
3012 3013
		eq = &adapter->eq_obj[i % adapter->num_evt_qs].q;
		rc = be_cmd_cq_create(adapter, cq, eq, false, 3);
3014
		if (rc)
S
Sathya Perla 已提交
3015
			return rc;
3016
	}
S
Sathya Perla 已提交
3017

S
Sathya Perla 已提交
3018
	dev_info(&adapter->pdev->dev,
3019
		 "created %d RX queue(s)\n", adapter->num_rx_qs);
S
Sathya Perla 已提交
3020
	return 0;
3021 3022
}

S
Sathya Perla 已提交
3023 3024
static irqreturn_t be_intx(int irq, void *dev)
{
3025 3026 3027
	struct be_eq_obj *eqo = dev;
	struct be_adapter *adapter = eqo->adapter;
	int num_evts = 0;
S
Sathya Perla 已提交
3028

3029 3030 3031 3032 3033 3034 3035
	/* IRQ is not expected when NAPI is scheduled as the EQ
	 * will not be armed.
	 * But, this can happen on Lancer INTx where it takes
	 * a while to de-assert INTx or in BE2 where occasionaly
	 * an interrupt may be raised even when EQ is unarmed.
	 * If NAPI is already scheduled, then counting & notifying
	 * events will orphan them.
3036
	 */
3037
	if (napi_schedule_prep(&eqo->napi)) {
3038
		num_evts = events_get(eqo);
3039 3040 3041 3042
		__napi_schedule(&eqo->napi);
		if (num_evts)
			eqo->spurious_intr = 0;
	}
3043
	be_eq_notify(adapter, eqo->q.id, false, true, num_evts, 0);
3044

3045 3046 3047
	/* Return IRQ_HANDLED only for the the first spurious intr
	 * after a valid intr to stop the kernel from branding
	 * this irq as a bad one!
3048
	 */
3049 3050 3051 3052
	if (num_evts || eqo->spurious_intr++ == 0)
		return IRQ_HANDLED;
	else
		return IRQ_NONE;
S
Sathya Perla 已提交
3053 3054
}

S
Sathya Perla 已提交
3055
static irqreturn_t be_msix(int irq, void *dev)
S
Sathya Perla 已提交
3056
{
S
Sathya Perla 已提交
3057
	struct be_eq_obj *eqo = dev;
S
Sathya Perla 已提交
3058

3059
	be_eq_notify(eqo->adapter, eqo->q.id, false, true, 0, 0);
3060
	napi_schedule(&eqo->napi);
S
Sathya Perla 已提交
3061 3062 3063
	return IRQ_HANDLED;
}

3064
static inline bool do_gro(struct be_rx_compl_info *rxcp)
S
Sathya Perla 已提交
3065
{
3066
	return (rxcp->tcpf && !rxcp->err && rxcp->l4_csum) ? true : false;
S
Sathya Perla 已提交
3067 3068
}

S
Sathya Perla 已提交
3069
static int be_process_rx(struct be_rx_obj *rxo, struct napi_struct *napi,
3070
			 int budget, int polling)
S
Sathya Perla 已提交
3071
{
3072 3073
	struct be_adapter *adapter = rxo->adapter;
	struct be_queue_info *rx_cq = &rxo->cq;
3074
	struct be_rx_compl_info *rxcp;
S
Sathya Perla 已提交
3075
	u32 work_done;
3076
	u32 frags_consumed = 0;
S
Sathya Perla 已提交
3077 3078

	for (work_done = 0; work_done < budget; work_done++) {
3079
		rxcp = be_rx_compl_get(rxo);
S
Sathya Perla 已提交
3080 3081 3082
		if (!rxcp)
			break;

3083 3084 3085 3086 3087 3088
		/* Is it a flush compl that has no data */
		if (unlikely(rxcp->num_rcvd == 0))
			goto loop_continue;

		/* Discard compl with partial DMA Lancer B0 */
		if (unlikely(!rxcp->pkt_size)) {
S
Sathya Perla 已提交
3089
			be_rx_compl_discard(rxo, rxcp);
3090 3091 3092 3093 3094 3095 3096
			goto loop_continue;
		}

		/* On BE drop pkts that arrive due to imperfect filtering in
		 * promiscuous mode on some skews
		 */
		if (unlikely(rxcp->port != adapter->port_num &&
3097
			     !lancer_chip(adapter))) {
S
Sathya Perla 已提交
3098
			be_rx_compl_discard(rxo, rxcp);
3099
			goto loop_continue;
3100
		}
3101

3102 3103
		/* Don't do gro when we're busy_polling */
		if (do_gro(rxcp) && polling != BUSY_POLLING)
S
Sathya Perla 已提交
3104
			be_rx_compl_process_gro(rxo, napi, rxcp);
3105
		else
3106 3107
			be_rx_compl_process(rxo, napi, rxcp);

3108
loop_continue:
3109
		frags_consumed += rxcp->num_rcvd;
3110
		be_rx_stats_update(rxo, rxcp);
S
Sathya Perla 已提交
3111 3112
	}

S
Sathya Perla 已提交
3113 3114
	if (work_done) {
		be_cq_notify(adapter, rx_cq->id, true, work_done);
3115

3116 3117 3118 3119 3120
		/* When an rx-obj gets into post_starved state, just
		 * let be_worker do the posting.
		 */
		if (atomic_read(&rxo->q.used) < RX_FRAGS_REFILL_WM &&
		    !rxo->rx_post_starved)
3121 3122 3123
			be_post_rx_frags(rxo, GFP_ATOMIC,
					 max_t(u32, MAX_RX_POST,
					       frags_consumed));
S
Sathya Perla 已提交
3124
	}
S
Sathya Perla 已提交
3125

S
Sathya Perla 已提交
3126 3127 3128
	return work_done;
}

3129
static inline void be_update_tx_err(struct be_tx_obj *txo, u8 status)
3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143
{
	switch (status) {
	case BE_TX_COMP_HDR_PARSE_ERR:
		tx_stats(txo)->tx_hdr_parse_err++;
		break;
	case BE_TX_COMP_NDMA_ERR:
		tx_stats(txo)->tx_dma_err++;
		break;
	case BE_TX_COMP_ACL_ERR:
		tx_stats(txo)->tx_spoof_check_err++;
		break;
	}
}

3144
static inline void lancer_update_tx_err(struct be_tx_obj *txo, u8 status)
3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165
{
	switch (status) {
	case LANCER_TX_COMP_LSO_ERR:
		tx_stats(txo)->tx_tso_err++;
		break;
	case LANCER_TX_COMP_HSW_DROP_MAC_ERR:
	case LANCER_TX_COMP_HSW_DROP_VLAN_ERR:
		tx_stats(txo)->tx_spoof_check_err++;
		break;
	case LANCER_TX_COMP_QINQ_ERR:
		tx_stats(txo)->tx_qinq_err++;
		break;
	case LANCER_TX_COMP_PARITY_ERR:
		tx_stats(txo)->tx_internal_parity_err++;
		break;
	case LANCER_TX_COMP_DMA_ERR:
		tx_stats(txo)->tx_dma_err++;
		break;
	}
}

S
Sathya Perla 已提交
3166 3167
static void be_process_tx(struct be_adapter *adapter, struct be_tx_obj *txo,
			  int idx)
S
Sathya Perla 已提交
3168
{
S
Sathya Perla 已提交
3169
	int num_wrbs = 0, work_done = 0;
3170
	struct be_tx_compl_info *txcp;
S
Sathya Perla 已提交
3171

3172 3173
	while ((txcp = be_tx_compl_get(txo))) {
		num_wrbs += be_tx_compl_process(adapter, txo, txcp->end_index);
S
Sathya Perla 已提交
3174
		work_done++;
3175

3176
		if (txcp->status) {
3177
			if (lancer_chip(adapter))
3178
				lancer_update_tx_err(txo, txcp->status);
3179
			else
3180
				be_update_tx_err(txo, txcp->status);
3181
		}
S
Sathya Perla 已提交
3182
	}
S
Sathya Perla 已提交
3183

S
Sathya Perla 已提交
3184 3185 3186
	if (work_done) {
		be_cq_notify(adapter, txo->cq.id, true, work_done);
		atomic_sub(num_wrbs, &txo->q.used);
3187

S
Sathya Perla 已提交
3188 3189 3190
		/* As Tx wrbs have been freed up, wake up netdev queue
		 * if it was stopped due to lack of tx wrbs.  */
		if (__netif_subqueue_stopped(adapter->netdev, idx) &&
3191
		    be_can_txq_wake(txo)) {
S
Sathya Perla 已提交
3192
			netif_wake_subqueue(adapter->netdev, idx);
3193
		}
S
Sathya Perla 已提交
3194 3195 3196 3197

		u64_stats_update_begin(&tx_stats(txo)->sync_compl);
		tx_stats(txo)->tx_compl += work_done;
		u64_stats_update_end(&tx_stats(txo)->sync_compl);
S
Sathya Perla 已提交
3198
	}
S
Sathya Perla 已提交
3199
}
S
Sathya Perla 已提交
3200

3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300
#ifdef CONFIG_NET_RX_BUSY_POLL
static inline bool be_lock_napi(struct be_eq_obj *eqo)
{
	bool status = true;

	spin_lock(&eqo->lock); /* BH is already disabled */
	if (eqo->state & BE_EQ_LOCKED) {
		WARN_ON(eqo->state & BE_EQ_NAPI);
		eqo->state |= BE_EQ_NAPI_YIELD;
		status = false;
	} else {
		eqo->state = BE_EQ_NAPI;
	}
	spin_unlock(&eqo->lock);
	return status;
}

static inline void be_unlock_napi(struct be_eq_obj *eqo)
{
	spin_lock(&eqo->lock); /* BH is already disabled */

	WARN_ON(eqo->state & (BE_EQ_POLL | BE_EQ_NAPI_YIELD));
	eqo->state = BE_EQ_IDLE;

	spin_unlock(&eqo->lock);
}

static inline bool be_lock_busy_poll(struct be_eq_obj *eqo)
{
	bool status = true;

	spin_lock_bh(&eqo->lock);
	if (eqo->state & BE_EQ_LOCKED) {
		eqo->state |= BE_EQ_POLL_YIELD;
		status = false;
	} else {
		eqo->state |= BE_EQ_POLL;
	}
	spin_unlock_bh(&eqo->lock);
	return status;
}

static inline void be_unlock_busy_poll(struct be_eq_obj *eqo)
{
	spin_lock_bh(&eqo->lock);

	WARN_ON(eqo->state & (BE_EQ_NAPI));
	eqo->state = BE_EQ_IDLE;

	spin_unlock_bh(&eqo->lock);
}

static inline void be_enable_busy_poll(struct be_eq_obj *eqo)
{
	spin_lock_init(&eqo->lock);
	eqo->state = BE_EQ_IDLE;
}

static inline void be_disable_busy_poll(struct be_eq_obj *eqo)
{
	local_bh_disable();

	/* It's enough to just acquire napi lock on the eqo to stop
	 * be_busy_poll() from processing any queueus.
	 */
	while (!be_lock_napi(eqo))
		mdelay(1);

	local_bh_enable();
}

#else /* CONFIG_NET_RX_BUSY_POLL */

static inline bool be_lock_napi(struct be_eq_obj *eqo)
{
	return true;
}

static inline void be_unlock_napi(struct be_eq_obj *eqo)
{
}

static inline bool be_lock_busy_poll(struct be_eq_obj *eqo)
{
	return false;
}

static inline void be_unlock_busy_poll(struct be_eq_obj *eqo)
{
}

static inline void be_enable_busy_poll(struct be_eq_obj *eqo)
{
}

static inline void be_disable_busy_poll(struct be_eq_obj *eqo)
{
}
#endif /* CONFIG_NET_RX_BUSY_POLL */

3301
int be_poll(struct napi_struct *napi, int budget)
S
Sathya Perla 已提交
3302 3303 3304
{
	struct be_eq_obj *eqo = container_of(napi, struct be_eq_obj, napi);
	struct be_adapter *adapter = eqo->adapter;
3305
	int max_work = 0, work, i, num_evts;
3306
	struct be_rx_obj *rxo;
3307
	struct be_tx_obj *txo;
3308
	u32 mult_enc = 0;
S
Sathya Perla 已提交
3309

3310 3311
	num_evts = events_get(eqo);

3312 3313
	for_all_tx_queues_on_eq(adapter, eqo, txo, i)
		be_process_tx(adapter, txo, i);
S
Sathya Perla 已提交
3314

3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326
	if (be_lock_napi(eqo)) {
		/* This loop will iterate twice for EQ0 in which
		 * completions of the last RXQ (default one) are also processed
		 * For other EQs the loop iterates only once
		 */
		for_all_rx_queues_on_eq(adapter, eqo, rxo, i) {
			work = be_process_rx(rxo, napi, budget, NAPI_POLLING);
			max_work = max(work, max_work);
		}
		be_unlock_napi(eqo);
	} else {
		max_work = budget;
S
Sathya Perla 已提交
3327
	}
S
Sathya Perla 已提交
3328

S
Sathya Perla 已提交
3329 3330
	if (is_mcc_eqo(eqo))
		be_process_mcc(adapter);
3331

S
Sathya Perla 已提交
3332 3333
	if (max_work < budget) {
		napi_complete(napi);
3334 3335 3336 3337 3338 3339 3340 3341 3342

		/* Skyhawk EQ_DB has a provision to set the rearm to interrupt
		 * delay via a delay multiplier encoding value
		 */
		if (skyhawk_chip(adapter))
			mult_enc = be_get_eq_delay_mult_enc(eqo);

		be_eq_notify(adapter, eqo->q.id, true, false, num_evts,
			     mult_enc);
S
Sathya Perla 已提交
3343 3344
	} else {
		/* As we'll continue in polling mode, count and clear events */
3345
		be_eq_notify(adapter, eqo->q.id, false, false, num_evts, 0);
3346
	}
S
Sathya Perla 已提交
3347
	return max_work;
S
Sathya Perla 已提交
3348 3349
}

3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371
#ifdef CONFIG_NET_RX_BUSY_POLL
static int be_busy_poll(struct napi_struct *napi)
{
	struct be_eq_obj *eqo = container_of(napi, struct be_eq_obj, napi);
	struct be_adapter *adapter = eqo->adapter;
	struct be_rx_obj *rxo;
	int i, work = 0;

	if (!be_lock_busy_poll(eqo))
		return LL_FLUSH_BUSY;

	for_all_rx_queues_on_eq(adapter, eqo, rxo, i) {
		work = be_process_rx(rxo, napi, 4, BUSY_POLLING);
		if (work)
			break;
	}

	be_unlock_busy_poll(eqo);
	return work;
}
#endif

3372
void be_detect_error(struct be_adapter *adapter)
3373
{
3374 3375
	u32 ue_lo = 0, ue_hi = 0, ue_lo_mask = 0, ue_hi_mask = 0;
	u32 sliport_status = 0, sliport_err1 = 0, sliport_err2 = 0;
3376
	u32 i;
3377
	struct device *dev = &adapter->pdev->dev;
3378

3379
	if (be_check_error(adapter, BE_ERROR_HW))
3380 3381
		return;

3382 3383 3384
	if (lancer_chip(adapter)) {
		sliport_status = ioread32(adapter->db + SLIPORT_STATUS_OFFSET);
		if (sliport_status & SLIPORT_STATUS_ERR_MASK) {
3385
			be_set_error(adapter, BE_ERROR_UE);
3386
			sliport_err1 = ioread32(adapter->db +
3387
						SLIPORT_ERROR1_OFFSET);
3388
			sliport_err2 = ioread32(adapter->db +
3389
						SLIPORT_ERROR2_OFFSET);
3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402
			/* Do not log error messages if its a FW reset */
			if (sliport_err1 == SLIPORT_ERROR_FW_RESET1 &&
			    sliport_err2 == SLIPORT_ERROR_FW_RESET2) {
				dev_info(dev, "Firmware update in progress\n");
			} else {
				dev_err(dev, "Error detected in the card\n");
				dev_err(dev, "ERR: sliport status 0x%x\n",
					sliport_status);
				dev_err(dev, "ERR: sliport error1 0x%x\n",
					sliport_err1);
				dev_err(dev, "ERR: sliport error2 0x%x\n",
					sliport_err2);
			}
3403 3404
		}
	} else {
3405 3406 3407 3408 3409 3410
		ue_lo = ioread32(adapter->pcicfg + PCICFG_UE_STATUS_LOW);
		ue_hi = ioread32(adapter->pcicfg + PCICFG_UE_STATUS_HIGH);
		ue_lo_mask = ioread32(adapter->pcicfg +
				      PCICFG_UE_STATUS_LOW_MASK);
		ue_hi_mask = ioread32(adapter->pcicfg +
				      PCICFG_UE_STATUS_HI_MASK);
3411

3412 3413
		ue_lo = (ue_lo & ~ue_lo_mask);
		ue_hi = (ue_hi & ~ue_hi_mask);
3414

3415 3416 3417 3418
		/* On certain platforms BE hardware can indicate spurious UEs.
		 * Allow HW to stop working completely in case of a real UE.
		 * Hence not setting the hw_error for UE detection.
		 */
3419

3420
		if (ue_lo || ue_hi) {
3421
			dev_err(dev, "Error detected in the adapter");
3422
			if (skyhawk_chip(adapter))
3423 3424
				be_set_error(adapter, BE_ERROR_UE);

3425 3426 3427 3428 3429 3430 3431 3432 3433 3434
			for (i = 0; ue_lo; ue_lo >>= 1, i++) {
				if (ue_lo & 1)
					dev_err(dev, "UE: %s bit set\n",
						ue_status_low_desc[i]);
			}
			for (i = 0; ue_hi; ue_hi >>= 1, i++) {
				if (ue_hi & 1)
					dev_err(dev, "UE: %s bit set\n",
						ue_status_hi_desc[i]);
			}
3435 3436 3437 3438
		}
	}
}

3439 3440
static void be_msix_disable(struct be_adapter *adapter)
{
3441
	if (msix_enabled(adapter)) {
3442
		pci_disable_msix(adapter->pdev);
3443
		adapter->num_msix_vec = 0;
3444
		adapter->num_msix_roce_vec = 0;
3445 3446 3447
	}
}

3448
static int be_msix_enable(struct be_adapter *adapter)
S
Sathya Perla 已提交
3449
{
3450
	unsigned int i, max_roce_eqs;
S
Sathya Perla 已提交
3451
	struct device *dev = &adapter->pdev->dev;
3452
	int num_vec;
S
Sathya Perla 已提交
3453

3454 3455 3456
	/* If RoCE is supported, program the max number of vectors that
	 * could be used for NIC and RoCE, else, just program the number
	 * we'll use initially.
3457
	 */
3458 3459 3460 3461 3462 3463 3464 3465 3466
	if (be_roce_supported(adapter)) {
		max_roce_eqs =
			be_max_func_eqs(adapter) - be_max_nic_eqs(adapter);
		max_roce_eqs = min(max_roce_eqs, num_online_cpus());
		num_vec = be_max_any_irqs(adapter) + max_roce_eqs;
	} else {
		num_vec = max(adapter->cfg_num_rx_irqs,
			      adapter->cfg_num_tx_irqs);
	}
3467

3468
	for (i = 0; i < num_vec; i++)
S
Sathya Perla 已提交
3469 3470
		adapter->msix_entries[i].entry = i;

3471 3472 3473 3474
	num_vec = pci_enable_msix_range(adapter->pdev, adapter->msix_entries,
					MIN_MSIX_VECTORS, num_vec);
	if (num_vec < 0)
		goto fail;
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485

	if (be_roce_supported(adapter) && num_vec > MIN_MSIX_VECTORS) {
		adapter->num_msix_roce_vec = num_vec / 2;
		dev_info(dev, "enabled %d MSI-x vector(s) for RoCE\n",
			 adapter->num_msix_roce_vec);
	}

	adapter->num_msix_vec = num_vec - adapter->num_msix_roce_vec;

	dev_info(dev, "enabled %d MSI-x vector(s) for NIC\n",
		 adapter->num_msix_vec);
3486
	return 0;
3487 3488 3489 3490 3491

fail:
	dev_warn(dev, "MSIx enable failed\n");

	/* INTx is not supported in VFs, so fail probe if enable_msix fails */
3492
	if (be_virtfn(adapter))
3493 3494
		return num_vec;
	return 0;
S
Sathya Perla 已提交
3495 3496
}

3497
static inline int be_msix_vec_get(struct be_adapter *adapter,
3498
				  struct be_eq_obj *eqo)
3499
{
S
Sathya Perla 已提交
3500
	return adapter->msix_entries[eqo->msix_idx].vector;
3501
}
S
Sathya Perla 已提交
3502

3503 3504
static int be_msix_register(struct be_adapter *adapter)
{
S
Sathya Perla 已提交
3505 3506 3507
	struct net_device *netdev = adapter->netdev;
	struct be_eq_obj *eqo;
	int status, i, vec;
S
Sathya Perla 已提交
3508

S
Sathya Perla 已提交
3509 3510 3511 3512
	for_all_evt_queues(adapter, eqo, i) {
		sprintf(eqo->desc, "%s-q%d", netdev->name, i);
		vec = be_msix_vec_get(adapter, eqo);
		status = request_irq(vec, be_msix, 0, eqo->desc, eqo);
3513 3514
		if (status)
			goto err_msix;
3515 3516

		irq_set_affinity_hint(vec, eqo->affinity_mask);
3517
	}
3518

S
Sathya Perla 已提交
3519
	return 0;
3520
err_msix:
3521 3522
	for (i--; i >= 0; i--) {
		eqo = &adapter->eq_obj[i];
S
Sathya Perla 已提交
3523
		free_irq(be_msix_vec_get(adapter, eqo), eqo);
3524
	}
S
Sathya Perla 已提交
3525
	dev_warn(&adapter->pdev->dev, "MSIX Request IRQ failed - err %d\n",
3526
		 status);
3527
	be_msix_disable(adapter);
S
Sathya Perla 已提交
3528 3529 3530 3531 3532 3533 3534 3535
	return status;
}

static int be_irq_register(struct be_adapter *adapter)
{
	struct net_device *netdev = adapter->netdev;
	int status;

3536
	if (msix_enabled(adapter)) {
S
Sathya Perla 已提交
3537 3538 3539
		status = be_msix_register(adapter);
		if (status == 0)
			goto done;
3540
		/* INTx is not supported for VF */
3541
		if (be_virtfn(adapter))
3542
			return status;
S
Sathya Perla 已提交
3543 3544
	}

3545
	/* INTx: only the first EQ is used */
S
Sathya Perla 已提交
3546 3547
	netdev->irq = adapter->pdev->irq;
	status = request_irq(netdev->irq, be_intx, IRQF_SHARED, netdev->name,
3548
			     &adapter->eq_obj[0]);
S
Sathya Perla 已提交
3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561
	if (status) {
		dev_err(&adapter->pdev->dev,
			"INTx request IRQ failed - err %d\n", status);
		return status;
	}
done:
	adapter->isr_registered = true;
	return 0;
}

static void be_irq_unregister(struct be_adapter *adapter)
{
	struct net_device *netdev = adapter->netdev;
S
Sathya Perla 已提交
3562
	struct be_eq_obj *eqo;
3563
	int i, vec;
S
Sathya Perla 已提交
3564 3565 3566 3567 3568

	if (!adapter->isr_registered)
		return;

	/* INTx */
3569
	if (!msix_enabled(adapter)) {
3570
		free_irq(netdev->irq, &adapter->eq_obj[0]);
S
Sathya Perla 已提交
3571 3572 3573 3574
		goto done;
	}

	/* MSIx */
3575 3576 3577 3578 3579
	for_all_evt_queues(adapter, eqo, i) {
		vec = be_msix_vec_get(adapter, eqo);
		irq_set_affinity_hint(vec, NULL);
		free_irq(vec, eqo);
	}
3580

S
Sathya Perla 已提交
3581 3582 3583 3584
done:
	adapter->isr_registered = false;
}

S
Sathya Perla 已提交
3585
static void be_rx_qs_destroy(struct be_adapter *adapter)
3586
{
3587
	struct rss_info *rss = &adapter->rss_info;
3588 3589 3590 3591 3592 3593 3594
	struct be_queue_info *q;
	struct be_rx_obj *rxo;
	int i;

	for_all_rx_queues(adapter, rxo, i) {
		q = &rxo->q;
		if (q->created) {
3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607
			/* If RXQs are destroyed while in an "out of buffer"
			 * state, there is a possibility of an HW stall on
			 * Lancer. So, post 64 buffers to each queue to relieve
			 * the "out of buffer" condition.
			 * Make sure there's space in the RXQ before posting.
			 */
			if (lancer_chip(adapter)) {
				be_rx_cq_clean(rxo);
				if (atomic_read(&q->used) == 0)
					be_post_rx_frags(rxo, GFP_KERNEL,
							 MAX_RX_POST);
			}

3608
			be_cmd_rxq_destroy(adapter, q);
S
Sathya Perla 已提交
3609
			be_rx_cq_clean(rxo);
3610
			be_rxq_clean(rxo);
3611
		}
S
Sathya Perla 已提交
3612
		be_queue_free(adapter, q);
3613
	}
3614 3615 3616 3617 3618 3619

	if (rss->rss_flags) {
		rss->rss_flags = RSS_ENABLE_NONE;
		be_cmd_rss_config(adapter, rss->rsstable, rss->rss_flags,
				  128, rss->rss_hkey);
	}
3620 3621
}

3622 3623
static void be_disable_if_filters(struct be_adapter *adapter)
{
3624
	be_dev_mac_del(adapter, adapter->pmac_id[0]);
3625
	be_clear_uc_list(adapter);
3626
	be_clear_mc_list(adapter);
3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647

	/* The IFACE flags are enabled in the open path and cleared
	 * in the close path. When a VF gets detached from the host and
	 * assigned to a VM the following happens:
	 *	- VF's IFACE flags get cleared in the detach path
	 *	- IFACE create is issued by the VF in the attach path
	 * Due to a bug in the BE3/Skyhawk-R FW
	 * (Lancer FW doesn't have the bug), the IFACE capability flags
	 * specified along with the IFACE create cmd issued by a VF are not
	 * honoured by FW.  As a consequence, if a *new* driver
	 * (that enables/disables IFACE flags in open/close)
	 * is loaded in the host and an *old* driver is * used by a VM/VF,
	 * the IFACE gets created *without* the needed flags.
	 * To avoid this, disable RX-filter flags only for Lancer.
	 */
	if (lancer_chip(adapter)) {
		be_cmd_rx_filter(adapter, BE_IF_ALL_FILT_FLAGS, OFF);
		adapter->if_flags &= ~BE_IF_ALL_FILT_FLAGS;
	}
}

3648 3649 3650
static int be_close(struct net_device *netdev)
{
	struct be_adapter *adapter = netdev_priv(netdev);
S
Sathya Perla 已提交
3651 3652
	struct be_eq_obj *eqo;
	int i;
3653

3654 3655 3656 3657 3658 3659
	/* This protection is needed as be_close() may be called even when the
	 * adapter is in cleared state (after eeh perm failure)
	 */
	if (!(adapter->flags & BE_FLAGS_SETUP_DONE))
		return 0;

3660 3661 3662 3663 3664
	/* Before attempting cleanup ensure all the pending cmds in the
	 * config_wq have finished execution
	 */
	flush_workqueue(be_wq);

3665 3666
	be_disable_if_filters(adapter);

3667 3668
	if (adapter->flags & BE_FLAGS_NAPI_ENABLED) {
		for_all_evt_queues(adapter, eqo, i) {
3669
			napi_disable(&eqo->napi);
3670 3671
			be_disable_busy_poll(eqo);
		}
3672
		adapter->flags &= ~BE_FLAGS_NAPI_ENABLED;
3673
	}
3674 3675 3676 3677 3678 3679

	be_async_mcc_disable(adapter);

	/* Wait for all pending tx completions to arrive so that
	 * all tx skbs are freed.
	 */
S
Sathya Perla 已提交
3680
	netif_tx_disable(netdev);
3681
	be_tx_compl_clean(adapter);
3682 3683

	be_rx_qs_destroy(adapter);
3684

3685
	for_all_evt_queues(adapter, eqo, i) {
S
Sathya Perla 已提交
3686 3687 3688 3689 3690
		if (msix_enabled(adapter))
			synchronize_irq(be_msix_vec_get(adapter, eqo));
		else
			synchronize_irq(netdev->irq);
		be_eq_clean(eqo);
3691 3692
	}

3693 3694
	be_irq_unregister(adapter);

3695 3696 3697
	return 0;
}

S
Sathya Perla 已提交
3698
static int be_rx_qs_create(struct be_adapter *adapter)
3699
{
3700 3701
	struct rss_info *rss = &adapter->rss_info;
	u8 rss_key[RSS_HASH_KEY_LEN];
3702
	struct be_rx_obj *rxo;
3703
	int rc, i, j;
3704 3705

	for_all_rx_queues(adapter, rxo, i) {
S
Sathya Perla 已提交
3706 3707 3708 3709 3710 3711
		rc = be_queue_alloc(adapter, &rxo->q, RX_Q_LEN,
				    sizeof(struct be_eth_rx_d));
		if (rc)
			return rc;
	}

3712 3713 3714 3715 3716 3717 3718 3719
	if (adapter->need_def_rxq || !adapter->num_rss_qs) {
		rxo = default_rxo(adapter);
		rc = be_cmd_rxq_create(adapter, &rxo->q, rxo->cq.id,
				       rx_frag_size, adapter->if_handle,
				       false, &rxo->rss_id);
		if (rc)
			return rc;
	}
S
Sathya Perla 已提交
3720 3721

	for_all_rss_queues(adapter, rxo, i) {
3722
		rc = be_cmd_rxq_create(adapter, &rxo->q, rxo->cq.id,
S
Sathya Perla 已提交
3723 3724
				       rx_frag_size, adapter->if_handle,
				       true, &rxo->rss_id);
3725 3726 3727 3728 3729
		if (rc)
			return rc;
	}

	if (be_multi_rxq(adapter)) {
3730
		for (j = 0; j < RSS_INDIR_TABLE_LEN; j += adapter->num_rss_qs) {
3731
			for_all_rss_queues(adapter, rxo, i) {
3732
				if ((j + i) >= RSS_INDIR_TABLE_LEN)
3733
					break;
3734 3735
				rss->rsstable[j + i] = rxo->rss_id;
				rss->rss_queue[j + i] = i;
3736 3737
			}
		}
3738 3739
		rss->rss_flags = RSS_ENABLE_TCP_IPV4 | RSS_ENABLE_IPV4 |
			RSS_ENABLE_TCP_IPV6 | RSS_ENABLE_IPV6;
3740 3741

		if (!BEx_chip(adapter))
3742 3743
			rss->rss_flags |= RSS_ENABLE_UDP_IPV4 |
				RSS_ENABLE_UDP_IPV6;
3744 3745 3746 3747 3748 3749 3750 3751 3752 3753

		netdev_rss_key_fill(rss_key, RSS_HASH_KEY_LEN);
		rc = be_cmd_rss_config(adapter, rss->rsstable, rss->rss_flags,
				       RSS_INDIR_TABLE_LEN, rss_key);
		if (rc) {
			rss->rss_flags = RSS_ENABLE_NONE;
			return rc;
		}

		memcpy(rss->rss_hkey, rss_key, RSS_HASH_KEY_LEN);
3754 3755
	} else {
		/* Disable RSS, if only default RX Q is created */
3756
		rss->rss_flags = RSS_ENABLE_NONE;
3757
	}
3758

3759

3760 3761 3762
	/* Post 1 less than RXQ-len to avoid head being equal to tail,
	 * which is a queue empty condition
	 */
S
Sathya Perla 已提交
3763
	for_all_rx_queues(adapter, rxo, i)
3764 3765
		be_post_rx_frags(rxo, GFP_KERNEL, RX_Q_LEN - 1);

3766 3767 3768
	return 0;
}

3769 3770 3771 3772
static int be_enable_if_filters(struct be_adapter *adapter)
{
	int status;

3773
	status = be_cmd_rx_filter(adapter, BE_IF_FILT_FLAGS_BASIC, ON);
3774 3775 3776 3777 3778
	if (status)
		return status;

	/* For BE3 VFs, the PF programs the initial MAC address */
	if (!(BEx_chip(adapter) && be_virtfn(adapter))) {
3779
		status = be_dev_mac_add(adapter, adapter->netdev->dev_addr);
3780 3781
		if (status)
			return status;
3782
		ether_addr_copy(adapter->dev_mac, adapter->netdev->dev_addr);
3783 3784 3785 3786 3787
	}

	if (adapter->vlans_added)
		be_vid_config(adapter);

3788
	__be_set_rx_mode(adapter);
3789 3790 3791 3792

	return 0;
}

S
Sathya Perla 已提交
3793 3794 3795
static int be_open(struct net_device *netdev)
{
	struct be_adapter *adapter = netdev_priv(netdev);
S
Sathya Perla 已提交
3796
	struct be_eq_obj *eqo;
3797
	struct be_rx_obj *rxo;
S
Sathya Perla 已提交
3798
	struct be_tx_obj *txo;
3799
	u8 link_status;
3800
	int status, i;
3801

S
Sathya Perla 已提交
3802
	status = be_rx_qs_create(adapter);
3803 3804 3805
	if (status)
		goto err;

3806 3807 3808 3809
	status = be_enable_if_filters(adapter);
	if (status)
		goto err;

3810 3811 3812
	status = be_irq_register(adapter);
	if (status)
		goto err;
3813

S
Sathya Perla 已提交
3814
	for_all_rx_queues(adapter, rxo, i)
3815
		be_cq_notify(adapter, rxo->cq.id, true, 0);
3816

S
Sathya Perla 已提交
3817 3818 3819
	for_all_tx_queues(adapter, txo, i)
		be_cq_notify(adapter, txo->cq.id, true, 0);

3820 3821
	be_async_mcc_enable(adapter);

S
Sathya Perla 已提交
3822 3823
	for_all_evt_queues(adapter, eqo, i) {
		napi_enable(&eqo->napi);
3824
		be_enable_busy_poll(eqo);
3825
		be_eq_notify(adapter, eqo->q.id, true, true, 0, 0);
S
Sathya Perla 已提交
3826
	}
3827
	adapter->flags |= BE_FLAGS_NAPI_ENABLED;
S
Sathya Perla 已提交
3828

3829
	status = be_cmd_link_status_query(adapter, NULL, &link_status, 0);
3830 3831 3832
	if (!status)
		be_link_status_update(adapter, link_status);

S
Sathya Perla 已提交
3833
	netif_tx_start_all_queues(netdev);
3834
	if (skyhawk_chip(adapter))
3835
		udp_tunnel_get_rx_info(netdev);
3836

3837 3838 3839 3840
	return 0;
err:
	be_close(adapter->netdev);
	return -EIO;
3841 3842
}

3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855
static void be_vf_eth_addr_generate(struct be_adapter *adapter, u8 *mac)
{
	u32 addr;

	addr = jhash(adapter->netdev->dev_addr, ETH_ALEN, 0);

	mac[5] = (u8)(addr & 0xFF);
	mac[4] = (u8)((addr >> 8) & 0xFF);
	mac[3] = (u8)((addr >> 16) & 0xFF);
	/* Use the OUI from the current MAC address */
	memcpy(mac, adapter->netdev->dev_addr, 3);
}

3856 3857 3858 3859 3860 3861
/*
 * Generate a seed MAC address from the PF MAC Address using jhash.
 * MAC Address for VFs are assigned incrementally starting from the seed.
 * These addresses are programmed in the ASIC by the PF and the VF driver
 * queries for the MAC address during its probe.
 */
3862
static int be_vf_eth_addr_config(struct be_adapter *adapter)
3863
{
3864
	u32 vf;
3865
	int status = 0;
3866
	u8 mac[ETH_ALEN];
3867
	struct be_vf_cfg *vf_cfg;
3868 3869 3870

	be_vf_eth_addr_generate(adapter, mac);

3871
	for_all_vfs(adapter, vf_cfg, vf) {
3872
		if (BEx_chip(adapter))
3873
			status = be_cmd_pmac_add(adapter, mac,
3874 3875
						 vf_cfg->if_handle,
						 &vf_cfg->pmac_id, vf + 1);
3876 3877 3878
		else
			status = be_cmd_set_mac(adapter, mac, vf_cfg->if_handle,
						vf + 1);
3879

3880 3881
		if (status)
			dev_err(&adapter->pdev->dev,
3882 3883
				"Mac address assignment failed for VF %d\n",
				vf);
3884
		else
3885
			memcpy(vf_cfg->mac_addr, mac, ETH_ALEN);
3886 3887 3888 3889 3890 3891

		mac[5] += 1;
	}
	return status;
}

3892 3893 3894 3895 3896 3897 3898
static int be_vfs_mac_query(struct be_adapter *adapter)
{
	int status, vf;
	u8 mac[ETH_ALEN];
	struct be_vf_cfg *vf_cfg;

	for_all_vfs(adapter, vf_cfg, vf) {
3899 3900 3901
		status = be_cmd_get_active_mac(adapter, vf_cfg->pmac_id,
					       mac, vf_cfg->if_handle,
					       false, vf+1);
3902 3903 3904 3905 3906 3907 3908
		if (status)
			return status;
		memcpy(vf_cfg->mac_addr, mac, ETH_ALEN);
	}
	return 0;
}

3909
static void be_vf_clear(struct be_adapter *adapter)
3910
{
3911
	struct be_vf_cfg *vf_cfg;
3912 3913
	u32 vf;

3914
	if (pci_vfs_assigned(adapter->pdev)) {
3915 3916
		dev_warn(&adapter->pdev->dev,
			 "VFs are assigned to VMs: not disabling VFs\n");
3917 3918 3919
		goto done;
	}

3920 3921
	pci_disable_sriov(adapter->pdev);

3922
	for_all_vfs(adapter, vf_cfg, vf) {
3923
		if (BEx_chip(adapter))
3924 3925
			be_cmd_pmac_del(adapter, vf_cfg->if_handle,
					vf_cfg->pmac_id, vf + 1);
3926 3927 3928
		else
			be_cmd_set_mac(adapter, NULL, vf_cfg->if_handle,
				       vf + 1);
3929

3930 3931
		be_cmd_if_destroy(adapter, vf_cfg->if_handle, vf + 1);
	}
3932 3933 3934 3935 3936

	if (BE3_chip(adapter))
		be_cmd_set_hsw_config(adapter, 0, 0,
				      adapter->if_handle,
				      PORT_FWD_TYPE_PASSTHRU, 0);
3937 3938 3939
done:
	kfree(adapter->vf_cfg);
	adapter->num_vfs = 0;
3940
	adapter->flags &= ~BE_FLAGS_SRIOV_ENABLED;
3941 3942
}

3943 3944 3945 3946 3947 3948 3949 3950
static void be_clear_queues(struct be_adapter *adapter)
{
	be_mcc_queues_destroy(adapter);
	be_rx_cqs_destroy(adapter);
	be_tx_queues_destroy(adapter);
	be_evt_queues_destroy(adapter);
}

3951
static void be_cancel_worker(struct be_adapter *adapter)
3952
{
3953 3954 3955 3956
	if (adapter->flags & BE_FLAGS_WORKER_SCHEDULED) {
		cancel_delayed_work_sync(&adapter->work);
		adapter->flags &= ~BE_FLAGS_WORKER_SCHEDULED;
	}
3957 3958
}

3959 3960
static void be_cancel_err_detection(struct be_adapter *adapter)
{
3961 3962 3963 3964 3965
	struct be_error_recovery *err_rec = &adapter->error_recovery;

	if (!be_err_recovery_workq)
		return;

3966
	if (adapter->flags & BE_FLAGS_ERR_DETECTION_SCHEDULED) {
3967
		cancel_delayed_work_sync(&err_rec->err_detection_work);
3968 3969 3970 3971
		adapter->flags &= ~BE_FLAGS_ERR_DETECTION_SCHEDULED;
	}
}

3972 3973
static void be_disable_vxlan_offloads(struct be_adapter *adapter)
{
3974 3975
	struct net_device *netdev = adapter->netdev;

3976 3977 3978 3979 3980 3981 3982 3983 3984
	if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS)
		be_cmd_manage_iface(adapter, adapter->if_handle,
				    OP_CONVERT_TUNNEL_TO_NORMAL);

	if (adapter->vxlan_port)
		be_cmd_set_vxlan_port(adapter, 0);

	adapter->flags &= ~BE_FLAGS_VXLAN_OFFLOADS;
	adapter->vxlan_port = 0;
3985 3986 3987

	netdev->hw_enc_features = 0;
	netdev->hw_features &= ~(NETIF_F_GSO_UDP_TUNNEL);
3988
	netdev->features &= ~(NETIF_F_GSO_UDP_TUNNEL);
3989 3990
}

3991 3992
static void be_calculate_vf_res(struct be_adapter *adapter, u16 num_vfs,
				struct be_resources *vft_res)
3993 3994
{
	struct be_resources res = adapter->pool_res;
3995 3996
	u32 vf_if_cap_flags = res.vf_if_cap_flags;
	struct be_resources res_mod = {0};
3997 3998
	u16 num_vf_qs = 1;

3999 4000 4001 4002 4003
	/* Distribute the queue resources among the PF and it's VFs */
	if (num_vfs) {
		/* Divide the rx queues evenly among the VFs and the PF, capped
		 * at VF-EQ-count. Any remainder queues belong to the PF.
		 */
4004 4005
		num_vf_qs = min(SH_VF_MAX_NIC_EQS,
				res.max_rss_qs / (num_vfs + 1));
4006

4007 4008 4009
		/* Skyhawk-R chip supports only MAX_PORT_RSS_TABLES
		 * RSS Tables per port. Provide RSS on VFs, only if number of
		 * VFs requested is less than it's PF Pool's RSS Tables limit.
4010
		 */
4011
		if (num_vfs >= be_max_pf_pool_rss_tables(adapter))
4012 4013
			num_vf_qs = 1;
	}
4014 4015 4016 4017

	/* Resource with fields set to all '1's by GET_PROFILE_CONFIG cmd,
	 * which are modifiable using SET_PROFILE_CONFIG cmd.
	 */
4018 4019
	be_cmd_get_profile_config(adapter, &res_mod, NULL, ACTIVE_PROFILE_TYPE,
				  RESOURCE_MODIFIABLE, 0);
4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064

	/* If RSS IFACE capability flags are modifiable for a VF, set the
	 * capability flag as valid and set RSS and DEFQ_RSS IFACE flags if
	 * more than 1 RSSQ is available for a VF.
	 * Otherwise, provision only 1 queue pair for VF.
	 */
	if (res_mod.vf_if_cap_flags & BE_IF_FLAGS_RSS) {
		vft_res->flags |= BIT(IF_CAPS_FLAGS_VALID_SHIFT);
		if (num_vf_qs > 1) {
			vf_if_cap_flags |= BE_IF_FLAGS_RSS;
			if (res.if_cap_flags & BE_IF_FLAGS_DEFQ_RSS)
				vf_if_cap_flags |= BE_IF_FLAGS_DEFQ_RSS;
		} else {
			vf_if_cap_flags &= ~(BE_IF_FLAGS_RSS |
					     BE_IF_FLAGS_DEFQ_RSS);
		}
	} else {
		num_vf_qs = 1;
	}

	if (res_mod.vf_if_cap_flags & BE_IF_FLAGS_VLAN_PROMISCUOUS) {
		vft_res->flags |= BIT(IF_CAPS_FLAGS_VALID_SHIFT);
		vf_if_cap_flags &= ~BE_IF_FLAGS_VLAN_PROMISCUOUS;
	}

	vft_res->vf_if_cap_flags = vf_if_cap_flags;
	vft_res->max_rx_qs = num_vf_qs;
	vft_res->max_rss_qs = num_vf_qs;
	vft_res->max_tx_qs = res.max_tx_qs / (num_vfs + 1);
	vft_res->max_cq_count = res.max_cq_count / (num_vfs + 1);

	/* Distribute unicast MACs, VLANs, IFACE count and MCCQ count equally
	 * among the PF and it's VFs, if the fields are changeable
	 */
	if (res_mod.max_uc_mac == FIELD_MODIFIABLE)
		vft_res->max_uc_mac = res.max_uc_mac / (num_vfs + 1);

	if (res_mod.max_vlans == FIELD_MODIFIABLE)
		vft_res->max_vlans = res.max_vlans / (num_vfs + 1);

	if (res_mod.max_iface_count == FIELD_MODIFIABLE)
		vft_res->max_iface_count = res.max_iface_count / (num_vfs + 1);

	if (res_mod.max_mcc_count == FIELD_MODIFIABLE)
		vft_res->max_mcc_count = res.max_mcc_count / (num_vfs + 1);
4065 4066
}

4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080
static void be_if_destroy(struct be_adapter *adapter)
{
	be_cmd_if_destroy(adapter, adapter->if_handle,  0);

	kfree(adapter->pmac_id);
	adapter->pmac_id = NULL;

	kfree(adapter->mc_list);
	adapter->mc_list = NULL;

	kfree(adapter->uc_list);
	adapter->uc_list = NULL;
}

4081 4082
static int be_clear(struct be_adapter *adapter)
{
4083
	struct pci_dev *pdev = adapter->pdev;
4084
	struct  be_resources vft_res = {0};
4085

4086
	be_cancel_worker(adapter);
4087

4088 4089
	flush_workqueue(be_wq);

4090
	if (sriov_enabled(adapter))
4091 4092
		be_vf_clear(adapter);

4093 4094 4095
	/* Re-configure FW to distribute resources evenly across max-supported
	 * number of VFs, only when VFs are not already enabled.
	 */
4096 4097
	if (skyhawk_chip(adapter) && be_physfn(adapter) &&
	    !pci_vfs_assigned(pdev)) {
4098 4099 4100
		be_calculate_vf_res(adapter,
				    pci_sriov_get_totalvfs(pdev),
				    &vft_res);
4101
		be_cmd_set_sriov_config(adapter, adapter->pool_res,
4102
					pci_sriov_get_totalvfs(pdev),
4103
					&vft_res);
4104
	}
4105

4106
	be_disable_vxlan_offloads(adapter);
4107

4108
	be_if_destroy(adapter);
4109

4110
	be_clear_queues(adapter);
4111

S
Sathya Perla 已提交
4112
	be_msix_disable(adapter);
4113
	adapter->flags &= ~BE_FLAGS_SETUP_DONE;
4114 4115 4116
	return 0;
}

4117
static int be_vfs_if_create(struct be_adapter *adapter)
4118
{
4119
	struct be_resources res = {0};
4120
	u32 cap_flags, en_flags, vf;
4121
	struct be_vf_cfg *vf_cfg;
4122
	int status;
4123

4124
	/* If a FW profile exists, then cap_flags are updated */
4125
	cap_flags = BE_VF_IF_EN_FLAGS;
4126

4127
	for_all_vfs(adapter, vf_cfg, vf) {
4128
		if (!BE3_chip(adapter)) {
4129 4130
			status = be_cmd_get_profile_config(adapter, &res, NULL,
							   ACTIVE_PROFILE_TYPE,
4131
							   RESOURCE_LIMITS,
4132
							   vf + 1);
4133
			if (!status) {
4134
				cap_flags = res.if_cap_flags;
4135 4136 4137 4138 4139
				/* Prevent VFs from enabling VLAN promiscuous
				 * mode
				 */
				cap_flags &= ~BE_IF_FLAGS_VLAN_PROMISCUOUS;
			}
4140
		}
4141

4142 4143
		/* PF should enable IF flags during proxy if_create call */
		en_flags = cap_flags & BE_VF_IF_EN_FLAGS;
4144 4145
		status = be_cmd_if_create(adapter, cap_flags, en_flags,
					  &vf_cfg->if_handle, vf + 1);
4146
		if (status)
4147
			return status;
4148
	}
4149 4150

	return 0;
4151 4152
}

4153
static int be_vf_setup_init(struct be_adapter *adapter)
4154
{
4155
	struct be_vf_cfg *vf_cfg;
4156 4157
	int vf;

4158 4159 4160 4161 4162
	adapter->vf_cfg = kcalloc(adapter->num_vfs, sizeof(*vf_cfg),
				  GFP_KERNEL);
	if (!adapter->vf_cfg)
		return -ENOMEM;

4163 4164 4165
	for_all_vfs(adapter, vf_cfg, vf) {
		vf_cfg->if_handle = -1;
		vf_cfg->pmac_id = -1;
4166
	}
4167
	return 0;
4168 4169
}

4170 4171
static int be_vf_setup(struct be_adapter *adapter)
{
4172
	struct device *dev = &adapter->pdev->dev;
4173
	struct be_vf_cfg *vf_cfg;
4174
	int status, old_vfs, vf;
4175
	bool spoofchk;
4176

4177
	old_vfs = pci_num_vf(adapter->pdev);
4178 4179 4180 4181

	status = be_vf_setup_init(adapter);
	if (status)
		goto err;
4182

4183 4184 4185 4186 4187 4188
	if (old_vfs) {
		for_all_vfs(adapter, vf_cfg, vf) {
			status = be_cmd_get_if_id(adapter, vf_cfg, vf);
			if (status)
				goto err;
		}
4189

4190 4191 4192 4193
		status = be_vfs_mac_query(adapter);
		if (status)
			goto err;
	} else {
4194 4195 4196 4197
		status = be_vfs_if_create(adapter);
		if (status)
			goto err;

4198 4199 4200 4201
		status = be_vf_eth_addr_config(adapter);
		if (status)
			goto err;
	}
4202

4203
	for_all_vfs(adapter, vf_cfg, vf) {
4204
		/* Allow VFs to programs MAC/VLAN filters */
4205 4206 4207
		status = be_cmd_get_fn_privileges(adapter, &vf_cfg->privileges,
						  vf + 1);
		if (!status && !(vf_cfg->privileges & BE_PRIV_FILTMGMT)) {
4208
			status = be_cmd_set_fn_privileges(adapter,
4209
							  vf_cfg->privileges |
4210 4211
							  BE_PRIV_FILTMGMT,
							  vf + 1);
4212 4213
			if (!status) {
				vf_cfg->privileges |= BE_PRIV_FILTMGMT;
4214 4215
				dev_info(dev, "VF%d has FILTMGMT privilege\n",
					 vf);
4216
			}
4217 4218
		}

4219 4220 4221
		/* Allow full available bandwidth */
		if (!old_vfs)
			be_cmd_config_qos(adapter, 0, 0, vf + 1);
4222

4223 4224 4225 4226 4227 4228
		status = be_cmd_get_hsw_config(adapter, NULL, vf + 1,
					       vf_cfg->if_handle, NULL,
					       &spoofchk);
		if (!status)
			vf_cfg->spoofchk = spoofchk;

4229
		if (!old_vfs) {
4230
			be_cmd_enable_vf(adapter, vf + 1);
4231 4232 4233 4234
			be_cmd_set_logical_link_config(adapter,
						       IFLA_VF_LINK_STATE_AUTO,
						       vf+1);
		}
4235
	}
4236 4237 4238 4239 4240 4241 4242 4243 4244

	if (!old_vfs) {
		status = pci_enable_sriov(adapter->pdev, adapter->num_vfs);
		if (status) {
			dev_err(dev, "SRIOV enable failed\n");
			adapter->num_vfs = 0;
			goto err;
		}
	}
4245

4246 4247 4248 4249 4250 4251 4252 4253 4254
	if (BE3_chip(adapter)) {
		/* On BE3, enable VEB only when SRIOV is enabled */
		status = be_cmd_set_hsw_config(adapter, 0, 0,
					       adapter->if_handle,
					       PORT_FWD_TYPE_VEB, 0);
		if (status)
			goto err;
	}

4255
	adapter->flags |= BE_FLAGS_SRIOV_ENABLED;
4256 4257
	return 0;
err:
4258 4259
	dev_err(dev, "VF setup failed\n");
	be_vf_clear(adapter);
4260 4261 4262
	return status;
}

4263 4264 4265 4266
/* Converting function_mode bits on BE3 to SH mc_type enums */

static u8 be_convert_mc_type(u32 function_mode)
{
4267
	if (function_mode & VNIC_MODE && function_mode & QNQ_MODE)
4268
		return vNIC1;
4269
	else if (function_mode & QNQ_MODE)
4270 4271 4272 4273 4274 4275 4276 4277 4278
		return FLEX10;
	else if (function_mode & VNIC_MODE)
		return vNIC2;
	else if (function_mode & UMC_ENABLED)
		return UMC;
	else
		return MC_NONE;
}

4279 4280 4281 4282
/* On BE2/BE3 FW does not suggest the supported limits */
static void BEx_get_resources(struct be_adapter *adapter,
			      struct be_resources *res)
{
4283
	bool use_sriov = adapter->num_vfs ? 1 : 0;
4284 4285 4286 4287 4288 4289

	if (be_physfn(adapter))
		res->max_uc_mac = BE_UC_PMAC_COUNT;
	else
		res->max_uc_mac = BE_VF_UC_PMAC_COUNT;

4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303
	adapter->mc_type = be_convert_mc_type(adapter->function_mode);

	if (be_is_mc(adapter)) {
		/* Assuming that there are 4 channels per port,
		 * when multi-channel is enabled
		 */
		if (be_is_qnq_mode(adapter))
			res->max_vlans = BE_NUM_VLANS_SUPPORTED/8;
		else
			/* In a non-qnq multichannel mode, the pvid
			 * takes up one vlan entry
			 */
			res->max_vlans = (BE_NUM_VLANS_SUPPORTED / 4) - 1;
	} else {
4304
		res->max_vlans = BE_NUM_VLANS_SUPPORTED;
4305 4306
	}

4307 4308
	res->max_mcast_mac = BE_MAX_MC;

4309 4310 4311 4312 4313
	/* 1) For BE3 1Gb ports, FW does not support multiple TXQs
	 * 2) Create multiple TX rings on a BE3-R multi-channel interface
	 *    *only* if it is RSS-capable.
	 */
	if (BE2_chip(adapter) || use_sriov ||  (adapter->port_num > 1) ||
4314 4315 4316
	    be_virtfn(adapter) ||
	    (be_is_mc(adapter) &&
	     !(adapter->function_caps & BE_FUNCTION_CAPS_RSS))) {
4317
		res->max_tx_qs = 1;
4318 4319 4320 4321 4322 4323
	} else if (adapter->function_caps & BE_FUNCTION_CAPS_SUPER_NIC) {
		struct be_resources super_nic_res = {0};

		/* On a SuperNIC profile, the driver needs to use the
		 * GET_PROFILE_CONFIG cmd to query the per-function TXQ limits
		 */
4324 4325 4326
		be_cmd_get_profile_config(adapter, &super_nic_res, NULL,
					  ACTIVE_PROFILE_TYPE, RESOURCE_LIMITS,
					  0);
4327 4328 4329
		/* Some old versions of BE3 FW don't report max_tx_qs value */
		res->max_tx_qs = super_nic_res.max_tx_qs ? : BE3_MAX_TX_QS;
	} else {
4330
		res->max_tx_qs = BE3_MAX_TX_QS;
4331
	}
4332 4333 4334 4335 4336 4337 4338

	if ((adapter->function_caps & BE_FUNCTION_CAPS_RSS) &&
	    !use_sriov && be_physfn(adapter))
		res->max_rss_qs = (adapter->be3_native) ?
					   BE3_MAX_RSS_QS : BE2_MAX_RSS_QS;
	res->max_rx_qs = res->max_rss_qs + 1;

4339
	if (be_physfn(adapter))
4340
		res->max_evt_qs = (be_max_vfs(adapter) > 0) ?
4341 4342 4343
					BE3_SRIOV_MAX_EVT_QS : BE3_MAX_EVT_QS;
	else
		res->max_evt_qs = 1;
4344 4345

	res->if_cap_flags = BE_IF_CAP_FLAGS_WANT;
4346
	res->if_cap_flags &= ~BE_IF_FLAGS_DEFQ_RSS;
4347 4348 4349 4350
	if (!(adapter->function_caps & BE_FUNCTION_CAPS_RSS))
		res->if_cap_flags &= ~BE_IF_FLAGS_RSS;
}

4351 4352 4353
static void be_setup_init(struct be_adapter *adapter)
{
	adapter->vlan_prio_bmap = 0xff;
A
Ajit Khaparde 已提交
4354
	adapter->phy.link_speed = -1;
4355 4356
	adapter->if_handle = -1;
	adapter->be3_native = false;
4357
	adapter->if_flags = 0;
4358
	adapter->phy_state = BE_UNKNOWN_PHY_STATE;
4359 4360 4361 4362
	if (be_physfn(adapter))
		adapter->cmd_privileges = MAX_PRIVILEGES;
	else
		adapter->cmd_privileges = MIN_PRIVILEGES;
4363 4364
}

4365 4366 4367 4368 4369 4370 4371
/* HW supports only MAX_PORT_RSS_TABLES RSS Policy Tables per port.
 * However, this HW limitation is not exposed to the host via any SLI cmd.
 * As a result, in the case of SRIOV and in particular multi-partition configs
 * the driver needs to calcuate a proportional share of RSS Tables per PF-pool
 * for distribution between the VFs. This self-imposed limit will determine the
 * no: of VFs for which RSS can be enabled.
 */
4372
static void be_calculate_pf_pool_rss_tables(struct be_adapter *adapter)
4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389
{
	struct be_port_resources port_res = {0};
	u8 rss_tables_on_port;
	u16 max_vfs = be_max_vfs(adapter);

	be_cmd_get_profile_config(adapter, NULL, &port_res, SAVED_PROFILE_TYPE,
				  RESOURCE_LIMITS, 0);

	rss_tables_on_port = MAX_PORT_RSS_TABLES - port_res.nic_pfs;

	/* Each PF Pool's RSS Tables limit =
	 * PF's Max VFs / Total_Max_VFs on Port * RSS Tables on Port
	 */
	adapter->pool_res.max_rss_tables =
		max_vfs * rss_tables_on_port / port_res.max_vfs;
}

4390 4391 4392
static int be_get_sriov_config(struct be_adapter *adapter)
{
	struct be_resources res = {0};
4393
	int max_vfs, old_vfs;
4394

4395 4396
	be_cmd_get_profile_config(adapter, &res, NULL, ACTIVE_PROFILE_TYPE,
				  RESOURCE_LIMITS, 0);
4397

4398
	/* Some old versions of BE3 FW don't report max_vfs value */
4399 4400 4401 4402 4403
	if (BE3_chip(adapter) && !res.max_vfs) {
		max_vfs = pci_sriov_get_totalvfs(adapter->pdev);
		res.max_vfs = max_vfs > 0 ? min(MAX_VFS, max_vfs) : 0;
	}

4404
	adapter->pool_res = res;
4405

4406 4407 4408 4409
	/* If during previous unload of the driver, the VFs were not disabled,
	 * then we cannot rely on the PF POOL limits for the TotalVFs value.
	 * Instead use the TotalVFs value stored in the pci-dev struct.
	 */
4410 4411
	old_vfs = pci_num_vf(adapter->pdev);
	if (old_vfs) {
4412 4413 4414 4415 4416
		dev_info(&adapter->pdev->dev, "%d VFs are already enabled\n",
			 old_vfs);

		adapter->pool_res.max_vfs =
			pci_sriov_get_totalvfs(adapter->pdev);
4417 4418 4419
		adapter->num_vfs = old_vfs;
	}

4420 4421 4422 4423 4424 4425
	if (skyhawk_chip(adapter) && be_max_vfs(adapter) && !old_vfs) {
		be_calculate_pf_pool_rss_tables(adapter);
		dev_info(&adapter->pdev->dev,
			 "RSS can be enabled for all VFs if num_vfs <= %d\n",
			 be_max_pf_pool_rss_tables(adapter));
	}
4426 4427 4428
	return 0;
}

4429 4430 4431
static void be_alloc_sriov_res(struct be_adapter *adapter)
{
	int old_vfs = pci_num_vf(adapter->pdev);
4432
	struct  be_resources vft_res = {0};
4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445
	int status;

	be_get_sriov_config(adapter);

	if (!old_vfs)
		pci_sriov_set_totalvfs(adapter->pdev, be_max_vfs(adapter));

	/* When the HW is in SRIOV capable configuration, the PF-pool
	 * resources are given to PF during driver load, if there are no
	 * old VFs. This facility is not available in BE3 FW.
	 * Also, this is done by FW in Lancer chip.
	 */
	if (skyhawk_chip(adapter) && be_max_vfs(adapter) && !old_vfs) {
4446
		be_calculate_vf_res(adapter, 0, &vft_res);
4447
		status = be_cmd_set_sriov_config(adapter, adapter->pool_res, 0,
4448
						 &vft_res);
4449 4450 4451 4452 4453 4454
		if (status)
			dev_err(&adapter->pdev->dev,
				"Failed to optimize SRIOV resources\n");
	}
}

4455
static int be_get_resources(struct be_adapter *adapter)
4456
{
4457 4458 4459
	struct device *dev = &adapter->pdev->dev;
	struct be_resources res = {0};
	int status;
4460

4461 4462 4463 4464
	/* For Lancer, SH etc read per-function resource limits from FW.
	 * GET_FUNC_CONFIG returns per function guaranteed limits.
	 * GET_PROFILE_CONFIG returns PCI-E related limits PF-pool limits
	 */
4465 4466 4467
	if (BEx_chip(adapter)) {
		BEx_get_resources(adapter, &res);
	} else {
4468 4469 4470
		status = be_cmd_get_func_config(adapter, &res);
		if (status)
			return status;
4471

4472 4473 4474 4475
		/* If a deafault RXQ must be created, we'll use up one RSSQ*/
		if (res.max_rss_qs && res.max_rss_qs == res.max_rx_qs &&
		    !(res.if_cap_flags & BE_IF_FLAGS_DEFQ_RSS))
			res.max_rss_qs -= 1;
4476
	}
4477

4478 4479 4480 4481 4482
	/* If RoCE is supported stash away half the EQs for RoCE */
	res.max_nic_evt_qs = be_roce_supported(adapter) ?
				res.max_evt_qs / 2 : res.max_evt_qs;
	adapter->res = res;

4483 4484 4485 4486 4487 4488
	/* If FW supports RSS default queue, then skip creating non-RSS
	 * queue for non-IP traffic.
	 */
	adapter->need_def_rxq = (be_if_cap_flags(adapter) &
				 BE_IF_FLAGS_DEFQ_RSS) ? 0 : 1;

S
Sathya Perla 已提交
4489 4490
	dev_info(dev, "Max: txqs %d, rxqs %d, rss %d, eqs %d, vfs %d\n",
		 be_max_txqs(adapter), be_max_rxqs(adapter),
4491
		 be_max_rss(adapter), be_max_nic_eqs(adapter),
S
Sathya Perla 已提交
4492 4493 4494 4495 4496
		 be_max_vfs(adapter));
	dev_info(dev, "Max: uc-macs %d, mc-macs %d, vlans %d\n",
		 be_max_uc(adapter), be_max_mc(adapter),
		 be_max_vlans(adapter));

4497 4498 4499 4500 4501
	/* Ensure RX and TX queues are created in pairs at init time */
	adapter->cfg_num_rx_irqs =
				min_t(u16, netif_get_num_default_rss_queues(),
				      be_max_qp_irqs(adapter));
	adapter->cfg_num_tx_irqs = adapter->cfg_num_rx_irqs;
4502
	return 0;
4503 4504
}

4505 4506
static int be_get_config(struct be_adapter *adapter)
{
4507
	int status, level;
4508
	u16 profile_id;
4509

4510 4511 4512 4513
	status = be_cmd_get_cntl_attributes(adapter);
	if (status)
		return status;

4514
	status = be_cmd_query_fw_cfg(adapter);
4515
	if (status)
4516
		return status;
4517

4518 4519 4520
	if (!lancer_chip(adapter) && be_physfn(adapter))
		be_cmd_get_fat_dump_len(adapter, &adapter->fat_dump_len);

4521 4522 4523 4524 4525 4526 4527
	if (BEx_chip(adapter)) {
		level = be_cmd_get_fw_log_level(adapter);
		adapter->msg_enable =
			level <= FW_LOG_LEVEL_DEFAULT ? NETIF_MSG_HW : 0;
	}

	be_cmd_get_acpi_wol_cap(adapter);
4528 4529
	pci_enable_wake(adapter->pdev, PCI_D3hot, adapter->wol_en);
	pci_enable_wake(adapter->pdev, PCI_D3cold, adapter->wol_en);
4530

4531 4532 4533
	be_cmd_query_port_name(adapter);

	if (be_physfn(adapter)) {
4534 4535 4536 4537
		status = be_cmd_get_active_profile(adapter, &profile_id);
		if (!status)
			dev_info(&adapter->pdev->dev,
				 "Using profile 0x%x\n", profile_id);
4538
	}
4539

4540
	return 0;
4541 4542
}

4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559
static int be_mac_setup(struct be_adapter *adapter)
{
	u8 mac[ETH_ALEN];
	int status;

	if (is_zero_ether_addr(adapter->netdev->dev_addr)) {
		status = be_cmd_get_perm_mac(adapter, mac);
		if (status)
			return status;

		memcpy(adapter->netdev->dev_addr, mac, ETH_ALEN);
		memcpy(adapter->netdev->perm_addr, mac, ETH_ALEN);
	}

	return 0;
}

4560 4561
static void be_schedule_worker(struct be_adapter *adapter)
{
4562
	queue_delayed_work(be_wq, &adapter->work, msecs_to_jiffies(1000));
4563 4564 4565
	adapter->flags |= BE_FLAGS_WORKER_SCHEDULED;
}

4566 4567 4568 4569 4570 4571 4572 4573 4574 4575
static void be_destroy_err_recovery_workq(void)
{
	if (!be_err_recovery_workq)
		return;

	flush_workqueue(be_err_recovery_workq);
	destroy_workqueue(be_err_recovery_workq);
	be_err_recovery_workq = NULL;
}

4576
static void be_schedule_err_detection(struct be_adapter *adapter, u32 delay)
4577
{
4578 4579 4580 4581 4582 4583 4584
	struct be_error_recovery *err_rec = &adapter->error_recovery;

	if (!be_err_recovery_workq)
		return;

	queue_delayed_work(be_err_recovery_workq, &err_rec->err_detection_work,
			   msecs_to_jiffies(delay));
4585 4586 4587
	adapter->flags |= BE_FLAGS_ERR_DETECTION_SCHEDULED;
}

4588
static int be_setup_queues(struct be_adapter *adapter)
4589
{
4590
	struct net_device *netdev = adapter->netdev;
S
Sathya Perla 已提交
4591
	int status;
4592

4593
	status = be_evt_queues_create(adapter);
4594 4595
	if (status)
		goto err;
4596

4597
	status = be_tx_qs_create(adapter);
4598 4599
	if (status)
		goto err;
S
Sathya Perla 已提交
4600

4601
	status = be_rx_cqs_create(adapter);
S
Sathya Perla 已提交
4602
	if (status)
4603
		goto err;
S
Sathya Perla 已提交
4604

4605
	status = be_mcc_queues_create(adapter);
S
Sathya Perla 已提交
4606 4607 4608
	if (status)
		goto err;

4609 4610 4611 4612 4613 4614 4615 4616
	status = netif_set_real_num_rx_queues(netdev, adapter->num_rx_qs);
	if (status)
		goto err;

	status = netif_set_real_num_tx_queues(netdev, adapter->num_tx_qs);
	if (status)
		goto err;

4617 4618 4619 4620 4621 4622
	return 0;
err:
	dev_err(&adapter->pdev->dev, "queue_setup failed\n");
	return status;
}

4623 4624 4625 4626 4627 4628
static int be_if_create(struct be_adapter *adapter)
{
	u32 en_flags = BE_IF_FLAGS_RSS | BE_IF_FLAGS_DEFQ_RSS;
	u32 cap_flags = be_if_cap_flags(adapter);
	int status;

4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644
	/* alloc required memory for other filtering fields */
	adapter->pmac_id = kcalloc(be_max_uc(adapter),
				   sizeof(*adapter->pmac_id), GFP_KERNEL);
	if (!adapter->pmac_id)
		return -ENOMEM;

	adapter->mc_list = kcalloc(be_max_mc(adapter),
				   sizeof(*adapter->mc_list), GFP_KERNEL);
	if (!adapter->mc_list)
		return -ENOMEM;

	adapter->uc_list = kcalloc(be_max_uc(adapter),
				   sizeof(*adapter->uc_list), GFP_KERNEL);
	if (!adapter->uc_list)
		return -ENOMEM;

4645
	if (adapter->cfg_num_rx_irqs == 1)
4646 4647 4648 4649 4650 4651 4652
		cap_flags &= ~(BE_IF_FLAGS_DEFQ_RSS | BE_IF_FLAGS_RSS);

	en_flags &= cap_flags;
	/* will enable all the needed filter flags in be_open() */
	status = be_cmd_if_create(adapter, be_if_cap_flags(adapter), en_flags,
				  &adapter->if_handle, 0);

4653 4654 4655 4656
	if (status)
		return status;

	return 0;
4657 4658
}

4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675
int be_update_queues(struct be_adapter *adapter)
{
	struct net_device *netdev = adapter->netdev;
	int status;

	if (netif_running(netdev))
		be_close(netdev);

	be_cancel_worker(adapter);

	/* If any vectors have been shared with RoCE we cannot re-program
	 * the MSIx table.
	 */
	if (!adapter->num_msix_roce_vec)
		be_msix_disable(adapter);

	be_clear_queues(adapter);
4676 4677 4678
	status = be_cmd_if_destroy(adapter, adapter->if_handle,  0);
	if (status)
		return status;
4679 4680 4681 4682 4683 4684 4685

	if (!msix_enabled(adapter)) {
		status = be_msix_enable(adapter);
		if (status)
			return status;
	}

4686 4687 4688 4689
	status = be_if_create(adapter);
	if (status)
		return status;

4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701
	status = be_setup_queues(adapter);
	if (status)
		return status;

	be_schedule_worker(adapter);

	if (netif_running(netdev))
		status = be_open(netdev);

	return status;
}

4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712
static inline int fw_major_num(const char *fw_ver)
{
	int fw_major = 0, i;

	i = sscanf(fw_ver, "%d.", &fw_major);
	if (i != 1)
		return 0;

	return fw_major;
}

4713 4714 4715
/* If it is error recovery, FLR the PF
 * Else if any VFs are already enabled don't FLR the PF
 */
4716 4717
static bool be_reset_required(struct be_adapter *adapter)
{
4718 4719 4720 4721
	if (be_error_recovering(adapter))
		return true;
	else
		return pci_num_vf(adapter->pdev) == 0;
4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732
}

/* Wait for the FW to be ready and perform the required initialization */
static int be_func_init(struct be_adapter *adapter)
{
	int status;

	status = be_fw_wait_ready(adapter);
	if (status)
		return status;

4733 4734 4735
	/* FW is now ready; clear errors to allow cmds/doorbell */
	be_clear_error(adapter, BE_CLEAR_ALL);

4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755
	if (be_reset_required(adapter)) {
		status = be_cmd_reset_function(adapter);
		if (status)
			return status;

		/* Wait for interrupts to quiesce after an FLR */
		msleep(100);
	}

	/* Tell FW we're ready to fire cmds */
	status = be_cmd_fw_init(adapter);
	if (status)
		return status;

	/* Allow interrupts for other ULPs running on NIC function */
	be_intr_set(adapter, true);

	return 0;
}

4756 4757 4758 4759 4760
static int be_setup(struct be_adapter *adapter)
{
	struct device *dev = &adapter->pdev->dev;
	int status;

4761 4762 4763 4764
	status = be_func_init(adapter);
	if (status)
		return status;

4765 4766 4767 4768 4769
	be_setup_init(adapter);

	if (!lancer_chip(adapter))
		be_cmd_req_native_mode(adapter);

4770 4771 4772 4773 4774 4775 4776 4777
	/* invoke this cmd first to get pf_num and vf_num which are needed
	 * for issuing profile related cmds
	 */
	if (!BEx_chip(adapter)) {
		status = be_cmd_get_func_config(adapter, NULL);
		if (status)
			return status;
	}
4778

4779 4780 4781 4782
	status = be_get_config(adapter);
	if (status)
		goto err;

4783 4784 4785
	if (!BE2_chip(adapter) && be_physfn(adapter))
		be_alloc_sriov_res(adapter);

4786
	status = be_get_resources(adapter);
S
Sathya Perla 已提交
4787
	if (status)
4788
		goto err;
S
Sathya Perla 已提交
4789

4790
	status = be_msix_enable(adapter);
S
Sathya Perla 已提交
4791
	if (status)
4792
		goto err;
S
Sathya Perla 已提交
4793

4794
	/* will enable all the needed filter flags in be_open() */
4795
	status = be_if_create(adapter);
4796
	if (status)
4797
		goto err;
S
Sathya Perla 已提交
4798

4799 4800
	/* Updating real_num_tx/rx_queues() requires rtnl_lock() */
	rtnl_lock();
4801
	status = be_setup_queues(adapter);
4802
	rtnl_unlock();
4803
	if (status)
4804 4805
		goto err;

4806 4807 4808
	be_cmd_get_fn_privileges(adapter, &adapter->cmd_privileges, 0);

	status = be_mac_setup(adapter);
S
Sathya Perla 已提交
4809 4810 4811
	if (status)
		goto err;

4812
	be_cmd_get_fw_ver(adapter);
S
Sathya Perla 已提交
4813
	dev_info(dev, "FW version is %s\n", adapter->fw_ver);
4814

4815
	if (BE2_chip(adapter) && fw_major_num(adapter->fw_ver) < 4) {
V
Vasundhara Volam 已提交
4816
		dev_err(dev, "Firmware on card is old(%s), IRQs may not work",
4817 4818 4819 4820
			adapter->fw_ver);
		dev_err(dev, "Please upgrade firmware to version >= 4.0\n");
	}

4821 4822 4823 4824 4825
	status = be_cmd_set_flow_control(adapter, adapter->tx_fc,
					 adapter->rx_fc);
	if (status)
		be_cmd_get_flow_control(adapter, &adapter->tx_fc,
					&adapter->rx_fc);
4826

4827 4828
	dev_info(&adapter->pdev->dev, "HW Flow control - TX:%d RX:%d\n",
		 adapter->tx_fc, adapter->rx_fc);
4829

4830 4831 4832 4833
	if (be_physfn(adapter))
		be_cmd_set_logical_link_config(adapter,
					       IFLA_VF_LINK_STATE_AUTO, 0);

4834 4835 4836 4837 4838 4839 4840 4841 4842
	/* BE3 EVB echoes broadcast/multicast packets back to PF's vport
	 * confusing a linux bridge or OVS that it might be connected to.
	 * Set the EVB to PASSTHRU mode which effectively disables the EVB
	 * when SRIOV is not enabled.
	 */
	if (BE3_chip(adapter))
		be_cmd_set_hsw_config(adapter, 0, 0, adapter->if_handle,
				      PORT_FWD_TYPE_PASSTHRU, 0);

4843 4844
	if (adapter->num_vfs)
		be_vf_setup(adapter);
4845

4846 4847
	status = be_cmd_get_phy_info(adapter);
	if (!status && be_pause_supported(adapter))
A
Ajit Khaparde 已提交
4848 4849
		adapter->phy.fc_autoneg = 1;

4850 4851 4852
	if (be_physfn(adapter) && !lancer_chip(adapter))
		be_cmd_set_features(adapter);

4853
	be_schedule_worker(adapter);
4854
	adapter->flags |= BE_FLAGS_SETUP_DONE;
4855
	return 0;
4856 4857 4858 4859
err:
	be_clear(adapter);
	return status;
}
S
Sathya Perla 已提交
4860

I
Ivan Vecera 已提交
4861 4862 4863 4864
#ifdef CONFIG_NET_POLL_CONTROLLER
static void be_netpoll(struct net_device *netdev)
{
	struct be_adapter *adapter = netdev_priv(netdev);
S
Sathya Perla 已提交
4865
	struct be_eq_obj *eqo;
I
Ivan Vecera 已提交
4866 4867
	int i;

4868
	for_all_evt_queues(adapter, eqo, i) {
4869
		be_eq_notify(eqo->adapter, eqo->q.id, false, true, 0, 0);
4870 4871
		napi_schedule(&eqo->napi);
	}
I
Ivan Vecera 已提交
4872 4873 4874
}
#endif

4875 4876 4877 4878 4879 4880 4881 4882
int be_load_fw(struct be_adapter *adapter, u8 *fw_file)
{
	const struct firmware *fw;
	int status;

	if (!netif_running(adapter->netdev)) {
		dev_err(&adapter->pdev->dev,
			"Firmware load not allowed (interface is down)\n");
4883
		return -ENETDOWN;
4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896
	}

	status = request_firmware(&fw, fw_file, &adapter->pdev->dev);
	if (status)
		goto fw_exit;

	dev_info(&adapter->pdev->dev, "Flashing firmware file %s\n", fw_file);

	if (lancer_chip(adapter))
		status = lancer_fw_download(adapter, fw);
	else
		status = be_fw_download(adapter, fw);

S
Somnath Kotur 已提交
4897
	if (!status)
4898
		be_cmd_get_fw_ver(adapter);
S
Somnath Kotur 已提交
4899

4900 4901 4902 4903 4904
fw_exit:
	release_firmware(fw);
	return status;
}

4905 4906
static int be_ndo_bridge_setlink(struct net_device *dev, struct nlmsghdr *nlh,
				 u16 flags)
4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917
{
	struct be_adapter *adapter = netdev_priv(dev);
	struct nlattr *attr, *br_spec;
	int rem;
	int status = 0;
	u16 mode = 0;

	if (!sriov_enabled(adapter))
		return -EOPNOTSUPP;

	br_spec = nlmsg_find_attr(nlh, sizeof(struct ifinfomsg), IFLA_AF_SPEC);
4918 4919
	if (!br_spec)
		return -EINVAL;
4920 4921 4922 4923 4924

	nla_for_each_nested(attr, br_spec, rem) {
		if (nla_type(attr) != IFLA_BRIDGE_MODE)
			continue;

4925 4926 4927
		if (nla_len(attr) < sizeof(mode))
			return -EINVAL;

4928
		mode = nla_get_u16(attr);
4929 4930 4931
		if (BE3_chip(adapter) && mode == BRIDGE_MODE_VEPA)
			return -EOPNOTSUPP;

4932 4933 4934 4935 4936 4937 4938
		if (mode != BRIDGE_MODE_VEPA && mode != BRIDGE_MODE_VEB)
			return -EINVAL;

		status = be_cmd_set_hsw_config(adapter, 0, 0,
					       adapter->if_handle,
					       mode == BRIDGE_MODE_VEPA ?
					       PORT_FWD_TYPE_VEPA :
4939
					       PORT_FWD_TYPE_VEB, 0);
4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955
		if (status)
			goto err;

		dev_info(&adapter->pdev->dev, "enabled switch mode: %s\n",
			 mode == BRIDGE_MODE_VEPA ? "VEPA" : "VEB");

		return status;
	}
err:
	dev_err(&adapter->pdev->dev, "Failed to set switch mode %s\n",
		mode == BRIDGE_MODE_VEPA ? "VEPA" : "VEB");

	return status;
}

static int be_ndo_bridge_getlink(struct sk_buff *skb, u32 pid, u32 seq,
4956 4957
				 struct net_device *dev, u32 filter_mask,
				 int nlflags)
4958 4959 4960 4961 4962 4963 4964
{
	struct be_adapter *adapter = netdev_priv(dev);
	int status = 0;
	u8 hsw_mode;

	/* BE and Lancer chips support VEB mode only */
	if (BEx_chip(adapter) || lancer_chip(adapter)) {
4965 4966 4967
		/* VEB is disabled in non-SR-IOV profiles on BE3/Lancer */
		if (!pci_sriov_get_totalvfs(adapter->pdev))
			return 0;
4968 4969 4970
		hsw_mode = PORT_FWD_TYPE_VEB;
	} else {
		status = be_cmd_get_hsw_config(adapter, NULL, 0,
4971 4972
					       adapter->if_handle, &hsw_mode,
					       NULL);
4973 4974
		if (status)
			return 0;
4975 4976 4977

		if (hsw_mode == PORT_FWD_TYPE_PASSTHRU)
			return 0;
4978 4979 4980 4981
	}

	return ndo_dflt_bridge_getlink(skb, pid, seq, dev,
				       hsw_mode == PORT_FWD_TYPE_VEPA ?
4982
				       BRIDGE_MODE_VEPA : BRIDGE_MODE_VEB,
4983
				       0, 0, nlflags, filter_mask, NULL);
4984 4985
}

4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002
static struct be_cmd_work *be_alloc_work(struct be_adapter *adapter,
					 void (*func)(struct work_struct *))
{
	struct be_cmd_work *work;

	work = kzalloc(sizeof(*work), GFP_ATOMIC);
	if (!work) {
		dev_err(&adapter->pdev->dev,
			"be_work memory allocation failed\n");
		return NULL;
	}

	INIT_WORK(&work->work, func);
	work->adapter = adapter;
	return work;
}

5003 5004 5005 5006 5007 5008
/* VxLAN offload Notes:
 *
 * The stack defines tunnel offload flags (hw_enc_features) for IP and doesn't
 * distinguish various types of transports (VxLAN, GRE, NVGRE ..). So, offload
 * is expected to work across all types of IP tunnels once exported. Skyhawk
 * supports offloads for either VxLAN or NVGRE, exclusively. So we export VxLAN
5009 5010 5011
 * offloads in hw_enc_features only when a VxLAN port is added. If other (non
 * VxLAN) tunnels are configured while VxLAN offloads are enabled, offloads for
 * those other tunnels are unexported on the fly through ndo_features_check().
5012 5013 5014 5015 5016
 *
 * Skyhawk supports VxLAN offloads only for one UDP dport. So, if the stack
 * adds more than one port, disable offloads and don't re-enable them again
 * until after all the tunnels are removed.
 */
5017
static void be_work_add_vxlan_port(struct work_struct *work)
5018
{
5019 5020 5021 5022
	struct be_cmd_work *cmd_work =
				container_of(work, struct be_cmd_work, work);
	struct be_adapter *adapter = cmd_work->adapter;
	struct net_device *netdev = adapter->netdev;
5023
	struct device *dev = &adapter->pdev->dev;
5024
	__be16 port = cmd_work->info.vxlan_port;
5025 5026
	int status;

5027 5028
	if (adapter->vxlan_port == port && adapter->vxlan_port_count) {
		adapter->vxlan_port_aliases++;
5029
		goto done;
5030 5031
	}

5032 5033 5034
	if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS) {
		dev_info(dev,
			 "Only one UDP port supported for VxLAN offloads\n");
5035 5036 5037
		dev_info(dev, "Disabling VxLAN offloads\n");
		adapter->vxlan_port_count++;
		goto err;
5038 5039
	}

5040
	if (adapter->vxlan_port_count++ >= 1)
5041
		goto done;
5042

5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057
	status = be_cmd_manage_iface(adapter, adapter->if_handle,
				     OP_CONVERT_NORMAL_TO_TUNNEL);
	if (status) {
		dev_warn(dev, "Failed to convert normal interface to tunnel\n");
		goto err;
	}

	status = be_cmd_set_vxlan_port(adapter, port);
	if (status) {
		dev_warn(dev, "Failed to add VxLAN port\n");
		goto err;
	}
	adapter->flags |= BE_FLAGS_VXLAN_OFFLOADS;
	adapter->vxlan_port = port;

5058 5059 5060 5061
	netdev->hw_enc_features |= NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
				   NETIF_F_TSO | NETIF_F_TSO6 |
				   NETIF_F_GSO_UDP_TUNNEL;
	netdev->hw_features |= NETIF_F_GSO_UDP_TUNNEL;
5062
	netdev->features |= NETIF_F_GSO_UDP_TUNNEL;
5063

5064 5065
	dev_info(dev, "Enabled VxLAN offloads for UDP port %d\n",
		 be16_to_cpu(port));
5066
	goto done;
5067 5068
err:
	be_disable_vxlan_offloads(adapter);
5069 5070
done:
	kfree(cmd_work);
5071 5072
}

5073
static void be_work_del_vxlan_port(struct work_struct *work)
5074
{
5075 5076 5077 5078
	struct be_cmd_work *cmd_work =
				container_of(work, struct be_cmd_work, work);
	struct be_adapter *adapter = cmd_work->adapter;
	__be16 port = cmd_work->info.vxlan_port;
5079 5080

	if (adapter->vxlan_port != port)
5081
		goto done;
5082

5083 5084
	if (adapter->vxlan_port_aliases) {
		adapter->vxlan_port_aliases--;
5085
		goto out;
5086 5087
	}

5088 5089 5090 5091 5092
	be_disable_vxlan_offloads(adapter);

	dev_info(&adapter->pdev->dev,
		 "Disabled VxLAN offloads for UDP port %d\n",
		 be16_to_cpu(port));
5093 5094
done:
	adapter->vxlan_port_count--;
5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128
out:
	kfree(cmd_work);
}

static void be_cfg_vxlan_port(struct net_device *netdev,
			      struct udp_tunnel_info *ti,
			      void (*func)(struct work_struct *))
{
	struct be_adapter *adapter = netdev_priv(netdev);
	struct be_cmd_work *cmd_work;

	if (ti->type != UDP_TUNNEL_TYPE_VXLAN)
		return;

	if (lancer_chip(adapter) || BEx_chip(adapter) || be_is_mc(adapter))
		return;

	cmd_work = be_alloc_work(adapter, func);
	if (cmd_work) {
		cmd_work->info.vxlan_port = ti->port;
		queue_work(be_wq, &cmd_work->work);
	}
}

static void be_del_vxlan_port(struct net_device *netdev,
			      struct udp_tunnel_info *ti)
{
	be_cfg_vxlan_port(netdev, ti, be_work_del_vxlan_port);
}

static void be_add_vxlan_port(struct net_device *netdev,
			      struct udp_tunnel_info *ti)
{
	be_cfg_vxlan_port(netdev, ti, be_work_add_vxlan_port);
5129
}
J
Joe Stringer 已提交
5130

5131 5132 5133
static netdev_features_t be_features_check(struct sk_buff *skb,
					   struct net_device *dev,
					   netdev_features_t features)
J
Joe Stringer 已提交
5134
{
5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166
	struct be_adapter *adapter = netdev_priv(dev);
	u8 l4_hdr = 0;

	/* The code below restricts offload features for some tunneled packets.
	 * Offload features for normal (non tunnel) packets are unchanged.
	 */
	if (!skb->encapsulation ||
	    !(adapter->flags & BE_FLAGS_VXLAN_OFFLOADS))
		return features;

	/* It's an encapsulated packet and VxLAN offloads are enabled. We
	 * should disable tunnel offload features if it's not a VxLAN packet,
	 * as tunnel offloads have been enabled only for VxLAN. This is done to
	 * allow other tunneled traffic like GRE work fine while VxLAN
	 * offloads are configured in Skyhawk-R.
	 */
	switch (vlan_get_protocol(skb)) {
	case htons(ETH_P_IP):
		l4_hdr = ip_hdr(skb)->protocol;
		break;
	case htons(ETH_P_IPV6):
		l4_hdr = ipv6_hdr(skb)->nexthdr;
		break;
	default:
		return features;
	}

	if (l4_hdr != IPPROTO_UDP ||
	    skb->inner_protocol_type != ENCAP_TYPE_ETHER ||
	    skb->inner_protocol != htons(ETH_P_TEB) ||
	    skb_inner_mac_header(skb) - skb_transport_header(skb) !=
	    sizeof(struct udphdr) + sizeof(struct vxlanhdr))
5167
		return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
5168 5169

	return features;
J
Joe Stringer 已提交
5170
}
5171

5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192
static int be_get_phys_port_id(struct net_device *dev,
			       struct netdev_phys_item_id *ppid)
{
	int i, id_len = CNTL_SERIAL_NUM_WORDS * CNTL_SERIAL_NUM_WORD_SZ + 1;
	struct be_adapter *adapter = netdev_priv(dev);
	u8 *id;

	if (MAX_PHYS_ITEM_ID_LEN < id_len)
		return -ENOSPC;

	ppid->id[0] = adapter->hba_port_num + 1;
	id = &ppid->id[1];
	for (i = CNTL_SERIAL_NUM_WORDS - 1; i >= 0;
	     i--, id += CNTL_SERIAL_NUM_WORD_SZ)
		memcpy(id, &adapter->serial_num[i], CNTL_SERIAL_NUM_WORD_SZ);

	ppid->id_len = id_len;

	return 0;
}

5193 5194 5195 5196 5197 5198 5199 5200 5201 5202
static void be_set_rx_mode(struct net_device *dev)
{
	struct be_adapter *adapter = netdev_priv(dev);
	struct be_cmd_work *work;

	work = be_alloc_work(adapter, be_work_set_rx_mode);
	if (work)
		queue_work(be_wq, &work->work);
}

5203
static const struct net_device_ops be_netdev_ops = {
S
Sathya Perla 已提交
5204 5205 5206
	.ndo_open		= be_open,
	.ndo_stop		= be_close,
	.ndo_start_xmit		= be_xmit,
5207
	.ndo_set_rx_mode	= be_set_rx_mode,
S
Sathya Perla 已提交
5208 5209
	.ndo_set_mac_address	= be_mac_addr_set,
	.ndo_change_mtu		= be_change_mtu,
5210
	.ndo_get_stats64	= be_get_stats64,
S
Sathya Perla 已提交
5211 5212 5213
	.ndo_validate_addr	= eth_validate_addr,
	.ndo_vlan_rx_add_vid	= be_vlan_add_vid,
	.ndo_vlan_rx_kill_vid	= be_vlan_rem_vid,
5214
	.ndo_set_vf_mac		= be_set_vf_mac,
5215
	.ndo_set_vf_vlan	= be_set_vf_vlan,
5216
	.ndo_set_vf_rate	= be_set_vf_tx_rate,
I
Ivan Vecera 已提交
5217
	.ndo_get_vf_config	= be_get_vf_config,
5218
	.ndo_set_vf_link_state  = be_set_vf_link_state,
5219
	.ndo_set_vf_spoofchk    = be_set_vf_spoofchk,
I
Ivan Vecera 已提交
5220 5221 5222
#ifdef CONFIG_NET_POLL_CONTROLLER
	.ndo_poll_controller	= be_netpoll,
#endif
5223 5224
	.ndo_bridge_setlink	= be_ndo_bridge_setlink,
	.ndo_bridge_getlink	= be_ndo_bridge_getlink,
5225
#ifdef CONFIG_NET_RX_BUSY_POLL
5226
	.ndo_busy_poll		= be_busy_poll,
5227
#endif
5228 5229
	.ndo_udp_tunnel_add	= be_add_vxlan_port,
	.ndo_udp_tunnel_del	= be_del_vxlan_port,
5230
	.ndo_features_check	= be_features_check,
5231
	.ndo_get_phys_port_id   = be_get_phys_port_id,
S
Sathya Perla 已提交
5232 5233 5234 5235 5236 5237
};

static void be_netdev_init(struct net_device *netdev)
{
	struct be_adapter *adapter = netdev_priv(netdev);

5238
	netdev->hw_features |= NETIF_F_SG | NETIF_F_TSO | NETIF_F_TSO6 |
5239
		NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_RXCSUM |
5240
		NETIF_F_HW_VLAN_CTAG_TX;
5241
	if ((be_if_cap_flags(adapter) & BE_IF_FLAGS_RSS))
5242
		netdev->hw_features |= NETIF_F_RXHASH;
5243 5244

	netdev->features |= netdev->hw_features |
5245
		NETIF_F_HW_VLAN_CTAG_RX | NETIF_F_HW_VLAN_CTAG_FILTER;
A
Ajit Khaparde 已提交
5246

5247
	netdev->vlan_features |= NETIF_F_SG | NETIF_F_TSO | NETIF_F_TSO6 |
5248
		NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM;
5249

5250 5251
	netdev->priv_flags |= IFF_UNICAST_FLT;

S
Sathya Perla 已提交
5252 5253
	netdev->flags |= IFF_MULTICAST;

5254
	netif_set_gso_max_size(netdev, BE_MAX_GSO_SIZE - ETH_HLEN);
5255

S
Sathya Perla 已提交
5256
	netdev->netdev_ops = &be_netdev_ops;
S
Sathya Perla 已提交
5257

5258
	netdev->ethtool_ops = &be_ethtool_ops;
S
Sathya Perla 已提交
5259 5260
}

5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273
static void be_cleanup(struct be_adapter *adapter)
{
	struct net_device *netdev = adapter->netdev;

	rtnl_lock();
	netif_device_detach(netdev);
	if (netif_running(netdev))
		be_close(netdev);
	rtnl_unlock();

	be_clear(adapter);
}

5274
static int be_resume(struct be_adapter *adapter)
5275
{
5276
	struct net_device *netdev = adapter->netdev;
5277 5278 5279 5280
	int status;

	status = be_setup(adapter);
	if (status)
5281
		return status;
5282

5283 5284
	rtnl_lock();
	if (netif_running(netdev))
5285
		status = be_open(netdev);
5286 5287 5288 5289
	rtnl_unlock();

	if (status)
		return status;
5290

5291 5292
	netif_device_attach(netdev);

5293 5294 5295
	return 0;
}

5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422
static void be_soft_reset(struct be_adapter *adapter)
{
	u32 val;

	dev_info(&adapter->pdev->dev, "Initiating chip soft reset\n");
	val = ioread32(adapter->pcicfg + SLIPORT_SOFTRESET_OFFSET);
	val |= SLIPORT_SOFTRESET_SR_MASK;
	iowrite32(val, adapter->pcicfg + SLIPORT_SOFTRESET_OFFSET);
}

static bool be_err_is_recoverable(struct be_adapter *adapter)
{
	struct be_error_recovery *err_rec = &adapter->error_recovery;
	unsigned long initial_idle_time =
		msecs_to_jiffies(ERR_RECOVERY_IDLE_TIME);
	unsigned long recovery_interval =
		msecs_to_jiffies(ERR_RECOVERY_INTERVAL);
	u16 ue_err_code;
	u32 val;

	val = be_POST_stage_get(adapter);
	if ((val & POST_STAGE_RECOVERABLE_ERR) != POST_STAGE_RECOVERABLE_ERR)
		return false;
	ue_err_code = val & POST_ERR_RECOVERY_CODE_MASK;
	if (ue_err_code == 0)
		return false;

	dev_err(&adapter->pdev->dev, "Recoverable HW error code: 0x%x\n",
		ue_err_code);

	if (jiffies - err_rec->probe_time <= initial_idle_time) {
		dev_err(&adapter->pdev->dev,
			"Cannot recover within %lu sec from driver load\n",
			jiffies_to_msecs(initial_idle_time) / MSEC_PER_SEC);
		return false;
	}

	if (err_rec->last_recovery_time &&
	    (jiffies - err_rec->last_recovery_time <= recovery_interval)) {
		dev_err(&adapter->pdev->dev,
			"Cannot recover within %lu sec from last recovery\n",
			jiffies_to_msecs(recovery_interval) / MSEC_PER_SEC);
		return false;
	}

	if (ue_err_code == err_rec->last_err_code) {
		dev_err(&adapter->pdev->dev,
			"Cannot recover from a consecutive TPE error\n");
		return false;
	}

	err_rec->last_recovery_time = jiffies;
	err_rec->last_err_code = ue_err_code;
	return true;
}

static int be_tpe_recover(struct be_adapter *adapter)
{
	struct be_error_recovery *err_rec = &adapter->error_recovery;
	int status = -EAGAIN;
	u32 val;

	switch (err_rec->recovery_state) {
	case ERR_RECOVERY_ST_NONE:
		err_rec->recovery_state = ERR_RECOVERY_ST_DETECT;
		err_rec->resched_delay = ERR_RECOVERY_UE_DETECT_DURATION;
		break;

	case ERR_RECOVERY_ST_DETECT:
		val = be_POST_stage_get(adapter);
		if ((val & POST_STAGE_RECOVERABLE_ERR) !=
		    POST_STAGE_RECOVERABLE_ERR) {
			dev_err(&adapter->pdev->dev,
				"Unrecoverable HW error detected: 0x%x\n", val);
			status = -EINVAL;
			err_rec->resched_delay = 0;
			break;
		}

		dev_err(&adapter->pdev->dev, "Recoverable HW error detected\n");

		/* Only PF0 initiates Chip Soft Reset. But PF0 must wait UE2SR
		 * milliseconds before it checks for final error status in
		 * SLIPORT_SEMAPHORE to determine if recovery criteria is met.
		 * If it does, then PF0 initiates a Soft Reset.
		 */
		if (adapter->pf_num == 0) {
			err_rec->recovery_state = ERR_RECOVERY_ST_RESET;
			err_rec->resched_delay = err_rec->ue_to_reset_time -
					ERR_RECOVERY_UE_DETECT_DURATION;
			break;
		}

		err_rec->recovery_state = ERR_RECOVERY_ST_PRE_POLL;
		err_rec->resched_delay = err_rec->ue_to_poll_time -
					ERR_RECOVERY_UE_DETECT_DURATION;
		break;

	case ERR_RECOVERY_ST_RESET:
		if (!be_err_is_recoverable(adapter)) {
			dev_err(&adapter->pdev->dev,
				"Failed to meet recovery criteria\n");
			status = -EIO;
			err_rec->resched_delay = 0;
			break;
		}
		be_soft_reset(adapter);
		err_rec->recovery_state = ERR_RECOVERY_ST_PRE_POLL;
		err_rec->resched_delay = err_rec->ue_to_poll_time -
					err_rec->ue_to_reset_time;
		break;

	case ERR_RECOVERY_ST_PRE_POLL:
		err_rec->recovery_state = ERR_RECOVERY_ST_REINIT;
		err_rec->resched_delay = 0;
		status = 0;			/* done */
		break;

	default:
		status = -EINVAL;
		err_rec->resched_delay = 0;
		break;
	}

	return status;
}

5423 5424 5425 5426
static int be_err_recover(struct be_adapter *adapter)
{
	int status;

5427 5428 5429 5430 5431 5432 5433 5434
	if (!lancer_chip(adapter)) {
		if (!adapter->error_recovery.recovery_supported ||
		    adapter->priv_flags & BE_DISABLE_TPE_RECOVERY)
			return -EIO;
		status = be_tpe_recover(adapter);
		if (status)
			goto err;
	}
5435 5436 5437 5438 5439 5440 5441 5442

	/* Wait for adapter to reach quiescent state before
	 * destroying queues
	 */
	status = be_fw_wait_ready(adapter);
	if (status)
		goto err;

5443 5444
	adapter->flags |= BE_FLAGS_TRY_RECOVERY;

5445 5446
	be_cleanup(adapter);

5447 5448 5449 5450
	status = be_resume(adapter);
	if (status)
		goto err;

5451 5452
	adapter->flags &= ~BE_FLAGS_TRY_RECOVERY;

5453 5454 5455 5456
err:
	return status;
}

5457
static void be_err_detection_task(struct work_struct *work)
5458
{
5459 5460 5461
	struct be_error_recovery *err_rec =
			container_of(work, struct be_error_recovery,
				     err_detection_work.work);
5462
	struct be_adapter *adapter =
5463 5464 5465
			container_of(err_rec, struct be_adapter,
				     error_recovery);
	u32 resched_delay = ERR_RECOVERY_DETECTION_DELAY;
5466 5467
	struct device *dev = &adapter->pdev->dev;
	int recovery_status;
5468 5469

	be_detect_error(adapter);
5470
	if (!be_check_error(adapter, BE_ERROR_HW))
5471 5472
		goto reschedule_task;

5473
	recovery_status = be_err_recover(adapter);
5474
	if (!recovery_status) {
5475 5476
		err_rec->recovery_retries = 0;
		err_rec->recovery_state = ERR_RECOVERY_ST_NONE;
5477 5478
		dev_info(dev, "Adapter recovery successful\n");
		goto reschedule_task;
5479 5480 5481 5482 5483 5484 5485 5486 5487
	} else if (!lancer_chip(adapter) && err_rec->resched_delay) {
		/* BEx/SH recovery state machine */
		if (adapter->pf_num == 0 &&
		    err_rec->recovery_state > ERR_RECOVERY_ST_DETECT)
			dev_err(&adapter->pdev->dev,
				"Adapter recovery in progress\n");
		resched_delay = err_rec->resched_delay;
		goto reschedule_task;
	} else if (lancer_chip(adapter) && be_virtfn(adapter)) {
5488 5489 5490 5491 5492
		/* For VFs, check if PF have allocated resources
		 * every second.
		 */
		dev_err(dev, "Re-trying adapter recovery\n");
		goto reschedule_task;
5493 5494
	} else if (lancer_chip(adapter) && err_rec->recovery_retries++ <
		   ERR_RECOVERY_MAX_RETRY_COUNT) {
5495 5496 5497 5498 5499
		/* In case of another error during recovery, it takes 30 sec
		 * for adapter to come out of error. Retry error recovery after
		 * this time interval.
		 */
		dev_err(&adapter->pdev->dev, "Re-trying adapter recovery\n");
5500
		resched_delay = ERR_RECOVERY_RETRY_DELAY;
5501
		goto reschedule_task;
5502 5503
	} else {
		dev_err(dev, "Adapter recovery failed\n");
5504
		dev_err(dev, "Please reboot server to recover\n");
5505 5506
	}

5507
	return;
5508

5509
reschedule_task:
5510
	be_schedule_err_detection(adapter, resched_delay);
5511 5512 5513 5514 5515 5516 5517 5518 5519
}

static void be_log_sfp_info(struct be_adapter *adapter)
{
	int status;

	status = be_cmd_query_sfp_info(adapter);
	if (!status) {
		dev_err(&adapter->pdev->dev,
5520 5521 5522 5523
			"Port %c: %s Vendor: %s part no: %s",
			adapter->port_name,
			be_misconfig_evt_port_state[adapter->phy_state],
			adapter->phy.vendor_name,
5524 5525
			adapter->phy.vendor_pn);
	}
5526
	adapter->flags &= ~BE_FLAGS_PHY_MISCONFIGURED;
5527 5528 5529 5530 5531 5532 5533 5534 5535
}

static void be_worker(struct work_struct *work)
{
	struct be_adapter *adapter =
		container_of(work, struct be_adapter, work.work);
	struct be_rx_obj *rxo;
	int i;

5536 5537 5538 5539
	if (be_physfn(adapter) &&
	    MODULO(adapter->work_counter, adapter->be_get_temp_freq) == 0)
		be_cmd_get_die_temperature(adapter);

5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565
	/* when interrupts are not yet enabled, just reap any pending
	 * mcc completions
	 */
	if (!netif_running(adapter->netdev)) {
		local_bh_disable();
		be_process_mcc(adapter);
		local_bh_enable();
		goto reschedule;
	}

	if (!adapter->stats_cmd_sent) {
		if (lancer_chip(adapter))
			lancer_cmd_get_pport_stats(adapter,
						   &adapter->stats_cmd);
		else
			be_cmd_get_stats(adapter, &adapter->stats_cmd);
	}

	for_all_rx_queues(adapter, rxo, i) {
		/* Replenish RX-queues starved due to memory
		 * allocation failures.
		 */
		if (rxo->rx_post_starved)
			be_post_rx_frags(rxo, GFP_KERNEL, MAX_RX_POST);
	}

5566 5567 5568
	/* EQ-delay update for Skyhawk is done while notifying EQ */
	if (!skyhawk_chip(adapter))
		be_eqd_update(adapter, false);
5569

5570
	if (adapter->flags & BE_FLAGS_PHY_MISCONFIGURED)
5571 5572 5573 5574
		be_log_sfp_info(adapter);

reschedule:
	adapter->work_counter++;
5575
	queue_delayed_work(be_wq, &adapter->work, msecs_to_jiffies(1000));
5576 5577
}

S
Sathya Perla 已提交
5578 5579
static void be_unmap_pci_bars(struct be_adapter *adapter)
{
5580 5581
	if (adapter->csr)
		pci_iounmap(adapter->pdev, adapter->csr);
5582
	if (adapter->db)
S
Sathya Perla 已提交
5583
		pci_iounmap(adapter->pdev, adapter->db);
5584 5585
	if (adapter->pcicfg && adapter->pcicfg_mapped)
		pci_iounmap(adapter->pdev, adapter->pcicfg);
5586 5587
}

S
Sathya Perla 已提交
5588 5589
static int db_bar(struct be_adapter *adapter)
{
5590
	if (lancer_chip(adapter) || be_virtfn(adapter))
S
Sathya Perla 已提交
5591 5592 5593 5594 5595 5596
		return 0;
	else
		return 4;
}

static int be_roce_map_pci_bars(struct be_adapter *adapter)
5597
{
S
Sathya Perla 已提交
5598
	if (skyhawk_chip(adapter)) {
S
Sathya Perla 已提交
5599 5600 5601 5602 5603 5604
		adapter->roce_db.size = 4096;
		adapter->roce_db.io_addr = pci_resource_start(adapter->pdev,
							      db_bar(adapter));
		adapter->roce_db.total_size = pci_resource_len(adapter->pdev,
							       db_bar(adapter));
	}
5605
	return 0;
S
Sathya Perla 已提交
5606 5607 5608 5609
}

static int be_map_pci_bars(struct be_adapter *adapter)
{
5610
	struct pci_dev *pdev = adapter->pdev;
S
Sathya Perla 已提交
5611
	u8 __iomem *addr;
5612 5613 5614 5615 5616 5617
	u32 sli_intf;

	pci_read_config_dword(adapter->pdev, SLI_INTF_REG_OFFSET, &sli_intf);
	adapter->sli_family = (sli_intf & SLI_INTF_FAMILY_MASK) >>
				SLI_INTF_FAMILY_SHIFT;
	adapter->virtfn = (sli_intf & SLI_INTF_FT_MASK) ? 1 : 0;
5618

5619
	if (BEx_chip(adapter) && be_physfn(adapter)) {
5620
		adapter->csr = pci_iomap(pdev, 2, 0);
K
Kalesh AP 已提交
5621
		if (!adapter->csr)
5622 5623 5624
			return -ENOMEM;
	}

5625
	addr = pci_iomap(pdev, db_bar(adapter), 0);
K
Kalesh AP 已提交
5626
	if (!addr)
S
Sathya Perla 已提交
5627
		goto pci_map_err;
5628
	adapter->db = addr;
S
Sathya Perla 已提交
5629

5630 5631 5632 5633 5634 5635 5636
	if (skyhawk_chip(adapter) || BEx_chip(adapter)) {
		if (be_physfn(adapter)) {
			/* PCICFG is the 2nd BAR in BE2 */
			addr = pci_iomap(pdev, BE2_chip(adapter) ? 1 : 0, 0);
			if (!addr)
				goto pci_map_err;
			adapter->pcicfg = addr;
5637
			adapter->pcicfg_mapped = true;
5638 5639
		} else {
			adapter->pcicfg = adapter->db + SRIOV_VF_PCICFG_OFFSET;
5640
			adapter->pcicfg_mapped = false;
5641 5642 5643
		}
	}

S
Sathya Perla 已提交
5644
	be_roce_map_pci_bars(adapter);
S
Sathya Perla 已提交
5645
	return 0;
S
Sathya Perla 已提交
5646

S
Sathya Perla 已提交
5647
pci_map_err:
5648
	dev_err(&pdev->dev, "Error in mapping PCI BARs\n");
S
Sathya Perla 已提交
5649 5650 5651 5652
	be_unmap_pci_bars(adapter);
	return -ENOMEM;
}

5653
static void be_drv_cleanup(struct be_adapter *adapter)
S
Sathya Perla 已提交
5654
{
5655
	struct be_dma_mem *mem = &adapter->mbox_mem_alloced;
5656
	struct device *dev = &adapter->pdev->dev;
S
Sathya Perla 已提交
5657 5658

	if (mem->va)
5659
		dma_free_coherent(dev, mem->size, mem->va, mem->dma);
5660

5661
	mem = &adapter->rx_filter;
5662
	if (mem->va)
5663 5664 5665 5666 5667
		dma_free_coherent(dev, mem->size, mem->va, mem->dma);

	mem = &adapter->stats_cmd;
	if (mem->va)
		dma_free_coherent(dev, mem->size, mem->va, mem->dma);
S
Sathya Perla 已提交
5668 5669
}

5670 5671
/* Allocate and initialize various fields in be_adapter struct */
static int be_drv_init(struct be_adapter *adapter)
S
Sathya Perla 已提交
5672
{
5673 5674
	struct be_dma_mem *mbox_mem_alloc = &adapter->mbox_mem_alloced;
	struct be_dma_mem *mbox_mem_align = &adapter->mbox_mem;
5675
	struct be_dma_mem *rx_filter = &adapter->rx_filter;
5676 5677 5678
	struct be_dma_mem *stats_cmd = &adapter->stats_cmd;
	struct device *dev = &adapter->pdev->dev;
	int status = 0;
S
Sathya Perla 已提交
5679 5680

	mbox_mem_alloc->size = sizeof(struct be_mcc_mailbox) + 16;
5681 5682 5683
	mbox_mem_alloc->va = dma_zalloc_coherent(dev, mbox_mem_alloc->size,
						 &mbox_mem_alloc->dma,
						 GFP_KERNEL);
5684 5685 5686
	if (!mbox_mem_alloc->va)
		return -ENOMEM;

S
Sathya Perla 已提交
5687 5688 5689
	mbox_mem_align->size = sizeof(struct be_mcc_mailbox);
	mbox_mem_align->va = PTR_ALIGN(mbox_mem_alloc->va, 16);
	mbox_mem_align->dma = PTR_ALIGN(mbox_mem_alloc->dma, 16);
5690

5691
	rx_filter->size = sizeof(struct be_cmd_req_rx_filter);
5692 5693
	rx_filter->va = dma_zalloc_coherent(dev, rx_filter->size,
					    &rx_filter->dma, GFP_KERNEL);
K
Kalesh AP 已提交
5694
	if (!rx_filter->va) {
5695 5696 5697
		status = -ENOMEM;
		goto free_mbox;
	}
5698

5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713
	if (lancer_chip(adapter))
		stats_cmd->size = sizeof(struct lancer_cmd_req_pport_stats);
	else if (BE2_chip(adapter))
		stats_cmd->size = sizeof(struct be_cmd_req_get_stats_v0);
	else if (BE3_chip(adapter))
		stats_cmd->size = sizeof(struct be_cmd_req_get_stats_v1);
	else
		stats_cmd->size = sizeof(struct be_cmd_req_get_stats_v2);
	stats_cmd->va = dma_zalloc_coherent(dev, stats_cmd->size,
					    &stats_cmd->dma, GFP_KERNEL);
	if (!stats_cmd->va) {
		status = -ENOMEM;
		goto free_rx_filter;
	}

5714
	mutex_init(&adapter->mbox_lock);
5715 5716
	mutex_init(&adapter->mcc_lock);
	mutex_init(&adapter->rx_filter_lock);
5717
	spin_lock_init(&adapter->mcc_cq_lock);
5718
	init_completion(&adapter->et_cmd_compl);
5719

5720
	pci_save_state(adapter->pdev);
S
Sathya Perla 已提交
5721

5722
	INIT_DELAYED_WORK(&adapter->work, be_worker);
5723 5724 5725 5726

	adapter->error_recovery.recovery_state = ERR_RECOVERY_ST_NONE;
	adapter->error_recovery.resched_delay = 0;
	INIT_DELAYED_WORK(&adapter->error_recovery.err_detection_work,
5727
			  be_err_detection_task);
S
Sathya Perla 已提交
5728

5729 5730
	adapter->rx_fc = true;
	adapter->tx_fc = true;
S
Sathya Perla 已提交
5731

5732 5733
	/* Must be a power of 2 or else MODULO will BUG_ON */
	adapter->be_get_temp_freq = 64;
5734

S
Sathya Perla 已提交
5735
	return 0;
5736 5737 5738 5739 5740 5741 5742

free_rx_filter:
	dma_free_coherent(dev, rx_filter->size, rx_filter->va, rx_filter->dma);
free_mbox:
	dma_free_coherent(dev, mbox_mem_alloc->size, mbox_mem_alloc->va,
			  mbox_mem_alloc->dma);
	return status;
S
Sathya Perla 已提交
5743 5744
}

B
Bill Pemberton 已提交
5745
static void be_remove(struct pci_dev *pdev)
S
Sathya Perla 已提交
5746 5747
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);
5748

S
Sathya Perla 已提交
5749 5750 5751
	if (!adapter)
		return;

5752
	be_roce_dev_remove(adapter);
5753
	be_intr_set(adapter, false);
5754

5755
	be_cancel_err_detection(adapter);
5756

S
Sathya Perla 已提交
5757 5758
	unregister_netdev(adapter->netdev);

5759 5760
	be_clear(adapter);

5761 5762 5763
	if (!pci_vfs_assigned(adapter->pdev))
		be_cmd_reset_function(adapter);

5764 5765 5766
	/* tell fw we're done with firing cmds */
	be_cmd_fw_clean(adapter);

5767 5768
	be_unmap_pci_bars(adapter);
	be_drv_cleanup(adapter);
S
Sathya Perla 已提交
5769

S
Sathya Perla 已提交
5770 5771
	pci_disable_pcie_error_reporting(pdev);

S
Sathya Perla 已提交
5772 5773 5774 5775 5776 5777
	pci_release_regions(pdev);
	pci_disable_device(pdev);

	free_netdev(adapter->netdev);
}

5778 5779 5780
static ssize_t be_hwmon_show_temp(struct device *dev,
				  struct device_attribute *dev_attr,
				  char *buf)
5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801
{
	struct be_adapter *adapter = dev_get_drvdata(dev);

	/* Unit: millidegree Celsius */
	if (adapter->hwmon_info.be_on_die_temp == BE_INVALID_DIE_TEMP)
		return -EIO;
	else
		return sprintf(buf, "%u\n",
			       adapter->hwmon_info.be_on_die_temp * 1000);
}

static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO,
			  be_hwmon_show_temp, NULL, 1);

static struct attribute *be_hwmon_attrs[] = {
	&sensor_dev_attr_temp1_input.dev_attr.attr,
	NULL
};

ATTRIBUTE_GROUPS(be_hwmon);

S
Sathya Perla 已提交
5802 5803
static char *mc_name(struct be_adapter *adapter)
{
5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829
	char *str = "";	/* default */

	switch (adapter->mc_type) {
	case UMC:
		str = "UMC";
		break;
	case FLEX10:
		str = "FLEX10";
		break;
	case vNIC1:
		str = "vNIC-1";
		break;
	case nPAR:
		str = "nPAR";
		break;
	case UFP:
		str = "UFP";
		break;
	case vNIC2:
		str = "vNIC-2";
		break;
	default:
		str = "";
	}

	return str;
S
Sathya Perla 已提交
5830 5831 5832 5833 5834 5835 5836
}

static inline char *func_name(struct be_adapter *adapter)
{
	return be_physfn(adapter) ? "PF" : "VF";
}

5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856
static inline char *nic_name(struct pci_dev *pdev)
{
	switch (pdev->device) {
	case OC_DEVICE_ID1:
		return OC_NAME;
	case OC_DEVICE_ID2:
		return OC_NAME_BE;
	case OC_DEVICE_ID3:
	case OC_DEVICE_ID4:
		return OC_NAME_LANCER;
	case BE_DEVICE_ID2:
		return BE3_NAME;
	case OC_DEVICE_ID5:
	case OC_DEVICE_ID6:
		return OC_NAME_SH;
	default:
		return BE_NAME;
	}
}

5857
static int be_probe(struct pci_dev *pdev, const struct pci_device_id *pdev_id)
S
Sathya Perla 已提交
5858 5859 5860
{
	struct be_adapter *adapter;
	struct net_device *netdev;
5861
	int status = 0;
S
Sathya Perla 已提交
5862

S
Sathya Perla 已提交
5863 5864
	dev_info(&pdev->dev, "%s version is %s\n", DRV_NAME, DRV_VER);

S
Sathya Perla 已提交
5865 5866 5867 5868 5869 5870 5871 5872 5873
	status = pci_enable_device(pdev);
	if (status)
		goto do_none;

	status = pci_request_regions(pdev, DRV_NAME);
	if (status)
		goto disable_dev;
	pci_set_master(pdev);

5874
	netdev = alloc_etherdev_mqs(sizeof(*adapter), MAX_TX_QS, MAX_RX_QS);
K
Kalesh AP 已提交
5875
	if (!netdev) {
S
Sathya Perla 已提交
5876 5877 5878 5879 5880 5881 5882
		status = -ENOMEM;
		goto rel_reg;
	}
	adapter = netdev_priv(netdev);
	adapter->pdev = pdev;
	pci_set_drvdata(pdev, adapter);
	adapter->netdev = netdev;
5883
	SET_NETDEV_DEV(netdev, &pdev->dev);
S
Sathya Perla 已提交
5884

5885
	status = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
S
Sathya Perla 已提交
5886 5887 5888
	if (!status) {
		netdev->features |= NETIF_F_HIGHDMA;
	} else {
5889
		status = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
S
Sathya Perla 已提交
5890 5891 5892 5893 5894 5895
		if (status) {
			dev_err(&pdev->dev, "Could not set PCI DMA Mask\n");
			goto free_netdev;
		}
	}

5896 5897 5898
	status = pci_enable_pcie_error_reporting(pdev);
	if (!status)
		dev_info(&pdev->dev, "PCIe error reporting enabled\n");
S
Sathya Perla 已提交
5899

5900
	status = be_map_pci_bars(adapter);
S
Sathya Perla 已提交
5901
	if (status)
5902
		goto free_netdev;
S
Sathya Perla 已提交
5903

5904 5905 5906 5907
	status = be_drv_init(adapter);
	if (status)
		goto unmap_bars;

5908 5909
	status = be_setup(adapter);
	if (status)
5910
		goto drv_cleanup;
5911

5912
	be_netdev_init(netdev);
S
Sathya Perla 已提交
5913 5914
	status = register_netdev(netdev);
	if (status != 0)
5915
		goto unsetup;
S
Sathya Perla 已提交
5916

5917 5918
	be_roce_dev_add(adapter);

5919
	be_schedule_err_detection(adapter, ERR_DETECTION_DELAY);
5920
	adapter->error_recovery.probe_time = jiffies;
5921

5922
	/* On Die temperature not supported for VF. */
5923
	if (be_physfn(adapter) && IS_ENABLED(CONFIG_BE2NET_HWMON)) {
5924 5925 5926 5927 5928 5929 5930 5931
		adapter->hwmon_info.hwmon_dev =
			devm_hwmon_device_register_with_groups(&pdev->dev,
							       DRV_NAME,
							       adapter,
							       be_hwmon_groups);
		adapter->hwmon_info.be_on_die_temp = BE_INVALID_DIE_TEMP;
	}

S
Sathya Perla 已提交
5932
	dev_info(&pdev->dev, "%s: %s %s port %c\n", nic_name(pdev),
5933
		 func_name(adapter), mc_name(adapter), adapter->port_name);
5934

S
Sathya Perla 已提交
5935 5936
	return 0;

5937 5938
unsetup:
	be_clear(adapter);
5939 5940 5941 5942
drv_cleanup:
	be_drv_cleanup(adapter);
unmap_bars:
	be_unmap_pci_bars(adapter);
5943
free_netdev:
5944
	free_netdev(netdev);
S
Sathya Perla 已提交
5945 5946 5947 5948 5949
rel_reg:
	pci_release_regions(pdev);
disable_dev:
	pci_disable_device(pdev);
do_none:
5950
	dev_err(&pdev->dev, "%s initialization failed\n", nic_name(pdev));
S
Sathya Perla 已提交
5951 5952 5953 5954 5955 5956 5957
	return status;
}

static int be_suspend(struct pci_dev *pdev, pm_message_t state)
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);

5958
	be_intr_set(adapter, false);
5959
	be_cancel_err_detection(adapter);
5960

5961
	be_cleanup(adapter);
S
Sathya Perla 已提交
5962 5963 5964 5965 5966 5967 5968

	pci_save_state(pdev);
	pci_disable_device(pdev);
	pci_set_power_state(pdev, pci_choose_state(pdev, state));
	return 0;
}

5969
static int be_pci_resume(struct pci_dev *pdev)
S
Sathya Perla 已提交
5970 5971
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);
5972
	int status = 0;
S
Sathya Perla 已提交
5973 5974 5975 5976 5977 5978 5979

	status = pci_enable_device(pdev);
	if (status)
		return status;

	pci_restore_state(pdev);

5980
	status = be_resume(adapter);
5981 5982 5983
	if (status)
		return status;

5984
	be_schedule_err_detection(adapter, ERR_DETECTION_DELAY);
5985

S
Sathya Perla 已提交
5986 5987 5988
	return 0;
}

5989 5990 5991 5992 5993 5994 5995
/*
 * An FLR will stop BE from DMAing any data.
 */
static void be_shutdown(struct pci_dev *pdev)
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);

5996 5997
	if (!adapter)
		return;
5998

5999
	be_roce_dev_shutdown(adapter);
6000
	cancel_delayed_work_sync(&adapter->work);
6001
	be_cancel_err_detection(adapter);
6002

6003
	netif_device_detach(adapter->netdev);
6004

6005 6006
	be_cmd_reset_function(adapter);

6007 6008 6009
	pci_disable_device(pdev);
}

6010
static pci_ers_result_t be_eeh_err_detected(struct pci_dev *pdev,
6011
					    pci_channel_state_t state)
6012 6013 6014 6015 6016
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);

	dev_err(&adapter->pdev->dev, "EEH error detected\n");

6017 6018
	be_roce_dev_remove(adapter);

6019 6020
	if (!be_check_error(adapter, BE_ERROR_EEH)) {
		be_set_error(adapter, BE_ERROR_EEH);
6021

6022
		be_cancel_err_detection(adapter);
6023

6024
		be_cleanup(adapter);
6025 6026 6027 6028 6029 6030 6031
	}

	if (state == pci_channel_io_perm_failure)
		return PCI_ERS_RESULT_DISCONNECT;

	pci_disable_device(pdev);

6032 6033
	/* The error could cause the FW to trigger a flash debug dump.
	 * Resetting the card while flash dump is in progress
6034 6035 6036
	 * can cause it not to recover; wait for it to finish.
	 * Wait only for first function as it is needed only once per
	 * adapter.
6037
	 */
6038 6039 6040
	if (pdev->devfn == 0)
		ssleep(30);

6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058
	return PCI_ERS_RESULT_NEED_RESET;
}

static pci_ers_result_t be_eeh_reset(struct pci_dev *pdev)
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);
	int status;

	dev_info(&adapter->pdev->dev, "EEH reset\n");

	status = pci_enable_device(pdev);
	if (status)
		return PCI_ERS_RESULT_DISCONNECT;

	pci_set_master(pdev);
	pci_restore_state(pdev);

	/* Check if card is ok and fw is ready */
6059 6060
	dev_info(&adapter->pdev->dev,
		 "Waiting for FW to be ready after EEH reset\n");
6061
	status = be_fw_wait_ready(adapter);
6062 6063 6064
	if (status)
		return PCI_ERS_RESULT_DISCONNECT;

S
Sathya Perla 已提交
6065
	pci_cleanup_aer_uncorrect_error_status(pdev);
6066
	be_clear_error(adapter, BE_CLEAR_ALL);
6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078
	return PCI_ERS_RESULT_RECOVERED;
}

static void be_eeh_resume(struct pci_dev *pdev)
{
	int status = 0;
	struct be_adapter *adapter = pci_get_drvdata(pdev);

	dev_info(&adapter->pdev->dev, "EEH resume\n");

	pci_save_state(pdev);

6079
	status = be_resume(adapter);
6080 6081 6082
	if (status)
		goto err;

6083 6084
	be_roce_dev_add(adapter);

6085
	be_schedule_err_detection(adapter, ERR_DETECTION_DELAY);
6086 6087 6088 6089 6090
	return;
err:
	dev_err(&adapter->pdev->dev, "EEH resume failed\n");
}

6091 6092 6093
static int be_pci_sriov_configure(struct pci_dev *pdev, int num_vfs)
{
	struct be_adapter *adapter = pci_get_drvdata(pdev);
6094
	struct be_resources vft_res = {0};
6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116
	int status;

	if (!num_vfs)
		be_vf_clear(adapter);

	adapter->num_vfs = num_vfs;

	if (adapter->num_vfs == 0 && pci_vfs_assigned(pdev)) {
		dev_warn(&pdev->dev,
			 "Cannot disable VFs while they are assigned\n");
		return -EBUSY;
	}

	/* When the HW is in SRIOV capable configuration, the PF-pool resources
	 * are equally distributed across the max-number of VFs. The user may
	 * request only a subset of the max-vfs to be enabled.
	 * Based on num_vfs, redistribute the resources across num_vfs so that
	 * each VF will have access to more number of resources.
	 * This facility is not available in BE3 FW.
	 * Also, this is done by FW in Lancer chip.
	 */
	if (skyhawk_chip(adapter) && !pci_num_vf(pdev)) {
6117 6118
		be_calculate_vf_res(adapter, adapter->num_vfs,
				    &vft_res);
6119
		status = be_cmd_set_sriov_config(adapter, adapter->pool_res,
6120
						 adapter->num_vfs, &vft_res);
6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 6141 6142 6143 6144 6145
		if (status)
			dev_err(&pdev->dev,
				"Failed to optimize SR-IOV resources\n");
	}

	status = be_get_resources(adapter);
	if (status)
		return be_cmd_status(status);

	/* Updating real_num_tx/rx_queues() requires rtnl_lock() */
	rtnl_lock();
	status = be_update_queues(adapter);
	rtnl_unlock();
	if (status)
		return be_cmd_status(status);

	if (adapter->num_vfs)
		status = be_vf_setup(adapter);

	if (!status)
		return adapter->num_vfs;

	return 0;
}

6146
static const struct pci_error_handlers be_eeh_handlers = {
6147 6148 6149 6150 6151
	.error_detected = be_eeh_err_detected,
	.slot_reset = be_eeh_reset,
	.resume = be_eeh_resume,
};

S
Sathya Perla 已提交
6152 6153 6154 6155 6156 6157
static struct pci_driver be_driver = {
	.name = DRV_NAME,
	.id_table = be_dev_ids,
	.probe = be_probe,
	.remove = be_remove,
	.suspend = be_suspend,
6158
	.resume = be_pci_resume,
6159
	.shutdown = be_shutdown,
6160
	.sriov_configure = be_pci_sriov_configure,
6161
	.err_handler = &be_eeh_handlers
S
Sathya Perla 已提交
6162 6163 6164 6165
};

static int __init be_init_module(void)
{
6166 6167
	int status;

6168 6169
	if (rx_frag_size != 8192 && rx_frag_size != 4096 &&
	    rx_frag_size != 2048) {
S
Sathya Perla 已提交
6170 6171 6172 6173 6174 6175
		printk(KERN_WARNING DRV_NAME
			" : Module param rx_frag_size must be 2048/4096/8192."
			" Using 2048\n");
		rx_frag_size = 2048;
	}

6176 6177 6178 6179 6180
	if (num_vfs > 0) {
		pr_info(DRV_NAME " : Module param num_vfs is obsolete.");
		pr_info(DRV_NAME " : Use sysfs method to enable VFs\n");
	}

6181 6182 6183 6184 6185 6186
	be_wq = create_singlethread_workqueue("be_wq");
	if (!be_wq) {
		pr_warn(DRV_NAME "workqueue creation failed\n");
		return -1;
	}

6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197
	be_err_recovery_workq =
		create_singlethread_workqueue("be_err_recover");
	if (!be_err_recovery_workq)
		pr_warn(DRV_NAME "Could not create error recovery workqueue\n");

	status = pci_register_driver(&be_driver);
	if (status) {
		destroy_workqueue(be_wq);
		be_destroy_err_recovery_workq();
	}
	return status;
S
Sathya Perla 已提交
6198 6199 6200 6201 6202 6203
}
module_init(be_init_module);

static void __exit be_exit_module(void)
{
	pci_unregister_driver(&be_driver);
6204

6205 6206
	be_destroy_err_recovery_workq();

6207 6208
	if (be_wq)
		destroy_workqueue(be_wq);
S
Sathya Perla 已提交
6209 6210
}
module_exit(be_exit_module);