enic_main.c 75.6 KB
Newer Older
1
/*
V
Vasanthy Kolluri 已提交
2
 * Copyright 2008-2010 Cisco Systems, Inc.  All rights reserved.
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
 * Copyright 2007 Nuova Systems, Inc.  All rights reserved.
 *
 * This program is free software; you may redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; version 2 of the License.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 *
 */

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/init.h>
26
#include <linux/interrupt.h>
27 28 29 30
#include <linux/workqueue.h>
#include <linux/pci.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
31
#include <linux/if.h>
32 33 34 35 36 37
#include <linux/if_ether.h>
#include <linux/if_vlan.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/tcp.h>
V
Vasanthy Kolluri 已提交
38
#include <linux/rtnetlink.h>
39
#include <linux/prefetch.h>
40
#include <net/ip6_checksum.h>
41
#include <linux/ktime.h>
42
#include <linux/numa.h>
43 44 45
#ifdef CONFIG_RFS_ACCEL
#include <linux/cpu_rmap.h>
#endif
46
#include <linux/crash_dump.h>
47
#include <net/busy_poll.h>
48
#include <net/vxlan.h>
49 50 51 52 53

#include "cq_enet_desc.h"
#include "vnic_dev.h"
#include "vnic_intr.h"
#include "vnic_stats.h"
54
#include "vnic_vic.h"
55 56
#include "enic_res.h"
#include "enic.h"
57
#include "enic_dev.h"
58
#include "enic_pp.h"
59
#include "enic_clsf.h"
60 61

#define ENIC_NOTIFY_TIMER_PERIOD	(2 * HZ)
62 63 64 65 66
#define WQ_ENET_MAX_DESC_LEN		(1 << WQ_ENET_LEN_BITS)
#define MAX_TSO				(1 << 16)
#define ENIC_DESC_MAX_SPLITS		(MAX_TSO / WQ_ENET_MAX_DESC_LEN + 1)

#define PCI_DEVICE_ID_CISCO_VIC_ENET         0x0043  /* ethernet vnic */
67
#define PCI_DEVICE_ID_CISCO_VIC_ENET_DYN     0x0044  /* enet dynamic vnic */
68
#define PCI_DEVICE_ID_CISCO_VIC_ENET_VF      0x0071  /* enet SRIOV VF */
69

70 71
#define RX_COPYBREAK_DEFAULT		256

72
/* Supported devices */
73
static const struct pci_device_id enic_id_table[] = {
74
	{ PCI_VDEVICE(CISCO, PCI_DEVICE_ID_CISCO_VIC_ENET) },
75
	{ PCI_VDEVICE(CISCO, PCI_DEVICE_ID_CISCO_VIC_ENET_DYN) },
76
	{ PCI_VDEVICE(CISCO, PCI_DEVICE_ID_CISCO_VIC_ENET_VF) },
77 78 79 80 81 82 83 84
	{ 0, }	/* end of table */
};

MODULE_DESCRIPTION(DRV_DESCRIPTION);
MODULE_AUTHOR("Scott Feldman <scofeldm@cisco.com>");
MODULE_LICENSE("GPL");
MODULE_DEVICE_TABLE(pci, enic_id_table);

85 86 87 88 89 90
#define ENIC_LARGE_PKT_THRESHOLD		1000
#define ENIC_MAX_COALESCE_TIMERS		10
/*  Interrupt moderation table, which will be used to decide the
 *  coalescing timer values
 *  {rx_rate in Mbps, mapping percentage of the range}
 */
91
static struct enic_intr_mod_table mod_table[ENIC_MAX_COALESCE_TIMERS + 1] = {
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107
	{4000,  0},
	{4400, 10},
	{5060, 20},
	{5230, 30},
	{5540, 40},
	{5820, 50},
	{6120, 60},
	{6435, 70},
	{6745, 80},
	{7000, 90},
	{0xFFFFFFFF, 100}
};

/* This table helps the driver to pick different ranges for rx coalescing
 * timer depending on the link speed.
 */
108
static struct enic_intr_mod_range mod_range[ENIC_MAX_LINK_SPEEDS] = {
109 110 111 112 113
	{0,  0}, /* 0  - 4  Gbps */
	{0,  3}, /* 4  - 10 Gbps */
	{3,  6}, /* 10 - 40 Gbps */
};

114 115 116 117 118 119 120
static void enic_init_affinity_hint(struct enic *enic)
{
	int numa_node = dev_to_node(&enic->pdev->dev);
	int i;

	for (i = 0; i < enic->intr_count; i++) {
		if (enic_is_err_intr(enic, i) || enic_is_notify_intr(enic, i) ||
121
		    (cpumask_available(enic->msix[i].affinity_mask) &&
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149
		     !cpumask_empty(enic->msix[i].affinity_mask)))
			continue;
		if (zalloc_cpumask_var(&enic->msix[i].affinity_mask,
				       GFP_KERNEL))
			cpumask_set_cpu(cpumask_local_spread(i, numa_node),
					enic->msix[i].affinity_mask);
	}
}

static void enic_free_affinity_hint(struct enic *enic)
{
	int i;

	for (i = 0; i < enic->intr_count; i++) {
		if (enic_is_err_intr(enic, i) || enic_is_notify_intr(enic, i))
			continue;
		free_cpumask_var(enic->msix[i].affinity_mask);
	}
}

static void enic_set_affinity_hint(struct enic *enic)
{
	int i;
	int err;

	for (i = 0; i < enic->intr_count; i++) {
		if (enic_is_err_intr(enic, i)		||
		    enic_is_notify_intr(enic, i)	||
150
		    !cpumask_available(enic->msix[i].affinity_mask) ||
151 152 153 154 155 156 157 158 159 160 161 162
		    cpumask_empty(enic->msix[i].affinity_mask))
			continue;
		err = irq_set_affinity_hint(enic->msix_entry[i].vector,
					    enic->msix[i].affinity_mask);
		if (err)
			netdev_warn(enic->netdev, "irq_set_affinity_hint failed, err %d\n",
				    err);
	}

	for (i = 0; i < enic->wq_count; i++) {
		int wq_intr = enic_msix_wq_intr(enic, i);

163
		if (cpumask_available(enic->msix[wq_intr].affinity_mask) &&
164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
		    !cpumask_empty(enic->msix[wq_intr].affinity_mask))
			netif_set_xps_queue(enic->netdev,
					    enic->msix[wq_intr].affinity_mask,
					    i);
	}
}

static void enic_unset_affinity_hint(struct enic *enic)
{
	int i;

	for (i = 0; i < enic->intr_count; i++)
		irq_set_affinity_hint(enic->msix_entry[i].vector, NULL);
}

179 180 181
static int enic_udp_tunnel_set_port(struct net_device *netdev,
				    unsigned int table, unsigned int entry,
				    struct udp_tunnel_info *ti)
182 183 184 185 186 187 188 189
{
	struct enic *enic = netdev_priv(netdev);
	int err;

	spin_lock_bh(&enic->devcmd_lock);

	err = vnic_dev_overlay_offload_cfg(enic->vdev,
					   OVERLAY_CFG_VXLAN_PORT_UPDATE,
190
					   ntohs(ti->port));
191 192 193 194 195 196 197 198
	if (err)
		goto error;

	err = vnic_dev_overlay_offload_ctrl(enic->vdev, OVERLAY_FEATURE_VXLAN,
					    enic->vxlan.patch_level);
	if (err)
		goto error;

199
	enic->vxlan.vxlan_udp_port_number = ntohs(ti->port);
200 201
error:
	spin_unlock_bh(&enic->devcmd_lock);
202 203

	return err;
204 205
}

206 207 208
static int enic_udp_tunnel_unset_port(struct net_device *netdev,
				      unsigned int table, unsigned int entry,
				      struct udp_tunnel_info *ti)
209 210 211 212 213 214 215 216
{
	struct enic *enic = netdev_priv(netdev);
	int err;

	spin_lock_bh(&enic->devcmd_lock);

	err = vnic_dev_overlay_offload_ctrl(enic->vdev, OVERLAY_FEATURE_VXLAN,
					    OVERLAY_OFFLOAD_DISABLE);
217
	if (err)
218 219 220 221 222 223
		goto unlock;

	enic->vxlan.vxlan_udp_port_number = 0;

unlock:
	spin_unlock_bh(&enic->devcmd_lock);
224 225

	return err;
226 227
}

228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
static const struct udp_tunnel_nic_info enic_udp_tunnels = {
	.set_port	= enic_udp_tunnel_set_port,
	.unset_port	= enic_udp_tunnel_unset_port,
	.tables		= {
		{ .n_entries = 1, .tunnel_types = UDP_TUNNEL_TYPE_VXLAN, },
	},
}, enic_udp_tunnels_v4 = {
	.set_port	= enic_udp_tunnel_set_port,
	.unset_port	= enic_udp_tunnel_unset_port,
	.flags		= UDP_TUNNEL_NIC_INFO_IPV4_ONLY,
	.tables		= {
		{ .n_entries = 1, .tunnel_types = UDP_TUNNEL_TYPE_VXLAN, },
	},
};

243 244 245 246 247 248 249 250
static netdev_features_t enic_features_check(struct sk_buff *skb,
					     struct net_device *dev,
					     netdev_features_t features)
{
	const struct ethhdr *eth = (struct ethhdr *)skb_inner_mac_header(skb);
	struct enic *enic = netdev_priv(dev);
	struct udphdr *udph;
	u16 port = 0;
251
	u8 proto;
252 253 254 255 256 257

	if (!skb->encapsulation)
		return features;

	features = vxlan_features_check(skb, features);

258 259 260 261 262 263 264 265 266 267
	switch (vlan_get_protocol(skb)) {
	case htons(ETH_P_IPV6):
		if (!(enic->vxlan.flags & ENIC_VXLAN_OUTER_IPV6))
			goto out;
		proto = ipv6_hdr(skb)->nexthdr;
		break;
	case htons(ETH_P_IP):
		proto = ip_hdr(skb)->protocol;
		break;
	default:
268
		goto out;
269
	}
270

271 272 273 274 275 276 277 278
	switch (eth->h_proto) {
	case ntohs(ETH_P_IPV6):
		if (!(enic->vxlan.flags & ENIC_VXLAN_INNER_IPV6))
			goto out;
		/* Fall through */
	case ntohs(ETH_P_IP):
		break;
	default:
279
		goto out;
280
	}
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299


	if (proto == IPPROTO_UDP) {
		udph = udp_hdr(skb);
		port = be16_to_cpu(udph->dest);
	}

	/* HW supports offload of only one UDP port. Remove CSUM and GSO MASK
	 * for other UDP port tunnels
	 */
	if (port  != enic->vxlan.vxlan_udp_port_number)
		goto out;

	return features;

out:
	return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
}

300
int enic_is_dynamic(struct enic *enic)
301 302 303 304
{
	return enic->pdev->device == PCI_DEVICE_ID_CISCO_VIC_ENET_DYN;
}

R
Roopa Prabhu 已提交
305 306 307 308 309
int enic_sriov_enabled(struct enic *enic)
{
	return (enic->priv_flags & ENIC_SRIOV_ENABLED) ? 1 : 0;
}

310 311 312 313 314
static int enic_is_sriov_vf(struct enic *enic)
{
	return enic->pdev->device == PCI_DEVICE_ID_CISCO_VIC_ENET_VF;
}

315 316 317 318 319 320 321 322 323
int enic_is_valid_vf(struct enic *enic, int vf)
{
#ifdef CONFIG_PCI_IOV
	return vf >= 0 && vf < enic->num_vfs;
#else
	return 0;
#endif
}

324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
static void enic_free_wq_buf(struct vnic_wq *wq, struct vnic_wq_buf *buf)
{
	struct enic *enic = vnic_dev_priv(wq->vdev);

	if (buf->sop)
		pci_unmap_single(enic->pdev, buf->dma_addr,
			buf->len, PCI_DMA_TODEVICE);
	else
		pci_unmap_page(enic->pdev, buf->dma_addr,
			buf->len, PCI_DMA_TODEVICE);

	if (buf->os_buf)
		dev_kfree_skb_any(buf->os_buf);
}

static void enic_wq_free_buf(struct vnic_wq *wq,
	struct cq_desc *cq_desc, struct vnic_wq_buf *buf, void *opaque)
{
	enic_free_wq_buf(wq, buf);
}

static int enic_wq_service(struct vnic_dev *vdev, struct cq_desc *cq_desc,
	u8 type, u16 q_number, u16 completed_index, void *opaque)
{
	struct enic *enic = vnic_dev_priv(vdev);

	spin_lock(&enic->wq_lock[q_number]);

	vnic_wq_service(&enic->wq[q_number], cq_desc,
		completed_index, enic_wq_free_buf,
		opaque);

356
	if (netif_tx_queue_stopped(netdev_get_tx_queue(enic->netdev, q_number)) &&
357 358
	    vnic_wq_desc_avail(&enic->wq[q_number]) >=
	    (MAX_SKB_FRAGS + ENIC_DESC_MAX_SPLITS))
359
		netif_wake_subqueue(enic->netdev, q_number);
360 361 362 363 364 365

	spin_unlock(&enic->wq_lock[q_number]);

	return 0;
}

366
static bool enic_log_q_error(struct enic *enic)
367 368 369
{
	unsigned int i;
	u32 error_status;
370
	bool err = false;
371 372 373

	for (i = 0; i < enic->wq_count; i++) {
		error_status = vnic_wq_error_status(&enic->wq[i]);
374
		err |= error_status;
375
		if (error_status)
376 377
			netdev_err(enic->netdev, "WQ[%d] error_status %d\n",
				i, error_status);
378 379 380 381
	}

	for (i = 0; i < enic->rq_count; i++) {
		error_status = vnic_rq_error_status(&enic->rq[i]);
382
		err |= error_status;
383
		if (error_status)
384 385
			netdev_err(enic->netdev, "RQ[%d] error_status %d\n",
				i, error_status);
386
	}
387 388

	return err;
389 390
}

391
static void enic_msglvl_check(struct enic *enic)
392
{
393
	u32 msg_enable = vnic_dev_msg_lvl(enic->vdev);
394

395
	if (msg_enable != enic->msg_enable) {
396 397
		netdev_info(enic->netdev, "msg lvl changed from 0x%x to 0x%x\n",
			enic->msg_enable, msg_enable);
398
		enic->msg_enable = msg_enable;
399 400 401 402 403 404
	}
}

static void enic_mtu_check(struct enic *enic)
{
	u32 mtu = vnic_dev_mtu(enic->vdev);
405
	struct net_device *netdev = enic->netdev;
406

407
	if (mtu && mtu != enic->port_mtu) {
408
		enic->port_mtu = mtu;
409
		if (enic_is_dynamic(enic) || enic_is_sriov_vf(enic)) {
410 411 412 413 414 415 416 417 418 419 420
			mtu = max_t(int, ENIC_MIN_MTU,
				min_t(int, ENIC_MAX_MTU, mtu));
			if (mtu != netdev->mtu)
				schedule_work(&enic->change_mtu_work);
		} else {
			if (mtu < netdev->mtu)
				netdev_warn(netdev,
					"interface MTU (%d) set higher "
					"than switch port MTU (%d)\n",
					netdev->mtu, mtu);
		}
421 422 423
	}
}

424
static void enic_link_check(struct enic *enic)
425
{
426 427
	int link_status = vnic_dev_link_status(enic->vdev);
	int carrier_ok = netif_carrier_ok(enic->netdev);
428

429
	if (link_status && !carrier_ok) {
430
		netdev_info(enic->netdev, "Link UP\n");
431 432
		netif_carrier_on(enic->netdev);
	} else if (!link_status && carrier_ok) {
433
		netdev_info(enic->netdev, "Link DOWN\n");
434
		netif_carrier_off(enic->netdev);
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
	}
}

static void enic_notify_check(struct enic *enic)
{
	enic_msglvl_check(enic);
	enic_mtu_check(enic);
	enic_link_check(enic);
}

#define ENIC_TEST_INTR(pba, i) (pba & (1 << i))

static irqreturn_t enic_isr_legacy(int irq, void *data)
{
	struct net_device *netdev = data;
	struct enic *enic = netdev_priv(netdev);
451 452 453
	unsigned int io_intr = enic_legacy_io_intr();
	unsigned int err_intr = enic_legacy_err_intr();
	unsigned int notify_intr = enic_legacy_notify_intr();
454 455
	u32 pba;

456
	vnic_intr_mask(&enic->intr[io_intr]);
457 458 459

	pba = vnic_intr_legacy_pba(enic->legacy_pba);
	if (!pba) {
460
		vnic_intr_unmask(&enic->intr[io_intr]);
461 462 463
		return IRQ_NONE;	/* not our interrupt */
	}

464
	if (ENIC_TEST_INTR(pba, notify_intr)) {
465
		enic_notify_check(enic);
466
		vnic_intr_return_all_credits(&enic->intr[notify_intr]);
467
	}
468

469 470
	if (ENIC_TEST_INTR(pba, err_intr)) {
		vnic_intr_return_all_credits(&enic->intr[err_intr]);
471 472 473 474 475 476
		enic_log_q_error(enic);
		/* schedule recovery from WQ/RQ error */
		schedule_work(&enic->reset);
		return IRQ_HANDLED;
	}

477 478 479
	if (ENIC_TEST_INTR(pba, io_intr))
		napi_schedule_irqoff(&enic->napi[0]);
	else
480
		vnic_intr_unmask(&enic->intr[io_intr]);
481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504

	return IRQ_HANDLED;
}

static irqreturn_t enic_isr_msi(int irq, void *data)
{
	struct enic *enic = data;

	/* With MSI, there is no sharing of interrupts, so this is
	 * our interrupt and there is no need to ack it.  The device
	 * is not providing per-vector masking, so the OS will not
	 * write to PCI config space to mask/unmask the interrupt.
	 * We're using mask_on_assertion for MSI, so the device
	 * automatically masks the interrupt when the interrupt is
	 * generated.  Later, when exiting polling, the interrupt
	 * will be unmasked (see enic_poll).
	 *
	 * Also, the device uses the same PCIe Traffic Class (TC)
	 * for Memory Write data and MSI, so there are no ordering
	 * issues; the MSI will always arrive at the Root Complex
	 * _after_ corresponding Memory Writes (i.e. descriptor
	 * writes).
	 */

505
	napi_schedule_irqoff(&enic->napi[0]);
506 507 508 509

	return IRQ_HANDLED;
}

510
static irqreturn_t enic_isr_msix(int irq, void *data)
511
{
512
	struct napi_struct *napi = data;
513

514
	napi_schedule_irqoff(napi);
515 516 517 518 519 520 521

	return IRQ_HANDLED;
}

static irqreturn_t enic_isr_msix_err(int irq, void *data)
{
	struct enic *enic = data;
522
	unsigned int intr = enic_msix_err_intr(enic);
523

524
	vnic_intr_return_all_credits(&enic->intr[intr]);
525

526 527 528
	if (enic_log_q_error(enic))
		/* schedule recovery from WQ/RQ error */
		schedule_work(&enic->reset);
529 530 531 532 533 534 535

	return IRQ_HANDLED;
}

static irqreturn_t enic_isr_msix_notify(int irq, void *data)
{
	struct enic *enic = data;
536
	unsigned int intr = enic_msix_notify_intr(enic);
537 538

	enic_notify_check(enic);
539
	vnic_intr_return_all_credits(&enic->intr[intr]);
540 541 542 543

	return IRQ_HANDLED;
}

544 545 546
static int enic_queue_wq_skb_cont(struct enic *enic, struct vnic_wq *wq,
				  struct sk_buff *skb, unsigned int len_left,
				  int loopback)
547
{
E
Eric Dumazet 已提交
548
	const skb_frag_t *frag;
549
	dma_addr_t dma_addr;
550 551 552

	/* Queue additional data fragments */
	for (frag = skb_shinfo(skb)->frags; len_left; frag++) {
E
Eric Dumazet 已提交
553
		len_left -= skb_frag_size(frag);
554 555 556 557 558 559 560 561
		dma_addr = skb_frag_dma_map(&enic->pdev->dev, frag, 0,
					    skb_frag_size(frag),
					    DMA_TO_DEVICE);
		if (unlikely(enic_dma_map_check(enic, dma_addr)))
			return -ENOMEM;
		enic_queue_wq_desc_cont(wq, skb, dma_addr, skb_frag_size(frag),
					(len_left == 0),	/* EOP? */
					loopback);
562
	}
563 564

	return 0;
565 566
}

567 568 569
static int enic_queue_wq_skb_vlan(struct enic *enic, struct vnic_wq *wq,
				  struct sk_buff *skb, int vlan_tag_insert,
				  unsigned int vlan_tag, int loopback)
570 571 572 573
{
	unsigned int head_len = skb_headlen(skb);
	unsigned int len_left = skb->len - head_len;
	int eop = (len_left == 0);
574 575 576 577 578 579 580
	dma_addr_t dma_addr;
	int err = 0;

	dma_addr = pci_map_single(enic->pdev, skb->data, head_len,
				  PCI_DMA_TODEVICE);
	if (unlikely(enic_dma_map_check(enic, dma_addr)))
		return -ENOMEM;
581

582 583 584 585 586
	/* Queue the main skb fragment. The fragments are no larger
	 * than max MTU(9000)+ETH_HDR_LEN(14) bytes, which is less
	 * than WQ_ENET_MAX_DESC_LEN length. So only one descriptor
	 * per fragment is queued.
	 */
587 588
	enic_queue_wq_desc(wq, skb, dma_addr, head_len,	vlan_tag_insert,
			   vlan_tag, eop, loopback);
589 590

	if (!eop)
591 592 593
		err = enic_queue_wq_skb_cont(enic, wq, skb, len_left, loopback);

	return err;
594 595
}

596 597 598
static int enic_queue_wq_skb_csum_l4(struct enic *enic, struct vnic_wq *wq,
				     struct sk_buff *skb, int vlan_tag_insert,
				     unsigned int vlan_tag, int loopback)
599 600 601
{
	unsigned int head_len = skb_headlen(skb);
	unsigned int len_left = skb->len - head_len;
602
	unsigned int hdr_len = skb_checksum_start_offset(skb);
603 604
	unsigned int csum_offset = hdr_len + skb->csum_offset;
	int eop = (len_left == 0);
605 606 607 608 609 610 611
	dma_addr_t dma_addr;
	int err = 0;

	dma_addr = pci_map_single(enic->pdev, skb->data, head_len,
				  PCI_DMA_TODEVICE);
	if (unlikely(enic_dma_map_check(enic, dma_addr)))
		return -ENOMEM;
612

613 614 615 616 617
	/* Queue the main skb fragment. The fragments are no larger
	 * than max MTU(9000)+ETH_HDR_LEN(14) bytes, which is less
	 * than WQ_ENET_MAX_DESC_LEN length. So only one descriptor
	 * per fragment is queued.
	 */
618 619 620
	enic_queue_wq_desc_csum_l4(wq, skb, dma_addr, head_len,	csum_offset,
				   hdr_len, vlan_tag_insert, vlan_tag, eop,
				   loopback);
621 622

	if (!eop)
623 624 625
		err = enic_queue_wq_skb_cont(enic, wq, skb, len_left, loopback);

	return err;
626 627
}

628
static void enic_preload_tcp_csum_encap(struct sk_buff *skb)
629
{
630 631 632 633
	const struct ethhdr *eth = (struct ethhdr *)skb_inner_mac_header(skb);

	switch (eth->h_proto) {
	case ntohs(ETH_P_IP):
634 635 636 637 638
		inner_ip_hdr(skb)->check = 0;
		inner_tcp_hdr(skb)->check =
			~csum_tcpudp_magic(inner_ip_hdr(skb)->saddr,
					   inner_ip_hdr(skb)->daddr, 0,
					   IPPROTO_TCP, 0);
639 640 641 642 643 644 645 646 647 648
		break;
	case ntohs(ETH_P_IPV6):
		inner_tcp_hdr(skb)->check =
			~csum_ipv6_magic(&inner_ipv6_hdr(skb)->saddr,
					 &inner_ipv6_hdr(skb)->daddr, 0,
					 IPPROTO_TCP, 0);
		break;
	default:
		WARN_ONCE(1, "Non ipv4/ipv6 inner pkt for encap offload");
		break;
649 650
	}
}
651

652 653
static void enic_preload_tcp_csum(struct sk_buff *skb)
{
654 655 656 657 658
	/* Preload TCP csum field with IP pseudo hdr calculated
	 * with IP length set to zero.  HW will later add in length
	 * to each TCP segment resulting from the TSO.
	 */

659
	if (skb->protocol == cpu_to_be16(ETH_P_IP)) {
660 661 662
		ip_hdr(skb)->check = 0;
		tcp_hdr(skb)->check = ~csum_tcpudp_magic(ip_hdr(skb)->saddr,
			ip_hdr(skb)->daddr, 0, IPPROTO_TCP, 0);
663
	} else if (skb->protocol == cpu_to_be16(ETH_P_IPV6)) {
664
		tcp_v6_gso_csum_prep(skb);
665
	}
666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
}

static int enic_queue_wq_skb_tso(struct enic *enic, struct vnic_wq *wq,
				 struct sk_buff *skb, unsigned int mss,
				 int vlan_tag_insert, unsigned int vlan_tag,
				 int loopback)
{
	unsigned int frag_len_left = skb_headlen(skb);
	unsigned int len_left = skb->len - frag_len_left;
	int eop = (len_left == 0);
	unsigned int offset = 0;
	unsigned int hdr_len;
	dma_addr_t dma_addr;
	unsigned int len;
	skb_frag_t *frag;

	if (skb->encapsulation) {
		hdr_len = skb_inner_transport_header(skb) - skb->data;
		hdr_len += inner_tcp_hdrlen(skb);
		enic_preload_tcp_csum_encap(skb);
	} else {
		hdr_len = skb_transport_offset(skb) + tcp_hdrlen(skb);
		enic_preload_tcp_csum(skb);
	}
690

691 692 693 694 695
	/* Queue WQ_ENET_MAX_DESC_LEN length descriptors
	 * for the main skb fragment
	 */
	while (frag_len_left) {
		len = min(frag_len_left, (unsigned int)WQ_ENET_MAX_DESC_LEN);
696 697 698 699 700 701 702
		dma_addr = pci_map_single(enic->pdev, skb->data + offset, len,
					  PCI_DMA_TODEVICE);
		if (unlikely(enic_dma_map_check(enic, dma_addr)))
			return -ENOMEM;
		enic_queue_wq_desc_tso(wq, skb, dma_addr, len, mss, hdr_len,
				       vlan_tag_insert, vlan_tag,
				       eop && (len == frag_len_left), loopback);
703 704 705
		frag_len_left -= len;
		offset += len;
	}
706

707
	if (eop)
708
		return 0;
709 710 711 712 713

	/* Queue WQ_ENET_MAX_DESC_LEN length descriptors
	 * for additional data fragments
	 */
	for (frag = skb_shinfo(skb)->frags; len_left; frag++) {
E
Eric Dumazet 已提交
714 715
		len_left -= skb_frag_size(frag);
		frag_len_left = skb_frag_size(frag);
716
		offset = 0;
717 718 719 720

		while (frag_len_left) {
			len = min(frag_len_left,
				(unsigned int)WQ_ENET_MAX_DESC_LEN);
721 722
			dma_addr = skb_frag_dma_map(&enic->pdev->dev, frag,
						    offset, len,
723
						    DMA_TO_DEVICE);
724 725 726 727 728 729
			if (unlikely(enic_dma_map_check(enic, dma_addr)))
				return -ENOMEM;
			enic_queue_wq_desc_cont(wq, skb, dma_addr, len,
						(len_left == 0) &&
						 (len == frag_len_left),/*EOP*/
						loopback);
730 731 732 733
			frag_len_left -= len;
			offset += len;
		}
	}
734 735

	return 0;
736 737
}

738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769
static inline int enic_queue_wq_skb_encap(struct enic *enic, struct vnic_wq *wq,
					  struct sk_buff *skb,
					  int vlan_tag_insert,
					  unsigned int vlan_tag, int loopback)
{
	unsigned int head_len = skb_headlen(skb);
	unsigned int len_left = skb->len - head_len;
	/* Hardware will overwrite the checksum fields, calculating from
	 * scratch and ignoring the value placed by software.
	 * Offload mode = 00
	 * mss[2], mss[1], mss[0] bits are set
	 */
	unsigned int mss_or_csum = 7;
	int eop = (len_left == 0);
	dma_addr_t dma_addr;
	int err = 0;

	dma_addr = pci_map_single(enic->pdev, skb->data, head_len,
				  PCI_DMA_TODEVICE);
	if (unlikely(enic_dma_map_check(enic, dma_addr)))
		return -ENOMEM;

	enic_queue_wq_desc_ex(wq, skb, dma_addr, head_len, mss_or_csum, 0,
			      vlan_tag_insert, vlan_tag,
			      WQ_ENET_OFFLOAD_MODE_CSUM, eop, 1 /* SOP */, eop,
			      loopback);
	if (!eop)
		err = enic_queue_wq_skb_cont(enic, wq, skb, len_left, loopback);

	return err;
}

770 771 772 773 774 775
static inline void enic_queue_wq_skb(struct enic *enic,
	struct vnic_wq *wq, struct sk_buff *skb)
{
	unsigned int mss = skb_shinfo(skb)->gso_size;
	unsigned int vlan_tag = 0;
	int vlan_tag_insert = 0;
776
	int loopback = 0;
777
	int err;
778

779
	if (skb_vlan_tag_present(skb)) {
780 781
		/* VLAN tag from trunking driver */
		vlan_tag_insert = 1;
782
		vlan_tag = skb_vlan_tag_get(skb);
783 784 785
	} else if (enic->loop_enable) {
		vlan_tag = enic->loop_tag;
		loopback = 1;
786 787 788
	}

	if (mss)
789 790 791
		err = enic_queue_wq_skb_tso(enic, wq, skb, mss,
					    vlan_tag_insert, vlan_tag,
					    loopback);
792 793 794
	else if (skb->encapsulation)
		err = enic_queue_wq_skb_encap(enic, wq, skb, vlan_tag_insert,
					      vlan_tag, loopback);
795
	else if	(skb->ip_summed == CHECKSUM_PARTIAL)
796 797
		err = enic_queue_wq_skb_csum_l4(enic, wq, skb, vlan_tag_insert,
						vlan_tag, loopback);
798
	else
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815
		err = enic_queue_wq_skb_vlan(enic, wq, skb, vlan_tag_insert,
					     vlan_tag, loopback);
	if (unlikely(err)) {
		struct vnic_wq_buf *buf;

		buf = wq->to_use->prev;
		/* while not EOP of previous pkt && queue not empty.
		 * For all non EOP bufs, os_buf is NULL.
		 */
		while (!buf->os_buf && (buf->next != wq->to_clean)) {
			enic_free_wq_buf(wq, buf);
			wq->ring.desc_avail++;
			buf = buf->prev;
		}
		wq->to_use = buf->next;
		dev_kfree_skb(skb);
	}
816 817
}

818
/* netif_tx_lock held, process context with BHs disabled, or BH */
819
static netdev_tx_t enic_hard_start_xmit(struct sk_buff *skb,
820
	struct net_device *netdev)
821 822
{
	struct enic *enic = netdev_priv(netdev);
823 824
	struct vnic_wq *wq;
	unsigned int txq_map;
825
	struct netdev_queue *txq;
826 827

	if (skb->len <= 0) {
828
		dev_kfree_skb_any(skb);
829 830 831
		return NETDEV_TX_OK;
	}

832 833
	txq_map = skb_get_queue_mapping(skb) % enic->wq_count;
	wq = &enic->wq[txq_map];
834
	txq = netdev_get_tx_queue(netdev, txq_map);
835

836 837 838 839 840 841 842 843
	/* Non-TSO sends must fit within ENIC_NON_TSO_MAX_DESC descs,
	 * which is very likely.  In the off chance it's going to take
	 * more than * ENIC_NON_TSO_MAX_DESC, linearize the skb.
	 */

	if (skb_shinfo(skb)->gso_size == 0 &&
	    skb_shinfo(skb)->nr_frags + 1 > ENIC_NON_TSO_MAX_DESC &&
	    skb_linearize(skb)) {
844
		dev_kfree_skb_any(skb);
845 846 847
		return NETDEV_TX_OK;
	}

848
	spin_lock(&enic->wq_lock[txq_map]);
849

850 851
	if (vnic_wq_desc_avail(wq) <
	    skb_shinfo(skb)->nr_frags + ENIC_DESC_MAX_SPLITS) {
852
		netif_tx_stop_queue(txq);
853
		/* This is a hard error, log it */
854
		netdev_err(netdev, "BUG! Tx ring full when queue awake!\n");
855
		spin_unlock(&enic->wq_lock[txq_map]);
856 857 858 859 860
		return NETDEV_TX_BUSY;
	}

	enic_queue_wq_skb(enic, wq, skb);

861
	if (vnic_wq_desc_avail(wq) < MAX_SKB_FRAGS + ENIC_DESC_MAX_SPLITS)
862
		netif_tx_stop_queue(txq);
863
	skb_tx_timestamp(skb);
864
	if (!netdev_xmit_more() || netif_xmit_stopped(txq))
865
		vnic_wq_doorbell(wq);
866

867
	spin_unlock(&enic->wq_lock[txq_map]);
868 869 870 871 872

	return NETDEV_TX_OK;
}

/* dev_base_lock rwlock held, nominally process context */
873 874
static void enic_get_stats(struct net_device *netdev,
			   struct rtnl_link_stats64 *net_stats)
875 876 877
{
	struct enic *enic = netdev_priv(netdev);
	struct vnic_stats *stats;
878
	int err;
879

880 881 882 883 884 885
	err = enic_dev_stats_dump(enic, &stats);
	/* return only when pci_zalloc_consistent fails in vnic_dev_stats_dump
	 * For other failures, like devcmd failure, we return previously
	 * recorded stats.
	 */
	if (err == -ENOMEM)
886
		return;
887

888 889 890 891
	net_stats->tx_packets = stats->tx.tx_frames_ok;
	net_stats->tx_bytes = stats->tx.tx_bytes_ok;
	net_stats->tx_errors = stats->tx.tx_errors;
	net_stats->tx_dropped = stats->tx.tx_drops;
892

893 894 895 896
	net_stats->rx_packets = stats->rx.rx_frames_ok;
	net_stats->rx_bytes = stats->rx.rx_bytes_ok;
	net_stats->rx_errors = stats->rx.rx_errors;
	net_stats->multicast = stats->rx.rx_multicast_frames_ok;
897
	net_stats->rx_over_errors = enic->rq_truncated_pkts;
898
	net_stats->rx_crc_errors = enic->rq_bad_fcs;
899
	net_stats->rx_dropped = stats->rx.rx_no_bufs + stats->rx.rx_drop;
900 901
}

902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959
static int enic_mc_sync(struct net_device *netdev, const u8 *mc_addr)
{
	struct enic *enic = netdev_priv(netdev);

	if (enic->mc_count == ENIC_MULTICAST_PERFECT_FILTERS) {
		unsigned int mc_count = netdev_mc_count(netdev);

		netdev_warn(netdev, "Registering only %d out of %d multicast addresses\n",
			    ENIC_MULTICAST_PERFECT_FILTERS, mc_count);

		return -ENOSPC;
	}

	enic_dev_add_addr(enic, mc_addr);
	enic->mc_count++;

	return 0;
}

static int enic_mc_unsync(struct net_device *netdev, const u8 *mc_addr)
{
	struct enic *enic = netdev_priv(netdev);

	enic_dev_del_addr(enic, mc_addr);
	enic->mc_count--;

	return 0;
}

static int enic_uc_sync(struct net_device *netdev, const u8 *uc_addr)
{
	struct enic *enic = netdev_priv(netdev);

	if (enic->uc_count == ENIC_UNICAST_PERFECT_FILTERS) {
		unsigned int uc_count = netdev_uc_count(netdev);

		netdev_warn(netdev, "Registering only %d out of %d unicast addresses\n",
			    ENIC_UNICAST_PERFECT_FILTERS, uc_count);

		return -ENOSPC;
	}

	enic_dev_add_addr(enic, uc_addr);
	enic->uc_count++;

	return 0;
}

static int enic_uc_unsync(struct net_device *netdev, const u8 *uc_addr)
{
	struct enic *enic = netdev_priv(netdev);

	enic_dev_del_addr(enic, uc_addr);
	enic->uc_count--;

	return 0;
}

960
void enic_reset_addr_lists(struct enic *enic)
961
{
962 963 964 965 966
	struct net_device *netdev = enic->netdev;

	__dev_uc_unsync(netdev, NULL);
	__dev_mc_unsync(netdev, NULL);

967
	enic->mc_count = 0;
968
	enic->uc_count = 0;
969
	enic->flags = 0;
970 971 972 973
}

static int enic_set_mac_addr(struct net_device *netdev, char *addr)
{
974 975
	struct enic *enic = netdev_priv(netdev);

976
	if (enic_is_dynamic(enic) || enic_is_sriov_vf(enic)) {
977 978 979 980 981 982
		if (!is_valid_ether_addr(addr) && !is_zero_ether_addr(addr))
			return -EADDRNOTAVAIL;
	} else {
		if (!is_valid_ether_addr(addr))
			return -EADDRNOTAVAIL;
	}
983 984 985 986 987 988

	memcpy(netdev->dev_addr, addr, netdev->addr_len);

	return 0;
}

989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016
static int enic_set_mac_address_dynamic(struct net_device *netdev, void *p)
{
	struct enic *enic = netdev_priv(netdev);
	struct sockaddr *saddr = p;
	char *addr = saddr->sa_data;
	int err;

	if (netif_running(enic->netdev)) {
		err = enic_dev_del_station_addr(enic);
		if (err)
			return err;
	}

	err = enic_set_mac_addr(netdev, addr);
	if (err)
		return err;

	if (netif_running(enic->netdev)) {
		err = enic_dev_add_station_addr(enic);
		if (err)
			return err;
	}

	return err;
}

static int enic_set_mac_address(struct net_device *netdev, void *p)
{
R
Roopa Prabhu 已提交
1017
	struct sockaddr *saddr = p;
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028
	char *addr = saddr->sa_data;
	struct enic *enic = netdev_priv(netdev);
	int err;

	err = enic_dev_del_station_addr(enic);
	if (err)
		return err;

	err = enic_set_mac_addr(netdev, addr);
	if (err)
		return err;
R
Roopa Prabhu 已提交
1029

1030
	return enic_dev_add_station_addr(enic);
1031 1032
}

1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
/* netif_tx_lock held, BHs disabled */
static void enic_set_rx_mode(struct net_device *netdev)
{
	struct enic *enic = netdev_priv(netdev);
	int directed = 1;
	int multicast = (netdev->flags & IFF_MULTICAST) ? 1 : 0;
	int broadcast = (netdev->flags & IFF_BROADCAST) ? 1 : 0;
	int promisc = (netdev->flags & IFF_PROMISC) ||
		netdev_uc_count(netdev) > ENIC_UNICAST_PERFECT_FILTERS;
	int allmulti = (netdev->flags & IFF_ALLMULTI) ||
		netdev_mc_count(netdev) > ENIC_MULTICAST_PERFECT_FILTERS;
	unsigned int flags = netdev->flags |
		(allmulti ? IFF_ALLMULTI : 0) |
		(promisc ? IFF_PROMISC : 0);

	if (enic->flags != flags) {
		enic->flags = flags;
		enic_dev_packet_filter(enic, directed,
			multicast, broadcast, promisc, allmulti);
	}

	if (!promisc) {
1055
		__dev_uc_sync(netdev, enic_uc_sync, enic_uc_unsync);
1056
		if (!allmulti)
1057
			__dev_mc_sync(netdev, enic_mc_sync, enic_mc_unsync);
1058 1059 1060
	}
}

1061
/* netif_tx_lock held, BHs disabled */
1062
static void enic_tx_timeout(struct net_device *netdev, unsigned int txqueue)
1063 1064
{
	struct enic *enic = netdev_priv(netdev);
1065
	schedule_work(&enic->tx_hang_reset);
1066 1067
}

1068 1069 1070
static int enic_set_vf_mac(struct net_device *netdev, int vf, u8 *mac)
{
	struct enic *enic = netdev_priv(netdev);
1071 1072
	struct enic_port_profile *pp;
	int err;
1073

1074 1075 1076
	ENIC_PP_BY_INDEX(enic, vf, pp, &err);
	if (err)
		return err;
1077

1078
	if (is_valid_ether_addr(mac) || is_zero_ether_addr(mac)) {
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
		if (vf == PORT_SELF_VF) {
			memcpy(pp->vf_mac, mac, ETH_ALEN);
			return 0;
		} else {
			/*
			 * For sriov vf's set the mac in hw
			 */
			ENIC_DEVCMD_PROXY_BY_INDEX(vf, err, enic,
				vnic_dev_set_mac_addr, mac);
			return enic_dev_status_to_errno(err);
		}
1090 1091 1092 1093
	} else
		return -EINVAL;
}

1094 1095 1096 1097
static int enic_set_vf_port(struct net_device *netdev, int vf,
	struct nlattr *port[])
{
	struct enic *enic = netdev_priv(netdev);
1098
	struct enic_port_profile prev_pp;
1099
	struct enic_port_profile *pp;
1100
	int err = 0, restore_pp = 1;
1101

1102 1103 1104
	ENIC_PP_BY_INDEX(enic, vf, pp, &err);
	if (err)
		return err;
1105

1106 1107 1108
	if (!port[IFLA_PORT_REQUEST])
		return -EOPNOTSUPP;

1109 1110
	memcpy(&prev_pp, pp, sizeof(*enic->pp));
	memset(pp, 0, sizeof(*enic->pp));
1111

1112 1113
	pp->set |= ENIC_SET_REQUEST;
	pp->request = nla_get_u8(port[IFLA_PORT_REQUEST]);
1114 1115

	if (port[IFLA_PORT_PROFILE]) {
1116 1117
		pp->set |= ENIC_SET_NAME;
		memcpy(pp->name, nla_data(port[IFLA_PORT_PROFILE]),
1118 1119 1120 1121
			PORT_PROFILE_MAX);
	}

	if (port[IFLA_PORT_INSTANCE_UUID]) {
1122 1123
		pp->set |= ENIC_SET_INSTANCE;
		memcpy(pp->instance_uuid,
1124 1125 1126 1127
			nla_data(port[IFLA_PORT_INSTANCE_UUID]), PORT_UUID_MAX);
	}

	if (port[IFLA_PORT_HOST_UUID]) {
1128 1129
		pp->set |= ENIC_SET_HOST;
		memcpy(pp->host_uuid,
1130 1131
			nla_data(port[IFLA_PORT_HOST_UUID]), PORT_UUID_MAX);
	}
1132

1133 1134 1135 1136
	if (vf == PORT_SELF_VF) {
		/* Special case handling: mac came from IFLA_VF_MAC */
		if (!is_zero_ether_addr(prev_pp.vf_mac))
			memcpy(pp->mac_addr, prev_pp.vf_mac, ETH_ALEN);
1137

1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
		if (is_zero_ether_addr(netdev->dev_addr))
			eth_hw_addr_random(netdev);
	} else {
		/* SR-IOV VF: get mac from adapter */
		ENIC_DEVCMD_PROXY_BY_INDEX(vf, err, enic,
			vnic_dev_get_mac_addr, pp->mac_addr);
		if (err) {
			netdev_err(netdev, "Error getting mac for vf %d\n", vf);
			memcpy(pp, &prev_pp, sizeof(*pp));
			return enic_dev_status_to_errno(err);
		}
	}
1150

1151
	err = enic_process_set_pp_request(enic, vf, &prev_pp, &restore_pp);
1152 1153 1154 1155 1156
	if (err) {
		if (restore_pp) {
			/* Things are still the way they were: Implicit
			 * DISASSOCIATE failed
			 */
1157
			memcpy(pp, &prev_pp, sizeof(*pp));
1158
		} else {
1159 1160
			memset(pp, 0, sizeof(*pp));
			if (vf == PORT_SELF_VF)
1161
				eth_zero_addr(netdev->dev_addr);
1162 1163 1164 1165 1166
		}
	} else {
		/* Set flag to indicate that the port assoc/disassoc
		 * request has been sent out to fw
		 */
1167
		pp->set |= ENIC_PORT_REQUEST_APPLIED;
1168 1169

		/* If DISASSOCIATE, clean up all assigned/saved macaddresses */
1170
		if (pp->request == PORT_REQUEST_DISASSOCIATE) {
1171
			eth_zero_addr(pp->mac_addr);
1172
			if (vf == PORT_SELF_VF)
1173
				eth_zero_addr(netdev->dev_addr);
1174 1175
		}
	}
1176

1177
	if (vf == PORT_SELF_VF)
1178
		eth_zero_addr(pp->vf_mac);
1179 1180

	return err;
1181 1182 1183 1184 1185 1186 1187
}

static int enic_get_vf_port(struct net_device *netdev, int vf,
	struct sk_buff *skb)
{
	struct enic *enic = netdev_priv(netdev);
	u16 response = PORT_PROFILE_RESPONSE_SUCCESS;
1188
	struct enic_port_profile *pp;
1189
	int err;
1190

1191 1192 1193 1194 1195
	ENIC_PP_BY_INDEX(enic, vf, pp, &err);
	if (err)
		return err;

	if (!(pp->set & ENIC_PORT_REQUEST_APPLIED))
1196
		return -ENODATA;
1197

1198
	err = enic_process_get_pp_request(enic, vf, pp->request, &response);
1199
	if (err)
1200
		return err;
1201

D
David S. Miller 已提交
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
	if (nla_put_u16(skb, IFLA_PORT_REQUEST, pp->request) ||
	    nla_put_u16(skb, IFLA_PORT_RESPONSE, response) ||
	    ((pp->set & ENIC_SET_NAME) &&
	     nla_put(skb, IFLA_PORT_PROFILE, PORT_PROFILE_MAX, pp->name)) ||
	    ((pp->set & ENIC_SET_INSTANCE) &&
	     nla_put(skb, IFLA_PORT_INSTANCE_UUID, PORT_UUID_MAX,
		     pp->instance_uuid)) ||
	    ((pp->set & ENIC_SET_HOST) &&
	     nla_put(skb, IFLA_PORT_HOST_UUID, PORT_UUID_MAX, pp->host_uuid)))
		goto nla_put_failure;
1212 1213 1214 1215 1216 1217
	return 0;

nla_put_failure:
	return -EMSGSIZE;
}

1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
static void enic_free_rq_buf(struct vnic_rq *rq, struct vnic_rq_buf *buf)
{
	struct enic *enic = vnic_dev_priv(rq->vdev);

	if (!buf->os_buf)
		return;

	pci_unmap_single(enic->pdev, buf->dma_addr,
		buf->len, PCI_DMA_FROMDEVICE);
	dev_kfree_skb_any(buf->os_buf);
1228
	buf->os_buf = NULL;
1229 1230 1231 1232 1233
}

static int enic_rq_alloc_buf(struct vnic_rq *rq)
{
	struct enic *enic = vnic_dev_priv(rq->vdev);
S
Scott Feldman 已提交
1234
	struct net_device *netdev = enic->netdev;
1235
	struct sk_buff *skb;
1236
	unsigned int len = netdev->mtu + VLAN_ETH_HLEN;
1237 1238
	unsigned int os_buf_index = 0;
	dma_addr_t dma_addr;
1239 1240 1241
	struct vnic_rq_buf *buf = rq->to_use;

	if (buf->os_buf) {
1242 1243
		enic_queue_rq_desc(rq, buf->os_buf, os_buf_index, buf->dma_addr,
				   buf->len);
1244

1245 1246
		return 0;
	}
1247
	skb = netdev_alloc_skb_ip_align(netdev, len);
1248 1249 1250
	if (!skb)
		return -ENOMEM;

1251 1252 1253 1254 1255 1256
	dma_addr = pci_map_single(enic->pdev, skb->data, len,
				  PCI_DMA_FROMDEVICE);
	if (unlikely(enic_dma_map_check(enic, dma_addr))) {
		dev_kfree_skb(skb);
		return -ENOMEM;
	}
1257 1258 1259 1260 1261 1262 1263

	enic_queue_rq_desc(rq, skb, os_buf_index,
		dma_addr, len);

	return 0;
}

1264 1265 1266 1267 1268 1269 1270 1271 1272
static void enic_intr_update_pkt_size(struct vnic_rx_bytes_counter *pkt_size,
				      u32 pkt_len)
{
	if (ENIC_LARGE_PKT_THRESHOLD <= pkt_len)
		pkt_size->large_pkt_bytes_cnt += pkt_len;
	else
		pkt_size->small_pkt_bytes_cnt += pkt_len;
}

1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
static bool enic_rxcopybreak(struct net_device *netdev, struct sk_buff **skb,
			     struct vnic_rq_buf *buf, u16 len)
{
	struct enic *enic = netdev_priv(netdev);
	struct sk_buff *new_skb;

	if (len > enic->rx_copybreak)
		return false;
	new_skb = netdev_alloc_skb_ip_align(netdev, len);
	if (!new_skb)
		return false;
	pci_dma_sync_single_for_cpu(enic->pdev, buf->dma_addr, len,
				    DMA_FROM_DEVICE);
	memcpy(new_skb->data, (*skb)->data, len);
	*skb = new_skb;

	return true;
}

1292 1293 1294 1295 1296
static void enic_rq_indicate_buf(struct vnic_rq *rq,
	struct cq_desc *cq_desc, struct vnic_rq_buf *buf,
	int skipped, void *opaque)
{
	struct enic *enic = vnic_dev_priv(rq->vdev);
S
Scott Feldman 已提交
1297
	struct net_device *netdev = enic->netdev;
1298
	struct sk_buff *skb;
1299
	struct vnic_cq *cq = &enic->cq[enic_cq_rq(enic, rq->index)];
1300 1301 1302 1303 1304 1305

	u8 type, color, eop, sop, ingress_port, vlan_stripped;
	u8 fcoe, fcoe_sof, fcoe_fc_crc_ok, fcoe_enc_error, fcoe_eof;
	u8 tcp_udp_csum_ok, udp, tcp, ipv4_csum_ok;
	u8 ipv6, ipv4, ipv4_fragment, fcs_ok, rss_type, csum_not_calc;
	u8 packet_error;
1306
	u16 q_number, completed_index, bytes_written, vlan_tci, checksum;
1307
	u32 rss_hash;
1308
	bool outer_csum_ok = true, encap = false;
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318

	if (skipped)
		return;

	skb = buf->os_buf;

	cq_enet_rq_desc_dec((struct cq_enet_rq_desc *)cq_desc,
		&type, &color, &q_number, &completed_index,
		&ingress_port, &fcoe, &eop, &sop, &rss_type,
		&csum_not_calc, &rss_hash, &bytes_written,
1319
		&packet_error, &vlan_stripped, &vlan_tci, &checksum,
1320 1321 1322 1323 1324 1325 1326
		&fcoe_sof, &fcoe_fc_crc_ok, &fcoe_enc_error,
		&fcoe_eof, &tcp_udp_csum_ok, &udp, &tcp,
		&ipv4_csum_ok, &ipv6, &ipv4, &ipv4_fragment,
		&fcs_ok);

	if (packet_error) {

1327 1328 1329 1330 1331 1332
		if (!fcs_ok) {
			if (bytes_written > 0)
				enic->rq_bad_fcs++;
			else if (bytes_written == 0)
				enic->rq_truncated_pkts++;
		}
1333

1334 1335
		pci_unmap_single(enic->pdev, buf->dma_addr, buf->len,
				 PCI_DMA_FROMDEVICE);
1336
		dev_kfree_skb_any(skb);
1337
		buf->os_buf = NULL;
1338 1339 1340 1341 1342 1343 1344 1345 1346

		return;
	}

	if (eop && bytes_written > 0) {

		/* Good receive
		 */

1347 1348 1349 1350 1351 1352 1353
		if (!enic_rxcopybreak(netdev, &skb, buf, bytes_written)) {
			buf->os_buf = NULL;
			pci_unmap_single(enic->pdev, buf->dma_addr, buf->len,
					 PCI_DMA_FROMDEVICE);
		}
		prefetch(skb->data - NET_IP_ALIGN);

1354
		skb_put(skb, bytes_written);
S
Scott Feldman 已提交
1355
		skb->protocol = eth_type_trans(skb, netdev);
1356
		skb_record_rx_queue(skb, q_number);
1357 1358
		if ((netdev->features & NETIF_F_RXHASH) && rss_hash &&
		    (type == 3)) {
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370
			switch (rss_type) {
			case CQ_ENET_RQ_DESC_RSS_TYPE_TCP_IPv4:
			case CQ_ENET_RQ_DESC_RSS_TYPE_TCP_IPv6:
			case CQ_ENET_RQ_DESC_RSS_TYPE_TCP_IPv6_EX:
				skb_set_hash(skb, rss_hash, PKT_HASH_TYPE_L4);
				break;
			case CQ_ENET_RQ_DESC_RSS_TYPE_IPv4:
			case CQ_ENET_RQ_DESC_RSS_TYPE_IPv6:
			case CQ_ENET_RQ_DESC_RSS_TYPE_IPv6_EX:
				skb_set_hash(skb, rss_hash, PKT_HASH_TYPE_L3);
				break;
			}
1371
		}
1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389
		if (enic->vxlan.vxlan_udp_port_number) {
			switch (enic->vxlan.patch_level) {
			case 0:
				if (fcoe) {
					encap = true;
					outer_csum_ok = fcoe_fc_crc_ok;
				}
				break;
			case 2:
				if ((type == 7) &&
				    (rss_hash & BIT(0))) {
					encap = true;
					outer_csum_ok = (rss_hash & BIT(1)) &&
							(rss_hash & BIT(2));
				}
				break;
			}
		}
1390

1391 1392 1393 1394
		/* Hardware does not provide whole packet checksum. It only
		 * provides pseudo checksum. Since hw validates the packet
		 * checksum but not provide us the checksum value. use
		 * CHECSUM_UNNECESSARY.
1395 1396 1397 1398
		 *
		 * In case of encap pkt tcp_udp_csum_ok/tcp_udp_csum_ok is
		 * inner csum_ok. outer_csum_ok is set by hw when outer udp
		 * csum is correct or is zero.
1399
		 */
1400
		if ((netdev->features & NETIF_F_RXCSUM) && !csum_not_calc &&
1401 1402
		    tcp_udp_csum_ok && outer_csum_ok &&
		    (ipv4_csum_ok || ipv6)) {
1403
			skb->ip_summed = CHECKSUM_UNNECESSARY;
1404 1405
			skb->csum_level = encap;
		}
1406

J
Jiri Pirko 已提交
1407
		if (vlan_stripped)
1408
			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), vlan_tci);
1409

1410
		skb_mark_napi_id(skb, &enic->napi[rq->index]);
1411
		if (!(netdev->features & NETIF_F_GRO))
J
Jiri Pirko 已提交
1412
			netif_receive_skb(skb);
1413 1414
		else
			napi_gro_receive(&enic->napi[q_number], skb);
1415 1416 1417
		if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
			enic_intr_update_pkt_size(&cq->pkt_size_counter,
						  bytes_written);
1418 1419 1420 1421 1422
	} else {

		/* Buffer overflow
		 */

1423 1424
		pci_unmap_single(enic->pdev, buf->dma_addr, buf->len,
				 PCI_DMA_FROMDEVICE);
1425
		dev_kfree_skb_any(skb);
1426
		buf->os_buf = NULL;
1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441
	}
}

static int enic_rq_service(struct vnic_dev *vdev, struct cq_desc *cq_desc,
	u8 type, u16 q_number, u16 completed_index, void *opaque)
{
	struct enic *enic = vnic_dev_priv(vdev);

	vnic_rq_service(&enic->rq[q_number], cq_desc,
		completed_index, VNIC_RQ_RETURN_DESC,
		enic_rq_indicate_buf, opaque);

	return 0;
}

1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499
static void enic_set_int_moderation(struct enic *enic, struct vnic_rq *rq)
{
	unsigned int intr = enic_msix_rq_intr(enic, rq->index);
	struct vnic_cq *cq = &enic->cq[enic_cq_rq(enic, rq->index)];
	u32 timer = cq->tobe_rx_coal_timeval;

	if (cq->tobe_rx_coal_timeval != cq->cur_rx_coal_timeval) {
		vnic_intr_coalescing_timer_set(&enic->intr[intr], timer);
		cq->cur_rx_coal_timeval = cq->tobe_rx_coal_timeval;
	}
}

static void enic_calc_int_moderation(struct enic *enic, struct vnic_rq *rq)
{
	struct enic_rx_coal *rx_coal = &enic->rx_coalesce_setting;
	struct vnic_cq *cq = &enic->cq[enic_cq_rq(enic, rq->index)];
	struct vnic_rx_bytes_counter *pkt_size_counter = &cq->pkt_size_counter;
	int index;
	u32 timer;
	u32 range_start;
	u32 traffic;
	u64 delta;
	ktime_t now = ktime_get();

	delta = ktime_us_delta(now, cq->prev_ts);
	if (delta < ENIC_AIC_TS_BREAK)
		return;
	cq->prev_ts = now;

	traffic = pkt_size_counter->large_pkt_bytes_cnt +
		  pkt_size_counter->small_pkt_bytes_cnt;
	/* The table takes Mbps
	 * traffic *= 8    => bits
	 * traffic *= (10^6 / delta)    => bps
	 * traffic /= 10^6     => Mbps
	 *
	 * Combining, traffic *= (8 / delta)
	 */

	traffic <<= 3;
	traffic = delta > UINT_MAX ? 0 : traffic / (u32)delta;

	for (index = 0; index < ENIC_MAX_COALESCE_TIMERS; index++)
		if (traffic < mod_table[index].rx_rate)
			break;
	range_start = (pkt_size_counter->small_pkt_bytes_cnt >
		       pkt_size_counter->large_pkt_bytes_cnt << 1) ?
		      rx_coal->small_pkt_range_start :
		      rx_coal->large_pkt_range_start;
	timer = range_start + ((rx_coal->range_end - range_start) *
			       mod_table[index].range_percent / 100);
	/* Damping */
	cq->tobe_rx_coal_timeval = (timer + cq->tobe_rx_coal_timeval) >> 1;

	pkt_size_counter->large_pkt_bytes_cnt = 0;
	pkt_size_counter->small_pkt_bytes_cnt = 0;
}

1500 1501
static int enic_poll(struct napi_struct *napi, int budget)
{
1502 1503 1504 1505 1506
	struct net_device *netdev = napi->dev;
	struct enic *enic = netdev_priv(netdev);
	unsigned int cq_rq = enic_cq_rq(enic, 0);
	unsigned int cq_wq = enic_cq_wq(enic, 0);
	unsigned int intr = enic_legacy_io_intr();
1507
	unsigned int rq_work_to_do = budget;
1508
	unsigned int wq_work_to_do = ENIC_WQ_NAPI_BUDGET;
1509
	unsigned int  work_done, rq_work_done = 0, wq_work_done;
1510
	int err;
1511

1512 1513 1514
	wq_work_done = vnic_cq_service(&enic->cq[cq_wq], wq_work_to_do,
				       enic_wq_service, NULL);

1515 1516 1517
	if (budget > 0)
		rq_work_done = vnic_cq_service(&enic->cq[cq_rq],
			rq_work_to_do, enic_rq_service, NULL);
1518 1519 1520 1521 1522 1523 1524 1525 1526

	/* Accumulate intr event credits for this polling
	 * cycle.  An intr event is the completion of a
	 * a WQ or RQ packet.
	 */

	work_done = rq_work_done + wq_work_done;

	if (work_done > 0)
1527
		vnic_intr_return_credits(&enic->intr[intr],
1528 1529 1530 1531
			work_done,
			0 /* don't unmask intr */,
			0 /* don't reset intr timer */);

1532
	err = vnic_rq_fill(&enic->rq[0], enic_rq_alloc_buf);
1533

1534 1535 1536
	/* Buffer allocation failed. Stay in polling
	 * mode so we can try to fill the ring again.
	 */
1537

1538 1539
	if (err)
		rq_work_done = rq_work_to_do;
1540 1541 1542 1543 1544
	if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
		/* Call the function which refreshes the intr coalescing timer
		 * value based on the traffic.
		 */
		enic_calc_int_moderation(enic, &enic->rq[0]);
1545

1546
	if ((rq_work_done < budget) && napi_complete_done(napi, rq_work_done)) {
1547

1548
		/* Some work done, but not enough to stay in polling,
1549
		 * exit polling
1550 1551
		 */

1552 1553
		if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
			enic_set_int_moderation(enic, &enic->rq[0]);
1554
		vnic_intr_unmask(&enic->intr[intr]);
1555 1556 1557 1558 1559
	}

	return rq_work_done;
}

1560 1561 1562 1563 1564 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
#ifdef CONFIG_RFS_ACCEL
static void enic_free_rx_cpu_rmap(struct enic *enic)
{
	free_irq_cpu_rmap(enic->netdev->rx_cpu_rmap);
	enic->netdev->rx_cpu_rmap = NULL;
}

static void enic_set_rx_cpu_rmap(struct enic *enic)
{
	int i, res;

	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX) {
		enic->netdev->rx_cpu_rmap = alloc_irq_cpu_rmap(enic->rq_count);
		if (unlikely(!enic->netdev->rx_cpu_rmap))
			return;
		for (i = 0; i < enic->rq_count; i++) {
			res = irq_cpu_rmap_add(enic->netdev->rx_cpu_rmap,
					       enic->msix_entry[i].vector);
			if (unlikely(res)) {
				enic_free_rx_cpu_rmap(enic);
				return;
			}
		}
	}
}

#else

static void enic_free_rx_cpu_rmap(struct enic *enic)
{
}

static void enic_set_rx_cpu_rmap(struct enic *enic)
{
}

#endif /* CONFIG_RFS_ACCEL */

1598 1599 1600 1601 1602 1603 1604 1605
static int enic_poll_msix_wq(struct napi_struct *napi, int budget)
{
	struct net_device *netdev = napi->dev;
	struct enic *enic = netdev_priv(netdev);
	unsigned int wq_index = (napi - &enic->napi[0]) - enic->rq_count;
	struct vnic_wq *wq = &enic->wq[wq_index];
	unsigned int cq;
	unsigned int intr;
1606
	unsigned int wq_work_to_do = ENIC_WQ_NAPI_BUDGET;
1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
	unsigned int wq_work_done;
	unsigned int wq_irq;

	wq_irq = wq->index;
	cq = enic_cq_wq(enic, wq_irq);
	intr = enic_msix_wq_intr(enic, wq_irq);
	wq_work_done = vnic_cq_service(&enic->cq[cq], wq_work_to_do,
				       enic_wq_service, NULL);

	vnic_intr_return_credits(&enic->intr[intr], wq_work_done,
				 0 /* don't unmask intr */,
				 1 /* reset intr timer */);
	if (!wq_work_done) {
		napi_complete(napi);
		vnic_intr_unmask(&enic->intr[intr]);
1622
		return 0;
1623 1624
	}

1625
	return budget;
1626 1627 1628
}

static int enic_poll_msix_rq(struct napi_struct *napi, int budget)
1629
{
1630 1631 1632 1633 1634
	struct net_device *netdev = napi->dev;
	struct enic *enic = netdev_priv(netdev);
	unsigned int rq = (napi - &enic->napi[0]);
	unsigned int cq = enic_cq_rq(enic, rq);
	unsigned int intr = enic_msix_rq_intr(enic, rq);
1635
	unsigned int work_to_do = budget;
1636
	unsigned int work_done = 0;
1637
	int err;
1638 1639 1640 1641

	/* Service RQ
	 */

1642 1643 1644
	if (budget > 0)
		work_done = vnic_cq_service(&enic->cq[cq],
			work_to_do, enic_rq_service, NULL);
1645

1646 1647 1648 1649
	/* Return intr event credits for this polling
	 * cycle.  An intr event is the completion of a
	 * RQ packet.
	 */
1650

1651
	if (work_done > 0)
1652
		vnic_intr_return_credits(&enic->intr[intr],
1653 1654 1655 1656
			work_done,
			0 /* don't unmask intr */,
			0 /* don't reset intr timer */);

1657
	err = vnic_rq_fill(&enic->rq[rq], enic_rq_alloc_buf);
1658 1659 1660 1661 1662 1663 1664

	/* Buffer allocation failed. Stay in polling mode
	 * so we can try to fill the ring again.
	 */

	if (err)
		work_done = work_to_do;
1665
	if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
1666 1667
		/* Call the function which refreshes the intr coalescing timer
		 * value based on the traffic.
1668 1669
		 */
		enic_calc_int_moderation(enic, &enic->rq[rq]);
1670

1671
	if ((work_done < budget) && napi_complete_done(napi, work_done)) {
1672 1673

		/* Some work done, but not enough to stay in polling,
1674
		 * exit polling
1675 1676
		 */

1677 1678
		if (enic->rx_coalesce_setting.use_adaptive_rx_coalesce)
			enic_set_int_moderation(enic, &enic->rq[rq]);
1679
		vnic_intr_unmask(&enic->intr[intr]);
1680 1681 1682 1683 1684
	}

	return work_done;
}

1685
static void enic_notify_timer(struct timer_list *t)
1686
{
1687
	struct enic *enic = from_timer(enic, t, notify_timer);
1688 1689 1690

	enic_notify_check(enic);

1691 1692
	mod_timer(&enic->notify_timer,
		round_jiffies(jiffies + ENIC_NOTIFY_TIMER_PERIOD));
1693 1694 1695 1696 1697 1698 1699
}

static void enic_free_intr(struct enic *enic)
{
	struct net_device *netdev = enic->netdev;
	unsigned int i;

1700
	enic_free_rx_cpu_rmap(enic);
1701 1702 1703 1704
	switch (vnic_dev_get_intr_mode(enic->vdev)) {
	case VNIC_DEV_INTR_MODE_INTX:
		free_irq(enic->pdev->irq, netdev);
		break;
1705 1706 1707
	case VNIC_DEV_INTR_MODE_MSI:
		free_irq(enic->pdev->irq, enic);
		break;
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721
	case VNIC_DEV_INTR_MODE_MSIX:
		for (i = 0; i < ARRAY_SIZE(enic->msix); i++)
			if (enic->msix[i].requested)
				free_irq(enic->msix_entry[i].vector,
					enic->msix[i].devid);
		break;
	default:
		break;
	}
}

static int enic_request_intr(struct enic *enic)
{
	struct net_device *netdev = enic->netdev;
1722
	unsigned int i, intr;
1723 1724
	int err = 0;

1725
	enic_set_rx_cpu_rmap(enic);
1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
	switch (vnic_dev_get_intr_mode(enic->vdev)) {

	case VNIC_DEV_INTR_MODE_INTX:

		err = request_irq(enic->pdev->irq, enic_isr_legacy,
			IRQF_SHARED, netdev->name, netdev);
		break;

	case VNIC_DEV_INTR_MODE_MSI:

		err = request_irq(enic->pdev->irq, enic_isr_msi,
			0, netdev->name, enic);
		break;

	case VNIC_DEV_INTR_MODE_MSIX:

1742 1743
		for (i = 0; i < enic->rq_count; i++) {
			intr = enic_msix_rq_intr(enic, i);
1744 1745
			snprintf(enic->msix[intr].devname,
				sizeof(enic->msix[intr].devname),
1746
				"%s-rx-%u", netdev->name, i);
1747
			enic->msix[intr].isr = enic_isr_msix;
1748 1749
			enic->msix[intr].devid = &enic->napi[i];
		}
1750

1751
		for (i = 0; i < enic->wq_count; i++) {
1752 1753
			int wq = enic_cq_wq(enic, i);

1754
			intr = enic_msix_wq_intr(enic, i);
1755 1756
			snprintf(enic->msix[intr].devname,
				sizeof(enic->msix[intr].devname),
1757
				"%s-tx-%u", netdev->name, i);
1758 1759
			enic->msix[intr].isr = enic_isr_msix;
			enic->msix[intr].devid = &enic->napi[wq];
1760
		}
1761

1762
		intr = enic_msix_err_intr(enic);
1763 1764
		snprintf(enic->msix[intr].devname,
			sizeof(enic->msix[intr].devname),
1765
			"%s-err", netdev->name);
1766 1767
		enic->msix[intr].isr = enic_isr_msix_err;
		enic->msix[intr].devid = enic;
1768

1769
		intr = enic_msix_notify_intr(enic);
1770 1771
		snprintf(enic->msix[intr].devname,
			sizeof(enic->msix[intr].devname),
1772
			"%s-notify", netdev->name);
1773 1774 1775 1776 1777
		enic->msix[intr].isr = enic_isr_msix_notify;
		enic->msix[intr].devid = enic;

		for (i = 0; i < ARRAY_SIZE(enic->msix); i++)
			enic->msix[i].requested = 0;
1778

1779
		for (i = 0; i < enic->intr_count; i++) {
1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799
			err = request_irq(enic->msix_entry[i].vector,
				enic->msix[i].isr, 0,
				enic->msix[i].devname,
				enic->msix[i].devid);
			if (err) {
				enic_free_intr(enic);
				break;
			}
			enic->msix[i].requested = 1;
		}

		break;

	default:
		break;
	}

	return err;
}

1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817
static void enic_synchronize_irqs(struct enic *enic)
{
	unsigned int i;

	switch (vnic_dev_get_intr_mode(enic->vdev)) {
	case VNIC_DEV_INTR_MODE_INTX:
	case VNIC_DEV_INTR_MODE_MSI:
		synchronize_irq(enic->pdev->irq);
		break;
	case VNIC_DEV_INTR_MODE_MSIX:
		for (i = 0; i < enic->intr_count; i++)
			synchronize_irq(enic->msix_entry[i].vector);
		break;
	default:
		break;
	}
}

1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
static void enic_set_rx_coal_setting(struct enic *enic)
{
	unsigned int speed;
	int index = -1;
	struct enic_rx_coal *rx_coal = &enic->rx_coalesce_setting;

	/* 1. Read the link speed from fw
	 * 2. Pick the default range for the speed
	 * 3. Update it in enic->rx_coalesce_setting
	 */
	speed = vnic_dev_port_speed(enic->vdev);
	if (ENIC_LINK_SPEED_10G < speed)
		index = ENIC_LINK_40G_INDEX;
	else if (ENIC_LINK_SPEED_4G < speed)
		index = ENIC_LINK_10G_INDEX;
	else
		index = ENIC_LINK_4G_INDEX;

	rx_coal->small_pkt_range_start = mod_range[index].small_pkt_range_start;
	rx_coal->large_pkt_range_start = mod_range[index].large_pkt_range_start;
	rx_coal->range_end = ENIC_RX_COALESCE_RANGE_END;

	/* Start with the value provided by UCSM */
	for (index = 0; index < enic->rq_count; index++)
		enic->cq[index].cur_rx_coal_timeval =
				enic->config.intr_timer_usec;

	rx_coal->use_adaptive_rx_coalesce = 1;
}

1848
static int enic_dev_notify_set(struct enic *enic)
1849 1850 1851
{
	int err;

1852
	spin_lock_bh(&enic->devcmd_lock);
1853 1854
	switch (vnic_dev_get_intr_mode(enic->vdev)) {
	case VNIC_DEV_INTR_MODE_INTX:
1855 1856
		err = vnic_dev_notify_set(enic->vdev,
			enic_legacy_notify_intr());
1857 1858
		break;
	case VNIC_DEV_INTR_MODE_MSIX:
1859 1860
		err = vnic_dev_notify_set(enic->vdev,
			enic_msix_notify_intr(enic));
1861 1862 1863 1864 1865
		break;
	default:
		err = vnic_dev_notify_set(enic->vdev, -1 /* no intr */);
		break;
	}
1866
	spin_unlock_bh(&enic->devcmd_lock);
1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879

	return err;
}

static void enic_notify_timer_start(struct enic *enic)
{
	switch (vnic_dev_get_intr_mode(enic->vdev)) {
	case VNIC_DEV_INTR_MODE_MSI:
		mod_timer(&enic->notify_timer, jiffies);
		break;
	default:
		/* Using intr for notification for INTx/MSI-X */
		break;
1880
	}
1881 1882 1883 1884 1885 1886 1887
}

/* rtnl lock is held, process context */
static int enic_open(struct net_device *netdev)
{
	struct enic *enic = netdev_priv(netdev);
	unsigned int i;
1888
	int err, ret;
1889

1890 1891
	err = enic_request_intr(enic);
	if (err) {
1892
		netdev_err(netdev, "Unable to request irq.\n");
1893 1894
		return err;
	}
1895 1896
	enic_init_affinity_hint(enic);
	enic_set_affinity_hint(enic);
1897

1898
	err = enic_dev_notify_set(enic);
1899
	if (err) {
1900 1901
		netdev_err(netdev,
			"Failed to alloc notify buffer, aborting.\n");
1902 1903 1904
		goto err_out_free_intr;
	}

1905
	for (i = 0; i < enic->rq_count; i++) {
1906 1907
		/* enable rq before updating rq desc */
		vnic_rq_enable(&enic->rq[i]);
1908
		vnic_rq_fill(&enic->rq[i], enic_rq_alloc_buf);
1909 1910
		/* Need at least one buffer on ring to get going */
		if (vnic_rq_desc_used(&enic->rq[i]) == 0) {
1911
			netdev_err(netdev, "Unable to alloc receive buffers\n");
1912
			err = -ENOMEM;
1913
			goto err_out_free_rq;
1914 1915 1916 1917 1918 1919
		}
	}

	for (i = 0; i < enic->wq_count; i++)
		vnic_wq_enable(&enic->wq[i]);

1920
	if (!enic_is_dynamic(enic) && !enic_is_sriov_vf(enic))
1921
		enic_dev_add_station_addr(enic);
1922

1923
	enic_set_rx_mode(netdev);
1924

1925
	netif_tx_wake_all_queues(netdev);
1926

1927
	for (i = 0; i < enic->rq_count; i++)
1928
		napi_enable(&enic->napi[i]);
1929

1930 1931 1932
	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX)
		for (i = 0; i < enic->wq_count; i++)
			napi_enable(&enic->napi[enic_cq_wq(enic, i)]);
1933
	enic_dev_enable(enic);
1934 1935 1936 1937 1938

	for (i = 0; i < enic->intr_count; i++)
		vnic_intr_unmask(&enic->intr[i]);

	enic_notify_timer_start(enic);
1939
	enic_rfs_timer_start(enic);
1940 1941

	return 0;
1942

1943
err_out_free_rq:
1944
	for (i = 0; i < enic->rq_count; i++) {
1945 1946 1947
		ret = vnic_rq_disable(&enic->rq[i]);
		if (!ret)
			vnic_rq_clean(&enic->rq[i], enic_free_rq_buf);
1948
	}
1949
	enic_dev_notify_unset(enic);
1950
err_out_free_intr:
1951
	enic_unset_affinity_hint(enic);
1952 1953 1954
	enic_free_intr(enic);

	return err;
1955 1956 1957 1958 1959 1960 1961 1962 1963
}

/* rtnl lock is held, process context */
static int enic_stop(struct net_device *netdev)
{
	struct enic *enic = netdev_priv(netdev);
	unsigned int i;
	int err;

V
Vasanthy Kolluri 已提交
1964
	for (i = 0; i < enic->intr_count; i++) {
1965
		vnic_intr_mask(&enic->intr[i]);
V
Vasanthy Kolluri 已提交
1966 1967
		(void)vnic_intr_masked(&enic->intr[i]); /* flush write */
	}
1968 1969 1970

	enic_synchronize_irqs(enic);

1971
	del_timer_sync(&enic->notify_timer);
1972
	enic_rfs_flw_tbl_free(enic);
1973

1974
	enic_dev_disable(enic);
1975

1976
	for (i = 0; i < enic->rq_count; i++)
1977 1978
		napi_disable(&enic->napi[i]);

1979
	netif_carrier_off(netdev);
1980 1981 1982
	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX)
		for (i = 0; i < enic->wq_count; i++)
			napi_disable(&enic->napi[enic_cq_wq(enic, i)]);
1983
	netif_tx_disable(netdev);
1984

1985
	if (!enic_is_dynamic(enic) && !enic_is_sriov_vf(enic))
1986
		enic_dev_del_station_addr(enic);
1987

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
	for (i = 0; i < enic->wq_count; i++) {
		err = vnic_wq_disable(&enic->wq[i]);
		if (err)
			return err;
	}
	for (i = 0; i < enic->rq_count; i++) {
		err = vnic_rq_disable(&enic->rq[i]);
		if (err)
			return err;
	}

1999
	enic_dev_notify_unset(enic);
2000
	enic_unset_affinity_hint(enic);
2001 2002
	enic_free_intr(enic);

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
	for (i = 0; i < enic->wq_count; i++)
		vnic_wq_clean(&enic->wq[i], enic_free_wq_buf);
	for (i = 0; i < enic->rq_count; i++)
		vnic_rq_clean(&enic->rq[i], enic_free_rq_buf);
	for (i = 0; i < enic->cq_count; i++)
		vnic_cq_clean(&enic->cq[i]);
	for (i = 0; i < enic->intr_count; i++)
		vnic_intr_clean(&enic->intr[i]);

	return 0;
}

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037
static int _enic_change_mtu(struct net_device *netdev, int new_mtu)
{
	bool running = netif_running(netdev);
	int err = 0;

	ASSERT_RTNL();
	if (running) {
		err = enic_stop(netdev);
		if (err)
			return err;
	}

	netdev->mtu = new_mtu;

	if (running) {
		err = enic_open(netdev);
		if (err)
			return err;
	}

	return 0;
}

2038 2039 2040 2041
static int enic_change_mtu(struct net_device *netdev, int new_mtu)
{
	struct enic *enic = netdev_priv(netdev);

2042
	if (enic_is_dynamic(enic) || enic_is_sriov_vf(enic))
2043 2044
		return -EOPNOTSUPP;

2045
	if (netdev->mtu > enic->port_mtu)
2046
		netdev_warn(netdev,
2047 2048
			    "interface MTU (%d) set higher than port MTU (%d)\n",
			    netdev->mtu, enic->port_mtu);
2049

2050
	return _enic_change_mtu(netdev, new_mtu);
2051 2052
}

2053 2054 2055 2056 2057 2058 2059
static void enic_change_mtu_work(struct work_struct *work)
{
	struct enic *enic = container_of(work, struct enic, change_mtu_work);
	struct net_device *netdev = enic->netdev;
	int new_mtu = vnic_dev_mtu(enic->vdev);

	rtnl_lock();
2060
	(void)_enic_change_mtu(netdev, new_mtu);
2061 2062 2063 2064 2065
	rtnl_unlock();

	netdev_info(netdev, "interface MTU set as %d\n", netdev->mtu);
}

2066 2067 2068 2069 2070
#ifdef CONFIG_NET_POLL_CONTROLLER
static void enic_poll_controller(struct net_device *netdev)
{
	struct enic *enic = netdev_priv(netdev);
	struct vnic_dev *vdev = enic->vdev;
2071
	unsigned int i, intr;
2072 2073 2074

	switch (vnic_dev_get_intr_mode(vdev)) {
	case VNIC_DEV_INTR_MODE_MSIX:
2075 2076
		for (i = 0; i < enic->rq_count; i++) {
			intr = enic_msix_rq_intr(enic, i);
2077 2078
			enic_isr_msix(enic->msix_entry[intr].vector,
				      &enic->napi[i]);
2079
		}
2080 2081 2082

		for (i = 0; i < enic->wq_count; i++) {
			intr = enic_msix_wq_intr(enic, i);
2083 2084
			enic_isr_msix(enic->msix_entry[intr].vector,
				      &enic->napi[enic_cq_wq(enic, i)]);
2085 2086
		}

2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137
		break;
	case VNIC_DEV_INTR_MODE_MSI:
		enic_isr_msi(enic->pdev->irq, enic);
		break;
	case VNIC_DEV_INTR_MODE_INTX:
		enic_isr_legacy(enic->pdev->irq, netdev);
		break;
	default:
		break;
	}
}
#endif

static int enic_dev_wait(struct vnic_dev *vdev,
	int (*start)(struct vnic_dev *, int),
	int (*finished)(struct vnic_dev *, int *),
	int arg)
{
	unsigned long time;
	int done;
	int err;

	BUG_ON(in_interrupt());

	err = start(vdev, arg);
	if (err)
		return err;

	/* Wait for func to complete...2 seconds max
	 */

	time = jiffies + (HZ * 2);
	do {

		err = finished(vdev, &done);
		if (err)
			return err;

		if (done)
			return 0;

		schedule_timeout_uninterruptible(HZ / 10);

	} while (time_after(time, jiffies));

	return -ETIMEDOUT;
}

static int enic_dev_open(struct enic *enic)
{
	int err;
2138
	u32 flags = CMD_OPENF_IG_DESCCACHE;
2139 2140

	err = enic_dev_wait(enic->vdev, vnic_dev_open,
2141
		vnic_dev_open_done, flags);
2142
	if (err)
2143 2144
		dev_err(enic_get_dev(enic), "vNIC device open failed, err %d\n",
			err);
2145 2146 2147 2148

	return err;
}

2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
static int enic_dev_soft_reset(struct enic *enic)
{
	int err;

	err = enic_dev_wait(enic->vdev, vnic_dev_soft_reset,
			    vnic_dev_soft_reset_done, 0);
	if (err)
		netdev_err(enic->netdev, "vNIC soft reset failed, err %d\n",
			   err);

	return err;
}

2162
static int enic_dev_hang_reset(struct enic *enic)
2163 2164 2165
{
	int err;

2166 2167
	err = enic_dev_wait(enic->vdev, vnic_dev_hang_reset,
		vnic_dev_hang_reset_done, 0);
2168
	if (err)
2169 2170
		netdev_err(enic->netdev, "vNIC hang reset failed, err %d\n",
			err);
2171 2172 2173 2174

	return err;
}

2175
int __enic_set_rsskey(struct enic *enic)
2176
{
2177
	union vnic_rss_key *rss_key_buf_va;
V
Vasanthy Kolluri 已提交
2178
	dma_addr_t rss_key_buf_pa;
2179
	int i, kidx, bidx, err;
2180

2181 2182 2183
	rss_key_buf_va = pci_zalloc_consistent(enic->pdev,
					       sizeof(union vnic_rss_key),
					       &rss_key_buf_pa);
2184 2185 2186
	if (!rss_key_buf_va)
		return -ENOMEM;

2187 2188 2189
	for (i = 0; i < ENIC_RSS_LEN; i++) {
		kidx = i / ENIC_RSS_BYTES_PER_KEY;
		bidx = i % ENIC_RSS_BYTES_PER_KEY;
2190
		rss_key_buf_va->key[kidx].b[bidx] = enic->rss_key[i];
2191
	}
2192
	spin_lock_bh(&enic->devcmd_lock);
2193 2194 2195
	err = enic_set_rss_key(enic,
		rss_key_buf_pa,
		sizeof(union vnic_rss_key));
2196
	spin_unlock_bh(&enic->devcmd_lock);
2197 2198 2199 2200 2201 2202 2203

	pci_free_consistent(enic->pdev, sizeof(union vnic_rss_key),
		rss_key_buf_va, rss_key_buf_pa);

	return err;
}

2204 2205 2206 2207 2208 2209 2210
static int enic_set_rsskey(struct enic *enic)
{
	netdev_rss_key_fill(enic->rss_key, ENIC_RSS_LEN);

	return __enic_set_rsskey(enic);
}

2211 2212
static int enic_set_rsscpu(struct enic *enic, u8 rss_hash_bits)
{
V
Vasanthy Kolluri 已提交
2213
	dma_addr_t rss_cpu_buf_pa;
2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225
	union vnic_rss_cpu *rss_cpu_buf_va = NULL;
	unsigned int i;
	int err;

	rss_cpu_buf_va = pci_alloc_consistent(enic->pdev,
		sizeof(union vnic_rss_cpu), &rss_cpu_buf_pa);
	if (!rss_cpu_buf_va)
		return -ENOMEM;

	for (i = 0; i < (1 << rss_hash_bits); i++)
		(*rss_cpu_buf_va).cpu[i/4].b[i%4] = i % enic->rq_count;

2226
	spin_lock_bh(&enic->devcmd_lock);
2227 2228 2229
	err = enic_set_rss_cpu(enic,
		rss_cpu_buf_pa,
		sizeof(union vnic_rss_cpu));
2230
	spin_unlock_bh(&enic->devcmd_lock);
2231 2232 2233 2234 2235 2236 2237 2238 2239

	pci_free_consistent(enic->pdev, sizeof(union vnic_rss_cpu),
		rss_cpu_buf_va, rss_cpu_buf_pa);

	return err;
}

static int enic_set_niccfg(struct enic *enic, u8 rss_default_cpu,
	u8 rss_hash_type, u8 rss_hash_bits, u8 rss_base_cpu, u8 rss_enable)
2240 2241 2242
{
	const u8 tso_ipid_split_en = 0;
	const u8 ig_vlan_strip_en = 1;
2243
	int err;
2244

2245 2246
	/* Enable VLAN tag stripping.
	*/
2247

2248
	spin_lock_bh(&enic->devcmd_lock);
2249
	err = enic_set_nic_cfg(enic,
2250 2251 2252 2253
		rss_default_cpu, rss_hash_type,
		rss_hash_bits, rss_base_cpu,
		rss_enable, tso_ipid_split_en,
		ig_vlan_strip_en);
2254
	spin_unlock_bh(&enic->devcmd_lock);
2255 2256 2257 2258

	return err;
}

2259 2260 2261 2262 2263 2264
static int enic_set_rss_nic_cfg(struct enic *enic)
{
	struct device *dev = enic_get_dev(enic);
	const u8 rss_default_cpu = 0;
	const u8 rss_hash_bits = 7;
	const u8 rss_base_cpu = 0;
2265 2266
	u8 rss_hash_type;
	int res;
2267 2268
	u8 rss_enable = ENIC_SETTING(enic, RSS) && (enic->rq_count > 1);

2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280
	spin_lock_bh(&enic->devcmd_lock);
	res = vnic_dev_capable_rss_hash_type(enic->vdev, &rss_hash_type);
	spin_unlock_bh(&enic->devcmd_lock);
	if (res) {
		/* defaults for old adapters
		 */
		rss_hash_type = NIC_CFG_RSS_HASH_TYPE_IPV4	|
				NIC_CFG_RSS_HASH_TYPE_TCP_IPV4	|
				NIC_CFG_RSS_HASH_TYPE_IPV6	|
				NIC_CFG_RSS_HASH_TYPE_TCP_IPV6;
	}

2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295
	if (rss_enable) {
		if (!enic_set_rsskey(enic)) {
			if (enic_set_rsscpu(enic, rss_hash_bits)) {
				rss_enable = 0;
				dev_warn(dev, "RSS disabled, "
					"Failed to set RSS cpu indirection table.");
			}
		} else {
			rss_enable = 0;
			dev_warn(dev, "RSS disabled, Failed to set RSS key.\n");
		}
	}

	return enic_set_niccfg(enic, rss_default_cpu, rss_hash_type,
		rss_hash_bits, rss_base_cpu, rss_enable);
2296 2297
}

2298 2299 2300 2301 2302 2303 2304 2305 2306
static void enic_reset(struct work_struct *work)
{
	struct enic *enic = container_of(work, struct enic, reset);

	if (!netif_running(enic->netdev))
		return;

	rtnl_lock();

2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326
	spin_lock(&enic->enic_api_lock);
	enic_stop(enic->netdev);
	enic_dev_soft_reset(enic);
	enic_reset_addr_lists(enic);
	enic_init_vnic_resources(enic);
	enic_set_rss_nic_cfg(enic);
	enic_dev_set_ig_vlan_rewrite_mode(enic);
	enic_open(enic->netdev);
	spin_unlock(&enic->enic_api_lock);
	call_netdevice_notifiers(NETDEV_REBOOT, enic->netdev);

	rtnl_unlock();
}

static void enic_tx_hang_reset(struct work_struct *work)
{
	struct enic *enic = container_of(work, struct enic, tx_hang_reset);

	rtnl_lock();

2327
	spin_lock(&enic->enic_api_lock);
2328
	enic_dev_hang_notify(enic);
2329
	enic_stop(enic->netdev);
2330
	enic_dev_hang_reset(enic);
2331
	enic_reset_addr_lists(enic);
2332
	enic_init_vnic_resources(enic);
2333
	enic_set_rss_nic_cfg(enic);
2334
	enic_dev_set_ig_vlan_rewrite_mode(enic);
2335
	enic_open(enic->netdev);
2336
	spin_unlock(&enic->enic_api_lock);
2337
	call_netdevice_notifiers(NETDEV_REBOOT, enic->netdev);
2338 2339 2340 2341 2342 2343

	rtnl_unlock();
}

static int enic_set_intr_mode(struct enic *enic)
{
2344
	unsigned int n = min_t(unsigned int, enic->rq_count, ENIC_RQ_MAX);
2345
	unsigned int m = min_t(unsigned int, enic->wq_count, ENIC_WQ_MAX);
2346 2347 2348
	unsigned int i;

	/* Set interrupt mode (INTx, MSI, MSI-X) depending
2349
	 * on system capabilities.
2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361
	 *
	 * Try MSI-X first
	 *
	 * We need n RQs, m WQs, n+m CQs, and n+m+2 INTRs
	 * (the second to last INTR is used for WQ/RQ errors)
	 * (the last INTR is used for notifications)
	 */

	BUG_ON(ARRAY_SIZE(enic->msix_entry) < n + m + 2);
	for (i = 0; i < n + m + 2; i++)
		enic->msix_entry[i].entry = i;

2362 2363 2364 2365 2366
	/* Use multiple RQs if RSS is enabled
	 */

	if (ENIC_SETTING(enic, RSS) &&
	    enic->config.intr_mode < 1 &&
2367 2368 2369
	    enic->rq_count >= n &&
	    enic->wq_count >= m &&
	    enic->cq_count >= n + m &&
2370
	    enic->intr_count >= n + m + 2) {
2371

2372 2373
		if (pci_enable_msix_range(enic->pdev, enic->msix_entry,
					  n + m + 2, n + m + 2) > 0) {
2374

2375 2376 2377 2378
			enic->rq_count = n;
			enic->wq_count = m;
			enic->cq_count = n + m;
			enic->intr_count = n + m + 2;
2379

2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391
			vnic_dev_set_intr_mode(enic->vdev,
				VNIC_DEV_INTR_MODE_MSIX);

			return 0;
		}
	}

	if (enic->config.intr_mode < 1 &&
	    enic->rq_count >= 1 &&
	    enic->wq_count >= m &&
	    enic->cq_count >= 1 + m &&
	    enic->intr_count >= 1 + m + 2) {
2392 2393
		if (pci_enable_msix_range(enic->pdev, enic->msix_entry,
					  1 + m + 2, 1 + m + 2) > 0) {
2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404

			enic->rq_count = 1;
			enic->wq_count = m;
			enic->cq_count = 1 + m;
			enic->intr_count = 1 + m + 2;

			vnic_dev_set_intr_mode(enic->vdev,
				VNIC_DEV_INTR_MODE_MSIX);

			return 0;
		}
2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473
	}

	/* Next try MSI
	 *
	 * We need 1 RQ, 1 WQ, 2 CQs, and 1 INTR
	 */

	if (enic->config.intr_mode < 2 &&
	    enic->rq_count >= 1 &&
	    enic->wq_count >= 1 &&
	    enic->cq_count >= 2 &&
	    enic->intr_count >= 1 &&
	    !pci_enable_msi(enic->pdev)) {

		enic->rq_count = 1;
		enic->wq_count = 1;
		enic->cq_count = 2;
		enic->intr_count = 1;

		vnic_dev_set_intr_mode(enic->vdev, VNIC_DEV_INTR_MODE_MSI);

		return 0;
	}

	/* Next try INTx
	 *
	 * We need 1 RQ, 1 WQ, 2 CQs, and 3 INTRs
	 * (the first INTR is used for WQ/RQ)
	 * (the second INTR is used for WQ/RQ errors)
	 * (the last INTR is used for notifications)
	 */

	if (enic->config.intr_mode < 3 &&
	    enic->rq_count >= 1 &&
	    enic->wq_count >= 1 &&
	    enic->cq_count >= 2 &&
	    enic->intr_count >= 3) {

		enic->rq_count = 1;
		enic->wq_count = 1;
		enic->cq_count = 2;
		enic->intr_count = 3;

		vnic_dev_set_intr_mode(enic->vdev, VNIC_DEV_INTR_MODE_INTX);

		return 0;
	}

	vnic_dev_set_intr_mode(enic->vdev, VNIC_DEV_INTR_MODE_UNKNOWN);

	return -EINVAL;
}

static void enic_clear_intr_mode(struct enic *enic)
{
	switch (vnic_dev_get_intr_mode(enic->vdev)) {
	case VNIC_DEV_INTR_MODE_MSIX:
		pci_disable_msix(enic->pdev);
		break;
	case VNIC_DEV_INTR_MODE_MSI:
		pci_disable_msi(enic->pdev);
		break;
	default:
		break;
	}

	vnic_dev_set_intr_mode(enic->vdev, VNIC_DEV_INTR_MODE_UNKNOWN);
}

2474 2475 2476 2477
static const struct net_device_ops enic_netdev_dynamic_ops = {
	.ndo_open		= enic_open,
	.ndo_stop		= enic_stop,
	.ndo_start_xmit		= enic_hard_start_xmit,
2478
	.ndo_get_stats64	= enic_get_stats,
2479
	.ndo_validate_addr	= eth_validate_addr,
2480
	.ndo_set_rx_mode	= enic_set_rx_mode,
2481 2482 2483 2484 2485 2486 2487
	.ndo_set_mac_address	= enic_set_mac_address_dynamic,
	.ndo_change_mtu		= enic_change_mtu,
	.ndo_vlan_rx_add_vid	= enic_vlan_rx_add_vid,
	.ndo_vlan_rx_kill_vid	= enic_vlan_rx_kill_vid,
	.ndo_tx_timeout		= enic_tx_timeout,
	.ndo_set_vf_port	= enic_set_vf_port,
	.ndo_get_vf_port	= enic_get_vf_port,
2488
	.ndo_set_vf_mac		= enic_set_vf_mac,
2489 2490 2491
#ifdef CONFIG_NET_POLL_CONTROLLER
	.ndo_poll_controller	= enic_poll_controller,
#endif
2492 2493 2494
#ifdef CONFIG_RFS_ACCEL
	.ndo_rx_flow_steer	= enic_rx_flow_steer,
#endif
2495 2496
	.ndo_udp_tunnel_add	= udp_tunnel_nic_add_port,
	.ndo_udp_tunnel_del	= udp_tunnel_nic_del_port,
2497
	.ndo_features_check	= enic_features_check,
2498 2499
};

2500 2501 2502
static const struct net_device_ops enic_netdev_ops = {
	.ndo_open		= enic_open,
	.ndo_stop		= enic_stop,
2503
	.ndo_start_xmit		= enic_hard_start_xmit,
2504
	.ndo_get_stats64	= enic_get_stats,
2505
	.ndo_validate_addr	= eth_validate_addr,
2506
	.ndo_set_mac_address	= enic_set_mac_address,
2507
	.ndo_set_rx_mode	= enic_set_rx_mode,
2508 2509 2510 2511
	.ndo_change_mtu		= enic_change_mtu,
	.ndo_vlan_rx_add_vid	= enic_vlan_rx_add_vid,
	.ndo_vlan_rx_kill_vid	= enic_vlan_rx_kill_vid,
	.ndo_tx_timeout		= enic_tx_timeout,
2512 2513 2514
	.ndo_set_vf_port	= enic_set_vf_port,
	.ndo_get_vf_port	= enic_get_vf_port,
	.ndo_set_vf_mac		= enic_set_vf_mac,
2515 2516 2517
#ifdef CONFIG_NET_POLL_CONTROLLER
	.ndo_poll_controller	= enic_poll_controller,
#endif
2518 2519 2520
#ifdef CONFIG_RFS_ACCEL
	.ndo_rx_flow_steer	= enic_rx_flow_steer,
#endif
2521 2522
	.ndo_udp_tunnel_add	= udp_tunnel_nic_add_port,
	.ndo_udp_tunnel_del	= udp_tunnel_nic_del_port,
2523
	.ndo_features_check	= enic_features_check,
2524 2525
};

2526
static void enic_dev_deinit(struct enic *enic)
2527
{
2528 2529
	unsigned int i;

2530 2531
	for (i = 0; i < enic->rq_count; i++) {
		napi_hash_del(&enic->napi[i]);
2532
		netif_napi_del(&enic->napi[i]);
2533
	}
2534 2535 2536
	if (vnic_dev_get_intr_mode(enic->vdev) == VNIC_DEV_INTR_MODE_MSIX)
		for (i = 0; i < enic->wq_count; i++)
			netif_napi_del(&enic->napi[enic_cq_wq(enic, i)]);
2537

2538 2539
	enic_free_vnic_resources(enic);
	enic_clear_intr_mode(enic);
2540
	enic_free_affinity_hint(enic);
2541 2542
}

2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554
static void enic_kdump_kernel_config(struct enic *enic)
{
	if (is_kdump_kernel()) {
		dev_info(enic_get_dev(enic), "Running from within kdump kernel. Using minimal resources\n");
		enic->rq_count = 1;
		enic->wq_count = 1;
		enic->config.rq_desc_count = ENIC_MIN_RQ_DESCS;
		enic->config.wq_desc_count = ENIC_MIN_WQ_DESCS;
		enic->config.mtu = min_t(u16, 1500, enic->config.mtu);
	}
}

2555
static int enic_dev_init(struct enic *enic)
2556
{
2557
	struct device *dev = enic_get_dev(enic);
2558
	struct net_device *netdev = enic->netdev;
2559
	unsigned int i;
2560 2561
	int err;

2562 2563 2564 2565 2566 2567 2568 2569
	/* Get interrupt coalesce timer info */
	err = enic_dev_intr_coal_timer_info(enic);
	if (err) {
		dev_warn(dev, "Using default conversion factor for "
			"interrupt coalesce timer\n");
		vnic_dev_intr_coal_timer_info_default(enic->vdev);
	}

2570 2571 2572 2573 2574
	/* Get vNIC configuration
	 */

	err = enic_get_vnic_config(enic);
	if (err) {
2575
		dev_err(dev, "Get vNIC configuration failed, aborting\n");
2576 2577 2578 2579 2580 2581 2582 2583
		return err;
	}

	/* Get available resource counts
	 */

	enic_get_res_counts(enic);

2584 2585 2586 2587
	/* modify resource count if we are in kdump_kernel
	 */
	enic_kdump_kernel_config(enic);

2588 2589 2590 2591 2592 2593
	/* Set interrupt mode based on resource counts and system
	 * capabilities
	 */

	err = enic_set_intr_mode(enic);
	if (err) {
2594 2595
		dev_err(dev, "Failed to set intr mode based on resource "
			"counts and system capabilities, aborting\n");
2596 2597 2598 2599 2600 2601 2602 2603
		return err;
	}

	/* Allocate and configure vNIC resources
	 */

	err = enic_alloc_vnic_resources(enic);
	if (err) {
2604
		dev_err(dev, "Failed to alloc vNIC resources, aborting\n");
2605 2606 2607 2608 2609
		goto err_out_free_vnic_resources;
	}

	enic_init_vnic_resources(enic);

2610
	err = enic_set_rss_nic_cfg(enic);
2611
	if (err) {
2612
		dev_err(dev, "Failed to config nic, aborting\n");
2613 2614 2615 2616 2617
		goto err_out_free_vnic_resources;
	}

	switch (vnic_dev_get_intr_mode(enic->vdev)) {
	default:
2618
		netif_napi_add(netdev, &enic->napi[0], enic_poll, 64);
2619 2620
		break;
	case VNIC_DEV_INTR_MODE_MSIX:
2621
		for (i = 0; i < enic->rq_count; i++) {
2622
			netif_napi_add(netdev, &enic->napi[i],
2623
				enic_poll_msix_rq, NAPI_POLL_WEIGHT);
2624
		}
2625 2626 2627
		for (i = 0; i < enic->wq_count; i++)
			netif_napi_add(netdev, &enic->napi[enic_cq_wq(enic, i)],
				       enic_poll_msix_wq, NAPI_POLL_WEIGHT);
2628 2629 2630 2631 2632 2633
		break;
	}

	return 0;

err_out_free_vnic_resources:
2634
	enic_free_affinity_hint(enic);
2635 2636 2637 2638 2639 2640
	enic_clear_intr_mode(enic);
	enic_free_vnic_resources(enic);

	return err;
}

2641 2642 2643 2644 2645 2646 2647 2648 2649
static void enic_iounmap(struct enic *enic)
{
	unsigned int i;

	for (i = 0; i < ARRAY_SIZE(enic->bar); i++)
		if (enic->bar[i].vaddr)
			iounmap(enic->bar[i].vaddr);
}

2650
static int enic_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
2651
{
2652
	struct device *dev = &pdev->dev;
2653 2654 2655 2656 2657
	struct net_device *netdev;
	struct enic *enic;
	int using_dac = 0;
	unsigned int i;
	int err;
R
Roopa Prabhu 已提交
2658 2659 2660
#ifdef CONFIG_PCI_IOV
	int pos = 0;
#endif
2661
	int num_pps = 1;
2662 2663 2664 2665 2666

	/* Allocate net device structure and initialize.  Private
	 * instance data is initialized to zero.
	 */

2667 2668
	netdev = alloc_etherdev_mqs(sizeof(struct enic),
				    ENIC_RQ_MAX, ENIC_WQ_MAX);
2669
	if (!netdev)
2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682
		return -ENOMEM;

	pci_set_drvdata(pdev, netdev);

	SET_NETDEV_DEV(netdev, &pdev->dev);

	enic = netdev_priv(netdev);
	enic->netdev = netdev;
	enic->pdev = pdev;

	/* Setup PCI resources
	 */

V
Vasanthy Kolluri 已提交
2683
	err = pci_enable_device_mem(pdev);
2684
	if (err) {
2685
		dev_err(dev, "Cannot enable PCI device, aborting\n");
2686 2687 2688 2689 2690
		goto err_out_free_netdev;
	}

	err = pci_request_regions(pdev, DRV_NAME);
	if (err) {
2691
		dev_err(dev, "Cannot request PCI regions, aborting\n");
2692 2693 2694 2695 2696 2697
		goto err_out_disable_device;
	}

	pci_set_master(pdev);

	/* Query PCI controller on system for DMA addressing
2698
	 * limitation for the device.  Try 47-bit first, and
2699 2700 2701
	 * fail to 32-bit.
	 */

2702
	err = pci_set_dma_mask(pdev, DMA_BIT_MASK(47));
2703
	if (err) {
2704
		err = pci_set_dma_mask(pdev, DMA_BIT_MASK(32));
2705
		if (err) {
2706
			dev_err(dev, "No usable DMA configuration, aborting\n");
2707 2708
			goto err_out_release_regions;
		}
2709
		err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32));
2710
		if (err) {
2711 2712
			dev_err(dev, "Unable to obtain %u-bit DMA "
				"for consistent allocations, aborting\n", 32);
2713 2714 2715
			goto err_out_release_regions;
		}
	} else {
2716
		err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(47));
2717
		if (err) {
2718
			dev_err(dev, "Unable to obtain %u-bit DMA "
2719
				"for consistent allocations, aborting\n", 47);
2720 2721 2722 2723 2724
			goto err_out_release_regions;
		}
		using_dac = 1;
	}

2725
	/* Map vNIC resources from BAR0-5
2726 2727
	 */

2728 2729 2730 2731 2732 2733
	for (i = 0; i < ARRAY_SIZE(enic->bar); i++) {
		if (!(pci_resource_flags(pdev, i) & IORESOURCE_MEM))
			continue;
		enic->bar[i].len = pci_resource_len(pdev, i);
		enic->bar[i].vaddr = pci_iomap(pdev, i, enic->bar[i].len);
		if (!enic->bar[i].vaddr) {
2734
			dev_err(dev, "Cannot memory-map BAR %d, aborting\n", i);
2735 2736 2737 2738
			err = -ENODEV;
			goto err_out_iounmap;
		}
		enic->bar[i].bus_addr = pci_resource_start(pdev, i);
2739 2740 2741 2742 2743
	}

	/* Register vNIC device
	 */

2744 2745
	enic->vdev = vnic_dev_register(NULL, enic, pdev, enic->bar,
		ARRAY_SIZE(enic->bar));
2746
	if (!enic->vdev) {
2747
		dev_err(dev, "vNIC registration failed, aborting\n");
2748 2749 2750 2751
		err = -ENODEV;
		goto err_out_iounmap;
	}

G
Govindarajulu Varadarajan 已提交
2752 2753 2754 2755 2756
	err = vnic_devcmd_init(enic->vdev);

	if (err)
		goto err_out_vnic_unregister;

R
Roopa Prabhu 已提交
2757 2758 2759 2760 2761
#ifdef CONFIG_PCI_IOV
	/* Get number of subvnics */
	pos = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_SRIOV);
	if (pos) {
		pci_read_config_word(pdev, pos + PCI_SRIOV_TOTAL_VF,
2762
			&enic->num_vfs);
R
Roopa Prabhu 已提交
2763 2764 2765 2766 2767 2768 2769 2770 2771
		if (enic->num_vfs) {
			err = pci_enable_sriov(pdev, enic->num_vfs);
			if (err) {
				dev_err(dev, "SRIOV enable failed, aborting."
					" pci_enable_sriov() returned %d\n",
					err);
				goto err_out_vnic_unregister;
			}
			enic->priv_flags |= ENIC_SRIOV_ENABLED;
2772
			num_pps = enic->num_vfs;
R
Roopa Prabhu 已提交
2773 2774 2775
		}
	}
#endif
2776

2777
	/* Allocate structure for port profiles */
2778
	enic->pp = kcalloc(num_pps, sizeof(*enic->pp), GFP_KERNEL);
2779 2780
	if (!enic->pp) {
		err = -ENOMEM;
2781
		goto err_out_disable_sriov_pp;
2782 2783
	}

2784 2785 2786 2787 2788
	/* Issue device open to get device in known state
	 */

	err = enic_dev_open(enic);
	if (err) {
2789
		dev_err(dev, "vNIC dev open failed, aborting\n");
2790
		goto err_out_disable_sriov;
2791 2792
	}

2793 2794 2795 2796
	/* Setup devcmd lock
	 */

	spin_lock_init(&enic->devcmd_lock);
2797
	spin_lock_init(&enic->enic_api_lock);
2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809

	/*
	 * Set ingress vlan rewrite mode before vnic initialization
	 */

	err = enic_dev_set_ig_vlan_rewrite_mode(enic);
	if (err) {
		dev_err(dev,
			"Failed to set ingress vlan rewrite mode, aborting.\n");
		goto err_out_dev_close;
	}

2810 2811 2812 2813 2814 2815 2816 2817 2818 2819
	/* Issue device init to initialize the vnic-to-switch link.
	 * We'll start with carrier off and wait for link UP
	 * notification later to turn on carrier.  We don't need
	 * to wait here for the vnic-to-switch link initialization
	 * to complete; link UP notification is the indication that
	 * the process is complete.
	 */

	netif_carrier_off(netdev);

2820 2821 2822 2823 2824
	/* Do not call dev_init for a dynamic vnic.
	 * For a dynamic vnic, init_prov_info will be
	 * called later by an upper layer.
	 */

2825
	if (!enic_is_dynamic(enic)) {
2826 2827
		err = vnic_dev_init(enic->vdev, 0);
		if (err) {
2828
			dev_err(dev, "vNIC dev init failed, aborting\n");
2829 2830
			goto err_out_dev_close;
		}
2831 2832
	}

2833
	err = enic_dev_init(enic);
2834
	if (err) {
2835
		dev_err(dev, "Device initialization failed, aborting\n");
2836 2837 2838
		goto err_out_dev_close;
	}

2839
	netif_set_real_num_tx_queues(netdev, enic->wq_count);
2840
	netif_set_real_num_rx_queues(netdev, enic->rq_count);
2841

2842
	/* Setup notification timer, HW reset task, and wq locks
2843 2844
	 */

2845
	timer_setup(&enic->notify_timer, enic_notify_timer, 0);
2846

2847
	enic_rfs_flw_tbl_init(enic);
2848
	enic_set_rx_coal_setting(enic);
2849
	INIT_WORK(&enic->reset, enic_reset);
2850
	INIT_WORK(&enic->tx_hang_reset, enic_tx_hang_reset);
2851
	INIT_WORK(&enic->change_mtu_work, enic_change_mtu_work);
2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862

	for (i = 0; i < enic->wq_count; i++)
		spin_lock_init(&enic->wq_lock[i]);

	/* Register net device
	 */

	enic->port_mtu = enic->config.mtu;

	err = enic_set_mac_addr(netdev, enic->mac_addr);
	if (err) {
2863
		dev_err(dev, "Invalid MAC address, aborting\n");
2864
		goto err_out_dev_deinit;
2865 2866
	}

2867
	enic->tx_coalesce_usecs = enic->config.intr_timer_usec;
2868 2869 2870
	/* rx coalesce time already got initialized. This gets used
	 * if adaptive coal is turned off
	 */
2871 2872
	enic->rx_coalesce_usecs = enic->tx_coalesce_usecs;

2873
	if (enic_is_dynamic(enic) || enic_is_sriov_vf(enic))
2874 2875 2876 2877
		netdev->netdev_ops = &enic_netdev_dynamic_ops;
	else
		netdev->netdev_ops = &enic_netdev_ops;

2878
	netdev->watchdog_timeo = 2 * HZ;
2879
	enic_set_ethtool_ops(netdev);
2880

2881
	netdev->features |= NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX;
2882
	if (ENIC_SETTING(enic, LOOP)) {
2883
		netdev->features &= ~NETIF_F_HW_VLAN_CTAG_TX;
2884 2885 2886 2887
		enic->loop_enable = 1;
		enic->loop_tag = enic->config.loop_tag;
		dev_info(dev, "loopback tag=0x%04x\n", enic->loop_tag);
	}
2888
	if (ENIC_SETTING(enic, TXCSUM))
2889
		netdev->hw_features |= NETIF_F_SG | NETIF_F_HW_CSUM;
2890
	if (ENIC_SETTING(enic, TSO))
2891
		netdev->hw_features |= NETIF_F_TSO |
2892
			NETIF_F_TSO6 | NETIF_F_TSO_ECN;
2893 2894
	if (ENIC_SETTING(enic, RSS))
		netdev->hw_features |= NETIF_F_RXHASH;
2895 2896
	if (ENIC_SETTING(enic, RXCSUM))
		netdev->hw_features |= NETIF_F_RXCSUM;
2897 2898
	if (ENIC_SETTING(enic, VXLAN)) {
		u64 patch_level;
2899
		u64 a1 = 0;
2900 2901 2902

		netdev->hw_enc_features |= NETIF_F_RXCSUM		|
					   NETIF_F_TSO			|
2903
					   NETIF_F_TSO6			|
2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921
					   NETIF_F_TSO_ECN		|
					   NETIF_F_GSO_UDP_TUNNEL	|
					   NETIF_F_HW_CSUM		|
					   NETIF_F_GSO_UDP_TUNNEL_CSUM;
		netdev->hw_features |= netdev->hw_enc_features;
		/* get bit mask from hw about supported offload bit level
		 * BIT(0) = fw supports patch_level 0
		 *	    fcoe bit = encap
		 *	    fcoe_fc_crc_ok = outer csum ok
		 * BIT(1) = always set by fw
		 * BIT(2) = fw supports patch_level 2
		 *	    BIT(0) in rss_hash = encap
		 *	    BIT(1,2) in rss_hash = outer_ip_csum_ok/
		 *				   outer_tcp_csum_ok
		 * used in enic_rq_indicate_buf
		 */
		err = vnic_dev_get_supported_feature_ver(enic->vdev,
							 VIC_FEATURE_VXLAN,
2922
							 &patch_level, &a1);
2923 2924
		if (err)
			patch_level = 0;
2925
		enic->vxlan.flags = (u8)a1;
2926 2927 2928 2929 2930 2931
		/* mask bits that are supported by driver
		 */
		patch_level &= BIT_ULL(0) | BIT_ULL(2);
		patch_level = fls(patch_level);
		patch_level = patch_level ? patch_level - 1 : 0;
		enic->vxlan.patch_level = patch_level;
2932 2933 2934 2935 2936 2937 2938

		if (vnic_dev_get_res_count(enic->vdev, RES_TYPE_WQ) == 1 ||
		    enic->vxlan.flags & ENIC_VXLAN_MULTI_WQ) {
			netdev->udp_tunnel_nic_info = &enic_udp_tunnels_v4;
			if (enic->vxlan.flags & ENIC_VXLAN_OUTER_IPV6)
				netdev->udp_tunnel_nic_info = &enic_udp_tunnels;
		}
2939
	}
2940 2941

	netdev->features |= netdev->hw_features;
2942
	netdev->vlan_features |= netdev->features;
2943

2944 2945 2946 2947
#ifdef CONFIG_RFS_ACCEL
	netdev->hw_features |= NETIF_F_NTUPLE;
#endif

2948 2949 2950
	if (using_dac)
		netdev->features |= NETIF_F_HIGHDMA;

2951 2952
	netdev->priv_flags |= IFF_UNICAST_FLT;

2953 2954 2955
	/* MTU range: 68 - 9000 */
	netdev->min_mtu = ENIC_MIN_MTU;
	netdev->max_mtu = ENIC_MAX_MTU;
2956
	netdev->mtu	= enic->port_mtu;
2957

2958 2959
	err = register_netdev(netdev);
	if (err) {
2960
		dev_err(dev, "Cannot register net device, aborting\n");
2961
		goto err_out_dev_deinit;
2962
	}
2963
	enic->rx_copybreak = RX_COPYBREAK_DEFAULT;
2964 2965 2966

	return 0;

2967 2968
err_out_dev_deinit:
	enic_dev_deinit(enic);
2969 2970
err_out_dev_close:
	vnic_dev_close(enic->vdev);
R
Roopa Prabhu 已提交
2971
err_out_disable_sriov:
2972 2973
	kfree(enic->pp);
err_out_disable_sriov_pp:
R
Roopa Prabhu 已提交
2974 2975 2976 2977 2978 2979
#ifdef CONFIG_PCI_IOV
	if (enic_sriov_enabled(enic)) {
		pci_disable_sriov(pdev);
		enic->priv_flags &= ~ENIC_SRIOV_ENABLED;
	}
#endif
2980
err_out_vnic_unregister:
2981
	vnic_dev_unregister(enic->vdev);
2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993
err_out_iounmap:
	enic_iounmap(enic);
err_out_release_regions:
	pci_release_regions(pdev);
err_out_disable_device:
	pci_disable_device(pdev);
err_out_free_netdev:
	free_netdev(netdev);

	return err;
}

B
Bill Pemberton 已提交
2994
static void enic_remove(struct pci_dev *pdev)
2995 2996 2997 2998 2999 3000
{
	struct net_device *netdev = pci_get_drvdata(pdev);

	if (netdev) {
		struct enic *enic = netdev_priv(netdev);

3001
		cancel_work_sync(&enic->reset);
3002
		cancel_work_sync(&enic->change_mtu_work);
3003
		unregister_netdev(netdev);
3004
		enic_dev_deinit(enic);
3005
		vnic_dev_close(enic->vdev);
R
Roopa Prabhu 已提交
3006 3007 3008 3009 3010 3011
#ifdef CONFIG_PCI_IOV
		if (enic_sriov_enabled(enic)) {
			pci_disable_sriov(pdev);
			enic->priv_flags &= ~ENIC_SRIOV_ENABLED;
		}
#endif
3012
		kfree(enic->pp);
3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024
		vnic_dev_unregister(enic->vdev);
		enic_iounmap(enic);
		pci_release_regions(pdev);
		pci_disable_device(pdev);
		free_netdev(netdev);
	}
}

static struct pci_driver enic_driver = {
	.name = DRV_NAME,
	.id_table = enic_id_table,
	.probe = enic_probe,
B
Bill Pemberton 已提交
3025
	.remove = enic_remove,
3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039
};

static int __init enic_init_module(void)
{
	return pci_register_driver(&enic_driver);
}

static void __exit enic_cleanup_module(void)
{
	pci_unregister_driver(&enic_driver);
}

module_init(enic_init_module);
module_exit(enic_cleanup_module);