summaryrefslogtreecommitdiffstats
path: root/drivers/net/virtio/virtio_user/vhost_kernel.c
blob: 8d0a1ab237b687f0bb85dc5b60d2e3ede0bdd432 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
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
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
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
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
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
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
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
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
/* SPDX-License-Identifier: BSD-3-Clause
 * Copyright(c) 2016 Intel Corporation
 */

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>

#include <rte_memory.h>
#include <rte_eal_memconfig.h>

#include "vhost.h"
#include "virtio_user_dev.h"
#include "vhost_kernel_tap.h"

struct vhost_memory_kernel {
	uint32_t nregions;
	uint32_t padding;
	struct vhost_memory_region regions[0];
};

/* vhost kernel ioctls */
#define VHOST_VIRTIO 0xAF
#define VHOST_GET_FEATURES _IOR(VHOST_VIRTIO, 0x00, __u64)
#define VHOST_SET_FEATURES _IOW(VHOST_VIRTIO, 0x00, __u64)
#define VHOST_SET_OWNER _IO(VHOST_VIRTIO, 0x01)
#define VHOST_RESET_OWNER _IO(VHOST_VIRTIO, 0x02)
#define VHOST_SET_MEM_TABLE _IOW(VHOST_VIRTIO, 0x03, struct vhost_memory_kernel)
#define VHOST_SET_LOG_BASE _IOW(VHOST_VIRTIO, 0x04, __u64)
#define VHOST_SET_LOG_FD _IOW(VHOST_VIRTIO, 0x07, int)
#define VHOST_SET_VRING_NUM _IOW(VHOST_VIRTIO, 0x10, struct vhost_vring_state)
#define VHOST_SET_VRING_ADDR _IOW(VHOST_VIRTIO, 0x11, struct vhost_vring_addr)
#define VHOST_SET_VRING_BASE _IOW(VHOST_VIRTIO, 0x12, struct vhost_vring_state)
#define VHOST_GET_VRING_BASE _IOWR(VHOST_VIRTIO, 0x12, struct vhost_vring_state)
#define VHOST_SET_VRING_KICK _IOW(VHOST_VIRTIO, 0x20, struct vhost_vring_file)
#define VHOST_SET_VRING_CALL _IOW(VHOST_VIRTIO, 0x21, struct vhost_vring_file)
#define VHOST_SET_VRING_ERR _IOW(VHOST_VIRTIO, 0x22, struct vhost_vring_file)
#define VHOST_NET_SET_BACKEND _IOW(VHOST_VIRTIO, 0x30, struct vhost_vring_file)

static uint64_t max_regions = 64;

static void
get_vhost_kernel_max_regions(void)
{
	int fd;
	char buf[20] = {'\0'};

	fd = open("/sys/module/vhost/parameters/max_mem_regions", O_RDONLY);
	if (fd < 0)
		return;

	if (read(fd, buf, sizeof(buf) - 1) > 0)
		max_regions = strtoull(buf, NULL, 10);

	close(fd);
}

static uint64_t vhost_req_user_to_kernel[] = {
	[VHOST_USER_SET_OWNER] = VHOST_SET_OWNER,
	[VHOST_USER_RESET_OWNER] = VHOST_RESET_OWNER,
	[VHOST_USER_SET_FEATURES] = VHOST_SET_FEATURES,
	[VHOST_USER_GET_FEATURES] = VHOST_GET_FEATURES,
	[VHOST_USER_SET_VRING_CALL] = VHOST_SET_VRING_CALL,
	[VHOST_USER_SET_VRING_NUM] = VHOST_SET_VRING_NUM,
	[VHOST_USER_SET_VRING_BASE] = VHOST_SET_VRING_BASE,
	[VHOST_USER_GET_VRING_BASE] = VHOST_GET_VRING_BASE,
	[VHOST_USER_SET_VRING_ADDR] = VHOST_SET_VRING_ADDR,
	[VHOST_USER_SET_VRING_KICK] = VHOST_SET_VRING_KICK,
	[VHOST_USER_SET_MEM_TABLE] = VHOST_SET_MEM_TABLE,
};

/* By default, vhost kernel module allows 64 regions, but DPDK allows
 * 256 segments. As a relief, below function merges those virtually
 * adjacent memsegs into one region.
 */
static struct vhost_memory_kernel *
prepare_vhost_memory_kernel(void)
{
	uint32_t i, j, k = 0;
	struct rte_memseg *seg;
	struct vhost_memory_region *mr;
	struct vhost_memory_kernel *vm;

	vm = malloc(sizeof(struct vhost_memory_kernel) +
		    max_regions *
		    sizeof(struct vhost_memory_region));
	if (!vm)
		return NULL;

	for (i = 0; i < RTE_MAX_MEMSEG; ++i) {
		seg = &rte_eal_get_configuration()->mem_config->memseg[i];
		if (!seg->addr)
			break;

		int new_region = 1;

		for (j = 0; j < k; ++j) {
			mr = &vm->regions[j];

			if (mr->userspace_addr + mr->memory_size ==
			    (uint64_t)(uintptr_t)seg->addr) {
				mr->memory_size += seg->len;
				new_region = 0;
				break;
			}

			if ((uint64_t)(uintptr_t)seg->addr + seg->len ==
			    mr->userspace_addr) {
				mr->guest_phys_addr =
					(uint64_t)(uintptr_t)seg->addr;
				mr->userspace_addr =
					(uint64_t)(uintptr_t)seg->addr;
				mr->memory_size += seg->len;
				new_region = 0;
				break;
			}
		}

		if (new_region == 0)
			continue;

		mr = &vm->regions[k++];
		/* use vaddr here! */
		mr->guest_phys_addr = (uint64_t)(uintptr_t)seg->addr;
		mr->userspace_addr = (uint64_t)(uintptr_t)seg->addr;
		mr->memory_size = seg->len;
		mr->mmap_offset = 0;

		if (k >= max_regions) {
			free(vm);
			return NULL;
		}
	}

	vm->nregions = k;
	vm->padding = 0;
	return vm;
}

/* with below features, vhost kernel does not need to do the checksum and TSO,
 * these info will be passed to virtio_user through virtio net header.
 */
#define VHOST_KERNEL_GUEST_OFFLOADS_MASK	\
	((1ULL << VIRTIO_NET_F_GUEST_CSUM) |	\
	 (1ULL << VIRTIO_NET_F_GUEST_TSO4) |	\
	 (1ULL << VIRTIO_NET_F_GUEST_TSO6) |	\
	 (1ULL << VIRTIO_NET_F_GUEST_ECN)  |	\
	 (1ULL << VIRTIO_NET_F_GUEST_UFO))

/* with below features, when flows from virtio_user to vhost kernel
 * (1) if flows goes up through the kernel networking stack, it does not need
 * to verify checksum, which can save CPU cycles;
 * (2) if flows goes through a Linux bridge and outside from an interface
 * (kernel driver), checksum and TSO will be done by GSO in kernel or even
 * offloaded into real physical device.
 */
#define VHOST_KERNEL_HOST_OFFLOADS_MASK		\
	((1ULL << VIRTIO_NET_F_HOST_TSO4) |	\
	 (1ULL << VIRTIO_NET_F_HOST_TSO6) |	\
	 (1ULL << VIRTIO_NET_F_CSUM))

static int
tap_supporte_mq(void)
{
	int tapfd;
	unsigned int tap_features;

	tapfd = open(PATH_NET_TUN, O_RDWR);
	if (tapfd < 0) {
		PMD_DRV_LOG(ERR, "fail to open %s: %s",
			    PATH_NET_TUN, strerror(errno));
		return -1;
	}

	if (ioctl(tapfd, TUNGETFEATURES, &tap_features) == -1) {
		PMD_DRV_LOG(ERR, "TUNGETFEATURES failed: %s", strerror(errno));
		close(tapfd);
		return -1;
	}

	close(tapfd);
	return tap_features & IFF_MULTI_QUEUE;
}

static int
vhost_kernel_ioctl(struct virtio_user_dev *dev,
		   enum vhost_user_request req,
		   void *arg)
{
	int ret = -1;
	unsigned int i;
	uint64_t req_kernel;
	struct vhost_memory_kernel *vm = NULL;
	int vhostfd;
	unsigned int queue_sel;

	PMD_DRV_LOG(INFO, "%s", vhost_msg_strings[req]);

	req_kernel = vhost_req_user_to_kernel[req];

	if (req_kernel == VHOST_SET_MEM_TABLE) {
		vm = prepare_vhost_memory_kernel();
		if (!vm)
			return -1;
		arg = (void *)vm;
	}

	if (req_kernel == VHOST_SET_FEATURES) {
		/* We don't need memory protection here */
		*(uint64_t *)arg &= ~(1ULL << VIRTIO_F_IOMMU_PLATFORM);

		/* VHOST kernel does not know about below flags */
		*(uint64_t *)arg &= ~VHOST_KERNEL_GUEST_OFFLOADS_MASK;
		*(uint64_t *)arg &= ~VHOST_KERNEL_HOST_OFFLOADS_MASK;

		*(uint64_t *)arg &= ~(1ULL << VIRTIO_NET_F_MQ);
	}

	switch (req_kernel) {
	case VHOST_SET_VRING_NUM:
	case VHOST_SET_VRING_ADDR:
	case VHOST_SET_VRING_BASE:
	case VHOST_GET_VRING_BASE:
	case VHOST_SET_VRING_KICK:
	case VHOST_SET_VRING_CALL:
		queue_sel = *(unsigned int *)arg;
		vhostfd = dev->vhostfds[queue_sel / 2];
		*(unsigned int *)arg = queue_sel % 2;
		PMD_DRV_LOG(DEBUG, "vhostfd=%d, index=%u",
			    vhostfd, *(unsigned int *)arg);
		break;
	default:
		vhostfd = -1;
	}
	if (vhostfd == -1) {
		for (i = 0; i < dev->max_queue_pairs; ++i) {
			if (dev->vhostfds[i] < 0)
				continue;

			ret = ioctl(dev->vhostfds[i], req_kernel, arg);
			if (ret < 0)
				break;
		}
	} else {
		ret = ioctl(vhostfd, req_kernel, arg);
	}

	if (!ret && req_kernel == VHOST_GET_FEATURES) {
		/* with tap as the backend, all these features are supported
		 * but not claimed by vhost-net, so we add them back when
		 * reporting to upper layer.
		 */
		*((uint64_t *)arg) |= VHOST_KERNEL_GUEST_OFFLOADS_MASK;
		*((uint64_t *)arg) |= VHOST_KERNEL_HOST_OFFLOADS_MASK;

		/* vhost_kernel will not declare this feature, but it does
		 * support multi-queue.
		 */
		if (tap_supporte_mq())
			*(uint64_t *)arg |= (1ull << VIRTIO_NET_F_MQ);
	}

	if (vm)
		free(vm);

	if (ret < 0)
		PMD_DRV_LOG(ERR, "%s failed: %s",
			    vhost_msg_strings[req], strerror(errno));

	return ret;
}

/**
 * Set up environment to talk with a vhost kernel backend.
 *
 * @return
 *   - (-1) if fail to set up;
 *   - (>=0) if successful.
 */
static int
vhost_kernel_setup(struct virtio_user_dev *dev)
{
	int vhostfd;
	uint32_t i;

	get_vhost_kernel_max_regions();

	for (i = 0; i < dev->max_queue_pairs; ++i) {
		vhostfd = open(dev->path, O_RDWR);
		if (vhostfd < 0) {
			PMD_DRV_LOG(ERR, "fail to open %s, %s",
				    dev->path, strerror(errno));
			return -1;
		}

		dev->vhostfds[i] = vhostfd;
	}

	return 0;
}

static int
vhost_kernel_set_backend(int vhostfd, int tapfd)
{
	struct vhost_vring_file f;

	f.fd = tapfd;
	f.index = 0;
	if (ioctl(vhostfd, VHOST_NET_SET_BACKEND, &f) < 0) {
		PMD_DRV_LOG(ERR, "VHOST_NET_SET_BACKEND fails, %s",
				strerror(errno));
		return -1;
	}

	f.index = 1;
	if (ioctl(vhostfd, VHOST_NET_SET_BACKEND, &f) < 0) {
		PMD_DRV_LOG(ERR, "VHOST_NET_SET_BACKEND fails, %s",
				strerror(errno));
		return -1;
	}

	return 0;
}

static int
vhost_kernel_enable_queue_pair(struct virtio_user_dev *dev,
			       uint16_t pair_idx,
			       int enable)
{
	int hdr_size;
	int vhostfd;
	int tapfd;
	int req_mq = (dev->max_queue_pairs > 1);

	vhostfd = dev->vhostfds[pair_idx];

	if (!enable) {
		if (dev->tapfds[pair_idx] >= 0) {
			close(dev->tapfds[pair_idx]);
			dev->tapfds[pair_idx] = -1;
		}
		return vhost_kernel_set_backend(vhostfd, -1);
	} else if (dev->tapfds[pair_idx] >= 0) {
		return 0;
	}

	if ((dev->features & (1ULL << VIRTIO_NET_F_MRG_RXBUF)) ||
	    (dev->features & (1ULL << VIRTIO_F_VERSION_1)))
		hdr_size = sizeof(struct virtio_net_hdr_mrg_rxbuf);
	else
		hdr_size = sizeof(struct virtio_net_hdr);

	tapfd = vhost_kernel_open_tap(&dev->ifname, hdr_size, req_mq);
	if (tapfd < 0) {
		PMD_DRV_LOG(ERR, "fail to open tap for vhost kernel");
		return -1;
	}

	if (vhost_kernel_set_backend(vhostfd, tapfd) < 0) {
		PMD_DRV_LOG(ERR, "fail to set backend for vhost kernel");
		close(tapfd);
		return -1;
	}

	dev->tapfds[pair_idx] = tapfd;
	return 0;
}

struct virtio_user_backend_ops ops_kernel = {
	.setup = vhost_kernel_setup,
	.send_request = vhost_kernel_ioctl,
	.enable_qp = vhost_kernel_enable_queue_pair
};