scsi.c 56.4 KB
Newer Older
1 2 3
/*******************************************************************************
 * Vhost kernel TCM fabric driver for virtio SCSI initiators
 *
4
 * (C) Copyright 2010-2013 Datera, Inc.
5 6 7 8
 * (C) Copyright 2010-2012 IBM Corp.
 *
 * Licensed to the Linux Foundation under the General Public License (GPL) version 2.
 *
9
 * Authors: Nicholas A. Bellinger <nab@daterainc.com>
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
 *          Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 ****************************************************************************/

#include <linux/module.h>
#include <linux/moduleparam.h>
#include <generated/utsrelease.h>
#include <linux/utsname.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/kthread.h>
#include <linux/types.h>
#include <linux/string.h>
#include <linux/configfs.h>
#include <linux/ctype.h>
#include <linux/compat.h>
#include <linux/eventfd.h>
#include <linux/fs.h>
38
#include <linux/vmalloc.h>
39 40
#include <linux/miscdevice.h>
#include <asm/unaligned.h>
41 42
#include <scsi/scsi_common.h>
#include <scsi/scsi_proto.h>
43 44 45 46
#include <target/target_core_base.h>
#include <target/target_core_fabric.h>
#include <linux/vhost.h>
#include <linux/virtio_scsi.h>
47
#include <linux/llist.h>
A
Asias He 已提交
48
#include <linux/bitmap.h>
49
#include <linux/percpu_ida.h>
50 51

#include "vhost.h"
M
Michael S. Tsirkin 已提交
52

53 54 55 56 57 58 59
#define VHOST_SCSI_VERSION  "v0.1"
#define VHOST_SCSI_NAMELEN 256
#define VHOST_SCSI_MAX_CDB_SIZE 32
#define VHOST_SCSI_DEFAULT_TAGS 256
#define VHOST_SCSI_PREALLOC_SGLS 2048
#define VHOST_SCSI_PREALLOC_UPAGES 2048
#define VHOST_SCSI_PREALLOC_PROT_SGLS 512
M
Michael S. Tsirkin 已提交
60 61 62 63 64 65 66 67

struct vhost_scsi_inflight {
	/* Wait for the flush operation to finish */
	struct completion comp;
	/* Refcount for the inflight reqs */
	struct kref kref;
};

68
struct vhost_scsi_cmd {
M
Michael S. Tsirkin 已提交
69 70 71 72
	/* Descriptor from vhost_get_vq_desc() for virt_queue segment */
	int tvc_vq_desc;
	/* virtio-scsi initiator task attribute */
	int tvc_task_attr;
73 74
	/* virtio-scsi response incoming iovecs */
	int tvc_in_iovs;
M
Michael S. Tsirkin 已提交
75 76 77 78 79 80 81 82
	/* virtio-scsi initiator data direction */
	enum dma_data_direction tvc_data_direction;
	/* Expected data transfer length from virtio-scsi header */
	u32 tvc_exp_data_len;
	/* The Tag from include/linux/virtio_scsi.h:struct virtio_scsi_cmd_req */
	u64 tvc_tag;
	/* The number of scatterlists associated with this cmd */
	u32 tvc_sgl_count;
83
	u32 tvc_prot_sgl_count;
84
	/* Saved unpacked SCSI LUN for vhost_scsi_submission_work() */
M
Michael S. Tsirkin 已提交
85 86 87
	u32 tvc_lun;
	/* Pointer to the SGL formatted memory from virtio-scsi */
	struct scatterlist *tvc_sgl;
88
	struct scatterlist *tvc_prot_sgl;
89
	struct page **tvc_upages;
90
	/* Pointer to response header iovec */
91
	struct iovec tvc_resp_iov;
M
Michael S. Tsirkin 已提交
92 93 94 95 96
	/* Pointer to vhost_scsi for our device */
	struct vhost_scsi *tvc_vhost;
	/* Pointer to vhost_virtqueue for the cmd */
	struct vhost_virtqueue *tvc_vq;
	/* Pointer to vhost nexus memory */
97
	struct vhost_scsi_nexus *tvc_nexus;
M
Michael S. Tsirkin 已提交
98 99
	/* The TCM I/O descriptor that is accessed via container_of() */
	struct se_cmd tvc_se_cmd;
100
	/* work item used for cmwq dispatch to vhost_scsi_submission_work() */
M
Michael S. Tsirkin 已提交
101 102
	struct work_struct work;
	/* Copy of the incoming SCSI command descriptor block (CDB) */
103
	unsigned char tvc_cdb[VHOST_SCSI_MAX_CDB_SIZE];
M
Michael S. Tsirkin 已提交
104 105 106 107 108 109 110 111
	/* Sense buffer that will be mapped into outgoing status */
	unsigned char tvc_sense_buf[TRANSPORT_SENSE_BUFFER];
	/* Completed commands list, serviced from vhost worker thread */
	struct llist_node tvc_completion_list;
	/* Used to track inflight cmd */
	struct vhost_scsi_inflight *inflight;
};

112
struct vhost_scsi_nexus {
M
Michael S. Tsirkin 已提交
113 114 115 116
	/* Pointer to TCM session for I_T Nexus */
	struct se_session *tvn_se_sess;
};

117
struct vhost_scsi_tpg {
M
Michael S. Tsirkin 已提交
118 119 120 121 122 123
	/* Vhost port target portal group tag for TCM */
	u16 tport_tpgt;
	/* Used to track number of TPG Port/Lun Links wrt to explict I_T Nexus shutdown */
	int tv_tpg_port_count;
	/* Used for vhost_scsi device reference to tpg_nexus, protected by tv_tpg_mutex */
	int tv_tpg_vhost_count;
124 125
	/* Used for enabling T10-PI with legacy devices */
	int tv_fabric_prot_type;
126
	/* list for vhost_scsi_list */
M
Michael S. Tsirkin 已提交
127 128 129 130
	struct list_head tv_tpg_list;
	/* Used to protect access for tpg_nexus */
	struct mutex tv_tpg_mutex;
	/* Pointer to the TCM VHost I_T Nexus for this TPG endpoint */
131 132 133 134
	struct vhost_scsi_nexus *tpg_nexus;
	/* Pointer back to vhost_scsi_tport */
	struct vhost_scsi_tport *tport;
	/* Returned by vhost_scsi_make_tpg() */
M
Michael S. Tsirkin 已提交
135 136 137 138 139
	struct se_portal_group se_tpg;
	/* Pointer back to vhost_scsi, protected by tv_tpg_mutex */
	struct vhost_scsi *vhost_scsi;
};

140
struct vhost_scsi_tport {
M
Michael S. Tsirkin 已提交
141 142 143 144 145
	/* SCSI protocol the tport is providing */
	u8 tport_proto_id;
	/* Binary World Wide unique Port Name for Vhost Target port */
	u64 tport_wwpn;
	/* ASCII formatted WWPN for Vhost Target port */
146 147
	char tport_name[VHOST_SCSI_NAMELEN];
	/* Returned by vhost_scsi_make_tport() */
M
Michael S. Tsirkin 已提交
148 149 150
	struct se_wwn tport_wwn;
};

151
struct vhost_scsi_evt {
M
Michael S. Tsirkin 已提交
152 153 154 155 156
	/* event to be sent to guest */
	struct virtio_scsi_event event;
	/* event list, serviced from vhost worker thread */
	struct llist_node list;
};
157

158 159 160 161 162 163
enum {
	VHOST_SCSI_VQ_CTL = 0,
	VHOST_SCSI_VQ_EVT = 1,
	VHOST_SCSI_VQ_IO = 2,
};

164
/* Note: can't set VIRTIO_F_VERSION_1 yet, since that implies ANY_LAYOUT. */
165
enum {
166
	VHOST_SCSI_FEATURES = VHOST_FEATURES | (1ULL << VIRTIO_SCSI_F_HOTPLUG) |
167
					       (1ULL << VIRTIO_SCSI_F_T10_PI)
168 169
};

A
Asias He 已提交
170 171
#define VHOST_SCSI_MAX_TARGET	256
#define VHOST_SCSI_MAX_VQ	128
172
#define VHOST_SCSI_MAX_EVENT	128
A
Asias He 已提交
173

174 175
struct vhost_scsi_virtqueue {
	struct vhost_virtqueue vq;
176 177 178 179 180
	/*
	 * Reference counting for inflight reqs, used for flush operation. At
	 * each time, one reference tracks new commands submitted, while we
	 * wait for another one to reach 0.
	 */
181
	struct vhost_scsi_inflight inflights[2];
182 183 184 185
	/*
	 * Indicate current inflight in use, protected by vq->mutex.
	 * Writers must also take dev mutex and flush under it.
	 */
186
	int inflight_idx;
187 188
};

189
struct vhost_scsi {
A
Asias He 已提交
190
	/* Protected by vhost_scsi->dev.mutex */
191
	struct vhost_scsi_tpg **vs_tpg;
A
Asias He 已提交
192 193
	char vs_vhost_wwpn[TRANSPORT_IQN_LEN];

194
	struct vhost_dev dev;
195
	struct vhost_scsi_virtqueue vqs[VHOST_SCSI_MAX_VQ];
196 197

	struct vhost_work vs_completion_work; /* cmd completion work item */
198
	struct llist_head vs_completion_list; /* cmd completion queue */
199 200 201 202 203 204

	struct vhost_work vs_event_work; /* evt injection work item */
	struct llist_head vs_event_list; /* evt injection queue */

	bool vs_events_missed; /* any missed events, protected by vq->mutex */
	int vs_events_nr; /* num of pending events, protected by vq->mutex */
205 206
};

207
static struct workqueue_struct *vhost_scsi_workqueue;
208

209 210 211
/* Global spinlock to protect vhost_scsi TPG list for vhost IOCTL access */
static DEFINE_MUTEX(vhost_scsi_mutex);
static LIST_HEAD(vhost_scsi_list);
212

213
static void vhost_scsi_done_inflight(struct kref *kref)
214 215 216 217 218 219 220
{
	struct vhost_scsi_inflight *inflight;

	inflight = container_of(kref, struct vhost_scsi_inflight, kref);
	complete(&inflight->comp);
}

221
static void vhost_scsi_init_inflight(struct vhost_scsi *vs,
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
				    struct vhost_scsi_inflight *old_inflight[])
{
	struct vhost_scsi_inflight *new_inflight;
	struct vhost_virtqueue *vq;
	int idx, i;

	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
		vq = &vs->vqs[i].vq;

		mutex_lock(&vq->mutex);

		/* store old infight */
		idx = vs->vqs[i].inflight_idx;
		if (old_inflight)
			old_inflight[i] = &vs->vqs[i].inflights[idx];

		/* setup new infight */
		vs->vqs[i].inflight_idx = idx ^ 1;
		new_inflight = &vs->vqs[i].inflights[idx ^ 1];
		kref_init(&new_inflight->kref);
		init_completion(&new_inflight->comp);

		mutex_unlock(&vq->mutex);
	}
}

static struct vhost_scsi_inflight *
249
vhost_scsi_get_inflight(struct vhost_virtqueue *vq)
250 251 252 253 254 255 256 257 258 259 260
{
	struct vhost_scsi_inflight *inflight;
	struct vhost_scsi_virtqueue *svq;

	svq = container_of(vq, struct vhost_scsi_virtqueue, vq);
	inflight = &svq->inflights[svq->inflight_idx];
	kref_get(&inflight->kref);

	return inflight;
}

261
static void vhost_scsi_put_inflight(struct vhost_scsi_inflight *inflight)
262
{
263
	kref_put(&inflight->kref, vhost_scsi_done_inflight);
264 265
}

266
static int vhost_scsi_check_true(struct se_portal_group *se_tpg)
267 268 269 270
{
	return 1;
}

271
static int vhost_scsi_check_false(struct se_portal_group *se_tpg)
272 273 274 275
{
	return 0;
}

276
static char *vhost_scsi_get_fabric_name(void)
277 278 279 280
{
	return "vhost";
}

281
static char *vhost_scsi_get_fabric_wwn(struct se_portal_group *se_tpg)
282
{
283 284 285
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
	struct vhost_scsi_tport *tport = tpg->tport;
286 287 288 289

	return &tport->tport_name[0];
}

290
static u16 vhost_scsi_get_tpgt(struct se_portal_group *se_tpg)
291
{
292 293
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
294 295 296
	return tpg->tport_tpgt;
}

297 298 299 300 301 302 303 304
static int vhost_scsi_check_prot_fabric_only(struct se_portal_group *se_tpg)
{
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);

	return tpg->tv_fabric_prot_type;
}

305
static u32 vhost_scsi_tpg_get_inst_index(struct se_portal_group *se_tpg)
306 307 308 309
{
	return 1;
}

310
static void vhost_scsi_release_cmd(struct se_cmd *se_cmd)
311
{
312 313
	struct vhost_scsi_cmd *tv_cmd = container_of(se_cmd,
				struct vhost_scsi_cmd, tvc_se_cmd);
314
	struct se_session *se_sess = tv_cmd->tvc_nexus->tvn_se_sess;
315
	int i;
316 317 318 319

	if (tv_cmd->tvc_sgl_count) {
		for (i = 0; i < tv_cmd->tvc_sgl_count; i++)
			put_page(sg_page(&tv_cmd->tvc_sgl[i]));
320
	}
321 322 323 324
	if (tv_cmd->tvc_prot_sgl_count) {
		for (i = 0; i < tv_cmd->tvc_prot_sgl_count; i++)
			put_page(sg_page(&tv_cmd->tvc_prot_sgl[i]));
	}
325

326
	vhost_scsi_put_inflight(tv_cmd->inflight);
327
	percpu_ida_free(&se_sess->sess_tag_pool, se_cmd->map_tag);
328 329
}

330
static u32 vhost_scsi_sess_get_index(struct se_session *se_sess)
331 332 333 334
{
	return 0;
}

335
static int vhost_scsi_write_pending(struct se_cmd *se_cmd)
336 337 338 339 340 341
{
	/* Go ahead and process the write immediately */
	target_execute_cmd(se_cmd);
	return 0;
}

342
static int vhost_scsi_write_pending_status(struct se_cmd *se_cmd)
343 344 345 346
{
	return 0;
}

347
static void vhost_scsi_set_default_node_attrs(struct se_node_acl *nacl)
348 349 350 351
{
	return;
}

352
static int vhost_scsi_get_cmd_state(struct se_cmd *se_cmd)
353 354 355 356
{
	return 0;
}

357
static void vhost_scsi_complete_cmd(struct vhost_scsi_cmd *cmd)
358
{
359
	struct vhost_scsi *vs = cmd->tvc_vhost;
360

361
	llist_add(&cmd->tvc_completion_list, &vs->vs_completion_list);
362 363 364

	vhost_work_queue(&vs->dev, &vs->vs_completion_work);
}
365

366
static int vhost_scsi_queue_data_in(struct se_cmd *se_cmd)
367
{
368 369
	struct vhost_scsi_cmd *cmd = container_of(se_cmd,
				struct vhost_scsi_cmd, tvc_se_cmd);
370
	vhost_scsi_complete_cmd(cmd);
371 372 373
	return 0;
}

374
static int vhost_scsi_queue_status(struct se_cmd *se_cmd)
375
{
376 377
	struct vhost_scsi_cmd *cmd = container_of(se_cmd,
				struct vhost_scsi_cmd, tvc_se_cmd);
378
	vhost_scsi_complete_cmd(cmd);
379 380 381
	return 0;
}

382
static void vhost_scsi_queue_tm_rsp(struct se_cmd *se_cmd)
383
{
384
	return;
385 386
}

387
static void vhost_scsi_aborted_task(struct se_cmd *se_cmd)
388 389 390 391
{
	return;
}

392
static void vhost_scsi_free_evt(struct vhost_scsi *vs, struct vhost_scsi_evt *evt)
393 394 395 396 397
{
	vs->vs_events_nr--;
	kfree(evt);
}

398 399
static struct vhost_scsi_evt *
vhost_scsi_allocate_evt(struct vhost_scsi *vs,
400
		       u32 event, u32 reason)
401
{
402
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
403
	struct vhost_scsi_evt *evt;
404 405 406 407 408 409 410 411

	if (vs->vs_events_nr > VHOST_SCSI_MAX_EVENT) {
		vs->vs_events_missed = true;
		return NULL;
	}

	evt = kzalloc(sizeof(*evt), GFP_KERNEL);
	if (!evt) {
412
		vq_err(vq, "Failed to allocate vhost_scsi_evt\n");
413 414 415 416
		vs->vs_events_missed = true;
		return NULL;
	}

417 418
	evt->event.event = cpu_to_vhost32(vq, event);
	evt->event.reason = cpu_to_vhost32(vq, reason);
419 420 421 422 423
	vs->vs_events_nr++;

	return evt;
}

424
static void vhost_scsi_free_cmd(struct vhost_scsi_cmd *cmd)
425
{
426
	struct se_cmd *se_cmd = &cmd->tvc_se_cmd;
427 428

	/* TODO locking against target/backend threads? */
429
	transport_generic_free_cmd(se_cmd, 0);
430

431
}
432

433 434
static int vhost_scsi_check_stop_free(struct se_cmd *se_cmd)
{
435
	return target_put_sess_cmd(se_cmd);
436 437
}

438
static void
439
vhost_scsi_do_evt_work(struct vhost_scsi *vs, struct vhost_scsi_evt *evt)
440
{
441
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
442 443 444 445 446 447 448 449 450 451 452 453
	struct virtio_scsi_event *event = &evt->event;
	struct virtio_scsi_event __user *eventp;
	unsigned out, in;
	int head, ret;

	if (!vq->private_data) {
		vs->vs_events_missed = true;
		return;
	}

again:
	vhost_disable_notify(&vs->dev, vq);
454
	head = vhost_get_vq_desc(vq, vq->iov,
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
			ARRAY_SIZE(vq->iov), &out, &in,
			NULL, NULL);
	if (head < 0) {
		vs->vs_events_missed = true;
		return;
	}
	if (head == vq->num) {
		if (vhost_enable_notify(&vs->dev, vq))
			goto again;
		vs->vs_events_missed = true;
		return;
	}

	if ((vq->iov[out].iov_len != sizeof(struct virtio_scsi_event))) {
		vq_err(vq, "Expecting virtio_scsi_event, got %zu bytes\n",
				vq->iov[out].iov_len);
		vs->vs_events_missed = true;
		return;
	}

	if (vs->vs_events_missed) {
476
		event->event |= cpu_to_vhost32(vq, VIRTIO_SCSI_T_EVENTS_MISSED);
477 478 479 480 481 482 483 484
		vs->vs_events_missed = false;
	}

	eventp = vq->iov[out].iov_base;
	ret = __copy_to_user(eventp, event, sizeof(*event));
	if (!ret)
		vhost_add_used_and_signal(&vs->dev, vq, head, 0);
	else
485
		vq_err(vq, "Faulted on vhost_scsi_send_event\n");
486 487
}

488
static void vhost_scsi_evt_work(struct vhost_work *work)
489 490 491
{
	struct vhost_scsi *vs = container_of(work, struct vhost_scsi,
					vs_event_work);
492
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
493
	struct vhost_scsi_evt *evt, *t;
494 495 496 497
	struct llist_node *llnode;

	mutex_lock(&vq->mutex);
	llnode = llist_del_all(&vs->vs_event_list);
498
	llist_for_each_entry_safe(evt, t, llnode, list) {
499 500
		vhost_scsi_do_evt_work(vs, evt);
		vhost_scsi_free_evt(vs, evt);
501 502 503 504
	}
	mutex_unlock(&vq->mutex);
}

505 506 507 508 509 510 511 512 513
/* Fill in status and signal that we are done processing this command
 *
 * This is scheduled in the vhost work queue so we are called with the owner
 * process mm and can access the vring.
 */
static void vhost_scsi_complete_cmd_work(struct vhost_work *work)
{
	struct vhost_scsi *vs = container_of(work, struct vhost_scsi,
					vs_completion_work);
A
Asias He 已提交
514
	DECLARE_BITMAP(signal, VHOST_SCSI_MAX_VQ);
515
	struct virtio_scsi_cmd_resp v_rsp;
516
	struct vhost_scsi_cmd *cmd, *t;
517 518
	struct llist_node *llnode;
	struct se_cmd *se_cmd;
519
	struct iov_iter iov_iter;
A
Asias He 已提交
520
	int ret, vq;
521

A
Asias He 已提交
522
	bitmap_zero(signal, VHOST_SCSI_MAX_VQ);
523
	llnode = llist_del_all(&vs->vs_completion_list);
524
	llist_for_each_entry_safe(cmd, t, llnode, tvc_completion_list) {
525
		se_cmd = &cmd->tvc_se_cmd;
526 527

		pr_debug("%s tv_cmd %p resid %u status %#02x\n", __func__,
528
			cmd, se_cmd->residual_count, se_cmd->scsi_status);
529 530

		memset(&v_rsp, 0, sizeof(v_rsp));
531
		v_rsp.resid = cpu_to_vhost32(cmd->tvc_vq, se_cmd->residual_count);
532 533
		/* TODO is status_qualifier field needed? */
		v_rsp.status = se_cmd->scsi_status;
534 535
		v_rsp.sense_len = cpu_to_vhost32(cmd->tvc_vq,
						 se_cmd->scsi_sense_length);
536
		memcpy(v_rsp.sense, cmd->tvc_sense_buf,
537
		       se_cmd->scsi_sense_length);
538

539
		iov_iter_init(&iov_iter, READ, &cmd->tvc_resp_iov,
540 541 542
			      cmd->tvc_in_iovs, sizeof(v_rsp));
		ret = copy_to_iter(&v_rsp, sizeof(v_rsp), &iov_iter);
		if (likely(ret == sizeof(v_rsp))) {
543
			struct vhost_scsi_virtqueue *q;
544 545
			vhost_add_used(cmd->tvc_vq, cmd->tvc_vq_desc, 0);
			q = container_of(cmd->tvc_vq, struct vhost_scsi_virtqueue, vq);
546
			vq = q - vs->vqs;
A
Asias He 已提交
547 548
			__set_bit(vq, signal);
		} else
549 550
			pr_err("Faulted on virtio_scsi_cmd_resp\n");

551
		vhost_scsi_free_cmd(cmd);
552 553
	}

A
Asias He 已提交
554 555 556
	vq = -1;
	while ((vq = find_next_bit(signal, VHOST_SCSI_MAX_VQ, vq + 1))
		< VHOST_SCSI_MAX_VQ)
557
		vhost_signal(&vs->dev, &vs->vqs[vq].vq);
558 559
}

560 561
static struct vhost_scsi_cmd *
vhost_scsi_get_tag(struct vhost_virtqueue *vq, struct vhost_scsi_tpg *tpg,
562 563
		   unsigned char *cdb, u64 scsi_tag, u16 lun, u8 task_attr,
		   u32 exp_data_len, int data_direction)
564
{
565 566
	struct vhost_scsi_cmd *cmd;
	struct vhost_scsi_nexus *tv_nexus;
567
	struct se_session *se_sess;
568
	struct scatterlist *sg, *prot_sg;
569
	struct page **pages;
570
	int tag;
571

572
	tv_nexus = tpg->tpg_nexus;
573
	if (!tv_nexus) {
574
		pr_err("Unable to locate active struct vhost_scsi_nexus\n");
575 576
		return ERR_PTR(-EIO);
	}
577
	se_sess = tv_nexus->tvn_se_sess;
578

579
	tag = percpu_ida_alloc(&se_sess->sess_tag_pool, TASK_RUNNING);
580
	if (tag < 0) {
581
		pr_err("Unable to obtain tag for vhost_scsi_cmd\n");
582 583 584
		return ERR_PTR(-ENOMEM);
	}

585
	cmd = &((struct vhost_scsi_cmd *)se_sess->sess_cmd_map)[tag];
586
	sg = cmd->tvc_sgl;
587
	prot_sg = cmd->tvc_prot_sgl;
588
	pages = cmd->tvc_upages;
589
	memset(cmd, 0, sizeof(struct vhost_scsi_cmd));
590

591
	cmd->tvc_sgl = sg;
592
	cmd->tvc_prot_sgl = prot_sg;
593
	cmd->tvc_upages = pages;
594
	cmd->tvc_se_cmd.map_tag = tag;
595 596 597
	cmd->tvc_tag = scsi_tag;
	cmd->tvc_lun = lun;
	cmd->tvc_task_attr = task_attr;
598 599 600
	cmd->tvc_exp_data_len = exp_data_len;
	cmd->tvc_data_direction = data_direction;
	cmd->tvc_nexus = tv_nexus;
601
	cmd->inflight = vhost_scsi_get_inflight(vq);
602

603
	memcpy(cmd->tvc_cdb, cdb, VHOST_SCSI_MAX_CDB_SIZE);
604

605
	return cmd;
606 607 608 609 610 611 612
}

/*
 * Map a user memory range into a scatterlist
 *
 * Returns the number of scatterlist entries used or -errno on error.
 */
613
static int
614
vhost_scsi_map_to_sgl(struct vhost_scsi_cmd *cmd,
615
		      struct iov_iter *iter,
616
		      struct scatterlist *sgl,
617
		      bool write)
618
{
619
	struct page **pages = cmd->tvc_upages;
620 621 622 623
	struct scatterlist *sg = sgl;
	ssize_t bytes;
	size_t offset;
	unsigned int npages = 0;
624

625 626
	bytes = iov_iter_get_pages(iter, pages, LONG_MAX,
				VHOST_SCSI_PREALLOC_UPAGES, &offset);
627
	/* No pages were pinned */
628 629
	if (bytes <= 0)
		return bytes < 0 ? bytes : -EFAULT;
630

631
	iov_iter_advance(iter, bytes);
632

633 634 635 636 637 638 639
	while (bytes) {
		unsigned n = min_t(unsigned, PAGE_SIZE - offset, bytes);
		sg_set_page(sg++, pages[npages++], n, offset);
		bytes -= n;
		offset = 0;
	}
	return npages;
640 641
}

642
static int
643
vhost_scsi_calc_sgls(struct iov_iter *iter, size_t bytes, int max_sgls)
644
{
645
	int sgl_count = 0;
646

647 648 649 650 651
	if (!iter || !iter->iov) {
		pr_err("%s: iter->iov is NULL, but expected bytes: %zu"
		       " present\n", __func__, bytes);
		return -EINVAL;
	}
652

653 654 655 656 657
	sgl_count = iov_iter_npages(iter, 0xffff);
	if (sgl_count > max_sgls) {
		pr_err("%s: requested sgl_count: %d exceeds pre-allocated"
		       " max_sgls: %d\n", __func__, sgl_count, max_sgls);
		return -EINVAL;
658
	}
659 660
	return sgl_count;
}
661

662
static int
663 664 665
vhost_scsi_iov_to_sgl(struct vhost_scsi_cmd *cmd, bool write,
		      struct iov_iter *iter,
		      struct scatterlist *sg, int sg_count)
666
{
667
	struct scatterlist *p = sg;
668
	int ret;
669

670 671
	while (iov_iter_count(iter)) {
		ret = vhost_scsi_map_to_sgl(cmd, iter, sg, write);
672
		if (ret < 0) {
673 674
			while (p < sg) {
				struct page *page = sg_page(p++);
675 676 677
				if (page)
					put_page(page);
			}
678 679 680 681 682 683 684
			return ret;
		}
		sg += ret;
	}
	return 0;
}

685
static int
686
vhost_scsi_mapal(struct vhost_scsi_cmd *cmd,
687 688 689 690 691 692 693 694
		 size_t prot_bytes, struct iov_iter *prot_iter,
		 size_t data_bytes, struct iov_iter *data_iter)
{
	int sgl_count, ret;
	bool write = (cmd->tvc_data_direction == DMA_FROM_DEVICE);

	if (prot_bytes) {
		sgl_count = vhost_scsi_calc_sgls(prot_iter, prot_bytes,
695
						 VHOST_SCSI_PREALLOC_PROT_SGLS);
696 697 698 699 700 701 702 703 704 705 706
		if (sgl_count < 0)
			return sgl_count;

		sg_init_table(cmd->tvc_prot_sgl, sgl_count);
		cmd->tvc_prot_sgl_count = sgl_count;
		pr_debug("%s prot_sg %p prot_sgl_count %u\n", __func__,
			 cmd->tvc_prot_sgl, cmd->tvc_prot_sgl_count);

		ret = vhost_scsi_iov_to_sgl(cmd, write, prot_iter,
					    cmd->tvc_prot_sgl,
					    cmd->tvc_prot_sgl_count);
707 708 709 710
		if (ret < 0) {
			cmd->tvc_prot_sgl_count = 0;
			return ret;
		}
711 712
	}
	sgl_count = vhost_scsi_calc_sgls(data_iter, data_bytes,
713
					 VHOST_SCSI_PREALLOC_SGLS);
714 715 716 717 718 719 720 721 722 723 724 725 726
	if (sgl_count < 0)
		return sgl_count;

	sg_init_table(cmd->tvc_sgl, sgl_count);
	cmd->tvc_sgl_count = sgl_count;
	pr_debug("%s data_sg %p data_sgl_count %u\n", __func__,
		  cmd->tvc_sgl, cmd->tvc_sgl_count);

	ret = vhost_scsi_iov_to_sgl(cmd, write, data_iter,
				    cmd->tvc_sgl, cmd->tvc_sgl_count);
	if (ret < 0) {
		cmd->tvc_sgl_count = 0;
		return ret;
727 728 729 730
	}
	return 0;
}

731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
static int vhost_scsi_to_tcm_attr(int attr)
{
	switch (attr) {
	case VIRTIO_SCSI_S_SIMPLE:
		return TCM_SIMPLE_TAG;
	case VIRTIO_SCSI_S_ORDERED:
		return TCM_ORDERED_TAG;
	case VIRTIO_SCSI_S_HEAD:
		return TCM_HEAD_TAG;
	case VIRTIO_SCSI_S_ACA:
		return TCM_ACA_TAG;
	default:
		break;
	}
	return TCM_SIMPLE_TAG;
}

748
static void vhost_scsi_submission_work(struct work_struct *work)
749
{
750 751 752
	struct vhost_scsi_cmd *cmd =
		container_of(work, struct vhost_scsi_cmd, work);
	struct vhost_scsi_nexus *tv_nexus;
753
	struct se_cmd *se_cmd = &cmd->tvc_se_cmd;
754 755
	struct scatterlist *sg_ptr, *sg_prot_ptr = NULL;
	int rc;
756

757
	/* FIXME: BIDI operation */
758 759
	if (cmd->tvc_sgl_count) {
		sg_ptr = cmd->tvc_sgl;
760 761 762 763 764

		if (cmd->tvc_prot_sgl_count)
			sg_prot_ptr = cmd->tvc_prot_sgl;
		else
			se_cmd->prot_pto = true;
765 766 767
	} else {
		sg_ptr = NULL;
	}
768
	tv_nexus = cmd->tvc_nexus;
769

770
	se_cmd->tag = 0;
771
	rc = target_submit_cmd_map_sgls(se_cmd, tv_nexus->tvn_se_sess,
772 773
			cmd->tvc_cdb, &cmd->tvc_sense_buf[0],
			cmd->tvc_lun, cmd->tvc_exp_data_len,
774 775 776 777
			vhost_scsi_to_tcm_attr(cmd->tvc_task_attr),
			cmd->tvc_data_direction, TARGET_SCF_ACK_KREF,
			sg_ptr, cmd->tvc_sgl_count, NULL, 0, sg_prot_ptr,
			cmd->tvc_prot_sgl_count);
778 779
	if (rc < 0) {
		transport_send_check_condition_and_sense(se_cmd,
780
				TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE, 0);
781 782 783 784
		transport_generic_free_cmd(se_cmd, 0);
	}
}

785 786 787 788
static void
vhost_scsi_send_bad_target(struct vhost_scsi *vs,
			   struct vhost_virtqueue *vq,
			   int head, unsigned out)
789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
{
	struct virtio_scsi_cmd_resp __user *resp;
	struct virtio_scsi_cmd_resp rsp;
	int ret;

	memset(&rsp, 0, sizeof(rsp));
	rsp.response = VIRTIO_SCSI_S_BAD_TARGET;
	resp = vq->iov[out].iov_base;
	ret = __copy_to_user(resp, &rsp, sizeof(rsp));
	if (!ret)
		vhost_add_used_and_signal(&vs->dev, vq, head, 0);
	else
		pr_err("Faulted on virtio_scsi_cmd_resp\n");
}

804 805
static void
vhost_scsi_handle_vq(struct vhost_scsi *vs, struct vhost_virtqueue *vq)
806
{
807
	struct vhost_scsi_tpg **vs_tpg, *tpg;
808
	struct virtio_scsi_cmd_req v_req;
809
	struct virtio_scsi_cmd_req_pi v_req_pi;
810
	struct vhost_scsi_cmd *cmd;
811
	struct iov_iter out_iter, in_iter, prot_iter, data_iter;
812
	u64 tag;
813
	u32 exp_data_len, data_direction;
814
	unsigned int out = 0, in = 0;
815 816 817
	int head, ret, prot_bytes;
	size_t req_size, rsp_size = sizeof(struct virtio_scsi_cmd_resp);
	size_t out_size, in_size;
818 819
	u16 lun;
	u8 *target, *lunp, task_attr;
820
	bool t10_pi = vhost_has_feature(vq, VIRTIO_SCSI_F_T10_PI);
821
	void *req, *cdb;
822

823
	mutex_lock(&vq->mutex);
824 825 826 827
	/*
	 * We can handle the vq only after the endpoint is setup by calling the
	 * VHOST_SCSI_SET_ENDPOINT ioctl.
	 */
828
	vs_tpg = vq->private_data;
829
	if (!vs_tpg)
830
		goto out;
831 832 833 834

	vhost_disable_notify(&vs->dev, vq);

	for (;;) {
835
		head = vhost_get_vq_desc(vq, vq->iov,
836 837
					 ARRAY_SIZE(vq->iov), &out, &in,
					 NULL, NULL);
838
		pr_debug("vhost_get_vq_desc: head: %d, out: %u in: %u\n",
839
			 head, out, in);
840 841 842 843 844 845 846 847 848 849 850 851
		/* On error, stop handling until the next kick. */
		if (unlikely(head < 0))
			break;
		/* Nothing new?  Wait for eventfd to tell us they refilled. */
		if (head == vq->num) {
			if (unlikely(vhost_enable_notify(&vs->dev, vq))) {
				vhost_disable_notify(&vs->dev, vq);
				continue;
			}
			break;
		}
		/*
852 853
		 * Check for a sane response buffer so we can report early
		 * errors back to the guest.
854
		 */
855 856 857
		if (unlikely(vq->iov[out].iov_len < rsp_size)) {
			vq_err(vq, "Expecting at least virtio_scsi_cmd_resp"
				" size, got %zu bytes\n", vq->iov[out].iov_len);
858 859
			break;
		}
860 861 862 863 864
		/*
		 * Setup pointers and values based upon different virtio-scsi
		 * request header if T10_PI is enabled in KVM guest.
		 */
		if (t10_pi) {
865
			req = &v_req_pi;
866
			req_size = sizeof(v_req_pi);
867 868 869 870
			lunp = &v_req_pi.lun[0];
			target = &v_req_pi.lun[1];
		} else {
			req = &v_req;
871
			req_size = sizeof(v_req);
872 873 874
			lunp = &v_req.lun[0];
			target = &v_req.lun[1];
		}
875 876 877 878 879
		/*
		 * FIXME: Not correct for BIDI operation
		 */
		out_size = iov_length(vq->iov, out);
		in_size = iov_length(&vq->iov[out], in);
880

881 882 883 884 885 886 887 888 889 890 891
		/*
		 * Copy over the virtio-scsi request header, which for a
		 * ANY_LAYOUT enabled guest may span multiple iovecs, or a
		 * single iovec may contain both the header + outgoing
		 * WRITE payloads.
		 *
		 * copy_from_iter() will advance out_iter, so that it will
		 * point at the start of the outgoing WRITE payload, if
		 * DMA_TO_DEVICE is set.
		 */
		iov_iter_init(&out_iter, WRITE, vq->iov, out, out_size);
892

893
		if (unlikely(!copy_from_iter_full(req, req_size, &out_iter))) {
894
			vq_err(vq, "Faulted on copy_from_iter\n");
895 896
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
897
		}
898
		/* virtio-scsi spec requires byte 0 of the lun to be 1 */
899
		if (unlikely(*lunp != 1)) {
900
			vq_err(vq, "Illegal virtio-scsi lun: %u\n", *lunp);
901 902 903 904
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
		}

905
		tpg = READ_ONCE(vs_tpg[*target]);
906
		if (unlikely(!tpg)) {
907
			/* Target does not exist, fail the request */
908
			vhost_scsi_send_bad_target(vs, vq, head, out);
A
Asias He 已提交
909 910
			continue;
		}
911
		/*
912 913 914
		 * Determine data_direction by calculating the total outgoing
		 * iovec sizes + incoming iovec sizes vs. virtio-scsi request +
		 * response headers respectively.
915
		 *
916 917 918 919 920 921
		 * For DMA_TO_DEVICE this is out_iter, which is already pointing
		 * to the right place.
		 *
		 * For DMA_FROM_DEVICE, the iovec will be just past the end
		 * of the virtio-scsi response header in either the same
		 * or immediately following iovec.
922
		 *
923 924
		 * Any associated T10_PI bytes for the outgoing / incoming
		 * payloads are included in calculation of exp_data_len here.
925
		 */
926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
		prot_bytes = 0;

		if (out_size > req_size) {
			data_direction = DMA_TO_DEVICE;
			exp_data_len = out_size - req_size;
			data_iter = out_iter;
		} else if (in_size > rsp_size) {
			data_direction = DMA_FROM_DEVICE;
			exp_data_len = in_size - rsp_size;

			iov_iter_init(&in_iter, READ, &vq->iov[out], in,
				      rsp_size + exp_data_len);
			iov_iter_advance(&in_iter, rsp_size);
			data_iter = in_iter;
		} else {
			data_direction = DMA_NONE;
			exp_data_len = 0;
		}
		/*
		 * If T10_PI header + payload is present, setup prot_iter values
		 * and recalculate data_iter for vhost_scsi_mapal() mapping to
		 * host scatterlists via get_user_pages_fast().
948
		 */
949
		if (t10_pi) {
950 951
			if (v_req_pi.pi_bytesout) {
				if (data_direction != DMA_TO_DEVICE) {
952 953
					vq_err(vq, "Received non zero pi_bytesout,"
						" but wrong data_direction\n");
954 955
					vhost_scsi_send_bad_target(vs, vq, head, out);
					continue;
956
				}
957
				prot_bytes = vhost32_to_cpu(vq, v_req_pi.pi_bytesout);
958 959
			} else if (v_req_pi.pi_bytesin) {
				if (data_direction != DMA_FROM_DEVICE) {
960 961
					vq_err(vq, "Received non zero pi_bytesin,"
						" but wrong data_direction\n");
962 963
					vhost_scsi_send_bad_target(vs, vq, head, out);
					continue;
964
				}
965
				prot_bytes = vhost32_to_cpu(vq, v_req_pi.pi_bytesin);
966
			}
967 968 969 970 971 972 973
			/*
			 * Set prot_iter to data_iter, and advance past any
			 * preceeding prot_bytes that may be present.
			 *
			 * Also fix up the exp_data_len to reflect only the
			 * actual data payload length.
			 */
974
			if (prot_bytes) {
975 976 977
				exp_data_len -= prot_bytes;
				prot_iter = data_iter;
				iov_iter_advance(&data_iter, prot_bytes);
978
			}
979
			tag = vhost64_to_cpu(vq, v_req_pi.tag);
980 981 982 983
			task_attr = v_req_pi.task_attr;
			cdb = &v_req_pi.cdb[0];
			lun = ((v_req_pi.lun[2] << 8) | v_req_pi.lun[3]) & 0x3FFF;
		} else {
984
			tag = vhost64_to_cpu(vq, v_req.tag);
985 986 987 988 989
			task_attr = v_req.task_attr;
			cdb = &v_req.cdb[0];
			lun = ((v_req.lun[2] << 8) | v_req.lun[3]) & 0x3FFF;
		}
		/*
990 991 992
		 * Check that the received CDB size does not exceeded our
		 * hardcoded max for vhost-scsi, then get a pre-allocated
		 * cmd descriptor for the new virtio-scsi tag.
993 994 995
		 *
		 * TODO what if cdb was too small for varlen cdb header?
		 */
996
		if (unlikely(scsi_command_size(cdb) > VHOST_SCSI_MAX_CDB_SIZE)) {
997 998
			vq_err(vq, "Received SCSI CDB with command_size: %d that"
				" exceeds SCSI_MAX_VARLEN_CDB_SIZE: %d\n",
999
				scsi_command_size(cdb), VHOST_SCSI_MAX_CDB_SIZE);
1000 1001
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
1002 1003
		}
		cmd = vhost_scsi_get_tag(vq, tpg, cdb, tag, lun, task_attr,
1004 1005
					 exp_data_len + prot_bytes,
					 data_direction);
1006
		if (IS_ERR(cmd)) {
1007
			vq_err(vq, "vhost_scsi_get_tag failed %ld\n",
1008
			       PTR_ERR(cmd));
1009 1010
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
1011
		}
1012 1013
		cmd->tvc_vhost = vs;
		cmd->tvc_vq = vq;
1014
		cmd->tvc_resp_iov = vq->iov[out];
1015
		cmd->tvc_in_iovs = in;
1016 1017

		pr_debug("vhost_scsi got command opcode: %#02x, lun: %d\n",
1018 1019 1020
			 cmd->tvc_cdb[0], cmd->tvc_lun);
		pr_debug("cmd: %p exp_data_len: %d, prot_bytes: %d data_direction:"
			 " %d\n", cmd, exp_data_len, prot_bytes, data_direction);
1021 1022

		if (data_direction != DMA_NONE) {
1023 1024 1025
			ret = vhost_scsi_mapal(cmd,
					       prot_bytes, &prot_iter,
					       exp_data_len, &data_iter);
1026 1027
			if (unlikely(ret)) {
				vq_err(vq, "Failed to map iov to sgl\n");
1028
				vhost_scsi_release_cmd(&cmd->tvc_se_cmd);
1029 1030
				vhost_scsi_send_bad_target(vs, vq, head, out);
				continue;
1031 1032 1033 1034 1035
			}
		}
		/*
		 * Save the descriptor from vhost_get_vq_desc() to be used to
		 * complete the virtio-scsi request in TCM callback context via
1036
		 * vhost_scsi_queue_data_in() and vhost_scsi_queue_status()
1037
		 */
1038
		cmd->tvc_vq_desc = head;
1039
		/*
1040 1041 1042 1043
		 * Dispatch cmd descriptor for cmwq execution in process
		 * context provided by vhost_scsi_workqueue.  This also ensures
		 * cmd is executed on the same kworker CPU as this vhost
		 * thread to gain positive L2 cache locality effects.
1044
		 */
1045 1046
		INIT_WORK(&cmd->work, vhost_scsi_submission_work);
		queue_work(vhost_scsi_workqueue, &cmd->work);
1047
	}
1048
out:
1049
	mutex_unlock(&vq->mutex);
1050 1051 1052 1053
}

static void vhost_scsi_ctl_handle_kick(struct vhost_work *work)
{
1054
	pr_debug("%s: The handling func for control queue.\n", __func__);
1055 1056
}

1057
static void
1058 1059
vhost_scsi_send_evt(struct vhost_scsi *vs,
		   struct vhost_scsi_tpg *tpg,
1060 1061 1062
		   struct se_lun *lun,
		   u32 event,
		   u32 reason)
1063
{
1064
	struct vhost_scsi_evt *evt;
1065

1066
	evt = vhost_scsi_allocate_evt(vs, event, reason);
1067 1068 1069 1070 1071 1072 1073 1074 1075 1076
	if (!evt)
		return;

	if (tpg && lun) {
		/* TODO: share lun setup code with virtio-scsi.ko */
		/*
		 * Note: evt->event is zeroed when we allocate it and
		 * lun[4-7] need to be zero according to virtio-scsi spec.
		 */
		evt->event.lun[0] = 0x01;
1077
		evt->event.lun[1] = tpg->tport_tpgt;
1078 1079 1080 1081 1082 1083 1084 1085 1086
		if (lun->unpacked_lun >= 256)
			evt->event.lun[2] = lun->unpacked_lun >> 8 | 0x40 ;
		evt->event.lun[3] = lun->unpacked_lun & 0xFF;
	}

	llist_add(&evt->list, &vs->vs_event_list);
	vhost_work_queue(&vs->dev, &vs->vs_event_work);
}

1087 1088
static void vhost_scsi_evt_handle_kick(struct vhost_work *work)
{
1089 1090 1091 1092 1093 1094 1095 1096 1097
	struct vhost_virtqueue *vq = container_of(work, struct vhost_virtqueue,
						poll.work);
	struct vhost_scsi *vs = container_of(vq->dev, struct vhost_scsi, dev);

	mutex_lock(&vq->mutex);
	if (!vq->private_data)
		goto out;

	if (vs->vs_events_missed)
1098
		vhost_scsi_send_evt(vs, NULL, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
1099 1100
out:
	mutex_unlock(&vq->mutex);
1101 1102 1103 1104 1105 1106 1107 1108
}

static void vhost_scsi_handle_kick(struct vhost_work *work)
{
	struct vhost_virtqueue *vq = container_of(work, struct vhost_virtqueue,
						poll.work);
	struct vhost_scsi *vs = container_of(vq->dev, struct vhost_scsi, dev);

A
Asias He 已提交
1109
	vhost_scsi_handle_vq(vs, vq);
1110 1111
}

1112 1113
static void vhost_scsi_flush_vq(struct vhost_scsi *vs, int index)
{
1114
	vhost_poll_flush(&vs->vqs[index].vq.poll);
1115 1116
}

1117
/* Callers must hold dev mutex */
1118 1119
static void vhost_scsi_flush(struct vhost_scsi *vs)
{
1120
	struct vhost_scsi_inflight *old_inflight[VHOST_SCSI_MAX_VQ];
1121 1122
	int i;

1123
	/* Init new inflight and remember the old inflight */
1124
	vhost_scsi_init_inflight(vs, old_inflight);
1125 1126 1127 1128 1129 1130 1131

	/*
	 * The inflight->kref was initialized to 1. We decrement it here to
	 * indicate the start of the flush operation so that it will reach 0
	 * when all the reqs are finished.
	 */
	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++)
1132
		kref_put(&old_inflight[i]->kref, vhost_scsi_done_inflight);
1133 1134

	/* Flush both the vhost poll and vhost work */
1135 1136 1137
	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++)
		vhost_scsi_flush_vq(vs, i);
	vhost_work_flush(&vs->dev, &vs->vs_completion_work);
1138
	vhost_work_flush(&vs->dev, &vs->vs_event_work);
1139 1140 1141 1142

	/* Wait for all reqs issued before the flush to be finished */
	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++)
		wait_for_completion(&old_inflight[i]->comp);
1143 1144
}

1145 1146
/*
 * Called from vhost_scsi_ioctl() context to walk the list of available
1147
 * vhost_scsi_tpg with an active struct vhost_scsi_nexus
1148 1149
 *
 *  The lock nesting rule is:
1150
 *    vhost_scsi_mutex -> vs->dev.mutex -> tpg->tv_tpg_mutex -> vq->mutex
1151
 */
1152 1153 1154
static int
vhost_scsi_set_endpoint(struct vhost_scsi *vs,
			struct vhost_scsi_target *t)
1155
{
1156
	struct se_portal_group *se_tpg;
1157 1158 1159
	struct vhost_scsi_tport *tv_tport;
	struct vhost_scsi_tpg *tpg;
	struct vhost_scsi_tpg **vs_tpg;
1160 1161
	struct vhost_virtqueue *vq;
	int index, ret, i, len;
A
Asias He 已提交
1162
	bool match = false;
1163

1164
	mutex_lock(&vhost_scsi_mutex);
1165
	mutex_lock(&vs->dev.mutex);
1166

1167 1168 1169
	/* Verify that ring has been setup correctly. */
	for (index = 0; index < vs->dev.nvqs; ++index) {
		/* Verify that ring has been setup correctly. */
1170
		if (!vhost_vq_access_ok(&vs->vqs[index].vq)) {
1171 1172
			ret = -EFAULT;
			goto out;
1173 1174 1175
		}
	}

1176 1177 1178
	len = sizeof(vs_tpg[0]) * VHOST_SCSI_MAX_TARGET;
	vs_tpg = kzalloc(len, GFP_KERNEL);
	if (!vs_tpg) {
1179 1180
		ret = -ENOMEM;
		goto out;
1181 1182 1183 1184
	}
	if (vs->vs_tpg)
		memcpy(vs_tpg, vs->vs_tpg, len);

1185
	list_for_each_entry(tpg, &vhost_scsi_list, tv_tpg_list) {
1186 1187 1188
		mutex_lock(&tpg->tv_tpg_mutex);
		if (!tpg->tpg_nexus) {
			mutex_unlock(&tpg->tv_tpg_mutex);
1189 1190
			continue;
		}
1191 1192
		if (tpg->tv_tpg_vhost_count != 0) {
			mutex_unlock(&tpg->tv_tpg_mutex);
1193 1194
			continue;
		}
1195
		tv_tport = tpg->tport;
1196

A
Asias He 已提交
1197
		if (!strcmp(tv_tport->tport_name, t->vhost_wwpn)) {
1198
			if (vs->vs_tpg && vs->vs_tpg[tpg->tport_tpgt]) {
1199
				kfree(vs_tpg);
1200
				mutex_unlock(&tpg->tv_tpg_mutex);
1201 1202
				ret = -EEXIST;
				goto out;
1203
			}
1204 1205 1206 1207 1208 1209 1210
			/*
			 * In order to ensure individual vhost-scsi configfs
			 * groups cannot be removed while in use by vhost ioctl,
			 * go ahead and take an explicit se_tpg->tpg_group.cg_item
			 * dependency now.
			 */
			se_tpg = &tpg->se_tpg;
1211
			ret = target_depend_item(&se_tpg->tpg_group.cg_item);
1212 1213 1214 1215 1216 1217
			if (ret) {
				pr_warn("configfs_depend_item() failed: %d\n", ret);
				kfree(vs_tpg);
				mutex_unlock(&tpg->tv_tpg_mutex);
				goto out;
			}
1218 1219 1220
			tpg->tv_tpg_vhost_count++;
			tpg->vhost_scsi = vs;
			vs_tpg[tpg->tport_tpgt] = tpg;
1221
			smp_mb__after_atomic();
A
Asias He 已提交
1222
			match = true;
1223
		}
1224
		mutex_unlock(&tpg->tv_tpg_mutex);
1225
	}
A
Asias He 已提交
1226 1227 1228 1229

	if (match) {
		memcpy(vs->vs_vhost_wwpn, t->vhost_wwpn,
		       sizeof(vs->vs_vhost_wwpn));
1230
		for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
1231
			vq = &vs->vqs[i].vq;
1232
			mutex_lock(&vq->mutex);
A
Asias He 已提交
1233
			vq->private_data = vs_tpg;
G
Greg Kurz 已提交
1234
			vhost_vq_init_access(vq);
1235 1236
			mutex_unlock(&vq->mutex);
		}
A
Asias He 已提交
1237 1238 1239 1240 1241
		ret = 0;
	} else {
		ret = -EEXIST;
	}

1242 1243 1244 1245 1246 1247 1248 1249
	/*
	 * Act as synchronize_rcu to make sure access to
	 * old vs->vs_tpg is finished.
	 */
	vhost_scsi_flush(vs);
	kfree(vs->vs_tpg);
	vs->vs_tpg = vs_tpg;

1250
out:
A
Asias He 已提交
1251
	mutex_unlock(&vs->dev.mutex);
1252
	mutex_unlock(&vhost_scsi_mutex);
A
Asias He 已提交
1253
	return ret;
1254 1255
}

1256 1257 1258
static int
vhost_scsi_clear_endpoint(struct vhost_scsi *vs,
			  struct vhost_scsi_target *t)
1259
{
1260
	struct se_portal_group *se_tpg;
1261 1262
	struct vhost_scsi_tport *tv_tport;
	struct vhost_scsi_tpg *tpg;
1263 1264
	struct vhost_virtqueue *vq;
	bool match = false;
A
Asias He 已提交
1265 1266
	int index, ret, i;
	u8 target;
1267

1268
	mutex_lock(&vhost_scsi_mutex);
1269 1270 1271
	mutex_lock(&vs->dev.mutex);
	/* Verify that ring has been setup correctly. */
	for (index = 0; index < vs->dev.nvqs; ++index) {
1272
		if (!vhost_vq_access_ok(&vs->vqs[index].vq)) {
1273
			ret = -EFAULT;
1274
			goto err_dev;
1275 1276
		}
	}
1277 1278

	if (!vs->vs_tpg) {
1279 1280
		ret = 0;
		goto err_dev;
1281 1282
	}

A
Asias He 已提交
1283 1284
	for (i = 0; i < VHOST_SCSI_MAX_TARGET; i++) {
		target = i;
1285 1286
		tpg = vs->vs_tpg[target];
		if (!tpg)
A
Asias He 已提交
1287 1288
			continue;

1289 1290
		mutex_lock(&tpg->tv_tpg_mutex);
		tv_tport = tpg->tport;
A
Asias He 已提交
1291 1292
		if (!tv_tport) {
			ret = -ENODEV;
1293
			goto err_tpg;
A
Asias He 已提交
1294 1295 1296
		}

		if (strcmp(tv_tport->tport_name, t->vhost_wwpn)) {
1297
			pr_warn("tv_tport->tport_name: %s, tpg->tport_tpgt: %hu"
A
Asias He 已提交
1298
				" does not match t->vhost_wwpn: %s, t->vhost_tpgt: %hu\n",
1299
				tv_tport->tport_name, tpg->tport_tpgt,
A
Asias He 已提交
1300 1301
				t->vhost_wwpn, t->vhost_tpgt);
			ret = -EINVAL;
1302
			goto err_tpg;
A
Asias He 已提交
1303
		}
1304 1305
		tpg->tv_tpg_vhost_count--;
		tpg->vhost_scsi = NULL;
A
Asias He 已提交
1306
		vs->vs_tpg[target] = NULL;
1307
		match = true;
1308
		mutex_unlock(&tpg->tv_tpg_mutex);
1309 1310 1311 1312 1313
		/*
		 * Release se_tpg->tpg_group.cg_item configfs dependency now
		 * to allow vhost-scsi WWPN se_tpg->tpg_group shutdown to occur.
		 */
		se_tpg = &tpg->se_tpg;
1314
		target_undepend_item(&se_tpg->tpg_group.cg_item);
1315
	}
1316 1317
	if (match) {
		for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
1318
			vq = &vs->vqs[i].vq;
1319
			mutex_lock(&vq->mutex);
A
Asias He 已提交
1320
			vq->private_data = NULL;
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330
			mutex_unlock(&vq->mutex);
		}
	}
	/*
	 * Act as synchronize_rcu to make sure access to
	 * old vs->vs_tpg is finished.
	 */
	vhost_scsi_flush(vs);
	kfree(vs->vs_tpg);
	vs->vs_tpg = NULL;
1331
	WARN_ON(vs->vs_events_nr);
1332
	mutex_unlock(&vs->dev.mutex);
1333
	mutex_unlock(&vhost_scsi_mutex);
1334
	return 0;
1335

1336
err_tpg:
1337
	mutex_unlock(&tpg->tv_tpg_mutex);
1338
err_dev:
1339
	mutex_unlock(&vs->dev.mutex);
1340
	mutex_unlock(&vhost_scsi_mutex);
1341
	return ret;
1342 1343
}

1344 1345
static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)
{
1346 1347 1348
	struct vhost_virtqueue *vq;
	int i;

1349 1350 1351 1352 1353 1354 1355 1356 1357
	if (features & ~VHOST_SCSI_FEATURES)
		return -EOPNOTSUPP;

	mutex_lock(&vs->dev.mutex);
	if ((features & (1 << VHOST_F_LOG_ALL)) &&
	    !vhost_log_access_ok(&vs->dev)) {
		mutex_unlock(&vs->dev.mutex);
		return -EFAULT;
	}
1358 1359 1360 1361 1362 1363 1364

	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
		vq = &vs->vqs[i].vq;
		mutex_lock(&vq->mutex);
		vq->acked_features = features;
		mutex_unlock(&vq->mutex);
	}
1365 1366 1367 1368
	mutex_unlock(&vs->dev.mutex);
	return 0;
}

1369 1370
static int vhost_scsi_open(struct inode *inode, struct file *f)
{
1371
	struct vhost_scsi *vs;
1372
	struct vhost_virtqueue **vqs;
1373
	int r = -ENOMEM, i;
1374

1375
	vs = kzalloc(sizeof(*vs), GFP_KERNEL | __GFP_NOWARN | __GFP_RETRY_MAYFAIL);
1376 1377 1378 1379 1380
	if (!vs) {
		vs = vzalloc(sizeof(*vs));
		if (!vs)
			goto err_vs;
	}
1381

1382
	vqs = kmalloc(VHOST_SCSI_MAX_VQ * sizeof(*vqs), GFP_KERNEL);
1383 1384
	if (!vqs)
		goto err_vqs;
1385

1386
	vhost_work_init(&vs->vs_completion_work, vhost_scsi_complete_cmd_work);
1387
	vhost_work_init(&vs->vs_event_work, vhost_scsi_evt_work);
1388

1389 1390
	vs->vs_events_nr = 0;
	vs->vs_events_missed = false;
1391

1392 1393 1394 1395
	vqs[VHOST_SCSI_VQ_CTL] = &vs->vqs[VHOST_SCSI_VQ_CTL].vq;
	vqs[VHOST_SCSI_VQ_EVT] = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
	vs->vqs[VHOST_SCSI_VQ_CTL].vq.handle_kick = vhost_scsi_ctl_handle_kick;
	vs->vqs[VHOST_SCSI_VQ_EVT].vq.handle_kick = vhost_scsi_evt_handle_kick;
1396
	for (i = VHOST_SCSI_VQ_IO; i < VHOST_SCSI_MAX_VQ; i++) {
1397 1398
		vqs[i] = &vs->vqs[i].vq;
		vs->vqs[i].vq.handle_kick = vhost_scsi_handle_kick;
1399
	}
Z
Zhi Yong Wu 已提交
1400
	vhost_dev_init(&vs->dev, vqs, VHOST_SCSI_MAX_VQ);
1401

1402
	vhost_scsi_init_inflight(vs, NULL);
1403

1404
	f->private_data = vs;
1405
	return 0;
1406 1407

err_vqs:
1408
	kvfree(vs);
1409 1410
err_vs:
	return r;
1411 1412 1413 1414
}

static int vhost_scsi_release(struct inode *inode, struct file *f)
{
1415
	struct vhost_scsi *vs = f->private_data;
A
Asias He 已提交
1416
	struct vhost_scsi_target t;
1417

1418 1419 1420 1421 1422 1423
	mutex_lock(&vs->dev.mutex);
	memcpy(t.vhost_wwpn, vs->vs_vhost_wwpn, sizeof(t.vhost_wwpn));
	mutex_unlock(&vs->dev.mutex);
	vhost_scsi_clear_endpoint(vs, &t);
	vhost_dev_stop(&vs->dev);
	vhost_dev_cleanup(&vs->dev, false);
1424
	/* Jobs can re-queue themselves in evt kick handler. Do extra flush. */
1425 1426
	vhost_scsi_flush(vs);
	kfree(vs->dev.vqs);
1427
	kvfree(vs);
1428 1429 1430
	return 0;
}

1431 1432 1433 1434
static long
vhost_scsi_ioctl(struct file *f,
		 unsigned int ioctl,
		 unsigned long arg)
1435 1436 1437 1438 1439
{
	struct vhost_scsi *vs = f->private_data;
	struct vhost_scsi_target backend;
	void __user *argp = (void __user *)arg;
	u64 __user *featurep = argp;
1440 1441
	u32 __user *eventsp = argp;
	u32 events_missed;
1442
	u64 features;
1443
	int r, abi_version = VHOST_SCSI_ABI_VERSION;
1444
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
1445 1446 1447 1448 1449

	switch (ioctl) {
	case VHOST_SCSI_SET_ENDPOINT:
		if (copy_from_user(&backend, argp, sizeof backend))
			return -EFAULT;
1450 1451
		if (backend.reserved != 0)
			return -EOPNOTSUPP;
1452 1453 1454 1455 1456

		return vhost_scsi_set_endpoint(vs, &backend);
	case VHOST_SCSI_CLEAR_ENDPOINT:
		if (copy_from_user(&backend, argp, sizeof backend))
			return -EFAULT;
1457 1458
		if (backend.reserved != 0)
			return -EOPNOTSUPP;
1459 1460 1461

		return vhost_scsi_clear_endpoint(vs, &backend);
	case VHOST_SCSI_GET_ABI_VERSION:
1462
		if (copy_to_user(argp, &abi_version, sizeof abi_version))
1463 1464
			return -EFAULT;
		return 0;
1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478
	case VHOST_SCSI_SET_EVENTS_MISSED:
		if (get_user(events_missed, eventsp))
			return -EFAULT;
		mutex_lock(&vq->mutex);
		vs->vs_events_missed = events_missed;
		mutex_unlock(&vq->mutex);
		return 0;
	case VHOST_SCSI_GET_EVENTS_MISSED:
		mutex_lock(&vq->mutex);
		events_missed = vs->vs_events_missed;
		mutex_unlock(&vq->mutex);
		if (put_user(events_missed, eventsp))
			return -EFAULT;
		return 0;
1479
	case VHOST_GET_FEATURES:
1480
		features = VHOST_SCSI_FEATURES;
1481 1482 1483 1484 1485 1486 1487 1488 1489
		if (copy_to_user(featurep, &features, sizeof features))
			return -EFAULT;
		return 0;
	case VHOST_SET_FEATURES:
		if (copy_from_user(&features, featurep, sizeof features))
			return -EFAULT;
		return vhost_scsi_set_features(vs, features);
	default:
		mutex_lock(&vs->dev.mutex);
1490 1491 1492 1493
		r = vhost_dev_ioctl(&vs->dev, ioctl, argp);
		/* TODO: flush backend after dev ioctl. */
		if (r == -ENOIOCTLCMD)
			r = vhost_vring_ioctl(&vs->dev, ioctl, argp);
1494 1495 1496 1497 1498
		mutex_unlock(&vs->dev.mutex);
		return r;
	}
}

1499 1500 1501 1502 1503 1504 1505 1506
#ifdef CONFIG_COMPAT
static long vhost_scsi_compat_ioctl(struct file *f, unsigned int ioctl,
				unsigned long arg)
{
	return vhost_scsi_ioctl(f, ioctl, (unsigned long)compat_ptr(arg));
}
#endif

1507 1508 1509 1510
static const struct file_operations vhost_scsi_fops = {
	.owner          = THIS_MODULE,
	.release        = vhost_scsi_release,
	.unlocked_ioctl = vhost_scsi_ioctl,
1511 1512 1513
#ifdef CONFIG_COMPAT
	.compat_ioctl	= vhost_scsi_compat_ioctl,
#endif
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528
	.open           = vhost_scsi_open,
	.llseek		= noop_llseek,
};

static struct miscdevice vhost_scsi_misc = {
	MISC_DYNAMIC_MINOR,
	"vhost-scsi",
	&vhost_scsi_fops,
};

static int __init vhost_scsi_register(void)
{
	return misc_register(&vhost_scsi_misc);
}

1529
static void vhost_scsi_deregister(void)
1530
{
1531
	misc_deregister(&vhost_scsi_misc);
1532 1533
}

1534
static char *vhost_scsi_dump_proto_id(struct vhost_scsi_tport *tport)
1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549
{
	switch (tport->tport_proto_id) {
	case SCSI_PROTOCOL_SAS:
		return "SAS";
	case SCSI_PROTOCOL_FCP:
		return "FCP";
	case SCSI_PROTOCOL_ISCSI:
		return "iSCSI";
	default:
		break;
	}

	return "Unknown";
}

1550
static void
1551
vhost_scsi_do_plug(struct vhost_scsi_tpg *tpg,
1552
		  struct se_lun *lun, bool plug)
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568
{

	struct vhost_scsi *vs = tpg->vhost_scsi;
	struct vhost_virtqueue *vq;
	u32 reason;

	if (!vs)
		return;

	mutex_lock(&vs->dev.mutex);

	if (plug)
		reason = VIRTIO_SCSI_EVT_RESET_RESCAN;
	else
		reason = VIRTIO_SCSI_EVT_RESET_REMOVED;

1569
	vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
1570
	mutex_lock(&vq->mutex);
1571
	if (vhost_has_feature(vq, VIRTIO_SCSI_F_HOTPLUG))
1572
		vhost_scsi_send_evt(vs, tpg, lun,
1573
				   VIRTIO_SCSI_T_TRANSPORT_RESET, reason);
1574 1575 1576 1577
	mutex_unlock(&vq->mutex);
	mutex_unlock(&vs->dev.mutex);
}

1578
static void vhost_scsi_hotplug(struct vhost_scsi_tpg *tpg, struct se_lun *lun)
1579
{
1580
	vhost_scsi_do_plug(tpg, lun, true);
1581 1582
}

1583
static void vhost_scsi_hotunplug(struct vhost_scsi_tpg *tpg, struct se_lun *lun)
1584
{
1585
	vhost_scsi_do_plug(tpg, lun, false);
1586 1587
}

1588
static int vhost_scsi_port_link(struct se_portal_group *se_tpg,
1589
			       struct se_lun *lun)
1590
{
1591 1592
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
1593

1594
	mutex_lock(&vhost_scsi_mutex);
1595

1596 1597 1598
	mutex_lock(&tpg->tv_tpg_mutex);
	tpg->tv_tpg_port_count++;
	mutex_unlock(&tpg->tv_tpg_mutex);
1599

1600
	vhost_scsi_hotplug(tpg, lun);
1601

1602
	mutex_unlock(&vhost_scsi_mutex);
1603

1604 1605 1606
	return 0;
}

1607
static void vhost_scsi_port_unlink(struct se_portal_group *se_tpg,
1608
				  struct se_lun *lun)
1609
{
1610 1611
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
1612

1613
	mutex_lock(&vhost_scsi_mutex);
1614

1615 1616 1617
	mutex_lock(&tpg->tv_tpg_mutex);
	tpg->tv_tpg_port_count--;
	mutex_unlock(&tpg->tv_tpg_mutex);
1618

1619
	vhost_scsi_hotunplug(tpg, lun);
1620

1621
	mutex_unlock(&vhost_scsi_mutex);
1622 1623
}

1624
static void vhost_scsi_free_cmd_map_res(struct se_session *se_sess)
1625
{
1626
	struct vhost_scsi_cmd *tv_cmd;
1627 1628 1629 1630 1631
	unsigned int i;

	if (!se_sess->sess_cmd_map)
		return;

1632 1633
	for (i = 0; i < VHOST_SCSI_DEFAULT_TAGS; i++) {
		tv_cmd = &((struct vhost_scsi_cmd *)se_sess->sess_cmd_map)[i];
1634 1635

		kfree(tv_cmd->tvc_sgl);
1636
		kfree(tv_cmd->tvc_prot_sgl);
1637 1638 1639 1640
		kfree(tv_cmd->tvc_upages);
	}
}

1641 1642
static ssize_t vhost_scsi_tpg_attrib_fabric_prot_type_store(
		struct config_item *item, const char *page, size_t count)
1643
{
1644
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
	unsigned long val;
	int ret = kstrtoul(page, 0, &val);

	if (ret) {
		pr_err("kstrtoul() returned %d for fabric_prot_type\n", ret);
		return ret;
	}
	if (val != 0 && val != 1 && val != 3) {
		pr_err("Invalid vhost_scsi fabric_prot_type: %lu\n", val);
		return -EINVAL;
	}
	tpg->tv_fabric_prot_type = val;

	return count;
}

1663 1664
static ssize_t vhost_scsi_tpg_attrib_fabric_prot_type_show(
		struct config_item *item, char *page)
1665
{
1666
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
1667 1668 1669 1670 1671
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);

	return sprintf(page, "%d\n", tpg->tv_fabric_prot_type);
}
1672 1673

CONFIGFS_ATTR(vhost_scsi_tpg_attrib_, fabric_prot_type);
1674 1675

static struct configfs_attribute *vhost_scsi_tpg_attrib_attrs[] = {
1676
	&vhost_scsi_tpg_attrib_attr_fabric_prot_type,
1677 1678 1679
	NULL,
};

1680 1681
static int vhost_scsi_nexus_cb(struct se_portal_group *se_tpg,
			       struct se_session *se_sess, void *p)
1682
{
1683
	struct vhost_scsi_cmd *tv_cmd;
1684
	unsigned int i;
1685

1686 1687
	for (i = 0; i < VHOST_SCSI_DEFAULT_TAGS; i++) {
		tv_cmd = &((struct vhost_scsi_cmd *)se_sess->sess_cmd_map)[i];
1688 1689

		tv_cmd->tvc_sgl = kzalloc(sizeof(struct scatterlist) *
1690
					VHOST_SCSI_PREALLOC_SGLS, GFP_KERNEL);
1691 1692 1693 1694 1695 1696
		if (!tv_cmd->tvc_sgl) {
			pr_err("Unable to allocate tv_cmd->tvc_sgl\n");
			goto out;
		}

		tv_cmd->tvc_upages = kzalloc(sizeof(struct page *) *
1697
				VHOST_SCSI_PREALLOC_UPAGES, GFP_KERNEL);
1698 1699 1700 1701
		if (!tv_cmd->tvc_upages) {
			pr_err("Unable to allocate tv_cmd->tvc_upages\n");
			goto out;
		}
1702 1703

		tv_cmd->tvc_prot_sgl = kzalloc(sizeof(struct scatterlist) *
1704
				VHOST_SCSI_PREALLOC_PROT_SGLS, GFP_KERNEL);
1705 1706 1707 1708
		if (!tv_cmd->tvc_prot_sgl) {
			pr_err("Unable to allocate tv_cmd->tvc_prot_sgl\n");
			goto out;
		}
1709
	}
1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
	return 0;
out:
	vhost_scsi_free_cmd_map_res(se_sess);
	return -ENOMEM;
}

static int vhost_scsi_make_nexus(struct vhost_scsi_tpg *tpg,
				const char *name)
{
	struct vhost_scsi_nexus *tv_nexus;

	mutex_lock(&tpg->tv_tpg_mutex);
	if (tpg->tpg_nexus) {
		mutex_unlock(&tpg->tv_tpg_mutex);
		pr_debug("tpg->tpg_nexus already exists\n");
		return -EEXIST;
	}

	tv_nexus = kzalloc(sizeof(struct vhost_scsi_nexus), GFP_KERNEL);
	if (!tv_nexus) {
		mutex_unlock(&tpg->tv_tpg_mutex);
		pr_err("Unable to allocate struct vhost_scsi_nexus\n");
		return -ENOMEM;
	}
1734 1735
	/*
	 * Since we are running in 'demo mode' this call with generate a
1736
	 * struct se_node_acl for the vhost_scsi struct se_portal_group with
1737 1738
	 * the SCSI Initiator port name of the passed configfs group 'name'.
	 */
1739 1740 1741 1742 1743 1744 1745
	tv_nexus->tvn_se_sess = target_alloc_session(&tpg->se_tpg,
					VHOST_SCSI_DEFAULT_TAGS,
					sizeof(struct vhost_scsi_cmd),
					TARGET_PROT_DIN_PASS | TARGET_PROT_DOUT_PASS,
					(unsigned char *)name, tv_nexus,
					vhost_scsi_nexus_cb);
	if (IS_ERR(tv_nexus->tvn_se_sess)) {
1746
		mutex_unlock(&tpg->tv_tpg_mutex);
1747 1748
		kfree(tv_nexus);
		return -ENOMEM;
1749
	}
1750
	tpg->tpg_nexus = tv_nexus;
1751

1752
	mutex_unlock(&tpg->tv_tpg_mutex);
1753 1754 1755
	return 0;
}

1756
static int vhost_scsi_drop_nexus(struct vhost_scsi_tpg *tpg)
1757 1758
{
	struct se_session *se_sess;
1759
	struct vhost_scsi_nexus *tv_nexus;
1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773

	mutex_lock(&tpg->tv_tpg_mutex);
	tv_nexus = tpg->tpg_nexus;
	if (!tv_nexus) {
		mutex_unlock(&tpg->tv_tpg_mutex);
		return -ENODEV;
	}

	se_sess = tv_nexus->tvn_se_sess;
	if (!se_sess) {
		mutex_unlock(&tpg->tv_tpg_mutex);
		return -ENODEV;
	}

1774
	if (tpg->tv_tpg_port_count != 0) {
1775
		mutex_unlock(&tpg->tv_tpg_mutex);
1776
		pr_err("Unable to remove TCM_vhost I_T Nexus with"
1777
			" active TPG port count: %d\n",
1778 1779
			tpg->tv_tpg_port_count);
		return -EBUSY;
1780 1781
	}

1782
	if (tpg->tv_tpg_vhost_count != 0) {
1783
		mutex_unlock(&tpg->tv_tpg_mutex);
1784
		pr_err("Unable to remove TCM_vhost I_T Nexus with"
1785
			" active TPG vhost count: %d\n",
1786 1787
			tpg->tv_tpg_vhost_count);
		return -EBUSY;
1788 1789
	}

1790
	pr_debug("TCM_vhost_ConfigFS: Removing I_T Nexus to emulated"
1791
		" %s Initiator Port: %s\n", vhost_scsi_dump_proto_id(tpg->tport),
1792
		tv_nexus->tvn_se_sess->se_node_acl->initiatorname);
1793

1794
	vhost_scsi_free_cmd_map_res(se_sess);
1795
	/*
1796
	 * Release the SCSI I_T Nexus to the emulated vhost Target Port
1797 1798 1799 1800 1801 1802 1803 1804 1805
	 */
	transport_deregister_session(tv_nexus->tvn_se_sess);
	tpg->tpg_nexus = NULL;
	mutex_unlock(&tpg->tv_tpg_mutex);

	kfree(tv_nexus);
	return 0;
}

1806
static ssize_t vhost_scsi_tpg_nexus_show(struct config_item *item, char *page)
1807
{
1808
	struct se_portal_group *se_tpg = to_tpg(item);
1809 1810 1811
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
	struct vhost_scsi_nexus *tv_nexus;
1812 1813
	ssize_t ret;

1814 1815
	mutex_lock(&tpg->tv_tpg_mutex);
	tv_nexus = tpg->tpg_nexus;
1816
	if (!tv_nexus) {
1817
		mutex_unlock(&tpg->tv_tpg_mutex);
1818 1819 1820 1821
		return -ENODEV;
	}
	ret = snprintf(page, PAGE_SIZE, "%s\n",
			tv_nexus->tvn_se_sess->se_node_acl->initiatorname);
1822
	mutex_unlock(&tpg->tv_tpg_mutex);
1823 1824 1825 1826

	return ret;
}

1827 1828
static ssize_t vhost_scsi_tpg_nexus_store(struct config_item *item,
		const char *page, size_t count)
1829
{
1830
	struct se_portal_group *se_tpg = to_tpg(item);
1831 1832 1833 1834
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
	struct vhost_scsi_tport *tport_wwn = tpg->tport;
	unsigned char i_port[VHOST_SCSI_NAMELEN], *ptr, *port_ptr;
1835 1836 1837 1838 1839
	int ret;
	/*
	 * Shutdown the active I_T nexus if 'NULL' is passed..
	 */
	if (!strncmp(page, "NULL", 4)) {
1840
		ret = vhost_scsi_drop_nexus(tpg);
1841 1842 1843 1844
		return (!ret) ? count : ret;
	}
	/*
	 * Otherwise make sure the passed virtual Initiator port WWN matches
1845 1846
	 * the fabric protocol_id set in vhost_scsi_make_tport(), and call
	 * vhost_scsi_make_nexus().
1847
	 */
1848
	if (strlen(page) >= VHOST_SCSI_NAMELEN) {
1849
		pr_err("Emulated NAA Sas Address: %s, exceeds"
1850
				" max: %d\n", page, VHOST_SCSI_NAMELEN);
1851 1852
		return -EINVAL;
	}
1853
	snprintf(&i_port[0], VHOST_SCSI_NAMELEN, "%s", page);
1854 1855 1856 1857 1858 1859

	ptr = strstr(i_port, "naa.");
	if (ptr) {
		if (tport_wwn->tport_proto_id != SCSI_PROTOCOL_SAS) {
			pr_err("Passed SAS Initiator Port %s does not"
				" match target port protoid: %s\n", i_port,
1860
				vhost_scsi_dump_proto_id(tport_wwn));
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870
			return -EINVAL;
		}
		port_ptr = &i_port[0];
		goto check_newline;
	}
	ptr = strstr(i_port, "fc.");
	if (ptr) {
		if (tport_wwn->tport_proto_id != SCSI_PROTOCOL_FCP) {
			pr_err("Passed FCP Initiator Port %s does not"
				" match target port protoid: %s\n", i_port,
1871
				vhost_scsi_dump_proto_id(tport_wwn));
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881
			return -EINVAL;
		}
		port_ptr = &i_port[3]; /* Skip over "fc." */
		goto check_newline;
	}
	ptr = strstr(i_port, "iqn.");
	if (ptr) {
		if (tport_wwn->tport_proto_id != SCSI_PROTOCOL_ISCSI) {
			pr_err("Passed iSCSI Initiator Port %s does not"
				" match target port protoid: %s\n", i_port,
1882
				vhost_scsi_dump_proto_id(tport_wwn));
1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897
			return -EINVAL;
		}
		port_ptr = &i_port[0];
		goto check_newline;
	}
	pr_err("Unable to locate prefix for emulated Initiator Port:"
			" %s\n", i_port);
	return -EINVAL;
	/*
	 * Clear any trailing newline for the NAA WWN
	 */
check_newline:
	if (i_port[strlen(i_port)-1] == '\n')
		i_port[strlen(i_port)-1] = '\0';

1898
	ret = vhost_scsi_make_nexus(tpg, port_ptr);
1899 1900 1901 1902 1903 1904
	if (ret < 0)
		return ret;

	return count;
}

1905
CONFIGFS_ATTR(vhost_scsi_tpg_, nexus);
1906

1907
static struct configfs_attribute *vhost_scsi_tpg_attrs[] = {
1908
	&vhost_scsi_tpg_attr_nexus,
1909 1910 1911
	NULL,
};

1912
static struct se_portal_group *
1913
vhost_scsi_make_tpg(struct se_wwn *wwn,
1914 1915
		   struct config_group *group,
		   const char *name)
1916
{
1917 1918
	struct vhost_scsi_tport *tport = container_of(wwn,
			struct vhost_scsi_tport, tport_wwn);
1919

1920
	struct vhost_scsi_tpg *tpg;
1921
	u16 tpgt;
1922 1923 1924 1925
	int ret;

	if (strstr(name, "tpgt_") != name)
		return ERR_PTR(-EINVAL);
1926
	if (kstrtou16(name + 5, 10, &tpgt) || tpgt >= VHOST_SCSI_MAX_TARGET)
1927 1928
		return ERR_PTR(-EINVAL);

1929
	tpg = kzalloc(sizeof(struct vhost_scsi_tpg), GFP_KERNEL);
1930
	if (!tpg) {
1931
		pr_err("Unable to allocate struct vhost_scsi_tpg");
1932 1933 1934 1935 1936 1937 1938
		return ERR_PTR(-ENOMEM);
	}
	mutex_init(&tpg->tv_tpg_mutex);
	INIT_LIST_HEAD(&tpg->tv_tpg_list);
	tpg->tport = tport;
	tpg->tport_tpgt = tpgt;

1939
	ret = core_tpg_register(wwn, &tpg->se_tpg, tport->tport_proto_id);
1940 1941 1942 1943
	if (ret < 0) {
		kfree(tpg);
		return NULL;
	}
1944 1945 1946
	mutex_lock(&vhost_scsi_mutex);
	list_add_tail(&tpg->tv_tpg_list, &vhost_scsi_list);
	mutex_unlock(&vhost_scsi_mutex);
1947 1948 1949 1950

	return &tpg->se_tpg;
}

1951
static void vhost_scsi_drop_tpg(struct se_portal_group *se_tpg)
1952
{
1953 1954
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
1955

1956
	mutex_lock(&vhost_scsi_mutex);
1957
	list_del(&tpg->tv_tpg_list);
1958
	mutex_unlock(&vhost_scsi_mutex);
1959
	/*
1960
	 * Release the virtual I_T Nexus for this vhost TPG
1961
	 */
1962
	vhost_scsi_drop_nexus(tpg);
1963 1964 1965 1966 1967 1968 1969
	/*
	 * Deregister the se_tpg from TCM..
	 */
	core_tpg_deregister(se_tpg);
	kfree(tpg);
}

1970
static struct se_wwn *
1971
vhost_scsi_make_tport(struct target_fabric_configfs *tf,
1972 1973
		     struct config_group *group,
		     const char *name)
1974
{
1975
	struct vhost_scsi_tport *tport;
1976 1977 1978 1979
	char *ptr;
	u64 wwpn = 0;
	int off = 0;

1980
	/* if (vhost_scsi_parse_wwn(name, &wwpn, 1) < 0)
1981 1982
		return ERR_PTR(-EINVAL); */

1983
	tport = kzalloc(sizeof(struct vhost_scsi_tport), GFP_KERNEL);
1984
	if (!tport) {
1985
		pr_err("Unable to allocate struct vhost_scsi_tport");
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
		return ERR_PTR(-ENOMEM);
	}
	tport->tport_wwpn = wwpn;
	/*
	 * Determine the emulated Protocol Identifier and Target Port Name
	 * based on the incoming configfs directory name.
	 */
	ptr = strstr(name, "naa.");
	if (ptr) {
		tport->tport_proto_id = SCSI_PROTOCOL_SAS;
		goto check_len;
	}
	ptr = strstr(name, "fc.");
	if (ptr) {
		tport->tport_proto_id = SCSI_PROTOCOL_FCP;
		off = 3; /* Skip over "fc." */
		goto check_len;
	}
	ptr = strstr(name, "iqn.");
	if (ptr) {
		tport->tport_proto_id = SCSI_PROTOCOL_ISCSI;
		goto check_len;
	}

	pr_err("Unable to locate prefix for emulated Target Port:"
			" %s\n", name);
	kfree(tport);
	return ERR_PTR(-EINVAL);

check_len:
2016
	if (strlen(name) >= VHOST_SCSI_NAMELEN) {
2017
		pr_err("Emulated %s Address: %s, exceeds"
2018 2019
			" max: %d\n", name, vhost_scsi_dump_proto_id(tport),
			VHOST_SCSI_NAMELEN);
2020 2021 2022
		kfree(tport);
		return ERR_PTR(-EINVAL);
	}
2023
	snprintf(&tport->tport_name[0], VHOST_SCSI_NAMELEN, "%s", &name[off]);
2024 2025

	pr_debug("TCM_VHost_ConfigFS: Allocated emulated Target"
2026
		" %s Address: %s\n", vhost_scsi_dump_proto_id(tport), name);
2027 2028 2029 2030

	return &tport->tport_wwn;
}

2031
static void vhost_scsi_drop_tport(struct se_wwn *wwn)
2032
{
2033 2034
	struct vhost_scsi_tport *tport = container_of(wwn,
				struct vhost_scsi_tport, tport_wwn);
2035 2036

	pr_debug("TCM_VHost_ConfigFS: Deallocating emulated Target"
2037
		" %s Address: %s\n", vhost_scsi_dump_proto_id(tport),
2038 2039 2040 2041 2042
		tport->tport_name);

	kfree(tport);
}

2043
static ssize_t
2044
vhost_scsi_wwn_version_show(struct config_item *item, char *page)
2045 2046
{
	return sprintf(page, "TCM_VHOST fabric module %s on %s/%s"
2047
		"on "UTS_RELEASE"\n", VHOST_SCSI_VERSION, utsname()->sysname,
2048 2049 2050
		utsname()->machine);
}

2051
CONFIGFS_ATTR_RO(vhost_scsi_wwn_, version);
2052

2053
static struct configfs_attribute *vhost_scsi_wwn_attrs[] = {
2054
	&vhost_scsi_wwn_attr_version,
2055 2056 2057
	NULL,
};

2058
static const struct target_core_fabric_ops vhost_scsi_ops = {
2059 2060
	.module				= THIS_MODULE,
	.name				= "vhost",
2061 2062 2063 2064 2065 2066 2067
	.get_fabric_name		= vhost_scsi_get_fabric_name,
	.tpg_get_wwn			= vhost_scsi_get_fabric_wwn,
	.tpg_get_tag			= vhost_scsi_get_tpgt,
	.tpg_check_demo_mode		= vhost_scsi_check_true,
	.tpg_check_demo_mode_cache	= vhost_scsi_check_true,
	.tpg_check_demo_mode_write_protect = vhost_scsi_check_false,
	.tpg_check_prod_mode_write_protect = vhost_scsi_check_false,
2068
	.tpg_check_prot_fabric_only	= vhost_scsi_check_prot_fabric_only,
2069 2070
	.tpg_get_inst_index		= vhost_scsi_tpg_get_inst_index,
	.release_cmd			= vhost_scsi_release_cmd,
2071
	.check_stop_free		= vhost_scsi_check_stop_free,
2072
	.sess_get_index			= vhost_scsi_sess_get_index,
2073
	.sess_get_initiator_sid		= NULL,
2074 2075 2076 2077 2078 2079 2080 2081
	.write_pending			= vhost_scsi_write_pending,
	.write_pending_status		= vhost_scsi_write_pending_status,
	.set_default_node_attributes	= vhost_scsi_set_default_node_attrs,
	.get_cmd_state			= vhost_scsi_get_cmd_state,
	.queue_data_in			= vhost_scsi_queue_data_in,
	.queue_status			= vhost_scsi_queue_status,
	.queue_tm_rsp			= vhost_scsi_queue_tm_rsp,
	.aborted_task			= vhost_scsi_aborted_task,
2082 2083 2084
	/*
	 * Setup callers for generic logic in target_core_fabric_configfs.c
	 */
2085 2086 2087 2088 2089 2090
	.fabric_make_wwn		= vhost_scsi_make_tport,
	.fabric_drop_wwn		= vhost_scsi_drop_tport,
	.fabric_make_tpg		= vhost_scsi_make_tpg,
	.fabric_drop_tpg		= vhost_scsi_drop_tpg,
	.fabric_post_link		= vhost_scsi_port_link,
	.fabric_pre_unlink		= vhost_scsi_port_unlink,
2091 2092 2093 2094

	.tfc_wwn_attrs			= vhost_scsi_wwn_attrs,
	.tfc_tpg_base_attrs		= vhost_scsi_tpg_attrs,
	.tfc_tpg_attrib_attrs		= vhost_scsi_tpg_attrib_attrs,
2095 2096
};

2097
static int __init vhost_scsi_init(void)
2098
{
2099
	int ret = -ENOMEM;
2100

2101
	pr_debug("TCM_VHOST fabric module %s on %s/%s"
2102
		" on "UTS_RELEASE"\n", VHOST_SCSI_VERSION, utsname()->sysname,
2103 2104
		utsname()->machine);

2105 2106 2107 2108
	/*
	 * Use our own dedicated workqueue for submitting I/O into
	 * target core to avoid contention within system_wq.
	 */
2109 2110
	vhost_scsi_workqueue = alloc_workqueue("vhost_scsi", 0, 0);
	if (!vhost_scsi_workqueue)
2111 2112 2113 2114 2115 2116
		goto out;

	ret = vhost_scsi_register();
	if (ret < 0)
		goto out_destroy_workqueue;

2117
	ret = target_register_template(&vhost_scsi_ops);
2118 2119 2120 2121 2122 2123 2124 2125
	if (ret < 0)
		goto out_vhost_scsi_deregister;

	return 0;

out_vhost_scsi_deregister:
	vhost_scsi_deregister();
out_destroy_workqueue:
2126
	destroy_workqueue(vhost_scsi_workqueue);
2127 2128 2129 2130
out:
	return ret;
};

2131
static void vhost_scsi_exit(void)
2132
{
2133
	target_unregister_template(&vhost_scsi_ops);
2134
	vhost_scsi_deregister();
2135
	destroy_workqueue(vhost_scsi_workqueue);
2136 2137
};

M
Michael S. Tsirkin 已提交
2138 2139
MODULE_DESCRIPTION("VHOST_SCSI series fabric driver");
MODULE_ALIAS("tcm_vhost");
2140
MODULE_LICENSE("GPL");
2141 2142
module_init(vhost_scsi_init);
module_exit(vhost_scsi_exit);