scsi.c 56.3 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 50

#include "vhost.h"
M
Michael S. Tsirkin 已提交
51

52 53 54 55 56 57
#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
58
#define VHOST_SCSI_PREALLOC_PROT_SGLS 2048
M
Michael S. Tsirkin 已提交
59 60 61 62 63 64 65 66

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

67
struct vhost_scsi_cmd {
M
Michael S. Tsirkin 已提交
68 69 70 71
	/* Descriptor from vhost_get_vq_desc() for virt_queue segment */
	int tvc_vq_desc;
	/* virtio-scsi initiator task attribute */
	int tvc_task_attr;
72 73
	/* virtio-scsi response incoming iovecs */
	int tvc_in_iovs;
M
Michael S. Tsirkin 已提交
74 75 76 77 78 79 80 81
	/* 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;
82
	u32 tvc_prot_sgl_count;
83
	/* Saved unpacked SCSI LUN for vhost_scsi_submission_work() */
M
Michael S. Tsirkin 已提交
84 85 86
	u32 tvc_lun;
	/* Pointer to the SGL formatted memory from virtio-scsi */
	struct scatterlist *tvc_sgl;
87
	struct scatterlist *tvc_prot_sgl;
88
	struct page **tvc_upages;
89
	/* Pointer to response header iovec */
90
	struct iovec tvc_resp_iov;
M
Michael S. Tsirkin 已提交
91 92 93 94 95
	/* 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 */
96
	struct vhost_scsi_nexus *tvc_nexus;
M
Michael S. Tsirkin 已提交
97 98
	/* The TCM I/O descriptor that is accessed via container_of() */
	struct se_cmd tvc_se_cmd;
99
	/* work item used for cmwq dispatch to vhost_scsi_submission_work() */
M
Michael S. Tsirkin 已提交
100 101
	struct work_struct work;
	/* Copy of the incoming SCSI command descriptor block (CDB) */
102
	unsigned char tvc_cdb[VHOST_SCSI_MAX_CDB_SIZE];
M
Michael S. Tsirkin 已提交
103 104 105 106 107 108 109 110
	/* 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;
};

111
struct vhost_scsi_nexus {
M
Michael S. Tsirkin 已提交
112 113 114 115
	/* Pointer to TCM session for I_T Nexus */
	struct se_session *tvn_se_sess;
};

116
struct vhost_scsi_tpg {
M
Michael S. Tsirkin 已提交
117 118 119 120 121 122
	/* 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;
123 124
	/* Used for enabling T10-PI with legacy devices */
	int tv_fabric_prot_type;
125
	/* list for vhost_scsi_list */
M
Michael S. Tsirkin 已提交
126 127 128 129
	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 */
130 131 132 133
	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
134 135 136 137 138
	struct se_portal_group se_tpg;
	/* Pointer back to vhost_scsi, protected by tv_tpg_mutex */
	struct vhost_scsi *vhost_scsi;
};

139
struct vhost_scsi_tport {
M
Michael S. Tsirkin 已提交
140 141 142 143 144
	/* 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 */
145 146
	char tport_name[VHOST_SCSI_NAMELEN];
	/* Returned by vhost_scsi_make_tport() */
M
Michael S. Tsirkin 已提交
147 148 149
	struct se_wwn tport_wwn;
};

150
struct vhost_scsi_evt {
M
Michael S. Tsirkin 已提交
151 152 153 154 155
	/* event to be sent to guest */
	struct virtio_scsi_event event;
	/* event list, serviced from vhost worker thread */
	struct llist_node list;
};
156

157 158 159 160 161 162
enum {
	VHOST_SCSI_VQ_CTL = 0,
	VHOST_SCSI_VQ_EVT = 1,
	VHOST_SCSI_VQ_IO = 2,
};

163
/* Note: can't set VIRTIO_F_VERSION_1 yet, since that implies ANY_LAYOUT. */
164
enum {
165
	VHOST_SCSI_FEATURES = VHOST_FEATURES | (1ULL << VIRTIO_SCSI_F_HOTPLUG) |
166
					       (1ULL << VIRTIO_SCSI_F_T10_PI)
167 168
};

A
Asias He 已提交
169 170
#define VHOST_SCSI_MAX_TARGET	256
#define VHOST_SCSI_MAX_VQ	128
171
#define VHOST_SCSI_MAX_EVENT	128
A
Asias He 已提交
172

173 174
struct vhost_scsi_virtqueue {
	struct vhost_virtqueue vq;
175 176 177 178 179
	/*
	 * 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.
	 */
180
	struct vhost_scsi_inflight inflights[2];
181 182 183 184
	/*
	 * Indicate current inflight in use, protected by vq->mutex.
	 * Writers must also take dev mutex and flush under it.
	 */
185
	int inflight_idx;
186 187
};

188
struct vhost_scsi {
A
Asias He 已提交
189
	/* Protected by vhost_scsi->dev.mutex */
190
	struct vhost_scsi_tpg **vs_tpg;
A
Asias He 已提交
191 192
	char vs_vhost_wwpn[TRANSPORT_IQN_LEN];

193
	struct vhost_dev dev;
194
	struct vhost_scsi_virtqueue vqs[VHOST_SCSI_MAX_VQ];
195 196

	struct vhost_work vs_completion_work; /* cmd completion work item */
197
	struct llist_head vs_completion_list; /* cmd completion queue */
198 199 200 201 202 203

	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 */
204 205
};

206
static struct workqueue_struct *vhost_scsi_workqueue;
207

208 209 210
/* Global spinlock to protect vhost_scsi TPG list for vhost IOCTL access */
static DEFINE_MUTEX(vhost_scsi_mutex);
static LIST_HEAD(vhost_scsi_list);
211

212
static void vhost_scsi_done_inflight(struct kref *kref)
213 214 215 216 217 218 219
{
	struct vhost_scsi_inflight *inflight;

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

220
static void vhost_scsi_init_inflight(struct vhost_scsi *vs,
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
				    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 *
248
vhost_scsi_get_inflight(struct vhost_virtqueue *vq)
249 250 251 252 253 254 255 256 257 258 259
{
	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;
}

260
static void vhost_scsi_put_inflight(struct vhost_scsi_inflight *inflight)
261
{
262
	kref_put(&inflight->kref, vhost_scsi_done_inflight);
263 264
}

265
static int vhost_scsi_check_true(struct se_portal_group *se_tpg)
266 267 268 269
{
	return 1;
}

270
static int vhost_scsi_check_false(struct se_portal_group *se_tpg)
271 272 273 274
{
	return 0;
}

275
static char *vhost_scsi_get_fabric_name(void)
276 277 278 279
{
	return "vhost";
}

280
static char *vhost_scsi_get_fabric_wwn(struct se_portal_group *se_tpg)
281
{
282 283 284
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
	struct vhost_scsi_tport *tport = tpg->tport;
285 286 287 288

	return &tport->tport_name[0];
}

289
static u16 vhost_scsi_get_tpgt(struct se_portal_group *se_tpg)
290
{
291 292
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
293 294 295
	return tpg->tport_tpgt;
}

296 297 298 299 300 301 302 303
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;
}

304
static u32 vhost_scsi_tpg_get_inst_index(struct se_portal_group *se_tpg)
305 306 307 308
{
	return 1;
}

309
static void vhost_scsi_release_cmd(struct se_cmd *se_cmd)
310
{
311 312
	struct vhost_scsi_cmd *tv_cmd = container_of(se_cmd,
				struct vhost_scsi_cmd, tvc_se_cmd);
313
	struct se_session *se_sess = tv_cmd->tvc_nexus->tvn_se_sess;
314
	int i;
315 316 317 318

	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]));
319
	}
320 321 322 323
	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]));
	}
324

325
	vhost_scsi_put_inflight(tv_cmd->inflight);
326
	target_free_tag(se_sess, se_cmd);
327 328
}

329
static u32 vhost_scsi_sess_get_index(struct se_session *se_sess)
330 331 332 333
{
	return 0;
}

334
static int vhost_scsi_write_pending(struct se_cmd *se_cmd)
335 336 337 338 339 340
{
	/* Go ahead and process the write immediately */
	target_execute_cmd(se_cmd);
	return 0;
}

341
static int vhost_scsi_write_pending_status(struct se_cmd *se_cmd)
342 343 344 345
{
	return 0;
}

346
static void vhost_scsi_set_default_node_attrs(struct se_node_acl *nacl)
347 348 349 350
{
	return;
}

351
static int vhost_scsi_get_cmd_state(struct se_cmd *se_cmd)
352 353 354 355
{
	return 0;
}

356
static void vhost_scsi_complete_cmd(struct vhost_scsi_cmd *cmd)
357
{
358
	struct vhost_scsi *vs = cmd->tvc_vhost;
359

360
	llist_add(&cmd->tvc_completion_list, &vs->vs_completion_list);
361 362 363

	vhost_work_queue(&vs->dev, &vs->vs_completion_work);
}
364

365
static int vhost_scsi_queue_data_in(struct se_cmd *se_cmd)
366
{
367 368
	struct vhost_scsi_cmd *cmd = container_of(se_cmd,
				struct vhost_scsi_cmd, tvc_se_cmd);
369
	vhost_scsi_complete_cmd(cmd);
370 371 372
	return 0;
}

373
static int vhost_scsi_queue_status(struct se_cmd *se_cmd)
374
{
375 376
	struct vhost_scsi_cmd *cmd = container_of(se_cmd,
				struct vhost_scsi_cmd, tvc_se_cmd);
377
	vhost_scsi_complete_cmd(cmd);
378 379 380
	return 0;
}

381
static void vhost_scsi_queue_tm_rsp(struct se_cmd *se_cmd)
382
{
383
	return;
384 385
}

386
static void vhost_scsi_aborted_task(struct se_cmd *se_cmd)
387 388 389 390
{
	return;
}

391
static void vhost_scsi_free_evt(struct vhost_scsi *vs, struct vhost_scsi_evt *evt)
392 393 394 395 396
{
	vs->vs_events_nr--;
	kfree(evt);
}

397 398
static struct vhost_scsi_evt *
vhost_scsi_allocate_evt(struct vhost_scsi *vs,
399
		       u32 event, u32 reason)
400
{
401
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
402
	struct vhost_scsi_evt *evt;
403 404 405 406 407 408 409 410

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

	evt = kzalloc(sizeof(*evt), GFP_KERNEL);
	if (!evt) {
411
		vq_err(vq, "Failed to allocate vhost_scsi_evt\n");
412 413 414 415
		vs->vs_events_missed = true;
		return NULL;
	}

416 417
	evt->event.event = cpu_to_vhost32(vq, event);
	evt->event.reason = cpu_to_vhost32(vq, reason);
418 419 420 421 422
	vs->vs_events_nr++;

	return evt;
}

423
static void vhost_scsi_free_cmd(struct vhost_scsi_cmd *cmd)
424
{
425
	struct se_cmd *se_cmd = &cmd->tvc_se_cmd;
426 427

	/* TODO locking against target/backend threads? */
428
	transport_generic_free_cmd(se_cmd, 0);
429

430
}
431

432 433
static int vhost_scsi_check_stop_free(struct se_cmd *se_cmd)
{
434
	return target_put_sess_cmd(se_cmd);
435 436
}

437
static void
438
vhost_scsi_do_evt_work(struct vhost_scsi *vs, struct vhost_scsi_evt *evt)
439
{
440
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
441 442 443 444 445 446 447 448 449 450 451 452
	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);
453
	head = vhost_get_vq_desc(vq, vq->iov,
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474
			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) {
475
		event->event |= cpu_to_vhost32(vq, VIRTIO_SCSI_T_EVENTS_MISSED);
476 477 478 479 480 481 482 483
		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
484
		vq_err(vq, "Faulted on vhost_scsi_send_event\n");
485 486
}

487
static void vhost_scsi_evt_work(struct vhost_work *work)
488 489 490
{
	struct vhost_scsi *vs = container_of(work, struct vhost_scsi,
					vs_event_work);
491
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
492
	struct vhost_scsi_evt *evt, *t;
493 494 495 496
	struct llist_node *llnode;

	mutex_lock(&vq->mutex);
	llnode = llist_del_all(&vs->vs_event_list);
497
	llist_for_each_entry_safe(evt, t, llnode, list) {
498 499
		vhost_scsi_do_evt_work(vs, evt);
		vhost_scsi_free_evt(vs, evt);
500 501 502 503
	}
	mutex_unlock(&vq->mutex);
}

504 505 506 507 508 509 510 511 512
/* 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
513
	DECLARE_BITMAP(signal, VHOST_SCSI_MAX_VQ);
514
	struct virtio_scsi_cmd_resp v_rsp;
515
	struct vhost_scsi_cmd *cmd, *t;
516 517
	struct llist_node *llnode;
	struct se_cmd *se_cmd;
518
	struct iov_iter iov_iter;
A
Asias He 已提交
519
	int ret, vq;
520

A
Asias He 已提交
521
	bitmap_zero(signal, VHOST_SCSI_MAX_VQ);
522
	llnode = llist_del_all(&vs->vs_completion_list);
523
	llist_for_each_entry_safe(cmd, t, llnode, tvc_completion_list) {
524
		se_cmd = &cmd->tvc_se_cmd;
525 526

		pr_debug("%s tv_cmd %p resid %u status %#02x\n", __func__,
527
			cmd, se_cmd->residual_count, se_cmd->scsi_status);
528 529

		memset(&v_rsp, 0, sizeof(v_rsp));
530
		v_rsp.resid = cpu_to_vhost32(cmd->tvc_vq, se_cmd->residual_count);
531 532
		/* TODO is status_qualifier field needed? */
		v_rsp.status = se_cmd->scsi_status;
533 534
		v_rsp.sense_len = cpu_to_vhost32(cmd->tvc_vq,
						 se_cmd->scsi_sense_length);
535
		memcpy(v_rsp.sense, cmd->tvc_sense_buf,
536
		       se_cmd->scsi_sense_length);
537

538
		iov_iter_init(&iov_iter, READ, &cmd->tvc_resp_iov,
539 540 541
			      cmd->tvc_in_iovs, sizeof(v_rsp));
		ret = copy_to_iter(&v_rsp, sizeof(v_rsp), &iov_iter);
		if (likely(ret == sizeof(v_rsp))) {
542
			struct vhost_scsi_virtqueue *q;
543 544
			vhost_add_used(cmd->tvc_vq, cmd->tvc_vq_desc, 0);
			q = container_of(cmd->tvc_vq, struct vhost_scsi_virtqueue, vq);
545
			vq = q - vs->vqs;
A
Asias He 已提交
546 547
			__set_bit(vq, signal);
		} else
548 549
			pr_err("Faulted on virtio_scsi_cmd_resp\n");

550
		vhost_scsi_free_cmd(cmd);
551 552
	}

A
Asias He 已提交
553 554 555
	vq = -1;
	while ((vq = find_next_bit(signal, VHOST_SCSI_MAX_VQ, vq + 1))
		< VHOST_SCSI_MAX_VQ)
556
		vhost_signal(&vs->dev, &vs->vqs[vq].vq);
557 558
}

559 560
static struct vhost_scsi_cmd *
vhost_scsi_get_tag(struct vhost_virtqueue *vq, struct vhost_scsi_tpg *tpg,
561 562
		   unsigned char *cdb, u64 scsi_tag, u16 lun, u8 task_attr,
		   u32 exp_data_len, int data_direction)
563
{
564 565
	struct vhost_scsi_cmd *cmd;
	struct vhost_scsi_nexus *tv_nexus;
566
	struct se_session *se_sess;
567
	struct scatterlist *sg, *prot_sg;
568
	struct page **pages;
569
	int tag, cpu;
570

571
	tv_nexus = tpg->tpg_nexus;
572
	if (!tv_nexus) {
573
		pr_err("Unable to locate active struct vhost_scsi_nexus\n");
574 575
		return ERR_PTR(-EIO);
	}
576
	se_sess = tv_nexus->tvn_se_sess;
577

578
	tag = sbitmap_queue_get(&se_sess->sess_tag_pool, &cpu);
579
	if (tag < 0) {
580
		pr_err("Unable to obtain tag for vhost_scsi_cmd\n");
581 582 583
		return ERR_PTR(-ENOMEM);
	}

584
	cmd = &((struct vhost_scsi_cmd *)se_sess->sess_cmd_map)[tag];
585
	sg = cmd->tvc_sgl;
586
	prot_sg = cmd->tvc_prot_sgl;
587
	pages = cmd->tvc_upages;
588
	memset(cmd, 0, sizeof(*cmd));
589
	cmd->tvc_sgl = sg;
590
	cmd->tvc_prot_sgl = prot_sg;
591
	cmd->tvc_upages = pages;
592
	cmd->tvc_se_cmd.map_tag = tag;
593
	cmd->tvc_se_cmd.map_cpu = cpu;
594 595 596
	cmd->tvc_tag = scsi_tag;
	cmd->tvc_lun = lun;
	cmd->tvc_task_attr = task_attr;
597 598 599
	cmd->tvc_exp_data_len = exp_data_len;
	cmd->tvc_data_direction = data_direction;
	cmd->tvc_nexus = tv_nexus;
600
	cmd->inflight = vhost_scsi_get_inflight(vq);
601

602
	memcpy(cmd->tvc_cdb, cdb, VHOST_SCSI_MAX_CDB_SIZE);
603

604
	return cmd;
605 606 607 608 609 610 611
}

/*
 * Map a user memory range into a scatterlist
 *
 * Returns the number of scatterlist entries used or -errno on error.
 */
612
static int
613
vhost_scsi_map_to_sgl(struct vhost_scsi_cmd *cmd,
614
		      struct iov_iter *iter,
615
		      struct scatterlist *sgl,
616
		      bool write)
617
{
618
	struct page **pages = cmd->tvc_upages;
619 620 621 622
	struct scatterlist *sg = sgl;
	ssize_t bytes;
	size_t offset;
	unsigned int npages = 0;
623

624 625
	bytes = iov_iter_get_pages(iter, pages, LONG_MAX,
				VHOST_SCSI_PREALLOC_UPAGES, &offset);
626
	/* No pages were pinned */
627 628
	if (bytes <= 0)
		return bytes < 0 ? bytes : -EFAULT;
629

630
	iov_iter_advance(iter, bytes);
631

632 633 634 635 636 637 638
	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;
639 640
}

641
static int
642
vhost_scsi_calc_sgls(struct iov_iter *iter, size_t bytes, int max_sgls)
643
{
644
	int sgl_count = 0;
645

646 647 648 649 650
	if (!iter || !iter->iov) {
		pr_err("%s: iter->iov is NULL, but expected bytes: %zu"
		       " present\n", __func__, bytes);
		return -EINVAL;
	}
651

652 653 654 655 656
	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;
657
	}
658 659
	return sgl_count;
}
660

661
static int
662 663 664
vhost_scsi_iov_to_sgl(struct vhost_scsi_cmd *cmd, bool write,
		      struct iov_iter *iter,
		      struct scatterlist *sg, int sg_count)
665
{
666
	struct scatterlist *p = sg;
667
	int ret;
668

669 670
	while (iov_iter_count(iter)) {
		ret = vhost_scsi_map_to_sgl(cmd, iter, sg, write);
671
		if (ret < 0) {
672 673
			while (p < sg) {
				struct page *page = sg_page(p++);
674 675 676
				if (page)
					put_page(page);
			}
677 678 679 680 681 682 683
			return ret;
		}
		sg += ret;
	}
	return 0;
}

684
static int
685
vhost_scsi_mapal(struct vhost_scsi_cmd *cmd,
686 687 688 689 690 691 692 693
		 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,
694
						 VHOST_SCSI_PREALLOC_PROT_SGLS);
695 696 697 698 699 700 701 702 703 704 705
		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);
706 707 708 709
		if (ret < 0) {
			cmd->tvc_prot_sgl_count = 0;
			return ret;
		}
710 711
	}
	sgl_count = vhost_scsi_calc_sgls(data_iter, data_bytes,
712
					 VHOST_SCSI_PREALLOC_SGLS);
713 714 715 716 717 718 719 720 721 722 723 724 725
	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;
726 727 728 729
	}
	return 0;
}

730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
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;
}

747
static void vhost_scsi_submission_work(struct work_struct *work)
748
{
749 750 751
	struct vhost_scsi_cmd *cmd =
		container_of(work, struct vhost_scsi_cmd, work);
	struct vhost_scsi_nexus *tv_nexus;
752
	struct se_cmd *se_cmd = &cmd->tvc_se_cmd;
753 754
	struct scatterlist *sg_ptr, *sg_prot_ptr = NULL;
	int rc;
755

756
	/* FIXME: BIDI operation */
757 758
	if (cmd->tvc_sgl_count) {
		sg_ptr = cmd->tvc_sgl;
759 760 761 762 763

		if (cmd->tvc_prot_sgl_count)
			sg_prot_ptr = cmd->tvc_prot_sgl;
		else
			se_cmd->prot_pto = true;
764 765 766
	} else {
		sg_ptr = NULL;
	}
767
	tv_nexus = cmd->tvc_nexus;
768

769
	se_cmd->tag = 0;
770
	rc = target_submit_cmd_map_sgls(se_cmd, tv_nexus->tvn_se_sess,
771 772
			cmd->tvc_cdb, &cmd->tvc_sense_buf[0],
			cmd->tvc_lun, cmd->tvc_exp_data_len,
773 774 775 776
			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);
777 778
	if (rc < 0) {
		transport_send_check_condition_and_sense(se_cmd,
779
				TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE, 0);
780 781 782 783
		transport_generic_free_cmd(se_cmd, 0);
	}
}

784 785 786 787
static void
vhost_scsi_send_bad_target(struct vhost_scsi *vs,
			   struct vhost_virtqueue *vq,
			   int head, unsigned out)
788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
{
	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");
}

803 804
static void
vhost_scsi_handle_vq(struct vhost_scsi *vs, struct vhost_virtqueue *vq)
805
{
806
	struct vhost_scsi_tpg **vs_tpg, *tpg;
807
	struct virtio_scsi_cmd_req v_req;
808
	struct virtio_scsi_cmd_req_pi v_req_pi;
809
	struct vhost_scsi_cmd *cmd;
810
	struct iov_iter out_iter, in_iter, prot_iter, data_iter;
811
	u64 tag;
812
	u32 exp_data_len, data_direction;
813
	unsigned int out = 0, in = 0;
814 815 816
	int head, ret, prot_bytes;
	size_t req_size, rsp_size = sizeof(struct virtio_scsi_cmd_resp);
	size_t out_size, in_size;
817 818
	u16 lun;
	u8 *target, *lunp, task_attr;
819
	bool t10_pi = vhost_has_feature(vq, VIRTIO_SCSI_F_T10_PI);
820
	void *req, *cdb;
821

822
	mutex_lock(&vq->mutex);
823 824 825 826
	/*
	 * We can handle the vq only after the endpoint is setup by calling the
	 * VHOST_SCSI_SET_ENDPOINT ioctl.
	 */
827
	vs_tpg = vq->private_data;
828
	if (!vs_tpg)
829
		goto out;
830 831 832 833

	vhost_disable_notify(&vs->dev, vq);

	for (;;) {
834
		head = vhost_get_vq_desc(vq, vq->iov,
835 836
					 ARRAY_SIZE(vq->iov), &out, &in,
					 NULL, NULL);
837
		pr_debug("vhost_get_vq_desc: head: %d, out: %u in: %u\n",
838
			 head, out, in);
839 840 841 842 843 844 845 846 847 848 849 850
		/* 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;
		}
		/*
851 852
		 * Check for a sane response buffer so we can report early
		 * errors back to the guest.
853
		 */
854 855 856
		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);
857 858
			break;
		}
859 860 861 862 863
		/*
		 * Setup pointers and values based upon different virtio-scsi
		 * request header if T10_PI is enabled in KVM guest.
		 */
		if (t10_pi) {
864
			req = &v_req_pi;
865
			req_size = sizeof(v_req_pi);
866 867 868 869
			lunp = &v_req_pi.lun[0];
			target = &v_req_pi.lun[1];
		} else {
			req = &v_req;
870
			req_size = sizeof(v_req);
871 872 873
			lunp = &v_req.lun[0];
			target = &v_req.lun[1];
		}
874 875 876 877 878
		/*
		 * FIXME: Not correct for BIDI operation
		 */
		out_size = iov_length(vq->iov, out);
		in_size = iov_length(&vq->iov[out], in);
879

880 881 882 883 884 885 886 887 888 889 890
		/*
		 * 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);
891

892
		if (unlikely(!copy_from_iter_full(req, req_size, &out_iter))) {
893
			vq_err(vq, "Faulted on copy_from_iter\n");
894 895
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
896
		}
897
		/* virtio-scsi spec requires byte 0 of the lun to be 1 */
898
		if (unlikely(*lunp != 1)) {
899
			vq_err(vq, "Illegal virtio-scsi lun: %u\n", *lunp);
900 901 902 903
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
		}

904
		tpg = READ_ONCE(vs_tpg[*target]);
905
		if (unlikely(!tpg)) {
906
			/* Target does not exist, fail the request */
907
			vhost_scsi_send_bad_target(vs, vq, head, out);
A
Asias He 已提交
908 909
			continue;
		}
910
		/*
911 912 913
		 * Determine data_direction by calculating the total outgoing
		 * iovec sizes + incoming iovec sizes vs. virtio-scsi request +
		 * response headers respectively.
914
		 *
915 916 917 918 919 920
		 * 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.
921
		 *
922 923
		 * Any associated T10_PI bytes for the outgoing / incoming
		 * payloads are included in calculation of exp_data_len here.
924
		 */
925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946
		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().
947
		 */
948
		if (t10_pi) {
949 950
			if (v_req_pi.pi_bytesout) {
				if (data_direction != DMA_TO_DEVICE) {
951 952
					vq_err(vq, "Received non zero pi_bytesout,"
						" but wrong data_direction\n");
953 954
					vhost_scsi_send_bad_target(vs, vq, head, out);
					continue;
955
				}
956
				prot_bytes = vhost32_to_cpu(vq, v_req_pi.pi_bytesout);
957 958
			} else if (v_req_pi.pi_bytesin) {
				if (data_direction != DMA_FROM_DEVICE) {
959 960
					vq_err(vq, "Received non zero pi_bytesin,"
						" but wrong data_direction\n");
961 962
					vhost_scsi_send_bad_target(vs, vq, head, out);
					continue;
963
				}
964
				prot_bytes = vhost32_to_cpu(vq, v_req_pi.pi_bytesin);
965
			}
966
			/*
967 968
			 * Set prot_iter to data_iter and truncate it to
			 * prot_bytes, and advance data_iter past any
969 970 971 972 973
			 * 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
				exp_data_len -= prot_bytes;
				prot_iter = data_iter;
977
				iov_iter_truncate(&prot_iter, prot_bytes);
978
				iov_iter_advance(&data_iter, prot_bytes);
979
			}
980
			tag = vhost64_to_cpu(vq, v_req_pi.tag);
981 982 983 984
			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 {
985
			tag = vhost64_to_cpu(vq, v_req.tag);
986 987 988 989 990
			task_attr = v_req.task_attr;
			cdb = &v_req.cdb[0];
			lun = ((v_req.lun[2] << 8) | v_req.lun[3]) & 0x3FFF;
		}
		/*
991 992 993
		 * 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.
994 995 996
		 *
		 * TODO what if cdb was too small for varlen cdb header?
		 */
997
		if (unlikely(scsi_command_size(cdb) > VHOST_SCSI_MAX_CDB_SIZE)) {
998 999
			vq_err(vq, "Received SCSI CDB with command_size: %d that"
				" exceeds SCSI_MAX_VARLEN_CDB_SIZE: %d\n",
1000
				scsi_command_size(cdb), VHOST_SCSI_MAX_CDB_SIZE);
1001 1002
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
1003 1004
		}
		cmd = vhost_scsi_get_tag(vq, tpg, cdb, tag, lun, task_attr,
1005 1006
					 exp_data_len + prot_bytes,
					 data_direction);
1007
		if (IS_ERR(cmd)) {
1008
			vq_err(vq, "vhost_scsi_get_tag failed %ld\n",
1009
			       PTR_ERR(cmd));
1010 1011
			vhost_scsi_send_bad_target(vs, vq, head, out);
			continue;
1012
		}
1013 1014
		cmd->tvc_vhost = vs;
		cmd->tvc_vq = vq;
1015
		cmd->tvc_resp_iov = vq->iov[out];
1016
		cmd->tvc_in_iovs = in;
1017 1018

		pr_debug("vhost_scsi got command opcode: %#02x, lun: %d\n",
1019 1020 1021
			 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);
1022 1023

		if (data_direction != DMA_NONE) {
1024 1025 1026
			ret = vhost_scsi_mapal(cmd,
					       prot_bytes, &prot_iter,
					       exp_data_len, &data_iter);
1027 1028
			if (unlikely(ret)) {
				vq_err(vq, "Failed to map iov to sgl\n");
1029
				vhost_scsi_release_cmd(&cmd->tvc_se_cmd);
1030 1031
				vhost_scsi_send_bad_target(vs, vq, head, out);
				continue;
1032 1033 1034 1035 1036
			}
		}
		/*
		 * Save the descriptor from vhost_get_vq_desc() to be used to
		 * complete the virtio-scsi request in TCM callback context via
1037
		 * vhost_scsi_queue_data_in() and vhost_scsi_queue_status()
1038
		 */
1039
		cmd->tvc_vq_desc = head;
1040
		/*
1041 1042 1043 1044
		 * 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.
1045
		 */
1046 1047
		INIT_WORK(&cmd->work, vhost_scsi_submission_work);
		queue_work(vhost_scsi_workqueue, &cmd->work);
1048
	}
1049
out:
1050
	mutex_unlock(&vq->mutex);
1051 1052 1053 1054
}

static void vhost_scsi_ctl_handle_kick(struct vhost_work *work)
{
1055
	pr_debug("%s: The handling func for control queue.\n", __func__);
1056 1057
}

1058
static void
1059 1060
vhost_scsi_send_evt(struct vhost_scsi *vs,
		   struct vhost_scsi_tpg *tpg,
1061 1062 1063
		   struct se_lun *lun,
		   u32 event,
		   u32 reason)
1064
{
1065
	struct vhost_scsi_evt *evt;
1066

1067
	evt = vhost_scsi_allocate_evt(vs, event, reason);
1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
	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;
1078
		evt->event.lun[1] = tpg->tport_tpgt;
1079 1080 1081 1082 1083 1084 1085 1086 1087
		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);
}

1088 1089
static void vhost_scsi_evt_handle_kick(struct vhost_work *work)
{
1090 1091 1092 1093 1094 1095 1096 1097 1098
	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)
1099
		vhost_scsi_send_evt(vs, NULL, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
1100 1101
out:
	mutex_unlock(&vq->mutex);
1102 1103 1104 1105 1106 1107 1108 1109
}

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
1110
	vhost_scsi_handle_vq(vs, vq);
1111 1112
}

1113 1114
static void vhost_scsi_flush_vq(struct vhost_scsi *vs, int index)
{
1115
	vhost_poll_flush(&vs->vqs[index].vq.poll);
1116 1117
}

1118
/* Callers must hold dev mutex */
1119 1120
static void vhost_scsi_flush(struct vhost_scsi *vs)
{
1121
	struct vhost_scsi_inflight *old_inflight[VHOST_SCSI_MAX_VQ];
1122 1123
	int i;

1124
	/* Init new inflight and remember the old inflight */
1125
	vhost_scsi_init_inflight(vs, old_inflight);
1126 1127 1128 1129 1130 1131 1132

	/*
	 * 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++)
1133
		kref_put(&old_inflight[i]->kref, vhost_scsi_done_inflight);
1134 1135

	/* Flush both the vhost poll and vhost work */
1136 1137 1138
	for (i = 0; i < VHOST_SCSI_MAX_VQ; i++)
		vhost_scsi_flush_vq(vs, i);
	vhost_work_flush(&vs->dev, &vs->vs_completion_work);
1139
	vhost_work_flush(&vs->dev, &vs->vs_event_work);
1140 1141 1142 1143

	/* 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);
1144 1145
}

1146 1147
/*
 * Called from vhost_scsi_ioctl() context to walk the list of available
1148
 * vhost_scsi_tpg with an active struct vhost_scsi_nexus
1149 1150
 *
 *  The lock nesting rule is:
1151
 *    vhost_scsi_mutex -> vs->dev.mutex -> tpg->tv_tpg_mutex -> vq->mutex
1152
 */
1153 1154 1155
static int
vhost_scsi_set_endpoint(struct vhost_scsi *vs,
			struct vhost_scsi_target *t)
1156
{
1157
	struct se_portal_group *se_tpg;
1158 1159 1160
	struct vhost_scsi_tport *tv_tport;
	struct vhost_scsi_tpg *tpg;
	struct vhost_scsi_tpg **vs_tpg;
1161 1162
	struct vhost_virtqueue *vq;
	int index, ret, i, len;
A
Asias He 已提交
1163
	bool match = false;
1164

1165
	mutex_lock(&vhost_scsi_mutex);
1166
	mutex_lock(&vs->dev.mutex);
1167

1168 1169 1170
	/* Verify that ring has been setup correctly. */
	for (index = 0; index < vs->dev.nvqs; ++index) {
		/* Verify that ring has been setup correctly. */
1171
		if (!vhost_vq_access_ok(&vs->vqs[index].vq)) {
1172 1173
			ret = -EFAULT;
			goto out;
1174 1175 1176
		}
	}

1177 1178 1179
	len = sizeof(vs_tpg[0]) * VHOST_SCSI_MAX_TARGET;
	vs_tpg = kzalloc(len, GFP_KERNEL);
	if (!vs_tpg) {
1180 1181
		ret = -ENOMEM;
		goto out;
1182 1183 1184 1185
	}
	if (vs->vs_tpg)
		memcpy(vs_tpg, vs->vs_tpg, len);

1186
	list_for_each_entry(tpg, &vhost_scsi_list, tv_tpg_list) {
1187 1188 1189
		mutex_lock(&tpg->tv_tpg_mutex);
		if (!tpg->tpg_nexus) {
			mutex_unlock(&tpg->tv_tpg_mutex);
1190 1191
			continue;
		}
1192 1193
		if (tpg->tv_tpg_vhost_count != 0) {
			mutex_unlock(&tpg->tv_tpg_mutex);
1194 1195
			continue;
		}
1196
		tv_tport = tpg->tport;
1197

A
Asias He 已提交
1198
		if (!strcmp(tv_tport->tport_name, t->vhost_wwpn)) {
1199
			if (vs->vs_tpg && vs->vs_tpg[tpg->tport_tpgt]) {
1200
				kfree(vs_tpg);
1201
				mutex_unlock(&tpg->tv_tpg_mutex);
1202 1203
				ret = -EEXIST;
				goto out;
1204
			}
1205 1206 1207 1208 1209 1210 1211
			/*
			 * 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;
1212
			ret = target_depend_item(&se_tpg->tpg_group.cg_item);
1213 1214 1215 1216 1217 1218
			if (ret) {
				pr_warn("configfs_depend_item() failed: %d\n", ret);
				kfree(vs_tpg);
				mutex_unlock(&tpg->tv_tpg_mutex);
				goto out;
			}
1219 1220 1221
			tpg->tv_tpg_vhost_count++;
			tpg->vhost_scsi = vs;
			vs_tpg[tpg->tport_tpgt] = tpg;
1222
			smp_mb__after_atomic();
A
Asias He 已提交
1223
			match = true;
1224
		}
1225
		mutex_unlock(&tpg->tv_tpg_mutex);
1226
	}
A
Asias He 已提交
1227 1228 1229 1230

	if (match) {
		memcpy(vs->vs_vhost_wwpn, t->vhost_wwpn,
		       sizeof(vs->vs_vhost_wwpn));
1231
		for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
1232
			vq = &vs->vqs[i].vq;
1233
			mutex_lock(&vq->mutex);
A
Asias He 已提交
1234
			vq->private_data = vs_tpg;
G
Greg Kurz 已提交
1235
			vhost_vq_init_access(vq);
1236 1237
			mutex_unlock(&vq->mutex);
		}
A
Asias He 已提交
1238 1239 1240 1241 1242
		ret = 0;
	} else {
		ret = -EEXIST;
	}

1243 1244 1245 1246 1247 1248 1249 1250
	/*
	 * 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;

1251
out:
A
Asias He 已提交
1252
	mutex_unlock(&vs->dev.mutex);
1253
	mutex_unlock(&vhost_scsi_mutex);
A
Asias He 已提交
1254
	return ret;
1255 1256
}

1257 1258 1259
static int
vhost_scsi_clear_endpoint(struct vhost_scsi *vs,
			  struct vhost_scsi_target *t)
1260
{
1261
	struct se_portal_group *se_tpg;
1262 1263
	struct vhost_scsi_tport *tv_tport;
	struct vhost_scsi_tpg *tpg;
1264 1265
	struct vhost_virtqueue *vq;
	bool match = false;
A
Asias He 已提交
1266 1267
	int index, ret, i;
	u8 target;
1268

1269
	mutex_lock(&vhost_scsi_mutex);
1270 1271 1272
	mutex_lock(&vs->dev.mutex);
	/* Verify that ring has been setup correctly. */
	for (index = 0; index < vs->dev.nvqs; ++index) {
1273
		if (!vhost_vq_access_ok(&vs->vqs[index].vq)) {
1274
			ret = -EFAULT;
1275
			goto err_dev;
1276 1277
		}
	}
1278 1279

	if (!vs->vs_tpg) {
1280 1281
		ret = 0;
		goto err_dev;
1282 1283
	}

A
Asias He 已提交
1284 1285
	for (i = 0; i < VHOST_SCSI_MAX_TARGET; i++) {
		target = i;
1286 1287
		tpg = vs->vs_tpg[target];
		if (!tpg)
A
Asias He 已提交
1288 1289
			continue;

1290 1291
		mutex_lock(&tpg->tv_tpg_mutex);
		tv_tport = tpg->tport;
A
Asias He 已提交
1292 1293
		if (!tv_tport) {
			ret = -ENODEV;
1294
			goto err_tpg;
A
Asias He 已提交
1295 1296 1297
		}

		if (strcmp(tv_tport->tport_name, t->vhost_wwpn)) {
1298
			pr_warn("tv_tport->tport_name: %s, tpg->tport_tpgt: %hu"
A
Asias He 已提交
1299
				" does not match t->vhost_wwpn: %s, t->vhost_tpgt: %hu\n",
1300
				tv_tport->tport_name, tpg->tport_tpgt,
A
Asias He 已提交
1301 1302
				t->vhost_wwpn, t->vhost_tpgt);
			ret = -EINVAL;
1303
			goto err_tpg;
A
Asias He 已提交
1304
		}
1305 1306
		tpg->tv_tpg_vhost_count--;
		tpg->vhost_scsi = NULL;
A
Asias He 已提交
1307
		vs->vs_tpg[target] = NULL;
1308
		match = true;
1309
		mutex_unlock(&tpg->tv_tpg_mutex);
1310 1311 1312 1313 1314
		/*
		 * 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;
1315
		target_undepend_item(&se_tpg->tpg_group.cg_item);
1316
	}
1317 1318
	if (match) {
		for (i = 0; i < VHOST_SCSI_MAX_VQ; i++) {
1319
			vq = &vs->vqs[i].vq;
1320
			mutex_lock(&vq->mutex);
A
Asias He 已提交
1321
			vq->private_data = NULL;
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331
			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;
1332
	WARN_ON(vs->vs_events_nr);
1333
	mutex_unlock(&vs->dev.mutex);
1334
	mutex_unlock(&vhost_scsi_mutex);
1335
	return 0;
1336

1337
err_tpg:
1338
	mutex_unlock(&tpg->tv_tpg_mutex);
1339
err_dev:
1340
	mutex_unlock(&vs->dev.mutex);
1341
	mutex_unlock(&vhost_scsi_mutex);
1342
	return ret;
1343 1344
}

1345 1346
static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)
{
1347 1348 1349
	struct vhost_virtqueue *vq;
	int i;

1350 1351 1352 1353 1354 1355 1356 1357 1358
	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;
	}
1359 1360 1361 1362 1363 1364 1365

	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);
	}
1366 1367 1368 1369
	mutex_unlock(&vs->dev.mutex);
	return 0;
}

1370 1371
static int vhost_scsi_open(struct inode *inode, struct file *f)
{
1372
	struct vhost_scsi *vs;
1373
	struct vhost_virtqueue **vqs;
1374
	int r = -ENOMEM, i;
1375

1376
	vs = kzalloc(sizeof(*vs), GFP_KERNEL | __GFP_NOWARN | __GFP_RETRY_MAYFAIL);
1377 1378 1379 1380 1381
	if (!vs) {
		vs = vzalloc(sizeof(*vs));
		if (!vs)
			goto err_vs;
	}
1382

1383
	vqs = kmalloc_array(VHOST_SCSI_MAX_VQ, sizeof(*vqs), GFP_KERNEL);
1384 1385
	if (!vqs)
		goto err_vqs;
1386

1387
	vhost_work_init(&vs->vs_completion_work, vhost_scsi_complete_cmd_work);
1388
	vhost_work_init(&vs->vs_event_work, vhost_scsi_evt_work);
1389

1390 1391
	vs->vs_events_nr = 0;
	vs->vs_events_missed = false;
1392

1393 1394 1395 1396
	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;
1397
	for (i = VHOST_SCSI_VQ_IO; i < VHOST_SCSI_MAX_VQ; i++) {
1398 1399
		vqs[i] = &vs->vqs[i].vq;
		vs->vqs[i].vq.handle_kick = vhost_scsi_handle_kick;
1400
	}
Z
Zhi Yong Wu 已提交
1401
	vhost_dev_init(&vs->dev, vqs, VHOST_SCSI_MAX_VQ);
1402

1403
	vhost_scsi_init_inflight(vs, NULL);
1404

1405
	f->private_data = vs;
1406
	return 0;
1407 1408

err_vqs:
1409
	kvfree(vs);
1410 1411
err_vs:
	return r;
1412 1413 1414 1415
}

static int vhost_scsi_release(struct inode *inode, struct file *f)
{
1416
	struct vhost_scsi *vs = f->private_data;
A
Asias He 已提交
1417
	struct vhost_scsi_target t;
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);
1424
	vhost_dev_cleanup(&vs->dev);
1425
	/* Jobs can re-queue themselves in evt kick handler. Do extra flush. */
1426 1427
	vhost_scsi_flush(vs);
	kfree(vs->dev.vqs);
1428
	kvfree(vs);
1429 1430 1431
	return 0;
}

1432 1433 1434 1435
static long
vhost_scsi_ioctl(struct file *f,
		 unsigned int ioctl,
		 unsigned long arg)
1436 1437 1438 1439 1440
{
	struct vhost_scsi *vs = f->private_data;
	struct vhost_scsi_target backend;
	void __user *argp = (void __user *)arg;
	u64 __user *featurep = argp;
1441 1442
	u32 __user *eventsp = argp;
	u32 events_missed;
1443
	u64 features;
1444
	int r, abi_version = VHOST_SCSI_ABI_VERSION;
1445
	struct vhost_virtqueue *vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
1446 1447 1448 1449 1450

	switch (ioctl) {
	case VHOST_SCSI_SET_ENDPOINT:
		if (copy_from_user(&backend, argp, sizeof backend))
			return -EFAULT;
1451 1452
		if (backend.reserved != 0)
			return -EOPNOTSUPP;
1453 1454 1455 1456 1457

		return vhost_scsi_set_endpoint(vs, &backend);
	case VHOST_SCSI_CLEAR_ENDPOINT:
		if (copy_from_user(&backend, argp, sizeof backend))
			return -EFAULT;
1458 1459
		if (backend.reserved != 0)
			return -EOPNOTSUPP;
1460 1461 1462

		return vhost_scsi_clear_endpoint(vs, &backend);
	case VHOST_SCSI_GET_ABI_VERSION:
1463
		if (copy_to_user(argp, &abi_version, sizeof abi_version))
1464 1465
			return -EFAULT;
		return 0;
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479
	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;
1480
	case VHOST_GET_FEATURES:
1481
		features = VHOST_SCSI_FEATURES;
1482 1483 1484 1485 1486 1487 1488 1489 1490
		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);
1491 1492 1493 1494
		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);
1495 1496 1497 1498 1499
		mutex_unlock(&vs->dev.mutex);
		return r;
	}
}

1500 1501 1502 1503 1504 1505 1506 1507
#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

1508 1509 1510 1511
static const struct file_operations vhost_scsi_fops = {
	.owner          = THIS_MODULE,
	.release        = vhost_scsi_release,
	.unlocked_ioctl = vhost_scsi_ioctl,
1512 1513 1514
#ifdef CONFIG_COMPAT
	.compat_ioctl	= vhost_scsi_compat_ioctl,
#endif
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529
	.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);
}

1530
static void vhost_scsi_deregister(void)
1531
{
1532
	misc_deregister(&vhost_scsi_misc);
1533 1534
}

1535
static char *vhost_scsi_dump_proto_id(struct vhost_scsi_tport *tport)
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
{
	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";
}

1551
static void
1552
vhost_scsi_do_plug(struct vhost_scsi_tpg *tpg,
1553
		  struct se_lun *lun, bool plug)
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
{

	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;

1570
	vq = &vs->vqs[VHOST_SCSI_VQ_EVT].vq;
1571
	mutex_lock(&vq->mutex);
1572
	if (vhost_has_feature(vq, VIRTIO_SCSI_F_HOTPLUG))
1573
		vhost_scsi_send_evt(vs, tpg, lun,
1574
				   VIRTIO_SCSI_T_TRANSPORT_RESET, reason);
1575 1576 1577 1578
	mutex_unlock(&vq->mutex);
	mutex_unlock(&vs->dev.mutex);
}

1579
static void vhost_scsi_hotplug(struct vhost_scsi_tpg *tpg, struct se_lun *lun)
1580
{
1581
	vhost_scsi_do_plug(tpg, lun, true);
1582 1583
}

1584
static void vhost_scsi_hotunplug(struct vhost_scsi_tpg *tpg, struct se_lun *lun)
1585
{
1586
	vhost_scsi_do_plug(tpg, lun, false);
1587 1588
}

1589
static int vhost_scsi_port_link(struct se_portal_group *se_tpg,
1590
			       struct se_lun *lun)
1591
{
1592 1593
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
1594

1595
	mutex_lock(&vhost_scsi_mutex);
1596

1597 1598 1599
	mutex_lock(&tpg->tv_tpg_mutex);
	tpg->tv_tpg_port_count++;
	mutex_unlock(&tpg->tv_tpg_mutex);
1600

1601
	vhost_scsi_hotplug(tpg, lun);
1602

1603
	mutex_unlock(&vhost_scsi_mutex);
1604

1605 1606 1607
	return 0;
}

1608
static void vhost_scsi_port_unlink(struct se_portal_group *se_tpg,
1609
				  struct se_lun *lun)
1610
{
1611 1612
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
1613

1614
	mutex_lock(&vhost_scsi_mutex);
1615

1616 1617 1618
	mutex_lock(&tpg->tv_tpg_mutex);
	tpg->tv_tpg_port_count--;
	mutex_unlock(&tpg->tv_tpg_mutex);
1619

1620
	vhost_scsi_hotunplug(tpg, lun);
1621

1622
	mutex_unlock(&vhost_scsi_mutex);
1623 1624
}

1625
static void vhost_scsi_free_cmd_map_res(struct se_session *se_sess)
1626
{
1627
	struct vhost_scsi_cmd *tv_cmd;
1628 1629 1630 1631 1632
	unsigned int i;

	if (!se_sess->sess_cmd_map)
		return;

1633 1634
	for (i = 0; i < VHOST_SCSI_DEFAULT_TAGS; i++) {
		tv_cmd = &((struct vhost_scsi_cmd *)se_sess->sess_cmd_map)[i];
1635 1636

		kfree(tv_cmd->tvc_sgl);
1637
		kfree(tv_cmd->tvc_prot_sgl);
1638 1639 1640 1641
		kfree(tv_cmd->tvc_upages);
	}
}

1642 1643
static ssize_t vhost_scsi_tpg_attrib_fabric_prot_type_store(
		struct config_item *item, const char *page, size_t count)
1644
{
1645
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663
	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;
}

1664 1665
static ssize_t vhost_scsi_tpg_attrib_fabric_prot_type_show(
		struct config_item *item, char *page)
1666
{
1667
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
1668 1669 1670 1671 1672
	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);
}
1673 1674

CONFIGFS_ATTR(vhost_scsi_tpg_attrib_, fabric_prot_type);
1675 1676

static struct configfs_attribute *vhost_scsi_tpg_attrib_attrs[] = {
1677
	&vhost_scsi_tpg_attrib_attr_fabric_prot_type,
1678 1679 1680
	NULL,
};

1681 1682
static int vhost_scsi_nexus_cb(struct se_portal_group *se_tpg,
			       struct se_session *se_sess, void *p)
1683
{
1684
	struct vhost_scsi_cmd *tv_cmd;
1685
	unsigned int i;
1686

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

K
Kees Cook 已提交
1690 1691 1692
		tv_cmd->tvc_sgl = kcalloc(VHOST_SCSI_PREALLOC_SGLS,
					  sizeof(struct scatterlist),
					  GFP_KERNEL);
1693 1694 1695 1696 1697
		if (!tv_cmd->tvc_sgl) {
			pr_err("Unable to allocate tv_cmd->tvc_sgl\n");
			goto out;
		}

K
Kees Cook 已提交
1698 1699 1700
		tv_cmd->tvc_upages = kcalloc(VHOST_SCSI_PREALLOC_UPAGES,
					     sizeof(struct page *),
					     GFP_KERNEL);
1701 1702 1703 1704
		if (!tv_cmd->tvc_upages) {
			pr_err("Unable to allocate tv_cmd->tvc_upages\n");
			goto out;
		}
1705

K
Kees Cook 已提交
1706 1707 1708
		tv_cmd->tvc_prot_sgl = kcalloc(VHOST_SCSI_PREALLOC_PROT_SGLS,
					       sizeof(struct scatterlist),
					       GFP_KERNEL);
1709 1710 1711 1712
		if (!tv_cmd->tvc_prot_sgl) {
			pr_err("Unable to allocate tv_cmd->tvc_prot_sgl\n");
			goto out;
		}
1713
	}
1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
	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;
	}

1732
	tv_nexus = kzalloc(sizeof(*tv_nexus), GFP_KERNEL);
1733 1734 1735 1736 1737
	if (!tv_nexus) {
		mutex_unlock(&tpg->tv_tpg_mutex);
		pr_err("Unable to allocate struct vhost_scsi_nexus\n");
		return -ENOMEM;
	}
1738 1739
	/*
	 * Since we are running in 'demo mode' this call with generate a
1740
	 * struct se_node_acl for the vhost_scsi struct se_portal_group with
1741 1742
	 * the SCSI Initiator port name of the passed configfs group 'name'.
	 */
1743
	tv_nexus->tvn_se_sess = target_setup_session(&tpg->se_tpg,
1744 1745 1746 1747 1748 1749
					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)) {
1750
		mutex_unlock(&tpg->tv_tpg_mutex);
1751 1752
		kfree(tv_nexus);
		return -ENOMEM;
1753
	}
1754
	tpg->tpg_nexus = tv_nexus;
1755

1756
	mutex_unlock(&tpg->tv_tpg_mutex);
1757 1758 1759
	return 0;
}

1760
static int vhost_scsi_drop_nexus(struct vhost_scsi_tpg *tpg)
1761 1762
{
	struct se_session *se_sess;
1763
	struct vhost_scsi_nexus *tv_nexus;
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777

	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;
	}

1778
	if (tpg->tv_tpg_port_count != 0) {
1779
		mutex_unlock(&tpg->tv_tpg_mutex);
1780
		pr_err("Unable to remove TCM_vhost I_T Nexus with"
1781
			" active TPG port count: %d\n",
1782 1783
			tpg->tv_tpg_port_count);
		return -EBUSY;
1784 1785
	}

1786
	if (tpg->tv_tpg_vhost_count != 0) {
1787
		mutex_unlock(&tpg->tv_tpg_mutex);
1788
		pr_err("Unable to remove TCM_vhost I_T Nexus with"
1789
			" active TPG vhost count: %d\n",
1790 1791
			tpg->tv_tpg_vhost_count);
		return -EBUSY;
1792 1793
	}

1794
	pr_debug("TCM_vhost_ConfigFS: Removing I_T Nexus to emulated"
1795
		" %s Initiator Port: %s\n", vhost_scsi_dump_proto_id(tpg->tport),
1796
		tv_nexus->tvn_se_sess->se_node_acl->initiatorname);
1797

1798
	vhost_scsi_free_cmd_map_res(se_sess);
1799
	/*
1800
	 * Release the SCSI I_T Nexus to the emulated vhost Target Port
1801
	 */
1802
	target_remove_session(se_sess);
1803 1804 1805 1806 1807 1808 1809
	tpg->tpg_nexus = NULL;
	mutex_unlock(&tpg->tv_tpg_mutex);

	kfree(tv_nexus);
	return 0;
}

1810
static ssize_t vhost_scsi_tpg_nexus_show(struct config_item *item, char *page)
1811
{
1812
	struct se_portal_group *se_tpg = to_tpg(item);
1813 1814 1815
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
	struct vhost_scsi_nexus *tv_nexus;
1816 1817
	ssize_t ret;

1818 1819
	mutex_lock(&tpg->tv_tpg_mutex);
	tv_nexus = tpg->tpg_nexus;
1820
	if (!tv_nexus) {
1821
		mutex_unlock(&tpg->tv_tpg_mutex);
1822 1823 1824 1825
		return -ENODEV;
	}
	ret = snprintf(page, PAGE_SIZE, "%s\n",
			tv_nexus->tvn_se_sess->se_node_acl->initiatorname);
1826
	mutex_unlock(&tpg->tv_tpg_mutex);
1827 1828 1829 1830

	return ret;
}

1831 1832
static ssize_t vhost_scsi_tpg_nexus_store(struct config_item *item,
		const char *page, size_t count)
1833
{
1834
	struct se_portal_group *se_tpg = to_tpg(item);
1835 1836 1837 1838
	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;
1839 1840 1841 1842 1843
	int ret;
	/*
	 * Shutdown the active I_T nexus if 'NULL' is passed..
	 */
	if (!strncmp(page, "NULL", 4)) {
1844
		ret = vhost_scsi_drop_nexus(tpg);
1845 1846 1847 1848
		return (!ret) ? count : ret;
	}
	/*
	 * Otherwise make sure the passed virtual Initiator port WWN matches
1849 1850
	 * the fabric protocol_id set in vhost_scsi_make_tport(), and call
	 * vhost_scsi_make_nexus().
1851
	 */
1852
	if (strlen(page) >= VHOST_SCSI_NAMELEN) {
1853
		pr_err("Emulated NAA Sas Address: %s, exceeds"
1854
				" max: %d\n", page, VHOST_SCSI_NAMELEN);
1855 1856
		return -EINVAL;
	}
1857
	snprintf(&i_port[0], VHOST_SCSI_NAMELEN, "%s", page);
1858 1859 1860 1861 1862 1863

	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,
1864
				vhost_scsi_dump_proto_id(tport_wwn));
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
			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,
1875
				vhost_scsi_dump_proto_id(tport_wwn));
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885
			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,
1886
				vhost_scsi_dump_proto_id(tport_wwn));
1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
			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';

1902
	ret = vhost_scsi_make_nexus(tpg, port_ptr);
1903 1904 1905 1906 1907 1908
	if (ret < 0)
		return ret;

	return count;
}

1909
CONFIGFS_ATTR(vhost_scsi_tpg_, nexus);
1910

1911
static struct configfs_attribute *vhost_scsi_tpg_attrs[] = {
1912
	&vhost_scsi_tpg_attr_nexus,
1913 1914 1915
	NULL,
};

1916
static struct se_portal_group *
1917
vhost_scsi_make_tpg(struct se_wwn *wwn, const char *name)
1918
{
1919 1920
	struct vhost_scsi_tport *tport = container_of(wwn,
			struct vhost_scsi_tport, tport_wwn);
1921

1922
	struct vhost_scsi_tpg *tpg;
1923
	u16 tpgt;
1924 1925 1926 1927
	int ret;

	if (strstr(name, "tpgt_") != name)
		return ERR_PTR(-EINVAL);
1928
	if (kstrtou16(name + 5, 10, &tpgt) || tpgt >= VHOST_SCSI_MAX_TARGET)
1929 1930
		return ERR_PTR(-EINVAL);

1931
	tpg = kzalloc(sizeof(*tpg), GFP_KERNEL);
1932
	if (!tpg) {
1933
		pr_err("Unable to allocate struct vhost_scsi_tpg");
1934 1935 1936 1937 1938 1939 1940
		return ERR_PTR(-ENOMEM);
	}
	mutex_init(&tpg->tv_tpg_mutex);
	INIT_LIST_HEAD(&tpg->tv_tpg_list);
	tpg->tport = tport;
	tpg->tport_tpgt = tpgt;

1941
	ret = core_tpg_register(wwn, &tpg->se_tpg, tport->tport_proto_id);
1942 1943 1944 1945
	if (ret < 0) {
		kfree(tpg);
		return NULL;
	}
1946 1947 1948
	mutex_lock(&vhost_scsi_mutex);
	list_add_tail(&tpg->tv_tpg_list, &vhost_scsi_list);
	mutex_unlock(&vhost_scsi_mutex);
1949 1950 1951 1952

	return &tpg->se_tpg;
}

1953
static void vhost_scsi_drop_tpg(struct se_portal_group *se_tpg)
1954
{
1955 1956
	struct vhost_scsi_tpg *tpg = container_of(se_tpg,
				struct vhost_scsi_tpg, se_tpg);
1957

1958
	mutex_lock(&vhost_scsi_mutex);
1959
	list_del(&tpg->tv_tpg_list);
1960
	mutex_unlock(&vhost_scsi_mutex);
1961
	/*
1962
	 * Release the virtual I_T Nexus for this vhost TPG
1963
	 */
1964
	vhost_scsi_drop_nexus(tpg);
1965 1966 1967 1968 1969 1970 1971
	/*
	 * Deregister the se_tpg from TCM..
	 */
	core_tpg_deregister(se_tpg);
	kfree(tpg);
}

1972
static struct se_wwn *
1973
vhost_scsi_make_tport(struct target_fabric_configfs *tf,
1974 1975
		     struct config_group *group,
		     const char *name)
1976
{
1977
	struct vhost_scsi_tport *tport;
1978 1979 1980 1981
	char *ptr;
	u64 wwpn = 0;
	int off = 0;

1982
	/* if (vhost_scsi_parse_wwn(name, &wwpn, 1) < 0)
1983 1984
		return ERR_PTR(-EINVAL); */

1985
	tport = kzalloc(sizeof(*tport), GFP_KERNEL);
1986
	if (!tport) {
1987
		pr_err("Unable to allocate struct vhost_scsi_tport");
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 2016 2017
		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:
2018
	if (strlen(name) >= VHOST_SCSI_NAMELEN) {
2019
		pr_err("Emulated %s Address: %s, exceeds"
2020 2021
			" max: %d\n", name, vhost_scsi_dump_proto_id(tport),
			VHOST_SCSI_NAMELEN);
2022 2023 2024
		kfree(tport);
		return ERR_PTR(-EINVAL);
	}
2025
	snprintf(&tport->tport_name[0], VHOST_SCSI_NAMELEN, "%s", &name[off]);
2026 2027

	pr_debug("TCM_VHost_ConfigFS: Allocated emulated Target"
2028
		" %s Address: %s\n", vhost_scsi_dump_proto_id(tport), name);
2029 2030 2031 2032

	return &tport->tport_wwn;
}

2033
static void vhost_scsi_drop_tport(struct se_wwn *wwn)
2034
{
2035 2036
	struct vhost_scsi_tport *tport = container_of(wwn,
				struct vhost_scsi_tport, tport_wwn);
2037 2038

	pr_debug("TCM_VHost_ConfigFS: Deallocating emulated Target"
2039
		" %s Address: %s\n", vhost_scsi_dump_proto_id(tport),
2040 2041 2042 2043 2044
		tport->tport_name);

	kfree(tport);
}

2045
static ssize_t
2046
vhost_scsi_wwn_version_show(struct config_item *item, char *page)
2047 2048
{
	return sprintf(page, "TCM_VHOST fabric module %s on %s/%s"
2049
		"on "UTS_RELEASE"\n", VHOST_SCSI_VERSION, utsname()->sysname,
2050 2051 2052
		utsname()->machine);
}

2053
CONFIGFS_ATTR_RO(vhost_scsi_wwn_, version);
2054

2055
static struct configfs_attribute *vhost_scsi_wwn_attrs[] = {
2056
	&vhost_scsi_wwn_attr_version,
2057 2058 2059
	NULL,
};

2060
static const struct target_core_fabric_ops vhost_scsi_ops = {
2061 2062
	.module				= THIS_MODULE,
	.name				= "vhost",
2063 2064 2065 2066 2067 2068 2069
	.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,
2070
	.tpg_check_prot_fabric_only	= vhost_scsi_check_prot_fabric_only,
2071 2072
	.tpg_get_inst_index		= vhost_scsi_tpg_get_inst_index,
	.release_cmd			= vhost_scsi_release_cmd,
2073
	.check_stop_free		= vhost_scsi_check_stop_free,
2074
	.sess_get_index			= vhost_scsi_sess_get_index,
2075
	.sess_get_initiator_sid		= NULL,
2076 2077 2078 2079 2080 2081 2082 2083
	.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,
2084 2085 2086
	/*
	 * Setup callers for generic logic in target_core_fabric_configfs.c
	 */
2087 2088 2089 2090 2091 2092
	.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,
2093 2094 2095 2096

	.tfc_wwn_attrs			= vhost_scsi_wwn_attrs,
	.tfc_tpg_base_attrs		= vhost_scsi_tpg_attrs,
	.tfc_tpg_attrib_attrs		= vhost_scsi_tpg_attrib_attrs,
2097 2098
};

2099
static int __init vhost_scsi_init(void)
2100
{
2101
	int ret = -ENOMEM;
2102

2103
	pr_debug("TCM_VHOST fabric module %s on %s/%s"
2104
		" on "UTS_RELEASE"\n", VHOST_SCSI_VERSION, utsname()->sysname,
2105 2106
		utsname()->machine);

2107 2108 2109 2110
	/*
	 * Use our own dedicated workqueue for submitting I/O into
	 * target core to avoid contention within system_wq.
	 */
2111 2112
	vhost_scsi_workqueue = alloc_workqueue("vhost_scsi", 0, 0);
	if (!vhost_scsi_workqueue)
2113 2114 2115 2116 2117 2118
		goto out;

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

2119
	ret = target_register_template(&vhost_scsi_ops);
2120 2121 2122 2123 2124 2125 2126 2127
	if (ret < 0)
		goto out_vhost_scsi_deregister;

	return 0;

out_vhost_scsi_deregister:
	vhost_scsi_deregister();
out_destroy_workqueue:
2128
	destroy_workqueue(vhost_scsi_workqueue);
2129 2130 2131 2132
out:
	return ret;
};

2133
static void vhost_scsi_exit(void)
2134
{
2135
	target_unregister_template(&vhost_scsi_ops);
2136
	vhost_scsi_deregister();
2137
	destroy_workqueue(vhost_scsi_workqueue);
2138 2139
};

M
Michael S. Tsirkin 已提交
2140 2141
MODULE_DESCRIPTION("VHOST_SCSI series fabric driver");
MODULE_ALIAS("tcm_vhost");
2142
MODULE_LICENSE("GPL");
2143 2144
module_init(vhost_scsi_init);
module_exit(vhost_scsi_exit);