virtio_scsi.c 28.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * Virtio SCSI HBA driver
 *
 * Copyright IBM Corp. 2010
 * Copyright Red Hat, Inc. 2011
 *
 * Authors:
 *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
 *  Paolo Bonzini   <pbonzini@redhat.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 *
 */

16 17
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

18 19 20 21 22 23 24
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/mempool.h>
#include <linux/virtio.h>
#include <linux/virtio_ids.h>
#include <linux/virtio_config.h>
#include <linux/virtio_scsi.h>
25
#include <linux/cpu.h>
26
#include <linux/blkdev.h>
27 28 29
#include <scsi/scsi_host.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_cmnd.h>
30
#include <scsi/scsi_tcq.h>
31
#include <linux/seqlock.h>
32 33

#define VIRTIO_SCSI_MEMPOOL_SZ 64
34
#define VIRTIO_SCSI_EVENT_LEN 8
35
#define VIRTIO_SCSI_VQ_BASE 2
36 37 38 39 40 41 42

/* Command queue element */
struct virtio_scsi_cmd {
	struct scsi_cmnd *sc;
	struct completion *comp;
	union {
		struct virtio_scsi_cmd_req       cmd;
43
		struct virtio_scsi_cmd_req_pi    cmd_pi;
44 45 46 47 48 49 50 51 52 53 54
		struct virtio_scsi_ctrl_tmf_req  tmf;
		struct virtio_scsi_ctrl_an_req   an;
	} req;
	union {
		struct virtio_scsi_cmd_resp      cmd;
		struct virtio_scsi_ctrl_tmf_resp tmf;
		struct virtio_scsi_ctrl_an_resp  an;
		struct virtio_scsi_event         evt;
	} resp;
} ____cacheline_aligned_in_smp;

55 56 57 58 59 60
struct virtio_scsi_event_node {
	struct virtio_scsi *vscsi;
	struct virtio_scsi_event event;
	struct work_struct work;
};

61 62 63 64 65 66 67
struct virtio_scsi_vq {
	/* Protects vq */
	spinlock_t vq_lock;

	struct virtqueue *vq;
};

68 69 70 71 72 73 74 75 76 77 78 79
/*
 * Per-target queue state.
 *
 * This struct holds the data needed by the queue steering policy.  When a
 * target is sent multiple requests, we need to drive them to the same queue so
 * that FIFO processing order is kept.  However, if a target was idle, we can
 * choose a queue arbitrarily.  In this case the queue is chosen according to
 * the current VCPU, so the driver expects the number of request queues to be
 * equal to the number of VCPUs.  This makes it easy and fast to select the
 * queue, and also lets the driver optimize the IRQ affinity for the virtqueues
 * (each virtqueue's affinity is set to the CPU that "owns" the queue).
 *
80
 * tgt_seq is held to serialize reading and writing req_vq.
81
 *
82 83 84
 * Decrements of reqs are never concurrent with writes of req_vq: before the
 * decrement reqs will be != 0; after the decrement the virtqueue completion
 * routine will not use the req_vq so it can be changed by a new request.
85
 * Thus they can happen outside the tgt_seq, provided of course we make reqs
86 87
 * an atomic_t.
 */
88
struct virtio_scsi_target_state {
89
	seqcount_t tgt_seq;
90 91 92 93 94 95

	/* Count of outstanding requests. */
	atomic_t reqs;

	/* Currently active virtqueue for requests sent to this target. */
	struct virtio_scsi_vq *req_vq;
96 97
};

98 99 100
/* Driver instance state */
struct virtio_scsi {
	struct virtio_device *vdev;
101

102 103
	/* Get some buffers ready for event vq */
	struct virtio_scsi_event_node event_list[VIRTIO_SCSI_EVENT_LEN];
104 105 106 107 108 109

	u32 num_queues;

	/* If the affinity hint is set for virtqueues */
	bool affinity_hint_set;

110 111 112
	/* CPU hotplug notifier */
	struct notifier_block nb;

113 114 115
	struct virtio_scsi_vq ctrl_vq;
	struct virtio_scsi_vq event_vq;
	struct virtio_scsi_vq req_vqs[];
116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
};

static struct kmem_cache *virtscsi_cmd_cache;
static mempool_t *virtscsi_cmd_pool;

static inline struct Scsi_Host *virtio_scsi_host(struct virtio_device *vdev)
{
	return vdev->priv;
}

static void virtscsi_compute_resid(struct scsi_cmnd *sc, u32 resid)
{
	if (!resid)
		return;

	if (!scsi_bidi_cmnd(sc)) {
		scsi_set_resid(sc, resid);
		return;
	}

	scsi_in(sc)->resid = min(resid, scsi_in(sc)->length);
	scsi_out(sc)->resid = resid - scsi_in(sc)->resid;
}

/**
 * virtscsi_complete_cmd - finish a scsi_cmd and invoke scsi_done
 *
 * Called with vq_lock held.
 */
145
static void virtscsi_complete_cmd(struct virtio_scsi *vscsi, void *buf)
146 147 148 149
{
	struct virtio_scsi_cmd *cmd = buf;
	struct scsi_cmnd *sc = cmd->sc;
	struct virtio_scsi_cmd_resp *resp = &cmd->resp.cmd;
150 151
	struct virtio_scsi_target_state *tgt =
				scsi_target(sc->device)->hostdata;
152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204

	dev_dbg(&sc->device->sdev_gendev,
		"cmd %p response %u status %#02x sense_len %u\n",
		sc, resp->response, resp->status, resp->sense_len);

	sc->result = resp->status;
	virtscsi_compute_resid(sc, resp->resid);
	switch (resp->response) {
	case VIRTIO_SCSI_S_OK:
		set_host_byte(sc, DID_OK);
		break;
	case VIRTIO_SCSI_S_OVERRUN:
		set_host_byte(sc, DID_ERROR);
		break;
	case VIRTIO_SCSI_S_ABORTED:
		set_host_byte(sc, DID_ABORT);
		break;
	case VIRTIO_SCSI_S_BAD_TARGET:
		set_host_byte(sc, DID_BAD_TARGET);
		break;
	case VIRTIO_SCSI_S_RESET:
		set_host_byte(sc, DID_RESET);
		break;
	case VIRTIO_SCSI_S_BUSY:
		set_host_byte(sc, DID_BUS_BUSY);
		break;
	case VIRTIO_SCSI_S_TRANSPORT_FAILURE:
		set_host_byte(sc, DID_TRANSPORT_DISRUPTED);
		break;
	case VIRTIO_SCSI_S_TARGET_FAILURE:
		set_host_byte(sc, DID_TARGET_FAILURE);
		break;
	case VIRTIO_SCSI_S_NEXUS_FAILURE:
		set_host_byte(sc, DID_NEXUS_FAILURE);
		break;
	default:
		scmd_printk(KERN_WARNING, sc, "Unknown response %d",
			    resp->response);
		/* fall through */
	case VIRTIO_SCSI_S_FAILURE:
		set_host_byte(sc, DID_ERROR);
		break;
	}

	WARN_ON(resp->sense_len > VIRTIO_SCSI_SENSE_SIZE);
	if (sc->sense_buffer) {
		memcpy(sc->sense_buffer, resp->sense,
		       min_t(u32, resp->sense_len, VIRTIO_SCSI_SENSE_SIZE));
		if (resp->sense_len)
			set_driver_byte(sc, DRIVER_SENSE);
	}

	sc->scsi_done(sc);
205 206

	atomic_dec(&tgt->reqs);
207 208
}

209 210
static void virtscsi_vq_done(struct virtio_scsi *vscsi,
			     struct virtio_scsi_vq *virtscsi_vq,
211
			     void (*fn)(struct virtio_scsi *vscsi, void *buf))
212 213 214
{
	void *buf;
	unsigned int len;
215 216
	unsigned long flags;
	struct virtqueue *vq = virtscsi_vq->vq;
217

218
	spin_lock_irqsave(&virtscsi_vq->vq_lock, flags);
219 220 221
	do {
		virtqueue_disable_cb(vq);
		while ((buf = virtqueue_get_buf(vq, &len)) != NULL)
222
			fn(vscsi, buf);
223 224 225

		if (unlikely(virtqueue_is_broken(vq)))
			break;
226
	} while (!virtqueue_enable_cb(vq));
227
	spin_unlock_irqrestore(&virtscsi_vq->vq_lock, flags);
228 229 230 231
}

static void virtscsi_req_done(struct virtqueue *vq)
{
232 233
	struct Scsi_Host *sh = virtio_scsi_host(vq->vdev);
	struct virtio_scsi *vscsi = shost_priv(sh);
234 235 236 237
	int index = vq->index - VIRTIO_SCSI_VQ_BASE;
	struct virtio_scsi_vq *req_vq = &vscsi->req_vqs[index];

	virtscsi_vq_done(vscsi, req_vq, virtscsi_complete_cmd);
238 239
};

240 241 242 243 244 245 246 247 248 249
static void virtscsi_poll_requests(struct virtio_scsi *vscsi)
{
	int i, num_vqs;

	num_vqs = vscsi->num_queues;
	for (i = 0; i < num_vqs; i++)
		virtscsi_vq_done(vscsi, &vscsi->req_vqs[i],
				 virtscsi_complete_cmd);
}

250
static void virtscsi_complete_free(struct virtio_scsi *vscsi, void *buf)
251 252 253 254 255 256 257 258 259
{
	struct virtio_scsi_cmd *cmd = buf;

	if (cmd->comp)
		complete_all(cmd->comp);
}

static void virtscsi_ctrl_done(struct virtqueue *vq)
{
260 261 262
	struct Scsi_Host *sh = virtio_scsi_host(vq->vdev);
	struct virtio_scsi *vscsi = shost_priv(sh);

263
	virtscsi_vq_done(vscsi, &vscsi->ctrl_vq, virtscsi_complete_free);
264 265
};

266 267
static void virtscsi_handle_event(struct work_struct *work);

268 269 270
static int virtscsi_kick_event(struct virtio_scsi *vscsi,
			       struct virtio_scsi_event_node *event_node)
{
271
	int err;
272 273 274
	struct scatterlist sg;
	unsigned long flags;

275
	INIT_WORK(&event_node->work, virtscsi_handle_event);
276
	sg_init_one(&sg, &event_node->event, sizeof(struct virtio_scsi_event));
277 278 279

	spin_lock_irqsave(&vscsi->event_vq.vq_lock, flags);

280 281
	err = virtqueue_add_inbuf(vscsi->event_vq.vq, &sg, 1, event_node,
				  GFP_ATOMIC);
282
	if (!err)
283 284 285 286
		virtqueue_kick(vscsi->event_vq.vq);

	spin_unlock_irqrestore(&vscsi->event_vq.vq_lock, flags);

287
	return err;
288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
}

static int virtscsi_kick_event_all(struct virtio_scsi *vscsi)
{
	int i;

	for (i = 0; i < VIRTIO_SCSI_EVENT_LEN; i++) {
		vscsi->event_list[i].vscsi = vscsi;
		virtscsi_kick_event(vscsi, &vscsi->event_list[i]);
	}

	return 0;
}

static void virtscsi_cancel_event_work(struct virtio_scsi *vscsi)
{
	int i;

	for (i = 0; i < VIRTIO_SCSI_EVENT_LEN; i++)
		cancel_work_sync(&vscsi->event_list[i].work);
}

static void virtscsi_handle_transport_reset(struct virtio_scsi *vscsi,
311
					    struct virtio_scsi_event *event)
312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
{
	struct scsi_device *sdev;
	struct Scsi_Host *shost = virtio_scsi_host(vscsi->vdev);
	unsigned int target = event->lun[1];
	unsigned int lun = (event->lun[2] << 8) | event->lun[3];

	switch (event->reason) {
	case VIRTIO_SCSI_EVT_RESET_RESCAN:
		scsi_add_device(shost, 0, target, lun);
		break;
	case VIRTIO_SCSI_EVT_RESET_REMOVED:
		sdev = scsi_device_lookup(shost, 0, target, lun);
		if (sdev) {
			scsi_remove_device(sdev);
			scsi_device_put(sdev);
		} else {
			pr_err("SCSI device %d 0 %d %d not found\n",
				shost->host_no, target, lun);
		}
		break;
	default:
		pr_info("Unsupport virtio scsi event reason %x\n", event->reason);
	}
}

337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
static void virtscsi_handle_param_change(struct virtio_scsi *vscsi,
					 struct virtio_scsi_event *event)
{
	struct scsi_device *sdev;
	struct Scsi_Host *shost = virtio_scsi_host(vscsi->vdev);
	unsigned int target = event->lun[1];
	unsigned int lun = (event->lun[2] << 8) | event->lun[3];
	u8 asc = event->reason & 255;
	u8 ascq = event->reason >> 8;

	sdev = scsi_device_lookup(shost, 0, target, lun);
	if (!sdev) {
		pr_err("SCSI device %d 0 %d %d not found\n",
			shost->host_no, target, lun);
		return;
	}

	/* Handle "Parameters changed", "Mode parameters changed", and
	   "Capacity data has changed".  */
	if (asc == 0x2a && (ascq == 0x00 || ascq == 0x01 || ascq == 0x09))
		scsi_rescan_device(&sdev->sdev_gendev);

	scsi_device_put(sdev);
}

362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379
static void virtscsi_handle_event(struct work_struct *work)
{
	struct virtio_scsi_event_node *event_node =
		container_of(work, struct virtio_scsi_event_node, work);
	struct virtio_scsi *vscsi = event_node->vscsi;
	struct virtio_scsi_event *event = &event_node->event;

	if (event->event & VIRTIO_SCSI_T_EVENTS_MISSED) {
		event->event &= ~VIRTIO_SCSI_T_EVENTS_MISSED;
		scsi_scan_host(virtio_scsi_host(vscsi->vdev));
	}

	switch (event->event) {
	case VIRTIO_SCSI_T_NO_EVENT:
		break;
	case VIRTIO_SCSI_T_TRANSPORT_RESET:
		virtscsi_handle_transport_reset(vscsi, event);
		break;
380 381 382
	case VIRTIO_SCSI_T_PARAM_CHANGE:
		virtscsi_handle_param_change(vscsi, event);
		break;
383 384 385 386 387 388
	default:
		pr_err("Unsupport virtio scsi event %x\n", event->event);
	}
	virtscsi_kick_event(vscsi, event_node);
}

389
static void virtscsi_complete_event(struct virtio_scsi *vscsi, void *buf)
390 391 392 393 394 395
{
	struct virtio_scsi_event_node *event_node = buf;

	schedule_work(&event_node->work);
}

396 397
static void virtscsi_event_done(struct virtqueue *vq)
{
398 399 400
	struct Scsi_Host *sh = virtio_scsi_host(vq->vdev);
	struct virtio_scsi *vscsi = shost_priv(sh);

401
	virtscsi_vq_done(vscsi, &vscsi->event_vq, virtscsi_complete_event);
402 403 404
};

/**
405 406
 * virtscsi_add_cmd - add a virtio_scsi_cmd to a virtqueue
 * @vq		: the struct virtqueue we're talking about
407 408 409 410
 * @cmd		: command structure
 * @req_size	: size of the request buffer
 * @resp_size	: size of the response buffer
 */
411 412
static int virtscsi_add_cmd(struct virtqueue *vq,
			    struct virtio_scsi_cmd *cmd,
413
			    size_t req_size, size_t resp_size)
414 415
{
	struct scsi_cmnd *sc = cmd->sc;
416
	struct scatterlist *sgs[6], req, resp;
417 418 419 420 421 422 423 424 425 426 427
	struct sg_table *out, *in;
	unsigned out_num = 0, in_num = 0;

	out = in = NULL;

	if (sc && sc->sc_data_direction != DMA_NONE) {
		if (sc->sc_data_direction != DMA_FROM_DEVICE)
			out = &scsi_out(sc)->table;
		if (sc->sc_data_direction != DMA_TO_DEVICE)
			in = &scsi_in(sc)->table;
	}
428 429

	/* Request header.  */
430 431
	sg_init_one(&req, &cmd->req, req_size);
	sgs[out_num++] = &req;
432 433

	/* Data-out buffer.  */
434 435 436 437
	if (out) {
		/* Place WRITE protection SGLs before Data OUT payload */
		if (scsi_prot_sg_count(sc))
			sgs[out_num++] = scsi_prot_sglist(sc);
438
		sgs[out_num++] = out->sgl;
439
	}
440 441

	/* Response header.  */
442 443
	sg_init_one(&resp, &cmd->resp, resp_size);
	sgs[out_num + in_num++] = &resp;
444 445

	/* Data-in buffer */
446 447 448 449
	if (in) {
		/* Place READ protection SGLs before Data IN payload */
		if (scsi_prot_sg_count(sc))
			sgs[out_num + in_num++] = scsi_prot_sglist(sc);
450
		sgs[out_num + in_num++] = in->sgl;
451
	}
452

453
	return virtqueue_add_sgs(vq, sgs, out_num, in_num, cmd, GFP_ATOMIC);
454 455
}

456
static int virtscsi_kick_cmd(struct virtio_scsi_vq *vq,
457
			     struct virtio_scsi_cmd *cmd,
458
			     size_t req_size, size_t resp_size)
459 460
{
	unsigned long flags;
461 462
	int err;
	bool needs_kick = false;
463

464
	spin_lock_irqsave(&vq->vq_lock, flags);
465
	err = virtscsi_add_cmd(vq->vq, cmd, req_size, resp_size);
466 467
	if (!err)
		needs_kick = virtqueue_kick_prepare(vq->vq);
468

469
	spin_unlock_irqrestore(&vq->vq_lock, flags);
470

471
	if (needs_kick)
472
		virtqueue_notify(vq->vq);
473
	return err;
474 475
}

476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507
static void virtio_scsi_init_hdr(struct virtio_scsi_cmd_req *cmd,
				 struct scsi_cmnd *sc)
{
	cmd->lun[0] = 1;
	cmd->lun[1] = sc->device->id;
	cmd->lun[2] = (sc->device->lun >> 8) | 0x40;
	cmd->lun[3] = sc->device->lun & 0xff;
	cmd->tag = (unsigned long)sc;
	cmd->task_attr = VIRTIO_SCSI_S_SIMPLE;
	cmd->prio = 0;
	cmd->crn = 0;
}

static void virtio_scsi_init_hdr_pi(struct virtio_scsi_cmd_req_pi *cmd_pi,
				    struct scsi_cmnd *sc)
{
	struct request *rq = sc->request;
	struct blk_integrity *bi;

	virtio_scsi_init_hdr((struct virtio_scsi_cmd_req *)cmd_pi, sc);

	if (!rq || !scsi_prot_sg_count(sc))
		return;

	bi = blk_get_integrity(rq->rq_disk);

	if (sc->sc_data_direction == DMA_TO_DEVICE)
		cmd_pi->pi_bytesout = blk_rq_sectors(rq) * bi->tuple_size;
	else if (sc->sc_data_direction == DMA_FROM_DEVICE)
		cmd_pi->pi_bytesin = blk_rq_sectors(rq) * bi->tuple_size;
}

508 509 510
static int virtscsi_queuecommand(struct virtio_scsi *vscsi,
				 struct virtio_scsi_vq *req_vq,
				 struct scsi_cmnd *sc)
511
{
512
	struct Scsi_Host *shost = virtio_scsi_host(vscsi->vdev);
C
Christoph Hellwig 已提交
513
	struct virtio_scsi_cmd *cmd = scsi_cmd_priv(sc);
514
	int req_size;
C
Christoph Hellwig 已提交
515

516 517 518 519 520
	BUG_ON(scsi_sg_count(sc) > shost->sg_tablesize);

	/* TODO: check feature bit and fail if unsupported?  */
	BUG_ON(sc->sc_data_direction == DMA_BIDIRECTIONAL);

521 522 523 524 525 526 527 528
	dev_dbg(&sc->device->sdev_gendev,
		"cmd %p CDB: %#02x\n", sc, sc->cmnd[0]);

	memset(cmd, 0, sizeof(*cmd));
	cmd->sc = sc;

	BUG_ON(sc->cmd_len > VIRTIO_SCSI_CDB_SIZE);

529 530 531 532 533 534 535 536 537 538
	if (virtio_has_feature(vscsi->vdev, VIRTIO_SCSI_F_T10_PI)) {
		virtio_scsi_init_hdr_pi(&cmd->req.cmd_pi, sc);
		memcpy(cmd->req.cmd_pi.cdb, sc->cmnd, sc->cmd_len);
		req_size = sizeof(cmd->req.cmd_pi);
	} else {
		virtio_scsi_init_hdr(&cmd->req.cmd, sc);
		memcpy(cmd->req.cmd.cdb, sc->cmnd, sc->cmd_len);
		req_size = sizeof(cmd->req.cmd);
	}

539
	if (virtscsi_kick_cmd(req_vq, cmd, req_size, sizeof(cmd->resp.cmd)) != 0)
C
Christoph Hellwig 已提交
540 541
		return SCSI_MLQUEUE_HOST_BUSY;
	return 0;
542 543
}

544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561
static int virtscsi_queuecommand_single(struct Scsi_Host *sh,
					struct scsi_cmnd *sc)
{
	struct virtio_scsi *vscsi = shost_priv(sh);
	struct virtio_scsi_target_state *tgt =
				scsi_target(sc->device)->hostdata;

	atomic_inc(&tgt->reqs);
	return virtscsi_queuecommand(vscsi, &vscsi->req_vqs[0], sc);
}

static struct virtio_scsi_vq *virtscsi_pick_vq(struct virtio_scsi *vscsi,
					       struct virtio_scsi_target_state *tgt)
{
	struct virtio_scsi_vq *vq;
	unsigned long flags;
	u32 queue_num;

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578
	local_irq_save(flags);
	if (atomic_inc_return(&tgt->reqs) > 1) {
		unsigned long seq;

		do {
			seq = read_seqcount_begin(&tgt->tgt_seq);
			vq = tgt->req_vq;
		} while (read_seqcount_retry(&tgt->tgt_seq, seq));
	} else {
		/* no writes can be concurrent because of atomic_t */
		write_seqcount_begin(&tgt->tgt_seq);

		/* keep previous req_vq if a reader just arrived */
		if (unlikely(atomic_read(&tgt->reqs) > 1)) {
			vq = tgt->req_vq;
			goto unlock;
		}
579 580 581 582 583

		queue_num = smp_processor_id();
		while (unlikely(queue_num >= vscsi->num_queues))
			queue_num -= vscsi->num_queues;
		tgt->req_vq = vq = &vscsi->req_vqs[queue_num];
584 585
 unlock:
		write_seqcount_end(&tgt->tgt_seq);
586
	}
587
	local_irq_restore(flags);
588 589 590 591 592 593 594 595 596 597 598 599 600 601 602

	return vq;
}

static int virtscsi_queuecommand_multi(struct Scsi_Host *sh,
				       struct scsi_cmnd *sc)
{
	struct virtio_scsi *vscsi = shost_priv(sh);
	struct virtio_scsi_target_state *tgt =
				scsi_target(sc->device)->hostdata;
	struct virtio_scsi_vq *req_vq = virtscsi_pick_vq(vscsi, tgt);

	return virtscsi_queuecommand(vscsi, req_vq, sc);
}

603 604 605
static int virtscsi_tmf(struct virtio_scsi *vscsi, struct virtio_scsi_cmd *cmd)
{
	DECLARE_COMPLETION_ONSTACK(comp);
606
	int ret = FAILED;
607 608

	cmd->comp = &comp;
609
	if (virtscsi_kick_cmd(&vscsi->ctrl_vq, cmd,
610
			      sizeof cmd->req.tmf, sizeof cmd->resp.tmf) < 0)
611
		goto out;
612 613

	wait_for_completion(&comp);
614 615 616
	if (cmd->resp.tmf.response == VIRTIO_SCSI_S_OK ||
	    cmd->resp.tmf.response == VIRTIO_SCSI_S_FUNCTION_SUCCEEDED)
		ret = SUCCESS;
617

618 619 620 621 622 623 624 625 626 627 628 629
	/*
	 * The spec guarantees that all requests related to the TMF have
	 * been completed, but the callback might not have run yet if
	 * we're using independent interrupts (e.g. MSI).  Poll the
	 * virtqueues once.
	 *
	 * In the abort case, sc->scsi_done will do nothing, because
	 * the block layer must have detected a timeout and as a result
	 * REQ_ATOM_COMPLETE has been set.
	 */
	virtscsi_poll_requests(vscsi);

630 631 632
out:
	mempool_free(cmd, virtscsi_cmd_pool);
	return ret;
633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657
}

static int virtscsi_device_reset(struct scsi_cmnd *sc)
{
	struct virtio_scsi *vscsi = shost_priv(sc->device->host);
	struct virtio_scsi_cmd *cmd;

	sdev_printk(KERN_INFO, sc->device, "device reset\n");
	cmd = mempool_alloc(virtscsi_cmd_pool, GFP_NOIO);
	if (!cmd)
		return FAILED;

	memset(cmd, 0, sizeof(*cmd));
	cmd->sc = sc;
	cmd->req.tmf = (struct virtio_scsi_ctrl_tmf_req){
		.type = VIRTIO_SCSI_T_TMF,
		.subtype = VIRTIO_SCSI_T_TMF_LOGICAL_UNIT_RESET,
		.lun[0] = 1,
		.lun[1] = sc->device->id,
		.lun[2] = (sc->device->lun >> 8) | 0x40,
		.lun[3] = sc->device->lun & 0xff,
	};
	return virtscsi_tmf(vscsi, cmd);
}

658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687
/**
 * virtscsi_change_queue_depth() - Change a virtscsi target's queue depth
 * @sdev:	Virtscsi target whose queue depth to change
 * @qdepth:	New queue depth
 * @reason:	Reason for the queue depth change.
 */
static int virtscsi_change_queue_depth(struct scsi_device *sdev,
				       int qdepth,
				       int reason)
{
	struct Scsi_Host *shost = sdev->host;
	int max_depth = shost->cmd_per_lun;

	switch (reason) {
	case SCSI_QDEPTH_QFULL: /* Drop qdepth in response to BUSY state */
		scsi_track_queue_full(sdev, qdepth);
		break;
	case SCSI_QDEPTH_RAMP_UP: /* Raise qdepth after BUSY state resolved */
	case SCSI_QDEPTH_DEFAULT: /* Manual change via sysfs */
		scsi_adjust_queue_depth(sdev,
					scsi_get_tag_type(sdev),
					min(max_depth, qdepth));
		break;
	default:
		return -EOPNOTSUPP;
	}

	return sdev->queue_depth;
}

688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711
static int virtscsi_abort(struct scsi_cmnd *sc)
{
	struct virtio_scsi *vscsi = shost_priv(sc->device->host);
	struct virtio_scsi_cmd *cmd;

	scmd_printk(KERN_INFO, sc, "abort\n");
	cmd = mempool_alloc(virtscsi_cmd_pool, GFP_NOIO);
	if (!cmd)
		return FAILED;

	memset(cmd, 0, sizeof(*cmd));
	cmd->sc = sc;
	cmd->req.tmf = (struct virtio_scsi_ctrl_tmf_req){
		.type = VIRTIO_SCSI_T_TMF,
		.subtype = VIRTIO_SCSI_T_TMF_ABORT_TASK,
		.lun[0] = 1,
		.lun[1] = sc->device->id,
		.lun[2] = (sc->device->lun >> 8) | 0x40,
		.lun[3] = sc->device->lun & 0xff,
		.tag = (unsigned long)sc,
	};
	return virtscsi_tmf(vscsi, cmd);
}

712 713
static int virtscsi_target_alloc(struct scsi_target *starget)
{
714 715 716
	struct Scsi_Host *sh = dev_to_shost(starget->dev.parent);
	struct virtio_scsi *vscsi = shost_priv(sh);

717 718 719 720 721
	struct virtio_scsi_target_state *tgt =
				kmalloc(sizeof(*tgt), GFP_KERNEL);
	if (!tgt)
		return -ENOMEM;

722
	seqcount_init(&tgt->tgt_seq);
723
	atomic_set(&tgt->reqs, 0);
724
	tgt->req_vq = &vscsi->req_vqs[0];
725 726 727 728 729 730 731 732 733 734 735

	starget->hostdata = tgt;
	return 0;
}

static void virtscsi_target_destroy(struct scsi_target *starget)
{
	struct virtio_scsi_target_state *tgt = starget->hostdata;
	kfree(tgt);
}

736 737 738 739 740
static struct scsi_host_template virtscsi_host_template_single = {
	.module = THIS_MODULE,
	.name = "Virtio SCSI HBA",
	.proc_name = "virtio_scsi",
	.this_id = -1,
C
Christoph Hellwig 已提交
741
	.cmd_size = sizeof(struct virtio_scsi_cmd),
742
	.queuecommand = virtscsi_queuecommand_single,
743
	.change_queue_depth = virtscsi_change_queue_depth,
744 745 746 747 748 749 750 751 752 753 754
	.eh_abort_handler = virtscsi_abort,
	.eh_device_reset_handler = virtscsi_device_reset,

	.can_queue = 1024,
	.dma_boundary = UINT_MAX,
	.use_clustering = ENABLE_CLUSTERING,
	.target_alloc = virtscsi_target_alloc,
	.target_destroy = virtscsi_target_destroy,
};

static struct scsi_host_template virtscsi_host_template_multi = {
755 756 757 758
	.module = THIS_MODULE,
	.name = "Virtio SCSI HBA",
	.proc_name = "virtio_scsi",
	.this_id = -1,
C
Christoph Hellwig 已提交
759
	.cmd_size = sizeof(struct virtio_scsi_cmd),
760
	.queuecommand = virtscsi_queuecommand_multi,
761
	.change_queue_depth = virtscsi_change_queue_depth,
762 763 764 765 766 767
	.eh_abort_handler = virtscsi_abort,
	.eh_device_reset_handler = virtscsi_device_reset,

	.can_queue = 1024,
	.dma_boundary = UINT_MAX,
	.use_clustering = ENABLE_CLUSTERING,
768 769
	.target_alloc = virtscsi_target_alloc,
	.target_destroy = virtscsi_target_destroy,
770 771 772 773 774
};

#define virtscsi_config_get(vdev, fld) \
	({ \
		typeof(((struct virtio_scsi_config *)0)->fld) __val; \
775
		virtio_cread(vdev, struct virtio_scsi_config, fld, &__val); \
776 777 778 779
		__val; \
	})

#define virtscsi_config_set(vdev, fld, val) \
780
	do { \
781
		typeof(((struct virtio_scsi_config *)0)->fld) __val = (val); \
782 783
		virtio_cwrite(vdev, struct virtio_scsi_config, fld, &__val); \
	} while(0)
784

785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
static void __virtscsi_set_affinity(struct virtio_scsi *vscsi, bool affinity)
{
	int i;
	int cpu;

	/* In multiqueue mode, when the number of cpu is equal
	 * to the number of request queues, we let the qeueues
	 * to be private to one cpu by setting the affinity hint
	 * to eliminate the contention.
	 */
	if ((vscsi->num_queues == 1 ||
	     vscsi->num_queues != num_online_cpus()) && affinity) {
		if (vscsi->affinity_hint_set)
			affinity = false;
		else
			return;
	}

	if (affinity) {
		i = 0;
		for_each_online_cpu(cpu) {
			virtqueue_set_affinity(vscsi->req_vqs[i].vq, cpu);
			i++;
		}

		vscsi->affinity_hint_set = true;
	} else {
812 813 814 815
		for (i = 0; i < vscsi->num_queues; i++) {
			if (!vscsi->req_vqs[i].vq)
				continue;

816
			virtqueue_set_affinity(vscsi->req_vqs[i].vq, -1);
817
		}
818 819 820 821 822 823 824 825 826 827 828 829

		vscsi->affinity_hint_set = false;
	}
}

static void virtscsi_set_affinity(struct virtio_scsi *vscsi, bool affinity)
{
	get_online_cpus();
	__virtscsi_set_affinity(vscsi, affinity);
	put_online_cpus();
}

830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
static int virtscsi_cpu_callback(struct notifier_block *nfb,
				 unsigned long action, void *hcpu)
{
	struct virtio_scsi *vscsi = container_of(nfb, struct virtio_scsi, nb);
	switch(action) {
	case CPU_ONLINE:
	case CPU_ONLINE_FROZEN:
	case CPU_DEAD:
	case CPU_DEAD_FROZEN:
		__virtscsi_set_affinity(vscsi, true);
		break;
	default:
		break;
	}
	return NOTIFY_OK;
}

847 848 849 850 851 852 853
static void virtscsi_init_vq(struct virtio_scsi_vq *virtscsi_vq,
			     struct virtqueue *vq)
{
	spin_lock_init(&virtscsi_vq->vq_lock);
	virtscsi_vq->vq = vq;
}

854 855 856 857 858 859 860
static void virtscsi_scan(struct virtio_device *vdev)
{
	struct Scsi_Host *shost = (struct Scsi_Host *)vdev->priv;

	scsi_scan_host(shost);
}

861 862
static void virtscsi_remove_vqs(struct virtio_device *vdev)
{
863 864 865 866 867
	struct Scsi_Host *sh = virtio_scsi_host(vdev);
	struct virtio_scsi *vscsi = shost_priv(sh);

	virtscsi_set_affinity(vscsi, false);

868 869 870 871 872 873
	/* Stop all the virtqueues. */
	vdev->config->reset(vdev);

	vdev->config->del_vqs(vdev);
}

874
static int virtscsi_init(struct virtio_device *vdev,
875
			 struct virtio_scsi *vscsi)
876 877
{
	int err;
878 879 880 881 882 883 884 885 886 887 888 889 890 891 892
	u32 i;
	u32 num_vqs;
	vq_callback_t **callbacks;
	const char **names;
	struct virtqueue **vqs;

	num_vqs = vscsi->num_queues + VIRTIO_SCSI_VQ_BASE;
	vqs = kmalloc(num_vqs * sizeof(struct virtqueue *), GFP_KERNEL);
	callbacks = kmalloc(num_vqs * sizeof(vq_callback_t *), GFP_KERNEL);
	names = kmalloc(num_vqs * sizeof(char *), GFP_KERNEL);

	if (!callbacks || !vqs || !names) {
		err = -ENOMEM;
		goto out;
	}
893

894 895 896 897 898 899 900 901
	callbacks[0] = virtscsi_ctrl_done;
	callbacks[1] = virtscsi_event_done;
	names[0] = "control";
	names[1] = "event";
	for (i = VIRTIO_SCSI_VQ_BASE; i < num_vqs; i++) {
		callbacks[i] = virtscsi_req_done;
		names[i] = "request";
	}
902 903

	/* Discover virtqueues and write information to configuration.  */
904
	err = vdev->config->find_vqs(vdev, num_vqs, vqs, callbacks, names);
905
	if (err)
906
		goto out;
907

908 909
	virtscsi_init_vq(&vscsi->ctrl_vq, vqs[0]);
	virtscsi_init_vq(&vscsi->event_vq, vqs[1]);
910 911 912 913 914
	for (i = VIRTIO_SCSI_VQ_BASE; i < num_vqs; i++)
		virtscsi_init_vq(&vscsi->req_vqs[i - VIRTIO_SCSI_VQ_BASE],
				 vqs[i]);

	virtscsi_set_affinity(vscsi, true);
915 916 917

	virtscsi_config_set(vdev, cdb_size, VIRTIO_SCSI_CDB_SIZE);
	virtscsi_config_set(vdev, sense_size, VIRTIO_SCSI_SENSE_SIZE);
918

919 920 921
	if (virtio_has_feature(vdev, VIRTIO_SCSI_F_HOTPLUG))
		virtscsi_kick_event_all(vscsi);

922 923 924 925 926 927 928 929
	err = 0;

out:
	kfree(names);
	kfree(callbacks);
	kfree(vqs);
	if (err)
		virtscsi_remove_vqs(vdev);
930
	return err;
931 932
}

933
static int virtscsi_probe(struct virtio_device *vdev)
934 935 936
{
	struct Scsi_Host *shost;
	struct virtio_scsi *vscsi;
937
	int err, host_prot;
938
	u32 sg_elems, num_targets;
939
	u32 cmd_per_lun;
940 941 942 943 944
	u32 num_queues;
	struct scsi_host_template *hostt;

	/* We need to know how many queues before we allocate. */
	num_queues = virtscsi_config_get(vdev, num_queues) ? : 1;
945

946
	num_targets = virtscsi_config_get(vdev, max_target) + 1;
947

948 949 950 951 952 953 954
	if (num_queues == 1)
		hostt = &virtscsi_host_template_single;
	else
		hostt = &virtscsi_host_template_multi;

	shost = scsi_host_alloc(hostt,
		sizeof(*vscsi) + sizeof(vscsi->req_vqs[0]) * num_queues);
955 956 957
	if (!shost)
		return -ENOMEM;

958
	sg_elems = virtscsi_config_get(vdev, seg_max) ?: 1;
959 960 961
	shost->sg_tablesize = sg_elems;
	vscsi = shost_priv(shost);
	vscsi->vdev = vdev;
962
	vscsi->num_queues = num_queues;
963 964
	vdev->priv = shost;

965
	err = virtscsi_init(vdev, vscsi);
966 967 968
	if (err)
		goto virtscsi_init_failed;

969 970 971 972 973 974 975
	vscsi->nb.notifier_call = &virtscsi_cpu_callback;
	err = register_hotcpu_notifier(&vscsi->nb);
	if (err) {
		pr_err("registering cpu notifier failed\n");
		goto scsi_add_host_failed;
	}

976 977 978
	cmd_per_lun = virtscsi_config_get(vdev, cmd_per_lun) ?: 1;
	shost->cmd_per_lun = min_t(u32, cmd_per_lun, shost->can_queue);
	shost->max_sectors = virtscsi_config_get(vdev, max_sectors) ?: 0xFFFF;
979 980 981 982 983

	/* LUNs > 256 are reported with format 1, so they go in the range
	 * 16640-32767.
	 */
	shost->max_lun = virtscsi_config_get(vdev, max_lun) + 1 + 0x4000;
984
	shost->max_id = num_targets;
985 986
	shost->max_channel = 0;
	shost->max_cmd_len = VIRTIO_SCSI_CDB_SIZE;
987 988 989 990 991 992 993 994 995 996

	if (virtio_has_feature(vdev, VIRTIO_SCSI_F_T10_PI)) {
		host_prot = SHOST_DIF_TYPE1_PROTECTION | SHOST_DIF_TYPE2_PROTECTION |
			    SHOST_DIF_TYPE3_PROTECTION | SHOST_DIX_TYPE1_PROTECTION |
			    SHOST_DIX_TYPE2_PROTECTION | SHOST_DIX_TYPE3_PROTECTION;

		scsi_host_set_prot(shost, host_prot);
		scsi_host_set_guard(shost, SHOST_DIX_GUARD_CRC);
	}

997 998 999
	err = scsi_add_host(shost, &vdev->dev);
	if (err)
		goto scsi_add_host_failed;
1000 1001 1002 1003
	/*
	 * scsi_scan_host() happens in virtscsi_scan() via virtio_driver->scan()
	 * after VIRTIO_CONFIG_S_DRIVER_OK has been set..
	 */
1004 1005 1006 1007 1008 1009 1010 1011 1012
	return 0;

scsi_add_host_failed:
	vdev->config->del_vqs(vdev);
virtscsi_init_failed:
	scsi_host_put(shost);
	return err;
}

1013
static void virtscsi_remove(struct virtio_device *vdev)
1014 1015
{
	struct Scsi_Host *shost = virtio_scsi_host(vdev);
1016 1017 1018 1019
	struct virtio_scsi *vscsi = shost_priv(shost);

	if (virtio_has_feature(vdev, VIRTIO_SCSI_F_HOTPLUG))
		virtscsi_cancel_event_work(vscsi);
1020 1021 1022

	scsi_remove_host(shost);

1023 1024
	unregister_hotcpu_notifier(&vscsi->nb);

1025 1026 1027 1028
	virtscsi_remove_vqs(vdev);
	scsi_host_put(shost);
}

1029
#ifdef CONFIG_PM_SLEEP
1030 1031
static int virtscsi_freeze(struct virtio_device *vdev)
{
1032 1033 1034 1035
	struct Scsi_Host *sh = virtio_scsi_host(vdev);
	struct virtio_scsi *vscsi = shost_priv(sh);

	unregister_hotcpu_notifier(&vscsi->nb);
1036 1037 1038 1039 1040 1041 1042 1043
	virtscsi_remove_vqs(vdev);
	return 0;
}

static int virtscsi_restore(struct virtio_device *vdev)
{
	struct Scsi_Host *sh = virtio_scsi_host(vdev);
	struct virtio_scsi *vscsi = shost_priv(sh);
1044 1045 1046 1047 1048 1049 1050 1051 1052
	int err;

	err = virtscsi_init(vdev, vscsi);
	if (err)
		return err;

	err = register_hotcpu_notifier(&vscsi->nb);
	if (err)
		vdev->config->del_vqs(vdev);
1053

1054
	return err;
1055 1056 1057 1058 1059 1060 1061 1062
}
#endif

static struct virtio_device_id id_table[] = {
	{ VIRTIO_ID_SCSI, VIRTIO_DEV_ANY_ID },
	{ 0 },
};

1063
static unsigned int features[] = {
1064 1065
	VIRTIO_SCSI_F_HOTPLUG,
	VIRTIO_SCSI_F_CHANGE,
1066
	VIRTIO_SCSI_F_T10_PI,
1067 1068
};

1069
static struct virtio_driver virtio_scsi_driver = {
1070 1071
	.feature_table = features,
	.feature_table_size = ARRAY_SIZE(features),
1072 1073 1074 1075
	.driver.name = KBUILD_MODNAME,
	.driver.owner = THIS_MODULE,
	.id_table = id_table,
	.probe = virtscsi_probe,
1076
	.scan = virtscsi_scan,
1077
#ifdef CONFIG_PM_SLEEP
1078 1079 1080
	.freeze = virtscsi_freeze,
	.restore = virtscsi_restore,
#endif
1081
	.remove = virtscsi_remove,
1082 1083 1084 1085 1086 1087 1088 1089
};

static int __init init(void)
{
	int ret = -ENOMEM;

	virtscsi_cmd_cache = KMEM_CACHE(virtio_scsi_cmd, 0);
	if (!virtscsi_cmd_cache) {
1090
		pr_err("kmem_cache_create() for virtscsi_cmd_cache failed\n");
1091 1092 1093 1094 1095 1096 1097 1098
		goto error;
	}


	virtscsi_cmd_pool =
		mempool_create_slab_pool(VIRTIO_SCSI_MEMPOOL_SZ,
					 virtscsi_cmd_cache);
	if (!virtscsi_cmd_pool) {
1099
		pr_err("mempool_create() for virtscsi_cmd_pool failed\n");
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
		goto error;
	}
	ret = register_virtio_driver(&virtio_scsi_driver);
	if (ret < 0)
		goto error;

	return 0;

error:
	if (virtscsi_cmd_pool) {
		mempool_destroy(virtscsi_cmd_pool);
		virtscsi_cmd_pool = NULL;
	}
	if (virtscsi_cmd_cache) {
		kmem_cache_destroy(virtscsi_cmd_cache);
		virtscsi_cmd_cache = NULL;
	}
	return ret;
}

static void __exit fini(void)
{
	unregister_virtio_driver(&virtio_scsi_driver);
	mempool_destroy(virtscsi_cmd_pool);
	kmem_cache_destroy(virtscsi_cmd_cache);
}
module_init(init);
module_exit(fini);

MODULE_DEVICE_TABLE(virtio, id_table);
MODULE_DESCRIPTION("Virtio SCSI HBA driver");
MODULE_LICENSE("GPL");