xhci-mem.c 74.2 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*
 * xHCI host controller driver
 *
 * Copyright (C) 2008 Intel Corp.
 *
 * Author: Sarah Sharp
 * Some code borrowed from the Linux EHCI driver.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * 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.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software Foundation,
 * Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

#include <linux/usb.h>
24
#include <linux/pci.h>
25
#include <linux/slab.h>
26
#include <linux/dmapool.h>
27
#include <linux/dma-mapping.h>
28 29 30

#include "xhci.h"

31 32 33 34 35 36 37
/*
 * Allocates a generic ring segment from the ring pool, sets the dma address,
 * initializes the segment to zero, and sets the private next pointer to NULL.
 *
 * Section 4.11.1.1:
 * "All components of all Command and Transfer TRBs shall be initialized to '0'"
 */
38 39
static struct xhci_segment *xhci_segment_alloc(struct xhci_hcd *xhci,
					unsigned int cycle_state, gfp_t flags)
40 41 42
{
	struct xhci_segment *seg;
	dma_addr_t	dma;
43
	int		i;
44 45 46

	seg = kzalloc(sizeof *seg, flags);
	if (!seg)
47
		return NULL;
48 49 50 51

	seg->trbs = dma_pool_alloc(xhci->segment_pool, flags, &dma);
	if (!seg->trbs) {
		kfree(seg);
52
		return NULL;
53 54
	}

55
	memset(seg->trbs, 0, TRB_SEGMENT_SIZE);
56 57 58 59 60
	/* If the cycle state is 0, set the cycle bit to 1 for all the TRBs */
	if (cycle_state == 0) {
		for (i = 0; i < TRBS_PER_SEGMENT; i++)
			seg->trbs[i].link.control |= TRB_CYCLE;
	}
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
	seg->dma = dma;
	seg->next = NULL;

	return seg;
}

static void xhci_segment_free(struct xhci_hcd *xhci, struct xhci_segment *seg)
{
	if (seg->trbs) {
		dma_pool_free(xhci->segment_pool, seg->trbs, seg->dma);
		seg->trbs = NULL;
	}
	kfree(seg);
}

76 77 78 79 80 81 82 83 84 85 86 87 88 89
static void xhci_free_segments_for_ring(struct xhci_hcd *xhci,
				struct xhci_segment *first)
{
	struct xhci_segment *seg;

	seg = first->next;
	while (seg != first) {
		struct xhci_segment *next = seg->next;
		xhci_segment_free(xhci, seg);
		seg = next;
	}
	xhci_segment_free(xhci, first);
}

90 91 92 93 94 95 96 97
/*
 * Make the prev segment point to the next segment.
 *
 * Change the last TRB in the prev segment to be a Link TRB which points to the
 * DMA address of the next segment.  The caller needs to set any Link TRB
 * related flags, such as End TRB, Toggle Cycle, and no snoop.
 */
static void xhci_link_segments(struct xhci_hcd *xhci, struct xhci_segment *prev,
A
Andiry Xu 已提交
98
		struct xhci_segment *next, enum xhci_ring_type type)
99 100 101 102 103 104
{
	u32 val;

	if (!prev || !next)
		return;
	prev->next = next;
A
Andiry Xu 已提交
105
	if (type != TYPE_EVENT) {
106 107
		prev->trbs[TRBS_PER_SEGMENT-1].link.segment_ptr =
			cpu_to_le64(next->dma);
108 109

		/* Set the last TRB in the segment to have a TRB type ID of Link TRB */
M
Matt Evans 已提交
110
		val = le32_to_cpu(prev->trbs[TRBS_PER_SEGMENT-1].link.control);
111 112
		val &= ~TRB_TYPE_BITMASK;
		val |= TRB_TYPE(TRB_LINK);
113
		/* Always set the chain bit with 0.95 hardware */
114 115
		/* Set chain bit for isoc rings on AMD 0.96 host */
		if (xhci_link_trb_quirk(xhci) ||
A
Andiry Xu 已提交
116 117
				(type == TYPE_ISOC &&
				 (xhci->quirks & XHCI_AMD_0x96_HOST)))
118
			val |= TRB_CHAIN;
M
Matt Evans 已提交
119
		prev->trbs[TRBS_PER_SEGMENT-1].link.control = cpu_to_le32(val);
120 121 122
	}
}

A
Andiry Xu 已提交
123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
/*
 * Link the ring to the new segments.
 * Set Toggle Cycle for the new ring if needed.
 */
static void xhci_link_rings(struct xhci_hcd *xhci, struct xhci_ring *ring,
		struct xhci_segment *first, struct xhci_segment *last,
		unsigned int num_segs)
{
	struct xhci_segment *next;

	if (!ring || !first || !last)
		return;

	next = ring->enq_seg->next;
	xhci_link_segments(xhci, ring->enq_seg, first, ring->type);
	xhci_link_segments(xhci, last, next, ring->type);
	ring->num_segs += num_segs;
	ring->num_trbs_free += (TRBS_PER_SEGMENT - 1) * num_segs;

	if (ring->type != TYPE_EVENT && ring->enq_seg == ring->last_seg) {
		ring->last_seg->trbs[TRBS_PER_SEGMENT-1].link.control
			&= ~cpu_to_le32(LINK_TOGGLE);
		last->trbs[TRBS_PER_SEGMENT-1].link.control
			|= cpu_to_le32(LINK_TOGGLE);
		ring->last_seg = last;
	}
}

151
/* XXX: Do we need the hcd structure in all these functions? */
152
void xhci_ring_free(struct xhci_hcd *xhci, struct xhci_ring *ring)
153
{
154
	if (!ring)
155
		return;
156 157 158 159

	if (ring->first_seg)
		xhci_free_segments_for_ring(xhci, ring->first_seg);

160 161 162
	kfree(ring);
}

163 164
static void xhci_initialize_ring_info(struct xhci_ring *ring,
					unsigned int cycle_state)
165 166 167 168 169 170 171 172 173
{
	/* The ring is empty, so the enqueue pointer == dequeue pointer */
	ring->enqueue = ring->first_seg->trbs;
	ring->enq_seg = ring->first_seg;
	ring->dequeue = ring->enqueue;
	ring->deq_seg = ring->first_seg;
	/* The ring is initialized to 0. The producer must write 1 to the cycle
	 * bit to handover ownership of the TRB, so PCS = 1.  The consumer must
	 * compare CCS to the cycle bit to check ownership, so CCS = 1.
174 175 176
	 *
	 * New rings are initialized with cycle state equal to 1; if we are
	 * handling ring expansion, set the cycle state equal to the old ring.
177
	 */
178
	ring->cycle_state = cycle_state;
179 180 181
	/* Not necessary for new rings, but needed for re-initialized rings */
	ring->enq_updates = 0;
	ring->deq_updates = 0;
182 183 184 185 186 187

	/*
	 * Each segment has a link TRB, and leave an extra TRB for SW
	 * accounting purpose
	 */
	ring->num_trbs_free = ring->num_segs * (TRBS_PER_SEGMENT - 1) - 1;
188 189
}

190 191 192
/* Allocate segments and link them for a ring */
static int xhci_alloc_segments_for_ring(struct xhci_hcd *xhci,
		struct xhci_segment **first, struct xhci_segment **last,
193 194
		unsigned int num_segs, unsigned int cycle_state,
		enum xhci_ring_type type, gfp_t flags)
195 196 197
{
	struct xhci_segment *prev;

198
	prev = xhci_segment_alloc(xhci, cycle_state, flags);
199 200 201 202 203 204 205 206
	if (!prev)
		return -ENOMEM;
	num_segs--;

	*first = prev;
	while (num_segs > 0) {
		struct xhci_segment	*next;

207
		next = xhci_segment_alloc(xhci, cycle_state, flags);
208
		if (!next) {
209 210 211 212 213 214
			prev = *first;
			while (prev) {
				next = prev->next;
				xhci_segment_free(xhci, prev);
				prev = next;
			}
215 216 217 218 219 220 221 222 223 224 225 226 227
			return -ENOMEM;
		}
		xhci_link_segments(xhci, prev, next, type);

		prev = next;
		num_segs--;
	}
	xhci_link_segments(xhci, prev, *first, type);
	*last = prev;

	return 0;
}

228 229 230 231 232 233 234 235
/**
 * Create a new ring with zero or more segments.
 *
 * Link each segment together into a ring.
 * Set the end flag and the cycle toggle bit on the last segment.
 * See section 4.9.1 and figures 15 and 16.
 */
static struct xhci_ring *xhci_ring_alloc(struct xhci_hcd *xhci,
236 237
		unsigned int num_segs, unsigned int cycle_state,
		enum xhci_ring_type type, gfp_t flags)
238 239
{
	struct xhci_ring	*ring;
240
	int ret;
241 242 243

	ring = kzalloc(sizeof *(ring), flags);
	if (!ring)
244
		return NULL;
245

246
	ring->num_segs = num_segs;
247
	INIT_LIST_HEAD(&ring->td_list);
A
Andiry Xu 已提交
248
	ring->type = type;
249 250 251
	if (num_segs == 0)
		return ring;

252
	ret = xhci_alloc_segments_for_ring(xhci, &ring->first_seg,
253
			&ring->last_seg, num_segs, cycle_state, type, flags);
254
	if (ret)
255 256
		goto fail;

A
Andiry Xu 已提交
257 258
	/* Only event ring does not use link TRB */
	if (type != TYPE_EVENT) {
259
		/* See section 4.9.2.1 and 6.4.4.1 */
260
		ring->last_seg->trbs[TRBS_PER_SEGMENT - 1].link.control |=
261
			cpu_to_le32(LINK_TOGGLE);
262
	}
263
	xhci_initialize_ring_info(ring, cycle_state);
264 265 266
	return ring;

fail:
267
	kfree(ring);
268
	return NULL;
269 270
}

271 272 273 274 275 276 277 278 279 280
void xhci_free_or_cache_endpoint_ring(struct xhci_hcd *xhci,
		struct xhci_virt_device *virt_dev,
		unsigned int ep_index)
{
	int rings_cached;

	rings_cached = virt_dev->num_rings_cached;
	if (rings_cached < XHCI_MAX_RINGS_CACHED) {
		virt_dev->ring_cache[rings_cached] =
			virt_dev->eps[ep_index].ring;
281
		virt_dev->num_rings_cached++;
282 283
		xhci_dbg(xhci, "Cached old ring, "
				"%d ring%s cached\n",
284 285
				virt_dev->num_rings_cached,
				(virt_dev->num_rings_cached > 1) ? "s" : "");
286 287 288 289 290 291 292 293 294
	} else {
		xhci_ring_free(xhci, virt_dev->eps[ep_index].ring);
		xhci_dbg(xhci, "Ring cache full (%d rings), "
				"freeing ring\n",
				virt_dev->num_rings_cached);
	}
	virt_dev->eps[ep_index].ring = NULL;
}

295 296 297 298
/* Zero an endpoint ring (except for link TRBs) and move the enqueue and dequeue
 * pointers to the beginning of the ring.
 */
static void xhci_reinit_cached_ring(struct xhci_hcd *xhci,
299 300
			struct xhci_ring *ring, unsigned int cycle_state,
			enum xhci_ring_type type)
301 302
{
	struct xhci_segment	*seg = ring->first_seg;
303 304
	int i;

305 306 307
	do {
		memset(seg->trbs, 0,
				sizeof(union xhci_trb)*TRBS_PER_SEGMENT);
308 309 310 311
		if (cycle_state == 0) {
			for (i = 0; i < TRBS_PER_SEGMENT; i++)
				seg->trbs[i].link.control |= TRB_CYCLE;
		}
312
		/* All endpoint rings have link TRBs */
A
Andiry Xu 已提交
313
		xhci_link_segments(xhci, seg, seg->next, type);
314 315
		seg = seg->next;
	} while (seg != ring->first_seg);
A
Andiry Xu 已提交
316
	ring->type = type;
317
	xhci_initialize_ring_info(ring, cycle_state);
318 319 320 321 322 323
	/* td list should be empty since all URBs have been cancelled,
	 * but just in case...
	 */
	INIT_LIST_HEAD(&ring->td_list);
}

A
Andiry Xu 已提交
324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
/*
 * Expand an existing ring.
 * Look for a cached ring or allocate a new ring which has same segment numbers
 * and link the two rings.
 */
int xhci_ring_expansion(struct xhci_hcd *xhci, struct xhci_ring *ring,
				unsigned int num_trbs, gfp_t flags)
{
	struct xhci_segment	*first;
	struct xhci_segment	*last;
	unsigned int		num_segs;
	unsigned int		num_segs_needed;
	int			ret;

	num_segs_needed = (num_trbs + (TRBS_PER_SEGMENT - 1) - 1) /
				(TRBS_PER_SEGMENT - 1);

	/* Allocate number of segments we needed, or double the ring size */
	num_segs = ring->num_segs > num_segs_needed ?
			ring->num_segs : num_segs_needed;

	ret = xhci_alloc_segments_for_ring(xhci, &first, &last,
			num_segs, ring->cycle_state, ring->type, flags);
	if (ret)
		return -ENOMEM;

	xhci_link_rings(xhci, ring, first, last, num_segs);
	xhci_dbg(xhci, "ring expansion succeed, now has %d segments\n",
			ring->num_segs);

	return 0;
}

357 358
#define CTX_SIZE(_hcc) (HCC_64BYTE_CONTEXT(_hcc) ? 64 : 32)

359
static struct xhci_container_ctx *xhci_alloc_container_ctx(struct xhci_hcd *xhci,
360 361
						    int type, gfp_t flags)
{
362 363 364 365 366 367
	struct xhci_container_ctx *ctx;

	if ((type != XHCI_CTX_TYPE_DEVICE) && (type != XHCI_CTX_TYPE_INPUT))
		return NULL;

	ctx = kzalloc(sizeof(*ctx), flags);
368 369 370 371 372 373 374 375 376
	if (!ctx)
		return NULL;

	ctx->type = type;
	ctx->size = HCC_64BYTE_CONTEXT(xhci->hcc_params) ? 2048 : 1024;
	if (type == XHCI_CTX_TYPE_INPUT)
		ctx->size += CTX_SIZE(xhci->hcc_params);

	ctx->bytes = dma_pool_alloc(xhci->device_pool, flags, &ctx->dma);
377 378 379 380
	if (!ctx->bytes) {
		kfree(ctx);
		return NULL;
	}
381 382 383 384
	memset(ctx->bytes, 0, ctx->size);
	return ctx;
}

385
static void xhci_free_container_ctx(struct xhci_hcd *xhci,
386 387
			     struct xhci_container_ctx *ctx)
{
388 389
	if (!ctx)
		return;
390 391 392 393 394 395 396
	dma_pool_free(xhci->device_pool, ctx->bytes, ctx->dma);
	kfree(ctx);
}

struct xhci_input_control_ctx *xhci_get_input_control_ctx(struct xhci_hcd *xhci,
					      struct xhci_container_ctx *ctx)
{
397 398 399
	if (ctx->type != XHCI_CTX_TYPE_INPUT)
		return NULL;

400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
	return (struct xhci_input_control_ctx *)ctx->bytes;
}

struct xhci_slot_ctx *xhci_get_slot_ctx(struct xhci_hcd *xhci,
					struct xhci_container_ctx *ctx)
{
	if (ctx->type == XHCI_CTX_TYPE_DEVICE)
		return (struct xhci_slot_ctx *)ctx->bytes;

	return (struct xhci_slot_ctx *)
		(ctx->bytes + CTX_SIZE(xhci->hcc_params));
}

struct xhci_ep_ctx *xhci_get_ep_ctx(struct xhci_hcd *xhci,
				    struct xhci_container_ctx *ctx,
				    unsigned int ep_index)
{
	/* increment ep index by offset of start of ep ctx array */
	ep_index++;
	if (ctx->type == XHCI_CTX_TYPE_INPUT)
		ep_index++;

	return (struct xhci_ep_ctx *)
		(ctx->bytes + (ep_index * CTX_SIZE(xhci->hcc_params)));
}

426 427 428

/***************** Streams structures manipulation *************************/

429
static void xhci_free_stream_ctx(struct xhci_hcd *xhci,
430 431 432 433 434 435
		unsigned int num_stream_ctxs,
		struct xhci_stream_ctx *stream_ctx, dma_addr_t dma)
{
	struct pci_dev *pdev = to_pci_dev(xhci_to_hcd(xhci)->self.controller);

	if (num_stream_ctxs > MEDIUM_STREAM_ARRAY_SIZE)
436
		dma_free_coherent(&pdev->dev,
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
				sizeof(struct xhci_stream_ctx)*num_stream_ctxs,
				stream_ctx, dma);
	else if (num_stream_ctxs <= SMALL_STREAM_ARRAY_SIZE)
		return dma_pool_free(xhci->small_streams_pool,
				stream_ctx, dma);
	else
		return dma_pool_free(xhci->medium_streams_pool,
				stream_ctx, dma);
}

/*
 * The stream context array for each endpoint with bulk streams enabled can
 * vary in size, based on:
 *  - how many streams the endpoint supports,
 *  - the maximum primary stream array size the host controller supports,
 *  - and how many streams the device driver asks for.
 *
 * The stream context array must be a power of 2, and can be as small as
 * 64 bytes or as large as 1MB.
 */
457
static struct xhci_stream_ctx *xhci_alloc_stream_ctx(struct xhci_hcd *xhci,
458 459 460 461 462 463
		unsigned int num_stream_ctxs, dma_addr_t *dma,
		gfp_t mem_flags)
{
	struct pci_dev *pdev = to_pci_dev(xhci_to_hcd(xhci)->self.controller);

	if (num_stream_ctxs > MEDIUM_STREAM_ARRAY_SIZE)
464
		return dma_alloc_coherent(&pdev->dev,
465
				sizeof(struct xhci_stream_ctx)*num_stream_ctxs,
466
				dma, mem_flags);
467 468 469 470 471 472 473 474
	else if (num_stream_ctxs <= SMALL_STREAM_ARRAY_SIZE)
		return dma_pool_alloc(xhci->small_streams_pool,
				mem_flags, dma);
	else
		return dma_pool_alloc(xhci->medium_streams_pool,
				mem_flags, dma);
}

475 476 477 478 479 480
struct xhci_ring *xhci_dma_to_transfer_ring(
		struct xhci_virt_ep *ep,
		u64 address)
{
	if (ep->ep_state & EP_HAS_STREAMS)
		return radix_tree_lookup(&ep->stream_info->trb_address_map,
481
				address >> TRB_SEGMENT_SHIFT);
482 483 484 485
	return ep->ring;
}

/* Only use this when you know stream_info is valid */
486
#ifdef CONFIG_USB_XHCI_HCD_DEBUGGING
487
static struct xhci_ring *dma_to_stream_ring(
488 489 490 491
		struct xhci_stream_info *stream_info,
		u64 address)
{
	return radix_tree_lookup(&stream_info->trb_address_map,
492
			address >> TRB_SEGMENT_SHIFT);
493 494 495
}
#endif	/* CONFIG_USB_XHCI_HCD_DEBUGGING */

496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
struct xhci_ring *xhci_stream_id_to_ring(
		struct xhci_virt_device *dev,
		unsigned int ep_index,
		unsigned int stream_id)
{
	struct xhci_virt_ep *ep = &dev->eps[ep_index];

	if (stream_id == 0)
		return ep->ring;
	if (!ep->stream_info)
		return NULL;

	if (stream_id > ep->stream_info->num_streams)
		return NULL;
	return ep->stream_info->stream_rings[stream_id];
}

513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
#ifdef CONFIG_USB_XHCI_HCD_DEBUGGING
static int xhci_test_radix_tree(struct xhci_hcd *xhci,
		unsigned int num_streams,
		struct xhci_stream_info *stream_info)
{
	u32 cur_stream;
	struct xhci_ring *cur_ring;
	u64 addr;

	for (cur_stream = 1; cur_stream < num_streams; cur_stream++) {
		struct xhci_ring *mapped_ring;
		int trb_size = sizeof(union xhci_trb);

		cur_ring = stream_info->stream_rings[cur_stream];
		for (addr = cur_ring->first_seg->dma;
528
				addr < cur_ring->first_seg->dma + TRB_SEGMENT_SIZE;
529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
				addr += trb_size) {
			mapped_ring = dma_to_stream_ring(stream_info, addr);
			if (cur_ring != mapped_ring) {
				xhci_warn(xhci, "WARN: DMA address 0x%08llx "
						"didn't map to stream ID %u; "
						"mapped to ring %p\n",
						(unsigned long long) addr,
						cur_stream,
						mapped_ring);
				return -EINVAL;
			}
		}
		/* One TRB after the end of the ring segment shouldn't return a
		 * pointer to the current ring (although it may be a part of a
		 * different ring).
		 */
		mapped_ring = dma_to_stream_ring(stream_info, addr);
		if (mapped_ring != cur_ring) {
			/* One TRB before should also fail */
			addr = cur_ring->first_seg->dma - trb_size;
			mapped_ring = dma_to_stream_ring(stream_info, addr);
		}
		if (mapped_ring == cur_ring) {
			xhci_warn(xhci, "WARN: Bad DMA address 0x%08llx "
					"mapped to valid stream ID %u; "
					"mapped ring = %p\n",
					(unsigned long long) addr,
					cur_stream,
					mapped_ring);
			return -EINVAL;
		}
	}
	return 0;
}
#endif	/* CONFIG_USB_XHCI_HCD_DEBUGGING */

/*
 * Change an endpoint's internal structure so it supports stream IDs.  The
 * number of requested streams includes stream 0, which cannot be used by device
 * drivers.
 *
 * The number of stream contexts in the stream context array may be bigger than
 * the number of streams the driver wants to use.  This is because the number of
 * stream context array entries must be a power of two.
 *
 * We need a radix tree for mapping physical addresses of TRBs to which stream
 * ID they belong to.  We need to do this because the host controller won't tell
 * us which stream ring the TRB came from.  We could store the stream ID in an
 * event data TRB, but that doesn't help us for the cancellation case, since the
 * endpoint may stop before it reaches that event data TRB.
 *
 * The radix tree maps the upper portion of the TRB DMA address to a ring
 * segment that has the same upper portion of DMA addresses.  For example, say I
 * have segments of size 1KB, that are always 64-byte aligned.  A segment may
 * start at 0x10c91000 and end at 0x10c913f0.  If I use the upper 10 bits, the
 * key to the stream ID is 0x43244.  I can use the DMA address of the TRB to
 * pass the radix tree a key to get the right stream ID:
 *
 * 	0x10c90fff >> 10 = 0x43243
 * 	0x10c912c0 >> 10 = 0x43244
 * 	0x10c91400 >> 10 = 0x43245
 *
 * Obviously, only those TRBs with DMA addresses that are within the segment
 * will make the radix tree return the stream ID for that ring.
 *
 * Caveats for the radix tree:
 *
 * The radix tree uses an unsigned long as a key pair.  On 32-bit systems, an
 * unsigned long will be 32-bits; on a 64-bit system an unsigned long will be
 * 64-bits.  Since we only request 32-bit DMA addresses, we can use that as the
 * key on 32-bit or 64-bit systems (it would also be fine if we asked for 64-bit
 * PCI DMA addresses on a 64-bit system).  There might be a problem on 32-bit
 * extended systems (where the DMA address can be bigger than 32-bits),
 * if we allow the PCI dma mask to be bigger than 32-bits.  So don't do that.
 */
struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
		unsigned int num_stream_ctxs,
		unsigned int num_streams, gfp_t mem_flags)
{
	struct xhci_stream_info *stream_info;
	u32 cur_stream;
	struct xhci_ring *cur_ring;
	unsigned long key;
	u64 addr;
	int ret;

	xhci_dbg(xhci, "Allocating %u streams and %u "
			"stream context array entries.\n",
			num_streams, num_stream_ctxs);
	if (xhci->cmd_ring_reserved_trbs == MAX_RSVD_CMD_TRBS) {
		xhci_dbg(xhci, "Command ring has no reserved TRBs available\n");
		return NULL;
	}
	xhci->cmd_ring_reserved_trbs++;

	stream_info = kzalloc(sizeof(struct xhci_stream_info), mem_flags);
	if (!stream_info)
		goto cleanup_trbs;

	stream_info->num_streams = num_streams;
	stream_info->num_stream_ctxs = num_stream_ctxs;

	/* Initialize the array of virtual pointers to stream rings. */
	stream_info->stream_rings = kzalloc(
			sizeof(struct xhci_ring *)*num_streams,
			mem_flags);
	if (!stream_info->stream_rings)
		goto cleanup_info;

	/* Initialize the array of DMA addresses for stream rings for the HW. */
	stream_info->stream_ctx_array = xhci_alloc_stream_ctx(xhci,
			num_stream_ctxs, &stream_info->ctx_array_dma,
			mem_flags);
	if (!stream_info->stream_ctx_array)
		goto cleanup_ctx;
	memset(stream_info->stream_ctx_array, 0,
			sizeof(struct xhci_stream_ctx)*num_stream_ctxs);

	/* Allocate everything needed to free the stream rings later */
	stream_info->free_streams_command =
		xhci_alloc_command(xhci, true, true, mem_flags);
	if (!stream_info->free_streams_command)
		goto cleanup_ctx;

	INIT_RADIX_TREE(&stream_info->trb_address_map, GFP_ATOMIC);

	/* Allocate rings for all the streams that the driver will use,
	 * and add their segment DMA addresses to the radix tree.
	 * Stream 0 is reserved.
	 */
	for (cur_stream = 1; cur_stream < num_streams; cur_stream++) {
		stream_info->stream_rings[cur_stream] =
661
			xhci_ring_alloc(xhci, 2, 1, TYPE_STREAM, mem_flags);
662 663 664
		cur_ring = stream_info->stream_rings[cur_stream];
		if (!cur_ring)
			goto cleanup_rings;
665
		cur_ring->stream_id = cur_stream;
666 667 668 669
		/* Set deq ptr, cycle bit, and stream context type */
		addr = cur_ring->first_seg->dma |
			SCT_FOR_CTX(SCT_PRI_TR) |
			cur_ring->cycle_state;
670 671
		stream_info->stream_ctx_array[cur_stream].stream_ring =
			cpu_to_le64(addr);
672 673 674 675
		xhci_dbg(xhci, "Setting stream %d ring ptr to 0x%08llx\n",
				cur_stream, (unsigned long long) addr);

		key = (unsigned long)
676
			(cur_ring->first_seg->dma >> TRB_SEGMENT_SHIFT);
677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706
		ret = radix_tree_insert(&stream_info->trb_address_map,
				key, cur_ring);
		if (ret) {
			xhci_ring_free(xhci, cur_ring);
			stream_info->stream_rings[cur_stream] = NULL;
			goto cleanup_rings;
		}
	}
	/* Leave the other unused stream ring pointers in the stream context
	 * array initialized to zero.  This will cause the xHC to give us an
	 * error if the device asks for a stream ID we don't have setup (if it
	 * was any other way, the host controller would assume the ring is
	 * "empty" and wait forever for data to be queued to that stream ID).
	 */
#if XHCI_DEBUG
	/* Do a little test on the radix tree to make sure it returns the
	 * correct values.
	 */
	if (xhci_test_radix_tree(xhci, num_streams, stream_info))
		goto cleanup_rings;
#endif

	return stream_info;

cleanup_rings:
	for (cur_stream = 1; cur_stream < num_streams; cur_stream++) {
		cur_ring = stream_info->stream_rings[cur_stream];
		if (cur_ring) {
			addr = cur_ring->first_seg->dma;
			radix_tree_delete(&stream_info->trb_address_map,
707
					addr >> TRB_SEGMENT_SHIFT);
708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736
			xhci_ring_free(xhci, cur_ring);
			stream_info->stream_rings[cur_stream] = NULL;
		}
	}
	xhci_free_command(xhci, stream_info->free_streams_command);
cleanup_ctx:
	kfree(stream_info->stream_rings);
cleanup_info:
	kfree(stream_info);
cleanup_trbs:
	xhci->cmd_ring_reserved_trbs--;
	return NULL;
}
/*
 * Sets the MaxPStreams field and the Linear Stream Array field.
 * Sets the dequeue pointer to the stream context array.
 */
void xhci_setup_streams_ep_input_ctx(struct xhci_hcd *xhci,
		struct xhci_ep_ctx *ep_ctx,
		struct xhci_stream_info *stream_info)
{
	u32 max_primary_streams;
	/* MaxPStreams is the number of stream context array entries, not the
	 * number we're actually using.  Must be in 2^(MaxPstreams + 1) format.
	 * fls(0) = 0, fls(0x1) = 1, fls(0x10) = 2, fls(0x100) = 3, etc.
	 */
	max_primary_streams = fls(stream_info->num_stream_ctxs) - 2;
	xhci_dbg(xhci, "Setting number of stream ctx array entries to %u\n",
			1 << (max_primary_streams + 1));
M
Matt Evans 已提交
737 738 739 740
	ep_ctx->ep_info &= cpu_to_le32(~EP_MAXPSTREAMS_MASK);
	ep_ctx->ep_info |= cpu_to_le32(EP_MAXPSTREAMS(max_primary_streams)
				       | EP_HAS_LSA);
	ep_ctx->deq  = cpu_to_le64(stream_info->ctx_array_dma);
741 742 743 744 745 746 747 748 749 750 751 752
}

/*
 * Sets the MaxPStreams field and the Linear Stream Array field to 0.
 * Reinstalls the "normal" endpoint ring (at its previous dequeue mark,
 * not at the beginning of the ring).
 */
void xhci_setup_no_streams_ep_input_ctx(struct xhci_hcd *xhci,
		struct xhci_ep_ctx *ep_ctx,
		struct xhci_virt_ep *ep)
{
	dma_addr_t addr;
M
Matt Evans 已提交
753
	ep_ctx->ep_info &= cpu_to_le32(~(EP_MAXPSTREAMS_MASK | EP_HAS_LSA));
754
	addr = xhci_trb_virt_to_dma(ep->ring->deq_seg, ep->ring->dequeue);
M
Matt Evans 已提交
755
	ep_ctx->deq  = cpu_to_le64(addr | ep->ring->cycle_state);
756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777
}

/* Frees all stream contexts associated with the endpoint,
 *
 * Caller should fix the endpoint context streams fields.
 */
void xhci_free_stream_info(struct xhci_hcd *xhci,
		struct xhci_stream_info *stream_info)
{
	int cur_stream;
	struct xhci_ring *cur_ring;
	dma_addr_t addr;

	if (!stream_info)
		return;

	for (cur_stream = 1; cur_stream < stream_info->num_streams;
			cur_stream++) {
		cur_ring = stream_info->stream_rings[cur_stream];
		if (cur_ring) {
			addr = cur_ring->first_seg->dma;
			radix_tree_delete(&stream_info->trb_address_map,
778
					addr >> TRB_SEGMENT_SHIFT);
779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
			xhci_ring_free(xhci, cur_ring);
			stream_info->stream_rings[cur_stream] = NULL;
		}
	}
	xhci_free_command(xhci, stream_info->free_streams_command);
	xhci->cmd_ring_reserved_trbs--;
	if (stream_info->stream_ctx_array)
		xhci_free_stream_ctx(xhci,
				stream_info->num_stream_ctxs,
				stream_info->stream_ctx_array,
				stream_info->ctx_array_dma);

	if (stream_info)
		kfree(stream_info->stream_rings);
	kfree(stream_info);
}


/***************** Device context manipulation *************************/

799 800 801 802 803 804 805 806 807
static void xhci_init_endpoint_timer(struct xhci_hcd *xhci,
		struct xhci_virt_ep *ep)
{
	init_timer(&ep->stop_cmd_timer);
	ep->stop_cmd_timer.data = (unsigned long) ep;
	ep->stop_cmd_timer.function = xhci_stop_endpoint_command_watchdog;
	ep->xhci = xhci;
}

808 809 810 811 812
static void xhci_free_tt_info(struct xhci_hcd *xhci,
		struct xhci_virt_device *virt_dev,
		int slot_id)
{
	struct list_head *tt_list_head;
813 814
	struct xhci_tt_bw_info *tt_info, *next;
	bool slot_found = false;
815 816 817 818 819 820 821 822 823 824 825

	/* If the device never made it past the Set Address stage,
	 * it may not have the real_port set correctly.
	 */
	if (virt_dev->real_port == 0 ||
			virt_dev->real_port > HCS_MAX_PORTS(xhci->hcs_params1)) {
		xhci_dbg(xhci, "Bad real port.\n");
		return;
	}

	tt_list_head = &(xhci->rh_bw[virt_dev->real_port - 1].tts);
826 827 828 829 830 831 832
	list_for_each_entry_safe(tt_info, next, tt_list_head, tt_list) {
		/* Multi-TT hubs will have more than one entry */
		if (tt_info->slot_id == slot_id) {
			slot_found = true;
			list_del(&tt_info->tt_list);
			kfree(tt_info);
		} else if (slot_found) {
833
			break;
834
		}
835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880
	}
}

int xhci_alloc_tt_info(struct xhci_hcd *xhci,
		struct xhci_virt_device *virt_dev,
		struct usb_device *hdev,
		struct usb_tt *tt, gfp_t mem_flags)
{
	struct xhci_tt_bw_info		*tt_info;
	unsigned int			num_ports;
	int				i, j;

	if (!tt->multi)
		num_ports = 1;
	else
		num_ports = hdev->maxchild;

	for (i = 0; i < num_ports; i++, tt_info++) {
		struct xhci_interval_bw_table *bw_table;

		tt_info = kzalloc(sizeof(*tt_info), mem_flags);
		if (!tt_info)
			goto free_tts;
		INIT_LIST_HEAD(&tt_info->tt_list);
		list_add(&tt_info->tt_list,
				&xhci->rh_bw[virt_dev->real_port - 1].tts);
		tt_info->slot_id = virt_dev->udev->slot_id;
		if (tt->multi)
			tt_info->ttport = i+1;
		bw_table = &tt_info->bw_table;
		for (j = 0; j < XHCI_MAX_INTERVAL; j++)
			INIT_LIST_HEAD(&bw_table->interval_bw[j].endpoints);
	}
	return 0;

free_tts:
	xhci_free_tt_info(xhci, virt_dev, virt_dev->udev->slot_id);
	return -ENOMEM;
}


/* All the xhci_tds in the ring's TD list should be freed at this point.
 * Should be called with xhci->lock held if there is any chance the TT lists
 * will be manipulated by the configure endpoint, allocate device, or update
 * hub functions while this function is removing the TT entries from the list.
 */
881 882 883 884
void xhci_free_virt_device(struct xhci_hcd *xhci, int slot_id)
{
	struct xhci_virt_device *dev;
	int i;
885
	int old_active_eps = 0;
886 887 888 889 890 891

	/* Slot ID 0 is reserved */
	if (slot_id == 0 || !xhci->devs[slot_id])
		return;

	dev = xhci->devs[slot_id];
892
	xhci->dcbaa->dev_context_ptrs[slot_id] = 0;
893 894 895
	if (!dev)
		return;

896 897 898
	if (dev->tt_info)
		old_active_eps = dev->tt_info->active_eps;

899
	for (i = 0; i < 31; ++i) {
900 901
		if (dev->eps[i].ring)
			xhci_ring_free(xhci, dev->eps[i].ring);
902 903 904
		if (dev->eps[i].stream_info)
			xhci_free_stream_info(xhci,
					dev->eps[i].stream_info);
905 906 907 908 909 910 911 912 913
		/* Endpoints on the TT/root port lists should have been removed
		 * when usb_disable_device() was called for the device.
		 * We can't drop them anyway, because the udev might have gone
		 * away by this point, and we can't tell what speed it was.
		 */
		if (!list_empty(&dev->eps[i].bw_endpoint_list))
			xhci_warn(xhci, "Slot %u endpoint %u "
					"not removed from BW list!\n",
					slot_id, i);
914
	}
915 916
	/* If this is a hub, free the TT(s) from the TT list */
	xhci_free_tt_info(xhci, dev, slot_id);
917 918
	/* If necessary, update the number of active TTs on this root port */
	xhci_update_tt_active_eps(xhci, dev, old_active_eps);
919

920 921 922 923 924 925
	if (dev->ring_cache) {
		for (i = 0; i < dev->num_rings_cached; i++)
			xhci_ring_free(xhci, dev->ring_cache[i]);
		kfree(dev->ring_cache);
	}

926
	if (dev->in_ctx)
927
		xhci_free_container_ctx(xhci, dev->in_ctx);
928
	if (dev->out_ctx)
929 930
		xhci_free_container_ctx(xhci, dev->out_ctx);

931
	kfree(xhci->devs[slot_id]);
932
	xhci->devs[slot_id] = NULL;
933 934 935 936 937 938
}

int xhci_alloc_virt_device(struct xhci_hcd *xhci, int slot_id,
		struct usb_device *udev, gfp_t flags)
{
	struct xhci_virt_device *dev;
939
	int i;
940 941 942 943 944 945 946 947 948 949 950 951

	/* Slot ID 0 is reserved */
	if (slot_id == 0 || xhci->devs[slot_id]) {
		xhci_warn(xhci, "Bad Slot ID %d\n", slot_id);
		return 0;
	}

	xhci->devs[slot_id] = kzalloc(sizeof(*xhci->devs[slot_id]), flags);
	if (!xhci->devs[slot_id])
		return 0;
	dev = xhci->devs[slot_id];

952 953
	/* Allocate the (output) device context that will be used in the HC. */
	dev->out_ctx = xhci_alloc_container_ctx(xhci, XHCI_CTX_TYPE_DEVICE, flags);
954 955
	if (!dev->out_ctx)
		goto fail;
956

957
	xhci_dbg(xhci, "Slot %d output ctx = 0x%llx (dma)\n", slot_id,
958
			(unsigned long long)dev->out_ctx->dma);
959 960

	/* Allocate the (input) device context for address device command */
961
	dev->in_ctx = xhci_alloc_container_ctx(xhci, XHCI_CTX_TYPE_INPUT, flags);
962 963
	if (!dev->in_ctx)
		goto fail;
964

965
	xhci_dbg(xhci, "Slot %d input ctx = 0x%llx (dma)\n", slot_id,
966
			(unsigned long long)dev->in_ctx->dma);
967

968 969 970
	/* Initialize the cancellation list and watchdog timers for each ep */
	for (i = 0; i < 31; i++) {
		xhci_init_endpoint_timer(xhci, &dev->eps[i]);
971
		INIT_LIST_HEAD(&dev->eps[i].cancelled_td_list);
972
		INIT_LIST_HEAD(&dev->eps[i].bw_endpoint_list);
973
	}
974

975
	/* Allocate endpoint 0 ring */
976
	dev->eps[0].ring = xhci_ring_alloc(xhci, 2, 1, TYPE_CTRL, flags);
977
	if (!dev->eps[0].ring)
978 979
		goto fail;

980 981 982 983 984 985 986 987
	/* Allocate pointers to the ring cache */
	dev->ring_cache = kzalloc(
			sizeof(struct xhci_ring *)*XHCI_MAX_RINGS_CACHED,
			flags);
	if (!dev->ring_cache)
		goto fail;
	dev->num_rings_cached = 0;

988
	init_completion(&dev->cmd_completion);
989
	INIT_LIST_HEAD(&dev->cmd_list);
990
	dev->udev = udev;
991

992
	/* Point to output device context in dcbaa. */
M
Matt Evans 已提交
993
	xhci->dcbaa->dev_context_ptrs[slot_id] = cpu_to_le64(dev->out_ctx->dma);
994
	xhci_dbg(xhci, "Set slot id %d dcbaa entry %p to 0x%llx\n",
M
Matt Evans 已提交
995 996
		 slot_id,
		 &xhci->dcbaa->dev_context_ptrs[slot_id],
997
		 le64_to_cpu(xhci->dcbaa->dev_context_ptrs[slot_id]));
998 999 1000 1001 1002 1003 1004

	return 1;
fail:
	xhci_free_virt_device(xhci, slot_id);
	return 0;
}

1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
void xhci_copy_ep0_dequeue_into_input_ctx(struct xhci_hcd *xhci,
		struct usb_device *udev)
{
	struct xhci_virt_device *virt_dev;
	struct xhci_ep_ctx	*ep0_ctx;
	struct xhci_ring	*ep_ring;

	virt_dev = xhci->devs[udev->slot_id];
	ep0_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, 0);
	ep_ring = virt_dev->eps[0].ring;
	/*
	 * FIXME we don't keep track of the dequeue pointer very well after a
	 * Set TR dequeue pointer, so we're setting the dequeue pointer of the
	 * host to our enqueue pointer.  This should only be called after a
	 * configured device has reset, so all control transfers should have
	 * been completed or cancelled before the reset.
	 */
M
Matt Evans 已提交
1022 1023 1024
	ep0_ctx->deq = cpu_to_le64(xhci_trb_virt_to_dma(ep_ring->enq_seg,
							ep_ring->enqueue)
				   | ep_ring->cycle_state);
1025 1026
}

1027 1028 1029 1030 1031 1032 1033 1034 1035
/*
 * The xHCI roothub may have ports of differing speeds in any order in the port
 * status registers.  xhci->port_array provides an array of the port speed for
 * each offset into the port status registers.
 *
 * The xHCI hardware wants to know the roothub port number that the USB device
 * is attached to (or the roothub port its ancestor hub is attached to).  All we
 * know is the index of that port under either the USB 2.0 or the USB 3.0
 * roothub, but that doesn't give us the real index into the HW port status
1036
 * registers. Call xhci_find_raw_port_number() to get real index.
1037 1038 1039 1040 1041
 */
static u32 xhci_find_real_port_number(struct xhci_hcd *xhci,
		struct usb_device *udev)
{
	struct usb_device *top_dev;
1042 1043 1044 1045 1046 1047
	struct usb_hcd *hcd;

	if (udev->speed == USB_SPEED_SUPER)
		hcd = xhci->shared_hcd;
	else
		hcd = xhci->main_hcd;
1048 1049 1050 1051 1052

	for (top_dev = udev; top_dev->parent && top_dev->parent->parent;
			top_dev = top_dev->parent)
		/* Found device below root hub */;

1053
	return	xhci_find_raw_port_number(hcd, top_dev->portnum);
1054 1055
}

1056 1057 1058 1059 1060
/* Setup an xHCI virtual device for a Set Address command */
int xhci_setup_addressable_virt_dev(struct xhci_hcd *xhci, struct usb_device *udev)
{
	struct xhci_virt_device *dev;
	struct xhci_ep_ctx	*ep0_ctx;
1061
	struct xhci_slot_ctx    *slot_ctx;
1062
	u32			port_num;
1063
	u32			max_packets;
1064
	struct usb_device *top_dev;
1065 1066 1067 1068 1069 1070 1071 1072

	dev = xhci->devs[udev->slot_id];
	/* Slot ID 0 is reserved */
	if (udev->slot_id == 0 || !dev) {
		xhci_warn(xhci, "Slot ID %d is not assigned to this device\n",
				udev->slot_id);
		return -EINVAL;
	}
1073 1074
	ep0_ctx = xhci_get_ep_ctx(xhci, dev->in_ctx, 0);
	slot_ctx = xhci_get_slot_ctx(xhci, dev->in_ctx);
1075 1076

	/* 3) Only the control endpoint is valid - one endpoint context */
1077
	slot_ctx->dev_info |= cpu_to_le32(LAST_CTX(1) | udev->route);
1078 1079
	switch (udev->speed) {
	case USB_SPEED_SUPER:
1080
		slot_ctx->dev_info |= cpu_to_le32(SLOT_SPEED_SS);
1081
		max_packets = MAX_PACKET(512);
1082 1083
		break;
	case USB_SPEED_HIGH:
1084
		slot_ctx->dev_info |= cpu_to_le32(SLOT_SPEED_HS);
1085
		max_packets = MAX_PACKET(64);
1086
		break;
1087
	/* USB core guesses at a 64-byte max packet first for FS devices */
1088
	case USB_SPEED_FULL:
1089
		slot_ctx->dev_info |= cpu_to_le32(SLOT_SPEED_FS);
1090
		max_packets = MAX_PACKET(64);
1091 1092
		break;
	case USB_SPEED_LOW:
1093
		slot_ctx->dev_info |= cpu_to_le32(SLOT_SPEED_LS);
1094
		max_packets = MAX_PACKET(8);
1095
		break;
1096
	case USB_SPEED_WIRELESS:
1097 1098 1099 1100 1101
		xhci_dbg(xhci, "FIXME xHCI doesn't support wireless speeds\n");
		return -EINVAL;
		break;
	default:
		/* Speed was set earlier, this shouldn't happen. */
1102
		return -EINVAL;
1103 1104
	}
	/* Find the root hub port this device is under */
1105 1106 1107
	port_num = xhci_find_real_port_number(xhci, udev);
	if (!port_num)
		return -EINVAL;
1108
	slot_ctx->dev_info2 |= cpu_to_le32(ROOT_HUB_PORT(port_num));
1109
	/* Set the port number in the virtual_device to the faked port number */
1110 1111 1112
	for (top_dev = udev; top_dev->parent && top_dev->parent->parent;
			top_dev = top_dev->parent)
		/* Found device below root hub */;
1113
	dev->fake_port = top_dev->portnum;
1114
	dev->real_port = port_num;
1115
	xhci_dbg(xhci, "Set root hub portnum to %d\n", port_num);
1116
	xhci_dbg(xhci, "Set fake root hub portnum to %d\n", dev->fake_port);
1117

1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
	/* Find the right bandwidth table that this device will be a part of.
	 * If this is a full speed device attached directly to a root port (or a
	 * decendent of one), it counts as a primary bandwidth domain, not a
	 * secondary bandwidth domain under a TT.  An xhci_tt_info structure
	 * will never be created for the HS root hub.
	 */
	if (!udev->tt || !udev->tt->hub->parent) {
		dev->bw_table = &xhci->rh_bw[port_num - 1].bw_table;
	} else {
		struct xhci_root_port_bw_info *rh_bw;
		struct xhci_tt_bw_info *tt_bw;

		rh_bw = &xhci->rh_bw[port_num - 1];
		/* Find the right TT. */
		list_for_each_entry(tt_bw, &rh_bw->tts, tt_list) {
			if (tt_bw->slot_id != udev->tt->hub->slot_id)
				continue;

			if (!dev->udev->tt->multi ||
					(udev->tt->multi &&
					 tt_bw->ttport == dev->udev->ttport)) {
				dev->bw_table = &tt_bw->bw_table;
				dev->tt_info = tt_bw;
				break;
			}
		}
		if (!dev->tt_info)
			xhci_warn(xhci, "WARN: Didn't find a matching TT\n");
	}

S
Sarah Sharp 已提交
1148 1149
	/* Is this a LS/FS device under an external HS hub? */
	if (udev->tt && udev->tt->hub->parent) {
M
Matt Evans 已提交
1150 1151
		slot_ctx->tt_info = cpu_to_le32(udev->tt->hub->slot_id |
						(udev->ttport << 8));
1152
		if (udev->tt->multi)
M
Matt Evans 已提交
1153
			slot_ctx->dev_info |= cpu_to_le32(DEV_MTT);
1154
	}
1155
	xhci_dbg(xhci, "udev->tt = %p\n", udev->tt);
1156 1157 1158 1159
	xhci_dbg(xhci, "udev->ttport = 0x%x\n", udev->ttport);

	/* Step 4 - ring already allocated */
	/* Step 5 */
M
Matt Evans 已提交
1160
	ep0_ctx->ep_info2 = cpu_to_le32(EP_TYPE(CTRL_EP));
1161

1162
	/* EP 0 can handle "burst" sizes of 1, so Max Burst Size field is 0 */
1163 1164
	ep0_ctx->ep_info2 |= cpu_to_le32(MAX_BURST(0) | ERROR_COUNT(3) |
					 max_packets);
1165

M
Matt Evans 已提交
1166 1167
	ep0_ctx->deq = cpu_to_le64(dev->eps[0].ring->first_seg->dma |
				   dev->eps[0].ring->cycle_state);
1168 1169 1170 1171 1172 1173

	/* Steps 7 and 8 were done in xhci_alloc_virt_device() */

	return 0;
}

1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186
/*
 * Convert interval expressed as 2^(bInterval - 1) == interval into
 * straight exponent value 2^n == interval.
 *
 */
static unsigned int xhci_parse_exponent_interval(struct usb_device *udev,
		struct usb_host_endpoint *ep)
{
	unsigned int interval;

	interval = clamp_val(ep->desc.bInterval, 1, 16) - 1;
	if (interval != ep->desc.bInterval - 1)
		dev_warn(&udev->dev,
1187
			 "ep %#x - rounding interval to %d %sframes\n",
1188
			 ep->desc.bEndpointAddress,
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
			 1 << interval,
			 udev->speed == USB_SPEED_FULL ? "" : "micro");

	if (udev->speed == USB_SPEED_FULL) {
		/*
		 * Full speed isoc endpoints specify interval in frames,
		 * not microframes. We are using microframes everywhere,
		 * so adjust accordingly.
		 */
		interval += 3;	/* 1 frame = 2^3 uframes */
	}
1200 1201 1202 1203 1204

	return interval;
}

/*
1205
 * Convert bInterval expressed in microframes (in 1-255 range) to exponent of
1206 1207
 * microframes, rounded down to nearest power of 2.
 */
1208 1209 1210
static unsigned int xhci_microframes_to_exponent(struct usb_device *udev,
		struct usb_host_endpoint *ep, unsigned int desc_interval,
		unsigned int min_exponent, unsigned int max_exponent)
1211 1212 1213
{
	unsigned int interval;

1214 1215 1216
	interval = fls(desc_interval) - 1;
	interval = clamp_val(interval, min_exponent, max_exponent);
	if ((1 << interval) != desc_interval)
1217 1218 1219 1220
		dev_warn(&udev->dev,
			 "ep %#x - rounding interval to %d microframes, ep desc says %d microframes\n",
			 ep->desc.bEndpointAddress,
			 1 << interval,
1221
			 desc_interval);
1222 1223 1224 1225

	return interval;
}

1226 1227 1228
static unsigned int xhci_parse_microframe_interval(struct usb_device *udev,
		struct usb_host_endpoint *ep)
{
1229 1230
	if (ep->desc.bInterval == 0)
		return 0;
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242
	return xhci_microframes_to_exponent(udev, ep,
			ep->desc.bInterval, 0, 15);
}


static unsigned int xhci_parse_frame_interval(struct usb_device *udev,
		struct usb_host_endpoint *ep)
{
	return xhci_microframes_to_exponent(udev, ep,
			ep->desc.bInterval * 8, 3, 10);
}

1243 1244 1245 1246 1247 1248 1249 1250
/* Return the polling or NAK interval.
 *
 * The polling interval is expressed in "microframes".  If xHCI's Interval field
 * is set to N, it will service the endpoint every 2^(Interval)*125us.
 *
 * The NAK interval is one NAK per 1 to 255 microframes, or no NAKs if interval
 * is set to 0.
 */
1251
static unsigned int xhci_get_endpoint_interval(struct usb_device *udev,
1252 1253 1254 1255 1256 1257 1258 1259
		struct usb_host_endpoint *ep)
{
	unsigned int interval = 0;

	switch (udev->speed) {
	case USB_SPEED_HIGH:
		/* Max NAK rate */
		if (usb_endpoint_xfer_control(&ep->desc) ||
1260
		    usb_endpoint_xfer_bulk(&ep->desc)) {
1261
			interval = xhci_parse_microframe_interval(udev, ep);
1262 1263
			break;
		}
1264
		/* Fall through - SS and HS isoc/int have same decoding */
1265

1266 1267
	case USB_SPEED_SUPER:
		if (usb_endpoint_xfer_int(&ep->desc) ||
1268 1269
		    usb_endpoint_xfer_isoc(&ep->desc)) {
			interval = xhci_parse_exponent_interval(udev, ep);
1270 1271
		}
		break;
1272

1273
	case USB_SPEED_FULL:
1274
		if (usb_endpoint_xfer_isoc(&ep->desc)) {
1275 1276 1277 1278
			interval = xhci_parse_exponent_interval(udev, ep);
			break;
		}
		/*
1279
		 * Fall through for interrupt endpoint interval decoding
1280 1281 1282 1283
		 * since it uses the same rules as low speed interrupt
		 * endpoints.
		 */

1284 1285
	case USB_SPEED_LOW:
		if (usb_endpoint_xfer_int(&ep->desc) ||
1286 1287 1288
		    usb_endpoint_xfer_isoc(&ep->desc)) {

			interval = xhci_parse_frame_interval(udev, ep);
1289 1290
		}
		break;
1291

1292 1293 1294 1295 1296 1297
	default:
		BUG();
	}
	return EP_INTERVAL(interval);
}

1298
/* The "Mult" field in the endpoint context is only set for SuperSpeed isoc eps.
1299 1300 1301 1302
 * High speed endpoint descriptors can define "the number of additional
 * transaction opportunities per microframe", but that goes in the Max Burst
 * endpoint context field.
 */
1303
static u32 xhci_get_endpoint_mult(struct usb_device *udev,
1304 1305
		struct usb_host_endpoint *ep)
{
1306 1307
	if (udev->speed != USB_SPEED_SUPER ||
			!usb_endpoint_xfer_isoc(&ep->desc))
1308
		return 0;
1309
	return ep->ss_ep_comp.bmAttributes;
1310 1311
}

1312
static u32 xhci_get_endpoint_type(struct usb_device *udev,
1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336
		struct usb_host_endpoint *ep)
{
	int in;
	u32 type;

	in = usb_endpoint_dir_in(&ep->desc);
	if (usb_endpoint_xfer_control(&ep->desc)) {
		type = EP_TYPE(CTRL_EP);
	} else if (usb_endpoint_xfer_bulk(&ep->desc)) {
		if (in)
			type = EP_TYPE(BULK_IN_EP);
		else
			type = EP_TYPE(BULK_OUT_EP);
	} else if (usb_endpoint_xfer_isoc(&ep->desc)) {
		if (in)
			type = EP_TYPE(ISOC_IN_EP);
		else
			type = EP_TYPE(ISOC_OUT_EP);
	} else if (usb_endpoint_xfer_int(&ep->desc)) {
		if (in)
			type = EP_TYPE(INT_IN_EP);
		else
			type = EP_TYPE(INT_OUT_EP);
	} else {
1337
		type = 0;
1338 1339 1340 1341
	}
	return type;
}

1342 1343 1344 1345
/* Return the maximum endpoint service interval time (ESIT) payload.
 * Basically, this is the maxpacket size, multiplied by the burst size
 * and mult size.
 */
1346
static u32 xhci_get_max_esit_payload(struct xhci_hcd *xhci,
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357
		struct usb_device *udev,
		struct usb_host_endpoint *ep)
{
	int max_burst;
	int max_packet;

	/* Only applies for interrupt or isochronous endpoints */
	if (usb_endpoint_xfer_control(&ep->desc) ||
			usb_endpoint_xfer_bulk(&ep->desc))
		return 0;

1358
	if (udev->speed == USB_SPEED_SUPER)
1359
		return le16_to_cpu(ep->ss_ep_comp.wBytesPerInterval);
1360

1361 1362
	max_packet = GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc));
	max_burst = (usb_endpoint_maxp(&ep->desc) & 0x1800) >> 11;
1363 1364 1365 1366
	/* A 0 in max burst means 1 transfer per ESIT */
	return max_packet * (max_burst + 1);
}

1367 1368 1369
/* Set up an endpoint with one ring segment.  Do not allocate stream rings.
 * Drivers will have to call usb_alloc_streams() to do that.
 */
1370 1371 1372
int xhci_endpoint_init(struct xhci_hcd *xhci,
		struct xhci_virt_device *virt_dev,
		struct usb_device *udev,
1373 1374
		struct usb_host_endpoint *ep,
		gfp_t mem_flags)
1375 1376 1377 1378 1379 1380
{
	unsigned int ep_index;
	struct xhci_ep_ctx *ep_ctx;
	struct xhci_ring *ep_ring;
	unsigned int max_packet;
	unsigned int max_burst;
A
Andiry Xu 已提交
1381
	enum xhci_ring_type type;
1382
	u32 max_esit_payload;
1383
	u32 endpoint_type;
1384 1385

	ep_index = xhci_get_endpoint_index(&ep->desc);
1386
	ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index);
1387

1388 1389 1390 1391 1392
	endpoint_type = xhci_get_endpoint_type(udev, ep);
	if (!endpoint_type)
		return -EINVAL;
	ep_ctx->ep_info2 = cpu_to_le32(endpoint_type);

A
Andiry Xu 已提交
1393
	type = usb_endpoint_type(&ep->desc);
1394
	/* Set up the endpoint ring */
A
Andiry Xu 已提交
1395
	virt_dev->eps[ep_index].new_ring =
1396
		xhci_ring_alloc(xhci, 2, 1, type, mem_flags);
1397 1398 1399 1400 1401 1402 1403 1404
	if (!virt_dev->eps[ep_index].new_ring) {
		/* Attempt to use the ring cache */
		if (virt_dev->num_rings_cached == 0)
			return -ENOMEM;
		virt_dev->eps[ep_index].new_ring =
			virt_dev->ring_cache[virt_dev->num_rings_cached];
		virt_dev->ring_cache[virt_dev->num_rings_cached] = NULL;
		virt_dev->num_rings_cached--;
1405
		xhci_reinit_cached_ring(xhci, virt_dev->eps[ep_index].new_ring,
1406
					1, type);
1407
	}
1408
	virt_dev->eps[ep_index].skip = false;
1409
	ep_ring = virt_dev->eps[ep_index].new_ring;
M
Matt Evans 已提交
1410
	ep_ctx->deq = cpu_to_le64(ep_ring->first_seg->dma | ep_ring->cycle_state);
1411

M
Matt Evans 已提交
1412 1413
	ep_ctx->ep_info = cpu_to_le32(xhci_get_endpoint_interval(udev, ep)
				      | EP_MULT(xhci_get_endpoint_mult(udev, ep)));
1414 1415 1416

	/* FIXME dig Mult and streams info out of ep companion desc */

1417
	/* Allow 3 retries for everything but isoc;
1418
	 * CErr shall be set to 0 for Isoch endpoints.
1419
	 */
1420
	if (!usb_endpoint_xfer_isoc(&ep->desc))
1421
		ep_ctx->ep_info2 |= cpu_to_le32(ERROR_COUNT(3));
1422
	else
1423
		ep_ctx->ep_info2 |= cpu_to_le32(ERROR_COUNT(0));
1424 1425

	/* Set the max packet size and max burst */
1426 1427
	max_packet = GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc));
	max_burst = 0;
1428 1429
	switch (udev->speed) {
	case USB_SPEED_SUPER:
S
Sarah Sharp 已提交
1430
		/* dig out max burst from ep companion desc */
1431
		max_burst = ep->ss_ep_comp.bMaxBurst;
1432 1433
		break;
	case USB_SPEED_HIGH:
1434 1435 1436
		/* Some devices get this wrong */
		if (usb_endpoint_xfer_bulk(&ep->desc))
			max_packet = 512;
1437 1438 1439 1440 1441
		/* bits 11:12 specify the number of additional transaction
		 * opportunities per microframe (USB 2.0, section 9.6.6)
		 */
		if (usb_endpoint_xfer_isoc(&ep->desc) ||
				usb_endpoint_xfer_int(&ep->desc)) {
1442
			max_burst = (usb_endpoint_maxp(&ep->desc)
M
Matt Evans 已提交
1443
				     & 0x1800) >> 11;
1444
		}
1445
		break;
1446 1447 1448 1449 1450 1451
	case USB_SPEED_FULL:
	case USB_SPEED_LOW:
		break;
	default:
		BUG();
	}
1452 1453
	ep_ctx->ep_info2 |= cpu_to_le32(MAX_PACKET(max_packet) |
			MAX_BURST(max_burst));
1454
	max_esit_payload = xhci_get_max_esit_payload(xhci, udev, ep);
M
Matt Evans 已提交
1455
	ep_ctx->tx_info = cpu_to_le32(MAX_ESIT_PAYLOAD_FOR_EP(max_esit_payload));
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470

	/*
	 * XXX no idea how to calculate the average TRB buffer length for bulk
	 * endpoints, as the driver gives us no clue how big each scatter gather
	 * list entry (or buffer) is going to be.
	 *
	 * For isochronous and interrupt endpoints, we set it to the max
	 * available, until we have new API in the USB core to allow drivers to
	 * declare how much bandwidth they actually need.
	 *
	 * Normally, it would be calculated by taking the total of the buffer
	 * lengths in the TD and then dividing by the number of TRBs in a TD,
	 * including link TRBs, No-op TRBs, and Event data TRBs.  Since we don't
	 * use Event Data TRBs, and we don't chain in a link TRB on short
	 * transfers, we're basically dividing by 1.
1471 1472 1473
	 *
	 * xHCI 1.0 specification indicates that the Average TRB Length should
	 * be set to 8 for control endpoints.
1474
	 */
1475 1476 1477 1478 1479
	if (usb_endpoint_xfer_control(&ep->desc) && xhci->hci_version == 0x100)
		ep_ctx->tx_info |= cpu_to_le32(AVG_TRB_LENGTH_FOR_EP(8));
	else
		ep_ctx->tx_info |=
			 cpu_to_le32(AVG_TRB_LENGTH_FOR_EP(max_esit_payload));
1480

1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
	/* FIXME Debug endpoint context */
	return 0;
}

void xhci_endpoint_zero(struct xhci_hcd *xhci,
		struct xhci_virt_device *virt_dev,
		struct usb_host_endpoint *ep)
{
	unsigned int ep_index;
	struct xhci_ep_ctx *ep_ctx;

	ep_index = xhci_get_endpoint_index(&ep->desc);
1493
	ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index);
1494 1495 1496

	ep_ctx->ep_info = 0;
	ep_ctx->ep_info2 = 0;
1497
	ep_ctx->deq = 0;
1498 1499 1500 1501 1502 1503
	ep_ctx->tx_info = 0;
	/* Don't free the endpoint ring until the set interface or configuration
	 * request succeeds.
	 */
}

1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
void xhci_clear_endpoint_bw_info(struct xhci_bw_info *bw_info)
{
	bw_info->ep_interval = 0;
	bw_info->mult = 0;
	bw_info->num_packets = 0;
	bw_info->max_packet_size = 0;
	bw_info->type = 0;
	bw_info->max_esit_payload = 0;
}

void xhci_update_bw_info(struct xhci_hcd *xhci,
		struct xhci_container_ctx *in_ctx,
		struct xhci_input_control_ctx *ctrl_ctx,
		struct xhci_virt_device *virt_dev)
{
	struct xhci_bw_info *bw_info;
	struct xhci_ep_ctx *ep_ctx;
	unsigned int ep_type;
	int i;

	for (i = 1; i < 31; ++i) {
		bw_info = &virt_dev->eps[i].bw_info;

		/* We can't tell what endpoint type is being dropped, but
		 * unconditionally clearing the bandwidth info for non-periodic
		 * endpoints should be harmless because the info will never be
		 * set in the first place.
		 */
		if (!EP_IS_ADDED(ctrl_ctx, i) && EP_IS_DROPPED(ctrl_ctx, i)) {
			/* Dropped endpoint */
			xhci_clear_endpoint_bw_info(bw_info);
			continue;
		}

		if (EP_IS_ADDED(ctrl_ctx, i)) {
			ep_ctx = xhci_get_ep_ctx(xhci, in_ctx, i);
			ep_type = CTX_TO_EP_TYPE(le32_to_cpu(ep_ctx->ep_info2));

			/* Ignore non-periodic endpoints */
			if (ep_type != ISOC_OUT_EP && ep_type != INT_OUT_EP &&
					ep_type != ISOC_IN_EP &&
					ep_type != INT_IN_EP)
				continue;

			/* Added or changed endpoint */
			bw_info->ep_interval = CTX_TO_EP_INTERVAL(
					le32_to_cpu(ep_ctx->ep_info));
1551 1552 1553
			/* Number of packets and mult are zero-based in the
			 * input context, but we want one-based for the
			 * interval table.
1554
			 */
1555 1556
			bw_info->mult = CTX_TO_EP_MULT(
					le32_to_cpu(ep_ctx->ep_info)) + 1;
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
			bw_info->num_packets = CTX_TO_MAX_BURST(
					le32_to_cpu(ep_ctx->ep_info2)) + 1;
			bw_info->max_packet_size = MAX_PACKET_DECODED(
					le32_to_cpu(ep_ctx->ep_info2));
			bw_info->type = ep_type;
			bw_info->max_esit_payload = CTX_TO_MAX_ESIT_PAYLOAD(
					le32_to_cpu(ep_ctx->tx_info));
		}
	}
}

1568 1569 1570 1571 1572
/* Copy output xhci_ep_ctx to the input xhci_ep_ctx copy.
 * Useful when you want to change one particular aspect of the endpoint and then
 * issue a configure endpoint command.
 */
void xhci_endpoint_copy(struct xhci_hcd *xhci,
1573 1574 1575
		struct xhci_container_ctx *in_ctx,
		struct xhci_container_ctx *out_ctx,
		unsigned int ep_index)
1576 1577 1578 1579
{
	struct xhci_ep_ctx *out_ep_ctx;
	struct xhci_ep_ctx *in_ep_ctx;

1580 1581
	out_ep_ctx = xhci_get_ep_ctx(xhci, out_ctx, ep_index);
	in_ep_ctx = xhci_get_ep_ctx(xhci, in_ctx, ep_index);
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593

	in_ep_ctx->ep_info = out_ep_ctx->ep_info;
	in_ep_ctx->ep_info2 = out_ep_ctx->ep_info2;
	in_ep_ctx->deq = out_ep_ctx->deq;
	in_ep_ctx->tx_info = out_ep_ctx->tx_info;
}

/* Copy output xhci_slot_ctx to the input xhci_slot_ctx.
 * Useful when you want to change one particular aspect of the endpoint and then
 * issue a configure endpoint command.  Only the context entries field matters,
 * but we'll copy the whole thing anyway.
 */
1594 1595 1596
void xhci_slot_copy(struct xhci_hcd *xhci,
		struct xhci_container_ctx *in_ctx,
		struct xhci_container_ctx *out_ctx)
1597 1598 1599 1600
{
	struct xhci_slot_ctx *in_slot_ctx;
	struct xhci_slot_ctx *out_slot_ctx;

1601 1602
	in_slot_ctx = xhci_get_slot_ctx(xhci, in_ctx);
	out_slot_ctx = xhci_get_slot_ctx(xhci, out_ctx);
1603 1604 1605 1606 1607 1608 1609

	in_slot_ctx->dev_info = out_slot_ctx->dev_info;
	in_slot_ctx->dev_info2 = out_slot_ctx->dev_info2;
	in_slot_ctx->tt_info = out_slot_ctx->tt_info;
	in_slot_ctx->dev_state = out_slot_ctx->dev_state;
}

1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
/* Set up the scratchpad buffer array and scratchpad buffers, if needed. */
static int scratchpad_alloc(struct xhci_hcd *xhci, gfp_t flags)
{
	int i;
	struct device *dev = xhci_to_hcd(xhci)->self.controller;
	int num_sp = HCS_MAX_SCRATCHPAD(xhci->hcs_params2);

	xhci_dbg(xhci, "Allocating %d scratchpad buffers\n", num_sp);

	if (!num_sp)
		return 0;

	xhci->scratchpad = kzalloc(sizeof(*xhci->scratchpad), flags);
	if (!xhci->scratchpad)
		goto fail_sp;

1626
	xhci->scratchpad->sp_array = dma_alloc_coherent(dev,
1627
				     num_sp * sizeof(u64),
1628
				     &xhci->scratchpad->sp_dma, flags);
1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
	if (!xhci->scratchpad->sp_array)
		goto fail_sp2;

	xhci->scratchpad->sp_buffers = kzalloc(sizeof(void *) * num_sp, flags);
	if (!xhci->scratchpad->sp_buffers)
		goto fail_sp3;

	xhci->scratchpad->sp_dma_buffers =
		kzalloc(sizeof(dma_addr_t) * num_sp, flags);

	if (!xhci->scratchpad->sp_dma_buffers)
		goto fail_sp4;

M
Matt Evans 已提交
1642
	xhci->dcbaa->dev_context_ptrs[0] = cpu_to_le64(xhci->scratchpad->sp_dma);
1643 1644
	for (i = 0; i < num_sp; i++) {
		dma_addr_t dma;
1645 1646
		void *buf = dma_alloc_coherent(dev, xhci->page_size, &dma,
				flags);
1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658
		if (!buf)
			goto fail_sp5;

		xhci->scratchpad->sp_array[i] = dma;
		xhci->scratchpad->sp_buffers[i] = buf;
		xhci->scratchpad->sp_dma_buffers[i] = dma;
	}

	return 0;

 fail_sp5:
	for (i = i - 1; i >= 0; i--) {
1659
		dma_free_coherent(dev, xhci->page_size,
1660 1661 1662 1663 1664 1665 1666 1667 1668
				    xhci->scratchpad->sp_buffers[i],
				    xhci->scratchpad->sp_dma_buffers[i]);
	}
	kfree(xhci->scratchpad->sp_dma_buffers);

 fail_sp4:
	kfree(xhci->scratchpad->sp_buffers);

 fail_sp3:
1669
	dma_free_coherent(dev, num_sp * sizeof(u64),
1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692
			    xhci->scratchpad->sp_array,
			    xhci->scratchpad->sp_dma);

 fail_sp2:
	kfree(xhci->scratchpad);
	xhci->scratchpad = NULL;

 fail_sp:
	return -ENOMEM;
}

static void scratchpad_free(struct xhci_hcd *xhci)
{
	int num_sp;
	int i;
	struct pci_dev	*pdev = to_pci_dev(xhci_to_hcd(xhci)->self.controller);

	if (!xhci->scratchpad)
		return;

	num_sp = HCS_MAX_SCRATCHPAD(xhci->hcs_params2);

	for (i = 0; i < num_sp; i++) {
1693
		dma_free_coherent(&pdev->dev, xhci->page_size,
1694 1695 1696 1697 1698
				    xhci->scratchpad->sp_buffers[i],
				    xhci->scratchpad->sp_dma_buffers[i]);
	}
	kfree(xhci->scratchpad->sp_dma_buffers);
	kfree(xhci->scratchpad->sp_buffers);
1699
	dma_free_coherent(&pdev->dev, num_sp * sizeof(u64),
1700 1701 1702 1703 1704 1705
			    xhci->scratchpad->sp_array,
			    xhci->scratchpad->sp_dma);
	kfree(xhci->scratchpad);
	xhci->scratchpad = NULL;
}

1706
struct xhci_command *xhci_alloc_command(struct xhci_hcd *xhci,
1707 1708
		bool allocate_in_ctx, bool allocate_completion,
		gfp_t mem_flags)
1709 1710 1711 1712 1713 1714 1715
{
	struct xhci_command *command;

	command = kzalloc(sizeof(*command), mem_flags);
	if (!command)
		return NULL;

1716 1717 1718 1719 1720 1721 1722 1723
	if (allocate_in_ctx) {
		command->in_ctx =
			xhci_alloc_container_ctx(xhci, XHCI_CTX_TYPE_INPUT,
					mem_flags);
		if (!command->in_ctx) {
			kfree(command);
			return NULL;
		}
1724
	}
1725 1726 1727 1728 1729 1730

	if (allocate_completion) {
		command->completion =
			kzalloc(sizeof(struct completion), mem_flags);
		if (!command->completion) {
			xhci_free_container_ctx(xhci, command->in_ctx);
1731
			kfree(command);
1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
			return NULL;
		}
		init_completion(command->completion);
	}

	command->status = 0;
	INIT_LIST_HEAD(&command->cmd_list);
	return command;
}

1742 1743
void xhci_urb_free_priv(struct xhci_hcd *xhci, struct urb_priv *urb_priv)
{
A
Andiry Xu 已提交
1744 1745 1746
	if (urb_priv) {
		kfree(urb_priv->td[0]);
		kfree(urb_priv);
1747 1748 1749
	}
}

1750 1751 1752 1753 1754 1755 1756 1757 1758
void xhci_free_command(struct xhci_hcd *xhci,
		struct xhci_command *command)
{
	xhci_free_container_ctx(xhci,
			command->in_ctx);
	kfree(command->completion);
	kfree(command);
}

1759 1760
void xhci_mem_cleanup(struct xhci_hcd *xhci)
{
1761
	struct pci_dev	*pdev = to_pci_dev(xhci_to_hcd(xhci)->self.controller);
A
Andiry Xu 已提交
1762
	struct dev_info	*dev_info, *next;
1763
	struct xhci_cd  *cur_cd, *next_cd;
A
Andiry Xu 已提交
1764
	unsigned long	flags;
1765
	int size;
1766
	int i, j, num_ports;
1767 1768 1769 1770

	/* Free the Event Ring Segment Table and the actual Event Ring */
	size = sizeof(struct xhci_erst_entry)*(xhci->erst.num_entries);
	if (xhci->erst.entries)
1771
		dma_free_coherent(&pdev->dev, size,
1772 1773 1774 1775 1776 1777 1778 1779
				xhci->erst.entries, xhci->erst.erst_dma_addr);
	xhci->erst.entries = NULL;
	xhci_dbg(xhci, "Freed ERST\n");
	if (xhci->event_ring)
		xhci_ring_free(xhci, xhci->event_ring);
	xhci->event_ring = NULL;
	xhci_dbg(xhci, "Freed event ring\n");

1780 1781
	if (xhci->lpm_command)
		xhci_free_command(xhci, xhci->lpm_command);
1782
	xhci->cmd_ring_reserved_trbs = 0;
1783 1784 1785 1786
	if (xhci->cmd_ring)
		xhci_ring_free(xhci, xhci->cmd_ring);
	xhci->cmd_ring = NULL;
	xhci_dbg(xhci, "Freed command ring\n");
1787 1788 1789 1790 1791
	list_for_each_entry_safe(cur_cd, next_cd,
			&xhci->cancel_cmd_list, cancel_cmd_list) {
		list_del(&cur_cd->cancel_cmd_list);
		kfree(cur_cd);
	}
1792 1793 1794 1795

	for (i = 1; i < MAX_HC_SLOTS; ++i)
		xhci_free_virt_device(xhci, i);

1796 1797 1798 1799
	if (xhci->segment_pool)
		dma_pool_destroy(xhci->segment_pool);
	xhci->segment_pool = NULL;
	xhci_dbg(xhci, "Freed segment pool\n");
1800 1801 1802 1803 1804 1805

	if (xhci->device_pool)
		dma_pool_destroy(xhci->device_pool);
	xhci->device_pool = NULL;
	xhci_dbg(xhci, "Freed device context pool\n");

1806 1807 1808 1809 1810 1811 1812 1813 1814 1815
	if (xhci->small_streams_pool)
		dma_pool_destroy(xhci->small_streams_pool);
	xhci->small_streams_pool = NULL;
	xhci_dbg(xhci, "Freed small stream array pool\n");

	if (xhci->medium_streams_pool)
		dma_pool_destroy(xhci->medium_streams_pool);
	xhci->medium_streams_pool = NULL;
	xhci_dbg(xhci, "Freed medium stream array pool\n");

1816
	if (xhci->dcbaa)
1817
		dma_free_coherent(&pdev->dev, sizeof(*xhci->dcbaa),
1818 1819
				xhci->dcbaa, xhci->dcbaa->dma);
	xhci->dcbaa = NULL;
1820

1821
	scratchpad_free(xhci);
1822

A
Andiry Xu 已提交
1823 1824 1825 1826 1827 1828 1829
	spin_lock_irqsave(&xhci->lock, flags);
	list_for_each_entry_safe(dev_info, next, &xhci->lpm_failed_devs, list) {
		list_del(&dev_info->list);
		kfree(dev_info);
	}
	spin_unlock_irqrestore(&xhci->lock, flags);

1830 1831 1832
	if (!xhci->rh_bw)
		goto no_bw;

1833 1834 1835 1836 1837 1838 1839
	num_ports = HCS_MAX_PORTS(xhci->hcs_params1);
	for (i = 0; i < num_ports; i++) {
		struct xhci_interval_bw_table *bwt = &xhci->rh_bw[i].bw_table;
		for (j = 0; j < XHCI_MAX_INTERVAL; j++) {
			struct list_head *ep = &bwt->interval_bw[j].endpoints;
			while (!list_empty(ep))
				list_del_init(ep->next);
1840 1841 1842
		}
	}

1843 1844 1845 1846 1847 1848
	for (i = 0; i < num_ports; i++) {
		struct xhci_tt_bw_info *tt, *n;
		list_for_each_entry_safe(tt, n, &xhci->rh_bw[i].tts, tt_list) {
			list_del(&tt->tt_list);
			kfree(tt);
		}
1849 1850
	}

1851
no_bw:
1852 1853
	xhci->num_usb2_ports = 0;
	xhci->num_usb3_ports = 0;
1854
	xhci->num_active_eps = 0;
1855 1856 1857
	kfree(xhci->usb2_ports);
	kfree(xhci->usb3_ports);
	kfree(xhci->port_array);
1858
	kfree(xhci->rh_bw);
1859
	kfree(xhci->ext_caps);
1860

1861 1862
	xhci->page_size = 0;
	xhci->page_shift = 0;
1863
	xhci->bus_state[0].bus_suspended = 0;
1864
	xhci->bus_state[1].bus_suspended = 0;
1865 1866
}

1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
static int xhci_test_trb_in_td(struct xhci_hcd *xhci,
		struct xhci_segment *input_seg,
		union xhci_trb *start_trb,
		union xhci_trb *end_trb,
		dma_addr_t input_dma,
		struct xhci_segment *result_seg,
		char *test_name, int test_number)
{
	unsigned long long start_dma;
	unsigned long long end_dma;
	struct xhci_segment *seg;

	start_dma = xhci_trb_virt_to_dma(input_seg, start_trb);
	end_dma = xhci_trb_virt_to_dma(input_seg, end_trb);

	seg = trb_in_td(input_seg, start_trb, end_trb, input_dma);
	if (seg != result_seg) {
		xhci_warn(xhci, "WARN: %s TRB math test %d failed!\n",
				test_name, test_number);
		xhci_warn(xhci, "Tested TRB math w/ seg %p and "
				"input DMA 0x%llx\n",
				input_seg,
				(unsigned long long) input_dma);
		xhci_warn(xhci, "starting TRB %p (0x%llx DMA), "
				"ending TRB %p (0x%llx DMA)\n",
				start_trb, start_dma,
				end_trb, end_dma);
		xhci_warn(xhci, "Expected seg %p, got seg %p\n",
				result_seg, seg);
		return -1;
	}
	return 0;
}

/* TRB math checks for xhci_trb_in_td(), using the command and event rings. */
static int xhci_check_trb_in_td_math(struct xhci_hcd *xhci, gfp_t mem_flags)
{
	struct {
		dma_addr_t		input_dma;
		struct xhci_segment	*result_seg;
	} simple_test_vector [] = {
		/* A zeroed DMA field should fail */
		{ 0, NULL },
		/* One TRB before the ring start should fail */
		{ xhci->event_ring->first_seg->dma - 16, NULL },
		/* One byte before the ring start should fail */
		{ xhci->event_ring->first_seg->dma - 1, NULL },
		/* Starting TRB should succeed */
		{ xhci->event_ring->first_seg->dma, xhci->event_ring->first_seg },
		/* Ending TRB should succeed */
		{ xhci->event_ring->first_seg->dma + (TRBS_PER_SEGMENT - 1)*16,
			xhci->event_ring->first_seg },
		/* One byte after the ring end should fail */
		{ xhci->event_ring->first_seg->dma + (TRBS_PER_SEGMENT - 1)*16 + 1, NULL },
		/* One TRB after the ring end should fail */
		{ xhci->event_ring->first_seg->dma + (TRBS_PER_SEGMENT)*16, NULL },
		/* An address of all ones should fail */
		{ (dma_addr_t) (~0), NULL },
	};
	struct {
		struct xhci_segment	*input_seg;
		union xhci_trb		*start_trb;
		union xhci_trb		*end_trb;
		dma_addr_t		input_dma;
		struct xhci_segment	*result_seg;
	} complex_test_vector [] = {
		/* Test feeding a valid DMA address from a different ring */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = xhci->event_ring->first_seg->trbs,
			.end_trb = &xhci->event_ring->first_seg->trbs[TRBS_PER_SEGMENT - 1],
			.input_dma = xhci->cmd_ring->first_seg->dma,
			.result_seg = NULL,
		},
		/* Test feeding a valid end TRB from a different ring */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = xhci->event_ring->first_seg->trbs,
			.end_trb = &xhci->cmd_ring->first_seg->trbs[TRBS_PER_SEGMENT - 1],
			.input_dma = xhci->cmd_ring->first_seg->dma,
			.result_seg = NULL,
		},
		/* Test feeding a valid start and end TRB from a different ring */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = xhci->cmd_ring->first_seg->trbs,
			.end_trb = &xhci->cmd_ring->first_seg->trbs[TRBS_PER_SEGMENT - 1],
			.input_dma = xhci->cmd_ring->first_seg->dma,
			.result_seg = NULL,
		},
		/* TRB in this ring, but after this TD */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = &xhci->event_ring->first_seg->trbs[0],
			.end_trb = &xhci->event_ring->first_seg->trbs[3],
			.input_dma = xhci->event_ring->first_seg->dma + 4*16,
			.result_seg = NULL,
		},
		/* TRB in this ring, but before this TD */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = &xhci->event_ring->first_seg->trbs[3],
			.end_trb = &xhci->event_ring->first_seg->trbs[6],
			.input_dma = xhci->event_ring->first_seg->dma + 2*16,
			.result_seg = NULL,
		},
		/* TRB in this ring, but after this wrapped TD */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = &xhci->event_ring->first_seg->trbs[TRBS_PER_SEGMENT - 3],
			.end_trb = &xhci->event_ring->first_seg->trbs[1],
			.input_dma = xhci->event_ring->first_seg->dma + 2*16,
			.result_seg = NULL,
		},
		/* TRB in this ring, but before this wrapped TD */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = &xhci->event_ring->first_seg->trbs[TRBS_PER_SEGMENT - 3],
			.end_trb = &xhci->event_ring->first_seg->trbs[1],
			.input_dma = xhci->event_ring->first_seg->dma + (TRBS_PER_SEGMENT - 4)*16,
			.result_seg = NULL,
		},
		/* TRB not in this ring, and we have a wrapped TD */
		{	.input_seg = xhci->event_ring->first_seg,
			.start_trb = &xhci->event_ring->first_seg->trbs[TRBS_PER_SEGMENT - 3],
			.end_trb = &xhci->event_ring->first_seg->trbs[1],
			.input_dma = xhci->cmd_ring->first_seg->dma + 2*16,
			.result_seg = NULL,
		},
	};

	unsigned int num_tests;
	int i, ret;

1994
	num_tests = ARRAY_SIZE(simple_test_vector);
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
	for (i = 0; i < num_tests; i++) {
		ret = xhci_test_trb_in_td(xhci,
				xhci->event_ring->first_seg,
				xhci->event_ring->first_seg->trbs,
				&xhci->event_ring->first_seg->trbs[TRBS_PER_SEGMENT - 1],
				simple_test_vector[i].input_dma,
				simple_test_vector[i].result_seg,
				"Simple", i);
		if (ret < 0)
			return ret;
	}

2007
	num_tests = ARRAY_SIZE(complex_test_vector);
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
	for (i = 0; i < num_tests; i++) {
		ret = xhci_test_trb_in_td(xhci,
				complex_test_vector[i].input_seg,
				complex_test_vector[i].start_trb,
				complex_test_vector[i].end_trb,
				complex_test_vector[i].input_dma,
				complex_test_vector[i].result_seg,
				"Complex", i);
		if (ret < 0)
			return ret;
	}
	xhci_dbg(xhci, "TRB math tests passed.\n");
	return 0;
}

2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045
static void xhci_set_hc_event_deq(struct xhci_hcd *xhci)
{
	u64 temp;
	dma_addr_t deq;

	deq = xhci_trb_virt_to_dma(xhci->event_ring->deq_seg,
			xhci->event_ring->dequeue);
	if (deq == 0 && !in_interrupt())
		xhci_warn(xhci, "WARN something wrong with SW event ring "
				"dequeue ptr.\n");
	/* Update HC event ring dequeue pointer */
	temp = xhci_read_64(xhci, &xhci->ir_set->erst_dequeue);
	temp &= ERST_PTR_MASK;
	/* Don't clear the EHB bit (which is RW1C) because
	 * there might be more events to service.
	 */
	temp &= ~ERST_EHB;
	xhci_dbg(xhci, "// Write event ring dequeue pointer, "
			"preserving EHB bit\n");
	xhci_write_64(xhci, ((u64) deq & (u64) ~ERST_PTR_MASK) | temp,
			&xhci->ir_set->erst_dequeue);
}

2046
static void xhci_add_in_port(struct xhci_hcd *xhci, unsigned int num_ports,
2047
		__le32 __iomem *addr, u8 major_revision, int max_caps)
2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070
{
	u32 temp, port_offset, port_count;
	int i;

	if (major_revision > 0x03) {
		xhci_warn(xhci, "Ignoring unknown port speed, "
				"Ext Cap %p, revision = 0x%x\n",
				addr, major_revision);
		/* Ignoring port protocol we can't understand. FIXME */
		return;
	}

	/* Port offset and count in the third dword, see section 7.2 */
	temp = xhci_readl(xhci, addr + 2);
	port_offset = XHCI_EXT_PORT_OFF(temp);
	port_count = XHCI_EXT_PORT_COUNT(temp);
	xhci_dbg(xhci, "Ext Cap %p, port offset = %u, "
			"count = %u, revision = 0x%x\n",
			addr, port_offset, port_count, major_revision);
	/* Port count includes the current port offset */
	if (port_offset == 0 || (port_offset + port_count - 1) > num_ports)
		/* WTF? "Valid values are ‘1’ to MaxPorts" */
		return;
A
Andiry Xu 已提交
2071

2072 2073 2074 2075
	/* cache usb2 port capabilities */
	if (major_revision < 0x03 && xhci->num_ext_caps < max_caps)
		xhci->ext_caps[xhci->num_ext_caps++] = temp;

A
Andiry Xu 已提交
2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091
	/* Check the host's USB2 LPM capability */
	if ((xhci->hci_version == 0x96) && (major_revision != 0x03) &&
			(temp & XHCI_L1C)) {
		xhci_dbg(xhci, "xHCI 0.96: support USB2 software lpm\n");
		xhci->sw_lpm_support = 1;
	}

	if ((xhci->hci_version >= 0x100) && (major_revision != 0x03)) {
		xhci_dbg(xhci, "xHCI 1.0: support USB2 software lpm\n");
		xhci->sw_lpm_support = 1;
		if (temp & XHCI_HLC) {
			xhci_dbg(xhci, "xHCI 1.0: support USB2 hardware lpm\n");
			xhci->hw_lpm_support = 1;
		}
	}

2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104
	port_offset--;
	for (i = port_offset; i < (port_offset + port_count); i++) {
		/* Duplicate entry.  Ignore the port if the revisions differ. */
		if (xhci->port_array[i] != 0) {
			xhci_warn(xhci, "Duplicate port entry, Ext Cap %p,"
					" port %u\n", addr, i);
			xhci_warn(xhci, "Port was marked as USB %u, "
					"duplicated as USB %u\n",
					xhci->port_array[i], major_revision);
			/* Only adjust the roothub port counts if we haven't
			 * found a similar duplicate.
			 */
			if (xhci->port_array[i] != major_revision &&
2105
				xhci->port_array[i] != DUPLICATE_ENTRY) {
2106 2107 2108 2109
				if (xhci->port_array[i] == 0x03)
					xhci->num_usb3_ports--;
				else
					xhci->num_usb2_ports--;
2110
				xhci->port_array[i] = DUPLICATE_ENTRY;
2111 2112
			}
			/* FIXME: Should we disable the port? */
2113
			continue;
2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
		}
		xhci->port_array[i] = major_revision;
		if (major_revision == 0x03)
			xhci->num_usb3_ports++;
		else
			xhci->num_usb2_ports++;
	}
	/* FIXME: Should we disable ports not in the Extended Capabilities? */
}

/*
 * Scan the Extended Capabilities for the "Supported Protocol Capabilities" that
 * specify what speeds each port is supposed to be.  We can't count on the port
 * speed bits in the PORTSC register being correct until a device is connected,
 * but we need to set up the two fake roothubs with the correct number of USB
 * 3.0 and USB 2.0 ports at host controller initialization time.
 */
static int xhci_setup_port_arrays(struct xhci_hcd *xhci, gfp_t flags)
{
2133 2134
	__le32 __iomem *addr, *tmp_addr;
	u32 offset, tmp_offset;
2135
	unsigned int num_ports;
2136
	int i, j, port_index;
2137
	int cap_count = 0;
2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151

	addr = &xhci->cap_regs->hcc_params;
	offset = XHCI_HCC_EXT_CAPS(xhci_readl(xhci, addr));
	if (offset == 0) {
		xhci_err(xhci, "No Extended Capability registers, "
				"unable to set up roothub.\n");
		return -ENODEV;
	}

	num_ports = HCS_MAX_PORTS(xhci->hcs_params1);
	xhci->port_array = kzalloc(sizeof(*xhci->port_array)*num_ports, flags);
	if (!xhci->port_array)
		return -ENOMEM;

2152 2153 2154
	xhci->rh_bw = kzalloc(sizeof(*xhci->rh_bw)*num_ports, flags);
	if (!xhci->rh_bw)
		return -ENOMEM;
2155 2156 2157
	for (i = 0; i < num_ports; i++) {
		struct xhci_interval_bw_table *bw_table;

2158
		INIT_LIST_HEAD(&xhci->rh_bw[i].tts);
2159 2160 2161 2162
		bw_table = &xhci->rh_bw[i].bw_table;
		for (j = 0; j < XHCI_MAX_INTERVAL; j++)
			INIT_LIST_HEAD(&bw_table->interval_bw[j].endpoints);
	}
2163

2164 2165 2166 2167 2168 2169
	/*
	 * For whatever reason, the first capability offset is from the
	 * capability register base, not from the HCCPARAMS register.
	 * See section 5.3.6 for offset calculation.
	 */
	addr = &xhci->cap_regs->hc_capbase + offset;
2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187

	tmp_addr = addr;
	tmp_offset = offset;

	/* count extended protocol capability entries for later caching */
	do {
		u32 cap_id;
		cap_id = xhci_readl(xhci, tmp_addr);
		if (XHCI_EXT_CAPS_ID(cap_id) == XHCI_EXT_CAPS_PROTOCOL)
			cap_count++;
		tmp_offset = XHCI_EXT_CAPS_NEXT(cap_id);
		tmp_addr += tmp_offset;
	} while (tmp_offset);

	xhci->ext_caps = kzalloc(sizeof(*xhci->ext_caps) * cap_count, flags);
	if (!xhci->ext_caps)
		return -ENOMEM;

2188 2189 2190 2191 2192 2193
	while (1) {
		u32 cap_id;

		cap_id = xhci_readl(xhci, addr);
		if (XHCI_EXT_CAPS_ID(cap_id) == XHCI_EXT_CAPS_PROTOCOL)
			xhci_add_in_port(xhci, num_ports, addr,
2194 2195
					(u8) XHCI_EXT_PORT_MAJOR(cap_id),
					cap_count);
2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212
		offset = XHCI_EXT_CAPS_NEXT(cap_id);
		if (!offset || (xhci->num_usb2_ports + xhci->num_usb3_ports)
				== num_ports)
			break;
		/*
		 * Once you're into the Extended Capabilities, the offset is
		 * always relative to the register holding the offset.
		 */
		addr += offset;
	}

	if (xhci->num_usb2_ports == 0 && xhci->num_usb3_ports == 0) {
		xhci_warn(xhci, "No ports on the roothubs?\n");
		return -ENODEV;
	}
	xhci_dbg(xhci, "Found %u USB 2.0 ports and %u USB 3.0 ports.\n",
			xhci->num_usb2_ports, xhci->num_usb3_ports);
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226

	/* Place limits on the number of roothub ports so that the hub
	 * descriptors aren't longer than the USB core will allocate.
	 */
	if (xhci->num_usb3_ports > 15) {
		xhci_dbg(xhci, "Limiting USB 3.0 roothub ports to 15.\n");
		xhci->num_usb3_ports = 15;
	}
	if (xhci->num_usb2_ports > USB_MAXCHILDREN) {
		xhci_dbg(xhci, "Limiting USB 2.0 roothub ports to %u.\n",
				USB_MAXCHILDREN);
		xhci->num_usb2_ports = USB_MAXCHILDREN;
	}

2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237
	/*
	 * Note we could have all USB 3.0 ports, or all USB 2.0 ports.
	 * Not sure how the USB core will handle a hub with no ports...
	 */
	if (xhci->num_usb2_ports) {
		xhci->usb2_ports = kmalloc(sizeof(*xhci->usb2_ports)*
				xhci->num_usb2_ports, flags);
		if (!xhci->usb2_ports)
			return -ENOMEM;

		port_index = 0;
2238 2239 2240
		for (i = 0; i < num_ports; i++) {
			if (xhci->port_array[i] == 0x03 ||
					xhci->port_array[i] == 0 ||
2241
					xhci->port_array[i] == DUPLICATE_ENTRY)
2242 2243 2244 2245 2246 2247 2248 2249 2250
				continue;

			xhci->usb2_ports[port_index] =
				&xhci->op_regs->port_status_base +
				NUM_PORT_REGS*i;
			xhci_dbg(xhci, "USB 2.0 port at index %u, "
					"addr = %p\n", i,
					xhci->usb2_ports[port_index]);
			port_index++;
2251 2252
			if (port_index == xhci->num_usb2_ports)
				break;
2253
		}
2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270
	}
	if (xhci->num_usb3_ports) {
		xhci->usb3_ports = kmalloc(sizeof(*xhci->usb3_ports)*
				xhci->num_usb3_ports, flags);
		if (!xhci->usb3_ports)
			return -ENOMEM;

		port_index = 0;
		for (i = 0; i < num_ports; i++)
			if (xhci->port_array[i] == 0x03) {
				xhci->usb3_ports[port_index] =
					&xhci->op_regs->port_status_base +
					NUM_PORT_REGS*i;
				xhci_dbg(xhci, "USB 3.0 port at index %u, "
						"addr = %p\n", i,
						xhci->usb3_ports[port_index]);
				port_index++;
2271 2272
				if (port_index == xhci->num_usb3_ports)
					break;
2273 2274 2275 2276
			}
	}
	return 0;
}
2277

2278 2279
int xhci_mem_init(struct xhci_hcd *xhci, gfp_t flags)
{
2280 2281
	dma_addr_t	dma;
	struct device	*dev = xhci_to_hcd(xhci)->self.controller;
2282
	unsigned int	val, val2;
2283
	u64		val_64;
2284
	struct xhci_segment	*seg;
2285
	u32 page_size, temp;
2286 2287
	int i;

2288 2289 2290
	INIT_LIST_HEAD(&xhci->lpm_failed_devs);
	INIT_LIST_HEAD(&xhci->cancel_cmd_list);

2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
	page_size = xhci_readl(xhci, &xhci->op_regs->page_size);
	xhci_dbg(xhci, "Supported page size register = 0x%x\n", page_size);
	for (i = 0; i < 16; i++) {
		if ((0x1 & page_size) != 0)
			break;
		page_size = page_size >> 1;
	}
	if (i < 16)
		xhci_dbg(xhci, "Supported page size of %iK\n", (1 << (i+12)) / 1024);
	else
		xhci_warn(xhci, "WARN: no supported page size\n");
	/* Use 4K pages, since that's common and the minimum the HC supports */
	xhci->page_shift = 12;
	xhci->page_size = 1 << xhci->page_shift;
	xhci_dbg(xhci, "HCD page size set to %iK\n", xhci->page_size / 1024);

	/*
	 * Program the Number of Device Slots Enabled field in the CONFIG
	 * register with the max value of slots the HC can handle.
	 */
	val = HCS_MAX_SLOTS(xhci_readl(xhci, &xhci->cap_regs->hcs_params1));
	xhci_dbg(xhci, "// xHC can handle at most %d device slots.\n",
			(unsigned int) val);
	val2 = xhci_readl(xhci, &xhci->op_regs->config_reg);
	val |= (val2 & ~HCS_SLOTS_MASK);
	xhci_dbg(xhci, "// Setting Max device slots reg = 0x%x.\n",
			(unsigned int) val);
	xhci_writel(xhci, val, &xhci->op_regs->config_reg);

2320 2321 2322 2323
	/*
	 * Section 5.4.8 - doorbell array must be
	 * "physically contiguous and 64-byte (cache line) aligned".
	 */
2324 2325
	xhci->dcbaa = dma_alloc_coherent(dev, sizeof(*xhci->dcbaa), &dma,
			GFP_KERNEL);
2326 2327 2328 2329
	if (!xhci->dcbaa)
		goto fail;
	memset(xhci->dcbaa, 0, sizeof *(xhci->dcbaa));
	xhci->dcbaa->dma = dma;
2330 2331
	xhci_dbg(xhci, "// Device context base array address = 0x%llx (DMA), %p (virt)\n",
			(unsigned long long)xhci->dcbaa->dma, xhci->dcbaa);
2332
	xhci_write_64(xhci, dma, &xhci->op_regs->dcbaa_ptr);
2333

2334 2335 2336 2337 2338 2339 2340
	/*
	 * Initialize the ring segment pool.  The ring must be a contiguous
	 * structure comprised of TRBs.  The TRBs must be 16 byte aligned,
	 * however, the command ring segment needs 64-byte aligned segments,
	 * so we pick the greater alignment need.
	 */
	xhci->segment_pool = dma_pool_create("xHCI ring segments", dev,
2341
			TRB_SEGMENT_SIZE, 64, xhci->page_size);
2342

2343 2344
	/* See Table 46 and Note on Figure 55 */
	xhci->device_pool = dma_pool_create("xHCI input/output contexts", dev,
2345
			2112, 64, xhci->page_size);
2346
	if (!xhci->segment_pool || !xhci->device_pool)
2347 2348
		goto fail;

2349 2350 2351 2352 2353 2354 2355 2356 2357 2358
	/* Linear stream context arrays don't have any boundary restrictions,
	 * and only need to be 16-byte aligned.
	 */
	xhci->small_streams_pool =
		dma_pool_create("xHCI 256 byte stream ctx arrays",
			dev, SMALL_STREAM_ARRAY_SIZE, 16, 0);
	xhci->medium_streams_pool =
		dma_pool_create("xHCI 1KB stream ctx arrays",
			dev, MEDIUM_STREAM_ARRAY_SIZE, 16, 0);
	/* Any stream context array bigger than MEDIUM_STREAM_ARRAY_SIZE
2359
	 * will be allocated with dma_alloc_coherent()
2360 2361 2362 2363 2364
	 */

	if (!xhci->small_streams_pool || !xhci->medium_streams_pool)
		goto fail;

2365
	/* Set up the command ring to have one segments for now. */
2366
	xhci->cmd_ring = xhci_ring_alloc(xhci, 1, 1, TYPE_COMMAND, flags);
2367 2368
	if (!xhci->cmd_ring)
		goto fail;
2369 2370 2371
	xhci_dbg(xhci, "Allocated command ring at %p\n", xhci->cmd_ring);
	xhci_dbg(xhci, "First segment DMA is 0x%llx\n",
			(unsigned long long)xhci->cmd_ring->first_seg->dma);
2372 2373

	/* Set the address in the Command Ring Control register */
2374 2375 2376
	val_64 = xhci_read_64(xhci, &xhci->op_regs->cmd_ring);
	val_64 = (val_64 & (u64) CMD_RING_RSVD_BITS) |
		(xhci->cmd_ring->first_seg->dma & (u64) ~CMD_RING_RSVD_BITS) |
2377
		xhci->cmd_ring->cycle_state;
2378 2379
	xhci_dbg(xhci, "// Setting command ring address to 0x%x\n", val);
	xhci_write_64(xhci, val_64, &xhci->op_regs->cmd_ring);
2380 2381
	xhci_dbg_cmd_ptrs(xhci);

2382 2383 2384 2385 2386 2387 2388 2389 2390 2391
	xhci->lpm_command = xhci_alloc_command(xhci, true, true, flags);
	if (!xhci->lpm_command)
		goto fail;

	/* Reserve one command ring TRB for disabling LPM.
	 * Since the USB core grabs the shared usb_bus bandwidth mutex before
	 * disabling LPM, we only need to reserve one TRB for all devices.
	 */
	xhci->cmd_ring_reserved_trbs++;

2392 2393 2394 2395
	val = xhci_readl(xhci, &xhci->cap_regs->db_off);
	val &= DBOFF_MASK;
	xhci_dbg(xhci, "// Doorbell array is located at offset 0x%x"
			" from cap regs base addr\n", val);
2396
	xhci->dba = (void __iomem *) xhci->cap_regs + val;
2397 2398 2399
	xhci_dbg_regs(xhci);
	xhci_print_run_regs(xhci);
	/* Set ir_set to interrupt register set 0 */
2400
	xhci->ir_set = &xhci->run_regs->ir_set[0];
2401 2402 2403 2404 2405 2406

	/*
	 * Event ring setup: Allocate a normal ring, but also setup
	 * the event ring segment table (ERST).  Section 4.9.3.
	 */
	xhci_dbg(xhci, "// Allocating event ring\n");
2407
	xhci->event_ring = xhci_ring_alloc(xhci, ERST_NUM_SEGS, 1, TYPE_EVENT,
2408
						flags);
2409 2410
	if (!xhci->event_ring)
		goto fail;
2411 2412
	if (xhci_check_trb_in_td_math(xhci, flags) < 0)
		goto fail;
2413

2414 2415 2416
	xhci->erst.entries = dma_alloc_coherent(dev,
			sizeof(struct xhci_erst_entry) * ERST_NUM_SEGS, &dma,
			GFP_KERNEL);
2417 2418
	if (!xhci->erst.entries)
		goto fail;
2419 2420
	xhci_dbg(xhci, "// Allocated event ring segment table at 0x%llx\n",
			(unsigned long long)dma);
2421 2422 2423 2424

	memset(xhci->erst.entries, 0, sizeof(struct xhci_erst_entry)*ERST_NUM_SEGS);
	xhci->erst.num_entries = ERST_NUM_SEGS;
	xhci->erst.erst_dma_addr = dma;
2425
	xhci_dbg(xhci, "Set ERST to 0; private num segs = %i, virt addr = %p, dma addr = 0x%llx\n",
2426
			xhci->erst.num_entries,
2427 2428
			xhci->erst.entries,
			(unsigned long long)xhci->erst.erst_dma_addr);
2429 2430 2431 2432

	/* set ring base address and size for each segment table entry */
	for (val = 0, seg = xhci->event_ring->first_seg; val < ERST_NUM_SEGS; val++) {
		struct xhci_erst_entry *entry = &xhci->erst.entries[val];
M
Matt Evans 已提交
2433 2434
		entry->seg_addr = cpu_to_le64(seg->dma);
		entry->seg_size = cpu_to_le32(TRBS_PER_SEGMENT);
2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
		entry->rsvd = 0;
		seg = seg->next;
	}

	/* set ERST count with the number of entries in the segment table */
	val = xhci_readl(xhci, &xhci->ir_set->erst_size);
	val &= ERST_SIZE_MASK;
	val |= ERST_NUM_SEGS;
	xhci_dbg(xhci, "// Write ERST size = %i to ir_set 0 (some bits preserved)\n",
			val);
	xhci_writel(xhci, val, &xhci->ir_set->erst_size);

	xhci_dbg(xhci, "// Set ERST entries to point to event ring.\n");
	/* set the segment table base address */
2449 2450
	xhci_dbg(xhci, "// Set ERST base address for ir_set 0 = 0x%llx\n",
			(unsigned long long)xhci->erst.erst_dma_addr);
2451 2452 2453 2454
	val_64 = xhci_read_64(xhci, &xhci->ir_set->erst_base);
	val_64 &= ERST_PTR_MASK;
	val_64 |= (xhci->erst.erst_dma_addr & (u64) ~ERST_PTR_MASK);
	xhci_write_64(xhci, val_64, &xhci->ir_set->erst_base);
2455 2456

	/* Set the event ring dequeue address */
2457
	xhci_set_hc_event_deq(xhci);
2458
	xhci_dbg(xhci, "Wrote ERST address to ir_set 0.\n");
2459
	xhci_print_ir_set(xhci, 0);
2460 2461 2462 2463 2464 2465

	/*
	 * XXX: Might need to set the Interrupter Moderation Register to
	 * something other than the default (~1ms minimum between interrupts).
	 * See section 5.5.1.2.
	 */
2466 2467
	init_completion(&xhci->addr_dev);
	for (i = 0; i < MAX_HC_SLOTS; ++i)
2468
		xhci->devs[i] = NULL;
2469
	for (i = 0; i < USB_MAXCHILDREN; ++i) {
2470
		xhci->bus_state[0].resume_done[i] = 0;
2471 2472
		xhci->bus_state[1].resume_done[i] = 0;
	}
2473

2474 2475
	if (scratchpad_alloc(xhci, flags))
		goto fail;
2476 2477
	if (xhci_setup_port_arrays(xhci, flags))
		goto fail;
2478

2479 2480 2481 2482 2483 2484 2485 2486 2487
	/* Enable USB 3.0 device notifications for function remote wake, which
	 * is necessary for allowing USB 3.0 devices to do remote wakeup from
	 * U3 (device suspend).
	 */
	temp = xhci_readl(xhci, &xhci->op_regs->dev_notification);
	temp &= ~DEV_NOTE_MASK;
	temp |= DEV_NOTE_FWAKE;
	xhci_writel(xhci, temp, &xhci->op_regs->dev_notification);

2488
	return 0;
2489

2490 2491
fail:
	xhci_warn(xhci, "Couldn't initialize memory\n");
2492 2493
	xhci_halt(xhci);
	xhci_reset(xhci);
2494 2495 2496
	xhci_mem_cleanup(xhci);
	return -ENOMEM;
}