intel_ringbuffer.c 22.5 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 24 25 26 27 28 29 30 31 32
/*
 * Copyright © 2008-2010 Intel Corporation
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 *
 * Authors:
 *    Eric Anholt <eric@anholt.net>
 *    Zou Nan hai <nanhai.zou@intel.com>
 *    Xiang Hai hao<haihao.xiang@intel.com>
 *
 */

#include "drmP.h"
#include "drm.h"
#include "i915_drv.h"
33
#include "i915_drm.h"
34 35
#include "i915_trace.h"

36 37 38 39 40 41 42 43 44 45 46 47 48 49
static u32 i915_gem_get_seqno(struct drm_device *dev)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	u32 seqno;

	seqno = dev_priv->next_seqno;

	/* reserve 0 for non-seqno */
	if (++dev_priv->next_seqno == 0)
		dev_priv->next_seqno = 1;

	return seqno;
}

50 51 52 53 54
static void
render_ring_flush(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		u32	invalidate_domains,
		u32	flush_domains)
55
{
56 57 58
	drm_i915_private_t *dev_priv = dev->dev_private;
	u32 cmd;

59 60 61 62
#if WATCH_EXEC
	DRM_INFO("%s: invalidate %08x flush %08x\n", __func__,
		  invalidate_domains, flush_domains);
#endif
63 64

	trace_i915_gem_request_flush(dev, dev_priv->next_seqno,
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
				     invalidate_domains, flush_domains);

	if ((invalidate_domains | flush_domains) & I915_GEM_GPU_DOMAINS) {
		/*
		 * read/write caches:
		 *
		 * I915_GEM_DOMAIN_RENDER is always invalidated, but is
		 * only flushed if MI_NO_WRITE_FLUSH is unset.  On 965, it is
		 * also flushed at 2d versus 3d pipeline switches.
		 *
		 * read-only caches:
		 *
		 * I915_GEM_DOMAIN_SAMPLER is flushed on pre-965 if
		 * MI_READ_FLUSH is set, and is always flushed on 965.
		 *
		 * I915_GEM_DOMAIN_COMMAND may not exist?
		 *
		 * I915_GEM_DOMAIN_INSTRUCTION, which exists on 965, is
		 * invalidated when MI_EXE_FLUSH is set.
		 *
		 * I915_GEM_DOMAIN_VERTEX, which exists on 965, is
		 * invalidated with every MI_FLUSH.
		 *
		 * TLBs:
		 *
		 * On 965, TLBs associated with I915_GEM_DOMAIN_COMMAND
		 * and I915_GEM_DOMAIN_CPU in are invalidated at PTE write and
		 * I915_GEM_DOMAIN_RENDER and I915_GEM_DOMAIN_SAMPLER
		 * are flushed at any MI_FLUSH.
		 */

		cmd = MI_FLUSH | MI_NO_WRITE_FLUSH;
		if ((invalidate_domains|flush_domains) &
		    I915_GEM_DOMAIN_RENDER)
			cmd &= ~MI_NO_WRITE_FLUSH;
100
		if (INTEL_INFO(dev)->gen < 4) {
101 102 103 104 105 106 107 108 109 110 111 112 113
			/*
			 * On the 965, the sampler cache always gets flushed
			 * and this bit is reserved.
			 */
			if (invalidate_domains & I915_GEM_DOMAIN_SAMPLER)
				cmd |= MI_READ_FLUSH;
		}
		if (invalidate_domains & I915_GEM_DOMAIN_INSTRUCTION)
			cmd |= MI_EXE_FLUSH;

#if WATCH_EXEC
		DRM_INFO("%s: queue flush %08x to ring\n", __func__, cmd);
#endif
114
		intel_ring_begin(dev, ring, 2);
115 116 117
		intel_ring_emit(dev, ring, cmd);
		intel_ring_emit(dev, ring, MI_NOOP);
		intel_ring_advance(dev, ring);
118
	}
119 120

	i915_gem_process_flushing_list(dev, flush_domains, ring);
121 122 123 124 125 126 127 128
}

static unsigned int render_ring_get_head(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	return I915_READ(PRB0_HEAD) & HEAD_ADDR;
}
129

130 131 132 133 134
static unsigned int render_ring_get_tail(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	return I915_READ(PRB0_TAIL) & TAIL_ADDR;
135
}
136 137 138 139 140

static unsigned int render_ring_get_active_head(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
141
	u32 acthd_reg = INTEL_INFO(dev)->gen ? ACTHD_I965 : ACTHD;
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224

	return I915_READ(acthd_reg);
}

static void render_ring_advance_ring(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	I915_WRITE(PRB0_TAIL, ring->tail);
}

static int init_ring_common(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	u32 head;
	drm_i915_private_t *dev_priv = dev->dev_private;
	struct drm_i915_gem_object *obj_priv;
	obj_priv = to_intel_bo(ring->gem_object);

	/* Stop the ring if it's running. */
	I915_WRITE(ring->regs.ctl, 0);
	I915_WRITE(ring->regs.head, 0);
	I915_WRITE(ring->regs.tail, 0);

	/* Initialize the ring. */
	I915_WRITE(ring->regs.start, obj_priv->gtt_offset);
	head = ring->get_head(dev, ring);

	/* G45 ring initialization fails to reset head to zero */
	if (head != 0) {
		DRM_ERROR("%s head not reset to zero "
				"ctl %08x head %08x tail %08x start %08x\n",
				ring->name,
				I915_READ(ring->regs.ctl),
				I915_READ(ring->regs.head),
				I915_READ(ring->regs.tail),
				I915_READ(ring->regs.start));

		I915_WRITE(ring->regs.head, 0);

		DRM_ERROR("%s head forced to zero "
				"ctl %08x head %08x tail %08x start %08x\n",
				ring->name,
				I915_READ(ring->regs.ctl),
				I915_READ(ring->regs.head),
				I915_READ(ring->regs.tail),
				I915_READ(ring->regs.start));
	}

	I915_WRITE(ring->regs.ctl,
			((ring->gem_object->size - PAGE_SIZE) & RING_NR_PAGES)
			| RING_NO_REPORT | RING_VALID);

	head = I915_READ(ring->regs.head) & HEAD_ADDR;
	/* If the head is still not zero, the ring is dead */
	if (head != 0) {
		DRM_ERROR("%s initialization failed "
				"ctl %08x head %08x tail %08x start %08x\n",
				ring->name,
				I915_READ(ring->regs.ctl),
				I915_READ(ring->regs.head),
				I915_READ(ring->regs.tail),
				I915_READ(ring->regs.start));
		return -EIO;
	}

	if (!drm_core_check_feature(dev, DRIVER_MODESET))
		i915_kernel_lost_context(dev);
	else {
		ring->head = ring->get_head(dev, ring);
		ring->tail = ring->get_tail(dev, ring);
		ring->space = ring->head - (ring->tail + 8);
		if (ring->space < 0)
			ring->space += ring->size;
	}
	return 0;
}

static int init_render_ring(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	int ret = init_ring_common(dev, ring);
225 226
	int mode;

227
	if (INTEL_INFO(dev)->gen > 3) {
228 229 230 231
		mode = VS_TIMER_DISPATCH << 16 | VS_TIMER_DISPATCH;
		if (IS_GEN6(dev))
			mode |= MI_FLUSH_ENABLE << 16 | MI_FLUSH_ENABLE;
		I915_WRITE(MI_MODE, mode);
232 233 234 235
	}
	return ret;
}

236
#define PIPE_CONTROL_FLUSH(addr)					\
237
do {									\
238
	OUT_RING(GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE |		\
239
		 PIPE_CONTROL_DEPTH_STALL | 2);				\
240 241 242
	OUT_RING(addr | PIPE_CONTROL_GLOBAL_GTT);			\
	OUT_RING(0);							\
	OUT_RING(0);							\
243
} while (0)
244 245 246 247 248 249 250 251 252

/**
 * Creates a new sequence number, emitting a write of it to the status page
 * plus an interrupt, which will trigger i915_user_interrupt_handler.
 *
 * Must be called with struct_lock held.
 *
 * Returned sequence numbers are nonzero on success.
 */
253 254 255 256 257
static u32
render_ring_add_request(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		struct drm_file *file_priv,
		u32 flush_domains)
258 259
{
	drm_i915_private_t *dev_priv = dev->dev_private;
260 261 262
	u32 seqno;

	seqno = i915_gem_get_seqno(dev);
263 264 265 266 267 268 269 270 271 272 273 274 275

	if (IS_GEN6(dev)) {
		BEGIN_LP_RING(6);
		OUT_RING(GFX_OP_PIPE_CONTROL | 3);
		OUT_RING(PIPE_CONTROL_QW_WRITE |
			 PIPE_CONTROL_WC_FLUSH | PIPE_CONTROL_IS_FLUSH |
			 PIPE_CONTROL_NOTIFY);
		OUT_RING(dev_priv->seqno_gfx_addr | PIPE_CONTROL_GLOBAL_GTT);
		OUT_RING(seqno);
		OUT_RING(0);
		OUT_RING(0);
		ADVANCE_LP_RING();
	} else if (HAS_PIPE_CONTROL(dev)) {
276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
		u32 scratch_addr = dev_priv->seqno_gfx_addr + 128;

		/*
		 * Workaround qword write incoherence by flushing the
		 * PIPE_NOTIFY buffers out to memory before requesting
		 * an interrupt.
		 */
		BEGIN_LP_RING(32);
		OUT_RING(GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE |
			 PIPE_CONTROL_WC_FLUSH | PIPE_CONTROL_TC_FLUSH);
		OUT_RING(dev_priv->seqno_gfx_addr | PIPE_CONTROL_GLOBAL_GTT);
		OUT_RING(seqno);
		OUT_RING(0);
		PIPE_CONTROL_FLUSH(scratch_addr);
		scratch_addr += 128; /* write to separate cachelines */
		PIPE_CONTROL_FLUSH(scratch_addr);
		scratch_addr += 128;
		PIPE_CONTROL_FLUSH(scratch_addr);
		scratch_addr += 128;
		PIPE_CONTROL_FLUSH(scratch_addr);
		scratch_addr += 128;
		PIPE_CONTROL_FLUSH(scratch_addr);
		scratch_addr += 128;
		PIPE_CONTROL_FLUSH(scratch_addr);
		OUT_RING(GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE |
			 PIPE_CONTROL_WC_FLUSH | PIPE_CONTROL_TC_FLUSH |
			 PIPE_CONTROL_NOTIFY);
		OUT_RING(dev_priv->seqno_gfx_addr | PIPE_CONTROL_GLOBAL_GTT);
		OUT_RING(seqno);
		OUT_RING(0);
		ADVANCE_LP_RING();
	} else {
		BEGIN_LP_RING(4);
		OUT_RING(MI_STORE_DWORD_INDEX);
		OUT_RING(I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
		OUT_RING(seqno);

		OUT_RING(MI_USER_INTERRUPT);
		ADVANCE_LP_RING();
	}
	return seqno;
}

319 320 321 322 323 324 325 326 327 328 329 330 331 332
static u32
render_ring_get_gem_seqno(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
	if (HAS_PIPE_CONTROL(dev))
		return ((volatile u32 *)(dev_priv->seqno_page))[0];
	else
		return intel_read_status_page(ring, I915_GEM_HWS_INDEX);
}

static void
render_ring_get_user_irq(struct drm_device *dev,
		struct intel_ring_buffer *ring)
333 334 335 336 337
{
	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
	unsigned long irqflags;

	spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags);
338
	if (dev->irq_enabled && (++ring->user_irq_refcount == 1)) {
339 340 341 342 343 344 345 346
		if (HAS_PCH_SPLIT(dev))
			ironlake_enable_graphics_irq(dev_priv, GT_PIPE_NOTIFY);
		else
			i915_enable_irq(dev_priv, I915_USER_INTERRUPT);
	}
	spin_unlock_irqrestore(&dev_priv->user_irq_lock, irqflags);
}

347 348 349
static void
render_ring_put_user_irq(struct drm_device *dev,
		struct intel_ring_buffer *ring)
350 351 352 353 354
{
	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
	unsigned long irqflags;

	spin_lock_irqsave(&dev_priv->user_irq_lock, irqflags);
355 356
	BUG_ON(dev->irq_enabled && ring->user_irq_refcount <= 0);
	if (dev->irq_enabled && (--ring->user_irq_refcount == 0)) {
357 358 359 360 361 362 363 364
		if (HAS_PCH_SPLIT(dev))
			ironlake_disable_graphics_irq(dev_priv, GT_PIPE_NOTIFY);
		else
			i915_disable_irq(dev_priv, I915_USER_INTERRUPT);
	}
	spin_unlock_irqrestore(&dev_priv->user_irq_lock, irqflags);
}

365 366 367 368 369 370 371 372 373 374 375 376 377 378
static void render_setup_status_page(struct drm_device *dev,
	struct	intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	if (IS_GEN6(dev)) {
		I915_WRITE(HWS_PGA_GEN6, ring->status_page.gfx_addr);
		I915_READ(HWS_PGA_GEN6); /* posting read */
	} else {
		I915_WRITE(HWS_PGA, ring->status_page.gfx_addr);
		I915_READ(HWS_PGA); /* posting read */
	}

}

379 380 381 382 383 384
void
bsd_ring_flush(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		u32     invalidate_domains,
		u32     flush_domains)
{
385
	intel_ring_begin(dev, ring, 2);
386 387 388
	intel_ring_emit(dev, ring, MI_FLUSH);
	intel_ring_emit(dev, ring, MI_NOOP);
	intel_ring_advance(dev, ring);
389 390

	i915_gem_process_flushing_list(dev, flush_domains, ring);
391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433
}

static inline unsigned int bsd_ring_get_head(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	return I915_READ(BSD_RING_HEAD) & HEAD_ADDR;
}

static inline unsigned int bsd_ring_get_tail(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	return I915_READ(BSD_RING_TAIL) & TAIL_ADDR;
}

static inline unsigned int bsd_ring_get_active_head(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	return I915_READ(BSD_RING_ACTHD);
}

static inline void bsd_ring_advance_ring(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	I915_WRITE(BSD_RING_TAIL, ring->tail);
}

static int init_bsd_ring(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	return init_ring_common(dev, ring);
}

static u32
bsd_ring_add_request(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		struct drm_file *file_priv,
		u32 flush_domains)
{
	u32 seqno;
434 435 436

	seqno = i915_gem_get_seqno(dev);

437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
	intel_ring_begin(dev, ring, 4);
	intel_ring_emit(dev, ring, MI_STORE_DWORD_INDEX);
	intel_ring_emit(dev, ring,
			I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
	intel_ring_emit(dev, ring, seqno);
	intel_ring_emit(dev, ring, MI_USER_INTERRUPT);
	intel_ring_advance(dev, ring);

	DRM_DEBUG_DRIVER("%s %d\n", ring->name, seqno);

	return seqno;
}

static void bsd_setup_status_page(struct drm_device *dev,
		struct  intel_ring_buffer *ring)
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	I915_WRITE(BSD_HWS_PGA, ring->status_page.gfx_addr);
	I915_READ(BSD_HWS_PGA);
}

static void
bsd_ring_get_user_irq(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	/* do nothing */
}
static void
bsd_ring_put_user_irq(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	/* do nothing */
}

static u32
bsd_ring_get_gem_seqno(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
	return intel_read_status_page(ring, I915_GEM_HWS_INDEX);
}

static int
bsd_ring_dispatch_gem_execbuffer(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		struct drm_i915_gem_execbuffer2 *exec,
		struct drm_clip_rect *cliprects,
		uint64_t exec_offset)
{
	uint32_t exec_start;
	exec_start = (uint32_t) exec_offset + exec->batch_start_offset;
	intel_ring_begin(dev, ring, 2);
	intel_ring_emit(dev, ring, MI_BATCH_BUFFER_START |
			(2 << 6) | MI_BATCH_NON_SECURE_I965);
	intel_ring_emit(dev, ring, exec_start);
	intel_ring_advance(dev, ring);
	return 0;
}


496 497 498 499 500 501
static int
render_ring_dispatch_gem_execbuffer(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		struct drm_i915_gem_execbuffer2 *exec,
		struct drm_clip_rect *cliprects,
		uint64_t exec_offset)
502 503 504 505 506 507 508 509
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	int nbox = exec->num_cliprects;
	int i = 0, count;
	uint32_t exec_start, exec_len;
	exec_start = (uint32_t) exec_offset + exec->batch_start_offset;
	exec_len = (uint32_t) exec->batch_len;

510
	trace_i915_gem_request_submit(dev, dev_priv->next_seqno + 1);
511 512 513 514 515 516 517 518 519 520 521 522

	count = nbox ? nbox : 1;

	for (i = 0; i < count; i++) {
		if (i < nbox) {
			int ret = i915_emit_box(dev, cliprects, i,
						exec->DR1, exec->DR4);
			if (ret)
				return ret;
		}

		if (IS_I830(dev) || IS_845G(dev)) {
523 524 525 526 527 528
			intel_ring_begin(dev, ring, 4);
			intel_ring_emit(dev, ring, MI_BATCH_BUFFER);
			intel_ring_emit(dev, ring,
					exec_start | MI_BATCH_NON_SECURE);
			intel_ring_emit(dev, ring, exec_start + exec_len - 4);
			intel_ring_emit(dev, ring, 0);
529
		} else {
530
			intel_ring_begin(dev, ring, 4);
531
			if (INTEL_INFO(dev)->gen >= 4) {
532 533 534 535
				intel_ring_emit(dev, ring,
						MI_BATCH_BUFFER_START | (2 << 6)
						| MI_BATCH_NON_SECURE_I965);
				intel_ring_emit(dev, ring, exec_start);
536
			} else {
537 538 539 540
				intel_ring_emit(dev, ring, MI_BATCH_BUFFER_START
						| (2 << 6));
				intel_ring_emit(dev, ring, exec_start |
						MI_BATCH_NON_SECURE);
541 542
			}
		}
543
		intel_ring_advance(dev, ring);
544 545
	}

546 547 548 549 550 551 552 553
	if (IS_G4X(dev) || IS_IRONLAKE(dev)) {
		intel_ring_begin(dev, ring, 2);
		intel_ring_emit(dev, ring, MI_FLUSH |
				MI_NO_WRITE_FLUSH |
				MI_INVALIDATE_ISP );
		intel_ring_emit(dev, ring, MI_NOOP);
		intel_ring_advance(dev, ring);
	}
554
	/* XXX breadcrumb */
555

556 557 558
	return 0;
}

559 560
static void cleanup_status_page(struct drm_device *dev,
		struct intel_ring_buffer *ring)
561 562 563 564 565
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	struct drm_gem_object *obj;
	struct drm_i915_gem_object *obj_priv;

566 567
	obj = ring->status_page.obj;
	if (obj == NULL)
568 569 570 571 572 573
		return;
	obj_priv = to_intel_bo(obj);

	kunmap(obj_priv->pages[0]);
	i915_gem_object_unpin(obj);
	drm_gem_object_unreference(obj);
574
	ring->status_page.obj = NULL;
575 576 577 578

	memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map));
}

579 580
static int init_status_page(struct drm_device *dev,
		struct intel_ring_buffer *ring)
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600
{
	drm_i915_private_t *dev_priv = dev->dev_private;
	struct drm_gem_object *obj;
	struct drm_i915_gem_object *obj_priv;
	int ret;

	obj = i915_gem_alloc_object(dev, 4096);
	if (obj == NULL) {
		DRM_ERROR("Failed to allocate status page\n");
		ret = -ENOMEM;
		goto err;
	}
	obj_priv = to_intel_bo(obj);
	obj_priv->agp_type = AGP_USER_CACHED_MEMORY;

	ret = i915_gem_object_pin(obj, 4096);
	if (ret != 0) {
		goto err_unref;
	}

601 602 603
	ring->status_page.gfx_addr = obj_priv->gtt_offset;
	ring->status_page.page_addr = kmap(obj_priv->pages[0]);
	if (ring->status_page.page_addr == NULL) {
604 605 606
		memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map));
		goto err_unpin;
	}
607 608
	ring->status_page.obj = obj;
	memset(ring->status_page.page_addr, 0, PAGE_SIZE);
609

610 611 612
	ring->setup_status_page(dev, ring);
	DRM_DEBUG_DRIVER("%s hws offset: 0x%08x\n",
			ring->name, ring->status_page.gfx_addr);
613 614 615 616 617 618 619 620

	return 0;

err_unpin:
	i915_gem_object_unpin(obj);
err_unref:
	drm_gem_object_unreference(obj);
err:
621
	return ret;
622 623
}

624 625 626

int intel_init_ring_buffer(struct drm_device *dev,
		struct intel_ring_buffer *ring)
627
{
628 629
	struct drm_i915_gem_object *obj_priv;
	struct drm_gem_object *obj;
630 631
	int ret;

632
	ring->dev = dev;
633

634 635 636 637 638
	if (I915_NEED_GFX_HWS(dev)) {
		ret = init_status_page(dev, ring);
		if (ret)
			return ret;
	}
639

640
	obj = i915_gem_alloc_object(dev, ring->size);
641 642
	if (obj == NULL) {
		DRM_ERROR("Failed to allocate ringbuffer\n");
643
		ret = -ENOMEM;
644
		goto err_hws;
645 646
	}

647 648 649
	ring->gem_object = obj;

	ret = i915_gem_object_pin(obj, ring->alignment);
650 651
	if (ret)
		goto err_unref;
652

653 654
	obj_priv = to_intel_bo(obj);
	ring->map.size = ring->size;
655 656 657 658 659 660 661 662
	ring->map.offset = dev->agp->base + obj_priv->gtt_offset;
	ring->map.type = 0;
	ring->map.flags = 0;
	ring->map.mtrr = 0;

	drm_core_ioremap_wc(&ring->map, dev);
	if (ring->map.handle == NULL) {
		DRM_ERROR("Failed to map ringbuffer.\n");
663
		ret = -EINVAL;
664
		goto err_unpin;
665 666
	}

667 668
	ring->virtual_start = ring->map.handle;
	ret = ring->init(dev, ring);
669 670
	if (ret)
		goto err_unmap;
671 672 673 674

	if (!drm_core_check_feature(dev, DRIVER_MODESET))
		i915_kernel_lost_context(dev);
	else {
675 676
		ring->head = ring->get_head(dev, ring);
		ring->tail = ring->get_tail(dev, ring);
677 678
		ring->space = ring->head - (ring->tail + 8);
		if (ring->space < 0)
679
			ring->space += ring->size;
680
	}
681 682 683
	INIT_LIST_HEAD(&ring->active_list);
	INIT_LIST_HEAD(&ring->request_list);
	return ret;
684 685 686 687 688 689 690 691 692

err_unmap:
	drm_core_ioremapfree(&ring->map, dev);
err_unpin:
	i915_gem_object_unpin(obj);
err_unref:
	drm_gem_object_unreference(obj);
	ring->gem_object = NULL;
err_hws:
693 694
	cleanup_status_page(dev, ring);
	return ret;
695 696
}

697 698
void intel_cleanup_ring_buffer(struct drm_device *dev,
		struct intel_ring_buffer *ring)
699
{
700
	if (ring->gem_object == NULL)
701 702
		return;

703
	drm_core_ioremapfree(&ring->map, dev);
704

705 706 707 708
	i915_gem_object_unpin(ring->gem_object);
	drm_gem_object_unreference(ring->gem_object);
	ring->gem_object = NULL;
	cleanup_status_page(dev, ring);
709 710
}

711 712
int intel_wrap_ring_buffer(struct drm_device *dev,
		struct intel_ring_buffer *ring)
713
{
714
	unsigned int *virt;
715
	int rem;
716
	rem = ring->size - ring->tail;
717

718 719
	if (ring->space < rem) {
		int ret = intel_wait_ring_buffer(dev, ring, rem);
720 721 722 723
		if (ret)
			return ret;
	}

724
	virt = (unsigned int *)(ring->virtual_start + ring->tail);
725 726
	rem /= 8;
	while (rem--) {
727
		*virt++ = MI_NOOP;
728 729
		*virt++ = MI_NOOP;
	}
730

731
	ring->tail = 0;
732
	ring->space = ring->head - 8;
733 734 735 736

	return 0;
}

737 738
int intel_wait_ring_buffer(struct drm_device *dev,
		struct intel_ring_buffer *ring, int n)
739
{
740
	unsigned long end;
741 742

	trace_i915_ring_wait_begin (dev);
743 744 745
	end = jiffies + 3 * HZ;
	do {
		ring->head = ring->get_head(dev, ring);
746 747
		ring->space = ring->head - (ring->tail + 8);
		if (ring->space < 0)
748
			ring->space += ring->size;
749 750 751 752 753 754 755 756 757 758
		if (ring->space >= n) {
			trace_i915_ring_wait_end (dev);
			return 0;
		}

		if (dev->primary->master) {
			struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv;
			if (master_priv->sarea_priv)
				master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT;
		}
759

760 761 762 763 764
		yield();
	} while (!time_after(jiffies, end));
	trace_i915_ring_wait_end (dev);
	return -EBUSY;
}
765

766
void intel_ring_begin(struct drm_device *dev,
767
		struct intel_ring_buffer *ring, int num_dwords)
768
{
769
	int n = 4*num_dwords;
770 771 772 773
	if (unlikely(ring->tail + n > ring->size))
		intel_wrap_ring_buffer(dev, ring);
	if (unlikely(ring->space < n))
		intel_wait_ring_buffer(dev, ring, n);
774 775

	ring->space -= n;
776
}
777

778 779 780
void intel_ring_advance(struct drm_device *dev,
		struct intel_ring_buffer *ring)
{
781
	ring->tail &= ring->size - 1;
782 783
	ring->advance_ring(dev, ring);
}
784

785 786 787 788 789 790 791
void intel_fill_struct(struct drm_device *dev,
		struct intel_ring_buffer *ring,
		void *data,
		unsigned int len)
{
	unsigned int *virt = ring->virtual_start + ring->tail;
	BUG_ON((len&~(4-1)) != 0);
792
	intel_ring_begin(dev, ring, len/4);
793 794 795 796 797 798
	memcpy(virt, data, len);
	ring->tail += len;
	ring->tail &= ring->size - 1;
	ring->space -= len;
	intel_ring_advance(dev, ring);
}
799

800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833
struct intel_ring_buffer render_ring = {
	.name			= "render ring",
	.regs                   = {
		.ctl = PRB0_CTL,
		.head = PRB0_HEAD,
		.tail = PRB0_TAIL,
		.start = PRB0_START
	},
	.size			= 32 * PAGE_SIZE,
	.alignment		= PAGE_SIZE,
	.virtual_start		= NULL,
	.dev			= NULL,
	.gem_object		= NULL,
	.head			= 0,
	.tail			= 0,
	.space			= 0,
	.user_irq_refcount	= 0,
	.irq_gem_seqno		= 0,
	.waiting_gem_seqno	= 0,
	.setup_status_page	= render_setup_status_page,
	.init			= init_render_ring,
	.get_head		= render_ring_get_head,
	.get_tail		= render_ring_get_tail,
	.get_active_head	= render_ring_get_active_head,
	.advance_ring		= render_ring_advance_ring,
	.flush			= render_ring_flush,
	.add_request		= render_ring_add_request,
	.get_gem_seqno		= render_ring_get_gem_seqno,
	.user_irq_get		= render_ring_get_user_irq,
	.user_irq_put		= render_ring_put_user_irq,
	.dispatch_gem_execbuffer = render_ring_dispatch_gem_execbuffer,
	.status_page		= {NULL, 0, NULL},
	.map			= {0,}
};
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

/* ring buffer for bit-stream decoder */

struct intel_ring_buffer bsd_ring = {
	.name                   = "bsd ring",
	.regs			= {
		.ctl = BSD_RING_CTL,
		.head = BSD_RING_HEAD,
		.tail = BSD_RING_TAIL,
		.start = BSD_RING_START
	},
	.size			= 32 * PAGE_SIZE,
	.alignment		= PAGE_SIZE,
	.virtual_start		= NULL,
	.dev			= NULL,
	.gem_object		= NULL,
	.head			= 0,
	.tail			= 0,
	.space			= 0,
	.user_irq_refcount	= 0,
	.irq_gem_seqno		= 0,
	.waiting_gem_seqno	= 0,
	.setup_status_page	= bsd_setup_status_page,
	.init			= init_bsd_ring,
	.get_head		= bsd_ring_get_head,
	.get_tail		= bsd_ring_get_tail,
	.get_active_head	= bsd_ring_get_active_head,
	.advance_ring		= bsd_ring_advance_ring,
	.flush			= bsd_ring_flush,
	.add_request		= bsd_ring_add_request,
	.get_gem_seqno		= bsd_ring_get_gem_seqno,
	.user_irq_get		= bsd_ring_get_user_irq,
	.user_irq_put		= bsd_ring_put_user_irq,
	.dispatch_gem_execbuffer = bsd_ring_dispatch_gem_execbuffer,
	.status_page		= {NULL, 0, NULL},
	.map			= {0,}
};