job.c 13.2 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 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
/*
 * Tegra host1x Job
 *
 * Copyright (c) 2010-2013, NVIDIA Corporation.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope 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, see <http://www.gnu.org/licenses/>.
 */

#include <linux/dma-mapping.h>
#include <linux/err.h>
#include <linux/kref.h>
#include <linux/module.h>
#include <linux/scatterlist.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <trace/events/host1x.h>

#include "channel.h"
#include "dev.h"
#include "host1x_bo.h"
#include "job.h"
#include "syncpt.h"

struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,
				    u32 num_cmdbufs, u32 num_relocs,
				    u32 num_waitchks)
{
	struct host1x_job *job = NULL;
	unsigned int num_unpins = num_cmdbufs + num_relocs;
	u64 total;
	void *mem;

	/* Check that we're not going to overflow */
	total = sizeof(struct host1x_job) +
		num_relocs * sizeof(struct host1x_reloc) +
		num_unpins * sizeof(struct host1x_job_unpin_data) +
		num_waitchks * sizeof(struct host1x_waitchk) +
		num_cmdbufs * sizeof(struct host1x_job_gather) +
		num_unpins * sizeof(dma_addr_t) +
		num_unpins * sizeof(u32 *);
	if (total > ULONG_MAX)
		return NULL;

	mem = job = kzalloc(total, GFP_KERNEL);
	if (!job)
		return NULL;

	kref_init(&job->ref);
	job->channel = ch;

	/* Redistribute memory to the structs  */
	mem += sizeof(struct host1x_job);
	job->relocarray = num_relocs ? mem : NULL;
	mem += num_relocs * sizeof(struct host1x_reloc);
	job->unpins = num_unpins ? mem : NULL;
	mem += num_unpins * sizeof(struct host1x_job_unpin_data);
	job->waitchk = num_waitchks ? mem : NULL;
	mem += num_waitchks * sizeof(struct host1x_waitchk);
	job->gathers = num_cmdbufs ? mem : NULL;
	mem += num_cmdbufs * sizeof(struct host1x_job_gather);
	job->addr_phys = num_unpins ? mem : NULL;

	job->reloc_addr_phys = job->addr_phys;
	job->gather_addr_phys = &job->addr_phys[num_relocs];

	return job;
}

struct host1x_job *host1x_job_get(struct host1x_job *job)
{
	kref_get(&job->ref);
	return job;
}

static void job_free(struct kref *ref)
{
	struct host1x_job *job = container_of(ref, struct host1x_job, ref);

	kfree(job);
}

void host1x_job_put(struct host1x_job *job)
{
	kref_put(&job->ref, job_free);
}

void host1x_job_add_gather(struct host1x_job *job, struct host1x_bo *bo,
			   u32 words, u32 offset)
{
	struct host1x_job_gather *cur_gather = &job->gathers[job->num_gathers];

	cur_gather->words = words;
	cur_gather->bo = bo;
	cur_gather->offset = offset;
	job->num_gathers++;
}

/*
 * NULL an already satisfied WAIT_SYNCPT host method, by patching its
 * args in the command stream. The method data is changed to reference
 * a reserved (never given out or incr) HOST1X_SYNCPT_RESERVED syncpt
 * with a matching threshold value of 0, so is guaranteed to be popped
 * by the host HW.
 */
static void host1x_syncpt_patch_offset(struct host1x_syncpt *sp,
				       struct host1x_bo *h, u32 offset)
{
	void *patch_addr = NULL;

	/* patch the wait */
	patch_addr = host1x_bo_kmap(h, offset >> PAGE_SHIFT);
	if (patch_addr) {
		host1x_syncpt_patch_wait(sp,
					 patch_addr + (offset & ~PAGE_MASK));
		host1x_bo_kunmap(h, offset >> PAGE_SHIFT, patch_addr);
	} else
		pr_err("Could not map cmdbuf for wait check\n");
}

/*
 * Check driver supplied waitchk structs for syncpt thresholds
 * that have already been satisfied and NULL the comparison (to
 * avoid a wrap condition in the HW).
 */
static int do_waitchks(struct host1x_job *job, struct host1x *host,
		       struct host1x_bo *patch)
{
	int i;

	/* compare syncpt vs wait threshold */
	for (i = 0; i < job->num_waitchk; i++) {
		struct host1x_waitchk *wait = &job->waitchk[i];
		struct host1x_syncpt *sp =
			host1x_syncpt_get(host, wait->syncpt_id);

		/* validate syncpt id */
		if (wait->syncpt_id > host1x_syncpt_nb_pts(host))
			continue;

		/* skip all other gathers */
		if (patch != wait->bo)
			continue;

		trace_host1x_syncpt_wait_check(wait->bo, wait->offset,
					       wait->syncpt_id, wait->thresh,
					       host1x_syncpt_read_min(sp));

		if (host1x_syncpt_is_expired(sp, wait->thresh)) {
			dev_dbg(host->dev,
				"drop WAIT id %d (%s) thresh 0x%x, min 0x%x\n",
				wait->syncpt_id, sp->name, wait->thresh,
				host1x_syncpt_read_min(sp));

			host1x_syncpt_patch_offset(sp, patch, wait->offset);
		}

		wait->bo = NULL;
	}

	return 0;
}

static unsigned int pin_job(struct host1x_job *job)
{
	unsigned int i;

	job->num_unpins = 0;

	for (i = 0; i < job->num_relocs; i++) {
		struct host1x_reloc *reloc = &job->relocarray[i];
		struct sg_table *sgt;
		dma_addr_t phys_addr;

		reloc->target = host1x_bo_get(reloc->target);
		if (!reloc->target)
			goto unpin;

		phys_addr = host1x_bo_pin(reloc->target, &sgt);
		if (!phys_addr)
			goto unpin;

		job->addr_phys[job->num_unpins] = phys_addr;
		job->unpins[job->num_unpins].bo = reloc->target;
		job->unpins[job->num_unpins].sgt = sgt;
		job->num_unpins++;
	}

	for (i = 0; i < job->num_gathers; i++) {
		struct host1x_job_gather *g = &job->gathers[i];
		struct sg_table *sgt;
		dma_addr_t phys_addr;

		g->bo = host1x_bo_get(g->bo);
		if (!g->bo)
			goto unpin;

		phys_addr = host1x_bo_pin(g->bo, &sgt);
		if (!phys_addr)
			goto unpin;

		job->addr_phys[job->num_unpins] = phys_addr;
		job->unpins[job->num_unpins].bo = g->bo;
		job->unpins[job->num_unpins].sgt = sgt;
		job->num_unpins++;
	}

	return job->num_unpins;

unpin:
	host1x_job_unpin(job);
	return 0;
}

static unsigned int do_relocs(struct host1x_job *job, struct host1x_bo *cmdbuf)
{
	int i = 0;
	u32 last_page = ~0;
	void *cmdbuf_page_addr = NULL;

	/* pin & patch the relocs for one gather */
	while (i < job->num_relocs) {
		struct host1x_reloc *reloc = &job->relocarray[i];
		u32 reloc_addr = (job->reloc_addr_phys[i] +
			reloc->target_offset) >> reloc->shift;
		u32 *target;

		/* skip all other gathers */
		if (!(reloc->cmdbuf && cmdbuf == reloc->cmdbuf)) {
			i++;
			continue;
		}

		if (last_page != reloc->cmdbuf_offset >> PAGE_SHIFT) {
			if (cmdbuf_page_addr)
				host1x_bo_kunmap(cmdbuf, last_page,
						 cmdbuf_page_addr);

			cmdbuf_page_addr = host1x_bo_kmap(cmdbuf,
					reloc->cmdbuf_offset >> PAGE_SHIFT);
			last_page = reloc->cmdbuf_offset >> PAGE_SHIFT;

			if (unlikely(!cmdbuf_page_addr)) {
				pr_err("Could not map cmdbuf for relocation\n");
				return -ENOMEM;
			}
		}

		target = cmdbuf_page_addr + (reloc->cmdbuf_offset & ~PAGE_MASK);
		*target = reloc_addr;

		/* mark this gather as handled */
		reloc->cmdbuf = 0;
	}

	if (cmdbuf_page_addr)
		host1x_bo_kunmap(cmdbuf, last_page, cmdbuf_page_addr);

	return 0;
}

271
static bool check_reloc(struct host1x_reloc *reloc, struct host1x_bo *cmdbuf,
272 273 274 275 276
		       unsigned int offset)
{
	offset *= sizeof(u32);

	if (reloc->cmdbuf != cmdbuf || reloc->cmdbuf_offset != offset)
277
		return false;
278

279
	return true;
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
}

struct host1x_firewall {
	struct host1x_job *job;
	struct device *dev;

	unsigned int num_relocs;
	struct host1x_reloc *reloc;

	struct host1x_bo *cmdbuf_id;
	unsigned int offset;

	u32 words;
	u32 class;
	u32 reg;
	u32 mask;
	u32 count;
};

static int check_mask(struct host1x_firewall *fw)
{
	u32 mask = fw->mask;
	u32 reg = fw->reg;

	while (mask) {
		if (fw->words == 0)
			return -EINVAL;

		if (mask & 1) {
			if (fw->job->is_addr_reg(fw->dev, fw->class, reg)) {
310 311 312 313
				if (!fw->num_relocs)
					return -EINVAL;
				if (!check_reloc(fw->reloc, fw->cmdbuf_id,
						 fw->offset))
314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332
					return -EINVAL;
				fw->reloc++;
				fw->num_relocs--;
			}
			fw->words--;
			fw->offset++;
		}
		mask >>= 1;
		reg++;
	}

	return 0;
}

static int check_incr(struct host1x_firewall *fw)
{
	u32 count = fw->count;
	u32 reg = fw->reg;

333
	while (count) {
334 335 336 337
		if (fw->words == 0)
			return -EINVAL;

		if (fw->job->is_addr_reg(fw->dev, fw->class, reg)) {
338 339 340
			if (!fw->num_relocs)
				return -EINVAL;
			if (!check_reloc(fw->reloc, fw->cmdbuf_id, fw->offset))
341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363
				return -EINVAL;
			fw->reloc++;
			fw->num_relocs--;
		}
		reg++;
		fw->words--;
		fw->offset++;
		count--;
	}

	return 0;
}

static int check_nonincr(struct host1x_firewall *fw)
{
	int is_addr_reg = fw->job->is_addr_reg(fw->dev, fw->class, fw->reg);
	u32 count = fw->count;

	while (count) {
		if (fw->words == 0)
			return -EINVAL;

		if (is_addr_reg) {
364 365 366
			if (!fw->num_relocs)
				return -EINVAL;
			if (!check_reloc(fw->reloc, fw->cmdbuf_id, fw->offset))
367 368 369 370 371 372 373 374 375 376 377 378
				return -EINVAL;
			fw->reloc++;
			fw->num_relocs--;
		}
		fw->words--;
		fw->offset++;
		count--;
	}

	return 0;
}

379
static int validate(struct host1x_firewall *fw, struct host1x_job_gather *g)
380 381 382 383
{
	u32 *cmdbuf_base;
	int err = 0;

384
	if (!fw->job->is_addr_reg)
385 386 387 388 389
		return 0;

	cmdbuf_base = host1x_bo_mmap(g->bo);
	if (!cmdbuf_base)
		return -ENOMEM;
390 391 392
	fw->words = g->words;
	fw->cmdbuf_id = g->bo;
	fw->offset = 0;
393

394 395
	while (fw->words && !err) {
		u32 word = cmdbuf_base[fw->offset];
396 397
		u32 opcode = (word & 0xf0000000) >> 28;

398 399 400 401 402
		fw->mask = 0;
		fw->reg = 0;
		fw->count = 0;
		fw->words--;
		fw->offset++;
403 404 405

		switch (opcode) {
		case 0:
406 407 408 409
			fw->class = word >> 6 & 0x3ff;
			fw->mask = word & 0x3f;
			fw->reg = word >> 16 & 0xfff;
			err = check_mask(fw);
410 411 412 413
			if (err)
				goto out;
			break;
		case 1:
414 415 416
			fw->reg = word >> 16 & 0xfff;
			fw->count = word & 0xffff;
			err = check_incr(fw);
417 418 419 420 421
			if (err)
				goto out;
			break;

		case 2:
422 423 424
			fw->reg = word >> 16 & 0xfff;
			fw->count = word & 0xffff;
			err = check_nonincr(fw);
425 426 427 428 429
			if (err)
				goto out;
			break;

		case 3:
430 431 432
			fw->mask = word & 0xffff;
			fw->reg = word >> 16 & 0xfff;
			err = check_mask(fw);
433 434 435 436 437 438 439 440 441 442 443 444 445 446
			if (err)
				goto out;
			break;
		case 4:
		case 5:
		case 14:
			break;
		default:
			err = -EINVAL;
			break;
		}
	}

	/* No relocs should remain at this point */
447
	if (fw->num_relocs)
448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499
		err = -EINVAL;

out:
	return err;
}

static inline int copy_gathers(struct host1x_job *job, struct device *dev)
{
	size_t size = 0;
	size_t offset = 0;
	int i;

	for (i = 0; i < job->num_gathers; i++) {
		struct host1x_job_gather *g = &job->gathers[i];
		size += g->words * sizeof(u32);
	}

	job->gather_copy_mapped = dma_alloc_writecombine(dev, size,
							 &job->gather_copy,
							 GFP_KERNEL);
	if (!job->gather_copy_mapped) {
		int err = PTR_ERR(job->gather_copy_mapped);
		job->gather_copy_mapped = NULL;
		return err;
	}

	job->gather_copy_size = size;

	for (i = 0; i < job->num_gathers; i++) {
		struct host1x_job_gather *g = &job->gathers[i];
		void *gather;

		gather = host1x_bo_mmap(g->bo);
		memcpy(job->gather_copy_mapped + offset, gather + g->offset,
		       g->words * sizeof(u32));
		host1x_bo_munmap(g->bo, gather);

		g->base = job->gather_copy;
		g->offset = offset;
		g->bo = NULL;

		offset += g->words * sizeof(u32);
	}

	return 0;
}

int host1x_job_pin(struct host1x_job *job, struct device *dev)
{
	int err;
	unsigned int i, j;
	struct host1x *host = dev_get_drvdata(dev->parent);
500
	struct host1x_firewall fw;
501 502
	DECLARE_BITMAP(waitchk_mask, host1x_syncpt_nb_pts(host));

503 504 505 506 507 508
	fw.job = job;
	fw.dev = dev;
	fw.reloc = job->relocarray;
	fw.num_relocs = job->num_relocs;
	fw.class = 0;

509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
	bitmap_zero(waitchk_mask, host1x_syncpt_nb_pts(host));
	for (i = 0; i < job->num_waitchk; i++) {
		u32 syncpt_id = job->waitchk[i].syncpt_id;
		if (syncpt_id < host1x_syncpt_nb_pts(host))
			set_bit(syncpt_id, waitchk_mask);
	}

	/* get current syncpt values for waitchk */
	for_each_set_bit(i, waitchk_mask, host1x_syncpt_nb_pts(host))
		host1x_syncpt_load(host->syncpt + i);

	/* pin memory */
	err = pin_job(job);
	if (!err)
		goto out;

	/* patch gathers */
	for (i = 0; i < job->num_gathers; i++) {
		struct host1x_job_gather *g = &job->gathers[i];

		/* process each gather mem only once */
		if (g->handled)
			continue;

		g->base = job->gather_addr_phys[i];

		for (j = 0; j < job->num_gathers; j++)
			if (job->gathers[j].bo == g->bo)
				job->gathers[j].handled = true;

		err = 0;

		if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL))
542
			err = validate(&fw, g);
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

		if (err)
			dev_err(dev, "Job invalid (err=%d)\n", err);

		if (!err)
			err = do_relocs(job, g->bo);

		if (!err)
			err = do_waitchks(job, host, g->bo);

		if (err)
			break;
	}

	if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL) && !err) {
		err = copy_gathers(job, dev);
		if (err) {
			host1x_job_unpin(job);
			return err;
		}
	}

out:
	wmb();

	return err;
}

void host1x_job_unpin(struct host1x_job *job)
{
	unsigned int i;

	for (i = 0; i < job->num_unpins; i++) {
		struct host1x_job_unpin_data *unpin = &job->unpins[i];
		host1x_bo_unpin(unpin->bo, unpin->sgt);
		host1x_bo_put(unpin->bo);
	}
	job->num_unpins = 0;

	if (job->gather_copy_size)
		dma_free_writecombine(job->channel->dev, job->gather_copy_size,
				      job->gather_copy_mapped,
				      job->gather_copy);
}

/*
 * Debug routine used to dump job entries
 */
void host1x_job_dump(struct device *dev, struct host1x_job *job)
{
	dev_dbg(dev, "    SYNCPT_ID   %d\n", job->syncpt_id);
	dev_dbg(dev, "    SYNCPT_VAL  %d\n", job->syncpt_end);
	dev_dbg(dev, "    FIRST_GET   0x%x\n", job->first_get);
	dev_dbg(dev, "    TIMEOUT     %d\n", job->timeout);
	dev_dbg(dev, "    NUM_SLOTS   %d\n", job->num_slots);
	dev_dbg(dev, "    NUM_HANDLES %d\n", job->num_unpins);
}