radeon_ring.c 10.4 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
/*
 * Copyright 2008 Advanced Micro Devices, Inc.
 * Copyright 2008 Red Hat Inc.
 * Copyright 2009 Jerome Glisse.
 *
 * 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 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 COPYRIGHT HOLDER(S) OR AUTHOR(S) 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: Dave Airlie
 *          Alex Deucher
 *          Jerome Glisse
 */
#include <linux/seq_file.h>
#include "drmP.h"
#include "radeon_drm.h"
#include "radeon_reg.h"
#include "radeon.h"
#include "atom.h"

int radeon_debugfs_ib_init(struct radeon_device *rdev);

/*
 * IB.
 */
int radeon_ib_get(struct radeon_device *rdev, struct radeon_ib **ib)
{
	struct radeon_fence *fence;
	struct radeon_ib *nib;
	unsigned long i;
	int r = 0;

	*ib = NULL;
	r = radeon_fence_create(rdev, &fence);
	if (r) {
		DRM_ERROR("failed to create fence for new IB\n");
		return r;
	}
	mutex_lock(&rdev->ib_pool.mutex);
	i = find_first_zero_bit(rdev->ib_pool.alloc_bm, RADEON_IB_POOL_SIZE);
	if (i < RADEON_IB_POOL_SIZE) {
		set_bit(i, rdev->ib_pool.alloc_bm);
		rdev->ib_pool.ibs[i].length_dw = 0;
		*ib = &rdev->ib_pool.ibs[i];
59
		mutex_unlock(&rdev->ib_pool.mutex);
60 61 62 63
		goto out;
	}
	if (list_empty(&rdev->ib_pool.scheduled_ibs)) {
		/* we go do nothings here */
64
		mutex_unlock(&rdev->ib_pool.mutex);
65 66 67 68 69 70 71 72 73
		DRM_ERROR("all IB allocated none scheduled.\n");
		r = -EINVAL;
		goto out;
	}
	/* get the first ib on the scheduled list */
	nib = list_entry(rdev->ib_pool.scheduled_ibs.next,
			 struct radeon_ib, list);
	if (nib->fence == NULL) {
		/* we go do nothings here */
74
		mutex_unlock(&rdev->ib_pool.mutex);
75 76 77 78
		DRM_ERROR("IB %lu scheduled without a fence.\n", nib->idx);
		r = -EINVAL;
		goto out;
	}
79 80
	mutex_unlock(&rdev->ib_pool.mutex);

81 82 83 84 85 86 87 88
	r = radeon_fence_wait(nib->fence, false);
	if (r) {
		DRM_ERROR("radeon: IB(%lu:0x%016lX:%u)\n", nib->idx,
			  (unsigned long)nib->gpu_addr, nib->length_dw);
		DRM_ERROR("radeon: GPU lockup detected, fail to get a IB\n");
		goto out;
	}
	radeon_fence_unref(&nib->fence);
89

90
	nib->length_dw = 0;
91 92 93

	/* scheduled list is accessed here */
	mutex_lock(&rdev->ib_pool.mutex);
94 95
	list_del(&nib->list);
	INIT_LIST_HEAD(&nib->list);
96 97
	mutex_unlock(&rdev->ib_pool.mutex);

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
	*ib = nib;
out:
	if (r) {
		radeon_fence_unref(&fence);
	} else {
		(*ib)->fence = fence;
	}
	return r;
}

void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib **ib)
{
	struct radeon_ib *tmp = *ib;

	*ib = NULL;
	if (tmp == NULL) {
		return;
	}
	mutex_lock(&rdev->ib_pool.mutex);
	if (!list_empty(&tmp->list) && !radeon_fence_signaled(tmp->fence)) {
		/* IB is scheduled & not signaled don't do anythings */
		mutex_unlock(&rdev->ib_pool.mutex);
		return;
	}
	list_del(&tmp->list);
123 124
	INIT_LIST_HEAD(&tmp->list);
	if (tmp->fence)
125
		radeon_fence_unref(&tmp->fence);
126

127 128 129 130 131 132 133 134 135 136 137 138 139 140
	tmp->length_dw = 0;
	clear_bit(tmp->idx, rdev->ib_pool.alloc_bm);
	mutex_unlock(&rdev->ib_pool.mutex);
}

int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib)
{
	int r = 0;

	if (!ib->length_dw || !rdev->cp.ready) {
		/* TODO: Nothings in the ib we should report. */
		DRM_ERROR("radeon: couldn't schedule IB(%lu).\n", ib->idx);
		return -EINVAL;
	}
141

142
	/* 64 dwords should be enough for fence too */
143 144 145 146 147
	r = radeon_ring_lock(rdev, 64);
	if (r) {
		DRM_ERROR("radeon: scheduling IB failled (%d).\n", r);
		return r;
	}
148
	radeon_ring_ib_execute(rdev, ib);
149
	radeon_fence_emit(rdev, ib->fence);
150
	mutex_lock(&rdev->ib_pool.mutex);
151 152
	list_add_tail(&ib->list, &rdev->ib_pool.scheduled_ibs);
	mutex_unlock(&rdev->ib_pool.mutex);
153
	radeon_ring_unlock_commit(rdev);
154 155 156 157 158 159 160 161 162 163
	return 0;
}

int radeon_ib_pool_init(struct radeon_device *rdev)
{
	void *ptr;
	uint64_t gpu_addr;
	int i;
	int r = 0;

164 165
	if (rdev->ib_pool.robj)
		return 0;
166 167
	/* Allocate 1M object buffer */
	INIT_LIST_HEAD(&rdev->ib_pool.scheduled_ibs);
168 169 170
	r = radeon_bo_create(rdev, NULL,  RADEON_IB_POOL_SIZE*64*1024,
				true, RADEON_GEM_DOMAIN_GTT,
				&rdev->ib_pool.robj);
171 172 173 174
	if (r) {
		DRM_ERROR("radeon: failed to ib pool (%d).\n", r);
		return r;
	}
175 176 177 178
	r = radeon_bo_reserve(rdev->ib_pool.robj, false);
	if (unlikely(r != 0))
		return r;
	r = radeon_bo_pin(rdev->ib_pool.robj, RADEON_GEM_DOMAIN_GTT, &gpu_addr);
179
	if (r) {
180
		radeon_bo_unreserve(rdev->ib_pool.robj);
181 182 183
		DRM_ERROR("radeon: failed to pin ib pool (%d).\n", r);
		return r;
	}
184 185
	r = radeon_bo_kmap(rdev->ib_pool.robj, &ptr);
	radeon_bo_unreserve(rdev->ib_pool.robj);
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
	if (r) {
		DRM_ERROR("radeon: failed to map ib poll (%d).\n", r);
		return r;
	}
	for (i = 0; i < RADEON_IB_POOL_SIZE; i++) {
		unsigned offset;

		offset = i * 64 * 1024;
		rdev->ib_pool.ibs[i].gpu_addr = gpu_addr + offset;
		rdev->ib_pool.ibs[i].ptr = ptr + offset;
		rdev->ib_pool.ibs[i].idx = i;
		rdev->ib_pool.ibs[i].length_dw = 0;
		INIT_LIST_HEAD(&rdev->ib_pool.ibs[i].list);
	}
	bitmap_zero(rdev->ib_pool.alloc_bm, RADEON_IB_POOL_SIZE);
	rdev->ib_pool.ready = true;
	DRM_INFO("radeon: ib pool ready.\n");
	if (radeon_debugfs_ib_init(rdev)) {
		DRM_ERROR("Failed to register debugfs file for IB !\n");
	}
	return r;
}

void radeon_ib_pool_fini(struct radeon_device *rdev)
{
211 212
	int r;

213 214 215 216 217 218
	if (!rdev->ib_pool.ready) {
		return;
	}
	mutex_lock(&rdev->ib_pool.mutex);
	bitmap_zero(rdev->ib_pool.alloc_bm, RADEON_IB_POOL_SIZE);
	if (rdev->ib_pool.robj) {
219 220 221 222 223 224 225
		r = radeon_bo_reserve(rdev->ib_pool.robj, false);
		if (likely(r == 0)) {
			radeon_bo_kunmap(rdev->ib_pool.robj);
			radeon_bo_unpin(rdev->ib_pool.robj);
			radeon_bo_unreserve(rdev->ib_pool.robj);
		}
		radeon_bo_unref(&rdev->ib_pool.robj);
226 227 228 229 230 231 232 233 234 235 236
		rdev->ib_pool.robj = NULL;
	}
	mutex_unlock(&rdev->ib_pool.mutex);
}


/*
 * Ring.
 */
void radeon_ring_free_size(struct radeon_device *rdev)
{
237 238 239 240
	if (rdev->family >= CHIP_R600)
		rdev->cp.rptr = RREG32(R600_CP_RB_RPTR);
	else
		rdev->cp.rptr = RREG32(RADEON_CP_RB_RPTR);
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
	/* This works because ring_size is a power of 2 */
	rdev->cp.ring_free_dw = (rdev->cp.rptr + (rdev->cp.ring_size / 4));
	rdev->cp.ring_free_dw -= rdev->cp.wptr;
	rdev->cp.ring_free_dw &= rdev->cp.ptr_mask;
	if (!rdev->cp.ring_free_dw) {
		rdev->cp.ring_free_dw = rdev->cp.ring_size / 4;
	}
}

int radeon_ring_lock(struct radeon_device *rdev, unsigned ndw)
{
	int r;

	/* Align requested size with padding so unlock_commit can
	 * pad safely */
	ndw = (ndw + rdev->cp.align_mask) & ~rdev->cp.align_mask;
	mutex_lock(&rdev->cp.mutex);
	while (ndw > (rdev->cp.ring_free_dw - 1)) {
		radeon_ring_free_size(rdev);
		if (ndw < rdev->cp.ring_free_dw) {
			break;
		}
		r = radeon_fence_wait_next(rdev);
		if (r) {
			mutex_unlock(&rdev->cp.mutex);
			return r;
		}
	}
	rdev->cp.count_dw = ndw;
	rdev->cp.wptr_old = rdev->cp.wptr;
	return 0;
}

void radeon_ring_unlock_commit(struct radeon_device *rdev)
{
	unsigned count_dw_pad;
	unsigned i;

	/* We pad to match fetch size */
	count_dw_pad = (rdev->cp.align_mask + 1) -
		       (rdev->cp.wptr & rdev->cp.align_mask);
	for (i = 0; i < count_dw_pad; i++) {
283
		radeon_ring_write(rdev, 2 << 30);
284 285
	}
	DRM_MEMORYBARRIER();
286
	radeon_cp_commit(rdev);
287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
	mutex_unlock(&rdev->cp.mutex);
}

void radeon_ring_unlock_undo(struct radeon_device *rdev)
{
	rdev->cp.wptr = rdev->cp.wptr_old;
	mutex_unlock(&rdev->cp.mutex);
}

int radeon_ring_init(struct radeon_device *rdev, unsigned ring_size)
{
	int r;

	rdev->cp.ring_size = ring_size;
	/* Allocate ring buffer */
	if (rdev->cp.ring_obj == NULL) {
303 304 305
		r = radeon_bo_create(rdev, NULL, rdev->cp.ring_size, true,
					RADEON_GEM_DOMAIN_GTT,
					&rdev->cp.ring_obj);
306
		if (r) {
307
			dev_err(rdev->dev, "(%d) ring create failed\n", r);
308 309
			return r;
		}
310 311 312 313 314
		r = radeon_bo_reserve(rdev->cp.ring_obj, false);
		if (unlikely(r != 0))
			return r;
		r = radeon_bo_pin(rdev->cp.ring_obj, RADEON_GEM_DOMAIN_GTT,
					&rdev->cp.gpu_addr);
315
		if (r) {
316 317
			radeon_bo_unreserve(rdev->cp.ring_obj);
			dev_err(rdev->dev, "(%d) ring pin failed\n", r);
318 319
			return r;
		}
320
		r = radeon_bo_kmap(rdev->cp.ring_obj,
321
				       (void **)&rdev->cp.ring);
322
		radeon_bo_unreserve(rdev->cp.ring_obj);
323
		if (r) {
324
			dev_err(rdev->dev, "(%d) ring map failed\n", r);
325 326 327 328 329 330 331 332 333 334
			return r;
		}
	}
	rdev->cp.ptr_mask = (rdev->cp.ring_size / 4) - 1;
	rdev->cp.ring_free_dw = rdev->cp.ring_size / 4;
	return 0;
}

void radeon_ring_fini(struct radeon_device *rdev)
{
335 336
	int r;

337 338
	mutex_lock(&rdev->cp.mutex);
	if (rdev->cp.ring_obj) {
339 340 341 342 343 344 345
		r = radeon_bo_reserve(rdev->cp.ring_obj, false);
		if (likely(r == 0)) {
			radeon_bo_kunmap(rdev->cp.ring_obj);
			radeon_bo_unpin(rdev->cp.ring_obj);
			radeon_bo_unreserve(rdev->cp.ring_obj);
		}
		radeon_bo_unref(&rdev->cp.ring_obj);
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
		rdev->cp.ring = NULL;
		rdev->cp.ring_obj = NULL;
	}
	mutex_unlock(&rdev->cp.mutex);
}


/*
 * Debugfs info
 */
#if defined(CONFIG_DEBUG_FS)
static int radeon_debugfs_ib_info(struct seq_file *m, void *data)
{
	struct drm_info_node *node = (struct drm_info_node *) m->private;
	struct radeon_ib *ib = node->info_ent->data;
	unsigned i;

	if (ib == NULL) {
		return 0;
	}
	seq_printf(m, "IB %04lu\n", ib->idx);
	seq_printf(m, "IB fence %p\n", ib->fence);
	seq_printf(m, "IB size %05u dwords\n", ib->length_dw);
	for (i = 0; i < ib->length_dw; i++) {
		seq_printf(m, "[%05u]=0x%08X\n", i, ib->ptr[i]);
	}
	return 0;
}

static struct drm_info_list radeon_debugfs_ib_list[RADEON_IB_POOL_SIZE];
static char radeon_debugfs_ib_names[RADEON_IB_POOL_SIZE][32];
#endif

int radeon_debugfs_ib_init(struct radeon_device *rdev)
{
#if defined(CONFIG_DEBUG_FS)
	unsigned i;

	for (i = 0; i < RADEON_IB_POOL_SIZE; i++) {
		sprintf(radeon_debugfs_ib_names[i], "radeon_ib_%04u", i);
		radeon_debugfs_ib_list[i].name = radeon_debugfs_ib_names[i];
		radeon_debugfs_ib_list[i].show = &radeon_debugfs_ib_info;
		radeon_debugfs_ib_list[i].driver_features = 0;
		radeon_debugfs_ib_list[i].data = &rdev->ib_pool.ibs[i];
	}
	return radeon_debugfs_add_files(rdev, radeon_debugfs_ib_list,
					RADEON_IB_POOL_SIZE);
#else
	return 0;
#endif
}