gmc_v6_0.c 31.1 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
/*
 * Copyright 2014 Advanced Micro Devices, Inc.
 *
 * 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.
 *
 */
#include <linux/firmware.h>
24
#include <drm/drmP.h>
25 26 27
#include "amdgpu.h"
#include "gmc_v6_0.h"
#include "amdgpu_ucode.h"
28 29 30 31 32 33 34 35 36 37

#include "bif/bif_3_0_d.h"
#include "bif/bif_3_0_sh_mask.h"
#include "oss/oss_1_0_d.h"
#include "oss/oss_1_0_sh_mask.h"
#include "gmc/gmc_6_0_d.h"
#include "gmc/gmc_6_0_sh_mask.h"
#include "dce/dce_6_0_d.h"
#include "dce/dce_6_0_sh_mask.h"
#include "si_enums.h"
38 39 40 41 42 43 44 45 46

static void gmc_v6_0_set_gart_funcs(struct amdgpu_device *adev);
static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev);
static int gmc_v6_0_wait_for_idle(void *handle);

MODULE_FIRMWARE("radeon/tahiti_mc.bin");
MODULE_FIRMWARE("radeon/pitcairn_mc.bin");
MODULE_FIRMWARE("radeon/verde_mc.bin");
MODULE_FIRMWARE("radeon/oland_mc.bin");
47
MODULE_FIRMWARE("radeon/si58_mc.bin");
48

49 50 51 52 53 54 55 56 57 58
#define MC_SEQ_MISC0__MT__MASK   0xf0000000
#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
#define MC_SEQ_MISC0__MT__DDR2   0x20000000
#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
#define MC_SEQ_MISC0__MT__HBM    0x60000000
#define MC_SEQ_MISC0__MT__DDR3   0xB0000000


59 60 61 62 63 64 65 66 67 68
static const u32 crtc_offsets[6] =
{
	SI_CRTC0_REGISTER_OFFSET,
	SI_CRTC1_REGISTER_OFFSET,
	SI_CRTC2_REGISTER_OFFSET,
	SI_CRTC3_REGISTER_OFFSET,
	SI_CRTC4_REGISTER_OFFSET,
	SI_CRTC5_REGISTER_OFFSET
};

69
static void gmc_v6_0_mc_stop(struct amdgpu_device *adev)
70 71 72 73 74
{
	u32 blackout;

	gmc_v6_0_wait_for_idle((void *)adev);

75 76
	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
77
		/* Block CPU access */
78
		WREG32(mmBIF_FB_EN, 0);
79 80
		/* blackout the MC */
		blackout = REG_SET_FIELD(blackout,
81 82
					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
83 84 85 86 87 88
	}
	/* wait for the MC to settle */
	udelay(100);

}

89
static void gmc_v6_0_mc_resume(struct amdgpu_device *adev)
90 91 92 93
{
	u32 tmp;

	/* unblackout the MC */
94 95 96
	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
97
	/* allow CPU access */
98 99 100
	tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
	tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
	WREG32(mmBIF_FB_EN, tmp);
101 102 103 104 105 106 107
}

static int gmc_v6_0_init_microcode(struct amdgpu_device *adev)
{
	const char *chip_name;
	char fw_name[30];
	int err;
108
	bool is_58_fw = false;
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

	DRM_DEBUG("\n");

	switch (adev->asic_type) {
	case CHIP_TAHITI:
		chip_name = "tahiti";
		break;
	case CHIP_PITCAIRN:
		chip_name = "pitcairn";
		break;
	case CHIP_VERDE:
		chip_name = "verde";
		break;
	case CHIP_OLAND:
		chip_name = "oland";
		break;
	case CHIP_HAINAN:
		chip_name = "hainan";
		break;
	default: BUG();
	}

131 132 133 134 135 136 137 138
	/* this memory configuration requires special firmware */
	if (((RREG32(mmMC_SEQ_MISC0) & 0xff000000) >> 24) == 0x58)
		is_58_fw = true;

	if (is_58_fw)
		snprintf(fw_name, sizeof(fw_name), "radeon/si58_mc.bin");
	else
		snprintf(fw_name, sizeof(fw_name), "radeon/%s_mc.bin", chip_name);
139 140 141 142 143 144 145 146
	err = request_firmware(&adev->mc.fw, fw_name, adev->dev);
	if (err)
		goto out;

	err = amdgpu_ucode_validate(adev->mc.fw);

out:
	if (err) {
147
		dev_err(adev->dev,
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
		       "si_mc: Failed to load firmware \"%s\"\n",
		       fw_name);
		release_firmware(adev->mc.fw);
		adev->mc.fw = NULL;
	}
	return err;
}

static int gmc_v6_0_mc_load_microcode(struct amdgpu_device *adev)
{
	const __le32 *new_fw_data = NULL;
	u32 running;
	const __le32 *new_io_mc_regs = NULL;
	int i, regs_size, ucode_size;
	const struct mc_firmware_header_v1_0 *hdr;

	if (!adev->mc.fw)
		return -EINVAL;

	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;

	amdgpu_ucode_print_mc_hdr(&hdr->header);

	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
	new_io_mc_regs = (const __le32 *)
		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
	new_fw_data = (const __le32 *)
		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));

179
	running = RREG32(mmMC_SEQ_SUP_CNTL) & MC_SEQ_SUP_CNTL__RUN_MASK;
180 181 182 183

	if (running == 0) {

		/* reset the engine and set to writable */
184 185
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
186 187 188

		/* load mc io regs */
		for (i = 0; i < regs_size; i++) {
189 190
			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(new_io_mc_regs++));
			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(new_io_mc_regs++));
191 192 193
		}
		/* load the MC ucode */
		for (i = 0; i < ucode_size; i++) {
194
			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(new_fw_data++));
195 196 197
		}

		/* put the engine back into the active state */
198 199 200
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
201 202 203

		/* wait for training to complete */
		for (i = 0; i < adev->usec_timeout; i++) {
204
			if (RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL) & MC_SEQ_TRAIN_WAKEUP_CNTL__TRAIN_DONE_D0_MASK)
205 206 207 208
				break;
			udelay(1);
		}
		for (i = 0; i < adev->usec_timeout; i++) {
209
			if (RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL) & MC_SEQ_TRAIN_WAKEUP_CNTL__TRAIN_DONE_D1_MASK)
210 211 212 213 214 215 216 217 218 219 220 221
				break;
			udelay(1);
		}

	}

	return 0;
}

static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev,
				       struct amdgpu_mc *mc)
{
222 223 224
	u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
	base <<= 24;

225 226 227 228 229
	if (mc->mc_vram_size > 0xFFC0000000ULL) {
		dev_warn(adev->dev, "limiting VRAM\n");
		mc->real_vram_size = 0xFFC0000000ULL;
		mc->mc_vram_size = 0xFFC0000000ULL;
	}
230
	amdgpu_vram_location(adev, &adev->mc, base);
231
	amdgpu_gart_location(adev, mc);
232 233 234 235 236 237 238 239 240 241 242 243 244 245
}

static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
{
	int i, j;

	/* Initialize HDP */
	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
		WREG32((0xb05 + j), 0x00000000);
		WREG32((0xb06 + j), 0x00000000);
		WREG32((0xb07 + j), 0x00000000);
		WREG32((0xb08 + j), 0x00000000);
		WREG32((0xb09 + j), 0x00000000);
	}
246
	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
247 248 249 250 251

	if (gmc_v6_0_wait_for_idle((void *)adev)) {
		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
	}

252 253 254 255 256 257 258 259 260 261 262 263 264
	if (adev->mode_info.num_crtc) {
		u32 tmp;

		/* Lockout access through VGA aperture*/
		tmp = RREG32(mmVGA_HDP_CONTROL);
		tmp |= VGA_HDP_CONTROL__VGA_MEMORY_DISABLE_MASK;
		WREG32(mmVGA_HDP_CONTROL, tmp);

		/* disable VGA render */
		tmp = RREG32(mmVGA_RENDER_CONTROL);
		tmp &= ~VGA_VSTATUS_CNTL;
		WREG32(mmVGA_RENDER_CONTROL, tmp);
	}
265
	/* Update configuration */
266
	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
267
	       adev->mc.vram_start >> 12);
268
	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
269
	       adev->mc.vram_end >> 12);
270
	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
271
	       adev->vram_scratch.gpu_addr >> 12);
272 273 274
	WREG32(mmMC_VM_AGP_BASE, 0);
	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
275 276 277 278 279 280 281 282 283 284 285

	if (gmc_v6_0_wait_for_idle((void *)adev)) {
		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
	}
}

static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
{

	u32 tmp;
	int chansize, numchan;
286
	int r;
287

288 289
	tmp = RREG32(mmMC_ARB_RAMCFG);
	if (tmp & (1 << 11)) {
290
		chansize = 16;
291
	} else if (tmp & MC_ARB_RAMCFG__CHANSIZE_MASK) {
292 293 294 295
		chansize = 64;
	} else {
		chansize = 32;
	}
296 297
	tmp = RREG32(mmMC_SHARED_CHMAP);
	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328
	case 0:
	default:
		numchan = 1;
		break;
	case 1:
		numchan = 2;
		break;
	case 2:
		numchan = 4;
		break;
	case 3:
		numchan = 8;
		break;
	case 4:
		numchan = 3;
		break;
	case 5:
		numchan = 6;
		break;
	case 6:
		numchan = 10;
		break;
	case 7:
		numchan = 12;
		break;
	case 8:
		numchan = 16;
		break;
	}
	adev->mc.vram_width = numchan * chansize;
	/* size in MB on si */
329 330
	adev->mc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
331 332 333 334 335 336 337 338

	if (!(adev->flags & AMD_IS_APU)) {
		r = amdgpu_device_resize_fb_bar(adev);
		if (r)
			return r;
	}
	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
339 340
	adev->mc.visible_vram_size = adev->mc.aper_size;

341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
	/* set the gart size */
	if (amdgpu_gart_size == -1) {
		switch (adev->asic_type) {
		case CHIP_HAINAN:    /* no MM engines */
		default:
			adev->mc.gart_size = 256ULL << 20;
			break;
		case CHIP_VERDE:    /* UVD, VCE do not support GPUVM */
		case CHIP_TAHITI:   /* UVD, VCE do not support GPUVM */
		case CHIP_PITCAIRN: /* UVD, VCE do not support GPUVM */
		case CHIP_OLAND:    /* UVD, VCE do not support GPUVM */
			adev->mc.gart_size = 1024ULL << 20;
			break;
		}
	} else {
		adev->mc.gart_size = (u64)amdgpu_gart_size << 20;
	}

359 360 361 362 363 364 365 366
	gmc_v6_0_vram_gtt_location(adev, &adev->mc);

	return 0;
}

static void gmc_v6_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
					uint32_t vmid)
{
367
	WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
368

369
	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
370 371 372 373 374 375
}

static int gmc_v6_0_gart_set_pte_pde(struct amdgpu_device *adev,
				     void *cpu_pt_addr,
				     uint32_t gpu_page_idx,
				     uint64_t addr,
376
				     uint64_t flags)
377 378 379 380 381 382 383 384 385 386 387
{
	void __iomem *ptr = (void *)cpu_pt_addr;
	uint64_t value;

	value = addr & 0xFFFFFFFFFFFFF000ULL;
	value |= flags;
	writeq(value, ptr + (gpu_page_idx * 8));

	return 0;
}

388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
static uint64_t gmc_v6_0_get_vm_pte_flags(struct amdgpu_device *adev,
					  uint32_t flags)
{
	uint64_t pte_flag = 0;

	if (flags & AMDGPU_VM_PAGE_READABLE)
		pte_flag |= AMDGPU_PTE_READABLE;
	if (flags & AMDGPU_VM_PAGE_WRITEABLE)
		pte_flag |= AMDGPU_PTE_WRITEABLE;
	if (flags & AMDGPU_VM_PAGE_PRT)
		pte_flag |= AMDGPU_PTE_PRT;

	return pte_flag;
}

403 404 405 406 407 408
static uint64_t gmc_v6_0_get_vm_pde(struct amdgpu_device *adev, uint64_t addr)
{
	BUG_ON(addr & 0xFFFFFF0000000FFFULL);
	return addr;
}

409 410 411 412 413
static void gmc_v6_0_set_fault_enable_default(struct amdgpu_device *adev,
					      bool value)
{
	u32 tmp;

414 415 416 417 418 419 420 421 422 423 424 425 426 427
	tmp = RREG32(mmVM_CONTEXT1_CNTL);
	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
	WREG32(mmVM_CONTEXT1_CNTL, tmp);
428 429
}

430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483
 /**
   + * gmc_v8_0_set_prt - set PRT VM fault
   + *
   + * @adev: amdgpu_device pointer
   + * @enable: enable/disable VM fault handling for PRT
   +*/
static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
{
	u32 tmp;

	if (enable && !adev->mc.prt_warning) {
		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
		adev->mc.prt_warning = true;
	}

	tmp = RREG32(mmVM_PRT_CNTL);
	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
			    CB_DISABLE_FAULT_ON_UNMAPPED_ACCESS,
			    enable);
	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
			    TC_DISABLE_FAULT_ON_UNMAPPED_ACCESS,
			    enable);
	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
			    L2_CACHE_STORE_INVALID_ENTRIES,
			    enable);
	tmp = REG_SET_FIELD(tmp, VM_PRT_CNTL,
			    L1_TLB_STORE_INVALID_ENTRIES,
			    enable);
	WREG32(mmVM_PRT_CNTL, tmp);

	if (enable) {
		uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
		uint32_t high = adev->vm_manager.max_pfn;

		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, low);
		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, low);
		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, high);
		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, high);
		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, high);
		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, high);
	} else {
		WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, 0xfffffff);
		WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, 0xfffffff);
		WREG32(mmVM_PRT_APERTURE2_LOW_ADDR, 0xfffffff);
		WREG32(mmVM_PRT_APERTURE3_LOW_ADDR, 0xfffffff);
		WREG32(mmVM_PRT_APERTURE0_HIGH_ADDR, 0x0);
		WREG32(mmVM_PRT_APERTURE1_HIGH_ADDR, 0x0);
		WREG32(mmVM_PRT_APERTURE2_HIGH_ADDR, 0x0);
		WREG32(mmVM_PRT_APERTURE3_HIGH_ADDR, 0x0);
	}
}

484 485 486
static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
{
	int r, i;
487
	u32 field;
488 489 490 491 492 493 494 495 496

	if (adev->gart.robj == NULL) {
		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
		return -EINVAL;
	}
	r = amdgpu_gart_table_vram_pin(adev);
	if (r)
		return r;
	/* Setup TLB control */
497
	WREG32(mmMC_VM_MX_L1_TLB_CNTL,
498
	       (0xA << 7) |
499 500 501 502 503
	       MC_VM_MX_L1_TLB_CNTL__ENABLE_L1_TLB_MASK |
	       MC_VM_MX_L1_TLB_CNTL__ENABLE_L1_FRAGMENT_PROCESSING_MASK |
	       MC_VM_MX_L1_TLB_CNTL__SYSTEM_ACCESS_MODE_MASK |
	       MC_VM_MX_L1_TLB_CNTL__ENABLE_ADVANCED_DRIVER_MODEL_MASK |
	       (0UL << MC_VM_MX_L1_TLB_CNTL__SYSTEM_APERTURE_UNMAPPED_ACCESS__SHIFT));
504
	/* Setup L2 cache */
505 506 507 508 509 510 511 512 513 514
	WREG32(mmVM_L2_CNTL,
	       VM_L2_CNTL__ENABLE_L2_CACHE_MASK |
	       VM_L2_CNTL__ENABLE_L2_FRAGMENT_PROCESSING_MASK |
	       VM_L2_CNTL__ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE_MASK |
	       VM_L2_CNTL__ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE_MASK |
	       (7UL << VM_L2_CNTL__EFFECTIVE_L2_QUEUE_SIZE__SHIFT) |
	       (1UL << VM_L2_CNTL__CONTEXT1_IDENTITY_ACCESS_MODE__SHIFT));
	WREG32(mmVM_L2_CNTL2,
	       VM_L2_CNTL2__INVALIDATE_ALL_L1_TLBS_MASK |
	       VM_L2_CNTL2__INVALIDATE_L2_CACHE_MASK);
515 516

	field = adev->vm_manager.fragment_size;
517 518
	WREG32(mmVM_L2_CNTL3,
	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
519 520
	       (field << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
	       (field << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
521
	/* setup context0 */
522 523
	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
524 525
	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
526
			(u32)(adev->dummy_page.addr >> 12));
527 528 529 530 531
	WREG32(mmVM_CONTEXT0_CNTL2, 0);
	WREG32(mmVM_CONTEXT0_CNTL,
	       VM_CONTEXT0_CNTL__ENABLE_CONTEXT_MASK |
	       (0UL << VM_CONTEXT0_CNTL__PAGE_TABLE_DEPTH__SHIFT) |
	       VM_CONTEXT0_CNTL__RANGE_PROTECTION_FAULT_ENABLE_DEFAULT_MASK);
532 533 534 535 536 537 538

	WREG32(0x575, 0);
	WREG32(0x576, 0);
	WREG32(0x577, 0);

	/* empty context1-15 */
	/* set vm size, must be a multiple of 4 */
539 540
	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
541 542 543 544 545 546
	/* Assign the pt base to something valid for now; the pts used for
	 * the VMs are determined by the application and setup and assigned
	 * on the fly in the vm part of radeon_gart.c
	 */
	for (i = 1; i < 16; i++) {
		if (i < 8)
547
			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
548 549
			       adev->gart.table_addr >> 12);
		else
550
			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
551 552 553 554
			       adev->gart.table_addr >> 12);
	}

	/* enable context1-15 */
555
	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
556
	       (u32)(adev->dummy_page.addr >> 12));
557 558 559 560
	WREG32(mmVM_CONTEXT1_CNTL2, 4);
	WREG32(mmVM_CONTEXT1_CNTL,
	       VM_CONTEXT1_CNTL__ENABLE_CONTEXT_MASK |
	       (1UL << VM_CONTEXT1_CNTL__PAGE_TABLE_DEPTH__SHIFT) |
561 562
	       ((adev->vm_manager.block_size - 9)
	       << VM_CONTEXT1_CNTL__PAGE_TABLE_BLOCK_SIZE__SHIFT));
563 564 565 566
	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
		gmc_v6_0_set_fault_enable_default(adev, false);
	else
		gmc_v6_0_set_fault_enable_default(adev, true);
567 568

	gmc_v6_0_gart_flush_gpu_tlb(adev, 0);
569
	dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n",
570
		 (unsigned)(adev->mc.gart_size >> 20),
571 572 573 574 575 576 577 578 579 580
		 (unsigned long long)adev->gart.table_addr);
	adev->gart.ready = true;
	return 0;
}

static int gmc_v6_0_gart_init(struct amdgpu_device *adev)
{
	int r;

	if (adev->gart.robj) {
581
		dev_warn(adev->dev, "gmc_v6_0 PCIE GART already initialized\n");
582 583 584 585 586 587
		return 0;
	}
	r = amdgpu_gart_init(adev);
	if (r)
		return r;
	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
588
	adev->gart.gart_pte_flags = 0;
589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605
	return amdgpu_gart_table_vram_alloc(adev);
}

static void gmc_v6_0_gart_disable(struct amdgpu_device *adev)
{
	/*unsigned i;

	for (i = 1; i < 16; ++i) {
		uint32_t reg;
		if (i < 8)
			reg = VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i ;
		else
			reg = VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (i - 8);
		adev->vm_manager.saved_table_addr[i] = RREG32(reg);
	}*/

	/* Disable all tables */
606 607
	WREG32(mmVM_CONTEXT0_CNTL, 0);
	WREG32(mmVM_CONTEXT1_CNTL, 0);
608
	/* Setup TLB control */
609 610 611
	WREG32(mmMC_VM_MX_L1_TLB_CNTL,
	       MC_VM_MX_L1_TLB_CNTL__SYSTEM_ACCESS_MODE_MASK |
	       (0UL << MC_VM_MX_L1_TLB_CNTL__SYSTEM_APERTURE_UNMAPPED_ACCESS__SHIFT));
612
	/* Setup L2 cache */
613 614 615 616 617 618 619 620 621
	WREG32(mmVM_L2_CNTL,
	       VM_L2_CNTL__ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE_MASK |
	       VM_L2_CNTL__ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE_MASK |
	       (7UL << VM_L2_CNTL__EFFECTIVE_L2_QUEUE_SIZE__SHIFT) |
	       (1UL << VM_L2_CNTL__CONTEXT1_IDENTITY_ACCESS_MODE__SHIFT));
	WREG32(mmVM_L2_CNTL2, 0);
	WREG32(mmVM_L2_CNTL3,
	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
	       (0UL << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
622 623 624 625 626 627 628 629 630 631 632 633 634
	amdgpu_gart_table_vram_unpin(adev);
}

static void gmc_v6_0_gart_fini(struct amdgpu_device *adev)
{
	amdgpu_gart_table_vram_free(adev);
	amdgpu_gart_fini(adev);
}

static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
				     u32 status, u32 addr, u32 mc_client)
{
	u32 mc_id;
635 636 637
	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
					PROTECTIONS);
638 639 640
	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };

641 642
	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
			      MEMORY_CLIENT_ID);
643

644
	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
645
	       protections, vmid, addr,
646 647
	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
			     MEMORY_CLIENT_RW) ?
648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749
	       "write" : "read", block, mc_client, mc_id);
}

/*
static const u32 mc_cg_registers[] = {
	MC_HUB_MISC_HUB_CG,
	MC_HUB_MISC_SIP_CG,
	MC_HUB_MISC_VM_CG,
	MC_XPB_CLK_GAT,
	ATC_MISC_CG,
	MC_CITF_MISC_WR_CG,
	MC_CITF_MISC_RD_CG,
	MC_CITF_MISC_VM_CG,
	VM_L2_CG,
};

static const u32 mc_cg_ls_en[] = {
	MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK,
	MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK,
	MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK,
	MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK,
	ATC_MISC_CG__MEM_LS_ENABLE_MASK,
	MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK,
	MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK,
	MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK,
	VM_L2_CG__MEM_LS_ENABLE_MASK,
};

static const u32 mc_cg_en[] = {
	MC_HUB_MISC_HUB_CG__ENABLE_MASK,
	MC_HUB_MISC_SIP_CG__ENABLE_MASK,
	MC_HUB_MISC_VM_CG__ENABLE_MASK,
	MC_XPB_CLK_GAT__ENABLE_MASK,
	ATC_MISC_CG__ENABLE_MASK,
	MC_CITF_MISC_WR_CG__ENABLE_MASK,
	MC_CITF_MISC_RD_CG__ENABLE_MASK,
	MC_CITF_MISC_VM_CG__ENABLE_MASK,
	VM_L2_CG__ENABLE_MASK,
};

static void gmc_v6_0_enable_mc_ls(struct amdgpu_device *adev,
				  bool enable)
{
	int i;
	u32 orig, data;

	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
		orig = data = RREG32(mc_cg_registers[i]);
		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_LS))
			data |= mc_cg_ls_en[i];
		else
			data &= ~mc_cg_ls_en[i];
		if (data != orig)
			WREG32(mc_cg_registers[i], data);
	}
}

static void gmc_v6_0_enable_mc_mgcg(struct amdgpu_device *adev,
				    bool enable)
{
	int i;
	u32 orig, data;

	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
		orig = data = RREG32(mc_cg_registers[i]);
		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_MGCG))
			data |= mc_cg_en[i];
		else
			data &= ~mc_cg_en[i];
		if (data != orig)
			WREG32(mc_cg_registers[i], data);
	}
}

static void gmc_v6_0_enable_bif_mgls(struct amdgpu_device *adev,
				     bool enable)
{
	u32 orig, data;

	orig = data = RREG32_PCIE(ixPCIE_CNTL2);

	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_BIF_LS)) {
		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 1);
		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 1);
		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 1);
		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 1);
	} else {
		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 0);
		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 0);
		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 0);
		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 0);
	}

	if (orig != data)
		WREG32_PCIE(ixPCIE_CNTL2, data);
}

static void gmc_v6_0_enable_hdp_mgcg(struct amdgpu_device *adev,
				     bool enable)
{
	u32 orig, data;

750
	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
751 752 753 754 755 756 757

	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_MGCG))
		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 0);
	else
		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 1);

	if (orig != data)
758
		WREG32(mmHDP_HOST_PATH_CNTL, data);
759 760 761 762 763 764 765
}

static void gmc_v6_0_enable_hdp_ls(struct amdgpu_device *adev,
				   bool enable)
{
	u32 orig, data;

766
	orig = data = RREG32(mmHDP_MEM_POWER_LS);
767 768 769 770 771 772 773

	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_LS))
		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 1);
	else
		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 0);

	if (orig != data)
774
		WREG32(mmHDP_MEM_POWER_LS, data);
775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
}
*/

static int gmc_v6_0_convert_vram_type(int mc_seq_vram_type)
{
	switch (mc_seq_vram_type) {
	case MC_SEQ_MISC0__MT__GDDR1:
		return AMDGPU_VRAM_TYPE_GDDR1;
	case MC_SEQ_MISC0__MT__DDR2:
		return AMDGPU_VRAM_TYPE_DDR2;
	case MC_SEQ_MISC0__MT__GDDR3:
		return AMDGPU_VRAM_TYPE_GDDR3;
	case MC_SEQ_MISC0__MT__GDDR4:
		return AMDGPU_VRAM_TYPE_GDDR4;
	case MC_SEQ_MISC0__MT__GDDR5:
		return AMDGPU_VRAM_TYPE_GDDR5;
	case MC_SEQ_MISC0__MT__DDR3:
		return AMDGPU_VRAM_TYPE_DDR3;
	default:
		return AMDGPU_VRAM_TYPE_UNKNOWN;
	}
}

static int gmc_v6_0_early_init(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

	gmc_v6_0_set_gart_funcs(adev);
	gmc_v6_0_set_irq_funcs(adev);

	return 0;
}

static int gmc_v6_0_late_init(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

812 813 814 815
	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
		return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
	else
		return 0;
816 817 818 819 820 821 822 823
}

static int gmc_v6_0_sw_init(void *handle)
{
	int r;
	int dma_bits;
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

824 825 826 827 828 829 830 831
	if (adev->flags & AMD_IS_APU) {
		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
	} else {
		u32 tmp = RREG32(mmMC_SEQ_MISC0);
		tmp &= MC_SEQ_MISC0__MT__MASK;
		adev->mc.vram_type = gmc_v6_0_convert_vram_type(tmp);
	}

832
	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->mc.vm_fault);
833 834 835
	if (r)
		return r;

836
	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->mc.vm_fault);
837 838 839
	if (r)
		return r;

840
	amdgpu_vm_adjust_size(adev, 64, 9);
841

842 843
	adev->mc.mc_mask = 0xffffffffffULL;

844 845
	adev->mc.stolen_size = 256 * 1024;

846 847 848 849 850 851
	adev->need_dma32 = false;
	dma_bits = adev->need_dma32 ? 32 : 40;
	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
	if (r) {
		adev->need_dma32 = true;
		dma_bits = 32;
852
		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
853 854 855 856
	}
	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
	if (r) {
		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
857
		dev_warn(adev->dev, "amdgpu: No coherent DMA available.\n");
858 859 860 861
	}

	r = gmc_v6_0_init_microcode(adev);
	if (r) {
862
		dev_err(adev->dev, "Failed to load mc firmware!\n");
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877
		return r;
	}

	r = gmc_v6_0_mc_init(adev);
	if (r)
		return r;

	r = amdgpu_bo_init(adev);
	if (r)
		return r;

	r = gmc_v6_0_gart_init(adev);
	if (r)
		return r;

878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895
	/*
	 * number of VMs
	 * VMID 0 is reserved for System
	 * amdgpu graphics/compute will use VMIDs 1-7
	 * amdkfd will use VMIDs 8-15
	 */
	adev->vm_manager.id_mgr[0].num_ids = AMDGPU_NUM_OF_VMIDS;
	adev->vm_manager.num_level = 1;
	amdgpu_vm_manager_init(adev);

	/* base offset of vram pages */
	if (adev->flags & AMD_IS_APU) {
		u64 tmp = RREG32(mmMC_VM_FB_OFFSET);

		tmp <<= 22;
		adev->vm_manager.vram_base_offset = tmp;
	} else {
		adev->vm_manager.vram_base_offset = 0;
896 897
	}

898
	return 0;
899 900 901 902 903 904
}

static int gmc_v6_0_sw_fini(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

905
	amdgpu_vm_manager_fini(adev);
906 907 908
	gmc_v6_0_gart_fini(adev);
	amdgpu_gem_force_release(adev);
	amdgpu_bo_fini(adev);
909 910
	release_firmware(adev->mc.fw);
	adev->mc.fw = NULL;
911 912 913 914 915 916 917 918 919 920 921 922 923 924

	return 0;
}

static int gmc_v6_0_hw_init(void *handle)
{
	int r;
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

	gmc_v6_0_mc_program(adev);

	if (!(adev->flags & AMD_IS_APU)) {
		r = gmc_v6_0_mc_load_microcode(adev);
		if (r) {
925
			dev_err(adev->dev, "Failed to load MC firmware!\n");
926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964
			return r;
		}
	}

	r = gmc_v6_0_gart_enable(adev);
	if (r)
		return r;

	return r;
}

static int gmc_v6_0_hw_fini(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
	gmc_v6_0_gart_disable(adev);

	return 0;
}

static int gmc_v6_0_suspend(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

	gmc_v6_0_hw_fini(adev);

	return 0;
}

static int gmc_v6_0_resume(void *handle)
{
	int r;
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

	r = gmc_v6_0_hw_init(adev);
	if (r)
		return r;

965
	amdgpu_vm_reset_all_ids(adev);
966

967
	return 0;
968 969 970 971 972
}

static bool gmc_v6_0_is_idle(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
973
	u32 tmp = RREG32(mmSRBM_STATUS);
974 975 976 977 978 979 980 981 982 983 984 985 986 987

	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK | SRBM_STATUS__VMC_BUSY_MASK))
		return false;

	return true;
}

static int gmc_v6_0_wait_for_idle(void *handle)
{
	unsigned i;
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

	for (i = 0; i < adev->usec_timeout; i++) {
988
		if (gmc_v6_0_is_idle(handle))
989 990 991 992 993 994 995 996 997 998 999
			return 0;
		udelay(1);
	}
	return -ETIMEDOUT;

}

static int gmc_v6_0_soft_reset(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
	u32 srbm_soft_reset = 0;
1000
	u32 tmp = RREG32(mmSRBM_STATUS);
1001 1002 1003

	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
1004
						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
1005 1006 1007 1008 1009

	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK)) {
		if (!(adev->flags & AMD_IS_APU))
			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
1010
							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
1011 1012 1013
	}

	if (srbm_soft_reset) {
1014
		gmc_v6_0_mc_stop(adev);
1015 1016 1017 1018 1019
		if (gmc_v6_0_wait_for_idle(adev)) {
			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
		}


1020
		tmp = RREG32(mmSRBM_SOFT_RESET);
1021 1022
		tmp |= srbm_soft_reset;
		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
1023 1024
		WREG32(mmSRBM_SOFT_RESET, tmp);
		tmp = RREG32(mmSRBM_SOFT_RESET);
1025 1026 1027 1028

		udelay(50);

		tmp &= ~srbm_soft_reset;
1029 1030
		WREG32(mmSRBM_SOFT_RESET, tmp);
		tmp = RREG32(mmSRBM_SOFT_RESET);
1031 1032 1033

		udelay(50);

1034
		gmc_v6_0_mc_resume(adev);
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
		udelay(50);
	}

	return 0;
}

static int gmc_v6_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
					     struct amdgpu_irq_src *src,
					     unsigned type,
					     enum amdgpu_interrupt_state state)
{
	u32 tmp;
	u32 bits = (VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
		    VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
		    VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
		    VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
		    VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
		    VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK);

	switch (state) {
	case AMDGPU_IRQ_STATE_DISABLE:
1056
		tmp = RREG32(mmVM_CONTEXT0_CNTL);
1057
		tmp &= ~bits;
1058 1059
		WREG32(mmVM_CONTEXT0_CNTL, tmp);
		tmp = RREG32(mmVM_CONTEXT1_CNTL);
1060
		tmp &= ~bits;
1061
		WREG32(mmVM_CONTEXT1_CNTL, tmp);
1062 1063
		break;
	case AMDGPU_IRQ_STATE_ENABLE:
1064
		tmp = RREG32(mmVM_CONTEXT0_CNTL);
1065
		tmp |= bits;
1066 1067
		WREG32(mmVM_CONTEXT0_CNTL, tmp);
		tmp = RREG32(mmVM_CONTEXT1_CNTL);
1068
		tmp |= bits;
1069
		WREG32(mmVM_CONTEXT1_CNTL, tmp);
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
		break;
	default:
		break;
	}

	return 0;
}

static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev,
				      struct amdgpu_irq_src *source,
				      struct amdgpu_iv_entry *entry)
{
	u32 addr, status;

1084 1085 1086
	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
1087 1088 1089 1090 1091 1092 1093

	if (!addr && !status)
		return 0;

	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
		gmc_v6_0_set_fault_enable_default(adev, false);

1094 1095
	if (printk_ratelimit()) {
		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
1096
			entry->src_id, entry->src_data[0]);
1097 1098 1099 1100 1101 1102
		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
			addr);
		dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
			status);
		gmc_v6_0_vm_decode_fault(adev, status, addr, 0);
	}
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118

	return 0;
}

static int gmc_v6_0_set_clockgating_state(void *handle,
					  enum amd_clockgating_state state)
{
	return 0;
}

static int gmc_v6_0_set_powergating_state(void *handle,
					  enum amd_powergating_state state)
{
	return 0;
}

1119
static const struct amd_ip_funcs gmc_v6_0_ip_funcs = {
1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138
	.name = "gmc_v6_0",
	.early_init = gmc_v6_0_early_init,
	.late_init = gmc_v6_0_late_init,
	.sw_init = gmc_v6_0_sw_init,
	.sw_fini = gmc_v6_0_sw_fini,
	.hw_init = gmc_v6_0_hw_init,
	.hw_fini = gmc_v6_0_hw_fini,
	.suspend = gmc_v6_0_suspend,
	.resume = gmc_v6_0_resume,
	.is_idle = gmc_v6_0_is_idle,
	.wait_for_idle = gmc_v6_0_wait_for_idle,
	.soft_reset = gmc_v6_0_soft_reset,
	.set_clockgating_state = gmc_v6_0_set_clockgating_state,
	.set_powergating_state = gmc_v6_0_set_powergating_state,
};

static const struct amdgpu_gart_funcs gmc_v6_0_gart_funcs = {
	.flush_gpu_tlb = gmc_v6_0_gart_flush_gpu_tlb,
	.set_pte_pde = gmc_v6_0_gart_set_pte_pde,
1139
	.set_prt = gmc_v6_0_set_prt,
1140
	.get_vm_pde = gmc_v6_0_get_vm_pde,
1141
	.get_vm_pte_flags = gmc_v6_0_get_vm_pte_flags
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160
};

static const struct amdgpu_irq_src_funcs gmc_v6_0_irq_funcs = {
	.set = gmc_v6_0_vm_fault_interrupt_state,
	.process = gmc_v6_0_process_interrupt,
};

static void gmc_v6_0_set_gart_funcs(struct amdgpu_device *adev)
{
	if (adev->gart.gart_funcs == NULL)
		adev->gart.gart_funcs = &gmc_v6_0_gart_funcs;
}

static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev)
{
	adev->mc.vm_fault.num_types = 1;
	adev->mc.vm_fault.funcs = &gmc_v6_0_irq_funcs;
}

1161 1162 1163 1164 1165 1166 1167 1168
const struct amdgpu_ip_block_version gmc_v6_0_ip_block =
{
	.type = AMD_IP_BLOCK_TYPE_GMC,
	.major = 6,
	.minor = 0,
	.rev = 0,
	.funcs = &gmc_v6_0_ip_funcs,
};