gmc_v6_0.c 32.1 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
#include <drm/drm_cache.h>
26 27 28
#include "amdgpu.h"
#include "gmc_v6_0.h"
#include "amdgpu_ucode.h"
29 30 31 32 33 34 35 36 37 38

#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"
39

40
static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev);
41 42 43
static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev);
static int gmc_v6_0_wait_for_idle(void *handle);

44 45 46 47 48
MODULE_FIRMWARE("amdgpu/tahiti_mc.bin");
MODULE_FIRMWARE("amdgpu/pitcairn_mc.bin");
MODULE_FIRMWARE("amdgpu/verde_mc.bin");
MODULE_FIRMWARE("amdgpu/oland_mc.bin");
MODULE_FIRMWARE("amdgpu/si58_mc.bin");
49

50 51 52 53 54 55 56 57 58 59
#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


60 61 62 63 64 65 66 67 68 69
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
};

70
static void gmc_v6_0_mc_stop(struct amdgpu_device *adev)
71 72 73 74 75
{
	u32 blackout;

	gmc_v6_0_wait_for_idle((void *)adev);

76 77
	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
78
		/* Block CPU access */
79
		WREG32(mmBIF_FB_EN, 0);
80 81
		/* blackout the MC */
		blackout = REG_SET_FIELD(blackout,
82 83
					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
84 85 86 87 88 89
	}
	/* wait for the MC to settle */
	udelay(100);

}

90
static void gmc_v6_0_mc_resume(struct amdgpu_device *adev)
91 92 93 94
{
	u32 tmp;

	/* unblackout the MC */
95 96 97
	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
98
	/* allow CPU access */
99 100 101
	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);
102 103 104 105 106 107 108
}

static int gmc_v6_0_init_microcode(struct amdgpu_device *adev)
{
	const char *chip_name;
	char fw_name[30];
	int err;
109
	bool is_58_fw = false;
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131

	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();
	}

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

	if (is_58_fw)
137
		snprintf(fw_name, sizeof(fw_name), "amdgpu/si58_mc.bin");
138
	else
139
		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
140
	err = request_firmware(&adev->gmc.fw, fw_name, adev->dev);
141 142 143
	if (err)
		goto out;

144
	err = amdgpu_ucode_validate(adev->gmc.fw);
145 146 147

out:
	if (err) {
148
		dev_err(adev->dev,
149 150
		       "si_mc: Failed to load firmware \"%s\"\n",
		       fw_name);
151 152
		release_firmware(adev->gmc.fw);
		adev->gmc.fw = NULL;
153 154 155 156 157 158 159 160 161 162 163 164
	}
	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;

165
	if (!adev->gmc.fw)
166 167
		return -EINVAL;

168
	hdr = (const struct mc_firmware_header_v1_0 *)adev->gmc.fw->data;
169 170 171

	amdgpu_ucode_print_mc_hdr(&hdr->header);

172
	adev->gmc.fw_version = le32_to_cpu(hdr->header.ucode_version);
173 174
	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
	new_io_mc_regs = (const __le32 *)
175
		(adev->gmc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
176 177
	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
	new_fw_data = (const __le32 *)
178
		(adev->gmc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
179

180
	running = RREG32(mmMC_SEQ_SUP_CNTL) & MC_SEQ_SUP_CNTL__RUN_MASK;
181 182 183 184

	if (running == 0) {

		/* reset the engine and set to writable */
185 186
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
187 188 189

		/* load mc io regs */
		for (i = 0; i < regs_size; i++) {
190 191
			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++));
192 193 194
		}
		/* load the MC ucode */
		for (i = 0; i < ucode_size; i++) {
195
			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(new_fw_data++));
196 197 198
		}

		/* put the engine back into the active state */
199 200 201
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
202 203 204

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

	}

	return 0;
}

static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev,
221
				       struct amdgpu_gmc *mc)
222
{
223 224 225
	u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
	base <<= 24;

226
	amdgpu_device_vram_location(adev, &adev->gmc, base);
227
	amdgpu_device_gart_location(adev, mc);
228 229 230 231 232 233 234 235 236 237 238 239 240 241
}

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);
	}
242
	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
243 244 245 246 247

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

248 249 250 251 252 253 254 255 256 257 258 259 260
	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);
	}
261
	/* Update configuration */
262
	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
263
	       adev->gmc.vram_start >> 12);
264
	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
265
	       adev->gmc.vram_end >> 12);
266
	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
267
	       adev->vram_scratch.gpu_addr >> 12);
268 269 270
	WREG32(mmMC_VM_AGP_BASE, 0);
	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
271 272 273 274 275 276 277 278 279 280 281

	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;
282
	int r;
283

284 285
	tmp = RREG32(mmMC_ARB_RAMCFG);
	if (tmp & (1 << 11)) {
286
		chansize = 16;
287
	} else if (tmp & MC_ARB_RAMCFG__CHANSIZE_MASK) {
288 289 290 291
		chansize = 64;
	} else {
		chansize = 32;
	}
292 293
	tmp = RREG32(mmMC_SHARED_CHMAP);
	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322
	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;
	}
323
	adev->gmc.vram_width = numchan * chansize;
324
	/* size in MB on si */
325 326
	adev->gmc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
	adev->gmc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
327 328 329 330 331 332

	if (!(adev->flags & AMD_IS_APU)) {
		r = amdgpu_device_resize_fb_bar(adev);
		if (r)
			return r;
	}
333 334 335
	adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
	adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
	adev->gmc.visible_vram_size = adev->gmc.aper_size;
336

337 338 339 340 341
	/* set the gart size */
	if (amdgpu_gart_size == -1) {
		switch (adev->asic_type) {
		case CHIP_HAINAN:    /* no MM engines */
		default:
342
			adev->gmc.gart_size = 256ULL << 20;
343 344 345 346 347
			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 */
348
			adev->gmc.gart_size = 1024ULL << 20;
349 350 351
			break;
		}
	} else {
352
		adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
353 354
	}

355
	gmc_v6_0_vram_gtt_location(adev, &adev->gmc);
356 357 358 359

	return 0;
}

360
static void gmc_v6_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid)
361
{
362
	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
363 364
}

365
static uint64_t gmc_v6_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
366
					    unsigned vmid, uint64_t pd_addr)
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
{
	uint32_t reg;

	/* write new base address */
	if (vmid < 8)
		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vmid;
	else
		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + (vmid - 8);
	amdgpu_ring_emit_wreg(ring, reg, pd_addr >> 12);

	/* bits 0-15 are the VM contexts0-15 */
	amdgpu_ring_emit_wreg(ring, mmVM_INVALIDATE_REQUEST, 1 << vmid);

	return pd_addr;
}

383 384 385
static int gmc_v6_0_set_pte_pde(struct amdgpu_device *adev, void *cpu_pt_addr,
				uint32_t gpu_page_idx, uint64_t addr,
				uint64_t flags)
386 387 388 389 390 391 392 393 394 395 396
{
	void __iomem *ptr = (void *)cpu_pt_addr;
	uint64_t value;

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

	return 0;
}

397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
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;
}

412 413
static void gmc_v6_0_get_vm_pde(struct amdgpu_device *adev, int level,
				uint64_t *addr, uint64_t *flags)
414
{
415
	BUG_ON(*addr & 0xFFFFFF0000000FFFULL);
416 417
}

418 419 420 421 422
static void gmc_v6_0_set_fault_enable_default(struct amdgpu_device *adev,
					      bool value)
{
	u32 tmp;

423 424 425 426 427 428 429 430 431 432 433 434 435 436
	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);
437 438
}

439 440 441 442 443 444 445 446 447 448
 /**
   + * 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;

449
	if (enable && !adev->gmc.prt_warning) {
450
		dev_warn(adev->dev, "Disabling VM faults because of PRT request!\n");
451
		adev->gmc.prt_warning = true;
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470
	}

	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;
471 472
		uint32_t high = adev->vm_manager.max_pfn -
			(AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493

		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);
	}
}

494 495
static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
{
496
	int r, i;
497
	u32 field;
498 499 500 501 502

	if (adev->gart.robj == NULL) {
		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
		return -EINVAL;
	}
503 504 505
	r = amdgpu_gart_table_vram_pin(adev);
	if (r)
		return r;
506
	/* Setup TLB control */
507
	WREG32(mmMC_VM_MX_L1_TLB_CNTL,
508
	       (0xA << 7) |
509 510 511 512 513
	       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));
514
	/* Setup L2 cache */
515 516 517 518 519 520 521 522 523 524
	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);
525 526

	field = adev->vm_manager.fragment_size;
527 528
	WREG32(mmVM_L2_CNTL3,
	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
529 530
	       (field << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
	       (field << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
531
	/* setup context0 */
532 533
	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->gmc.gart_start >> 12);
	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->gmc.gart_end >> 12);
534 535
	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
536
			(u32)(adev->dummy_page_addr >> 12));
537 538 539 540 541
	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);
542 543 544 545 546 547 548

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

	/* empty context1-15 */
	/* set vm size, must be a multiple of 4 */
549 550
	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
551 552 553 554 555 556
	/* 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)
557
			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
558 559
			       adev->gart.table_addr >> 12);
		else
560
			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
561 562 563 564
			       adev->gart.table_addr >> 12);
	}

	/* enable context1-15 */
565
	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
566
	       (u32)(adev->dummy_page_addr >> 12));
567 568 569 570
	WREG32(mmVM_CONTEXT1_CNTL2, 4);
	WREG32(mmVM_CONTEXT1_CNTL,
	       VM_CONTEXT1_CNTL__ENABLE_CONTEXT_MASK |
	       (1UL << VM_CONTEXT1_CNTL__PAGE_TABLE_DEPTH__SHIFT) |
571 572
	       ((adev->vm_manager.block_size - 9)
	       << VM_CONTEXT1_CNTL__PAGE_TABLE_BLOCK_SIZE__SHIFT));
573 574 575 576
	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);
577

578
	gmc_v6_0_flush_gpu_tlb(adev, 0);
579
	dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n",
580
		 (unsigned)(adev->gmc.gart_size >> 20),
581 582 583 584 585 586 587 588 589 590
		 (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) {
591
		dev_warn(adev->dev, "gmc_v6_0 PCIE GART already initialized\n");
592 593 594 595 596 597
		return 0;
	}
	r = amdgpu_gart_init(adev);
	if (r)
		return r;
	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
598
	adev->gart.gart_pte_flags = 0;
599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
	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 */
616 617
	WREG32(mmVM_CONTEXT0_CNTL, 0);
	WREG32(mmVM_CONTEXT1_CNTL, 0);
618
	/* Setup TLB control */
619 620 621
	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));
622
	/* Setup L2 cache */
623 624 625 626 627 628 629 630 631
	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));
632
	amdgpu_gart_table_vram_unpin(adev);
633 634 635 636 637 638
}

static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
				     u32 status, u32 addr, u32 mc_client)
{
	u32 mc_id;
639 640 641
	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
					PROTECTIONS);
642 643 644
	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };

645 646
	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
			      MEMORY_CLIENT_ID);
647

648
	dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
649
	       protections, vmid, addr,
650 651
	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
			     MEMORY_CLIENT_RW) ?
652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753
	       "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;

754
	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
755 756 757 758 759 760 761

	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)
762
		WREG32(mmHDP_HOST_PATH_CNTL, data);
763 764 765 766 767 768 769
}

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

770
	orig = data = RREG32(mmHDP_MEM_POWER_LS);
771 772 773 774 775 776 777

	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)
778
		WREG32(mmHDP_MEM_POWER_LS, data);
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
}
*/

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;

806
	gmc_v6_0_set_gmc_funcs(adev);
807 808 809 810 811 812 813 814 815
	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;

816 817
	amdgpu_bo_late_init(adev);

818
	if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
819
		return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
820 821
	else
		return 0;
822 823
}

824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842
static unsigned gmc_v6_0_get_vbios_fb_size(struct amdgpu_device *adev)
{
	u32 d1vga_control = RREG32(mmD1VGA_CONTROL);
	unsigned size;

	if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
		size = 9 * 1024 * 1024; /* reserve 8MB for vga emulator and 1 MB for FB */
	} else {
		u32 viewport = RREG32(mmVIEWPORT_SIZE);
		size = (REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_HEIGHT) *
			REG_GET_FIELD(viewport, VIEWPORT_SIZE, VIEWPORT_WIDTH) *
			4);
	}
	/* return 0 if the pre-OS buffer uses up most of vram */
	if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024))
		return 0;
	return size;
}

843 844 845 846 847 848
static int gmc_v6_0_sw_init(void *handle)
{
	int r;
	int dma_bits;
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;

849
	if (adev->flags & AMD_IS_APU) {
850
		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
851 852 853
	} else {
		u32 tmp = RREG32(mmMC_SEQ_MISC0);
		tmp &= MC_SEQ_MISC0__MT__MASK;
854
		adev->gmc.vram_type = gmc_v6_0_convert_vram_type(tmp);
855 856
	}

857
	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 146, &adev->gmc.vm_fault);
858 859 860
	if (r)
		return r;

861
	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_LEGACY, 147, &adev->gmc.vm_fault);
862 863 864
	if (r)
		return r;

865
	amdgpu_vm_adjust_size(adev, 64, 9, 1, 40);
866

867
	adev->gmc.mc_mask = 0xffffffffffULL;
868 869 870 871 872 873 874

	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;
875
		dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n");
876 877 878 879
	}
	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));
880
		dev_warn(adev->dev, "amdgpu: No coherent DMA available.\n");
881
	}
882
	adev->need_swiotlb = drm_get_max_iomem() > ((u64)1 << dma_bits);
883 884 885

	r = gmc_v6_0_init_microcode(adev);
	if (r) {
886
		dev_err(adev->dev, "Failed to load mc firmware!\n");
887 888 889 890 891 892 893
		return r;
	}

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

894 895
	adev->gmc.stolen_size = gmc_v6_0_get_vbios_fb_size(adev);

896 897 898 899 900 901 902 903
	r = amdgpu_bo_init(adev);
	if (r)
		return r;

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

904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920
	/*
	 * 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;
	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;
921 922
	}

923
	return 0;
924 925 926 927 928 929
}

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

930
	amdgpu_gem_force_release(adev);
931
	amdgpu_vm_manager_fini(adev);
932
	amdgpu_gart_table_vram_free(adev);
933
	amdgpu_bo_fini(adev);
934
	amdgpu_gart_fini(adev);
935 936
	release_firmware(adev->gmc.fw);
	adev->gmc.fw = NULL;
937 938 939 940 941 942 943 944 945 946 947 948 949 950

	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) {
951
			dev_err(adev->dev, "Failed to load MC firmware!\n");
952 953 954 955 956 957 958 959 960 961 962 963 964 965 966
			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;

967
	amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990
	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;

991
	amdgpu_vmid_reset_all(adev);
992

993
	return 0;
994 995 996 997 998
}

static bool gmc_v6_0_is_idle(void *handle)
{
	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
999
	u32 tmp = RREG32(mmSRBM_STATUS);
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013

	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++) {
1014
		if (gmc_v6_0_is_idle(handle))
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
			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;
1026
	u32 tmp = RREG32(mmSRBM_STATUS);
1027 1028 1029

	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
1030
						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
1031 1032 1033 1034 1035

	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,
1036
							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
1037 1038 1039
	}

	if (srbm_soft_reset) {
1040
		gmc_v6_0_mc_stop(adev);
1041 1042 1043 1044 1045
		if (gmc_v6_0_wait_for_idle(adev)) {
			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
		}


1046
		tmp = RREG32(mmSRBM_SOFT_RESET);
1047 1048
		tmp |= srbm_soft_reset;
		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
1049 1050
		WREG32(mmSRBM_SOFT_RESET, tmp);
		tmp = RREG32(mmSRBM_SOFT_RESET);
1051 1052 1053 1054

		udelay(50);

		tmp &= ~srbm_soft_reset;
1055 1056
		WREG32(mmSRBM_SOFT_RESET, tmp);
		tmp = RREG32(mmSRBM_SOFT_RESET);
1057 1058 1059

		udelay(50);

1060
		gmc_v6_0_mc_resume(adev);
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
		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:
1082
		tmp = RREG32(mmVM_CONTEXT0_CNTL);
1083
		tmp &= ~bits;
1084 1085
		WREG32(mmVM_CONTEXT0_CNTL, tmp);
		tmp = RREG32(mmVM_CONTEXT1_CNTL);
1086
		tmp &= ~bits;
1087
		WREG32(mmVM_CONTEXT1_CNTL, tmp);
1088 1089
		break;
	case AMDGPU_IRQ_STATE_ENABLE:
1090
		tmp = RREG32(mmVM_CONTEXT0_CNTL);
1091
		tmp |= bits;
1092 1093
		WREG32(mmVM_CONTEXT0_CNTL, tmp);
		tmp = RREG32(mmVM_CONTEXT1_CNTL);
1094
		tmp |= bits;
1095
		WREG32(mmVM_CONTEXT1_CNTL, tmp);
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
		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;

1110 1111 1112
	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
1113 1114 1115 1116 1117 1118 1119

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

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

1120 1121
	if (printk_ratelimit()) {
		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
1122
			entry->src_id, entry->src_data[0]);
1123 1124 1125 1126 1127 1128
		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);
	}
1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144

	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;
}

1145
static const struct amd_ip_funcs gmc_v6_0_ip_funcs = {
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161
	.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,
};

1162 1163
static const struct amdgpu_gmc_funcs gmc_v6_0_gmc_funcs = {
	.flush_gpu_tlb = gmc_v6_0_flush_gpu_tlb,
1164
	.emit_flush_gpu_tlb = gmc_v6_0_emit_flush_gpu_tlb,
1165
	.set_pte_pde = gmc_v6_0_set_pte_pde,
1166
	.set_prt = gmc_v6_0_set_prt,
1167
	.get_vm_pde = gmc_v6_0_get_vm_pde,
1168
	.get_vm_pte_flags = gmc_v6_0_get_vm_pte_flags
1169 1170 1171 1172 1173 1174 1175
};

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,
};

1176
static void gmc_v6_0_set_gmc_funcs(struct amdgpu_device *adev)
1177
{
1178 1179
	if (adev->gmc.gmc_funcs == NULL)
		adev->gmc.gmc_funcs = &gmc_v6_0_gmc_funcs;
1180 1181 1182 1183
}

static void gmc_v6_0_set_irq_funcs(struct amdgpu_device *adev)
{
1184 1185
	adev->gmc.vm_fault.num_types = 1;
	adev->gmc.vm_fault.funcs = &gmc_v6_0_irq_funcs;
1186 1187
}

1188 1189 1190 1191 1192 1193 1194 1195
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,
};