r600.c 111.9 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
/*
 * 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
 */
28
#include <linux/slab.h>
29 30 31
#include <linux/seq_file.h>
#include <linux/firmware.h>
#include <linux/platform_device.h>
32
#include <linux/module.h>
33
#include "drmP.h"
34
#include "radeon_drm.h"
35
#include "radeon.h"
36
#include "radeon_asic.h"
37 38 39
#include "radeon_mode.h"
#include "r600d.h"
#include "atom.h"
40
#include "avivod.h"
41

42 43
#define PFP_UCODE_SIZE 576
#define PM4_UCODE_SIZE 1792
44
#define RLC_UCODE_SIZE 768
45 46
#define R700_PFP_UCODE_SIZE 848
#define R700_PM4_UCODE_SIZE 1360
47
#define R700_RLC_UCODE_SIZE 1024
48 49
#define EVERGREEN_PFP_UCODE_SIZE 1120
#define EVERGREEN_PM4_UCODE_SIZE 1376
50
#define EVERGREEN_RLC_UCODE_SIZE 768
51
#define CAYMAN_RLC_UCODE_SIZE 1024
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

/* Firmware Names */
MODULE_FIRMWARE("radeon/R600_pfp.bin");
MODULE_FIRMWARE("radeon/R600_me.bin");
MODULE_FIRMWARE("radeon/RV610_pfp.bin");
MODULE_FIRMWARE("radeon/RV610_me.bin");
MODULE_FIRMWARE("radeon/RV630_pfp.bin");
MODULE_FIRMWARE("radeon/RV630_me.bin");
MODULE_FIRMWARE("radeon/RV620_pfp.bin");
MODULE_FIRMWARE("radeon/RV620_me.bin");
MODULE_FIRMWARE("radeon/RV635_pfp.bin");
MODULE_FIRMWARE("radeon/RV635_me.bin");
MODULE_FIRMWARE("radeon/RV670_pfp.bin");
MODULE_FIRMWARE("radeon/RV670_me.bin");
MODULE_FIRMWARE("radeon/RS780_pfp.bin");
MODULE_FIRMWARE("radeon/RS780_me.bin");
MODULE_FIRMWARE("radeon/RV770_pfp.bin");
MODULE_FIRMWARE("radeon/RV770_me.bin");
MODULE_FIRMWARE("radeon/RV730_pfp.bin");
MODULE_FIRMWARE("radeon/RV730_me.bin");
MODULE_FIRMWARE("radeon/RV710_pfp.bin");
MODULE_FIRMWARE("radeon/RV710_me.bin");
74 75
MODULE_FIRMWARE("radeon/R600_rlc.bin");
MODULE_FIRMWARE("radeon/R700_rlc.bin");
76 77
MODULE_FIRMWARE("radeon/CEDAR_pfp.bin");
MODULE_FIRMWARE("radeon/CEDAR_me.bin");
78
MODULE_FIRMWARE("radeon/CEDAR_rlc.bin");
79 80
MODULE_FIRMWARE("radeon/REDWOOD_pfp.bin");
MODULE_FIRMWARE("radeon/REDWOOD_me.bin");
81
MODULE_FIRMWARE("radeon/REDWOOD_rlc.bin");
82 83
MODULE_FIRMWARE("radeon/JUNIPER_pfp.bin");
MODULE_FIRMWARE("radeon/JUNIPER_me.bin");
84
MODULE_FIRMWARE("radeon/JUNIPER_rlc.bin");
85
MODULE_FIRMWARE("radeon/CYPRESS_pfp.bin");
86
MODULE_FIRMWARE("radeon/CYPRESS_me.bin");
87
MODULE_FIRMWARE("radeon/CYPRESS_rlc.bin");
88 89 90
MODULE_FIRMWARE("radeon/PALM_pfp.bin");
MODULE_FIRMWARE("radeon/PALM_me.bin");
MODULE_FIRMWARE("radeon/SUMO_rlc.bin");
91 92 93 94
MODULE_FIRMWARE("radeon/SUMO_pfp.bin");
MODULE_FIRMWARE("radeon/SUMO_me.bin");
MODULE_FIRMWARE("radeon/SUMO2_pfp.bin");
MODULE_FIRMWARE("radeon/SUMO2_me.bin");
95 96

int r600_debugfs_mc_info_init(struct radeon_device *rdev);
97

98
/* r600,rv610,rv630,rv620,rv635,rv670 */
99 100
int r600_mc_wait_for_idle(struct radeon_device *rdev);
void r600_gpu_init(struct radeon_device *rdev);
101
void r600_fini(struct radeon_device *rdev);
102
void r600_irq_disable(struct radeon_device *rdev);
103
static void r600_pcie_gen2_enable(struct radeon_device *rdev);
104

105
/* get temperature in millidegrees */
106
int rv6xx_get_temp(struct radeon_device *rdev)
107 108 109
{
	u32 temp = (RREG32(CG_THERMAL_STATUS) & ASIC_T_MASK) >>
		ASIC_T_SHIFT;
110
	int actual_temp = temp & 0xff;
111

112 113 114 115
	if (temp & 0x100)
		actual_temp -= 256;

	return actual_temp * 1000;
116 117
}

118
void r600_pm_get_dynpm_state(struct radeon_device *rdev)
119 120 121
{
	int i;

122 123
	rdev->pm.dynpm_can_upclock = true;
	rdev->pm.dynpm_can_downclock = true;
124 125 126 127 128 129 130 131

	/* power state array is low to high, default is first */
	if ((rdev->flags & RADEON_IS_IGP) || (rdev->family == CHIP_R600)) {
		int min_power_state_index = 0;

		if (rdev->pm.num_power_states > 2)
			min_power_state_index = 1;

132 133
		switch (rdev->pm.dynpm_planned_action) {
		case DYNPM_ACTION_MINIMUM:
134 135
			rdev->pm.requested_power_state_index = min_power_state_index;
			rdev->pm.requested_clock_mode_index = 0;
136
			rdev->pm.dynpm_can_downclock = false;
137
			break;
138
		case DYNPM_ACTION_DOWNCLOCK:
139 140
			if (rdev->pm.current_power_state_index == min_power_state_index) {
				rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index;
141
				rdev->pm.dynpm_can_downclock = false;
142 143 144
			} else {
				if (rdev->pm.active_crtc_count > 1) {
					for (i = 0; i < rdev->pm.num_power_states; i++) {
145
						if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY)
146 147 148 149 150 151 152 153 154 155
							continue;
						else if (i >= rdev->pm.current_power_state_index) {
							rdev->pm.requested_power_state_index =
								rdev->pm.current_power_state_index;
							break;
						} else {
							rdev->pm.requested_power_state_index = i;
							break;
						}
					}
156 157 158 159 160 161 162 163
				} else {
					if (rdev->pm.current_power_state_index == 0)
						rdev->pm.requested_power_state_index =
							rdev->pm.num_power_states - 1;
					else
						rdev->pm.requested_power_state_index =
							rdev->pm.current_power_state_index - 1;
				}
164 165
			}
			rdev->pm.requested_clock_mode_index = 0;
166 167 168 169 170 171 172
			/* don't use the power state if crtcs are active and no display flag is set */
			if ((rdev->pm.active_crtc_count > 0) &&
			    (rdev->pm.power_state[rdev->pm.requested_power_state_index].
			     clock_info[rdev->pm.requested_clock_mode_index].flags &
			     RADEON_PM_MODE_NO_DISPLAY)) {
				rdev->pm.requested_power_state_index++;
			}
173
			break;
174
		case DYNPM_ACTION_UPCLOCK:
175 176
			if (rdev->pm.current_power_state_index == (rdev->pm.num_power_states - 1)) {
				rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index;
177
				rdev->pm.dynpm_can_upclock = false;
178 179 180
			} else {
				if (rdev->pm.active_crtc_count > 1) {
					for (i = (rdev->pm.num_power_states - 1); i >= 0; i--) {
181
						if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY)
182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
							continue;
						else if (i <= rdev->pm.current_power_state_index) {
							rdev->pm.requested_power_state_index =
								rdev->pm.current_power_state_index;
							break;
						} else {
							rdev->pm.requested_power_state_index = i;
							break;
						}
					}
				} else
					rdev->pm.requested_power_state_index =
						rdev->pm.current_power_state_index + 1;
			}
			rdev->pm.requested_clock_mode_index = 0;
			break;
198
		case DYNPM_ACTION_DEFAULT:
199 200
			rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index;
			rdev->pm.requested_clock_mode_index = 0;
201
			rdev->pm.dynpm_can_upclock = false;
202
			break;
203
		case DYNPM_ACTION_NONE:
204 205 206 207 208 209 210 211 212 213 214 215
		default:
			DRM_ERROR("Requested mode for not defined action\n");
			return;
		}
	} else {
		/* XXX select a power state based on AC/DC, single/dualhead, etc. */
		/* for now just select the first power state and switch between clock modes */
		/* power state array is low to high, default is first (0) */
		if (rdev->pm.active_crtc_count > 1) {
			rdev->pm.requested_power_state_index = -1;
			/* start at 1 as we don't want the default mode */
			for (i = 1; i < rdev->pm.num_power_states; i++) {
216
				if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY)
217 218 219 220 221 222 223 224 225 226 227 228 229
					continue;
				else if ((rdev->pm.power_state[i].type == POWER_STATE_TYPE_PERFORMANCE) ||
					 (rdev->pm.power_state[i].type == POWER_STATE_TYPE_BATTERY)) {
					rdev->pm.requested_power_state_index = i;
					break;
				}
			}
			/* if nothing selected, grab the default state. */
			if (rdev->pm.requested_power_state_index == -1)
				rdev->pm.requested_power_state_index = 0;
		} else
			rdev->pm.requested_power_state_index = 1;

230 231
		switch (rdev->pm.dynpm_planned_action) {
		case DYNPM_ACTION_MINIMUM:
232
			rdev->pm.requested_clock_mode_index = 0;
233
			rdev->pm.dynpm_can_downclock = false;
234
			break;
235
		case DYNPM_ACTION_DOWNCLOCK:
236 237 238
			if (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index) {
				if (rdev->pm.current_clock_mode_index == 0) {
					rdev->pm.requested_clock_mode_index = 0;
239
					rdev->pm.dynpm_can_downclock = false;
240 241 242 243 244
				} else
					rdev->pm.requested_clock_mode_index =
						rdev->pm.current_clock_mode_index - 1;
			} else {
				rdev->pm.requested_clock_mode_index = 0;
245
				rdev->pm.dynpm_can_downclock = false;
246
			}
247 248 249 250 251 252 253
			/* don't use the power state if crtcs are active and no display flag is set */
			if ((rdev->pm.active_crtc_count > 0) &&
			    (rdev->pm.power_state[rdev->pm.requested_power_state_index].
			     clock_info[rdev->pm.requested_clock_mode_index].flags &
			     RADEON_PM_MODE_NO_DISPLAY)) {
				rdev->pm.requested_clock_mode_index++;
			}
254
			break;
255
		case DYNPM_ACTION_UPCLOCK:
256 257 258 259
			if (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index) {
				if (rdev->pm.current_clock_mode_index ==
				    (rdev->pm.power_state[rdev->pm.requested_power_state_index].num_clock_modes - 1)) {
					rdev->pm.requested_clock_mode_index = rdev->pm.current_clock_mode_index;
260
					rdev->pm.dynpm_can_upclock = false;
261 262 263 264 265 266
				} else
					rdev->pm.requested_clock_mode_index =
						rdev->pm.current_clock_mode_index + 1;
			} else {
				rdev->pm.requested_clock_mode_index =
					rdev->pm.power_state[rdev->pm.requested_power_state_index].num_clock_modes - 1;
267
				rdev->pm.dynpm_can_upclock = false;
268 269
			}
			break;
270
		case DYNPM_ACTION_DEFAULT:
271 272
			rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index;
			rdev->pm.requested_clock_mode_index = 0;
273
			rdev->pm.dynpm_can_upclock = false;
274
			break;
275
		case DYNPM_ACTION_NONE:
276 277 278 279 280 281
		default:
			DRM_ERROR("Requested mode for not defined action\n");
			return;
		}
	}

282
	DRM_DEBUG_DRIVER("Requested: e: %d m: %d p: %d\n",
283 284 285 286 287 288
		  rdev->pm.power_state[rdev->pm.requested_power_state_index].
		  clock_info[rdev->pm.requested_clock_mode_index].sclk,
		  rdev->pm.power_state[rdev->pm.requested_power_state_index].
		  clock_info[rdev->pm.requested_clock_mode_index].mclk,
		  rdev->pm.power_state[rdev->pm.requested_power_state_index].
		  pcie_lanes);
289 290
}

291 292 293 294 295 296 297 298 299 300 301 302 303
void rs780_pm_init_profile(struct radeon_device *rdev)
{
	if (rdev->pm.num_power_states == 2) {
		/* default */
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
		/* low sh */
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
304 305 306 307 308
		/* mid sh */
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
309 310 311 312 313 314 315 316 317 318
		/* high sh */
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
		/* low mh */
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
319 320 321 322 323
		/* mid mh */
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
		/* high mh */
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
	} else if (rdev->pm.num_power_states == 3) {
		/* default */
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
		/* low sh */
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
340 341 342 343 344
		/* mid sh */
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
345 346 347 348 349 350 351 352 353 354
		/* high sh */
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
		/* low mh */
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
355 356 357 358 359
		/* mid mh */
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
		/* high mh */
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 1;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
	} else {
		/* default */
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
		/* low sh */
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
376 377 378 379 380
		/* mid sh */
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
381 382 383 384 385 386 387 388 389 390
		/* high sh */
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 3;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
		/* low mh */
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
391 392 393 394 395
		/* mid mh */
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
396 397 398 399 400 401 402
		/* high mh */
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 2;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 3;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
	}
}
403

404 405
void r600_pm_init_profile(struct radeon_device *rdev)
{
406 407
	int idx;

408 409 410 411 412 413
	if (rdev->family == CHIP_R600) {
		/* XXX */
		/* default */
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
414
		rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0;
415 416 417 418
		/* low sh */
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
419
		rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
420 421 422 423 424
		/* mid sh */
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0;
425 426 427 428
		/* high sh */
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
429
		rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0;
430 431 432 433
		/* low mh */
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
434
		rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
435 436 437 438 439
		/* mid mh */
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
		rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0;
440 441 442 443
		/* high mh */
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
444
		rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0;
445 446 447 448 449 450 451 452
	} else {
		if (rdev->pm.num_power_states < 4) {
			/* default */
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 2;
			/* low sh */
453 454
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 1;
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 1;
455
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
456 457 458 459 460 461
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
			/* mid sh */
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 1;
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1;
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1;
462
			/* high sh */
463 464
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 1;
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 1;
465 466 467
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 2;
			/* low mh */
468 469
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 2;
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 2;
470
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
471 472 473 474 475 476
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
			/* low mh */
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 2;
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 2;
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1;
477
			/* high mh */
478 479
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 2;
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 2;
480 481 482 483 484 485 486 487 488
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 2;
		} else {
			/* default */
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index;
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index;
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 2;
			/* low sh */
489 490 491 492 493 494 495 496
			if (rdev->flags & RADEON_IS_MOBILITY)
				idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 0);
			else
				idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0);
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0;
497
			/* mid sh */
498 499 500 501
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1;
502
			/* high sh */
503 504 505
			idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0);
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = idx;
506 507 508
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 2;
			/* low mh */
509 510 511 512 513 514 515 516
			if (rdev->flags & RADEON_IS_MOBILITY)
				idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 1);
			else
				idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1);
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0;
517
			/* mid mh */
518 519 520 521
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1;
522
			/* high mh */
523 524 525
			idx = radeon_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1);
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = idx;
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = idx;
526 527 528 529
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0;
			rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 2;
		}
	}
530 531
}

532 533
void r600_pm_misc(struct radeon_device *rdev)
{
534 535 536 537
	int req_ps_idx = rdev->pm.requested_power_state_index;
	int req_cm_idx = rdev->pm.requested_clock_mode_index;
	struct radeon_power_state *ps = &rdev->pm.power_state[req_ps_idx];
	struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage;
538

539
	if ((voltage->type == VOLTAGE_SW) && voltage->voltage) {
540 541 542
		/* 0xff01 is a flag rather then an actual voltage */
		if (voltage->voltage == 0xff01)
			return;
543
		if (voltage->voltage != rdev->pm.current_vddc) {
544
			radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC);
545
			rdev->pm.current_vddc = voltage->voltage;
546
			DRM_DEBUG_DRIVER("Setting: v: %d\n", voltage->voltage);
547 548
		}
	}
549 550
}

551 552 553 554 555 556 557 558
bool r600_gui_idle(struct radeon_device *rdev)
{
	if (RREG32(GRBM_STATUS) & GUI_ACTIVE)
		return false;
	else
		return true;
}

559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
/* hpd for digital panel detect/disconnect */
bool r600_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd)
{
	bool connected = false;

	if (ASIC_IS_DCE3(rdev)) {
		switch (hpd) {
		case RADEON_HPD_1:
			if (RREG32(DC_HPD1_INT_STATUS) & DC_HPDx_SENSE)
				connected = true;
			break;
		case RADEON_HPD_2:
			if (RREG32(DC_HPD2_INT_STATUS) & DC_HPDx_SENSE)
				connected = true;
			break;
		case RADEON_HPD_3:
			if (RREG32(DC_HPD3_INT_STATUS) & DC_HPDx_SENSE)
				connected = true;
			break;
		case RADEON_HPD_4:
			if (RREG32(DC_HPD4_INT_STATUS) & DC_HPDx_SENSE)
				connected = true;
			break;
			/* DCE 3.2 */
		case RADEON_HPD_5:
			if (RREG32(DC_HPD5_INT_STATUS) & DC_HPDx_SENSE)
				connected = true;
			break;
		case RADEON_HPD_6:
			if (RREG32(DC_HPD6_INT_STATUS) & DC_HPDx_SENSE)
				connected = true;
			break;
		default:
			break;
		}
	} else {
		switch (hpd) {
		case RADEON_HPD_1:
			if (RREG32(DC_HOT_PLUG_DETECT1_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE)
				connected = true;
			break;
		case RADEON_HPD_2:
			if (RREG32(DC_HOT_PLUG_DETECT2_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE)
				connected = true;
			break;
		case RADEON_HPD_3:
			if (RREG32(DC_HOT_PLUG_DETECT3_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE)
				connected = true;
			break;
		default:
			break;
		}
	}
	return connected;
}

void r600_hpd_set_polarity(struct radeon_device *rdev,
616
			   enum radeon_hpd_id hpd)
617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711
{
	u32 tmp;
	bool connected = r600_hpd_sense(rdev, hpd);

	if (ASIC_IS_DCE3(rdev)) {
		switch (hpd) {
		case RADEON_HPD_1:
			tmp = RREG32(DC_HPD1_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HPDx_INT_POLARITY;
			else
				tmp |= DC_HPDx_INT_POLARITY;
			WREG32(DC_HPD1_INT_CONTROL, tmp);
			break;
		case RADEON_HPD_2:
			tmp = RREG32(DC_HPD2_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HPDx_INT_POLARITY;
			else
				tmp |= DC_HPDx_INT_POLARITY;
			WREG32(DC_HPD2_INT_CONTROL, tmp);
			break;
		case RADEON_HPD_3:
			tmp = RREG32(DC_HPD3_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HPDx_INT_POLARITY;
			else
				tmp |= DC_HPDx_INT_POLARITY;
			WREG32(DC_HPD3_INT_CONTROL, tmp);
			break;
		case RADEON_HPD_4:
			tmp = RREG32(DC_HPD4_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HPDx_INT_POLARITY;
			else
				tmp |= DC_HPDx_INT_POLARITY;
			WREG32(DC_HPD4_INT_CONTROL, tmp);
			break;
		case RADEON_HPD_5:
			tmp = RREG32(DC_HPD5_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HPDx_INT_POLARITY;
			else
				tmp |= DC_HPDx_INT_POLARITY;
			WREG32(DC_HPD5_INT_CONTROL, tmp);
			break;
			/* DCE 3.2 */
		case RADEON_HPD_6:
			tmp = RREG32(DC_HPD6_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HPDx_INT_POLARITY;
			else
				tmp |= DC_HPDx_INT_POLARITY;
			WREG32(DC_HPD6_INT_CONTROL, tmp);
			break;
		default:
			break;
		}
	} else {
		switch (hpd) {
		case RADEON_HPD_1:
			tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY;
			else
				tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY;
			WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp);
			break;
		case RADEON_HPD_2:
			tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY;
			else
				tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY;
			WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp);
			break;
		case RADEON_HPD_3:
			tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL);
			if (connected)
				tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY;
			else
				tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY;
			WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp);
			break;
		default:
			break;
		}
	}
}

void r600_hpd_init(struct radeon_device *rdev)
{
	struct drm_device *dev = rdev->ddev;
	struct drm_connector *connector;

712 713 714 715 716 717 718
	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
		struct radeon_connector *radeon_connector = to_radeon_connector(connector);

		if (ASIC_IS_DCE3(rdev)) {
			u32 tmp = DC_HPDx_CONNECTION_TIMER(0x9c4) | DC_HPDx_RX_INT_TIMER(0xfa);
			if (ASIC_IS_DCE32(rdev))
				tmp |= DC_HPDx_EN;
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

			switch (radeon_connector->hpd.hpd) {
			case RADEON_HPD_1:
				WREG32(DC_HPD1_CONTROL, tmp);
				rdev->irq.hpd[0] = true;
				break;
			case RADEON_HPD_2:
				WREG32(DC_HPD2_CONTROL, tmp);
				rdev->irq.hpd[1] = true;
				break;
			case RADEON_HPD_3:
				WREG32(DC_HPD3_CONTROL, tmp);
				rdev->irq.hpd[2] = true;
				break;
			case RADEON_HPD_4:
				WREG32(DC_HPD4_CONTROL, tmp);
				rdev->irq.hpd[3] = true;
				break;
				/* DCE 3.2 */
			case RADEON_HPD_5:
				WREG32(DC_HPD5_CONTROL, tmp);
				rdev->irq.hpd[4] = true;
				break;
			case RADEON_HPD_6:
				WREG32(DC_HPD6_CONTROL, tmp);
				rdev->irq.hpd[5] = true;
				break;
			default:
				break;
			}
749
		} else {
750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766
			switch (radeon_connector->hpd.hpd) {
			case RADEON_HPD_1:
				WREG32(DC_HOT_PLUG_DETECT1_CONTROL, DC_HOT_PLUG_DETECTx_EN);
				rdev->irq.hpd[0] = true;
				break;
			case RADEON_HPD_2:
				WREG32(DC_HOT_PLUG_DETECT2_CONTROL, DC_HOT_PLUG_DETECTx_EN);
				rdev->irq.hpd[1] = true;
				break;
			case RADEON_HPD_3:
				WREG32(DC_HOT_PLUG_DETECT3_CONTROL, DC_HOT_PLUG_DETECTx_EN);
				rdev->irq.hpd[2] = true;
				break;
			default:
				break;
			}
		}
767
		radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd);
768
	}
769 770
	if (rdev->irq.installed)
		r600_irq_set(rdev);
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833
}

void r600_hpd_fini(struct radeon_device *rdev)
{
	struct drm_device *dev = rdev->ddev;
	struct drm_connector *connector;

	if (ASIC_IS_DCE3(rdev)) {
		list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
			struct radeon_connector *radeon_connector = to_radeon_connector(connector);
			switch (radeon_connector->hpd.hpd) {
			case RADEON_HPD_1:
				WREG32(DC_HPD1_CONTROL, 0);
				rdev->irq.hpd[0] = false;
				break;
			case RADEON_HPD_2:
				WREG32(DC_HPD2_CONTROL, 0);
				rdev->irq.hpd[1] = false;
				break;
			case RADEON_HPD_3:
				WREG32(DC_HPD3_CONTROL, 0);
				rdev->irq.hpd[2] = false;
				break;
			case RADEON_HPD_4:
				WREG32(DC_HPD4_CONTROL, 0);
				rdev->irq.hpd[3] = false;
				break;
				/* DCE 3.2 */
			case RADEON_HPD_5:
				WREG32(DC_HPD5_CONTROL, 0);
				rdev->irq.hpd[4] = false;
				break;
			case RADEON_HPD_6:
				WREG32(DC_HPD6_CONTROL, 0);
				rdev->irq.hpd[5] = false;
				break;
			default:
				break;
			}
		}
	} else {
		list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
			struct radeon_connector *radeon_connector = to_radeon_connector(connector);
			switch (radeon_connector->hpd.hpd) {
			case RADEON_HPD_1:
				WREG32(DC_HOT_PLUG_DETECT1_CONTROL, 0);
				rdev->irq.hpd[0] = false;
				break;
			case RADEON_HPD_2:
				WREG32(DC_HOT_PLUG_DETECT2_CONTROL, 0);
				rdev->irq.hpd[1] = false;
				break;
			case RADEON_HPD_3:
				WREG32(DC_HOT_PLUG_DETECT3_CONTROL, 0);
				rdev->irq.hpd[2] = false;
				break;
			default:
				break;
			}
		}
	}
}

834
/*
835
 * R600 PCIE GART
836
 */
837 838 839 840 841
void r600_pcie_gart_tlb_flush(struct radeon_device *rdev)
{
	unsigned i;
	u32 tmp;

842
	/* flush hdp cache so updates hit vram */
843 844
	if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) &&
	    !(rdev->flags & RADEON_IS_AGP)) {
845
		void __iomem *ptr = (void *)rdev->gart.ptr;
846 847 848 849
		u32 tmp;

		/* r7xx hw bug.  write to HDP_DEBUG1 followed by fb read
		 * rather than write to HDP_REG_COHERENCY_FLUSH_CNTL
850 851
		 * This seems to cause problems on some AGP cards. Just use the old
		 * method for them.
852 853 854 855 856
		 */
		WREG32(HDP_DEBUG1, 0);
		tmp = readl((void __iomem *)ptr);
	} else
		WREG32(R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1);
857

858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
	WREG32(VM_CONTEXT0_INVALIDATION_LOW_ADDR, rdev->mc.gtt_start >> 12);
	WREG32(VM_CONTEXT0_INVALIDATION_HIGH_ADDR, (rdev->mc.gtt_end - 1) >> 12);
	WREG32(VM_CONTEXT0_REQUEST_RESPONSE, REQUEST_TYPE(1));
	for (i = 0; i < rdev->usec_timeout; i++) {
		/* read MC_STATUS */
		tmp = RREG32(VM_CONTEXT0_REQUEST_RESPONSE);
		tmp = (tmp & RESPONSE_TYPE_MASK) >> RESPONSE_TYPE_SHIFT;
		if (tmp == 2) {
			printk(KERN_WARNING "[drm] r600 flush TLB failed\n");
			return;
		}
		if (tmp) {
			return;
		}
		udelay(1);
	}
}

876
int r600_pcie_gart_init(struct radeon_device *rdev)
877
{
878
	int r;
879

880
	if (rdev->gart.robj) {
J
Joe Perches 已提交
881
		WARN(1, "R600 PCIE GART already initialized\n");
882 883
		return 0;
	}
884 885
	/* Initialize common gart structure */
	r = radeon_gart_init(rdev);
886
	if (r)
887 888
		return r;
	rdev->gart.table_size = rdev->gart.num_gpu_pages * 8;
889 890 891 892 893 894 895 896
	return radeon_gart_table_vram_alloc(rdev);
}

int r600_pcie_gart_enable(struct radeon_device *rdev)
{
	u32 tmp;
	int r, i;

897
	if (rdev->gart.robj == NULL) {
898 899
		dev_err(rdev->dev, "No VRAM object for PCIE GART.\n");
		return -EINVAL;
900
	}
901 902 903
	r = radeon_gart_table_vram_pin(rdev);
	if (r)
		return r;
904
	radeon_gart_restore(rdev);
905

906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931
	/* Setup L2 cache */
	WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING |
				ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE |
				EFFECTIVE_L2_QUEUE_SIZE(7));
	WREG32(VM_L2_CNTL2, 0);
	WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1));
	/* Setup TLB control */
	tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING |
		SYSTEM_ACCESS_MODE_NOT_IN_SYS |
		EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) |
		ENABLE_WAIT_L2_QUERY;
	WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING);
	WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
	WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
	WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12);
932
	WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12);
933 934 935 936 937 938 939
	WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12);
	WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) |
				RANGE_PROTECTION_FAULT_ENABLE_DEFAULT);
	WREG32(VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
			(u32)(rdev->dummy_page.addr >> 12));
	for (i = 1; i < 7; i++)
		WREG32(VM_CONTEXT0_CNTL + (i * 4), 0);
940

941
	r600_pcie_gart_tlb_flush(rdev);
942 943 944
	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
		 (unsigned)(rdev->mc.gtt_size >> 20),
		 (unsigned long long)rdev->gart.table_addr);
945
	rdev->gart.ready = true;
946 947 948
	return 0;
}

949
void r600_pcie_gart_disable(struct radeon_device *rdev)
950
{
951
	u32 tmp;
952
	int i;
953

954 955 956
	/* Disable all tables */
	for (i = 0; i < 7; i++)
		WREG32(VM_CONTEXT0_CNTL + (i * 4), 0);
957

958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978
	/* Disable L2 cache */
	WREG32(VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING |
				EFFECTIVE_L2_QUEUE_SIZE(7));
	WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1));
	/* Setup L1 TLB control */
	tmp = EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) |
		ENABLE_WAIT_L2_QUERY;
	WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp);
979
	radeon_gart_table_vram_unpin(rdev);
980 981 982 983
}

void r600_pcie_gart_fini(struct radeon_device *rdev)
{
984
	radeon_gart_fini(rdev);
985 986
	r600_pcie_gart_disable(rdev);
	radeon_gart_table_vram_free(rdev);
987 988
}

989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
void r600_agp_enable(struct radeon_device *rdev)
{
	u32 tmp;
	int i;

	/* Setup L2 cache */
	WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING |
				ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE |
				EFFECTIVE_L2_QUEUE_SIZE(7));
	WREG32(VM_L2_CNTL2, 0);
	WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1));
	/* Setup TLB control */
	tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING |
		SYSTEM_ACCESS_MODE_NOT_IN_SYS |
		EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) |
		ENABLE_WAIT_L2_QUERY;
	WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING);
	WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp);
	WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
	WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE);
	for (i = 0; i < 7; i++)
		WREG32(VM_CONTEXT0_CNTL + (i * 4), 0);
}

1023 1024
int r600_mc_wait_for_idle(struct radeon_device *rdev)
{
1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
	unsigned i;
	u32 tmp;

	for (i = 0; i < rdev->usec_timeout; i++) {
		/* read MC_STATUS */
		tmp = RREG32(R_000E50_SRBM_STATUS) & 0x3F00;
		if (!tmp)
			return 0;
		udelay(1);
	}
	return -1;
1036 1037
}

1038
static void r600_mc_program(struct radeon_device *rdev)
1039
{
1040
	struct rv515_mc_save save;
1041 1042
	u32 tmp;
	int i, j;
1043

1044 1045 1046 1047 1048 1049 1050 1051 1052
	/* Initialize HDP */
	for (i = 0, j = 0; i < 32; i++, j += 0x18) {
		WREG32((0x2c14 + j), 0x00000000);
		WREG32((0x2c18 + j), 0x00000000);
		WREG32((0x2c1c + j), 0x00000000);
		WREG32((0x2c20 + j), 0x00000000);
		WREG32((0x2c24 + j), 0x00000000);
	}
	WREG32(HDP_REG_COHERENCY_FLUSH_CNTL, 0);
1053

1054
	rv515_mc_stop(rdev, &save);
1055
	if (r600_mc_wait_for_idle(rdev)) {
1056
		dev_warn(rdev->dev, "Wait for MC idle timedout !\n");
1057
	}
1058
	/* Lockout access through VGA aperture (doesn't exist before R600) */
1059 1060
	WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE);
	/* Update configuration */
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
	if (rdev->flags & RADEON_IS_AGP) {
		if (rdev->mc.vram_start < rdev->mc.gtt_start) {
			/* VRAM before AGP */
			WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR,
				rdev->mc.vram_start >> 12);
			WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
				rdev->mc.gtt_end >> 12);
		} else {
			/* VRAM after AGP */
			WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR,
				rdev->mc.gtt_start >> 12);
			WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR,
				rdev->mc.vram_end >> 12);
		}
	} else {
		WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start >> 12);
		WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, rdev->mc.vram_end >> 12);
	}
1079
	WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, rdev->vram_scratch.gpu_addr >> 12);
1080
	tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16;
1081 1082 1083 1084
	tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF);
	WREG32(MC_VM_FB_LOCATION, tmp);
	WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8));
	WREG32(HDP_NONSURFACE_INFO, (2 << 7));
1085
	WREG32(HDP_NONSURFACE_SIZE, 0x3FFFFFFF);
1086
	if (rdev->flags & RADEON_IS_AGP) {
1087 1088
		WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 22);
		WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 22);
1089 1090 1091 1092 1093 1094 1095
		WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22);
	} else {
		WREG32(MC_VM_AGP_BASE, 0);
		WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF);
		WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF);
	}
	if (r600_mc_wait_for_idle(rdev)) {
1096
		dev_warn(rdev->dev, "Wait for MC idle timedout !\n");
1097
	}
1098
	rv515_mc_resume(rdev, &save);
1099 1100
	/* we need to own VRAM, so turn off the VGA renderer here
	 * to stop it overwriting our objects */
1101
	rv515_vga_render_disable(rdev);
1102 1103
}

1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
/**
 * r600_vram_gtt_location - try to find VRAM & GTT location
 * @rdev: radeon device structure holding all necessary informations
 * @mc: memory controller structure holding memory informations
 *
 * Function will place try to place VRAM at same place as in CPU (PCI)
 * address space as some GPU seems to have issue when we reprogram at
 * different address space.
 *
 * If there is not enough space to fit the unvisible VRAM after the
 * aperture then we limit the VRAM size to the aperture.
 *
 * If we are using AGP then place VRAM adjacent to AGP aperture are we need
 * them to be in one from GPU point of view so that we can program GPU to
 * catch access outside them (weird GPU policy see ??).
 *
 * This function will never fails, worst case are limiting VRAM or GTT.
 *
 * Note: GTT start, end, size should be initialized before calling this
 * function on AGP platform.
 */
1125
static void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc)
1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
{
	u64 size_bf, size_af;

	if (mc->mc_vram_size > 0xE0000000) {
		/* leave room for at least 512M GTT */
		dev_warn(rdev->dev, "limiting VRAM\n");
		mc->real_vram_size = 0xE0000000;
		mc->mc_vram_size = 0xE0000000;
	}
	if (rdev->flags & RADEON_IS_AGP) {
		size_bf = mc->gtt_start;
		size_af = 0xFFFFFFFF - mc->gtt_end + 1;
		if (size_bf > size_af) {
			if (mc->mc_vram_size > size_bf) {
				dev_warn(rdev->dev, "limiting VRAM\n");
				mc->real_vram_size = size_bf;
				mc->mc_vram_size = size_bf;
			}
			mc->vram_start = mc->gtt_start - mc->mc_vram_size;
		} else {
			if (mc->mc_vram_size > size_af) {
				dev_warn(rdev->dev, "limiting VRAM\n");
				mc->real_vram_size = size_af;
				mc->mc_vram_size = size_af;
			}
			mc->vram_start = mc->gtt_end;
		}
		mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
		dev_info(rdev->dev, "VRAM: %lluM 0x%08llX - 0x%08llX (%lluM used)\n",
				mc->mc_vram_size >> 20, mc->vram_start,
				mc->vram_end, mc->real_vram_size >> 20);
	} else {
		u64 base = 0;
1159 1160 1161 1162
		if (rdev->flags & RADEON_IS_IGP) {
			base = RREG32(MC_VM_FB_LOCATION) & 0xFFFF;
			base <<= 24;
		}
1163
		radeon_vram_location(rdev, &rdev->mc, base);
1164
		rdev->mc.gtt_base_align = 0;
1165 1166 1167 1168
		radeon_gtt_location(rdev, mc);
	}
}

1169
int r600_mc_init(struct radeon_device *rdev)
1170
{
1171
	u32 tmp;
1172
	int chansize, numchan;
1173

1174
	/* Get VRAM informations */
1175
	rdev->mc.vram_is_ddr = true;
1176 1177
	tmp = RREG32(RAMCFG);
	if (tmp & CHANSIZE_OVERRIDE) {
1178
		chansize = 16;
1179
	} else if (tmp & CHANSIZE_MASK) {
1180 1181 1182 1183
		chansize = 64;
	} else {
		chansize = 32;
	}
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
	tmp = RREG32(CHMAP);
	switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) {
	case 0:
	default:
		numchan = 1;
		break;
	case 1:
		numchan = 2;
		break;
	case 2:
		numchan = 4;
		break;
	case 3:
		numchan = 8;
		break;
1199
	}
1200
	rdev->mc.vram_width = numchan * chansize;
1201
	/* Could aper size report 0 ? */
1202 1203
	rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0);
	rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0);
1204 1205 1206
	/* Setup GPU memory space */
	rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE);
	rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE);
1207
	rdev->mc.visible_vram_size = rdev->mc.aper_size;
1208
	r600_vram_gtt_location(rdev, &rdev->mc);
1209

1210 1211
	if (rdev->flags & RADEON_IS_IGP) {
		rs690_pm_info(rdev);
1212
		rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev);
1213
	}
1214
	radeon_update_bandwidth_info(rdev);
1215
	return 0;
1216 1217
}

1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264
int r600_vram_scratch_init(struct radeon_device *rdev)
{
	int r;

	if (rdev->vram_scratch.robj == NULL) {
		r = radeon_bo_create(rdev, RADEON_GPU_PAGE_SIZE,
				     PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
				     &rdev->vram_scratch.robj);
		if (r) {
			return r;
		}
	}

	r = radeon_bo_reserve(rdev->vram_scratch.robj, false);
	if (unlikely(r != 0))
		return r;
	r = radeon_bo_pin(rdev->vram_scratch.robj,
			  RADEON_GEM_DOMAIN_VRAM, &rdev->vram_scratch.gpu_addr);
	if (r) {
		radeon_bo_unreserve(rdev->vram_scratch.robj);
		return r;
	}
	r = radeon_bo_kmap(rdev->vram_scratch.robj,
				(void **)&rdev->vram_scratch.ptr);
	if (r)
		radeon_bo_unpin(rdev->vram_scratch.robj);
	radeon_bo_unreserve(rdev->vram_scratch.robj);

	return r;
}

void r600_vram_scratch_fini(struct radeon_device *rdev)
{
	int r;

	if (rdev->vram_scratch.robj == NULL) {
		return;
	}
	r = radeon_bo_reserve(rdev->vram_scratch.robj, false);
	if (likely(r == 0)) {
		radeon_bo_kunmap(rdev->vram_scratch.robj);
		radeon_bo_unpin(rdev->vram_scratch.robj);
		radeon_bo_unreserve(rdev->vram_scratch.robj);
	}
	radeon_bo_unref(&rdev->vram_scratch.robj);
}

1265 1266 1267 1268 1269
/* We doesn't check that the GPU really needs a reset we simply do the
 * reset, it's up to the caller to determine if the GPU needs one. We
 * might add an helper function to check that.
 */
int r600_gpu_soft_reset(struct radeon_device *rdev)
1270
{
1271
	struct rv515_mc_save save;
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287
	u32 grbm_busy_mask = S_008010_VC_BUSY(1) | S_008010_VGT_BUSY_NO_DMA(1) |
				S_008010_VGT_BUSY(1) | S_008010_TA03_BUSY(1) |
				S_008010_TC_BUSY(1) | S_008010_SX_BUSY(1) |
				S_008010_SH_BUSY(1) | S_008010_SPI03_BUSY(1) |
				S_008010_SMX_BUSY(1) | S_008010_SC_BUSY(1) |
				S_008010_PA_BUSY(1) | S_008010_DB03_BUSY(1) |
				S_008010_CR_BUSY(1) | S_008010_CB03_BUSY(1) |
				S_008010_GUI_ACTIVE(1);
	u32 grbm2_busy_mask = S_008014_SPI0_BUSY(1) | S_008014_SPI1_BUSY(1) |
			S_008014_SPI2_BUSY(1) | S_008014_SPI3_BUSY(1) |
			S_008014_TA0_BUSY(1) | S_008014_TA1_BUSY(1) |
			S_008014_TA2_BUSY(1) | S_008014_TA3_BUSY(1) |
			S_008014_DB0_BUSY(1) | S_008014_DB1_BUSY(1) |
			S_008014_DB2_BUSY(1) | S_008014_DB3_BUSY(1) |
			S_008014_CB0_BUSY(1) | S_008014_CB1_BUSY(1) |
			S_008014_CB2_BUSY(1) | S_008014_CB3_BUSY(1);
1288
	u32 tmp;
1289

1290 1291 1292
	if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE))
		return 0;

1293 1294 1295 1296
	dev_info(rdev->dev, "GPU softreset \n");
	dev_info(rdev->dev, "  R_008010_GRBM_STATUS=0x%08X\n",
		RREG32(R_008010_GRBM_STATUS));
	dev_info(rdev->dev, "  R_008014_GRBM_STATUS2=0x%08X\n",
1297
		RREG32(R_008014_GRBM_STATUS2));
1298 1299
	dev_info(rdev->dev, "  R_000E50_SRBM_STATUS=0x%08X\n",
		RREG32(R_000E50_SRBM_STATUS));
1300 1301 1302 1303
	rv515_mc_stop(rdev, &save);
	if (r600_mc_wait_for_idle(rdev)) {
		dev_warn(rdev->dev, "Wait for MC idle timedout !\n");
	}
1304
	/* Disable CP parsing/prefetching */
1305
	WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1));
1306 1307 1308
	/* Check if any of the rendering block is busy and reset it */
	if ((RREG32(R_008010_GRBM_STATUS) & grbm_busy_mask) ||
	    (RREG32(R_008014_GRBM_STATUS2) & grbm2_busy_mask)) {
1309
		tmp = S_008020_SOFT_RESET_CR(1) |
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320
			S_008020_SOFT_RESET_DB(1) |
			S_008020_SOFT_RESET_CB(1) |
			S_008020_SOFT_RESET_PA(1) |
			S_008020_SOFT_RESET_SC(1) |
			S_008020_SOFT_RESET_SMX(1) |
			S_008020_SOFT_RESET_SPI(1) |
			S_008020_SOFT_RESET_SX(1) |
			S_008020_SOFT_RESET_SH(1) |
			S_008020_SOFT_RESET_TC(1) |
			S_008020_SOFT_RESET_TA(1) |
			S_008020_SOFT_RESET_VC(1) |
1321
			S_008020_SOFT_RESET_VGT(1);
1322
		dev_info(rdev->dev, "  R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp);
1323
		WREG32(R_008020_GRBM_SOFT_RESET, tmp);
1324 1325
		RREG32(R_008020_GRBM_SOFT_RESET);
		mdelay(15);
1326 1327 1328
		WREG32(R_008020_GRBM_SOFT_RESET, 0);
	}
	/* Reset CP (we always reset CP) */
1329 1330 1331
	tmp = S_008020_SOFT_RESET_CP(1);
	dev_info(rdev->dev, "R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp);
	WREG32(R_008020_GRBM_SOFT_RESET, tmp);
1332 1333
	RREG32(R_008020_GRBM_SOFT_RESET);
	mdelay(15);
1334 1335
	WREG32(R_008020_GRBM_SOFT_RESET, 0);
	/* Wait a little for things to settle down */
1336
	mdelay(1);
1337 1338 1339 1340 1341 1342
	dev_info(rdev->dev, "  R_008010_GRBM_STATUS=0x%08X\n",
		RREG32(R_008010_GRBM_STATUS));
	dev_info(rdev->dev, "  R_008014_GRBM_STATUS2=0x%08X\n",
		RREG32(R_008014_GRBM_STATUS2));
	dev_info(rdev->dev, "  R_000E50_SRBM_STATUS=0x%08X\n",
		RREG32(R_000E50_SRBM_STATUS));
1343
	rv515_mc_resume(rdev, &save);
1344 1345 1346
	return 0;
}

1347 1348 1349 1350 1351
bool r600_gpu_is_lockup(struct radeon_device *rdev)
{
	u32 srbm_status;
	u32 grbm_status;
	u32 grbm_status2;
1352
	struct r100_gpu_lockup *lockup;
1353 1354
	int r;

1355 1356 1357 1358 1359
	if (rdev->family >= CHIP_RV770)
		lockup = &rdev->config.rv770.lockup;
	else
		lockup = &rdev->config.r600.lockup;

1360 1361 1362 1363
	srbm_status = RREG32(R_000E50_SRBM_STATUS);
	grbm_status = RREG32(R_008010_GRBM_STATUS);
	grbm_status2 = RREG32(R_008014_GRBM_STATUS2);
	if (!G_008010_GUI_ACTIVE(grbm_status)) {
1364
		r100_gpu_lockup_update(lockup, &rdev->cp);
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
		return false;
	}
	/* force CP activities */
	r = radeon_ring_lock(rdev, 2);
	if (!r) {
		/* PACKET2 NOP */
		radeon_ring_write(rdev, 0x80000000);
		radeon_ring_write(rdev, 0x80000000);
		radeon_ring_unlock_commit(rdev);
	}
	rdev->cp.rptr = RREG32(R600_CP_RB_RPTR);
1376
	return r100_gpu_cp_is_lockup(rdev, lockup, &rdev->cp);
1377 1378
}

1379
int r600_asic_reset(struct radeon_device *rdev)
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501
{
	return r600_gpu_soft_reset(rdev);
}

static u32 r600_get_tile_pipe_to_backend_map(u32 num_tile_pipes,
					     u32 num_backends,
					     u32 backend_disable_mask)
{
	u32 backend_map = 0;
	u32 enabled_backends_mask;
	u32 enabled_backends_count;
	u32 cur_pipe;
	u32 swizzle_pipe[R6XX_MAX_PIPES];
	u32 cur_backend;
	u32 i;

	if (num_tile_pipes > R6XX_MAX_PIPES)
		num_tile_pipes = R6XX_MAX_PIPES;
	if (num_tile_pipes < 1)
		num_tile_pipes = 1;
	if (num_backends > R6XX_MAX_BACKENDS)
		num_backends = R6XX_MAX_BACKENDS;
	if (num_backends < 1)
		num_backends = 1;

	enabled_backends_mask = 0;
	enabled_backends_count = 0;
	for (i = 0; i < R6XX_MAX_BACKENDS; ++i) {
		if (((backend_disable_mask >> i) & 1) == 0) {
			enabled_backends_mask |= (1 << i);
			++enabled_backends_count;
		}
		if (enabled_backends_count == num_backends)
			break;
	}

	if (enabled_backends_count == 0) {
		enabled_backends_mask = 1;
		enabled_backends_count = 1;
	}

	if (enabled_backends_count != num_backends)
		num_backends = enabled_backends_count;

	memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * R6XX_MAX_PIPES);
	switch (num_tile_pipes) {
	case 1:
		swizzle_pipe[0] = 0;
		break;
	case 2:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 1;
		break;
	case 3:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 1;
		swizzle_pipe[2] = 2;
		break;
	case 4:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 1;
		swizzle_pipe[2] = 2;
		swizzle_pipe[3] = 3;
		break;
	case 5:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 1;
		swizzle_pipe[2] = 2;
		swizzle_pipe[3] = 3;
		swizzle_pipe[4] = 4;
		break;
	case 6:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 2;
		swizzle_pipe[2] = 4;
		swizzle_pipe[3] = 5;
		swizzle_pipe[4] = 1;
		swizzle_pipe[5] = 3;
		break;
	case 7:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 2;
		swizzle_pipe[2] = 4;
		swizzle_pipe[3] = 6;
		swizzle_pipe[4] = 1;
		swizzle_pipe[5] = 3;
		swizzle_pipe[6] = 5;
		break;
	case 8:
		swizzle_pipe[0] = 0;
		swizzle_pipe[1] = 2;
		swizzle_pipe[2] = 4;
		swizzle_pipe[3] = 6;
		swizzle_pipe[4] = 1;
		swizzle_pipe[5] = 3;
		swizzle_pipe[6] = 5;
		swizzle_pipe[7] = 7;
		break;
	}

	cur_backend = 0;
	for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) {
		while (((1 << cur_backend) & enabled_backends_mask) == 0)
			cur_backend = (cur_backend + 1) % R6XX_MAX_BACKENDS;

		backend_map |= (u32)(((cur_backend & 3) << (swizzle_pipe[cur_pipe] * 2)));

		cur_backend = (cur_backend + 1) % R6XX_MAX_BACKENDS;
	}

	return backend_map;
}

int r600_count_pipe_bits(uint32_t val)
{
	int i, ret = 0;

	for (i = 0; i < 32; i++) {
		ret += val & 1;
		val >>= 1;
	}
	return ret;
1502 1503
}

1504 1505 1506 1507
void r600_gpu_init(struct radeon_device *rdev)
{
	u32 tiling_config;
	u32 ramcfg;
1508 1509 1510
	u32 backend_map;
	u32 cc_rb_backend_disable;
	u32 cc_gc_shader_pipe_config;
1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619
	u32 tmp;
	int i, j;
	u32 sq_config;
	u32 sq_gpr_resource_mgmt_1 = 0;
	u32 sq_gpr_resource_mgmt_2 = 0;
	u32 sq_thread_resource_mgmt = 0;
	u32 sq_stack_resource_mgmt_1 = 0;
	u32 sq_stack_resource_mgmt_2 = 0;

	/* FIXME: implement */
	switch (rdev->family) {
	case CHIP_R600:
		rdev->config.r600.max_pipes = 4;
		rdev->config.r600.max_tile_pipes = 8;
		rdev->config.r600.max_simds = 4;
		rdev->config.r600.max_backends = 4;
		rdev->config.r600.max_gprs = 256;
		rdev->config.r600.max_threads = 192;
		rdev->config.r600.max_stack_entries = 256;
		rdev->config.r600.max_hw_contexts = 8;
		rdev->config.r600.max_gs_threads = 16;
		rdev->config.r600.sx_max_export_size = 128;
		rdev->config.r600.sx_max_export_pos_size = 16;
		rdev->config.r600.sx_max_export_smx_size = 128;
		rdev->config.r600.sq_num_cf_insts = 2;
		break;
	case CHIP_RV630:
	case CHIP_RV635:
		rdev->config.r600.max_pipes = 2;
		rdev->config.r600.max_tile_pipes = 2;
		rdev->config.r600.max_simds = 3;
		rdev->config.r600.max_backends = 1;
		rdev->config.r600.max_gprs = 128;
		rdev->config.r600.max_threads = 192;
		rdev->config.r600.max_stack_entries = 128;
		rdev->config.r600.max_hw_contexts = 8;
		rdev->config.r600.max_gs_threads = 4;
		rdev->config.r600.sx_max_export_size = 128;
		rdev->config.r600.sx_max_export_pos_size = 16;
		rdev->config.r600.sx_max_export_smx_size = 128;
		rdev->config.r600.sq_num_cf_insts = 2;
		break;
	case CHIP_RV610:
	case CHIP_RV620:
	case CHIP_RS780:
	case CHIP_RS880:
		rdev->config.r600.max_pipes = 1;
		rdev->config.r600.max_tile_pipes = 1;
		rdev->config.r600.max_simds = 2;
		rdev->config.r600.max_backends = 1;
		rdev->config.r600.max_gprs = 128;
		rdev->config.r600.max_threads = 192;
		rdev->config.r600.max_stack_entries = 128;
		rdev->config.r600.max_hw_contexts = 4;
		rdev->config.r600.max_gs_threads = 4;
		rdev->config.r600.sx_max_export_size = 128;
		rdev->config.r600.sx_max_export_pos_size = 16;
		rdev->config.r600.sx_max_export_smx_size = 128;
		rdev->config.r600.sq_num_cf_insts = 1;
		break;
	case CHIP_RV670:
		rdev->config.r600.max_pipes = 4;
		rdev->config.r600.max_tile_pipes = 4;
		rdev->config.r600.max_simds = 4;
		rdev->config.r600.max_backends = 4;
		rdev->config.r600.max_gprs = 192;
		rdev->config.r600.max_threads = 192;
		rdev->config.r600.max_stack_entries = 256;
		rdev->config.r600.max_hw_contexts = 8;
		rdev->config.r600.max_gs_threads = 16;
		rdev->config.r600.sx_max_export_size = 128;
		rdev->config.r600.sx_max_export_pos_size = 16;
		rdev->config.r600.sx_max_export_smx_size = 128;
		rdev->config.r600.sq_num_cf_insts = 2;
		break;
	default:
		break;
	}

	/* Initialize HDP */
	for (i = 0, j = 0; i < 32; i++, j += 0x18) {
		WREG32((0x2c14 + j), 0x00000000);
		WREG32((0x2c18 + j), 0x00000000);
		WREG32((0x2c1c + j), 0x00000000);
		WREG32((0x2c20 + j), 0x00000000);
		WREG32((0x2c24 + j), 0x00000000);
	}

	WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff));

	/* Setup tiling */
	tiling_config = 0;
	ramcfg = RREG32(RAMCFG);
	switch (rdev->config.r600.max_tile_pipes) {
	case 1:
		tiling_config |= PIPE_TILING(0);
		break;
	case 2:
		tiling_config |= PIPE_TILING(1);
		break;
	case 4:
		tiling_config |= PIPE_TILING(2);
		break;
	case 8:
		tiling_config |= PIPE_TILING(3);
		break;
	default:
		break;
	}
1620
	rdev->config.r600.tiling_npipes = rdev->config.r600.max_tile_pipes;
1621
	rdev->config.r600.tiling_nbanks = 4 << ((ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT);
1622
	tiling_config |= BANK_TILING((ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT);
1623 1624 1625 1626 1627
	tiling_config |= GROUP_SIZE((ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT);
	if ((ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT)
		rdev->config.r600.tiling_group_size = 512;
	else
		rdev->config.r600.tiling_group_size = 256;
1628 1629 1630 1631 1632 1633 1634 1635 1636
	tmp = (ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT;
	if (tmp > 3) {
		tiling_config |= ROW_TILING(3);
		tiling_config |= SAMPLE_SPLIT(3);
	} else {
		tiling_config |= ROW_TILING(tmp);
		tiling_config |= SAMPLE_SPLIT(tmp);
	}
	tiling_config |= BANK_SWAPS(1);
1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652

	cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE) & 0x00ff0000;
	cc_rb_backend_disable |=
		BACKEND_DISABLE((R6XX_MAX_BACKENDS_MASK << rdev->config.r600.max_backends) & R6XX_MAX_BACKENDS_MASK);

	cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG) & 0xffffff00;
	cc_gc_shader_pipe_config |=
		INACTIVE_QD_PIPES((R6XX_MAX_PIPES_MASK << rdev->config.r600.max_pipes) & R6XX_MAX_PIPES_MASK);
	cc_gc_shader_pipe_config |=
		INACTIVE_SIMDS((R6XX_MAX_SIMDS_MASK << rdev->config.r600.max_simds) & R6XX_MAX_SIMDS_MASK);

	backend_map = r600_get_tile_pipe_to_backend_map(rdev->config.r600.max_tile_pipes,
							(R6XX_MAX_BACKENDS -
							 r600_count_pipe_bits((cc_rb_backend_disable &
									       R6XX_MAX_BACKENDS_MASK) >> 16)),
							(cc_rb_backend_disable >> 16));
1653
	rdev->config.r600.tile_config = tiling_config;
1654
	rdev->config.r600.backend_map = backend_map;
1655
	tiling_config |= BACKEND_MAP(backend_map);
1656 1657 1658 1659 1660
	WREG32(GB_TILING_CONFIG, tiling_config);
	WREG32(DCP_TILING_CONFIG, tiling_config & 0xffff);
	WREG32(HDP_TILING_CONFIG, tiling_config & 0xffff);

	/* Setup pipes */
1661 1662
	WREG32(CC_RB_BACKEND_DISABLE, cc_rb_backend_disable);
	WREG32(CC_GC_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config);
1663
	WREG32(GC_USER_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config);
1664

1665
	tmp = R6XX_MAX_PIPES - r600_count_pipe_bits((cc_gc_shader_pipe_config & INACTIVE_QD_PIPES_MASK) >> 8);
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688
	WREG32(VGT_OUT_DEALLOC_CNTL, (tmp * 4) & DEALLOC_DIST_MASK);
	WREG32(VGT_VERTEX_REUSE_BLOCK_CNTL, ((tmp * 4) - 2) & VTX_REUSE_DEPTH_MASK);

	/* Setup some CP states */
	WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | ROQ_IB2_START(0x2b)));
	WREG32(CP_MEQ_THRESHOLDS, (MEQ_END(0x40) | ROQ_END(0x40)));

	WREG32(TA_CNTL_AUX, (DISABLE_CUBE_ANISO | SYNC_GRADIENT |
			     SYNC_WALKER | SYNC_ALIGNER));
	/* Setup various GPU states */
	if (rdev->family == CHIP_RV670)
		WREG32(ARB_GDEC_RD_CNTL, 0x00000021);

	tmp = RREG32(SX_DEBUG_1);
	tmp |= SMX_EVENT_RELEASE;
	if ((rdev->family > CHIP_R600))
		tmp |= ENABLE_NEW_SMX_ADDRESS;
	WREG32(SX_DEBUG_1, tmp);

	if (((rdev->family) == CHIP_R600) ||
	    ((rdev->family) == CHIP_RV630) ||
	    ((rdev->family) == CHIP_RV610) ||
	    ((rdev->family) == CHIP_RV620) ||
1689 1690
	    ((rdev->family) == CHIP_RS780) ||
	    ((rdev->family) == CHIP_RS880)) {
1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
		WREG32(DB_DEBUG, PREZ_MUST_WAIT_FOR_POSTZ_DONE);
	} else {
		WREG32(DB_DEBUG, 0);
	}
	WREG32(DB_WATERMARKS, (DEPTH_FREE(4) | DEPTH_CACHELINE_FREE(16) |
			       DEPTH_FLUSH(16) | DEPTH_PENDING_FREE(4)));

	WREG32(PA_SC_MULTI_CHIP_CNTL, 0);
	WREG32(VGT_NUM_INSTANCES, 0);

	WREG32(SPI_CONFIG_CNTL, GPR_WRITE_PRIORITY(0));
	WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(0));

	tmp = RREG32(SQ_MS_FIFO_SIZES);
	if (((rdev->family) == CHIP_RV610) ||
	    ((rdev->family) == CHIP_RV620) ||
1707 1708
	    ((rdev->family) == CHIP_RS780) ||
	    ((rdev->family) == CHIP_RS880)) {
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750
		tmp = (CACHE_FIFO_SIZE(0xa) |
		       FETCH_FIFO_HIWATER(0xa) |
		       DONE_FIFO_HIWATER(0xe0) |
		       ALU_UPDATE_FIFO_HIWATER(0x8));
	} else if (((rdev->family) == CHIP_R600) ||
		   ((rdev->family) == CHIP_RV630)) {
		tmp &= ~DONE_FIFO_HIWATER(0xff);
		tmp |= DONE_FIFO_HIWATER(0x4);
	}
	WREG32(SQ_MS_FIFO_SIZES, tmp);

	/* SQ_CONFIG, SQ_GPR_RESOURCE_MGMT, SQ_THREAD_RESOURCE_MGMT, SQ_STACK_RESOURCE_MGMT
	 * should be adjusted as needed by the 2D/3D drivers.  This just sets default values
	 */
	sq_config = RREG32(SQ_CONFIG);
	sq_config &= ~(PS_PRIO(3) |
		       VS_PRIO(3) |
		       GS_PRIO(3) |
		       ES_PRIO(3));
	sq_config |= (DX9_CONSTS |
		      VC_ENABLE |
		      PS_PRIO(0) |
		      VS_PRIO(1) |
		      GS_PRIO(2) |
		      ES_PRIO(3));

	if ((rdev->family) == CHIP_R600) {
		sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(124) |
					  NUM_VS_GPRS(124) |
					  NUM_CLAUSE_TEMP_GPRS(4));
		sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(0) |
					  NUM_ES_GPRS(0));
		sq_thread_resource_mgmt = (NUM_PS_THREADS(136) |
					   NUM_VS_THREADS(48) |
					   NUM_GS_THREADS(4) |
					   NUM_ES_THREADS(4));
		sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(128) |
					    NUM_VS_STACK_ENTRIES(128));
		sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(0) |
					    NUM_ES_STACK_ENTRIES(0));
	} else if (((rdev->family) == CHIP_RV610) ||
		   ((rdev->family) == CHIP_RV620) ||
1751 1752
		   ((rdev->family) == CHIP_RS780) ||
		   ((rdev->family) == CHIP_RS880)) {
1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
		/* no vertex cache */
		sq_config &= ~VC_ENABLE;

		sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) |
					  NUM_VS_GPRS(44) |
					  NUM_CLAUSE_TEMP_GPRS(2));
		sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(17) |
					  NUM_ES_GPRS(17));
		sq_thread_resource_mgmt = (NUM_PS_THREADS(79) |
					   NUM_VS_THREADS(78) |
					   NUM_GS_THREADS(4) |
					   NUM_ES_THREADS(31));
		sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(40) |
					    NUM_VS_STACK_ENTRIES(40));
		sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(32) |
					    NUM_ES_STACK_ENTRIES(16));
	} else if (((rdev->family) == CHIP_RV630) ||
		   ((rdev->family) == CHIP_RV635)) {
		sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) |
					  NUM_VS_GPRS(44) |
					  NUM_CLAUSE_TEMP_GPRS(2));
		sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(18) |
					  NUM_ES_GPRS(18));
		sq_thread_resource_mgmt = (NUM_PS_THREADS(79) |
					   NUM_VS_THREADS(78) |
					   NUM_GS_THREADS(4) |
					   NUM_ES_THREADS(31));
		sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(40) |
					    NUM_VS_STACK_ENTRIES(40));
		sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(32) |
					    NUM_ES_STACK_ENTRIES(16));
	} else if ((rdev->family) == CHIP_RV670) {
		sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) |
					  NUM_VS_GPRS(44) |
					  NUM_CLAUSE_TEMP_GPRS(2));
		sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(17) |
					  NUM_ES_GPRS(17));
		sq_thread_resource_mgmt = (NUM_PS_THREADS(79) |
					   NUM_VS_THREADS(78) |
					   NUM_GS_THREADS(4) |
					   NUM_ES_THREADS(31));
		sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(64) |
					    NUM_VS_STACK_ENTRIES(64));
		sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(64) |
					    NUM_ES_STACK_ENTRIES(64));
	}

	WREG32(SQ_CONFIG, sq_config);
	WREG32(SQ_GPR_RESOURCE_MGMT_1,  sq_gpr_resource_mgmt_1);
	WREG32(SQ_GPR_RESOURCE_MGMT_2,  sq_gpr_resource_mgmt_2);
	WREG32(SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt);
	WREG32(SQ_STACK_RESOURCE_MGMT_1, sq_stack_resource_mgmt_1);
	WREG32(SQ_STACK_RESOURCE_MGMT_2, sq_stack_resource_mgmt_2);

	if (((rdev->family) == CHIP_RV610) ||
	    ((rdev->family) == CHIP_RV620) ||
1809 1810
	    ((rdev->family) == CHIP_RS780) ||
	    ((rdev->family) == CHIP_RS880)) {
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836
		WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(TC_ONLY));
	} else {
		WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(VC_AND_TC));
	}

	/* More default values. 2D/3D driver should adjust as needed */
	WREG32(PA_SC_AA_SAMPLE_LOCS_2S, (S0_X(0xc) | S0_Y(0x4) |
					 S1_X(0x4) | S1_Y(0xc)));
	WREG32(PA_SC_AA_SAMPLE_LOCS_4S, (S0_X(0xe) | S0_Y(0xe) |
					 S1_X(0x2) | S1_Y(0x2) |
					 S2_X(0xa) | S2_Y(0x6) |
					 S3_X(0x6) | S3_Y(0xa)));
	WREG32(PA_SC_AA_SAMPLE_LOCS_8S_WD0, (S0_X(0xe) | S0_Y(0xb) |
					     S1_X(0x4) | S1_Y(0xc) |
					     S2_X(0x1) | S2_Y(0x6) |
					     S3_X(0xa) | S3_Y(0xe)));
	WREG32(PA_SC_AA_SAMPLE_LOCS_8S_WD1, (S4_X(0x6) | S4_Y(0x1) |
					     S5_X(0x0) | S5_Y(0x0) |
					     S6_X(0xb) | S6_Y(0x4) |
					     S7_X(0x7) | S7_Y(0x8)));

	WREG32(VGT_STRMOUT_EN, 0);
	tmp = rdev->config.r600.max_pipes * 16;
	switch (rdev->family) {
	case CHIP_RV610:
	case CHIP_RV620:
1837 1838
	case CHIP_RS780:
	case CHIP_RS880:
1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879
		tmp += 32;
		break;
	case CHIP_RV670:
		tmp += 128;
		break;
	default:
		break;
	}
	if (tmp > 256) {
		tmp = 256;
	}
	WREG32(VGT_ES_PER_GS, 128);
	WREG32(VGT_GS_PER_ES, tmp);
	WREG32(VGT_GS_PER_VS, 2);
	WREG32(VGT_GS_VERTEX_REUSE, 16);

	/* more default values. 2D/3D driver should adjust as needed */
	WREG32(PA_SC_LINE_STIPPLE_STATE, 0);
	WREG32(VGT_STRMOUT_EN, 0);
	WREG32(SX_MISC, 0);
	WREG32(PA_SC_MODE_CNTL, 0);
	WREG32(PA_SC_AA_CONFIG, 0);
	WREG32(PA_SC_LINE_STIPPLE, 0);
	WREG32(SPI_INPUT_Z, 0);
	WREG32(SPI_PS_IN_CONTROL_0, NUM_INTERP(2));
	WREG32(CB_COLOR7_FRAG, 0);

	/* Clear render buffer base addresses */
	WREG32(CB_COLOR0_BASE, 0);
	WREG32(CB_COLOR1_BASE, 0);
	WREG32(CB_COLOR2_BASE, 0);
	WREG32(CB_COLOR3_BASE, 0);
	WREG32(CB_COLOR4_BASE, 0);
	WREG32(CB_COLOR5_BASE, 0);
	WREG32(CB_COLOR6_BASE, 0);
	WREG32(CB_COLOR7_BASE, 0);
	WREG32(CB_COLOR7_FRAG, 0);

	switch (rdev->family) {
	case CHIP_RV610:
	case CHIP_RV620:
1880 1881
	case CHIP_RS780:
	case CHIP_RS880:
1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910
		tmp = TC_L2_SIZE(8);
		break;
	case CHIP_RV630:
	case CHIP_RV635:
		tmp = TC_L2_SIZE(4);
		break;
	case CHIP_R600:
		tmp = TC_L2_SIZE(0) | L2_DISABLE_LATE_HIT;
		break;
	default:
		tmp = TC_L2_SIZE(0);
		break;
	}
	WREG32(TC_CNTL, tmp);

	tmp = RREG32(HDP_HOST_PATH_CNTL);
	WREG32(HDP_HOST_PATH_CNTL, tmp);

	tmp = RREG32(ARB_POP);
	tmp |= ENABLE_TC128;
	WREG32(ARB_POP, tmp);

	WREG32(PA_SC_MULTI_CHIP_CNTL, 0);
	WREG32(PA_CL_ENHANCE, (CLIP_VTX_REORDER_ENA |
			       NUM_CLIP_SEQ(3)));
	WREG32(PA_SC_ENHANCE, FORCE_EOV_MAX_CLK_CNT(4095));
}


1911 1912 1913
/*
 * Indirect registers accessor
 */
1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
u32 r600_pciep_rreg(struct radeon_device *rdev, u32 reg)
{
	u32 r;

	WREG32(PCIE_PORT_INDEX, ((reg) & 0xff));
	(void)RREG32(PCIE_PORT_INDEX);
	r = RREG32(PCIE_PORT_DATA);
	return r;
}

void r600_pciep_wreg(struct radeon_device *rdev, u32 reg, u32 v)
{
	WREG32(PCIE_PORT_INDEX, ((reg) & 0xff));
	(void)RREG32(PCIE_PORT_INDEX);
	WREG32(PCIE_PORT_DATA, (v));
	(void)RREG32(PCIE_PORT_DATA);
}

/*
 * CP & Ring
 */
void r600_cp_stop(struct radeon_device *rdev)
{
1937
	radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size);
1938
	WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1));
1939
	WREG32(SCRATCH_UMSK, 0);
1940 1941
}

1942
int r600_init_microcode(struct radeon_device *rdev)
1943 1944 1945
{
	struct platform_device *pdev;
	const char *chip_name;
1946 1947
	const char *rlc_chip_name;
	size_t pfp_req_size, me_req_size, rlc_req_size;
1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
	char fw_name[30];
	int err;

	DRM_DEBUG("\n");

	pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0);
	err = IS_ERR(pdev);
	if (err) {
		printk(KERN_ERR "radeon_cp: Failed to register firmware\n");
		return -EINVAL;
	}

	switch (rdev->family) {
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
	case CHIP_R600:
		chip_name = "R600";
		rlc_chip_name = "R600";
		break;
	case CHIP_RV610:
		chip_name = "RV610";
		rlc_chip_name = "R600";
		break;
	case CHIP_RV630:
		chip_name = "RV630";
		rlc_chip_name = "R600";
		break;
	case CHIP_RV620:
		chip_name = "RV620";
		rlc_chip_name = "R600";
		break;
	case CHIP_RV635:
		chip_name = "RV635";
		rlc_chip_name = "R600";
		break;
	case CHIP_RV670:
		chip_name = "RV670";
		rlc_chip_name = "R600";
		break;
1985
	case CHIP_RS780:
1986 1987 1988 1989 1990 1991 1992 1993
	case CHIP_RS880:
		chip_name = "RS780";
		rlc_chip_name = "R600";
		break;
	case CHIP_RV770:
		chip_name = "RV770";
		rlc_chip_name = "R700";
		break;
1994
	case CHIP_RV730:
1995 1996 1997 1998 1999 2000 2001 2002
	case CHIP_RV740:
		chip_name = "RV730";
		rlc_chip_name = "R700";
		break;
	case CHIP_RV710:
		chip_name = "RV710";
		rlc_chip_name = "R700";
		break;
2003 2004
	case CHIP_CEDAR:
		chip_name = "CEDAR";
2005
		rlc_chip_name = "CEDAR";
2006 2007 2008
		break;
	case CHIP_REDWOOD:
		chip_name = "REDWOOD";
2009
		rlc_chip_name = "REDWOOD";
2010 2011 2012
		break;
	case CHIP_JUNIPER:
		chip_name = "JUNIPER";
2013
		rlc_chip_name = "JUNIPER";
2014 2015 2016 2017
		break;
	case CHIP_CYPRESS:
	case CHIP_HEMLOCK:
		chip_name = "CYPRESS";
2018
		rlc_chip_name = "CYPRESS";
2019
		break;
2020 2021 2022 2023
	case CHIP_PALM:
		chip_name = "PALM";
		rlc_chip_name = "SUMO";
		break;
2024 2025 2026 2027 2028 2029 2030 2031
	case CHIP_SUMO:
		chip_name = "SUMO";
		rlc_chip_name = "SUMO";
		break;
	case CHIP_SUMO2:
		chip_name = "SUMO2";
		rlc_chip_name = "SUMO";
		break;
2032 2033 2034
	default: BUG();
	}

2035 2036 2037
	if (rdev->family >= CHIP_CEDAR) {
		pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4;
		me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4;
2038
		rlc_req_size = EVERGREEN_RLC_UCODE_SIZE * 4;
2039
	} else if (rdev->family >= CHIP_RV770) {
2040 2041
		pfp_req_size = R700_PFP_UCODE_SIZE * 4;
		me_req_size = R700_PM4_UCODE_SIZE * 4;
2042
		rlc_req_size = R700_RLC_UCODE_SIZE * 4;
2043 2044 2045
	} else {
		pfp_req_size = PFP_UCODE_SIZE * 4;
		me_req_size = PM4_UCODE_SIZE * 12;
2046
		rlc_req_size = RLC_UCODE_SIZE * 4;
2047 2048
	}

2049
	DRM_INFO("Loading %s Microcode\n", chip_name);
2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072

	snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name);
	err = request_firmware(&rdev->pfp_fw, fw_name, &pdev->dev);
	if (err)
		goto out;
	if (rdev->pfp_fw->size != pfp_req_size) {
		printk(KERN_ERR
		       "r600_cp: Bogus length %zu in firmware \"%s\"\n",
		       rdev->pfp_fw->size, fw_name);
		err = -EINVAL;
		goto out;
	}

	snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name);
	err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev);
	if (err)
		goto out;
	if (rdev->me_fw->size != me_req_size) {
		printk(KERN_ERR
		       "r600_cp: Bogus length %zu in firmware \"%s\"\n",
		       rdev->me_fw->size, fw_name);
		err = -EINVAL;
	}
2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084

	snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", rlc_chip_name);
	err = request_firmware(&rdev->rlc_fw, fw_name, &pdev->dev);
	if (err)
		goto out;
	if (rdev->rlc_fw->size != rlc_req_size) {
		printk(KERN_ERR
		       "r600_rlc: Bogus length %zu in firmware \"%s\"\n",
		       rdev->rlc_fw->size, fw_name);
		err = -EINVAL;
	}

2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096
out:
	platform_device_unregister(pdev);

	if (err) {
		if (err != -EINVAL)
			printk(KERN_ERR
			       "r600_cp: Failed to load firmware \"%s\"\n",
			       fw_name);
		release_firmware(rdev->pfp_fw);
		rdev->pfp_fw = NULL;
		release_firmware(rdev->me_fw);
		rdev->me_fw = NULL;
2097 2098
		release_firmware(rdev->rlc_fw);
		rdev->rlc_fw = NULL;
2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112
	}
	return err;
}

static int r600_cp_load_microcode(struct radeon_device *rdev)
{
	const __be32 *fw_data;
	int i;

	if (!rdev->me_fw || !rdev->pfp_fw)
		return -EINVAL;

	r600_cp_stop(rdev);

2113 2114 2115 2116 2117
	WREG32(CP_RB_CNTL,
#ifdef __BIG_ENDIAN
	       BUF_SWAP_32BIT |
#endif
	       RB_NO_UPDATE | RB_BLKSZ(15) | RB_BUFSZ(3));
2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156

	/* Reset cp */
	WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP);
	RREG32(GRBM_SOFT_RESET);
	mdelay(15);
	WREG32(GRBM_SOFT_RESET, 0);

	WREG32(CP_ME_RAM_WADDR, 0);

	fw_data = (const __be32 *)rdev->me_fw->data;
	WREG32(CP_ME_RAM_WADDR, 0);
	for (i = 0; i < PM4_UCODE_SIZE * 3; i++)
		WREG32(CP_ME_RAM_DATA,
		       be32_to_cpup(fw_data++));

	fw_data = (const __be32 *)rdev->pfp_fw->data;
	WREG32(CP_PFP_UCODE_ADDR, 0);
	for (i = 0; i < PFP_UCODE_SIZE; i++)
		WREG32(CP_PFP_UCODE_DATA,
		       be32_to_cpup(fw_data++));

	WREG32(CP_PFP_UCODE_ADDR, 0);
	WREG32(CP_ME_RAM_WADDR, 0);
	WREG32(CP_ME_RAM_RADDR, 0);
	return 0;
}

int r600_cp_start(struct radeon_device *rdev)
{
	int r;
	uint32_t cp_me;

	r = radeon_ring_lock(rdev, 7);
	if (r) {
		DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r);
		return r;
	}
	radeon_ring_write(rdev, PACKET3(PACKET3_ME_INITIALIZE, 5));
	radeon_ring_write(rdev, 0x1);
2157
	if (rdev->family >= CHIP_RV770) {
2158 2159
		radeon_ring_write(rdev, 0x0);
		radeon_ring_write(rdev, rdev->config.rv770.max_hw_contexts - 1);
2160 2161 2162
	} else {
		radeon_ring_write(rdev, 0x3);
		radeon_ring_write(rdev, rdev->config.r600.max_hw_contexts - 1);
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187
	}
	radeon_ring_write(rdev, PACKET3_ME_INITIALIZE_DEVICE_ID(1));
	radeon_ring_write(rdev, 0);
	radeon_ring_write(rdev, 0);
	radeon_ring_unlock_commit(rdev);

	cp_me = 0xff;
	WREG32(R_0086D8_CP_ME_CNTL, cp_me);
	return 0;
}

int r600_cp_resume(struct radeon_device *rdev)
{
	u32 tmp;
	u32 rb_bufsz;
	int r;

	/* Reset cp */
	WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP);
	RREG32(GRBM_SOFT_RESET);
	mdelay(15);
	WREG32(GRBM_SOFT_RESET, 0);

	/* Set ring buffer size */
	rb_bufsz = drm_order(rdev->cp.ring_size / 8);
2188
	tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
2189
#ifdef __BIG_ENDIAN
2190
	tmp |= BUF_SWAP_32BIT;
2191
#endif
2192
	WREG32(CP_RB_CNTL, tmp);
2193 2194 2195 2196 2197 2198 2199 2200
	WREG32(CP_SEM_WAIT_TIMER, 0x4);

	/* Set the write pointer delay */
	WREG32(CP_RB_WPTR_DELAY, 0);

	/* Initialize the ring buffer's read and write pointers */
	WREG32(CP_RB_CNTL, tmp | RB_RPTR_WR_ENA);
	WREG32(CP_RB_RPTR_WR, 0);
2201 2202
	rdev->cp.wptr = 0;
	WREG32(CP_RB_WPTR, rdev->cp.wptr);
2203 2204

	/* set the wb address whether it's enabled or not */
2205 2206
	WREG32(CP_RB_RPTR_ADDR,
	       ((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC));
2207 2208 2209 2210 2211 2212 2213 2214 2215 2216
	WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF);
	WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF);

	if (rdev->wb.enabled)
		WREG32(SCRATCH_UMSK, 0xff);
	else {
		tmp |= RB_NO_UPDATE;
		WREG32(SCRATCH_UMSK, 0);
	}

2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
	mdelay(1);
	WREG32(CP_RB_CNTL, tmp);

	WREG32(CP_RB_BASE, rdev->cp.gpu_addr >> 8);
	WREG32(CP_DEBUG, (1 << 27) | (1 << 28));

	rdev->cp.rptr = RREG32(CP_RB_RPTR);

	r600_cp_start(rdev);
	rdev->cp.ready = true;
	r = radeon_ring_test(rdev);
	if (r) {
		rdev->cp.ready = false;
		return r;
	}
	return 0;
}

void r600_cp_commit(struct radeon_device *rdev)
{
	WREG32(CP_RB_WPTR, rdev->cp.wptr);
	(void)RREG32(CP_RB_WPTR);
}

void r600_ring_init(struct radeon_device *rdev, unsigned ring_size)
{
	u32 rb_bufsz;

	/* Align ring size */
	rb_bufsz = drm_order(ring_size / 8);
	ring_size = (1 << (rb_bufsz + 1)) * 4;
	rdev->cp.ring_size = ring_size;
	rdev->cp.align_mask = 16 - 1;
}

2252 2253 2254 2255 2256 2257
void r600_cp_fini(struct radeon_device *rdev)
{
	r600_cp_stop(rdev);
	radeon_ring_fini(rdev);
}

2258 2259 2260 2261 2262 2263 2264 2265 2266

/*
 * GPU scratch registers helpers function.
 */
void r600_scratch_init(struct radeon_device *rdev)
{
	int i;

	rdev->scratch.num_reg = 7;
2267
	rdev->scratch.reg_base = SCRATCH_REG0;
2268 2269
	for (i = 0; i < rdev->scratch.num_reg; i++) {
		rdev->scratch.free[i] = true;
2270
		rdev->scratch.reg[i] = rdev->scratch.reg_base + (i * 4);
2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316
	}
}

int r600_ring_test(struct radeon_device *rdev)
{
	uint32_t scratch;
	uint32_t tmp = 0;
	unsigned i;
	int r;

	r = radeon_scratch_get(rdev, &scratch);
	if (r) {
		DRM_ERROR("radeon: cp failed to get scratch reg (%d).\n", r);
		return r;
	}
	WREG32(scratch, 0xCAFEDEAD);
	r = radeon_ring_lock(rdev, 3);
	if (r) {
		DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r);
		radeon_scratch_free(rdev, scratch);
		return r;
	}
	radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1));
	radeon_ring_write(rdev, ((scratch - PACKET3_SET_CONFIG_REG_OFFSET) >> 2));
	radeon_ring_write(rdev, 0xDEADBEEF);
	radeon_ring_unlock_commit(rdev);
	for (i = 0; i < rdev->usec_timeout; i++) {
		tmp = RREG32(scratch);
		if (tmp == 0xDEADBEEF)
			break;
		DRM_UDELAY(1);
	}
	if (i < rdev->usec_timeout) {
		DRM_INFO("ring test succeeded in %d usecs\n", i);
	} else {
		DRM_ERROR("radeon: ring test failed (scratch(0x%04X)=0x%08X)\n",
			  scratch, tmp);
		r = -EINVAL;
	}
	radeon_scratch_free(rdev, scratch);
	return r;
}

void r600_fence_ring_emit(struct radeon_device *rdev,
			  struct radeon_fence *fence)
{
2317 2318
	if (rdev->wb.use_event) {
		u64 addr = rdev->wb.gpu_addr + R600_WB_EVENT_OFFSET +
2319
			(u64)(rdev->fence_drv[fence->ring].scratch_reg - rdev->scratch.reg_base);
2320 2321 2322 2323 2324 2325 2326 2327
		/* flush read cache over gart */
		radeon_ring_write(rdev, PACKET3(PACKET3_SURFACE_SYNC, 3));
		radeon_ring_write(rdev, PACKET3_TC_ACTION_ENA |
					PACKET3_VC_ACTION_ENA |
					PACKET3_SH_ACTION_ENA);
		radeon_ring_write(rdev, 0xFFFFFFFF);
		radeon_ring_write(rdev, 0);
		radeon_ring_write(rdev, 10); /* poll interval */
2328 2329 2330 2331 2332 2333 2334 2335
		/* EVENT_WRITE_EOP - flush caches, send int */
		radeon_ring_write(rdev, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
		radeon_ring_write(rdev, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT_TS) | EVENT_INDEX(5));
		radeon_ring_write(rdev, addr & 0xffffffff);
		radeon_ring_write(rdev, (upper_32_bits(addr) & 0xff) | DATA_SEL(1) | INT_SEL(2));
		radeon_ring_write(rdev, fence->seq);
		radeon_ring_write(rdev, 0);
	} else {
2336 2337 2338 2339 2340 2341 2342 2343
		/* flush read cache over gart */
		radeon_ring_write(rdev, PACKET3(PACKET3_SURFACE_SYNC, 3));
		radeon_ring_write(rdev, PACKET3_TC_ACTION_ENA |
					PACKET3_VC_ACTION_ENA |
					PACKET3_SH_ACTION_ENA);
		radeon_ring_write(rdev, 0xFFFFFFFF);
		radeon_ring_write(rdev, 0);
		radeon_ring_write(rdev, 10); /* poll interval */
2344 2345 2346 2347 2348 2349 2350 2351
		radeon_ring_write(rdev, PACKET3(PACKET3_EVENT_WRITE, 0));
		radeon_ring_write(rdev, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT) | EVENT_INDEX(0));
		/* wait for 3D idle clean */
		radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1));
		radeon_ring_write(rdev, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2);
		radeon_ring_write(rdev, WAIT_3D_IDLE_bit | WAIT_3D_IDLECLEAN_bit);
		/* Emit fence sequence & fire IRQ */
		radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1));
2352
		radeon_ring_write(rdev, ((rdev->fence_drv[fence->ring].scratch_reg - PACKET3_SET_CONFIG_REG_OFFSET) >> 2));
2353 2354 2355 2356 2357
		radeon_ring_write(rdev, fence->seq);
		/* CP_INTERRUPT packet 3 no longer exists, use packet 0 */
		radeon_ring_write(rdev, PACKET0(CP_INT_STATUS, 0));
		radeon_ring_write(rdev, RB_INT_STAT);
	}
2358 2359 2360
}

int r600_copy_blit(struct radeon_device *rdev,
2361 2362 2363 2364
		   uint64_t src_offset,
		   uint64_t dst_offset,
		   unsigned num_gpu_pages,
		   struct radeon_fence *fence)
2365
{
2366 2367 2368 2369
	int r;

	mutex_lock(&rdev->r600_blit.mutex);
	rdev->r600_blit.vb_ib = NULL;
2370
	r = r600_blit_prepare_copy(rdev, num_gpu_pages);
2371 2372 2373 2374 2375 2376
	if (r) {
		if (rdev->r600_blit.vb_ib)
			radeon_ib_free(rdev, &rdev->r600_blit.vb_ib);
		mutex_unlock(&rdev->r600_blit.mutex);
		return r;
	}
2377
	r600_kms_blit_copy(rdev, src_offset, dst_offset, num_gpu_pages);
2378
	r600_blit_done_copy(rdev, fence);
2379
	mutex_unlock(&rdev->r600_blit.mutex);
2380 2381 2382
	return 0;
}

2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396
void r600_blit_suspend(struct radeon_device *rdev)
{
	int r;

	/* unpin shaders bo */
	if (rdev->r600_blit.shader_obj) {
		r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
		if (!r) {
			radeon_bo_unpin(rdev->r600_blit.shader_obj);
			radeon_bo_unreserve(rdev->r600_blit.shader_obj);
		}
	}
}

2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409
int r600_set_surface_reg(struct radeon_device *rdev, int reg,
			 uint32_t tiling_flags, uint32_t pitch,
			 uint32_t offset, uint32_t obj_size)
{
	/* FIXME: implement */
	return 0;
}

void r600_clear_surface_reg(struct radeon_device *rdev, int reg)
{
	/* FIXME: implement */
}

2410
int r600_startup(struct radeon_device *rdev)
2411 2412 2413
{
	int r;

2414 2415 2416
	/* enable pcie gen2 link */
	r600_pcie_gen2_enable(rdev);

2417 2418 2419 2420 2421 2422 2423 2424
	if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) {
		r = r600_init_microcode(rdev);
		if (r) {
			DRM_ERROR("Failed to load firmware!\n");
			return r;
		}
	}

2425 2426 2427 2428
	r = r600_vram_scratch_init(rdev);
	if (r)
		return r;

2429
	r600_mc_program(rdev);
2430 2431 2432 2433 2434 2435 2436
	if (rdev->flags & RADEON_IS_AGP) {
		r600_agp_enable(rdev);
	} else {
		r = r600_pcie_gart_enable(rdev);
		if (r)
			return r;
	}
2437
	r600_gpu_init(rdev);
2438 2439 2440 2441 2442 2443
	r = r600_blit_init(rdev);
	if (r) {
		r600_blit_fini(rdev);
		rdev->asic->copy = NULL;
		dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r);
	}
2444

2445 2446 2447 2448 2449
	/* allocate wb buffer */
	r = radeon_wb_init(rdev);
	if (r)
		return r;

2450 2451 2452 2453 2454 2455 2456 2457 2458
	/* Enable IRQ */
	r = r600_irq_init(rdev);
	if (r) {
		DRM_ERROR("radeon: IH init failed (%d).\n", r);
		radeon_irq_kms_fini(rdev);
		return r;
	}
	r600_irq_set(rdev);

2459 2460 2461 2462 2463 2464 2465 2466 2467
	r = radeon_ring_init(rdev, rdev->cp.ring_size);
	if (r)
		return r;
	r = r600_cp_load_microcode(rdev);
	if (r)
		return r;
	r = r600_cp_resume(rdev);
	if (r)
		return r;
2468

2469 2470 2471
	return 0;
}

2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485
void r600_vga_set_state(struct radeon_device *rdev, bool state)
{
	uint32_t temp;

	temp = RREG32(CONFIG_CNTL);
	if (state == false) {
		temp &= ~(1<<0);
		temp |= (1<<1);
	} else {
		temp &= ~(1<<1);
	}
	WREG32(CONFIG_CNTL, temp);
}

2486 2487 2488 2489
int r600_resume(struct radeon_device *rdev)
{
	int r;

2490 2491 2492 2493
	/* Do not reset GPU before posting, on r600 hw unlike on r500 hw,
	 * posting will perform necessary task to bring back GPU into good
	 * shape.
	 */
2494
	/* post card */
2495
	atom_asic_init(rdev->mode_info.atom_context);
2496 2497 2498 2499 2500 2501 2502

	r = r600_startup(rdev);
	if (r) {
		DRM_ERROR("r600 startup failed on resume\n");
		return r;
	}

2503
	r = r600_ib_test(rdev);
2504
	if (r) {
P
Paul Bolle 已提交
2505
		DRM_ERROR("radeon: failed testing IB (%d).\n", r);
2506 2507
		return r;
	}
2508 2509 2510 2511 2512 2513 2514

	r = r600_audio_init(rdev);
	if (r) {
		DRM_ERROR("radeon: audio resume failed\n");
		return r;
	}

2515 2516 2517
	return r;
}

2518 2519
int r600_suspend(struct radeon_device *rdev)
{
2520
	r600_audio_fini(rdev);
2521 2522
	/* FIXME: we should wait for ring to be empty */
	r600_cp_stop(rdev);
2523
	rdev->cp.ready = false;
2524
	r600_irq_suspend(rdev);
2525
	radeon_wb_disable(rdev);
2526
	r600_pcie_gart_disable(rdev);
2527 2528
	r600_blit_suspend(rdev);

2529 2530 2531 2532 2533 2534 2535 2536 2537 2538
	return 0;
}

/* Plan is to move initialization in that function and use
 * helper function so that radeon_device_init pretty much
 * do nothing more than calling asic specific function. This
 * should also allow to remove a bunch of callback function
 * like vram_info.
 */
int r600_init(struct radeon_device *rdev)
2539
{
2540
	int r;
2541

2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554
	if (r600_debugfs_mc_info_init(rdev)) {
		DRM_ERROR("Failed to register debugfs file for mc !\n");
	}
	/* This don't do much */
	r = radeon_gem_init(rdev);
	if (r)
		return r;
	/* Read BIOS */
	if (!radeon_get_bios(rdev)) {
		if (ASIC_IS_AVIVO(rdev))
			return -EINVAL;
	}
	/* Must be an ATOMBIOS */
2555 2556
	if (!rdev->is_atom_bios) {
		dev_err(rdev->dev, "Expecting atombios for R600 GPU\n");
2557
		return -EINVAL;
2558
	}
2559 2560 2561 2562
	r = radeon_atombios_init(rdev);
	if (r)
		return r;
	/* Post card if necessary */
2563
	if (!radeon_card_posted(rdev)) {
2564 2565 2566 2567
		if (!rdev->bios) {
			dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n");
			return -EINVAL;
		}
2568 2569 2570 2571 2572 2573 2574
		DRM_INFO("GPU not posted. posting now...\n");
		atom_asic_init(rdev->mode_info.atom_context);
	}
	/* Initialize scratch registers */
	r600_scratch_init(rdev);
	/* Initialize surface registers */
	radeon_surface_init(rdev);
2575
	/* Initialize clocks */
2576
	radeon_get_clock_info(rdev->ddev);
2577
	/* Fence driver */
2578
	r = radeon_fence_driver_init(rdev, 1);
2579 2580
	if (r)
		return r;
2581 2582 2583 2584 2585
	if (rdev->flags & RADEON_IS_AGP) {
		r = radeon_agp_init(rdev);
		if (r)
			radeon_agp_disable(rdev);
	}
2586
	r = r600_mc_init(rdev);
2587
	if (r)
2588 2589
		return r;
	/* Memory manager */
2590
	r = radeon_bo_init(rdev);
2591 2592
	if (r)
		return r;
2593 2594 2595 2596 2597

	r = radeon_irq_kms_init(rdev);
	if (r)
		return r;

2598 2599 2600
	rdev->cp.ring_obj = NULL;
	r600_ring_init(rdev, 1024 * 1024);

2601 2602
	rdev->ih.ring_obj = NULL;
	r600_ih_ring_init(rdev, 64 * 1024);
2603

2604 2605 2606 2607
	r = r600_pcie_gart_init(rdev);
	if (r)
		return r;

2608
	rdev->accel_working = true;
2609
	r = r600_startup(rdev);
2610
	if (r) {
2611 2612 2613
		dev_err(rdev->dev, "disabling GPU acceleration\n");
		r600_cp_fini(rdev);
		r600_irq_fini(rdev);
2614
		radeon_wb_fini(rdev);
2615
		radeon_irq_kms_fini(rdev);
2616
		r600_pcie_gart_fini(rdev);
2617
		rdev->accel_working = false;
2618
	}
2619 2620 2621
	if (rdev->accel_working) {
		r = radeon_ib_pool_init(rdev);
		if (r) {
2622
			dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
2623
			rdev->accel_working = false;
2624 2625 2626 2627 2628 2629
		} else {
			r = r600_ib_test(rdev);
			if (r) {
				dev_err(rdev->dev, "IB test failed (%d).\n", r);
				rdev->accel_working = false;
			}
2630
		}
2631
	}
2632 2633 2634 2635

	r = r600_audio_init(rdev);
	if (r)
		return r; /* TODO error handling */
2636 2637 2638 2639 2640
	return 0;
}

void r600_fini(struct radeon_device *rdev)
{
2641
	r600_audio_fini(rdev);
2642
	r600_blit_fini(rdev);
2643
	r600_cp_fini(rdev);
2644
	r600_irq_fini(rdev);
2645
	radeon_wb_fini(rdev);
2646
	radeon_ib_pool_fini(rdev);
2647
	radeon_irq_kms_fini(rdev);
2648
	r600_pcie_gart_fini(rdev);
2649
	r600_vram_scratch_fini(rdev);
2650
	radeon_agp_fini(rdev);
2651 2652
	radeon_gem_fini(rdev);
	radeon_fence_driver_fini(rdev);
2653
	radeon_bo_fini(rdev);
2654
	radeon_atombios_fini(rdev);
2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666
	kfree(rdev->bios);
	rdev->bios = NULL;
}


/*
 * CS stuff
 */
void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
{
	/* FIXME: implement */
	radeon_ring_write(rdev, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
2667 2668 2669 2670 2671
	radeon_ring_write(rdev,
#ifdef __BIG_ENDIAN
			  (2 << 0) |
#endif
			  (ib->gpu_addr & 0xFFFFFFFC));
2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732
	radeon_ring_write(rdev, upper_32_bits(ib->gpu_addr) & 0xFF);
	radeon_ring_write(rdev, ib->length_dw);
}

int r600_ib_test(struct radeon_device *rdev)
{
	struct radeon_ib *ib;
	uint32_t scratch;
	uint32_t tmp = 0;
	unsigned i;
	int r;

	r = radeon_scratch_get(rdev, &scratch);
	if (r) {
		DRM_ERROR("radeon: failed to get scratch reg (%d).\n", r);
		return r;
	}
	WREG32(scratch, 0xCAFEDEAD);
	r = radeon_ib_get(rdev, &ib);
	if (r) {
		DRM_ERROR("radeon: failed to get ib (%d).\n", r);
		return r;
	}
	ib->ptr[0] = PACKET3(PACKET3_SET_CONFIG_REG, 1);
	ib->ptr[1] = ((scratch - PACKET3_SET_CONFIG_REG_OFFSET) >> 2);
	ib->ptr[2] = 0xDEADBEEF;
	ib->ptr[3] = PACKET2(0);
	ib->ptr[4] = PACKET2(0);
	ib->ptr[5] = PACKET2(0);
	ib->ptr[6] = PACKET2(0);
	ib->ptr[7] = PACKET2(0);
	ib->ptr[8] = PACKET2(0);
	ib->ptr[9] = PACKET2(0);
	ib->ptr[10] = PACKET2(0);
	ib->ptr[11] = PACKET2(0);
	ib->ptr[12] = PACKET2(0);
	ib->ptr[13] = PACKET2(0);
	ib->ptr[14] = PACKET2(0);
	ib->ptr[15] = PACKET2(0);
	ib->length_dw = 16;
	r = radeon_ib_schedule(rdev, ib);
	if (r) {
		radeon_scratch_free(rdev, scratch);
		radeon_ib_free(rdev, &ib);
		DRM_ERROR("radeon: failed to schedule ib (%d).\n", r);
		return r;
	}
	r = radeon_fence_wait(ib->fence, false);
	if (r) {
		DRM_ERROR("radeon: fence wait failed (%d).\n", r);
		return r;
	}
	for (i = 0; i < rdev->usec_timeout; i++) {
		tmp = RREG32(scratch);
		if (tmp == 0xDEADBEEF)
			break;
		DRM_UDELAY(1);
	}
	if (i < rdev->usec_timeout) {
		DRM_INFO("ib test succeeded in %u usecs\n", i);
	} else {
D
Daniel J Blueman 已提交
2733
		DRM_ERROR("radeon: ib test failed (scratch(0x%04X)=0x%08X)\n",
2734 2735 2736 2737 2738
			  scratch, tmp);
		r = -EINVAL;
	}
	radeon_scratch_free(rdev, scratch);
	radeon_ib_free(rdev, &ib);
2739 2740 2741
	return r;
}

2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760
/*
 * Interrupts
 *
 * Interrupts use a ring buffer on r6xx/r7xx hardware.  It works pretty
 * the same as the CP ring buffer, but in reverse.  Rather than the CPU
 * writing to the ring and the GPU consuming, the GPU writes to the ring
 * and host consumes.  As the host irq handler processes interrupts, it
 * increments the rptr.  When the rptr catches up with the wptr, all the
 * current interrupts have been processed.
 */

void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size)
{
	u32 rb_bufsz;

	/* Align ring size */
	rb_bufsz = drm_order(ring_size / 4);
	ring_size = (1 << rb_bufsz) * 4;
	rdev->ih.ring_size = ring_size;
2761 2762
	rdev->ih.ptr_mask = rdev->ih.ring_size - 1;
	rdev->ih.rptr = 0;
2763 2764
}

2765
static int r600_ih_ring_alloc(struct radeon_device *rdev)
2766 2767 2768 2769 2770
{
	int r;

	/* Allocate ring buffer */
	if (rdev->ih.ring_obj == NULL) {
2771
		r = radeon_bo_create(rdev, rdev->ih.ring_size,
2772
				     PAGE_SIZE, true,
2773 2774
				     RADEON_GEM_DOMAIN_GTT,
				     &rdev->ih.ring_obj);
2775 2776 2777 2778
		if (r) {
			DRM_ERROR("radeon: failed to create ih ring buffer (%d).\n", r);
			return r;
		}
2779 2780 2781 2782 2783 2784
		r = radeon_bo_reserve(rdev->ih.ring_obj, false);
		if (unlikely(r != 0))
			return r;
		r = radeon_bo_pin(rdev->ih.ring_obj,
				  RADEON_GEM_DOMAIN_GTT,
				  &rdev->ih.gpu_addr);
2785
		if (r) {
2786
			radeon_bo_unreserve(rdev->ih.ring_obj);
2787 2788 2789
			DRM_ERROR("radeon: failed to pin ih ring buffer (%d).\n", r);
			return r;
		}
2790 2791 2792
		r = radeon_bo_kmap(rdev->ih.ring_obj,
				   (void **)&rdev->ih.ring);
		radeon_bo_unreserve(rdev->ih.ring_obj);
2793 2794 2795 2796 2797 2798 2799 2800 2801 2802
		if (r) {
			DRM_ERROR("radeon: failed to map ih ring buffer (%d).\n", r);
			return r;
		}
	}
	return 0;
}

static void r600_ih_ring_fini(struct radeon_device *rdev)
{
2803
	int r;
2804
	if (rdev->ih.ring_obj) {
2805 2806 2807 2808 2809 2810 2811
		r = radeon_bo_reserve(rdev->ih.ring_obj, false);
		if (likely(r == 0)) {
			radeon_bo_kunmap(rdev->ih.ring_obj);
			radeon_bo_unpin(rdev->ih.ring_obj);
			radeon_bo_unreserve(rdev->ih.ring_obj);
		}
		radeon_bo_unref(&rdev->ih.ring_obj);
2812 2813 2814 2815 2816
		rdev->ih.ring = NULL;
		rdev->ih.ring_obj = NULL;
	}
}

2817
void r600_rlc_stop(struct radeon_device *rdev)
2818 2819
{

2820 2821
	if ((rdev->family >= CHIP_RV770) &&
	    (rdev->family <= CHIP_RV740)) {
2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
		/* r7xx asics need to soft reset RLC before halting */
		WREG32(SRBM_SOFT_RESET, SOFT_RESET_RLC);
		RREG32(SRBM_SOFT_RESET);
		udelay(15000);
		WREG32(SRBM_SOFT_RESET, 0);
		RREG32(SRBM_SOFT_RESET);
	}

	WREG32(RLC_CNTL, 0);
}

static void r600_rlc_start(struct radeon_device *rdev)
{
	WREG32(RLC_CNTL, RLC_ENABLE);
}

static int r600_rlc_init(struct radeon_device *rdev)
{
	u32 i;
	const __be32 *fw_data;

	if (!rdev->rlc_fw)
		return -EINVAL;

	r600_rlc_stop(rdev);

	WREG32(RLC_HB_BASE, 0);
	WREG32(RLC_HB_CNTL, 0);
	WREG32(RLC_HB_RPTR, 0);
	WREG32(RLC_HB_WPTR, 0);
2852 2853 2854 2855
	if (rdev->family <= CHIP_CAICOS) {
		WREG32(RLC_HB_WPTR_LSB_ADDR, 0);
		WREG32(RLC_HB_WPTR_MSB_ADDR, 0);
	}
2856 2857 2858 2859
	WREG32(RLC_MC_CNTL, 0);
	WREG32(RLC_UCODE_CNTL, 0);

	fw_data = (const __be32 *)rdev->rlc_fw->data;
2860 2861 2862 2863 2864 2865
	if (rdev->family >= CHIP_CAYMAN) {
		for (i = 0; i < CAYMAN_RLC_UCODE_SIZE; i++) {
			WREG32(RLC_UCODE_ADDR, i);
			WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
		}
	} else if (rdev->family >= CHIP_CEDAR) {
2866 2867 2868 2869 2870
		for (i = 0; i < EVERGREEN_RLC_UCODE_SIZE; i++) {
			WREG32(RLC_UCODE_ADDR, i);
			WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
		}
	} else if (rdev->family >= CHIP_RV770) {
2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899
		for (i = 0; i < R700_RLC_UCODE_SIZE; i++) {
			WREG32(RLC_UCODE_ADDR, i);
			WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
		}
	} else {
		for (i = 0; i < RLC_UCODE_SIZE; i++) {
			WREG32(RLC_UCODE_ADDR, i);
			WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++));
		}
	}
	WREG32(RLC_UCODE_ADDR, 0);

	r600_rlc_start(rdev);

	return 0;
}

static void r600_enable_interrupts(struct radeon_device *rdev)
{
	u32 ih_cntl = RREG32(IH_CNTL);
	u32 ih_rb_cntl = RREG32(IH_RB_CNTL);

	ih_cntl |= ENABLE_INTR;
	ih_rb_cntl |= IH_RB_ENABLE;
	WREG32(IH_CNTL, ih_cntl);
	WREG32(IH_RB_CNTL, ih_rb_cntl);
	rdev->ih.enabled = true;
}

2900
void r600_disable_interrupts(struct radeon_device *rdev)
2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916
{
	u32 ih_rb_cntl = RREG32(IH_RB_CNTL);
	u32 ih_cntl = RREG32(IH_CNTL);

	ih_rb_cntl &= ~IH_RB_ENABLE;
	ih_cntl &= ~ENABLE_INTR;
	WREG32(IH_RB_CNTL, ih_rb_cntl);
	WREG32(IH_CNTL, ih_cntl);
	/* set rptr, wptr to 0 */
	WREG32(IH_RB_RPTR, 0);
	WREG32(IH_RB_WPTR, 0);
	rdev->ih.enabled = false;
	rdev->ih.wptr = 0;
	rdev->ih.rptr = 0;
}

2917 2918 2919 2920
static void r600_disable_interrupt_state(struct radeon_device *rdev)
{
	u32 tmp;

2921
	WREG32(CP_INT_CNTL, CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE);
2922 2923
	WREG32(GRBM_INT_CNTL, 0);
	WREG32(DxMODE_INT_MASK, 0);
2924 2925
	WREG32(D1GRPH_INTERRUPT_CONTROL, 0);
	WREG32(D2GRPH_INTERRUPT_CONTROL, 0);
2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938
	if (ASIC_IS_DCE3(rdev)) {
		WREG32(DCE3_DACA_AUTODETECT_INT_CONTROL, 0);
		WREG32(DCE3_DACB_AUTODETECT_INT_CONTROL, 0);
		tmp = RREG32(DC_HPD1_INT_CONTROL) & DC_HPDx_INT_POLARITY;
		WREG32(DC_HPD1_INT_CONTROL, tmp);
		tmp = RREG32(DC_HPD2_INT_CONTROL) & DC_HPDx_INT_POLARITY;
		WREG32(DC_HPD2_INT_CONTROL, tmp);
		tmp = RREG32(DC_HPD3_INT_CONTROL) & DC_HPDx_INT_POLARITY;
		WREG32(DC_HPD3_INT_CONTROL, tmp);
		tmp = RREG32(DC_HPD4_INT_CONTROL) & DC_HPDx_INT_POLARITY;
		WREG32(DC_HPD4_INT_CONTROL, tmp);
		if (ASIC_IS_DCE32(rdev)) {
			tmp = RREG32(DC_HPD5_INT_CONTROL) & DC_HPDx_INT_POLARITY;
2939
			WREG32(DC_HPD5_INT_CONTROL, tmp);
2940
			tmp = RREG32(DC_HPD6_INT_CONTROL) & DC_HPDx_INT_POLARITY;
2941
			WREG32(DC_HPD6_INT_CONTROL, tmp);
2942 2943 2944 2945 2946
		}
	} else {
		WREG32(DACA_AUTODETECT_INT_CONTROL, 0);
		WREG32(DACB_AUTODETECT_INT_CONTROL, 0);
		tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY;
2947
		WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp);
2948
		tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY;
2949
		WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp);
2950
		tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY;
2951
		WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp);
2952 2953 2954
	}
}

2955 2956 2957 2958 2959 2960 2961
int r600_irq_init(struct radeon_device *rdev)
{
	int ret = 0;
	int rb_bufsz;
	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;

	/* allocate ring */
2962
	ret = r600_ih_ring_alloc(rdev);
2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993
	if (ret)
		return ret;

	/* disable irqs */
	r600_disable_interrupts(rdev);

	/* init rlc */
	ret = r600_rlc_init(rdev);
	if (ret) {
		r600_ih_ring_fini(rdev);
		return ret;
	}

	/* setup interrupt control */
	/* set dummy read address to ring address */
	WREG32(INTERRUPT_CNTL2, rdev->ih.gpu_addr >> 8);
	interrupt_cntl = RREG32(INTERRUPT_CNTL);
	/* IH_DUMMY_RD_OVERRIDE=0 - dummy read disabled with msi, enabled without msi
	 * IH_DUMMY_RD_OVERRIDE=1 - dummy read controlled by IH_DUMMY_RD_EN
	 */
	interrupt_cntl &= ~IH_DUMMY_RD_OVERRIDE;
	/* IH_REQ_NONSNOOP_EN=1 if ring is in non-cacheable memory, e.g., vram */
	interrupt_cntl &= ~IH_REQ_NONSNOOP_EN;
	WREG32(INTERRUPT_CNTL, interrupt_cntl);

	WREG32(IH_RB_BASE, rdev->ih.gpu_addr >> 8);
	rb_bufsz = drm_order(rdev->ih.ring_size / 4);

	ih_rb_cntl = (IH_WPTR_OVERFLOW_ENABLE |
		      IH_WPTR_OVERFLOW_CLEAR |
		      (rb_bufsz << 1));
2994 2995 2996 2997 2998 2999 3000

	if (rdev->wb.enabled)
		ih_rb_cntl |= IH_WPTR_WRITEBACK_ENABLE;

	/* set the writeback address whether it's enabled or not */
	WREG32(IH_RB_WPTR_ADDR_LO, (rdev->wb.gpu_addr + R600_WB_IH_WPTR_OFFSET) & 0xFFFFFFFC);
	WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + R600_WB_IH_WPTR_OFFSET) & 0xFF);
3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015

	WREG32(IH_RB_CNTL, ih_rb_cntl);

	/* set rptr, wptr to 0 */
	WREG32(IH_RB_RPTR, 0);
	WREG32(IH_RB_WPTR, 0);

	/* Default settings for IH_CNTL (disabled at first) */
	ih_cntl = MC_WRREQ_CREDIT(0x10) | MC_WR_CLEAN_CNT(0x10);
	/* RPTR_REARM only works if msi's are enabled */
	if (rdev->msi_enabled)
		ih_cntl |= RPTR_REARM;
	WREG32(IH_CNTL, ih_cntl);

	/* force the active interrupt state to all disabled */
3016 3017 3018 3019
	if (rdev->family >= CHIP_CEDAR)
		evergreen_disable_interrupt_state(rdev);
	else
		r600_disable_interrupt_state(rdev);
3020 3021 3022 3023 3024 3025 3026

	/* enable irqs */
	r600_enable_interrupts(rdev);

	return ret;
}

3027
void r600_irq_suspend(struct radeon_device *rdev)
3028
{
3029
	r600_irq_disable(rdev);
3030
	r600_rlc_stop(rdev);
3031 3032 3033 3034 3035
}

void r600_irq_fini(struct radeon_device *rdev)
{
	r600_irq_suspend(rdev);
3036 3037 3038 3039 3040
	r600_ih_ring_fini(rdev);
}

int r600_irq_set(struct radeon_device *rdev)
{
3041 3042 3043
	u32 cp_int_cntl = CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE;
	u32 mode_int = 0;
	u32 hpd1, hpd2, hpd3, hpd4 = 0, hpd5 = 0, hpd6 = 0;
3044
	u32 grbm_int_cntl = 0;
3045
	u32 hdmi1, hdmi2;
3046
	u32 d1grph = 0, d2grph = 0;
3047

3048
	if (!rdev->irq.installed) {
J
Joe Perches 已提交
3049
		WARN(1, "Can't enable IRQ/MSI because no handler is installed\n");
3050 3051
		return -EINVAL;
	}
3052
	/* don't enable anything if the ih is disabled */
3053 3054 3055 3056
	if (!rdev->ih.enabled) {
		r600_disable_interrupts(rdev);
		/* force the active interrupt state to all disabled */
		r600_disable_interrupt_state(rdev);
3057
		return 0;
3058
	}
3059

3060
	hdmi1 = RREG32(R600_HDMI_BLOCK1 + R600_HDMI_CNTL) & ~R600_HDMI_INT_EN;
3061
	if (ASIC_IS_DCE3(rdev)) {
3062
		hdmi2 = RREG32(R600_HDMI_BLOCK3 + R600_HDMI_CNTL) & ~R600_HDMI_INT_EN;
3063 3064 3065 3066 3067 3068 3069 3070 3071
		hpd1 = RREG32(DC_HPD1_INT_CONTROL) & ~DC_HPDx_INT_EN;
		hpd2 = RREG32(DC_HPD2_INT_CONTROL) & ~DC_HPDx_INT_EN;
		hpd3 = RREG32(DC_HPD3_INT_CONTROL) & ~DC_HPDx_INT_EN;
		hpd4 = RREG32(DC_HPD4_INT_CONTROL) & ~DC_HPDx_INT_EN;
		if (ASIC_IS_DCE32(rdev)) {
			hpd5 = RREG32(DC_HPD5_INT_CONTROL) & ~DC_HPDx_INT_EN;
			hpd6 = RREG32(DC_HPD6_INT_CONTROL) & ~DC_HPDx_INT_EN;
		}
	} else {
3072
		hdmi2 = RREG32(R600_HDMI_BLOCK2 + R600_HDMI_CNTL) & ~R600_HDMI_INT_EN;
3073 3074 3075 3076 3077
		hpd1 = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL) & ~DC_HPDx_INT_EN;
		hpd2 = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL) & ~DC_HPDx_INT_EN;
		hpd3 = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL) & ~DC_HPDx_INT_EN;
	}

3078 3079 3080
	if (rdev->irq.sw_int) {
		DRM_DEBUG("r600_irq_set: sw int\n");
		cp_int_cntl |= RB_INT_ENABLE;
3081
		cp_int_cntl |= TIME_STAMP_INT_ENABLE;
3082
	}
3083 3084
	if (rdev->irq.crtc_vblank_int[0] ||
	    rdev->irq.pflip[0]) {
3085 3086 3087
		DRM_DEBUG("r600_irq_set: vblank 0\n");
		mode_int |= D1MODE_VBLANK_INT_MASK;
	}
3088 3089
	if (rdev->irq.crtc_vblank_int[1] ||
	    rdev->irq.pflip[1]) {
3090 3091 3092
		DRM_DEBUG("r600_irq_set: vblank 1\n");
		mode_int |= D2MODE_VBLANK_INT_MASK;
	}
3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116
	if (rdev->irq.hpd[0]) {
		DRM_DEBUG("r600_irq_set: hpd 1\n");
		hpd1 |= DC_HPDx_INT_EN;
	}
	if (rdev->irq.hpd[1]) {
		DRM_DEBUG("r600_irq_set: hpd 2\n");
		hpd2 |= DC_HPDx_INT_EN;
	}
	if (rdev->irq.hpd[2]) {
		DRM_DEBUG("r600_irq_set: hpd 3\n");
		hpd3 |= DC_HPDx_INT_EN;
	}
	if (rdev->irq.hpd[3]) {
		DRM_DEBUG("r600_irq_set: hpd 4\n");
		hpd4 |= DC_HPDx_INT_EN;
	}
	if (rdev->irq.hpd[4]) {
		DRM_DEBUG("r600_irq_set: hpd 5\n");
		hpd5 |= DC_HPDx_INT_EN;
	}
	if (rdev->irq.hpd[5]) {
		DRM_DEBUG("r600_irq_set: hpd 6\n");
		hpd6 |= DC_HPDx_INT_EN;
	}
3117 3118 3119 3120 3121 3122 3123 3124
	if (rdev->irq.hdmi[0]) {
		DRM_DEBUG("r600_irq_set: hdmi 1\n");
		hdmi1 |= R600_HDMI_INT_EN;
	}
	if (rdev->irq.hdmi[1]) {
		DRM_DEBUG("r600_irq_set: hdmi 2\n");
		hdmi2 |= R600_HDMI_INT_EN;
	}
3125 3126 3127 3128
	if (rdev->irq.gui_idle) {
		DRM_DEBUG("gui idle\n");
		grbm_int_cntl |= GUI_IDLE_INT_ENABLE;
	}
3129 3130 3131

	WREG32(CP_INT_CNTL, cp_int_cntl);
	WREG32(DxMODE_INT_MASK, mode_int);
3132 3133
	WREG32(D1GRPH_INTERRUPT_CONTROL, d1grph);
	WREG32(D2GRPH_INTERRUPT_CONTROL, d2grph);
3134
	WREG32(GRBM_INT_CNTL, grbm_int_cntl);
3135
	WREG32(R600_HDMI_BLOCK1 + R600_HDMI_CNTL, hdmi1);
3136
	if (ASIC_IS_DCE3(rdev)) {
3137
		WREG32(R600_HDMI_BLOCK3 + R600_HDMI_CNTL, hdmi2);
3138 3139 3140 3141 3142 3143 3144 3145 3146
		WREG32(DC_HPD1_INT_CONTROL, hpd1);
		WREG32(DC_HPD2_INT_CONTROL, hpd2);
		WREG32(DC_HPD3_INT_CONTROL, hpd3);
		WREG32(DC_HPD4_INT_CONTROL, hpd4);
		if (ASIC_IS_DCE32(rdev)) {
			WREG32(DC_HPD5_INT_CONTROL, hpd5);
			WREG32(DC_HPD6_INT_CONTROL, hpd6);
		}
	} else {
3147
		WREG32(R600_HDMI_BLOCK2 + R600_HDMI_CNTL, hdmi2);
3148 3149 3150 3151
		WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, hpd1);
		WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, hpd2);
		WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, hpd3);
	}
3152 3153 3154 3155

	return 0;
}

3156
static void r600_irq_ack(struct radeon_device *rdev)
3157
{
3158 3159 3160
	u32 tmp;

	if (ASIC_IS_DCE3(rdev)) {
3161 3162 3163
		rdev->irq.stat_regs.r600.disp_int = RREG32(DCE3_DISP_INTERRUPT_STATUS);
		rdev->irq.stat_regs.r600.disp_int_cont = RREG32(DCE3_DISP_INTERRUPT_STATUS_CONTINUE);
		rdev->irq.stat_regs.r600.disp_int_cont2 = RREG32(DCE3_DISP_INTERRUPT_STATUS_CONTINUE2);
3164
	} else {
3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176
		rdev->irq.stat_regs.r600.disp_int = RREG32(DISP_INTERRUPT_STATUS);
		rdev->irq.stat_regs.r600.disp_int_cont = RREG32(DISP_INTERRUPT_STATUS_CONTINUE);
		rdev->irq.stat_regs.r600.disp_int_cont2 = 0;
	}
	rdev->irq.stat_regs.r600.d1grph_int = RREG32(D1GRPH_INTERRUPT_STATUS);
	rdev->irq.stat_regs.r600.d2grph_int = RREG32(D2GRPH_INTERRUPT_STATUS);

	if (rdev->irq.stat_regs.r600.d1grph_int & DxGRPH_PFLIP_INT_OCCURRED)
		WREG32(D1GRPH_INTERRUPT_STATUS, DxGRPH_PFLIP_INT_CLEAR);
	if (rdev->irq.stat_regs.r600.d2grph_int & DxGRPH_PFLIP_INT_OCCURRED)
		WREG32(D2GRPH_INTERRUPT_STATUS, DxGRPH_PFLIP_INT_CLEAR);
	if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VBLANK_INTERRUPT)
3177
		WREG32(D1MODE_VBLANK_STATUS, DxMODE_VBLANK_ACK);
3178
	if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VLINE_INTERRUPT)
3179
		WREG32(D1MODE_VLINE_STATUS, DxMODE_VLINE_ACK);
3180
	if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VBLANK_INTERRUPT)
3181
		WREG32(D2MODE_VBLANK_STATUS, DxMODE_VBLANK_ACK);
3182
	if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VLINE_INTERRUPT)
3183
		WREG32(D2MODE_VLINE_STATUS, DxMODE_VLINE_ACK);
3184
	if (rdev->irq.stat_regs.r600.disp_int & DC_HPD1_INTERRUPT) {
3185 3186 3187 3188 3189 3190 3191 3192 3193 3194
		if (ASIC_IS_DCE3(rdev)) {
			tmp = RREG32(DC_HPD1_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HPD1_INT_CONTROL, tmp);
		} else {
			tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp);
		}
	}
3195
	if (rdev->irq.stat_regs.r600.disp_int & DC_HPD2_INTERRUPT) {
3196 3197 3198 3199 3200 3201 3202 3203 3204 3205
		if (ASIC_IS_DCE3(rdev)) {
			tmp = RREG32(DC_HPD2_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HPD2_INT_CONTROL, tmp);
		} else {
			tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp);
		}
	}
3206
	if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD3_INTERRUPT) {
3207 3208 3209 3210 3211 3212 3213 3214 3215 3216
		if (ASIC_IS_DCE3(rdev)) {
			tmp = RREG32(DC_HPD3_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HPD3_INT_CONTROL, tmp);
		} else {
			tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp);
		}
	}
3217
	if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD4_INTERRUPT) {
3218 3219 3220 3221 3222
		tmp = RREG32(DC_HPD4_INT_CONTROL);
		tmp |= DC_HPDx_INT_ACK;
		WREG32(DC_HPD4_INT_CONTROL, tmp);
	}
	if (ASIC_IS_DCE32(rdev)) {
3223
		if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD5_INTERRUPT) {
3224 3225 3226 3227
			tmp = RREG32(DC_HPD5_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HPD5_INT_CONTROL, tmp);
		}
3228
		if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD6_INTERRUPT) {
3229 3230 3231 3232 3233
			tmp = RREG32(DC_HPD5_INT_CONTROL);
			tmp |= DC_HPDx_INT_ACK;
			WREG32(DC_HPD6_INT_CONTROL, tmp);
		}
	}
3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245
	if (RREG32(R600_HDMI_BLOCK1 + R600_HDMI_STATUS) & R600_HDMI_INT_PENDING) {
		WREG32_P(R600_HDMI_BLOCK1 + R600_HDMI_CNTL, R600_HDMI_INT_ACK, ~R600_HDMI_INT_ACK);
	}
	if (ASIC_IS_DCE3(rdev)) {
		if (RREG32(R600_HDMI_BLOCK3 + R600_HDMI_STATUS) & R600_HDMI_INT_PENDING) {
			WREG32_P(R600_HDMI_BLOCK3 + R600_HDMI_CNTL, R600_HDMI_INT_ACK, ~R600_HDMI_INT_ACK);
		}
	} else {
		if (RREG32(R600_HDMI_BLOCK2 + R600_HDMI_STATUS) & R600_HDMI_INT_PENDING) {
			WREG32_P(R600_HDMI_BLOCK2 + R600_HDMI_CNTL, R600_HDMI_INT_ACK, ~R600_HDMI_INT_ACK);
		}
	}
3246 3247 3248 3249 3250 3251 3252
}

void r600_irq_disable(struct radeon_device *rdev)
{
	r600_disable_interrupts(rdev);
	/* Wait and acknowledge irq */
	mdelay(1);
3253
	r600_irq_ack(rdev);
3254
	r600_disable_interrupt_state(rdev);
3255 3256
}

3257
static u32 r600_get_ih_wptr(struct radeon_device *rdev)
3258 3259
{
	u32 wptr, tmp;
3260

3261
	if (rdev->wb.enabled)
3262
		wptr = le32_to_cpu(rdev->wb.wb[R600_WB_IH_WPTR_OFFSET/4]);
3263 3264
	else
		wptr = RREG32(IH_RB_WPTR);
3265

3266
	if (wptr & RB_OVERFLOW) {
3267 3268 3269 3270 3271 3272 3273
		/* When a ring buffer overflow happen start parsing interrupt
		 * from the last not overwritten vector (wptr + 16). Hopefully
		 * this should allow us to catchup.
		 */
		dev_warn(rdev->dev, "IH ring buffer overflow (0x%08X, %d, %d)\n",
			wptr, rdev->ih.rptr, (wptr + 16) + rdev->ih.ptr_mask);
		rdev->ih.rptr = (wptr + 16) & rdev->ih.ptr_mask;
3274 3275 3276 3277
		tmp = RREG32(IH_RB_CNTL);
		tmp |= IH_WPTR_OVERFLOW_CLEAR;
		WREG32(IH_RB_CNTL, tmp);
	}
3278
	return (wptr & rdev->ih.ptr_mask);
3279
}
3280

3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298
/*        r600 IV Ring
 * Each IV ring entry is 128 bits:
 * [7:0]    - interrupt source id
 * [31:8]   - reserved
 * [59:32]  - interrupt source data
 * [127:60]  - reserved
 *
 * The basic interrupt vector entries
 * are decoded as follows:
 * src_id  src_data  description
 *      1         0  D1 Vblank
 *      1         1  D1 Vline
 *      5         0  D2 Vblank
 *      5         1  D2 Vline
 *     19         0  FP Hot plug detection A
 *     19         1  FP Hot plug detection B
 *     19         2  DAC A auto-detection
 *     19         3  DAC B auto-detection
3299 3300
 *     21         4  HDMI block A
 *     21         5  HDMI block B
3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312
 *    176         -  CP_INT RB
 *    177         -  CP_INT IB1
 *    178         -  CP_INT IB2
 *    181         -  EOP Interrupt
 *    233         -  GUI Idle
 *
 * Note, these are based on r600 and may need to be
 * adjusted or added to on newer asics
 */

int r600_irq_process(struct radeon_device *rdev)
{
3313 3314
	u32 wptr;
	u32 rptr;
3315
	u32 src_id, src_data;
3316
	u32 ring_index;
3317
	unsigned long flags;
A
Alex Deucher 已提交
3318
	bool queue_hotplug = false;
3319

3320
	if (!rdev->ih.enabled || rdev->shutdown)
3321
		return IRQ_NONE;
3322

3323 3324 3325 3326
	/* No MSIs, need a dummy read to flush PCI DMAs */
	if (!rdev->msi_enabled)
		RREG32(IH_RB_WPTR);

3327 3328 3329 3330
	wptr = r600_get_ih_wptr(rdev);
	rptr = rdev->ih.rptr;
	DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr);

3331 3332 3333 3334 3335 3336 3337 3338
	spin_lock_irqsave(&rdev->ih.lock, flags);

	if (rptr == wptr) {
		spin_unlock_irqrestore(&rdev->ih.lock, flags);
		return IRQ_NONE;
	}

restart_ih:
3339 3340 3341
	/* Order reading of wptr vs. reading of IH ring data */
	rmb();

3342
	/* display interrupts */
3343
	r600_irq_ack(rdev);
3344 3345 3346 3347 3348

	rdev->ih.wptr = wptr;
	while (rptr != wptr) {
		/* wptr/rptr are in bytes! */
		ring_index = rptr / 4;
3349 3350
		src_id = le32_to_cpu(rdev->ih.ring[ring_index]) & 0xff;
		src_data = le32_to_cpu(rdev->ih.ring[ring_index + 1]) & 0xfffffff;
3351 3352 3353 3354 3355

		switch (src_id) {
		case 1: /* D1 vblank/vline */
			switch (src_data) {
			case 0: /* D1 vblank */
3356 3357 3358 3359 3360 3361
				if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VBLANK_INTERRUPT) {
					if (rdev->irq.crtc_vblank_int[0]) {
						drm_handle_vblank(rdev->ddev, 0);
						rdev->pm.vblank_sync = true;
						wake_up(&rdev->irq.vblank_queue);
					}
3362 3363
					if (rdev->irq.pflip[0])
						radeon_crtc_handle_flip(rdev, 0);
3364
					rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VBLANK_INTERRUPT;
3365 3366 3367 3368
					DRM_DEBUG("IH: D1 vblank\n");
				}
				break;
			case 1: /* D1 vline */
3369 3370
				if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VLINE_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VLINE_INTERRUPT;
3371 3372 3373 3374
					DRM_DEBUG("IH: D1 vline\n");
				}
				break;
			default:
3375
				DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
3376 3377 3378 3379 3380 3381
				break;
			}
			break;
		case 5: /* D2 vblank/vline */
			switch (src_data) {
			case 0: /* D2 vblank */
3382 3383 3384 3385 3386 3387
				if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VBLANK_INTERRUPT) {
					if (rdev->irq.crtc_vblank_int[1]) {
						drm_handle_vblank(rdev->ddev, 1);
						rdev->pm.vblank_sync = true;
						wake_up(&rdev->irq.vblank_queue);
					}
3388 3389
					if (rdev->irq.pflip[1])
						radeon_crtc_handle_flip(rdev, 1);
3390
					rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VBLANK_INTERRUPT;
3391 3392 3393 3394
					DRM_DEBUG("IH: D2 vblank\n");
				}
				break;
			case 1: /* D1 vline */
3395 3396
				if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VLINE_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VLINE_INTERRUPT;
3397 3398 3399 3400
					DRM_DEBUG("IH: D2 vline\n");
				}
				break;
			default:
3401
				DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
3402 3403 3404
				break;
			}
			break;
3405 3406 3407
		case 19: /* HPD/DAC hotplug */
			switch (src_data) {
			case 0:
3408 3409
				if (rdev->irq.stat_regs.r600.disp_int & DC_HPD1_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int &= ~DC_HPD1_INTERRUPT;
A
Alex Deucher 已提交
3410 3411
					queue_hotplug = true;
					DRM_DEBUG("IH: HPD1\n");
3412 3413 3414
				}
				break;
			case 1:
3415 3416
				if (rdev->irq.stat_regs.r600.disp_int & DC_HPD2_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int &= ~DC_HPD2_INTERRUPT;
A
Alex Deucher 已提交
3417 3418
					queue_hotplug = true;
					DRM_DEBUG("IH: HPD2\n");
3419 3420 3421
				}
				break;
			case 4:
3422 3423
				if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD3_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int_cont &= ~DC_HPD3_INTERRUPT;
A
Alex Deucher 已提交
3424 3425
					queue_hotplug = true;
					DRM_DEBUG("IH: HPD3\n");
3426 3427 3428
				}
				break;
			case 5:
3429 3430
				if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD4_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int_cont &= ~DC_HPD4_INTERRUPT;
A
Alex Deucher 已提交
3431 3432
					queue_hotplug = true;
					DRM_DEBUG("IH: HPD4\n");
3433 3434 3435
				}
				break;
			case 10:
3436 3437
				if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD5_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int_cont2 &= ~DC_HPD5_INTERRUPT;
A
Alex Deucher 已提交
3438 3439
					queue_hotplug = true;
					DRM_DEBUG("IH: HPD5\n");
3440 3441 3442
				}
				break;
			case 12:
3443 3444
				if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD6_INTERRUPT) {
					rdev->irq.stat_regs.r600.disp_int_cont2 &= ~DC_HPD6_INTERRUPT;
A
Alex Deucher 已提交
3445 3446
					queue_hotplug = true;
					DRM_DEBUG("IH: HPD6\n");
3447 3448 3449
				}
				break;
			default:
3450
				DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
3451 3452 3453
				break;
			}
			break;
3454 3455 3456 3457
		case 21: /* HDMI */
			DRM_DEBUG("IH: HDMI: 0x%x\n", src_data);
			r600_audio_schedule_polling(rdev);
			break;
3458 3459 3460 3461
		case 176: /* CP_INT in ring buffer */
		case 177: /* CP_INT in IB1 */
		case 178: /* CP_INT in IB2 */
			DRM_DEBUG("IH: CP int: 0x%08x\n", src_data);
3462
			radeon_fence_process(rdev, RADEON_RING_TYPE_GFX_INDEX);
3463 3464 3465
			break;
		case 181: /* CP EOP event */
			DRM_DEBUG("IH: CP EOP\n");
3466
			radeon_fence_process(rdev, RADEON_RING_TYPE_GFX_INDEX);
3467
			break;
3468
		case 233: /* GUI IDLE */
3469
			DRM_DEBUG("IH: GUI idle\n");
3470 3471 3472
			rdev->pm.gui_idle = true;
			wake_up(&rdev->irq.idle_queue);
			break;
3473
		default:
3474
			DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data);
3475 3476 3477 3478
			break;
		}

		/* wptr/rptr are in bytes! */
3479 3480
		rptr += 16;
		rptr &= rdev->ih.ptr_mask;
3481 3482 3483 3484 3485
	}
	/* make sure wptr hasn't changed while processing */
	wptr = r600_get_ih_wptr(rdev);
	if (wptr != rdev->ih.wptr)
		goto restart_ih;
A
Alex Deucher 已提交
3486
	if (queue_hotplug)
3487
		schedule_work(&rdev->hotplug_work);
3488 3489 3490 3491 3492
	rdev->ih.rptr = rptr;
	WREG32(IH_RB_RPTR, rdev->ih.rptr);
	spin_unlock_irqrestore(&rdev->ih.lock, flags);
	return IRQ_HANDLED;
}
3493 3494 3495 3496 3497 3498 3499

/*
 * Debugfs info
 */
#if defined(CONFIG_DEBUG_FS)

static int r600_debugfs_cp_ring_info(struct seq_file *m, void *data)
3500
{
3501 3502 3503 3504 3505 3506
	struct drm_info_node *node = (struct drm_info_node *) m->private;
	struct drm_device *dev = node->minor->dev;
	struct radeon_device *rdev = dev->dev_private;
	unsigned count, i, j;

	radeon_ring_free_size(rdev);
3507
	count = (rdev->cp.ring_size / 4) - rdev->cp.ring_free_dw;
3508
	seq_printf(m, "CP_STAT 0x%08x\n", RREG32(CP_STAT));
3509 3510 3511 3512
	seq_printf(m, "CP_RB_WPTR 0x%08x\n", RREG32(CP_RB_WPTR));
	seq_printf(m, "CP_RB_RPTR 0x%08x\n", RREG32(CP_RB_RPTR));
	seq_printf(m, "driver's copy of the CP_RB_WPTR 0x%08x\n", rdev->cp.wptr);
	seq_printf(m, "driver's copy of the CP_RB_RPTR 0x%08x\n", rdev->cp.rptr);
3513 3514
	seq_printf(m, "%u free dwords in ring\n", rdev->cp.ring_free_dw);
	seq_printf(m, "%u dwords in ring\n", count);
3515
	i = rdev->cp.rptr;
3516 3517
	for (j = 0; j <= count; j++) {
		seq_printf(m, "r[%04d]=0x%08x\n", i, rdev->cp.ring[i]);
3518
		i = (i + 1) & rdev->cp.ptr_mask;
3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546
	}
	return 0;
}

static int r600_debugfs_mc_info(struct seq_file *m, void *data)
{
	struct drm_info_node *node = (struct drm_info_node *) m->private;
	struct drm_device *dev = node->minor->dev;
	struct radeon_device *rdev = dev->dev_private;

	DREG32_SYS(m, rdev, R_000E50_SRBM_STATUS);
	DREG32_SYS(m, rdev, VM_L2_STATUS);
	return 0;
}

static struct drm_info_list r600_mc_info_list[] = {
	{"r600_mc_info", r600_debugfs_mc_info, 0, NULL},
	{"r600_ring_info", r600_debugfs_cp_ring_info, 0, NULL},
};
#endif

int r600_debugfs_mc_info_init(struct radeon_device *rdev)
{
#if defined(CONFIG_DEBUG_FS)
	return radeon_debugfs_add_files(rdev, r600_mc_info_list, ARRAY_SIZE(r600_mc_info_list));
#else
	return 0;
#endif
3547
}
3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560

/**
 * r600_ioctl_wait_idle - flush host path cache on wait idle ioctl
 * rdev: radeon device structure
 * bo: buffer object struct which userspace is waiting for idle
 *
 * Some R6XX/R7XX doesn't seems to take into account HDP flush performed
 * through ring buffer, this leads to corruption in rendering, see
 * http://bugzilla.kernel.org/show_bug.cgi?id=15186 to avoid this we
 * directly perform HDP flush by writing register through MMIO.
 */
void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo)
{
3561
	/* r7xx hw bug.  write to HDP_DEBUG1 followed by fb read
3562 3563 3564
	 * rather than write to HDP_REG_COHERENCY_FLUSH_CNTL.
	 * This seems to cause problems on some AGP cards. Just use the old
	 * method for them.
3565
	 */
3566
	if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) &&
3567
	    rdev->vram_scratch.ptr && !(rdev->flags & RADEON_IS_AGP)) {
3568
		void __iomem *ptr = (void *)rdev->vram_scratch.ptr;
3569 3570 3571 3572 3573 3574
		u32 tmp;

		WREG32(HDP_DEBUG1, 0);
		tmp = readl((void __iomem *)ptr);
	} else
		WREG32(R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1);
3575
}
3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693

void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes)
{
	u32 link_width_cntl, mask, target_reg;

	if (rdev->flags & RADEON_IS_IGP)
		return;

	if (!(rdev->flags & RADEON_IS_PCIE))
		return;

	/* x2 cards have a special sequence */
	if (ASIC_IS_X2(rdev))
		return;

	/* FIXME wait for idle */

	switch (lanes) {
	case 0:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X0;
		break;
	case 1:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X1;
		break;
	case 2:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X2;
		break;
	case 4:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X4;
		break;
	case 8:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X8;
		break;
	case 12:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X12;
		break;
	case 16:
	default:
		mask = RADEON_PCIE_LC_LINK_WIDTH_X16;
		break;
	}

	link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL);

	if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) ==
	    (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT))
		return;

	if (link_width_cntl & R600_PCIE_LC_UPCONFIGURE_DIS)
		return;

	link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK |
			     RADEON_PCIE_LC_RECONFIG_NOW |
			     R600_PCIE_LC_RENEGOTIATE_EN |
			     R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE);
	link_width_cntl |= mask;

	WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);

        /* some northbridges can renegotiate the link rather than requiring                                  
         * a complete re-config.                                                                             
         * e.g., AMD 780/790 northbridges (pci ids: 0x5956, 0x5957, 0x5958, etc.)                            
         */
        if (link_width_cntl & R600_PCIE_LC_RENEGOTIATION_SUPPORT)
		link_width_cntl |= R600_PCIE_LC_RENEGOTIATE_EN | R600_PCIE_LC_UPCONFIGURE_SUPPORT;
        else
		link_width_cntl |= R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE;

	WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl |
						       RADEON_PCIE_LC_RECONFIG_NOW));

        if (rdev->family >= CHIP_RV770)
		target_reg = R700_TARGET_AND_CURRENT_PROFILE_INDEX;
        else
		target_reg = R600_TARGET_AND_CURRENT_PROFILE_INDEX;

        /* wait for lane set to complete */
        link_width_cntl = RREG32(target_reg);
        while (link_width_cntl == 0xffffffff)
		link_width_cntl = RREG32(target_reg);

}

int r600_get_pcie_lanes(struct radeon_device *rdev)
{
	u32 link_width_cntl;

	if (rdev->flags & RADEON_IS_IGP)
		return 0;

	if (!(rdev->flags & RADEON_IS_PCIE))
		return 0;

	/* x2 cards have a special sequence */
	if (ASIC_IS_X2(rdev))
		return 0;

	/* FIXME wait for idle */

	link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL);

	switch ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) >> RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT) {
	case RADEON_PCIE_LC_LINK_WIDTH_X0:
		return 0;
	case RADEON_PCIE_LC_LINK_WIDTH_X1:
		return 1;
	case RADEON_PCIE_LC_LINK_WIDTH_X2:
		return 2;
	case RADEON_PCIE_LC_LINK_WIDTH_X4:
		return 4;
	case RADEON_PCIE_LC_LINK_WIDTH_X8:
		return 8;
	case RADEON_PCIE_LC_LINK_WIDTH_X16:
	default:
		return 16;
	}
}

3694 3695 3696 3697 3698
static void r600_pcie_gen2_enable(struct radeon_device *rdev)
{
	u32 link_width_cntl, lanes, speed_cntl, training_cntl, tmp;
	u16 link_cntl2;

3699 3700 3701
	if (radeon_pcie_gen2 == 0)
		return;

3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794
	if (rdev->flags & RADEON_IS_IGP)
		return;

	if (!(rdev->flags & RADEON_IS_PCIE))
		return;

	/* x2 cards have a special sequence */
	if (ASIC_IS_X2(rdev))
		return;

	/* only RV6xx+ chips are supported */
	if (rdev->family <= CHIP_R600)
		return;

	/* 55 nm r6xx asics */
	if ((rdev->family == CHIP_RV670) ||
	    (rdev->family == CHIP_RV620) ||
	    (rdev->family == CHIP_RV635)) {
		/* advertise upconfig capability */
		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
		link_width_cntl &= ~LC_UPCONFIGURE_DIS;
		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
		if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) {
			lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT;
			link_width_cntl &= ~(LC_LINK_WIDTH_MASK |
					     LC_RECONFIG_ARC_MISSING_ESCAPE);
			link_width_cntl |= lanes | LC_RECONFIG_NOW | LC_RENEGOTIATE_EN;
			WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
		} else {
			link_width_cntl |= LC_UPCONFIGURE_DIS;
			WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
		}
	}

	speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
	if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) &&
	    (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) {

		/* 55 nm r6xx asics */
		if ((rdev->family == CHIP_RV670) ||
		    (rdev->family == CHIP_RV620) ||
		    (rdev->family == CHIP_RV635)) {
			WREG32(MM_CFGREGS_CNTL, 0x8);
			link_cntl2 = RREG32(0x4088);
			WREG32(MM_CFGREGS_CNTL, 0);
			/* not supported yet */
			if (link_cntl2 & SELECTABLE_DEEMPHASIS)
				return;
		}

		speed_cntl &= ~LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK;
		speed_cntl |= (0x3 << LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT);
		speed_cntl &= ~LC_VOLTAGE_TIMER_SEL_MASK;
		speed_cntl &= ~LC_FORCE_DIS_HW_SPEED_CHANGE;
		speed_cntl |= LC_FORCE_EN_HW_SPEED_CHANGE;
		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);

		tmp = RREG32(0x541c);
		WREG32(0x541c, tmp | 0x8);
		WREG32(MM_CFGREGS_CNTL, MM_WR_TO_CFG_EN);
		link_cntl2 = RREG16(0x4088);
		link_cntl2 &= ~TARGET_LINK_SPEED_MASK;
		link_cntl2 |= 0x2;
		WREG16(0x4088, link_cntl2);
		WREG32(MM_CFGREGS_CNTL, 0);

		if ((rdev->family == CHIP_RV670) ||
		    (rdev->family == CHIP_RV620) ||
		    (rdev->family == CHIP_RV635)) {
			training_cntl = RREG32_PCIE_P(PCIE_LC_TRAINING_CNTL);
			training_cntl &= ~LC_POINT_7_PLUS_EN;
			WREG32_PCIE_P(PCIE_LC_TRAINING_CNTL, training_cntl);
		} else {
			speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
			speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN;
			WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
		}

		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
		speed_cntl |= LC_GEN2_EN_STRAP;
		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);

	} else {
		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
		/* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */
		if (1)
			link_width_cntl |= LC_UPCONFIGURE_DIS;
		else
			link_width_cntl &= ~LC_UPCONFIGURE_DIS;
		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
	}
}