1/*
2 * Copyright © 2012 Intel Corporation
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21 * IN THE SOFTWARE.
22 *
23 * Authors:
24 *    Eugeni Dodonov <eugeni.dodonov@intel.com>
25 *
26 */
27
28#include <linux/module.h>
29#include <linux/pm_runtime.h>
30
31#include <drm/drm_atomic_helper.h>
32#include <drm/drm_fourcc.h>
33#include <drm/drm_plane_helper.h>
34
35#include "display/intel_atomic.h"
36#include "display/intel_bw.h"
37#include "display/intel_display_types.h"
38#include "display/intel_fbc.h"
39#include "display/intel_sprite.h"
40
41#include "gt/intel_llc.h"
42
43#include "i915_drv.h"
44#include "i915_fixed.h"
45#include "i915_irq.h"
46#include "i915_trace.h"
47#include "intel_pm.h"
48#include "intel_sideband.h"
49#include "../../../platform/x86/intel_ips.h"
50
51/* Stores plane specific WM parameters */
52struct skl_wm_params {
53	bool x_tiled, y_tiled;
54	bool rc_surface;
55	bool is_planar;
56	u32 width;
57	u8 cpp;
58	u32 plane_pixel_rate;
59	u32 y_min_scanlines;
60	u32 plane_bytes_per_line;
61	uint_fixed_16_16_t plane_blocks_per_line;
62	uint_fixed_16_16_t y_tile_minimum;
63	u32 linetime_us;
64	u32 dbuf_block_size;
65};
66
67/* used in computing the new watermarks state */
68struct intel_wm_config {
69	unsigned int num_pipes_active;
70	bool sprites_enabled;
71	bool sprites_scaled;
72};
73
74static void gen9_init_clock_gating(struct drm_i915_private *dev_priv)
75{
76	if (HAS_LLC(dev_priv)) {
77		/*
78		 * WaCompressedResourceDisplayNewHashMode:skl,kbl
79		 * Display WA #0390: skl,kbl
80		 *
81		 * Must match Sampler, Pixel Back End, and Media. See
82		 * WaCompressedResourceSamplerPbeMediaNewHashMode.
83		 */
84		I915_WRITE(CHICKEN_PAR1_1,
85			   I915_READ(CHICKEN_PAR1_1) |
86			   SKL_DE_COMPRESSED_HASH_MODE);
87	}
88
89	/* See Bspec note for PSR2_CTL bit 31, Wa#828:skl,bxt,kbl,cfl */
90	I915_WRITE(CHICKEN_PAR1_1,
91		   I915_READ(CHICKEN_PAR1_1) | SKL_EDP_PSR_FIX_RDWRAP);
92
93	/* WaEnableChickenDCPR:skl,bxt,kbl,glk,cfl */
94	I915_WRITE(GEN8_CHICKEN_DCPR_1,
95		   I915_READ(GEN8_CHICKEN_DCPR_1) | MASK_WAKEMEM);
96
97	/*
98	 * WaFbcWakeMemOn:skl,bxt,kbl,glk,cfl
99	 * Display WA #0859: skl,bxt,kbl,glk,cfl
100	 */
101	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
102		   DISP_FBC_MEMORY_WAKE);
103}
104
105static void bxt_init_clock_gating(struct drm_i915_private *dev_priv)
106{
107	gen9_init_clock_gating(dev_priv);
108
109	/* WaDisableSDEUnitClockGating:bxt */
110	I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
111		   GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
112
113	/*
114	 * FIXME:
115	 * GEN8_HDCUNIT_CLOCK_GATE_DISABLE_HDCREQ applies on 3x6 GT SKUs only.
116	 */
117	I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
118		   GEN8_HDCUNIT_CLOCK_GATE_DISABLE_HDCREQ);
119
120	/*
121	 * Wa: Backlight PWM may stop in the asserted state, causing backlight
122	 * to stay fully on.
123	 */
124	I915_WRITE(GEN9_CLKGATE_DIS_0, I915_READ(GEN9_CLKGATE_DIS_0) |
125		   PWM1_GATING_DIS | PWM2_GATING_DIS);
126
127	/*
128	 * Lower the display internal timeout.
129	 * This is needed to avoid any hard hangs when DSI port PLL
130	 * is off and a MMIO access is attempted by any privilege
131	 * application, using batch buffers or any other means.
132	 */
133	I915_WRITE(RM_TIMEOUT, MMIO_TIMEOUT_US(950));
134
135	/*
136	 * WaFbcTurnOffFbcWatermark:bxt
137	 * Display WA #0562: bxt
138	 */
139	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
140		   DISP_FBC_WM_DIS);
141
142	/*
143	 * WaFbcHighMemBwCorruptionAvoidance:bxt
144	 * Display WA #0883: bxt
145	 */
146	I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
147		   ILK_DPFC_DISABLE_DUMMY0);
148}
149
150static void glk_init_clock_gating(struct drm_i915_private *dev_priv)
151{
152	gen9_init_clock_gating(dev_priv);
153
154	/*
155	 * WaDisablePWMClockGating:glk
156	 * Backlight PWM may stop in the asserted state, causing backlight
157	 * to stay fully on.
158	 */
159	I915_WRITE(GEN9_CLKGATE_DIS_0, I915_READ(GEN9_CLKGATE_DIS_0) |
160		   PWM1_GATING_DIS | PWM2_GATING_DIS);
161}
162
163static void pnv_get_mem_freq(struct drm_i915_private *dev_priv)
164{
165	u32 tmp;
166
167	tmp = I915_READ(CLKCFG);
168
169	switch (tmp & CLKCFG_FSB_MASK) {
170	case CLKCFG_FSB_533:
171		dev_priv->fsb_freq = 533; /* 133*4 */
172		break;
173	case CLKCFG_FSB_800:
174		dev_priv->fsb_freq = 800; /* 200*4 */
175		break;
176	case CLKCFG_FSB_667:
177		dev_priv->fsb_freq =  667; /* 167*4 */
178		break;
179	case CLKCFG_FSB_400:
180		dev_priv->fsb_freq = 400; /* 100*4 */
181		break;
182	}
183
184	switch (tmp & CLKCFG_MEM_MASK) {
185	case CLKCFG_MEM_533:
186		dev_priv->mem_freq = 533;
187		break;
188	case CLKCFG_MEM_667:
189		dev_priv->mem_freq = 667;
190		break;
191	case CLKCFG_MEM_800:
192		dev_priv->mem_freq = 800;
193		break;
194	}
195
196	/* detect pineview DDR3 setting */
197	tmp = I915_READ(CSHRDDR3CTL);
198	dev_priv->is_ddr3 = (tmp & CSHRDDR3CTL_DDR3) ? 1 : 0;
199}
200
201static void ilk_get_mem_freq(struct drm_i915_private *dev_priv)
202{
203	u16 ddrpll, csipll;
204
205	ddrpll = intel_uncore_read16(&dev_priv->uncore, DDRMPLL1);
206	csipll = intel_uncore_read16(&dev_priv->uncore, CSIPLL0);
207
208	switch (ddrpll & 0xff) {
209	case 0xc:
210		dev_priv->mem_freq = 800;
211		break;
212	case 0x10:
213		dev_priv->mem_freq = 1066;
214		break;
215	case 0x14:
216		dev_priv->mem_freq = 1333;
217		break;
218	case 0x18:
219		dev_priv->mem_freq = 1600;
220		break;
221	default:
222		drm_dbg(&dev_priv->drm, "unknown memory frequency 0x%02x\n",
223			ddrpll & 0xff);
224		dev_priv->mem_freq = 0;
225		break;
226	}
227
228	switch (csipll & 0x3ff) {
229	case 0x00c:
230		dev_priv->fsb_freq = 3200;
231		break;
232	case 0x00e:
233		dev_priv->fsb_freq = 3733;
234		break;
235	case 0x010:
236		dev_priv->fsb_freq = 4266;
237		break;
238	case 0x012:
239		dev_priv->fsb_freq = 4800;
240		break;
241	case 0x014:
242		dev_priv->fsb_freq = 5333;
243		break;
244	case 0x016:
245		dev_priv->fsb_freq = 5866;
246		break;
247	case 0x018:
248		dev_priv->fsb_freq = 6400;
249		break;
250	default:
251		drm_dbg(&dev_priv->drm, "unknown fsb frequency 0x%04x\n",
252			csipll & 0x3ff);
253		dev_priv->fsb_freq = 0;
254		break;
255	}
256}
257
258static const struct cxsr_latency cxsr_latency_table[] = {
259	{1, 0, 800, 400, 3382, 33382, 3983, 33983},    /* DDR2-400 SC */
260	{1, 0, 800, 667, 3354, 33354, 3807, 33807},    /* DDR2-667 SC */
261	{1, 0, 800, 800, 3347, 33347, 3763, 33763},    /* DDR2-800 SC */
262	{1, 1, 800, 667, 6420, 36420, 6873, 36873},    /* DDR3-667 SC */
263	{1, 1, 800, 800, 5902, 35902, 6318, 36318},    /* DDR3-800 SC */
264
265	{1, 0, 667, 400, 3400, 33400, 4021, 34021},    /* DDR2-400 SC */
266	{1, 0, 667, 667, 3372, 33372, 3845, 33845},    /* DDR2-667 SC */
267	{1, 0, 667, 800, 3386, 33386, 3822, 33822},    /* DDR2-800 SC */
268	{1, 1, 667, 667, 6438, 36438, 6911, 36911},    /* DDR3-667 SC */
269	{1, 1, 667, 800, 5941, 35941, 6377, 36377},    /* DDR3-800 SC */
270
271	{1, 0, 400, 400, 3472, 33472, 4173, 34173},    /* DDR2-400 SC */
272	{1, 0, 400, 667, 3443, 33443, 3996, 33996},    /* DDR2-667 SC */
273	{1, 0, 400, 800, 3430, 33430, 3946, 33946},    /* DDR2-800 SC */
274	{1, 1, 400, 667, 6509, 36509, 7062, 37062},    /* DDR3-667 SC */
275	{1, 1, 400, 800, 5985, 35985, 6501, 36501},    /* DDR3-800 SC */
276
277	{0, 0, 800, 400, 3438, 33438, 4065, 34065},    /* DDR2-400 SC */
278	{0, 0, 800, 667, 3410, 33410, 3889, 33889},    /* DDR2-667 SC */
279	{0, 0, 800, 800, 3403, 33403, 3845, 33845},    /* DDR2-800 SC */
280	{0, 1, 800, 667, 6476, 36476, 6955, 36955},    /* DDR3-667 SC */
281	{0, 1, 800, 800, 5958, 35958, 6400, 36400},    /* DDR3-800 SC */
282
283	{0, 0, 667, 400, 3456, 33456, 4103, 34106},    /* DDR2-400 SC */
284	{0, 0, 667, 667, 3428, 33428, 3927, 33927},    /* DDR2-667 SC */
285	{0, 0, 667, 800, 3443, 33443, 3905, 33905},    /* DDR2-800 SC */
286	{0, 1, 667, 667, 6494, 36494, 6993, 36993},    /* DDR3-667 SC */
287	{0, 1, 667, 800, 5998, 35998, 6460, 36460},    /* DDR3-800 SC */
288
289	{0, 0, 400, 400, 3528, 33528, 4255, 34255},    /* DDR2-400 SC */
290	{0, 0, 400, 667, 3500, 33500, 4079, 34079},    /* DDR2-667 SC */
291	{0, 0, 400, 800, 3487, 33487, 4029, 34029},    /* DDR2-800 SC */
292	{0, 1, 400, 667, 6566, 36566, 7145, 37145},    /* DDR3-667 SC */
293	{0, 1, 400, 800, 6042, 36042, 6584, 36584},    /* DDR3-800 SC */
294};
295
296static const struct cxsr_latency *intel_get_cxsr_latency(bool is_desktop,
297							 bool is_ddr3,
298							 int fsb,
299							 int mem)
300{
301	const struct cxsr_latency *latency;
302	int i;
303
304	if (fsb == 0 || mem == 0)
305		return NULL;
306
307	for (i = 0; i < ARRAY_SIZE(cxsr_latency_table); i++) {
308		latency = &cxsr_latency_table[i];
309		if (is_desktop == latency->is_desktop &&
310		    is_ddr3 == latency->is_ddr3 &&
311		    fsb == latency->fsb_freq && mem == latency->mem_freq)
312			return latency;
313	}
314
315	DRM_DEBUG_KMS("Unknown FSB/MEM found, disable CxSR\n");
316
317	return NULL;
318}
319
320static void chv_set_memory_dvfs(struct drm_i915_private *dev_priv, bool enable)
321{
322	u32 val;
323
324	vlv_punit_get(dev_priv);
325
326	val = vlv_punit_read(dev_priv, PUNIT_REG_DDR_SETUP2);
327	if (enable)
328		val &= ~FORCE_DDR_HIGH_FREQ;
329	else
330		val |= FORCE_DDR_HIGH_FREQ;
331	val &= ~FORCE_DDR_LOW_FREQ;
332	val |= FORCE_DDR_FREQ_REQ_ACK;
333	vlv_punit_write(dev_priv, PUNIT_REG_DDR_SETUP2, val);
334
335	if (wait_for((vlv_punit_read(dev_priv, PUNIT_REG_DDR_SETUP2) &
336		      FORCE_DDR_FREQ_REQ_ACK) == 0, 3))
337		drm_err(&dev_priv->drm,
338			"timed out waiting for Punit DDR DVFS request\n");
339
340	vlv_punit_put(dev_priv);
341}
342
343static void chv_set_memory_pm5(struct drm_i915_private *dev_priv, bool enable)
344{
345	u32 val;
346
347	vlv_punit_get(dev_priv);
348
349	val = vlv_punit_read(dev_priv, PUNIT_REG_DSPSSPM);
350	if (enable)
351		val |= DSP_MAXFIFO_PM5_ENABLE;
352	else
353		val &= ~DSP_MAXFIFO_PM5_ENABLE;
354	vlv_punit_write(dev_priv, PUNIT_REG_DSPSSPM, val);
355
356	vlv_punit_put(dev_priv);
357}
358
359#define FW_WM(value, plane) \
360	(((value) << DSPFW_ ## plane ## _SHIFT) & DSPFW_ ## plane ## _MASK)
361
362static bool _intel_set_memory_cxsr(struct drm_i915_private *dev_priv, bool enable)
363{
364	bool was_enabled;
365	u32 val;
366
367	if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) {
368		was_enabled = I915_READ(FW_BLC_SELF_VLV) & FW_CSPWRDWNEN;
369		I915_WRITE(FW_BLC_SELF_VLV, enable ? FW_CSPWRDWNEN : 0);
370		POSTING_READ(FW_BLC_SELF_VLV);
371	} else if (IS_G4X(dev_priv) || IS_I965GM(dev_priv)) {
372		was_enabled = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN;
373		I915_WRITE(FW_BLC_SELF, enable ? FW_BLC_SELF_EN : 0);
374		POSTING_READ(FW_BLC_SELF);
375	} else if (IS_PINEVIEW(dev_priv)) {
376		val = I915_READ(DSPFW3);
377		was_enabled = val & PINEVIEW_SELF_REFRESH_EN;
378		if (enable)
379			val |= PINEVIEW_SELF_REFRESH_EN;
380		else
381			val &= ~PINEVIEW_SELF_REFRESH_EN;
382		I915_WRITE(DSPFW3, val);
383		POSTING_READ(DSPFW3);
384	} else if (IS_I945G(dev_priv) || IS_I945GM(dev_priv)) {
385		was_enabled = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN;
386		val = enable ? _MASKED_BIT_ENABLE(FW_BLC_SELF_EN) :
387			       _MASKED_BIT_DISABLE(FW_BLC_SELF_EN);
388		I915_WRITE(FW_BLC_SELF, val);
389		POSTING_READ(FW_BLC_SELF);
390	} else if (IS_I915GM(dev_priv)) {
391		/*
392		 * FIXME can't find a bit like this for 915G, and
393		 * and yet it does have the related watermark in
394		 * FW_BLC_SELF. What's going on?
395		 */
396		was_enabled = I915_READ(INSTPM) & INSTPM_SELF_EN;
397		val = enable ? _MASKED_BIT_ENABLE(INSTPM_SELF_EN) :
398			       _MASKED_BIT_DISABLE(INSTPM_SELF_EN);
399		I915_WRITE(INSTPM, val);
400		POSTING_READ(INSTPM);
401	} else {
402		return false;
403	}
404
405	trace_intel_memory_cxsr(dev_priv, was_enabled, enable);
406
407	drm_dbg_kms(&dev_priv->drm, "memory self-refresh is %s (was %s)\n",
408		    enableddisabled(enable),
409		    enableddisabled(was_enabled));
410
411	return was_enabled;
412}
413
414/**
415 * intel_set_memory_cxsr - Configure CxSR state
416 * @dev_priv: i915 device
417 * @enable: Allow vs. disallow CxSR
418 *
419 * Allow or disallow the system to enter a special CxSR
420 * (C-state self refresh) state. What typically happens in CxSR mode
421 * is that several display FIFOs may get combined into a single larger
422 * FIFO for a particular plane (so called max FIFO mode) to allow the
423 * system to defer memory fetches longer, and the memory will enter
424 * self refresh.
425 *
426 * Note that enabling CxSR does not guarantee that the system enter
427 * this special mode, nor does it guarantee that the system stays
428 * in that mode once entered. So this just allows/disallows the system
429 * to autonomously utilize the CxSR mode. Other factors such as core
430 * C-states will affect when/if the system actually enters/exits the
431 * CxSR mode.
432 *
433 * Note that on VLV/CHV this actually only controls the max FIFO mode,
434 * and the system is free to enter/exit memory self refresh at any time
435 * even when the use of CxSR has been disallowed.
436 *
437 * While the system is actually in the CxSR/max FIFO mode, some plane
438 * control registers will not get latched on vblank. Thus in order to
439 * guarantee the system will respond to changes in the plane registers
440 * we must always disallow CxSR prior to making changes to those registers.
441 * Unfortunately the system will re-evaluate the CxSR conditions at
442 * frame start which happens after vblank start (which is when the plane
443 * registers would get latched), so we can't proceed with the plane update
444 * during the same frame where we disallowed CxSR.
445 *
446 * Certain platforms also have a deeper HPLL SR mode. Fortunately the
447 * HPLL SR mode depends on CxSR itself, so we don't have to hand hold
448 * the hardware w.r.t. HPLL SR when writing to plane registers.
449 * Disallowing just CxSR is sufficient.
450 */
451bool intel_set_memory_cxsr(struct drm_i915_private *dev_priv, bool enable)
452{
453	bool ret;
454
455	mutex_lock(&dev_priv->wm.wm_mutex);
456	ret = _intel_set_memory_cxsr(dev_priv, enable);
457	if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
458		dev_priv->wm.vlv.cxsr = enable;
459	else if (IS_G4X(dev_priv))
460		dev_priv->wm.g4x.cxsr = enable;
461	mutex_unlock(&dev_priv->wm.wm_mutex);
462
463	return ret;
464}
465
466/*
467 * Latency for FIFO fetches is dependent on several factors:
468 *   - memory configuration (speed, channels)
469 *   - chipset
470 *   - current MCH state
471 * It can be fairly high in some situations, so here we assume a fairly
472 * pessimal value.  It's a tradeoff between extra memory fetches (if we
473 * set this value too high, the FIFO will fetch frequently to stay full)
474 * and power consumption (set it too low to save power and we might see
475 * FIFO underruns and display "flicker").
476 *
477 * A value of 5us seems to be a good balance; safe for very low end
478 * platforms but not overly aggressive on lower latency configs.
479 */
480static const int pessimal_latency_ns = 5000;
481
482#define VLV_FIFO_START(dsparb, dsparb2, lo_shift, hi_shift) \
483	((((dsparb) >> (lo_shift)) & 0xff) | ((((dsparb2) >> (hi_shift)) & 0x1) << 8))
484
485static void vlv_get_fifo_size(struct intel_crtc_state *crtc_state)
486{
487	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
488	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
489	struct vlv_fifo_state *fifo_state = &crtc_state->wm.vlv.fifo_state;
490	enum pipe pipe = crtc->pipe;
491	int sprite0_start, sprite1_start;
492	u32 dsparb, dsparb2, dsparb3;
493
494	switch (pipe) {
495	case PIPE_A:
496		dsparb = I915_READ(DSPARB);
497		dsparb2 = I915_READ(DSPARB2);
498		sprite0_start = VLV_FIFO_START(dsparb, dsparb2, 0, 0);
499		sprite1_start = VLV_FIFO_START(dsparb, dsparb2, 8, 4);
500		break;
501	case PIPE_B:
502		dsparb = I915_READ(DSPARB);
503		dsparb2 = I915_READ(DSPARB2);
504		sprite0_start = VLV_FIFO_START(dsparb, dsparb2, 16, 8);
505		sprite1_start = VLV_FIFO_START(dsparb, dsparb2, 24, 12);
506		break;
507	case PIPE_C:
508		dsparb2 = I915_READ(DSPARB2);
509		dsparb3 = I915_READ(DSPARB3);
510		sprite0_start = VLV_FIFO_START(dsparb3, dsparb2, 0, 16);
511		sprite1_start = VLV_FIFO_START(dsparb3, dsparb2, 8, 20);
512		break;
513	default:
514		MISSING_CASE(pipe);
515		return;
516	}
517
518	fifo_state->plane[PLANE_PRIMARY] = sprite0_start;
519	fifo_state->plane[PLANE_SPRITE0] = sprite1_start - sprite0_start;
520	fifo_state->plane[PLANE_SPRITE1] = 511 - sprite1_start;
521	fifo_state->plane[PLANE_CURSOR] = 63;
522}
523
524static int i9xx_get_fifo_size(struct drm_i915_private *dev_priv,
525			      enum i9xx_plane_id i9xx_plane)
526{
527	u32 dsparb = I915_READ(DSPARB);
528	int size;
529
530	size = dsparb & 0x7f;
531	if (i9xx_plane == PLANE_B)
532		size = ((dsparb >> DSPARB_CSTART_SHIFT) & 0x7f) - size;
533
534	drm_dbg_kms(&dev_priv->drm, "FIFO size - (0x%08x) %c: %d\n",
535		    dsparb, plane_name(i9xx_plane), size);
536
537	return size;
538}
539
540static int i830_get_fifo_size(struct drm_i915_private *dev_priv,
541			      enum i9xx_plane_id i9xx_plane)
542{
543	u32 dsparb = I915_READ(DSPARB);
544	int size;
545
546	size = dsparb & 0x1ff;
547	if (i9xx_plane == PLANE_B)
548		size = ((dsparb >> DSPARB_BEND_SHIFT) & 0x1ff) - size;
549	size >>= 1; /* Convert to cachelines */
550
551	drm_dbg_kms(&dev_priv->drm, "FIFO size - (0x%08x) %c: %d\n",
552		    dsparb, plane_name(i9xx_plane), size);
553
554	return size;
555}
556
557static int i845_get_fifo_size(struct drm_i915_private *dev_priv,
558			      enum i9xx_plane_id i9xx_plane)
559{
560	u32 dsparb = I915_READ(DSPARB);
561	int size;
562
563	size = dsparb & 0x7f;
564	size >>= 2; /* Convert to cachelines */
565
566	drm_dbg_kms(&dev_priv->drm, "FIFO size - (0x%08x) %c: %d\n",
567		    dsparb, plane_name(i9xx_plane), size);
568
569	return size;
570}
571
572/* Pineview has different values for various configs */
573static const struct intel_watermark_params pnv_display_wm = {
574	.fifo_size = PINEVIEW_DISPLAY_FIFO,
575	.max_wm = PINEVIEW_MAX_WM,
576	.default_wm = PINEVIEW_DFT_WM,
577	.guard_size = PINEVIEW_GUARD_WM,
578	.cacheline_size = PINEVIEW_FIFO_LINE_SIZE,
579};
580
581static const struct intel_watermark_params pnv_display_hplloff_wm = {
582	.fifo_size = PINEVIEW_DISPLAY_FIFO,
583	.max_wm = PINEVIEW_MAX_WM,
584	.default_wm = PINEVIEW_DFT_HPLLOFF_WM,
585	.guard_size = PINEVIEW_GUARD_WM,
586	.cacheline_size = PINEVIEW_FIFO_LINE_SIZE,
587};
588
589static const struct intel_watermark_params pnv_cursor_wm = {
590	.fifo_size = PINEVIEW_CURSOR_FIFO,
591	.max_wm = PINEVIEW_CURSOR_MAX_WM,
592	.default_wm = PINEVIEW_CURSOR_DFT_WM,
593	.guard_size = PINEVIEW_CURSOR_GUARD_WM,
594	.cacheline_size = PINEVIEW_FIFO_LINE_SIZE,
595};
596
597static const struct intel_watermark_params pnv_cursor_hplloff_wm = {
598	.fifo_size = PINEVIEW_CURSOR_FIFO,
599	.max_wm = PINEVIEW_CURSOR_MAX_WM,
600	.default_wm = PINEVIEW_CURSOR_DFT_WM,
601	.guard_size = PINEVIEW_CURSOR_GUARD_WM,
602	.cacheline_size = PINEVIEW_FIFO_LINE_SIZE,
603};
604
605static const struct intel_watermark_params i965_cursor_wm_info = {
606	.fifo_size = I965_CURSOR_FIFO,
607	.max_wm = I965_CURSOR_MAX_WM,
608	.default_wm = I965_CURSOR_DFT_WM,
609	.guard_size = 2,
610	.cacheline_size = I915_FIFO_LINE_SIZE,
611};
612
613static const struct intel_watermark_params i945_wm_info = {
614	.fifo_size = I945_FIFO_SIZE,
615	.max_wm = I915_MAX_WM,
616	.default_wm = 1,
617	.guard_size = 2,
618	.cacheline_size = I915_FIFO_LINE_SIZE,
619};
620
621static const struct intel_watermark_params i915_wm_info = {
622	.fifo_size = I915_FIFO_SIZE,
623	.max_wm = I915_MAX_WM,
624	.default_wm = 1,
625	.guard_size = 2,
626	.cacheline_size = I915_FIFO_LINE_SIZE,
627};
628
629static const struct intel_watermark_params i830_a_wm_info = {
630	.fifo_size = I855GM_FIFO_SIZE,
631	.max_wm = I915_MAX_WM,
632	.default_wm = 1,
633	.guard_size = 2,
634	.cacheline_size = I830_FIFO_LINE_SIZE,
635};
636
637static const struct intel_watermark_params i830_bc_wm_info = {
638	.fifo_size = I855GM_FIFO_SIZE,
639	.max_wm = I915_MAX_WM/2,
640	.default_wm = 1,
641	.guard_size = 2,
642	.cacheline_size = I830_FIFO_LINE_SIZE,
643};
644
645static const struct intel_watermark_params i845_wm_info = {
646	.fifo_size = I830_FIFO_SIZE,
647	.max_wm = I915_MAX_WM,
648	.default_wm = 1,
649	.guard_size = 2,
650	.cacheline_size = I830_FIFO_LINE_SIZE,
651};
652
653/**
654 * intel_wm_method1 - Method 1 / "small buffer" watermark formula
655 * @pixel_rate: Pipe pixel rate in kHz
656 * @cpp: Plane bytes per pixel
657 * @latency: Memory wakeup latency in 0.1us units
658 *
659 * Compute the watermark using the method 1 or "small buffer"
660 * formula. The caller may additonally add extra cachelines
661 * to account for TLB misses and clock crossings.
662 *
663 * This method is concerned with the short term drain rate
664 * of the FIFO, ie. it does not account for blanking periods
665 * which would effectively reduce the average drain rate across
666 * a longer period. The name "small" refers to the fact the
667 * FIFO is relatively small compared to the amount of data
668 * fetched.
669 *
670 * The FIFO level vs. time graph might look something like:
671 *
672 *   |\   |\
673 *   | \  | \
674 * __---__---__ (- plane active, _ blanking)
675 * -> time
676 *
677 * or perhaps like this:
678 *
679 *   |\|\  |\|\
680 * __----__----__ (- plane active, _ blanking)
681 * -> time
682 *
683 * Returns:
684 * The watermark in bytes
685 */
686static unsigned int intel_wm_method1(unsigned int pixel_rate,
687				     unsigned int cpp,
688				     unsigned int latency)
689{
690	u64 ret;
691
692	ret = mul_u32_u32(pixel_rate, cpp * latency);
693	ret = DIV_ROUND_UP_ULL(ret, 10000);
694
695	return ret;
696}
697
698/**
699 * intel_wm_method2 - Method 2 / "large buffer" watermark formula
700 * @pixel_rate: Pipe pixel rate in kHz
701 * @htotal: Pipe horizontal total
702 * @width: Plane width in pixels
703 * @cpp: Plane bytes per pixel
704 * @latency: Memory wakeup latency in 0.1us units
705 *
706 * Compute the watermark using the method 2 or "large buffer"
707 * formula. The caller may additonally add extra cachelines
708 * to account for TLB misses and clock crossings.
709 *
710 * This method is concerned with the long term drain rate
711 * of the FIFO, ie. it does account for blanking periods
712 * which effectively reduce the average drain rate across
713 * a longer period. The name "large" refers to the fact the
714 * FIFO is relatively large compared to the amount of data
715 * fetched.
716 *
717 * The FIFO level vs. time graph might look something like:
718 *
719 *    |\___       |\___
720 *    |    \___   |    \___
721 *    |        \  |        \
722 * __ --__--__--__--__--__--__ (- plane active, _ blanking)
723 * -> time
724 *
725 * Returns:
726 * The watermark in bytes
727 */
728static unsigned int intel_wm_method2(unsigned int pixel_rate,
729				     unsigned int htotal,
730				     unsigned int width,
731				     unsigned int cpp,
732				     unsigned int latency)
733{
734	unsigned int ret;
735
736	/*
737	 * FIXME remove once all users are computing
738	 * watermarks in the correct place.
739	 */
740	if (WARN_ON_ONCE(htotal == 0))
741		htotal = 1;
742
743	ret = (latency * pixel_rate) / (htotal * 10000);
744	ret = (ret + 1) * width * cpp;
745
746	return ret;
747}
748
749/**
750 * intel_calculate_wm - calculate watermark level
751 * @pixel_rate: pixel clock
752 * @wm: chip FIFO params
753 * @fifo_size: size of the FIFO buffer
754 * @cpp: bytes per pixel
755 * @latency_ns: memory latency for the platform
756 *
757 * Calculate the watermark level (the level at which the display plane will
758 * start fetching from memory again).  Each chip has a different display
759 * FIFO size and allocation, so the caller needs to figure that out and pass
760 * in the correct intel_watermark_params structure.
761 *
762 * As the pixel clock runs, the FIFO will be drained at a rate that depends
763 * on the pixel size.  When it reaches the watermark level, it'll start
764 * fetching FIFO line sized based chunks from memory until the FIFO fills
765 * past the watermark point.  If the FIFO drains completely, a FIFO underrun
766 * will occur, and a display engine hang could result.
767 */
768static unsigned int intel_calculate_wm(int pixel_rate,
769				       const struct intel_watermark_params *wm,
770				       int fifo_size, int cpp,
771				       unsigned int latency_ns)
772{
773	int entries, wm_size;
774
775	/*
776	 * Note: we need to make sure we don't overflow for various clock &
777	 * latency values.
778	 * clocks go from a few thousand to several hundred thousand.
779	 * latency is usually a few thousand
780	 */
781	entries = intel_wm_method1(pixel_rate, cpp,
782				   latency_ns / 100);
783	entries = DIV_ROUND_UP(entries, wm->cacheline_size) +
784		wm->guard_size;
785	DRM_DEBUG_KMS("FIFO entries required for mode: %d\n", entries);
786
787	wm_size = fifo_size - entries;
788	DRM_DEBUG_KMS("FIFO watermark level: %d\n", wm_size);
789
790	/* Don't promote wm_size to unsigned... */
791	if (wm_size > wm->max_wm)
792		wm_size = wm->max_wm;
793	if (wm_size <= 0)
794		wm_size = wm->default_wm;
795
796	/*
797	 * Bspec seems to indicate that the value shouldn't be lower than
798	 * 'burst size + 1'. Certainly 830 is quite unhappy with low values.
799	 * Lets go for 8 which is the burst size since certain platforms
800	 * already use a hardcoded 8 (which is what the spec says should be
801	 * done).
802	 */
803	if (wm_size <= 8)
804		wm_size = 8;
805
806	return wm_size;
807}
808
809static bool is_disabling(int old, int new, int threshold)
810{
811	return old >= threshold && new < threshold;
812}
813
814static bool is_enabling(int old, int new, int threshold)
815{
816	return old < threshold && new >= threshold;
817}
818
819static int intel_wm_num_levels(struct drm_i915_private *dev_priv)
820{
821	return dev_priv->wm.max_level + 1;
822}
823
824static bool intel_wm_plane_visible(const struct intel_crtc_state *crtc_state,
825				   const struct intel_plane_state *plane_state)
826{
827	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
828
829	/* FIXME check the 'enable' instead */
830	if (!crtc_state->hw.active)
831		return false;
832
833	/*
834	 * Treat cursor with fb as always visible since cursor updates
835	 * can happen faster than the vrefresh rate, and the current
836	 * watermark code doesn't handle that correctly. Cursor updates
837	 * which set/clear the fb or change the cursor size are going
838	 * to get throttled by intel_legacy_cursor_update() to work
839	 * around this problem with the watermark code.
840	 */
841	if (plane->id == PLANE_CURSOR)
842		return plane_state->hw.fb != NULL;
843	else
844		return plane_state->uapi.visible;
845}
846
847static bool intel_crtc_active(struct intel_crtc *crtc)
848{
849	/* Be paranoid as we can arrive here with only partial
850	 * state retrieved from the hardware during setup.
851	 *
852	 * We can ditch the adjusted_mode.crtc_clock check as soon
853	 * as Haswell has gained clock readout/fastboot support.
854	 *
855	 * We can ditch the crtc->primary->state->fb check as soon as we can
856	 * properly reconstruct framebuffers.
857	 *
858	 * FIXME: The intel_crtc->active here should be switched to
859	 * crtc->state->active once we have proper CRTC states wired up
860	 * for atomic.
861	 */
862	return crtc->active && crtc->base.primary->state->fb &&
863		crtc->config->hw.adjusted_mode.crtc_clock;
864}
865
866static struct intel_crtc *single_enabled_crtc(struct drm_i915_private *dev_priv)
867{
868	struct intel_crtc *crtc, *enabled = NULL;
869
870	for_each_intel_crtc(&dev_priv->drm, crtc) {
871		if (intel_crtc_active(crtc)) {
872			if (enabled)
873				return NULL;
874			enabled = crtc;
875		}
876	}
877
878	return enabled;
879}
880
881static void pnv_update_wm(struct intel_crtc *unused_crtc)
882{
883	struct drm_i915_private *dev_priv = to_i915(unused_crtc->base.dev);
884	struct intel_crtc *crtc;
885	const struct cxsr_latency *latency;
886	u32 reg;
887	unsigned int wm;
888
889	latency = intel_get_cxsr_latency(!IS_MOBILE(dev_priv),
890					 dev_priv->is_ddr3,
891					 dev_priv->fsb_freq,
892					 dev_priv->mem_freq);
893	if (!latency) {
894		drm_dbg_kms(&dev_priv->drm,
895			    "Unknown FSB/MEM found, disable CxSR\n");
896		intel_set_memory_cxsr(dev_priv, false);
897		return;
898	}
899
900	crtc = single_enabled_crtc(dev_priv);
901	if (crtc) {
902		const struct drm_display_mode *adjusted_mode =
903			&crtc->config->hw.adjusted_mode;
904		const struct drm_framebuffer *fb =
905			crtc->base.primary->state->fb;
906		int cpp = fb->format->cpp[0];
907		int clock = adjusted_mode->crtc_clock;
908
909		/* Display SR */
910		wm = intel_calculate_wm(clock, &pnv_display_wm,
911					pnv_display_wm.fifo_size,
912					cpp, latency->display_sr);
913		reg = I915_READ(DSPFW1);
914		reg &= ~DSPFW_SR_MASK;
915		reg |= FW_WM(wm, SR);
916		I915_WRITE(DSPFW1, reg);
917		drm_dbg_kms(&dev_priv->drm, "DSPFW1 register is %x\n", reg);
918
919		/* cursor SR */
920		wm = intel_calculate_wm(clock, &pnv_cursor_wm,
921					pnv_display_wm.fifo_size,
922					4, latency->cursor_sr);
923		reg = I915_READ(DSPFW3);
924		reg &= ~DSPFW_CURSOR_SR_MASK;
925		reg |= FW_WM(wm, CURSOR_SR);
926		I915_WRITE(DSPFW3, reg);
927
928		/* Display HPLL off SR */
929		wm = intel_calculate_wm(clock, &pnv_display_hplloff_wm,
930					pnv_display_hplloff_wm.fifo_size,
931					cpp, latency->display_hpll_disable);
932		reg = I915_READ(DSPFW3);
933		reg &= ~DSPFW_HPLL_SR_MASK;
934		reg |= FW_WM(wm, HPLL_SR);
935		I915_WRITE(DSPFW3, reg);
936
937		/* cursor HPLL off SR */
938		wm = intel_calculate_wm(clock, &pnv_cursor_hplloff_wm,
939					pnv_display_hplloff_wm.fifo_size,
940					4, latency->cursor_hpll_disable);
941		reg = I915_READ(DSPFW3);
942		reg &= ~DSPFW_HPLL_CURSOR_MASK;
943		reg |= FW_WM(wm, HPLL_CURSOR);
944		I915_WRITE(DSPFW3, reg);
945		drm_dbg_kms(&dev_priv->drm, "DSPFW3 register is %x\n", reg);
946
947		intel_set_memory_cxsr(dev_priv, true);
948	} else {
949		intel_set_memory_cxsr(dev_priv, false);
950	}
951}
952
953/*
954 * Documentation says:
955 * "If the line size is small, the TLB fetches can get in the way of the
956 *  data fetches, causing some lag in the pixel data return which is not
957 *  accounted for in the above formulas. The following adjustment only
958 *  needs to be applied if eight whole lines fit in the buffer at once.
959 *  The WM is adjusted upwards by the difference between the FIFO size
960 *  and the size of 8 whole lines. This adjustment is always performed
961 *  in the actual pixel depth regardless of whether FBC is enabled or not."
962 */
963static unsigned int g4x_tlb_miss_wa(int fifo_size, int width, int cpp)
964{
965	int tlb_miss = fifo_size * 64 - width * cpp * 8;
966
967	return max(0, tlb_miss);
968}
969
970static void g4x_write_wm_values(struct drm_i915_private *dev_priv,
971				const struct g4x_wm_values *wm)
972{
973	enum pipe pipe;
974
975	for_each_pipe(dev_priv, pipe)
976		trace_g4x_wm(intel_get_crtc_for_pipe(dev_priv, pipe), wm);
977
978	I915_WRITE(DSPFW1,
979		   FW_WM(wm->sr.plane, SR) |
980		   FW_WM(wm->pipe[PIPE_B].plane[PLANE_CURSOR], CURSORB) |
981		   FW_WM(wm->pipe[PIPE_B].plane[PLANE_PRIMARY], PLANEB) |
982		   FW_WM(wm->pipe[PIPE_A].plane[PLANE_PRIMARY], PLANEA));
983	I915_WRITE(DSPFW2,
984		   (wm->fbc_en ? DSPFW_FBC_SR_EN : 0) |
985		   FW_WM(wm->sr.fbc, FBC_SR) |
986		   FW_WM(wm->hpll.fbc, FBC_HPLL_SR) |
987		   FW_WM(wm->pipe[PIPE_B].plane[PLANE_SPRITE0], SPRITEB) |
988		   FW_WM(wm->pipe[PIPE_A].plane[PLANE_CURSOR], CURSORA) |
989		   FW_WM(wm->pipe[PIPE_A].plane[PLANE_SPRITE0], SPRITEA));
990	I915_WRITE(DSPFW3,
991		   (wm->hpll_en ? DSPFW_HPLL_SR_EN : 0) |
992		   FW_WM(wm->sr.cursor, CURSOR_SR) |
993		   FW_WM(wm->hpll.cursor, HPLL_CURSOR) |
994		   FW_WM(wm->hpll.plane, HPLL_SR));
995
996	POSTING_READ(DSPFW1);
997}
998
999#define FW_WM_VLV(value, plane) \
1000	(((value) << DSPFW_ ## plane ## _SHIFT) & DSPFW_ ## plane ## _MASK_VLV)
1001
1002static void vlv_write_wm_values(struct drm_i915_private *dev_priv,
1003				const struct vlv_wm_values *wm)
1004{
1005	enum pipe pipe;
1006
1007	for_each_pipe(dev_priv, pipe) {
1008		trace_vlv_wm(intel_get_crtc_for_pipe(dev_priv, pipe), wm);
1009
1010		I915_WRITE(VLV_DDL(pipe),
1011			   (wm->ddl[pipe].plane[PLANE_CURSOR] << DDL_CURSOR_SHIFT) |
1012			   (wm->ddl[pipe].plane[PLANE_SPRITE1] << DDL_SPRITE_SHIFT(1)) |
1013			   (wm->ddl[pipe].plane[PLANE_SPRITE0] << DDL_SPRITE_SHIFT(0)) |
1014			   (wm->ddl[pipe].plane[PLANE_PRIMARY] << DDL_PLANE_SHIFT));
1015	}
1016
1017	/*
1018	 * Zero the (unused) WM1 watermarks, and also clear all the
1019	 * high order bits so that there are no out of bounds values
1020	 * present in the registers during the reprogramming.
1021	 */
1022	I915_WRITE(DSPHOWM, 0);
1023	I915_WRITE(DSPHOWM1, 0);
1024	I915_WRITE(DSPFW4, 0);
1025	I915_WRITE(DSPFW5, 0);
1026	I915_WRITE(DSPFW6, 0);
1027
1028	I915_WRITE(DSPFW1,
1029		   FW_WM(wm->sr.plane, SR) |
1030		   FW_WM(wm->pipe[PIPE_B].plane[PLANE_CURSOR], CURSORB) |
1031		   FW_WM_VLV(wm->pipe[PIPE_B].plane[PLANE_PRIMARY], PLANEB) |
1032		   FW_WM_VLV(wm->pipe[PIPE_A].plane[PLANE_PRIMARY], PLANEA));
1033	I915_WRITE(DSPFW2,
1034		   FW_WM_VLV(wm->pipe[PIPE_A].plane[PLANE_SPRITE1], SPRITEB) |
1035		   FW_WM(wm->pipe[PIPE_A].plane[PLANE_CURSOR], CURSORA) |
1036		   FW_WM_VLV(wm->pipe[PIPE_A].plane[PLANE_SPRITE0], SPRITEA));
1037	I915_WRITE(DSPFW3,
1038		   FW_WM(wm->sr.cursor, CURSOR_SR));
1039
1040	if (IS_CHERRYVIEW(dev_priv)) {
1041		I915_WRITE(DSPFW7_CHV,
1042			   FW_WM_VLV(wm->pipe[PIPE_B].plane[PLANE_SPRITE1], SPRITED) |
1043			   FW_WM_VLV(wm->pipe[PIPE_B].plane[PLANE_SPRITE0], SPRITEC));
1044		I915_WRITE(DSPFW8_CHV,
1045			   FW_WM_VLV(wm->pipe[PIPE_C].plane[PLANE_SPRITE1], SPRITEF) |
1046			   FW_WM_VLV(wm->pipe[PIPE_C].plane[PLANE_SPRITE0], SPRITEE));
1047		I915_WRITE(DSPFW9_CHV,
1048			   FW_WM_VLV(wm->pipe[PIPE_C].plane[PLANE_PRIMARY], PLANEC) |
1049			   FW_WM(wm->pipe[PIPE_C].plane[PLANE_CURSOR], CURSORC));
1050		I915_WRITE(DSPHOWM,
1051			   FW_WM(wm->sr.plane >> 9, SR_HI) |
1052			   FW_WM(wm->pipe[PIPE_C].plane[PLANE_SPRITE1] >> 8, SPRITEF_HI) |
1053			   FW_WM(wm->pipe[PIPE_C].plane[PLANE_SPRITE0] >> 8, SPRITEE_HI) |
1054			   FW_WM(wm->pipe[PIPE_C].plane[PLANE_PRIMARY] >> 8, PLANEC_HI) |
1055			   FW_WM(wm->pipe[PIPE_B].plane[PLANE_SPRITE1] >> 8, SPRITED_HI) |
1056			   FW_WM(wm->pipe[PIPE_B].plane[PLANE_SPRITE0] >> 8, SPRITEC_HI) |
1057			   FW_WM(wm->pipe[PIPE_B].plane[PLANE_PRIMARY] >> 8, PLANEB_HI) |
1058			   FW_WM(wm->pipe[PIPE_A].plane[PLANE_SPRITE1] >> 8, SPRITEB_HI) |
1059			   FW_WM(wm->pipe[PIPE_A].plane[PLANE_SPRITE0] >> 8, SPRITEA_HI) |
1060			   FW_WM(wm->pipe[PIPE_A].plane[PLANE_PRIMARY] >> 8, PLANEA_HI));
1061	} else {
1062		I915_WRITE(DSPFW7,
1063			   FW_WM_VLV(wm->pipe[PIPE_B].plane[PLANE_SPRITE1], SPRITED) |
1064			   FW_WM_VLV(wm->pipe[PIPE_B].plane[PLANE_SPRITE0], SPRITEC));
1065		I915_WRITE(DSPHOWM,
1066			   FW_WM(wm->sr.plane >> 9, SR_HI) |
1067			   FW_WM(wm->pipe[PIPE_B].plane[PLANE_SPRITE1] >> 8, SPRITED_HI) |
1068			   FW_WM(wm->pipe[PIPE_B].plane[PLANE_SPRITE0] >> 8, SPRITEC_HI) |
1069			   FW_WM(wm->pipe[PIPE_B].plane[PLANE_PRIMARY] >> 8, PLANEB_HI) |
1070			   FW_WM(wm->pipe[PIPE_A].plane[PLANE_SPRITE1] >> 8, SPRITEB_HI) |
1071			   FW_WM(wm->pipe[PIPE_A].plane[PLANE_SPRITE0] >> 8, SPRITEA_HI) |
1072			   FW_WM(wm->pipe[PIPE_A].plane[PLANE_PRIMARY] >> 8, PLANEA_HI));
1073	}
1074
1075	POSTING_READ(DSPFW1);
1076}
1077
1078#undef FW_WM_VLV
1079
1080static void g4x_setup_wm_latency(struct drm_i915_private *dev_priv)
1081{
1082	/* all latencies in usec */
1083	dev_priv->wm.pri_latency[G4X_WM_LEVEL_NORMAL] = 5;
1084	dev_priv->wm.pri_latency[G4X_WM_LEVEL_SR] = 12;
1085	dev_priv->wm.pri_latency[G4X_WM_LEVEL_HPLL] = 35;
1086
1087	dev_priv->wm.max_level = G4X_WM_LEVEL_HPLL;
1088}
1089
1090static int g4x_plane_fifo_size(enum plane_id plane_id, int level)
1091{
1092	/*
1093	 * DSPCNTR[13] supposedly controls whether the
1094	 * primary plane can use the FIFO space otherwise
1095	 * reserved for the sprite plane. It's not 100% clear
1096	 * what the actual FIFO size is, but it looks like we
1097	 * can happily set both primary and sprite watermarks
1098	 * up to 127 cachelines. So that would seem to mean
1099	 * that either DSPCNTR[13] doesn't do anything, or that
1100	 * the total FIFO is >= 256 cachelines in size. Either
1101	 * way, we don't seem to have to worry about this
1102	 * repartitioning as the maximum watermark value the
1103	 * register can hold for each plane is lower than the
1104	 * minimum FIFO size.
1105	 */
1106	switch (plane_id) {
1107	case PLANE_CURSOR:
1108		return 63;
1109	case PLANE_PRIMARY:
1110		return level == G4X_WM_LEVEL_NORMAL ? 127 : 511;
1111	case PLANE_SPRITE0:
1112		return level == G4X_WM_LEVEL_NORMAL ? 127 : 0;
1113	default:
1114		MISSING_CASE(plane_id);
1115		return 0;
1116	}
1117}
1118
1119static int g4x_fbc_fifo_size(int level)
1120{
1121	switch (level) {
1122	case G4X_WM_LEVEL_SR:
1123		return 7;
1124	case G4X_WM_LEVEL_HPLL:
1125		return 15;
1126	default:
1127		MISSING_CASE(level);
1128		return 0;
1129	}
1130}
1131
1132static u16 g4x_compute_wm(const struct intel_crtc_state *crtc_state,
1133			  const struct intel_plane_state *plane_state,
1134			  int level)
1135{
1136	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
1137	struct drm_i915_private *dev_priv = to_i915(plane->base.dev);
1138	const struct drm_display_mode *adjusted_mode =
1139		&crtc_state->hw.adjusted_mode;
1140	unsigned int latency = dev_priv->wm.pri_latency[level] * 10;
1141	unsigned int clock, htotal, cpp, width, wm;
1142
1143	if (latency == 0)
1144		return USHRT_MAX;
1145
1146	if (!intel_wm_plane_visible(crtc_state, plane_state))
1147		return 0;
1148
1149	cpp = plane_state->hw.fb->format->cpp[0];
1150
1151	/*
1152	 * Not 100% sure which way ELK should go here as the
1153	 * spec only says CL/CTG should assume 32bpp and BW
1154	 * doesn't need to. But as these things followed the
1155	 * mobile vs. desktop lines on gen3 as well, let's
1156	 * assume ELK doesn't need this.
1157	 *
1158	 * The spec also fails to list such a restriction for
1159	 * the HPLL watermark, which seems a little strange.
1160	 * Let's use 32bpp for the HPLL watermark as well.
1161	 */
1162	if (IS_GM45(dev_priv) && plane->id == PLANE_PRIMARY &&
1163	    level != G4X_WM_LEVEL_NORMAL)
1164		cpp = max(cpp, 4u);
1165
1166	clock = adjusted_mode->crtc_clock;
1167	htotal = adjusted_mode->crtc_htotal;
1168
1169	width = drm_rect_width(&plane_state->uapi.dst);
1170
1171	if (plane->id == PLANE_CURSOR) {
1172		wm = intel_wm_method2(clock, htotal, width, cpp, latency);
1173	} else if (plane->id == PLANE_PRIMARY &&
1174		   level == G4X_WM_LEVEL_NORMAL) {
1175		wm = intel_wm_method1(clock, cpp, latency);
1176	} else {
1177		unsigned int small, large;
1178
1179		small = intel_wm_method1(clock, cpp, latency);
1180		large = intel_wm_method2(clock, htotal, width, cpp, latency);
1181
1182		wm = min(small, large);
1183	}
1184
1185	wm += g4x_tlb_miss_wa(g4x_plane_fifo_size(plane->id, level),
1186			      width, cpp);
1187
1188	wm = DIV_ROUND_UP(wm, 64) + 2;
1189
1190	return min_t(unsigned int, wm, USHRT_MAX);
1191}
1192
1193static bool g4x_raw_plane_wm_set(struct intel_crtc_state *crtc_state,
1194				 int level, enum plane_id plane_id, u16 value)
1195{
1196	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
1197	bool dirty = false;
1198
1199	for (; level < intel_wm_num_levels(dev_priv); level++) {
1200		struct g4x_pipe_wm *raw = &crtc_state->wm.g4x.raw[level];
1201
1202		dirty |= raw->plane[plane_id] != value;
1203		raw->plane[plane_id] = value;
1204	}
1205
1206	return dirty;
1207}
1208
1209static bool g4x_raw_fbc_wm_set(struct intel_crtc_state *crtc_state,
1210			       int level, u16 value)
1211{
1212	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
1213	bool dirty = false;
1214
1215	/* NORMAL level doesn't have an FBC watermark */
1216	level = max(level, G4X_WM_LEVEL_SR);
1217
1218	for (; level < intel_wm_num_levels(dev_priv); level++) {
1219		struct g4x_pipe_wm *raw = &crtc_state->wm.g4x.raw[level];
1220
1221		dirty |= raw->fbc != value;
1222		raw->fbc = value;
1223	}
1224
1225	return dirty;
1226}
1227
1228static u32 ilk_compute_fbc_wm(const struct intel_crtc_state *crtc_state,
1229			      const struct intel_plane_state *plane_state,
1230			      u32 pri_val);
1231
1232static bool g4x_raw_plane_wm_compute(struct intel_crtc_state *crtc_state,
1233				     const struct intel_plane_state *plane_state)
1234{
1235	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
1236	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
1237	int num_levels = intel_wm_num_levels(to_i915(plane->base.dev));
1238	enum plane_id plane_id = plane->id;
1239	bool dirty = false;
1240	int level;
1241
1242	if (!intel_wm_plane_visible(crtc_state, plane_state)) {
1243		dirty |= g4x_raw_plane_wm_set(crtc_state, 0, plane_id, 0);
1244		if (plane_id == PLANE_PRIMARY)
1245			dirty |= g4x_raw_fbc_wm_set(crtc_state, 0, 0);
1246		goto out;
1247	}
1248
1249	for (level = 0; level < num_levels; level++) {
1250		struct g4x_pipe_wm *raw = &crtc_state->wm.g4x.raw[level];
1251		int wm, max_wm;
1252
1253		wm = g4x_compute_wm(crtc_state, plane_state, level);
1254		max_wm = g4x_plane_fifo_size(plane_id, level);
1255
1256		if (wm > max_wm)
1257			break;
1258
1259		dirty |= raw->plane[plane_id] != wm;
1260		raw->plane[plane_id] = wm;
1261
1262		if (plane_id != PLANE_PRIMARY ||
1263		    level == G4X_WM_LEVEL_NORMAL)
1264			continue;
1265
1266		wm = ilk_compute_fbc_wm(crtc_state, plane_state,
1267					raw->plane[plane_id]);
1268		max_wm = g4x_fbc_fifo_size(level);
1269
1270		/*
1271		 * FBC wm is not mandatory as we
1272		 * can always just disable its use.
1273		 */
1274		if (wm > max_wm)
1275			wm = USHRT_MAX;
1276
1277		dirty |= raw->fbc != wm;
1278		raw->fbc = wm;
1279	}
1280
1281	/* mark watermarks as invalid */
1282	dirty |= g4x_raw_plane_wm_set(crtc_state, level, plane_id, USHRT_MAX);
1283
1284	if (plane_id == PLANE_PRIMARY)
1285		dirty |= g4x_raw_fbc_wm_set(crtc_state, level, USHRT_MAX);
1286
1287 out:
1288	if (dirty) {
1289		drm_dbg_kms(&dev_priv->drm,
1290			    "%s watermarks: normal=%d, SR=%d, HPLL=%d\n",
1291			    plane->base.name,
1292			    crtc_state->wm.g4x.raw[G4X_WM_LEVEL_NORMAL].plane[plane_id],
1293			    crtc_state->wm.g4x.raw[G4X_WM_LEVEL_SR].plane[plane_id],
1294			    crtc_state->wm.g4x.raw[G4X_WM_LEVEL_HPLL].plane[plane_id]);
1295
1296		if (plane_id == PLANE_PRIMARY)
1297			drm_dbg_kms(&dev_priv->drm,
1298				    "FBC watermarks: SR=%d, HPLL=%d\n",
1299				    crtc_state->wm.g4x.raw[G4X_WM_LEVEL_SR].fbc,
1300				    crtc_state->wm.g4x.raw[G4X_WM_LEVEL_HPLL].fbc);
1301	}
1302
1303	return dirty;
1304}
1305
1306static bool g4x_raw_plane_wm_is_valid(const struct intel_crtc_state *crtc_state,
1307				      enum plane_id plane_id, int level)
1308{
1309	const struct g4x_pipe_wm *raw = &crtc_state->wm.g4x.raw[level];
1310
1311	return raw->plane[plane_id] <= g4x_plane_fifo_size(plane_id, level);
1312}
1313
1314static bool g4x_raw_crtc_wm_is_valid(const struct intel_crtc_state *crtc_state,
1315				     int level)
1316{
1317	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
1318
1319	if (level > dev_priv->wm.max_level)
1320		return false;
1321
1322	return g4x_raw_plane_wm_is_valid(crtc_state, PLANE_PRIMARY, level) &&
1323		g4x_raw_plane_wm_is_valid(crtc_state, PLANE_SPRITE0, level) &&
1324		g4x_raw_plane_wm_is_valid(crtc_state, PLANE_CURSOR, level);
1325}
1326
1327/* mark all levels starting from 'level' as invalid */
1328static void g4x_invalidate_wms(struct intel_crtc *crtc,
1329			       struct g4x_wm_state *wm_state, int level)
1330{
1331	if (level <= G4X_WM_LEVEL_NORMAL) {
1332		enum plane_id plane_id;
1333
1334		for_each_plane_id_on_crtc(crtc, plane_id)
1335			wm_state->wm.plane[plane_id] = USHRT_MAX;
1336	}
1337
1338	if (level <= G4X_WM_LEVEL_SR) {
1339		wm_state->cxsr = false;
1340		wm_state->sr.cursor = USHRT_MAX;
1341		wm_state->sr.plane = USHRT_MAX;
1342		wm_state->sr.fbc = USHRT_MAX;
1343	}
1344
1345	if (level <= G4X_WM_LEVEL_HPLL) {
1346		wm_state->hpll_en = false;
1347		wm_state->hpll.cursor = USHRT_MAX;
1348		wm_state->hpll.plane = USHRT_MAX;
1349		wm_state->hpll.fbc = USHRT_MAX;
1350	}
1351}
1352
1353static bool g4x_compute_fbc_en(const struct g4x_wm_state *wm_state,
1354			       int level)
1355{
1356	if (level < G4X_WM_LEVEL_SR)
1357		return false;
1358
1359	if (level >= G4X_WM_LEVEL_SR &&
1360	    wm_state->sr.fbc > g4x_fbc_fifo_size(G4X_WM_LEVEL_SR))
1361		return false;
1362
1363	if (level >= G4X_WM_LEVEL_HPLL &&
1364	    wm_state->hpll.fbc > g4x_fbc_fifo_size(G4X_WM_LEVEL_HPLL))
1365		return false;
1366
1367	return true;
1368}
1369
1370static int g4x_compute_pipe_wm(struct intel_crtc_state *crtc_state)
1371{
1372	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
1373	struct intel_atomic_state *state =
1374		to_intel_atomic_state(crtc_state->uapi.state);
1375	struct g4x_wm_state *wm_state = &crtc_state->wm.g4x.optimal;
1376	int num_active_planes = hweight8(crtc_state->active_planes &
1377					 ~BIT(PLANE_CURSOR));
1378	const struct g4x_pipe_wm *raw;
1379	const struct intel_plane_state *old_plane_state;
1380	const struct intel_plane_state *new_plane_state;
1381	struct intel_plane *plane;
1382	enum plane_id plane_id;
1383	int i, level;
1384	unsigned int dirty = 0;
1385
1386	for_each_oldnew_intel_plane_in_state(state, plane,
1387					     old_plane_state,
1388					     new_plane_state, i) {
1389		if (new_plane_state->hw.crtc != &crtc->base &&
1390		    old_plane_state->hw.crtc != &crtc->base)
1391			continue;
1392
1393		if (g4x_raw_plane_wm_compute(crtc_state, new_plane_state))
1394			dirty |= BIT(plane->id);
1395	}
1396
1397	if (!dirty)
1398		return 0;
1399
1400	level = G4X_WM_LEVEL_NORMAL;
1401	if (!g4x_raw_crtc_wm_is_valid(crtc_state, level))
1402		goto out;
1403
1404	raw = &crtc_state->wm.g4x.raw[level];
1405	for_each_plane_id_on_crtc(crtc, plane_id)
1406		wm_state->wm.plane[plane_id] = raw->plane[plane_id];
1407
1408	level = G4X_WM_LEVEL_SR;
1409	if (!g4x_raw_crtc_wm_is_valid(crtc_state, level))
1410		goto out;
1411
1412	raw = &crtc_state->wm.g4x.raw[level];
1413	wm_state->sr.plane = raw->plane[PLANE_PRIMARY];
1414	wm_state->sr.cursor = raw->plane[PLANE_CURSOR];
1415	wm_state->sr.fbc = raw->fbc;
1416
1417	wm_state->cxsr = num_active_planes == BIT(PLANE_PRIMARY);
1418
1419	level = G4X_WM_LEVEL_HPLL;
1420	if (!g4x_raw_crtc_wm_is_valid(crtc_state, level))
1421		goto out;
1422
1423	raw = &crtc_state->wm.g4x.raw[level];
1424	wm_state->hpll.plane = raw->plane[PLANE_PRIMARY];
1425	wm_state->hpll.cursor = raw->plane[PLANE_CURSOR];
1426	wm_state->hpll.fbc = raw->fbc;
1427
1428	wm_state->hpll_en = wm_state->cxsr;
1429
1430	level++;
1431
1432 out:
1433	if (level == G4X_WM_LEVEL_NORMAL)
1434		return -EINVAL;
1435
1436	/* invalidate the higher levels */
1437	g4x_invalidate_wms(crtc, wm_state, level);
1438
1439	/*
1440	 * Determine if the FBC watermark(s) can be used. IF
1441	 * this isn't the case we prefer to disable the FBC
1442	 * watermark(s) rather than disable the SR/HPLL
1443	 * level(s) entirely. 'level-1' is the highest valid
1444	 * level here.
1445	 */
1446	wm_state->fbc_en = g4x_compute_fbc_en(wm_state, level - 1);
1447
1448	return 0;
1449}
1450
1451static int g4x_compute_intermediate_wm(struct intel_crtc_state *new_crtc_state)
1452{
1453	struct intel_crtc *crtc = to_intel_crtc(new_crtc_state->uapi.crtc);
1454	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
1455	struct g4x_wm_state *intermediate = &new_crtc_state->wm.g4x.intermediate;
1456	const struct g4x_wm_state *optimal = &new_crtc_state->wm.g4x.optimal;
1457	struct intel_atomic_state *intel_state =
1458		to_intel_atomic_state(new_crtc_state->uapi.state);
1459	const struct intel_crtc_state *old_crtc_state =
1460		intel_atomic_get_old_crtc_state(intel_state, crtc);
1461	const struct g4x_wm_state *active = &old_crtc_state->wm.g4x.optimal;
1462	enum plane_id plane_id;
1463
1464	if (!new_crtc_state->hw.active || drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi)) {
1465		*intermediate = *optimal;
1466
1467		intermediate->cxsr = false;
1468		intermediate->hpll_en = false;
1469		goto out;
1470	}
1471
1472	intermediate->cxsr = optimal->cxsr && active->cxsr &&
1473		!new_crtc_state->disable_cxsr;
1474	intermediate->hpll_en = optimal->hpll_en && active->hpll_en &&
1475		!new_crtc_state->disable_cxsr;
1476	intermediate->fbc_en = optimal->fbc_en && active->fbc_en;
1477
1478	for_each_plane_id_on_crtc(crtc, plane_id) {
1479		intermediate->wm.plane[plane_id] =
1480			max(optimal->wm.plane[plane_id],
1481			    active->wm.plane[plane_id]);
1482
1483		drm_WARN_ON(&dev_priv->drm, intermediate->wm.plane[plane_id] >
1484			    g4x_plane_fifo_size(plane_id, G4X_WM_LEVEL_NORMAL));
1485	}
1486
1487	intermediate->sr.plane = max(optimal->sr.plane,
1488				     active->sr.plane);
1489	intermediate->sr.cursor = max(optimal->sr.cursor,
1490				      active->sr.cursor);
1491	intermediate->sr.fbc = max(optimal->sr.fbc,
1492				   active->sr.fbc);
1493
1494	intermediate->hpll.plane = max(optimal->hpll.plane,
1495				       active->hpll.plane);
1496	intermediate->hpll.cursor = max(optimal->hpll.cursor,
1497					active->hpll.cursor);
1498	intermediate->hpll.fbc = max(optimal->hpll.fbc,
1499				     active->hpll.fbc);
1500
1501	drm_WARN_ON(&dev_priv->drm,
1502		    (intermediate->sr.plane >
1503		     g4x_plane_fifo_size(PLANE_PRIMARY, G4X_WM_LEVEL_SR) ||
1504		     intermediate->sr.cursor >
1505		     g4x_plane_fifo_size(PLANE_CURSOR, G4X_WM_LEVEL_SR)) &&
1506		    intermediate->cxsr);
1507	drm_WARN_ON(&dev_priv->drm,
1508		    (intermediate->sr.plane >
1509		     g4x_plane_fifo_size(PLANE_PRIMARY, G4X_WM_LEVEL_HPLL) ||
1510		     intermediate->sr.cursor >
1511		     g4x_plane_fifo_size(PLANE_CURSOR, G4X_WM_LEVEL_HPLL)) &&
1512		    intermediate->hpll_en);
1513
1514	drm_WARN_ON(&dev_priv->drm,
1515		    intermediate->sr.fbc > g4x_fbc_fifo_size(1) &&
1516		    intermediate->fbc_en && intermediate->cxsr);
1517	drm_WARN_ON(&dev_priv->drm,
1518		    intermediate->hpll.fbc > g4x_fbc_fifo_size(2) &&
1519		    intermediate->fbc_en && intermediate->hpll_en);
1520
1521out:
1522	/*
1523	 * If our intermediate WM are identical to the final WM, then we can
1524	 * omit the post-vblank programming; only update if it's different.
1525	 */
1526	if (memcmp(intermediate, optimal, sizeof(*intermediate)) != 0)
1527		new_crtc_state->wm.need_postvbl_update = true;
1528
1529	return 0;
1530}
1531
1532static void g4x_merge_wm(struct drm_i915_private *dev_priv,
1533			 struct g4x_wm_values *wm)
1534{
1535	struct intel_crtc *crtc;
1536	int num_active_pipes = 0;
1537
1538	wm->cxsr = true;
1539	wm->hpll_en = true;
1540	wm->fbc_en = true;
1541
1542	for_each_intel_crtc(&dev_priv->drm, crtc) {
1543		const struct g4x_wm_state *wm_state = &crtc->wm.active.g4x;
1544
1545		if (!crtc->active)
1546			continue;
1547
1548		if (!wm_state->cxsr)
1549			wm->cxsr = false;
1550		if (!wm_state->hpll_en)
1551			wm->hpll_en = false;
1552		if (!wm_state->fbc_en)
1553			wm->fbc_en = false;
1554
1555		num_active_pipes++;
1556	}
1557
1558	if (num_active_pipes != 1) {
1559		wm->cxsr = false;
1560		wm->hpll_en = false;
1561		wm->fbc_en = false;
1562	}
1563
1564	for_each_intel_crtc(&dev_priv->drm, crtc) {
1565		const struct g4x_wm_state *wm_state = &crtc->wm.active.g4x;
1566		enum pipe pipe = crtc->pipe;
1567
1568		wm->pipe[pipe] = wm_state->wm;
1569		if (crtc->active && wm->cxsr)
1570			wm->sr = wm_state->sr;
1571		if (crtc->active && wm->hpll_en)
1572			wm->hpll = wm_state->hpll;
1573	}
1574}
1575
1576static void g4x_program_watermarks(struct drm_i915_private *dev_priv)
1577{
1578	struct g4x_wm_values *old_wm = &dev_priv->wm.g4x;
1579	struct g4x_wm_values new_wm = {};
1580
1581	g4x_merge_wm(dev_priv, &new_wm);
1582
1583	if (memcmp(old_wm, &new_wm, sizeof(new_wm)) == 0)
1584		return;
1585
1586	if (is_disabling(old_wm->cxsr, new_wm.cxsr, true))
1587		_intel_set_memory_cxsr(dev_priv, false);
1588
1589	g4x_write_wm_values(dev_priv, &new_wm);
1590
1591	if (is_enabling(old_wm->cxsr, new_wm.cxsr, true))
1592		_intel_set_memory_cxsr(dev_priv, true);
1593
1594	*old_wm = new_wm;
1595}
1596
1597static void g4x_initial_watermarks(struct intel_atomic_state *state,
1598				   struct intel_crtc *crtc)
1599{
1600	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
1601	const struct intel_crtc_state *crtc_state =
1602		intel_atomic_get_new_crtc_state(state, crtc);
1603
1604	mutex_lock(&dev_priv->wm.wm_mutex);
1605	crtc->wm.active.g4x = crtc_state->wm.g4x.intermediate;
1606	g4x_program_watermarks(dev_priv);
1607	mutex_unlock(&dev_priv->wm.wm_mutex);
1608}
1609
1610static void g4x_optimize_watermarks(struct intel_atomic_state *state,
1611				    struct intel_crtc *crtc)
1612{
1613	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
1614	const struct intel_crtc_state *crtc_state =
1615		intel_atomic_get_new_crtc_state(state, crtc);
1616
1617	if (!crtc_state->wm.need_postvbl_update)
1618		return;
1619
1620	mutex_lock(&dev_priv->wm.wm_mutex);
1621	crtc->wm.active.g4x = crtc_state->wm.g4x.optimal;
1622	g4x_program_watermarks(dev_priv);
1623	mutex_unlock(&dev_priv->wm.wm_mutex);
1624}
1625
1626/* latency must be in 0.1us units. */
1627static unsigned int vlv_wm_method2(unsigned int pixel_rate,
1628				   unsigned int htotal,
1629				   unsigned int width,
1630				   unsigned int cpp,
1631				   unsigned int latency)
1632{
1633	unsigned int ret;
1634
1635	ret = intel_wm_method2(pixel_rate, htotal,
1636			       width, cpp, latency);
1637	ret = DIV_ROUND_UP(ret, 64);
1638
1639	return ret;
1640}
1641
1642static void vlv_setup_wm_latency(struct drm_i915_private *dev_priv)
1643{
1644	/* all latencies in usec */
1645	dev_priv->wm.pri_latency[VLV_WM_LEVEL_PM2] = 3;
1646
1647	dev_priv->wm.max_level = VLV_WM_LEVEL_PM2;
1648
1649	if (IS_CHERRYVIEW(dev_priv)) {
1650		dev_priv->wm.pri_latency[VLV_WM_LEVEL_PM5] = 12;
1651		dev_priv->wm.pri_latency[VLV_WM_LEVEL_DDR_DVFS] = 33;
1652
1653		dev_priv->wm.max_level = VLV_WM_LEVEL_DDR_DVFS;
1654	}
1655}
1656
1657static u16 vlv_compute_wm_level(const struct intel_crtc_state *crtc_state,
1658				const struct intel_plane_state *plane_state,
1659				int level)
1660{
1661	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
1662	struct drm_i915_private *dev_priv = to_i915(plane->base.dev);
1663	const struct drm_display_mode *adjusted_mode =
1664		&crtc_state->hw.adjusted_mode;
1665	unsigned int clock, htotal, cpp, width, wm;
1666
1667	if (dev_priv->wm.pri_latency[level] == 0)
1668		return USHRT_MAX;
1669
1670	if (!intel_wm_plane_visible(crtc_state, plane_state))
1671		return 0;
1672
1673	cpp = plane_state->hw.fb->format->cpp[0];
1674	clock = adjusted_mode->crtc_clock;
1675	htotal = adjusted_mode->crtc_htotal;
1676	width = crtc_state->pipe_src_w;
1677
1678	if (plane->id == PLANE_CURSOR) {
1679		/*
1680		 * FIXME the formula gives values that are
1681		 * too big for the cursor FIFO, and hence we
1682		 * would never be able to use cursors. For
1683		 * now just hardcode the watermark.
1684		 */
1685		wm = 63;
1686	} else {
1687		wm = vlv_wm_method2(clock, htotal, width, cpp,
1688				    dev_priv->wm.pri_latency[level] * 10);
1689	}
1690
1691	return min_t(unsigned int, wm, USHRT_MAX);
1692}
1693
1694static bool vlv_need_sprite0_fifo_workaround(unsigned int active_planes)
1695{
1696	return (active_planes & (BIT(PLANE_SPRITE0) |
1697				 BIT(PLANE_SPRITE1))) == BIT(PLANE_SPRITE1);
1698}
1699
1700static int vlv_compute_fifo(struct intel_crtc_state *crtc_state)
1701{
1702	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
1703	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
1704	const struct g4x_pipe_wm *raw =
1705		&crtc_state->wm.vlv.raw[VLV_WM_LEVEL_PM2];
1706	struct vlv_fifo_state *fifo_state = &crtc_state->wm.vlv.fifo_state;
1707	unsigned int active_planes = crtc_state->active_planes & ~BIT(PLANE_CURSOR);
1708	int num_active_planes = hweight8(active_planes);
1709	const int fifo_size = 511;
1710	int fifo_extra, fifo_left = fifo_size;
1711	int sprite0_fifo_extra = 0;
1712	unsigned int total_rate;
1713	enum plane_id plane_id;
1714
1715	/*
1716	 * When enabling sprite0 after sprite1 has already been enabled
1717	 * we tend to get an underrun unless sprite0 already has some
1718	 * FIFO space allcoated. Hence we always allocate at least one
1719	 * cacheline for sprite0 whenever sprite1 is enabled.
1720	 *
1721	 * All other plane enable sequences appear immune to this problem.
1722	 */
1723	if (vlv_need_sprite0_fifo_workaround(active_planes))
1724		sprite0_fifo_extra = 1;
1725
1726	total_rate = raw->plane[PLANE_PRIMARY] +
1727		raw->plane[PLANE_SPRITE0] +
1728		raw->plane[PLANE_SPRITE1] +
1729		sprite0_fifo_extra;
1730
1731	if (total_rate > fifo_size)
1732		return -EINVAL;
1733
1734	if (total_rate == 0)
1735		total_rate = 1;
1736
1737	for_each_plane_id_on_crtc(crtc, plane_id) {
1738		unsigned int rate;
1739
1740		if ((active_planes & BIT(plane_id)) == 0) {
1741			fifo_state->plane[plane_id] = 0;
1742			continue;
1743		}
1744
1745		rate = raw->plane[plane_id];
1746		fifo_state->plane[plane_id] = fifo_size * rate / total_rate;
1747		fifo_left -= fifo_state->plane[plane_id];
1748	}
1749
1750	fifo_state->plane[PLANE_SPRITE0] += sprite0_fifo_extra;
1751	fifo_left -= sprite0_fifo_extra;
1752
1753	fifo_state->plane[PLANE_CURSOR] = 63;
1754
1755	fifo_extra = DIV_ROUND_UP(fifo_left, num_active_planes ?: 1);
1756
1757	/* spread the remainder evenly */
1758	for_each_plane_id_on_crtc(crtc, plane_id) {
1759		int plane_extra;
1760
1761		if (fifo_left == 0)
1762			break;
1763
1764		if ((active_planes & BIT(plane_id)) == 0)
1765			continue;
1766
1767		plane_extra = min(fifo_extra, fifo_left);
1768		fifo_state->plane[plane_id] += plane_extra;
1769		fifo_left -= plane_extra;
1770	}
1771
1772	drm_WARN_ON(&dev_priv->drm, active_planes != 0 && fifo_left != 0);
1773
1774	/* give it all to the first plane if none are active */
1775	if (active_planes == 0) {
1776		drm_WARN_ON(&dev_priv->drm, fifo_left != fifo_size);
1777		fifo_state->plane[PLANE_PRIMARY] = fifo_left;
1778	}
1779
1780	return 0;
1781}
1782
1783/* mark all levels starting from 'level' as invalid */
1784static void vlv_invalidate_wms(struct intel_crtc *crtc,
1785			       struct vlv_wm_state *wm_state, int level)
1786{
1787	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
1788
1789	for (; level < intel_wm_num_levels(dev_priv); level++) {
1790		enum plane_id plane_id;
1791
1792		for_each_plane_id_on_crtc(crtc, plane_id)
1793			wm_state->wm[level].plane[plane_id] = USHRT_MAX;
1794
1795		wm_state->sr[level].cursor = USHRT_MAX;
1796		wm_state->sr[level].plane = USHRT_MAX;
1797	}
1798}
1799
1800static u16 vlv_invert_wm_value(u16 wm, u16 fifo_size)
1801{
1802	if (wm > fifo_size)
1803		return USHRT_MAX;
1804	else
1805		return fifo_size - wm;
1806}
1807
1808/*
1809 * Starting from 'level' set all higher
1810 * levels to 'value' in the "raw" watermarks.
1811 */
1812static bool vlv_raw_plane_wm_set(struct intel_crtc_state *crtc_state,
1813				 int level, enum plane_id plane_id, u16 value)
1814{
1815	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
1816	int num_levels = intel_wm_num_levels(dev_priv);
1817	bool dirty = false;
1818
1819	for (; level < num_levels; level++) {
1820		struct g4x_pipe_wm *raw = &crtc_state->wm.vlv.raw[level];
1821
1822		dirty |= raw->plane[plane_id] != value;
1823		raw->plane[plane_id] = value;
1824	}
1825
1826	return dirty;
1827}
1828
1829static bool vlv_raw_plane_wm_compute(struct intel_crtc_state *crtc_state,
1830				     const struct intel_plane_state *plane_state)
1831{
1832	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
1833	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
1834	enum plane_id plane_id = plane->id;
1835	int num_levels = intel_wm_num_levels(to_i915(plane->base.dev));
1836	int level;
1837	bool dirty = false;
1838
1839	if (!intel_wm_plane_visible(crtc_state, plane_state)) {
1840		dirty |= vlv_raw_plane_wm_set(crtc_state, 0, plane_id, 0);
1841		goto out;
1842	}
1843
1844	for (level = 0; level < num_levels; level++) {
1845		struct g4x_pipe_wm *raw = &crtc_state->wm.vlv.raw[level];
1846		int wm = vlv_compute_wm_level(crtc_state, plane_state, level);
1847		int max_wm = plane_id == PLANE_CURSOR ? 63 : 511;
1848
1849		if (wm > max_wm)
1850			break;
1851
1852		dirty |= raw->plane[plane_id] != wm;
1853		raw->plane[plane_id] = wm;
1854	}
1855
1856	/* mark all higher levels as invalid */
1857	dirty |= vlv_raw_plane_wm_set(crtc_state, level, plane_id, USHRT_MAX);
1858
1859out:
1860	if (dirty)
1861		drm_dbg_kms(&dev_priv->drm,
1862			    "%s watermarks: PM2=%d, PM5=%d, DDR DVFS=%d\n",
1863			    plane->base.name,
1864			    crtc_state->wm.vlv.raw[VLV_WM_LEVEL_PM2].plane[plane_id],
1865			    crtc_state->wm.vlv.raw[VLV_WM_LEVEL_PM5].plane[plane_id],
1866			    crtc_state->wm.vlv.raw[VLV_WM_LEVEL_DDR_DVFS].plane[plane_id]);
1867
1868	return dirty;
1869}
1870
1871static bool vlv_raw_plane_wm_is_valid(const struct intel_crtc_state *crtc_state,
1872				      enum plane_id plane_id, int level)
1873{
1874	const struct g4x_pipe_wm *raw =
1875		&crtc_state->wm.vlv.raw[level];
1876	const struct vlv_fifo_state *fifo_state =
1877		&crtc_state->wm.vlv.fifo_state;
1878
1879	return raw->plane[plane_id] <= fifo_state->plane[plane_id];
1880}
1881
1882static bool vlv_raw_crtc_wm_is_valid(const struct intel_crtc_state *crtc_state, int level)
1883{
1884	return vlv_raw_plane_wm_is_valid(crtc_state, PLANE_PRIMARY, level) &&
1885		vlv_raw_plane_wm_is_valid(crtc_state, PLANE_SPRITE0, level) &&
1886		vlv_raw_plane_wm_is_valid(crtc_state, PLANE_SPRITE1, level) &&
1887		vlv_raw_plane_wm_is_valid(crtc_state, PLANE_CURSOR, level);
1888}
1889
1890static int vlv_compute_pipe_wm(struct intel_crtc_state *crtc_state)
1891{
1892	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
1893	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
1894	struct intel_atomic_state *state =
1895		to_intel_atomic_state(crtc_state->uapi.state);
1896	struct vlv_wm_state *wm_state = &crtc_state->wm.vlv.optimal;
1897	const struct vlv_fifo_state *fifo_state =
1898		&crtc_state->wm.vlv.fifo_state;
1899	int num_active_planes = hweight8(crtc_state->active_planes &
1900					 ~BIT(PLANE_CURSOR));
1901	bool needs_modeset = drm_atomic_crtc_needs_modeset(&crtc_state->uapi);
1902	const struct intel_plane_state *old_plane_state;
1903	const struct intel_plane_state *new_plane_state;
1904	struct intel_plane *plane;
1905	enum plane_id plane_id;
1906	int level, ret, i;
1907	unsigned int dirty = 0;
1908
1909	for_each_oldnew_intel_plane_in_state(state, plane,
1910					     old_plane_state,
1911					     new_plane_state, i) {
1912		if (new_plane_state->hw.crtc != &crtc->base &&
1913		    old_plane_state->hw.crtc != &crtc->base)
1914			continue;
1915
1916		if (vlv_raw_plane_wm_compute(crtc_state, new_plane_state))
1917			dirty |= BIT(plane->id);
1918	}
1919
1920	/*
1921	 * DSPARB registers may have been reset due to the
1922	 * power well being turned off. Make sure we restore
1923	 * them to a consistent state even if no primary/sprite
1924	 * planes are initially active.
1925	 */
1926	if (needs_modeset)
1927		crtc_state->fifo_changed = true;
1928
1929	if (!dirty)
1930		return 0;
1931
1932	/* cursor changes don't warrant a FIFO recompute */
1933	if (dirty & ~BIT(PLANE_CURSOR)) {
1934		const struct intel_crtc_state *old_crtc_state =
1935			intel_atomic_get_old_crtc_state(state, crtc);
1936		const struct vlv_fifo_state *old_fifo_state =
1937			&old_crtc_state->wm.vlv.fifo_state;
1938
1939		ret = vlv_compute_fifo(crtc_state);
1940		if (ret)
1941			return ret;
1942
1943		if (needs_modeset ||
1944		    memcmp(old_fifo_state, fifo_state,
1945			   sizeof(*fifo_state)) != 0)
1946			crtc_state->fifo_changed = true;
1947	}
1948
1949	/* initially allow all levels */
1950	wm_state->num_levels = intel_wm_num_levels(dev_priv);
1951	/*
1952	 * Note that enabling cxsr with no primary/sprite planes
1953	 * enabled can wedge the pipe. Hence we only allow cxsr
1954	 * with exactly one enabled primary/sprite plane.
1955	 */
1956	wm_state->cxsr = crtc->pipe != PIPE_C && num_active_planes == 1;
1957
1958	for (level = 0; level < wm_state->num_levels; level++) {
1959		const struct g4x_pipe_wm *raw = &crtc_state->wm.vlv.raw[level];
1960		const int sr_fifo_size = INTEL_NUM_PIPES(dev_priv) * 512 - 1;
1961
1962		if (!vlv_raw_crtc_wm_is_valid(crtc_state, level))
1963			break;
1964
1965		for_each_plane_id_on_crtc(crtc, plane_id) {
1966			wm_state->wm[level].plane[plane_id] =
1967				vlv_invert_wm_value(raw->plane[plane_id],
1968						    fifo_state->plane[plane_id]);
1969		}
1970
1971		wm_state->sr[level].plane =
1972			vlv_invert_wm_value(max3(raw->plane[PLANE_PRIMARY],
1973						 raw->plane[PLANE_SPRITE0],
1974						 raw->plane[PLANE_SPRITE1]),
1975					    sr_fifo_size);
1976
1977		wm_state->sr[level].cursor =
1978			vlv_invert_wm_value(raw->plane[PLANE_CURSOR],
1979					    63);
1980	}
1981
1982	if (level == 0)
1983		return -EINVAL;
1984
1985	/* limit to only levels we can actually handle */
1986	wm_state->num_levels = level;
1987
1988	/* invalidate the higher levels */
1989	vlv_invalidate_wms(crtc, wm_state, level);
1990
1991	return 0;
1992}
1993
1994#define VLV_FIFO(plane, value) \
1995	(((value) << DSPARB_ ## plane ## _SHIFT_VLV) & DSPARB_ ## plane ## _MASK_VLV)
1996
1997static void vlv_atomic_update_fifo(struct intel_atomic_state *state,
1998				   struct intel_crtc *crtc)
1999{
2000	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
2001	struct intel_uncore *uncore = &dev_priv->uncore;
2002	const struct intel_crtc_state *crtc_state =
2003		intel_atomic_get_new_crtc_state(state, crtc);
2004	const struct vlv_fifo_state *fifo_state =
2005		&crtc_state->wm.vlv.fifo_state;
2006	int sprite0_start, sprite1_start, fifo_size;
2007	u32 dsparb, dsparb2, dsparb3;
2008
2009	if (!crtc_state->fifo_changed)
2010		return;
2011
2012	sprite0_start = fifo_state->plane[PLANE_PRIMARY];
2013	sprite1_start = fifo_state->plane[PLANE_SPRITE0] + sprite0_start;
2014	fifo_size = fifo_state->plane[PLANE_SPRITE1] + sprite1_start;
2015
2016	drm_WARN_ON(&dev_priv->drm, fifo_state->plane[PLANE_CURSOR] != 63);
2017	drm_WARN_ON(&dev_priv->drm, fifo_size != 511);
2018
2019	trace_vlv_fifo_size(crtc, sprite0_start, sprite1_start, fifo_size);
2020
2021	/*
2022	 * uncore.lock serves a double purpose here. It allows us to
2023	 * use the less expensive I915_{READ,WRITE}_FW() functions, and
2024	 * it protects the DSPARB registers from getting clobbered by
2025	 * parallel updates from multiple pipes.
2026	 *
2027	 * intel_pipe_update_start() has already disabled interrupts
2028	 * for us, so a plain spin_lock() is sufficient here.
2029	 */
2030	spin_lock(&uncore->lock);
2031
2032	switch (crtc->pipe) {
2033	case PIPE_A:
2034		dsparb = intel_uncore_read_fw(uncore, DSPARB);
2035		dsparb2 = intel_uncore_read_fw(uncore, DSPARB2);
2036
2037		dsparb &= ~(VLV_FIFO(SPRITEA, 0xff) |
2038			    VLV_FIFO(SPRITEB, 0xff));
2039		dsparb |= (VLV_FIFO(SPRITEA, sprite0_start) |
2040			   VLV_FIFO(SPRITEB, sprite1_start));
2041
2042		dsparb2 &= ~(VLV_FIFO(SPRITEA_HI, 0x1) |
2043			     VLV_FIFO(SPRITEB_HI, 0x1));
2044		dsparb2 |= (VLV_FIFO(SPRITEA_HI, sprite0_start >> 8) |
2045			   VLV_FIFO(SPRITEB_HI, sprite1_start >> 8));
2046
2047		intel_uncore_write_fw(uncore, DSPARB, dsparb);
2048		intel_uncore_write_fw(uncore, DSPARB2, dsparb2);
2049		break;
2050	case PIPE_B:
2051		dsparb = intel_uncore_read_fw(uncore, DSPARB);
2052		dsparb2 = intel_uncore_read_fw(uncore, DSPARB2);
2053
2054		dsparb &= ~(VLV_FIFO(SPRITEC, 0xff) |
2055			    VLV_FIFO(SPRITED, 0xff));
2056		dsparb |= (VLV_FIFO(SPRITEC, sprite0_start) |
2057			   VLV_FIFO(SPRITED, sprite1_start));
2058
2059		dsparb2 &= ~(VLV_FIFO(SPRITEC_HI, 0xff) |
2060			     VLV_FIFO(SPRITED_HI, 0xff));
2061		dsparb2 |= (VLV_FIFO(SPRITEC_HI, sprite0_start >> 8) |
2062			   VLV_FIFO(SPRITED_HI, sprite1_start >> 8));
2063
2064		intel_uncore_write_fw(uncore, DSPARB, dsparb);
2065		intel_uncore_write_fw(uncore, DSPARB2, dsparb2);
2066		break;
2067	case PIPE_C:
2068		dsparb3 = intel_uncore_read_fw(uncore, DSPARB3);
2069		dsparb2 = intel_uncore_read_fw(uncore, DSPARB2);
2070
2071		dsparb3 &= ~(VLV_FIFO(SPRITEE, 0xff) |
2072			     VLV_FIFO(SPRITEF, 0xff));
2073		dsparb3 |= (VLV_FIFO(SPRITEE, sprite0_start) |
2074			    VLV_FIFO(SPRITEF, sprite1_start));
2075
2076		dsparb2 &= ~(VLV_FIFO(SPRITEE_HI, 0xff) |
2077			     VLV_FIFO(SPRITEF_HI, 0xff));
2078		dsparb2 |= (VLV_FIFO(SPRITEE_HI, sprite0_start >> 8) |
2079			   VLV_FIFO(SPRITEF_HI, sprite1_start >> 8));
2080
2081		intel_uncore_write_fw(uncore, DSPARB3, dsparb3);
2082		intel_uncore_write_fw(uncore, DSPARB2, dsparb2);
2083		break;
2084	default:
2085		break;
2086	}
2087
2088	intel_uncore_posting_read_fw(uncore, DSPARB);
2089
2090	spin_unlock(&uncore->lock);
2091}
2092
2093#undef VLV_FIFO
2094
2095static int vlv_compute_intermediate_wm(struct intel_crtc_state *new_crtc_state)
2096{
2097	struct intel_crtc *crtc = to_intel_crtc(new_crtc_state->uapi.crtc);
2098	struct vlv_wm_state *intermediate = &new_crtc_state->wm.vlv.intermediate;
2099	const struct vlv_wm_state *optimal = &new_crtc_state->wm.vlv.optimal;
2100	struct intel_atomic_state *intel_state =
2101		to_intel_atomic_state(new_crtc_state->uapi.state);
2102	const struct intel_crtc_state *old_crtc_state =
2103		intel_atomic_get_old_crtc_state(intel_state, crtc);
2104	const struct vlv_wm_state *active = &old_crtc_state->wm.vlv.optimal;
2105	int level;
2106
2107	if (!new_crtc_state->hw.active || drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi)) {
2108		*intermediate = *optimal;
2109
2110		intermediate->cxsr = false;
2111		goto out;
2112	}
2113
2114	intermediate->num_levels = min(optimal->num_levels, active->num_levels);
2115	intermediate->cxsr = optimal->cxsr && active->cxsr &&
2116		!new_crtc_state->disable_cxsr;
2117
2118	for (level = 0; level < intermediate->num_levels; level++) {
2119		enum plane_id plane_id;
2120
2121		for_each_plane_id_on_crtc(crtc, plane_id) {
2122			intermediate->wm[level].plane[plane_id] =
2123				min(optimal->wm[level].plane[plane_id],
2124				    active->wm[level].plane[plane_id]);
2125		}
2126
2127		intermediate->sr[level].plane = min(optimal->sr[level].plane,
2128						    active->sr[level].plane);
2129		intermediate->sr[level].cursor = min(optimal->sr[level].cursor,
2130						     active->sr[level].cursor);
2131	}
2132
2133	vlv_invalidate_wms(crtc, intermediate, level);
2134
2135out:
2136	/*
2137	 * If our intermediate WM are identical to the final WM, then we can
2138	 * omit the post-vblank programming; only update if it's different.
2139	 */
2140	if (memcmp(intermediate, optimal, sizeof(*intermediate)) != 0)
2141		new_crtc_state->wm.need_postvbl_update = true;
2142
2143	return 0;
2144}
2145
2146static void vlv_merge_wm(struct drm_i915_private *dev_priv,
2147			 struct vlv_wm_values *wm)
2148{
2149	struct intel_crtc *crtc;
2150	int num_active_pipes = 0;
2151
2152	wm->level = dev_priv->wm.max_level;
2153	wm->cxsr = true;
2154
2155	for_each_intel_crtc(&dev_priv->drm, crtc) {
2156		const struct vlv_wm_state *wm_state = &crtc->wm.active.vlv;
2157
2158		if (!crtc->active)
2159			continue;
2160
2161		if (!wm_state->cxsr)
2162			wm->cxsr = false;
2163
2164		num_active_pipes++;
2165		wm->level = min_t(int, wm->level, wm_state->num_levels - 1);
2166	}
2167
2168	if (num_active_pipes != 1)
2169		wm->cxsr = false;
2170
2171	if (num_active_pipes > 1)
2172		wm->level = VLV_WM_LEVEL_PM2;
2173
2174	for_each_intel_crtc(&dev_priv->drm, crtc) {
2175		const struct vlv_wm_state *wm_state = &crtc->wm.active.vlv;
2176		enum pipe pipe = crtc->pipe;
2177
2178		wm->pipe[pipe] = wm_state->wm[wm->level];
2179		if (crtc->active && wm->cxsr)
2180			wm->sr = wm_state->sr[wm->level];
2181
2182		wm->ddl[pipe].plane[PLANE_PRIMARY] = DDL_PRECISION_HIGH | 2;
2183		wm->ddl[pipe].plane[PLANE_SPRITE0] = DDL_PRECISION_HIGH | 2;
2184		wm->ddl[pipe].plane[PLANE_SPRITE1] = DDL_PRECISION_HIGH | 2;
2185		wm->ddl[pipe].plane[PLANE_CURSOR] = DDL_PRECISION_HIGH | 2;
2186	}
2187}
2188
2189static void vlv_program_watermarks(struct drm_i915_private *dev_priv)
2190{
2191	struct vlv_wm_values *old_wm = &dev_priv->wm.vlv;
2192	struct vlv_wm_values new_wm = {};
2193
2194	vlv_merge_wm(dev_priv, &new_wm);
2195
2196	if (memcmp(old_wm, &new_wm, sizeof(new_wm)) == 0)
2197		return;
2198
2199	if (is_disabling(old_wm->level, new_wm.level, VLV_WM_LEVEL_DDR_DVFS))
2200		chv_set_memory_dvfs(dev_priv, false);
2201
2202	if (is_disabling(old_wm->level, new_wm.level, VLV_WM_LEVEL_PM5))
2203		chv_set_memory_pm5(dev_priv, false);
2204
2205	if (is_disabling(old_wm->cxsr, new_wm.cxsr, true))
2206		_intel_set_memory_cxsr(dev_priv, false);
2207
2208	vlv_write_wm_values(dev_priv, &new_wm);
2209
2210	if (is_enabling(old_wm->cxsr, new_wm.cxsr, true))
2211		_intel_set_memory_cxsr(dev_priv, true);
2212
2213	if (is_enabling(old_wm->level, new_wm.level, VLV_WM_LEVEL_PM5))
2214		chv_set_memory_pm5(dev_priv, true);
2215
2216	if (is_enabling(old_wm->level, new_wm.level, VLV_WM_LEVEL_DDR_DVFS))
2217		chv_set_memory_dvfs(dev_priv, true);
2218
2219	*old_wm = new_wm;
2220}
2221
2222static void vlv_initial_watermarks(struct intel_atomic_state *state,
2223				   struct intel_crtc *crtc)
2224{
2225	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
2226	const struct intel_crtc_state *crtc_state =
2227		intel_atomic_get_new_crtc_state(state, crtc);
2228
2229	mutex_lock(&dev_priv->wm.wm_mutex);
2230	crtc->wm.active.vlv = crtc_state->wm.vlv.intermediate;
2231	vlv_program_watermarks(dev_priv);
2232	mutex_unlock(&dev_priv->wm.wm_mutex);
2233}
2234
2235static void vlv_optimize_watermarks(struct intel_atomic_state *state,
2236				    struct intel_crtc *crtc)
2237{
2238	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
2239	const struct intel_crtc_state *crtc_state =
2240		intel_atomic_get_new_crtc_state(state, crtc);
2241
2242	if (!crtc_state->wm.need_postvbl_update)
2243		return;
2244
2245	mutex_lock(&dev_priv->wm.wm_mutex);
2246	crtc->wm.active.vlv = crtc_state->wm.vlv.optimal;
2247	vlv_program_watermarks(dev_priv);
2248	mutex_unlock(&dev_priv->wm.wm_mutex);
2249}
2250
2251static void i965_update_wm(struct intel_crtc *unused_crtc)
2252{
2253	struct drm_i915_private *dev_priv = to_i915(unused_crtc->base.dev);
2254	struct intel_crtc *crtc;
2255	int srwm = 1;
2256	int cursor_sr = 16;
2257	bool cxsr_enabled;
2258
2259	/* Calc sr entries for one plane configs */
2260	crtc = single_enabled_crtc(dev_priv);
2261	if (crtc) {
2262		/* self-refresh has much higher latency */
2263		static const int sr_latency_ns = 12000;
2264		const struct drm_display_mode *adjusted_mode =
2265			&crtc->config->hw.adjusted_mode;
2266		const struct drm_framebuffer *fb =
2267			crtc->base.primary->state->fb;
2268		int clock = adjusted_mode->crtc_clock;
2269		int htotal = adjusted_mode->crtc_htotal;
2270		int hdisplay = crtc->config->pipe_src_w;
2271		int cpp = fb->format->cpp[0];
2272		int entries;
2273
2274		entries = intel_wm_method2(clock, htotal,
2275					   hdisplay, cpp, sr_latency_ns / 100);
2276		entries = DIV_ROUND_UP(entries, I915_FIFO_LINE_SIZE);
2277		srwm = I965_FIFO_SIZE - entries;
2278		if (srwm < 0)
2279			srwm = 1;
2280		srwm &= 0x1ff;
2281		drm_dbg_kms(&dev_priv->drm,
2282			    "self-refresh entries: %d, wm: %d\n",
2283			    entries, srwm);
2284
2285		entries = intel_wm_method2(clock, htotal,
2286					   crtc->base.cursor->state->crtc_w, 4,
2287					   sr_latency_ns / 100);
2288		entries = DIV_ROUND_UP(entries,
2289				       i965_cursor_wm_info.cacheline_size) +
2290			i965_cursor_wm_info.guard_size;
2291
2292		cursor_sr = i965_cursor_wm_info.fifo_size - entries;
2293		if (cursor_sr > i965_cursor_wm_info.max_wm)
2294			cursor_sr = i965_cursor_wm_info.max_wm;
2295
2296		drm_dbg_kms(&dev_priv->drm,
2297			    "self-refresh watermark: display plane %d "
2298			    "cursor %d\n", srwm, cursor_sr);
2299
2300		cxsr_enabled = true;
2301	} else {
2302		cxsr_enabled = false;
2303		/* Turn off self refresh if both pipes are enabled */
2304		intel_set_memory_cxsr(dev_priv, false);
2305	}
2306
2307	drm_dbg_kms(&dev_priv->drm,
2308		    "Setting FIFO watermarks - A: 8, B: 8, C: 8, SR %d\n",
2309		    srwm);
2310
2311	/* 965 has limitations... */
2312	I915_WRITE(DSPFW1, FW_WM(srwm, SR) |
2313		   FW_WM(8, CURSORB) |
2314		   FW_WM(8, PLANEB) |
2315		   FW_WM(8, PLANEA));
2316	I915_WRITE(DSPFW2, FW_WM(8, CURSORA) |
2317		   FW_WM(8, PLANEC_OLD));
2318	/* update cursor SR watermark */
2319	I915_WRITE(DSPFW3, FW_WM(cursor_sr, CURSOR_SR));
2320
2321	if (cxsr_enabled)
2322		intel_set_memory_cxsr(dev_priv, true);
2323}
2324
2325#undef FW_WM
2326
2327static void i9xx_update_wm(struct intel_crtc *unused_crtc)
2328{
2329	struct drm_i915_private *dev_priv = to_i915(unused_crtc->base.dev);
2330	const struct intel_watermark_params *wm_info;
2331	u32 fwater_lo;
2332	u32 fwater_hi;
2333	int cwm, srwm = 1;
2334	int fifo_size;
2335	int planea_wm, planeb_wm;
2336	struct intel_crtc *crtc, *enabled = NULL;
2337
2338	if (IS_I945GM(dev_priv))
2339		wm_info = &i945_wm_info;
2340	else if (!IS_GEN(dev_priv, 2))
2341		wm_info = &i915_wm_info;
2342	else
2343		wm_info = &i830_a_wm_info;
2344
2345	fifo_size = dev_priv->display.get_fifo_size(dev_priv, PLANE_A);
2346	crtc = intel_get_crtc_for_plane(dev_priv, PLANE_A);
2347	if (intel_crtc_active(crtc)) {
2348		const struct drm_display_mode *adjusted_mode =
2349			&crtc->config->hw.adjusted_mode;
2350		const struct drm_framebuffer *fb =
2351			crtc->base.primary->state->fb;
2352		int cpp;
2353
2354		if (IS_GEN(dev_priv, 2))
2355			cpp = 4;
2356		else
2357			cpp = fb->format->cpp[0];
2358
2359		planea_wm = intel_calculate_wm(adjusted_mode->crtc_clock,
2360					       wm_info, fifo_size, cpp,
2361					       pessimal_latency_ns);
2362		enabled = crtc;
2363	} else {
2364		planea_wm = fifo_size - wm_info->guard_size;
2365		if (planea_wm > (long)wm_info->max_wm)
2366			planea_wm = wm_info->max_wm;
2367	}
2368
2369	if (IS_GEN(dev_priv, 2))
2370		wm_info = &i830_bc_wm_info;
2371
2372	fifo_size = dev_priv->display.get_fifo_size(dev_priv, PLANE_B);
2373	crtc = intel_get_crtc_for_plane(dev_priv, PLANE_B);
2374	if (intel_crtc_active(crtc)) {
2375		const struct drm_display_mode *adjusted_mode =
2376			&crtc->config->hw.adjusted_mode;
2377		const struct drm_framebuffer *fb =
2378			crtc->base.primary->state->fb;
2379		int cpp;
2380
2381		if (IS_GEN(dev_priv, 2))
2382			cpp = 4;
2383		else
2384			cpp = fb->format->cpp[0];
2385
2386		planeb_wm = intel_calculate_wm(adjusted_mode->crtc_clock,
2387					       wm_info, fifo_size, cpp,
2388					       pessimal_latency_ns);
2389		if (enabled == NULL)
2390			enabled = crtc;
2391		else
2392			enabled = NULL;
2393	} else {
2394		planeb_wm = fifo_size - wm_info->guard_size;
2395		if (planeb_wm > (long)wm_info->max_wm)
2396			planeb_wm = wm_info->max_wm;
2397	}
2398
2399	drm_dbg_kms(&dev_priv->drm,
2400		    "FIFO watermarks - A: %d, B: %d\n", planea_wm, planeb_wm);
2401
2402	if (IS_I915GM(dev_priv) && enabled) {
2403		struct drm_i915_gem_object *obj;
2404
2405		obj = intel_fb_obj(enabled->base.primary->state->fb);
2406
2407		/* self-refresh seems busted with untiled */
2408		if (!i915_gem_object_is_tiled(obj))
2409			enabled = NULL;
2410	}
2411
2412	/*
2413	 * Overlay gets an aggressive default since video jitter is bad.
2414	 */
2415	cwm = 2;
2416
2417	/* Play safe and disable self-refresh before adjusting watermarks. */
2418	intel_set_memory_cxsr(dev_priv, false);
2419
2420	/* Calc sr entries for one plane configs */
2421	if (HAS_FW_BLC(dev_priv) && enabled) {
2422		/* self-refresh has much higher latency */
2423		static const int sr_latency_ns = 6000;
2424		const struct drm_display_mode *adjusted_mode =
2425			&enabled->config->hw.adjusted_mode;
2426		const struct drm_framebuffer *fb =
2427			enabled->base.primary->state->fb;
2428		int clock = adjusted_mode->crtc_clock;
2429		int htotal = adjusted_mode->crtc_htotal;
2430		int hdisplay = enabled->config->pipe_src_w;
2431		int cpp;
2432		int entries;
2433
2434		if (IS_I915GM(dev_priv) || IS_I945GM(dev_priv))
2435			cpp = 4;
2436		else
2437			cpp = fb->format->cpp[0];
2438
2439		entries = intel_wm_method2(clock, htotal, hdisplay, cpp,
2440					   sr_latency_ns / 100);
2441		entries = DIV_ROUND_UP(entries, wm_info->cacheline_size);
2442		drm_dbg_kms(&dev_priv->drm,
2443			    "self-refresh entries: %d\n", entries);
2444		srwm = wm_info->fifo_size - entries;
2445		if (srwm < 0)
2446			srwm = 1;
2447
2448		if (IS_I945G(dev_priv) || IS_I945GM(dev_priv))
2449			I915_WRITE(FW_BLC_SELF,
2450				   FW_BLC_SELF_FIFO_MASK | (srwm & 0xff));
2451		else
2452			I915_WRITE(FW_BLC_SELF, srwm & 0x3f);
2453	}
2454
2455	drm_dbg_kms(&dev_priv->drm,
2456		    "Setting FIFO watermarks - A: %d, B: %d, C: %d, SR %d\n",
2457		     planea_wm, planeb_wm, cwm, srwm);
2458
2459	fwater_lo = ((planeb_wm & 0x3f) << 16) | (planea_wm & 0x3f);
2460	fwater_hi = (cwm & 0x1f);
2461
2462	/* Set request length to 8 cachelines per fetch */
2463	fwater_lo = fwater_lo | (1 << 24) | (1 << 8);
2464	fwater_hi = fwater_hi | (1 << 8);
2465
2466	I915_WRITE(FW_BLC, fwater_lo);
2467	I915_WRITE(FW_BLC2, fwater_hi);
2468
2469	if (enabled)
2470		intel_set_memory_cxsr(dev_priv, true);
2471}
2472
2473static void i845_update_wm(struct intel_crtc *unused_crtc)
2474{
2475	struct drm_i915_private *dev_priv = to_i915(unused_crtc->base.dev);
2476	struct intel_crtc *crtc;
2477	const struct drm_display_mode *adjusted_mode;
2478	u32 fwater_lo;
2479	int planea_wm;
2480
2481	crtc = single_enabled_crtc(dev_priv);
2482	if (crtc == NULL)
2483		return;
2484
2485	adjusted_mode = &crtc->config->hw.adjusted_mode;
2486	planea_wm = intel_calculate_wm(adjusted_mode->crtc_clock,
2487				       &i845_wm_info,
2488				       dev_priv->display.get_fifo_size(dev_priv, PLANE_A),
2489				       4, pessimal_latency_ns);
2490	fwater_lo = I915_READ(FW_BLC) & ~0xfff;
2491	fwater_lo |= (3<<8) | planea_wm;
2492
2493	drm_dbg_kms(&dev_priv->drm,
2494		    "Setting FIFO watermarks - A: %d\n", planea_wm);
2495
2496	I915_WRITE(FW_BLC, fwater_lo);
2497}
2498
2499/* latency must be in 0.1us units. */
2500static unsigned int ilk_wm_method1(unsigned int pixel_rate,
2501				   unsigned int cpp,
2502				   unsigned int latency)
2503{
2504	unsigned int ret;
2505
2506	ret = intel_wm_method1(pixel_rate, cpp, latency);
2507	ret = DIV_ROUND_UP(ret, 64) + 2;
2508
2509	return ret;
2510}
2511
2512/* latency must be in 0.1us units. */
2513static unsigned int ilk_wm_method2(unsigned int pixel_rate,
2514				   unsigned int htotal,
2515				   unsigned int width,
2516				   unsigned int cpp,
2517				   unsigned int latency)
2518{
2519	unsigned int ret;
2520
2521	ret = intel_wm_method2(pixel_rate, htotal,
2522			       width, cpp, latency);
2523	ret = DIV_ROUND_UP(ret, 64) + 2;
2524
2525	return ret;
2526}
2527
2528static u32 ilk_wm_fbc(u32 pri_val, u32 horiz_pixels, u8 cpp)
2529{
2530	/*
2531	 * Neither of these should be possible since this function shouldn't be
2532	 * called if the CRTC is off or the plane is invisible.  But let's be
2533	 * extra paranoid to avoid a potential divide-by-zero if we screw up
2534	 * elsewhere in the driver.
2535	 */
2536	if (WARN_ON(!cpp))
2537		return 0;
2538	if (WARN_ON(!horiz_pixels))
2539		return 0;
2540
2541	return DIV_ROUND_UP(pri_val * 64, horiz_pixels * cpp) + 2;
2542}
2543
2544struct ilk_wm_maximums {
2545	u16 pri;
2546	u16 spr;
2547	u16 cur;
2548	u16 fbc;
2549};
2550
2551/*
2552 * For both WM_PIPE and WM_LP.
2553 * mem_value must be in 0.1us units.
2554 */
2555static u32 ilk_compute_pri_wm(const struct intel_crtc_state *crtc_state,
2556			      const struct intel_plane_state *plane_state,
2557			      u32 mem_value, bool is_lp)
2558{
2559	u32 method1, method2;
2560	int cpp;
2561
2562	if (mem_value == 0)
2563		return U32_MAX;
2564
2565	if (!intel_wm_plane_visible(crtc_state, plane_state))
2566		return 0;
2567
2568	cpp = plane_state->hw.fb->format->cpp[0];
2569
2570	method1 = ilk_wm_method1(crtc_state->pixel_rate, cpp, mem_value);
2571
2572	if (!is_lp)
2573		return method1;
2574
2575	method2 = ilk_wm_method2(crtc_state->pixel_rate,
2576				 crtc_state->hw.adjusted_mode.crtc_htotal,
2577				 drm_rect_width(&plane_state->uapi.dst),
2578				 cpp, mem_value);
2579
2580	return min(method1, method2);
2581}
2582
2583/*
2584 * For both WM_PIPE and WM_LP.
2585 * mem_value must be in 0.1us units.
2586 */
2587static u32 ilk_compute_spr_wm(const struct intel_crtc_state *crtc_state,
2588			      const struct intel_plane_state *plane_state,
2589			      u32 mem_value)
2590{
2591	u32 method1, method2;
2592	int cpp;
2593
2594	if (mem_value == 0)
2595		return U32_MAX;
2596
2597	if (!intel_wm_plane_visible(crtc_state, plane_state))
2598		return 0;
2599
2600	cpp = plane_state->hw.fb->format->cpp[0];
2601
2602	method1 = ilk_wm_method1(crtc_state->pixel_rate, cpp, mem_value);
2603	method2 = ilk_wm_method2(crtc_state->pixel_rate,
2604				 crtc_state->hw.adjusted_mode.crtc_htotal,
2605				 drm_rect_width(&plane_state->uapi.dst),
2606				 cpp, mem_value);
2607	return min(method1, method2);
2608}
2609
2610/*
2611 * For both WM_PIPE and WM_LP.
2612 * mem_value must be in 0.1us units.
2613 */
2614static u32 ilk_compute_cur_wm(const struct intel_crtc_state *crtc_state,
2615			      const struct intel_plane_state *plane_state,
2616			      u32 mem_value)
2617{
2618	int cpp;
2619
2620	if (mem_value == 0)
2621		return U32_MAX;
2622
2623	if (!intel_wm_plane_visible(crtc_state, plane_state))
2624		return 0;
2625
2626	cpp = plane_state->hw.fb->format->cpp[0];
2627
2628	return ilk_wm_method2(crtc_state->pixel_rate,
2629			      crtc_state->hw.adjusted_mode.crtc_htotal,
2630			      drm_rect_width(&plane_state->uapi.dst),
2631			      cpp, mem_value);
2632}
2633
2634/* Only for WM_LP. */
2635static u32 ilk_compute_fbc_wm(const struct intel_crtc_state *crtc_state,
2636			      const struct intel_plane_state *plane_state,
2637			      u32 pri_val)
2638{
2639	int cpp;
2640
2641	if (!intel_wm_plane_visible(crtc_state, plane_state))
2642		return 0;
2643
2644	cpp = plane_state->hw.fb->format->cpp[0];
2645
2646	return ilk_wm_fbc(pri_val, drm_rect_width(&plane_state->uapi.dst),
2647			  cpp);
2648}
2649
2650static unsigned int
2651ilk_display_fifo_size(const struct drm_i915_private *dev_priv)
2652{
2653	if (INTEL_GEN(dev_priv) >= 8)
2654		return 3072;
2655	else if (INTEL_GEN(dev_priv) >= 7)
2656		return 768;
2657	else
2658		return 512;
2659}
2660
2661static unsigned int
2662ilk_plane_wm_reg_max(const struct drm_i915_private *dev_priv,
2663		     int level, bool is_sprite)
2664{
2665	if (INTEL_GEN(dev_priv) >= 8)
2666		/* BDW primary/sprite plane watermarks */
2667		return level == 0 ? 255 : 2047;
2668	else if (INTEL_GEN(dev_priv) >= 7)
2669		/* IVB/HSW primary/sprite plane watermarks */
2670		return level == 0 ? 127 : 1023;
2671	else if (!is_sprite)
2672		/* ILK/SNB primary plane watermarks */
2673		return level == 0 ? 127 : 511;
2674	else
2675		/* ILK/SNB sprite plane watermarks */
2676		return level == 0 ? 63 : 255;
2677}
2678
2679static unsigned int
2680ilk_cursor_wm_reg_max(const struct drm_i915_private *dev_priv, int level)
2681{
2682	if (INTEL_GEN(dev_priv) >= 7)
2683		return level == 0 ? 63 : 255;
2684	else
2685		return level == 0 ? 31 : 63;
2686}
2687
2688static unsigned int ilk_fbc_wm_reg_max(const struct drm_i915_private *dev_priv)
2689{
2690	if (INTEL_GEN(dev_priv) >= 8)
2691		return 31;
2692	else
2693		return 15;
2694}
2695
2696/* Calculate the maximum primary/sprite plane watermark */
2697static unsigned int ilk_plane_wm_max(const struct drm_i915_private *dev_priv,
2698				     int level,
2699				     const struct intel_wm_config *config,
2700				     enum intel_ddb_partitioning ddb_partitioning,
2701				     bool is_sprite)
2702{
2703	unsigned int fifo_size = ilk_display_fifo_size(dev_priv);
2704
2705	/* if sprites aren't enabled, sprites get nothing */
2706	if (is_sprite && !config->sprites_enabled)
2707		return 0;
2708
2709	/* HSW allows LP1+ watermarks even with multiple pipes */
2710	if (level == 0 || config->num_pipes_active > 1) {
2711		fifo_size /= INTEL_NUM_PIPES(dev_priv);
2712
2713		/*
2714		 * For some reason the non self refresh
2715		 * FIFO size is only half of the self
2716		 * refresh FIFO size on ILK/SNB.
2717		 */
2718		if (INTEL_GEN(dev_priv) <= 6)
2719			fifo_size /= 2;
2720	}
2721
2722	if (config->sprites_enabled) {
2723		/* level 0 is always calculated with 1:1 split */
2724		if (level > 0 && ddb_partitioning == INTEL_DDB_PART_5_6) {
2725			if (is_sprite)
2726				fifo_size *= 5;
2727			fifo_size /= 6;
2728		} else {
2729			fifo_size /= 2;
2730		}
2731	}
2732
2733	/* clamp to max that the registers can hold */
2734	return min(fifo_size, ilk_plane_wm_reg_max(dev_priv, level, is_sprite));
2735}
2736
2737/* Calculate the maximum cursor plane watermark */
2738static unsigned int ilk_cursor_wm_max(const struct drm_i915_private *dev_priv,
2739				      int level,
2740				      const struct intel_wm_config *config)
2741{
2742	/* HSW LP1+ watermarks w/ multiple pipes */
2743	if (level > 0 && config->num_pipes_active > 1)
2744		return 64;
2745
2746	/* otherwise just report max that registers can hold */
2747	return ilk_cursor_wm_reg_max(dev_priv, level);
2748}
2749
2750static void ilk_compute_wm_maximums(const struct drm_i915_private *dev_priv,
2751				    int level,
2752				    const struct intel_wm_config *config,
2753				    enum intel_ddb_partitioning ddb_partitioning,
2754				    struct ilk_wm_maximums *max)
2755{
2756	max->pri = ilk_plane_wm_max(dev_priv, level, config, ddb_partitioning, false);
2757	max->spr = ilk_plane_wm_max(dev_priv, level, config, ddb_partitioning, true);
2758	max->cur = ilk_cursor_wm_max(dev_priv, level, config);
2759	max->fbc = ilk_fbc_wm_reg_max(dev_priv);
2760}
2761
2762static void ilk_compute_wm_reg_maximums(const struct drm_i915_private *dev_priv,
2763					int level,
2764					struct ilk_wm_maximums *max)
2765{
2766	max->pri = ilk_plane_wm_reg_max(dev_priv, level, false);
2767	max->spr = ilk_plane_wm_reg_max(dev_priv, level, true);
2768	max->cur = ilk_cursor_wm_reg_max(dev_priv, level);
2769	max->fbc = ilk_fbc_wm_reg_max(dev_priv);
2770}
2771
2772static bool ilk_validate_wm_level(int level,
2773				  const struct ilk_wm_maximums *max,
2774				  struct intel_wm_level *result)
2775{
2776	bool ret;
2777
2778	/* already determined to be invalid? */
2779	if (!result->enable)
2780		return false;
2781
2782	result->enable = result->pri_val <= max->pri &&
2783			 result->spr_val <= max->spr &&
2784			 result->cur_val <= max->cur;
2785
2786	ret = result->enable;
2787
2788	/*
2789	 * HACK until we can pre-compute everything,
2790	 * and thus fail gracefully if LP0 watermarks
2791	 * are exceeded...
2792	 */
2793	if (level == 0 && !result->enable) {
2794		if (result->pri_val > max->pri)
2795			DRM_DEBUG_KMS("Primary WM%d too large %u (max %u)\n",
2796				      level, result->pri_val, max->pri);
2797		if (result->spr_val > max->spr)
2798			DRM_DEBUG_KMS("Sprite WM%d too large %u (max %u)\n",
2799				      level, result->spr_val, max->spr);
2800		if (result->cur_val > max->cur)
2801			DRM_DEBUG_KMS("Cursor WM%d too large %u (max %u)\n",
2802				      level, result->cur_val, max->cur);
2803
2804		result->pri_val = min_t(u32, result->pri_val, max->pri);
2805		result->spr_val = min_t(u32, result->spr_val, max->spr);
2806		result->cur_val = min_t(u32, result->cur_val, max->cur);
2807		result->enable = true;
2808	}
2809
2810	return ret;
2811}
2812
2813static void ilk_compute_wm_level(const struct drm_i915_private *dev_priv,
2814				 const struct intel_crtc *crtc,
2815				 int level,
2816				 struct intel_crtc_state *crtc_state,
2817				 const struct intel_plane_state *pristate,
2818				 const struct intel_plane_state *sprstate,
2819				 const struct intel_plane_state *curstate,
2820				 struct intel_wm_level *result)
2821{
2822	u16 pri_latency = dev_priv->wm.pri_latency[level];
2823	u16 spr_latency = dev_priv->wm.spr_latency[level];
2824	u16 cur_latency = dev_priv->wm.cur_latency[level];
2825
2826	/* WM1+ latency values stored in 0.5us units */
2827	if (level > 0) {
2828		pri_latency *= 5;
2829		spr_latency *= 5;
2830		cur_latency *= 5;
2831	}
2832
2833	if (pristate) {
2834		result->pri_val = ilk_compute_pri_wm(crtc_state, pristate,
2835						     pri_latency, level);
2836		result->fbc_val = ilk_compute_fbc_wm(crtc_state, pristate, result->pri_val);
2837	}
2838
2839	if (sprstate)
2840		result->spr_val = ilk_compute_spr_wm(crtc_state, sprstate, spr_latency);
2841
2842	if (curstate)
2843		result->cur_val = ilk_compute_cur_wm(crtc_state, curstate, cur_latency);
2844
2845	result->enable = true;
2846}
2847
2848static void intel_read_wm_latency(struct drm_i915_private *dev_priv,
2849				  u16 wm[])
2850{
2851	struct intel_uncore *uncore = &dev_priv->uncore;
2852
2853	if (INTEL_GEN(dev_priv) >= 9) {
2854		u32 val;
2855		int ret, i;
2856		int level, max_level = ilk_wm_max_level(dev_priv);
2857
2858		/* read the first set of memory latencies[0:3] */
2859		val = 0; /* data0 to be programmed to 0 for first set */
2860		ret = sandybridge_pcode_read(dev_priv,
2861					     GEN9_PCODE_READ_MEM_LATENCY,
2862					     &val, NULL);
2863
2864		if (ret) {
2865			drm_err(&dev_priv->drm,
2866				"SKL Mailbox read error = %d\n", ret);
2867			return;
2868		}
2869
2870		wm[0] = val & GEN9_MEM_LATENCY_LEVEL_MASK;
2871		wm[1] = (val >> GEN9_MEM_LATENCY_LEVEL_1_5_SHIFT) &
2872				GEN9_MEM_LATENCY_LEVEL_MASK;
2873		wm[2] = (val >> GEN9_MEM_LATENCY_LEVEL_2_6_SHIFT) &
2874				GEN9_MEM_LATENCY_LEVEL_MASK;
2875		wm[3] = (val >> GEN9_MEM_LATENCY_LEVEL_3_7_SHIFT) &
2876				GEN9_MEM_LATENCY_LEVEL_MASK;
2877
2878		/* read the second set of memory latencies[4:7] */
2879		val = 1; /* data0 to be programmed to 1 for second set */
2880		ret = sandybridge_pcode_read(dev_priv,
2881					     GEN9_PCODE_READ_MEM_LATENCY,
2882					     &val, NULL);
2883		if (ret) {
2884			drm_err(&dev_priv->drm,
2885				"SKL Mailbox read error = %d\n", ret);
2886			return;
2887		}
2888
2889		wm[4] = val & GEN9_MEM_LATENCY_LEVEL_MASK;
2890		wm[5] = (val >> GEN9_MEM_LATENCY_LEVEL_1_5_SHIFT) &
2891				GEN9_MEM_LATENCY_LEVEL_MASK;
2892		wm[6] = (val >> GEN9_MEM_LATENCY_LEVEL_2_6_SHIFT) &
2893				GEN9_MEM_LATENCY_LEVEL_MASK;
2894		wm[7] = (val >> GEN9_MEM_LATENCY_LEVEL_3_7_SHIFT) &
2895				GEN9_MEM_LATENCY_LEVEL_MASK;
2896
2897		/*
2898		 * If a level n (n > 1) has a 0us latency, all levels m (m >= n)
2899		 * need to be disabled. We make sure to sanitize the values out
2900		 * of the punit to satisfy this requirement.
2901		 */
2902		for (level = 1; level <= max_level; level++) {
2903			if (wm[level] == 0) {
2904				for (i = level + 1; i <= max_level; i++)
2905					wm[i] = 0;
2906				break;
2907			}
2908		}
2909
2910		/*
2911		 * WaWmMemoryReadLatency:skl+,glk
2912		 *
2913		 * punit doesn't take into account the read latency so we need
2914		 * to add 2us to the various latency levels we retrieve from the
2915		 * punit when level 0 response data us 0us.
2916		 */
2917		if (wm[0] == 0) {
2918			wm[0] += 2;
2919			for (level = 1; level <= max_level; level++) {
2920				if (wm[level] == 0)
2921					break;
2922				wm[level] += 2;
2923			}
2924		}
2925
2926		/*
2927		 * WA Level-0 adjustment for 16GB DIMMs: SKL+
2928		 * If we could not get dimm info enable this WA to prevent from
2929		 * any underrun. If not able to get Dimm info assume 16GB dimm
2930		 * to avoid any underrun.
2931		 */
2932		if (dev_priv->dram_info.is_16gb_dimm)
2933			wm[0] += 1;
2934
2935	} else if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv)) {
2936		u64 sskpd = intel_uncore_read64(uncore, MCH_SSKPD);
2937
2938		wm[0] = (sskpd >> 56) & 0xFF;
2939		if (wm[0] == 0)
2940			wm[0] = sskpd & 0xF;
2941		wm[1] = (sskpd >> 4) & 0xFF;
2942		wm[2] = (sskpd >> 12) & 0xFF;
2943		wm[3] = (sskpd >> 20) & 0x1FF;
2944		wm[4] = (sskpd >> 32) & 0x1FF;
2945	} else if (INTEL_GEN(dev_priv) >= 6) {
2946		u32 sskpd = intel_uncore_read(uncore, MCH_SSKPD);
2947
2948		wm[0] = (sskpd >> SSKPD_WM0_SHIFT) & SSKPD_WM_MASK;
2949		wm[1] = (sskpd >> SSKPD_WM1_SHIFT) & SSKPD_WM_MASK;
2950		wm[2] = (sskpd >> SSKPD_WM2_SHIFT) & SSKPD_WM_MASK;
2951		wm[3] = (sskpd >> SSKPD_WM3_SHIFT) & SSKPD_WM_MASK;
2952	} else if (INTEL_GEN(dev_priv) >= 5) {
2953		u32 mltr = intel_uncore_read(uncore, MLTR_ILK);
2954
2955		/* ILK primary LP0 latency is 700 ns */
2956		wm[0] = 7;
2957		wm[1] = (mltr >> MLTR_WM1_SHIFT) & ILK_SRLT_MASK;
2958		wm[2] = (mltr >> MLTR_WM2_SHIFT) & ILK_SRLT_MASK;
2959	} else {
2960		MISSING_CASE(INTEL_DEVID(dev_priv));
2961	}
2962}
2963
2964static void intel_fixup_spr_wm_latency(struct drm_i915_private *dev_priv,
2965				       u16 wm[5])
2966{
2967	/* ILK sprite LP0 latency is 1300 ns */
2968	if (IS_GEN(dev_priv, 5))
2969		wm[0] = 13;
2970}
2971
2972static void intel_fixup_cur_wm_latency(struct drm_i915_private *dev_priv,
2973				       u16 wm[5])
2974{
2975	/* ILK cursor LP0 latency is 1300 ns */
2976	if (IS_GEN(dev_priv, 5))
2977		wm[0] = 13;
2978}
2979
2980int ilk_wm_max_level(const struct drm_i915_private *dev_priv)
2981{
2982	/* how many WM levels are we expecting */
2983	if (INTEL_GEN(dev_priv) >= 9)
2984		return 7;
2985	else if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv))
2986		return 4;
2987	else if (INTEL_GEN(dev_priv) >= 6)
2988		return 3;
2989	else
2990		return 2;
2991}
2992
2993static void intel_print_wm_latency(struct drm_i915_private *dev_priv,
2994				   const char *name,
2995				   const u16 wm[])
2996{
2997	int level, max_level = ilk_wm_max_level(dev_priv);
2998
2999	for (level = 0; level <= max_level; level++) {
3000		unsigned int latency = wm[level];
3001
3002		if (latency == 0) {
3003			drm_dbg_kms(&dev_priv->drm,
3004				    "%s WM%d latency not provided\n",
3005				    name, level);
3006			continue;
3007		}
3008
3009		/*
3010		 * - latencies are in us on gen9.
3011		 * - before then, WM1+ latency values are in 0.5us units
3012		 */
3013		if (INTEL_GEN(dev_priv) >= 9)
3014			latency *= 10;
3015		else if (level > 0)
3016			latency *= 5;
3017
3018		drm_dbg_kms(&dev_priv->drm,
3019			    "%s WM%d latency %u (%u.%u usec)\n", name, level,
3020			    wm[level], latency / 10, latency % 10);
3021	}
3022}
3023
3024static bool ilk_increase_wm_latency(struct drm_i915_private *dev_priv,
3025				    u16 wm[5], u16 min)
3026{
3027	int level, max_level = ilk_wm_max_level(dev_priv);
3028
3029	if (wm[0] >= min)
3030		return false;
3031
3032	wm[0] = max(wm[0], min);
3033	for (level = 1; level <= max_level; level++)
3034		wm[level] = max_t(u16, wm[level], DIV_ROUND_UP(min, 5));
3035
3036	return true;
3037}
3038
3039static void snb_wm_latency_quirk(struct drm_i915_private *dev_priv)
3040{
3041	bool changed;
3042
3043	/*
3044	 * The BIOS provided WM memory latency values are often
3045	 * inadequate for high resolution displays. Adjust them.
3046	 */
3047	changed = ilk_increase_wm_latency(dev_priv, dev_priv->wm.pri_latency, 12);
3048	changed |= ilk_increase_wm_latency(dev_priv, dev_priv->wm.spr_latency, 12);
3049	changed |= ilk_increase_wm_latency(dev_priv, dev_priv->wm.cur_latency, 12);
3050
3051	if (!changed)
3052		return;
3053
3054	drm_dbg_kms(&dev_priv->drm,
3055		    "WM latency values increased to avoid potential underruns\n");
3056	intel_print_wm_latency(dev_priv, "Primary", dev_priv->wm.pri_latency);
3057	intel_print_wm_latency(dev_priv, "Sprite", dev_priv->wm.spr_latency);
3058	intel_print_wm_latency(dev_priv, "Cursor", dev_priv->wm.cur_latency);
3059}
3060
3061static void snb_wm_lp3_irq_quirk(struct drm_i915_private *dev_priv)
3062{
3063	/*
3064	 * On some SNB machines (Thinkpad X220 Tablet at least)
3065	 * LP3 usage can cause vblank interrupts to be lost.
3066	 * The DEIIR bit will go high but it looks like the CPU
3067	 * never gets interrupted.
3068	 *
3069	 * It's not clear whether other interrupt source could
3070	 * be affected or if this is somehow limited to vblank
3071	 * interrupts only. To play it safe we disable LP3
3072	 * watermarks entirely.
3073	 */
3074	if (dev_priv->wm.pri_latency[3] == 0 &&
3075	    dev_priv->wm.spr_latency[3] == 0 &&
3076	    dev_priv->wm.cur_latency[3] == 0)
3077		return;
3078
3079	dev_priv->wm.pri_latency[3] = 0;
3080	dev_priv->wm.spr_latency[3] = 0;
3081	dev_priv->wm.cur_latency[3] = 0;
3082
3083	drm_dbg_kms(&dev_priv->drm,
3084		    "LP3 watermarks disabled due to potential for lost interrupts\n");
3085	intel_print_wm_latency(dev_priv, "Primary", dev_priv->wm.pri_latency);
3086	intel_print_wm_latency(dev_priv, "Sprite", dev_priv->wm.spr_latency);
3087	intel_print_wm_latency(dev_priv, "Cursor", dev_priv->wm.cur_latency);
3088}
3089
3090static void ilk_setup_wm_latency(struct drm_i915_private *dev_priv)
3091{
3092	intel_read_wm_latency(dev_priv, dev_priv->wm.pri_latency);
3093
3094	memcpy(dev_priv->wm.spr_latency, dev_priv->wm.pri_latency,
3095	       sizeof(dev_priv->wm.pri_latency));
3096	memcpy(dev_priv->wm.cur_latency, dev_priv->wm.pri_latency,
3097	       sizeof(dev_priv->wm.pri_latency));
3098
3099	intel_fixup_spr_wm_latency(dev_priv, dev_priv->wm.spr_latency);
3100	intel_fixup_cur_wm_latency(dev_priv, dev_priv->wm.cur_latency);
3101
3102	intel_print_wm_latency(dev_priv, "Primary", dev_priv->wm.pri_latency);
3103	intel_print_wm_latency(dev_priv, "Sprite", dev_priv->wm.spr_latency);
3104	intel_print_wm_latency(dev_priv, "Cursor", dev_priv->wm.cur_latency);
3105
3106	if (IS_GEN(dev_priv, 6)) {
3107		snb_wm_latency_quirk(dev_priv);
3108		snb_wm_lp3_irq_quirk(dev_priv);
3109	}
3110}
3111
3112static void skl_setup_wm_latency(struct drm_i915_private *dev_priv)
3113{
3114	intel_read_wm_latency(dev_priv, dev_priv->wm.skl_latency);
3115	intel_print_wm_latency(dev_priv, "Gen9 Plane", dev_priv->wm.skl_latency);
3116}
3117
3118static bool ilk_validate_pipe_wm(const struct drm_i915_private *dev_priv,
3119				 struct intel_pipe_wm *pipe_wm)
3120{
3121	/* LP0 watermark maximums depend on this pipe alone */
3122	const struct intel_wm_config config = {
3123		.num_pipes_active = 1,
3124		.sprites_enabled = pipe_wm->sprites_enabled,
3125		.sprites_scaled = pipe_wm->sprites_scaled,
3126	};
3127	struct ilk_wm_maximums max;
3128
3129	/* LP0 watermarks always use 1/2 DDB partitioning */
3130	ilk_compute_wm_maximums(dev_priv, 0, &config, INTEL_DDB_PART_1_2, &max);
3131
3132	/* At least LP0 must be valid */
3133	if (!ilk_validate_wm_level(0, &max, &pipe_wm->wm[0])) {
3134		drm_dbg_kms(&dev_priv->drm, "LP0 watermark invalid\n");
3135		return false;
3136	}
3137
3138	return true;
3139}
3140
3141/* Compute new watermarks for the pipe */
3142static int ilk_compute_pipe_wm(struct intel_crtc_state *crtc_state)
3143{
3144	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
3145	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
3146	struct intel_pipe_wm *pipe_wm;
3147	struct intel_plane *plane;
3148	const struct intel_plane_state *plane_state;
3149	const struct intel_plane_state *pristate = NULL;
3150	const struct intel_plane_state *sprstate = NULL;
3151	const struct intel_plane_state *curstate = NULL;
3152	int level, max_level = ilk_wm_max_level(dev_priv), usable_level;
3153	struct ilk_wm_maximums max;
3154
3155	pipe_wm = &crtc_state->wm.ilk.optimal;
3156
3157	intel_atomic_crtc_state_for_each_plane_state(plane, plane_state, crtc_state) {
3158		if (plane->base.type == DRM_PLANE_TYPE_PRIMARY)
3159			pristate = plane_state;
3160		else if (plane->base.type == DRM_PLANE_TYPE_OVERLAY)
3161			sprstate = plane_state;
3162		else if (plane->base.type == DRM_PLANE_TYPE_CURSOR)
3163			curstate = plane_state;
3164	}
3165
3166	pipe_wm->pipe_enabled = crtc_state->hw.active;
3167	if (sprstate) {
3168		pipe_wm->sprites_enabled = sprstate->uapi.visible;
3169		pipe_wm->sprites_scaled = sprstate->uapi.visible &&
3170			(drm_rect_width(&sprstate->uapi.dst) != drm_rect_width(&sprstate->uapi.src) >> 16 ||
3171			 drm_rect_height(&sprstate->uapi.dst) != drm_rect_height(&sprstate->uapi.src) >> 16);
3172	}
3173
3174	usable_level = max_level;
3175
3176	/* ILK/SNB: LP2+ watermarks only w/o sprites */
3177	if (INTEL_GEN(dev_priv) <= 6 && pipe_wm->sprites_enabled)
3178		usable_level = 1;
3179
3180	/* ILK/SNB/IVB: LP1+ watermarks only w/o scaling */
3181	if (pipe_wm->sprites_scaled)
3182		usable_level = 0;
3183
3184	memset(&pipe_wm->wm, 0, sizeof(pipe_wm->wm));
3185	ilk_compute_wm_level(dev_priv, crtc, 0, crtc_state,
3186			     pristate, sprstate, curstate, &pipe_wm->wm[0]);
3187
3188	if (!ilk_validate_pipe_wm(dev_priv, pipe_wm))
3189		return -EINVAL;
3190
3191	ilk_compute_wm_reg_maximums(dev_priv, 1, &max);
3192
3193	for (level = 1; level <= usable_level; level++) {
3194		struct intel_wm_level *wm = &pipe_wm->wm[level];
3195
3196		ilk_compute_wm_level(dev_priv, crtc, level, crtc_state,
3197				     pristate, sprstate, curstate, wm);
3198
3199		/*
3200		 * Disable any watermark level that exceeds the
3201		 * register maximums since such watermarks are
3202		 * always invalid.
3203		 */
3204		if (!ilk_validate_wm_level(level, &max, wm)) {
3205			memset(wm, 0, sizeof(*wm));
3206			break;
3207		}
3208	}
3209
3210	return 0;
3211}
3212
3213/*
3214 * Build a set of 'intermediate' watermark values that satisfy both the old
3215 * state and the new state.  These can be programmed to the hardware
3216 * immediately.
3217 */
3218static int ilk_compute_intermediate_wm(struct intel_crtc_state *newstate)
3219{
3220	struct intel_crtc *intel_crtc = to_intel_crtc(newstate->uapi.crtc);
3221	struct drm_i915_private *dev_priv = to_i915(intel_crtc->base.dev);
3222	struct intel_pipe_wm *a = &newstate->wm.ilk.intermediate;
3223	struct intel_atomic_state *intel_state =
3224		to_intel_atomic_state(newstate->uapi.state);
3225	const struct intel_crtc_state *oldstate =
3226		intel_atomic_get_old_crtc_state(intel_state, intel_crtc);
3227	const struct intel_pipe_wm *b = &oldstate->wm.ilk.optimal;
3228	int level, max_level = ilk_wm_max_level(dev_priv);
3229
3230	/*
3231	 * Start with the final, target watermarks, then combine with the
3232	 * currently active watermarks to get values that are safe both before
3233	 * and after the vblank.
3234	 */
3235	*a = newstate->wm.ilk.optimal;
3236	if (!newstate->hw.active || drm_atomic_crtc_needs_modeset(&newstate->uapi) ||
3237	    intel_state->skip_intermediate_wm)
3238		return 0;
3239
3240	a->pipe_enabled |= b->pipe_enabled;
3241	a->sprites_enabled |= b->sprites_enabled;
3242	a->sprites_scaled |= b->sprites_scaled;
3243
3244	for (level = 0; level <= max_level; level++) {
3245		struct intel_wm_level *a_wm = &a->wm[level];
3246		const struct intel_wm_level *b_wm = &b->wm[level];
3247
3248		a_wm->enable &= b_wm->enable;
3249		a_wm->pri_val = max(a_wm->pri_val, b_wm->pri_val);
3250		a_wm->spr_val = max(a_wm->spr_val, b_wm->spr_val);
3251		a_wm->cur_val = max(a_wm->cur_val, b_wm->cur_val);
3252		a_wm->fbc_val = max(a_wm->fbc_val, b_wm->fbc_val);
3253	}
3254
3255	/*
3256	 * We need to make sure that these merged watermark values are
3257	 * actually a valid configuration themselves.  If they're not,
3258	 * there's no safe way to transition from the old state to
3259	 * the new state, so we need to fail the atomic transaction.
3260	 */
3261	if (!ilk_validate_pipe_wm(dev_priv, a))
3262		return -EINVAL;
3263
3264	/*
3265	 * If our intermediate WM are identical to the final WM, then we can
3266	 * omit the post-vblank programming; only update if it's different.
3267	 */
3268	if (memcmp(a, &newstate->wm.ilk.optimal, sizeof(*a)) != 0)
3269		newstate->wm.need_postvbl_update = true;
3270
3271	return 0;
3272}
3273
3274/*
3275 * Merge the watermarks from all active pipes for a specific level.
3276 */
3277static void ilk_merge_wm_level(struct drm_i915_private *dev_priv,
3278			       int level,
3279			       struct intel_wm_level *ret_wm)
3280{
3281	const struct intel_crtc *intel_crtc;
3282
3283	ret_wm->enable = true;
3284
3285	for_each_intel_crtc(&dev_priv->drm, intel_crtc) {
3286		const struct intel_pipe_wm *active = &intel_crtc->wm.active.ilk;
3287		const struct intel_wm_level *wm = &active->wm[level];
3288
3289		if (!active->pipe_enabled)
3290			continue;
3291
3292		/*
3293		 * The watermark values may have been used in the past,
3294		 * so we must maintain them in the registers for some
3295		 * time even if the level is now disabled.
3296		 */
3297		if (!wm->enable)
3298			ret_wm->enable = false;
3299
3300		ret_wm->pri_val = max(ret_wm->pri_val, wm->pri_val);
3301		ret_wm->spr_val = max(ret_wm->spr_val, wm->spr_val);
3302		ret_wm->cur_val = max(ret_wm->cur_val, wm->cur_val);
3303		ret_wm->fbc_val = max(ret_wm->fbc_val, wm->fbc_val);
3304	}
3305}
3306
3307/*
3308 * Merge all low power watermarks for all active pipes.
3309 */
3310static void ilk_wm_merge(struct drm_i915_private *dev_priv,
3311			 const struct intel_wm_config *config,
3312			 const struct ilk_wm_maximums *max,
3313			 struct intel_pipe_wm *merged)
3314{
3315	int level, max_level = ilk_wm_max_level(dev_priv);
3316	int last_enabled_level = max_level;
3317
3318	/* ILK/SNB/IVB: LP1+ watermarks only w/ single pipe */
3319	if ((INTEL_GEN(dev_priv) <= 6 || IS_IVYBRIDGE(dev_priv)) &&
3320	    config->num_pipes_active > 1)
3321		last_enabled_level = 0;
3322
3323	/* ILK: FBC WM must be disabled always */
3324	merged->fbc_wm_enabled = INTEL_GEN(dev_priv) >= 6;
3325
3326	/* merge each WM1+ level */
3327	for (level = 1; level <= max_level; level++) {
3328		struct intel_wm_level *wm = &merged->wm[level];
3329
3330		ilk_merge_wm_level(dev_priv, level, wm);
3331
3332		if (level > last_enabled_level)
3333			wm->enable = false;
3334		else if (!ilk_validate_wm_level(level, max, wm))
3335			/* make sure all following levels get disabled */
3336			last_enabled_level = level - 1;
3337
3338		/*
3339		 * The spec says it is preferred to disable
3340		 * FBC WMs instead of disabling a WM level.
3341		 */
3342		if (wm->fbc_val > max->fbc) {
3343			if (wm->enable)
3344				merged->fbc_wm_enabled = false;
3345			wm->fbc_val = 0;
3346		}
3347	}
3348
3349	/* ILK: LP2+ must be disabled when FBC WM is disabled but FBC enabled */
3350	/*
3351	 * FIXME this is racy. FBC might get enabled later.
3352	 * What we should check here is whether FBC can be
3353	 * enabled sometime later.
3354	 */
3355	if (IS_GEN(dev_priv, 5) && !merged->fbc_wm_enabled &&
3356	    intel_fbc_is_active(dev_priv)) {
3357		for (level = 2; level <= max_level; level++) {
3358			struct intel_wm_level *wm = &merged->wm[level];
3359
3360			wm->enable = false;
3361		}
3362	}
3363}
3364
3365static int ilk_wm_lp_to_level(int wm_lp, const struct intel_pipe_wm *pipe_wm)
3366{
3367	/* LP1,LP2,LP3 levels are either 1,2,3 or 1,3,4 */
3368	return wm_lp + (wm_lp >= 2 && pipe_wm->wm[4].enable);
3369}
3370
3371/* The value we need to program into the WM_LPx latency field */
3372static unsigned int ilk_wm_lp_latency(struct drm_i915_private *dev_priv,
3373				      int level)
3374{
3375	if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv))
3376		return 2 * level;
3377	else
3378		return dev_priv->wm.pri_latency[level];
3379}
3380
3381static void ilk_compute_wm_results(struct drm_i915_private *dev_priv,
3382				   const struct intel_pipe_wm *merged,
3383				   enum intel_ddb_partitioning partitioning,
3384				   struct ilk_wm_values *results)
3385{
3386	struct intel_crtc *intel_crtc;
3387	int level, wm_lp;
3388
3389	results->enable_fbc_wm = merged->fbc_wm_enabled;
3390	results->partitioning = partitioning;
3391
3392	/* LP1+ register values */
3393	for (wm_lp = 1; wm_lp <= 3; wm_lp++) {
3394		const struct intel_wm_level *r;
3395
3396		level = ilk_wm_lp_to_level(wm_lp, merged);
3397
3398		r = &merged->wm[level];
3399
3400		/*
3401		 * Maintain the watermark values even if the level is
3402		 * disabled. Doing otherwise could cause underruns.
3403		 */
3404		results->wm_lp[wm_lp - 1] =
3405			(ilk_wm_lp_latency(dev_priv, level) << WM1_LP_LATENCY_SHIFT) |
3406			(r->pri_val << WM1_LP_SR_SHIFT) |
3407			r->cur_val;
3408
3409		if (r->enable)
3410			results->wm_lp[wm_lp - 1] |= WM1_LP_SR_EN;
3411
3412		if (INTEL_GEN(dev_priv) >= 8)
3413			results->wm_lp[wm_lp - 1] |=
3414				r->fbc_val << WM1_LP_FBC_SHIFT_BDW;
3415		else
3416			results->wm_lp[wm_lp - 1] |=
3417				r->fbc_val << WM1_LP_FBC_SHIFT;
3418
3419		/*
3420		 * Always set WM1S_LP_EN when spr_val != 0, even if the
3421		 * level is disabled. Doing otherwise could cause underruns.
3422		 */
3423		if (INTEL_GEN(dev_priv) <= 6 && r->spr_val) {
3424			drm_WARN_ON(&dev_priv->drm, wm_lp != 1);
3425			results->wm_lp_spr[wm_lp - 1] = WM1S_LP_EN | r->spr_val;
3426		} else
3427			results->wm_lp_spr[wm_lp - 1] = r->spr_val;
3428	}
3429
3430	/* LP0 register values */
3431	for_each_intel_crtc(&dev_priv->drm, intel_crtc) {
3432		enum pipe pipe = intel_crtc->pipe;
3433		const struct intel_pipe_wm *pipe_wm = &intel_crtc->wm.active.ilk;
3434		const struct intel_wm_level *r = &pipe_wm->wm[0];
3435
3436		if (drm_WARN_ON(&dev_priv->drm, !r->enable))
3437			continue;
3438
3439		results->wm_pipe[pipe] =
3440			(r->pri_val << WM0_PIPE_PLANE_SHIFT) |
3441			(r->spr_val << WM0_PIPE_SPRITE_SHIFT) |
3442			r->cur_val;
3443	}
3444}
3445
3446/* Find the result with the highest level enabled. Check for enable_fbc_wm in
3447 * case both are at the same level. Prefer r1 in case they're the same. */
3448static struct intel_pipe_wm *
3449ilk_find_best_result(struct drm_i915_private *dev_priv,
3450		     struct intel_pipe_wm *r1,
3451		     struct intel_pipe_wm *r2)
3452{
3453	int level, max_level = ilk_wm_max_level(dev_priv);
3454	int level1 = 0, level2 = 0;
3455
3456	for (level = 1; level <= max_level; level++) {
3457		if (r1->wm[level].enable)
3458			level1 = level;
3459		if (r2->wm[level].enable)
3460			level2 = level;
3461	}
3462
3463	if (level1 == level2) {
3464		if (r2->fbc_wm_enabled && !r1->fbc_wm_enabled)
3465			return r2;
3466		else
3467			return r1;
3468	} else if (level1 > level2) {
3469		return r1;
3470	} else {
3471		return r2;
3472	}
3473}
3474
3475/* dirty bits used to track which watermarks need changes */
3476#define WM_DIRTY_PIPE(pipe) (1 << (pipe))
3477#define WM_DIRTY_LP(wm_lp) (1 << (15 + (wm_lp)))
3478#define WM_DIRTY_LP_ALL (WM_DIRTY_LP(1) | WM_DIRTY_LP(2) | WM_DIRTY_LP(3))
3479#define WM_DIRTY_FBC (1 << 24)
3480#define WM_DIRTY_DDB (1 << 25)
3481
3482static unsigned int ilk_compute_wm_dirty(struct drm_i915_private *dev_priv,
3483					 const struct ilk_wm_values *old,
3484					 const struct ilk_wm_values *new)
3485{
3486	unsigned int dirty = 0;
3487	enum pipe pipe;
3488	int wm_lp;
3489
3490	for_each_pipe(dev_priv, pipe) {
3491		if (old->wm_pipe[pipe] != new->wm_pipe[pipe]) {
3492			dirty |= WM_DIRTY_PIPE(pipe);
3493			/* Must disable LP1+ watermarks too */
3494			dirty |= WM_DIRTY_LP_ALL;
3495		}
3496	}
3497
3498	if (old->enable_fbc_wm != new->enable_fbc_wm) {
3499		dirty |= WM_DIRTY_FBC;
3500		/* Must disable LP1+ watermarks too */
3501		dirty |= WM_DIRTY_LP_ALL;
3502	}
3503
3504	if (old->partitioning != new->partitioning) {
3505		dirty |= WM_DIRTY_DDB;
3506		/* Must disable LP1+ watermarks too */
3507		dirty |= WM_DIRTY_LP_ALL;
3508	}
3509
3510	/* LP1+ watermarks already deemed dirty, no need to continue */
3511	if (dirty & WM_DIRTY_LP_ALL)
3512		return dirty;
3513
3514	/* Find the lowest numbered LP1+ watermark in need of an update... */
3515	for (wm_lp = 1; wm_lp <= 3; wm_lp++) {
3516		if (old->wm_lp[wm_lp - 1] != new->wm_lp[wm_lp - 1] ||
3517		    old->wm_lp_spr[wm_lp - 1] != new->wm_lp_spr[wm_lp - 1])
3518			break;
3519	}
3520
3521	/* ...and mark it and all higher numbered LP1+ watermarks as dirty */
3522	for (; wm_lp <= 3; wm_lp++)
3523		dirty |= WM_DIRTY_LP(wm_lp);
3524
3525	return dirty;
3526}
3527
3528static bool _ilk_disable_lp_wm(struct drm_i915_private *dev_priv,
3529			       unsigned int dirty)
3530{
3531	struct ilk_wm_values *previous = &dev_priv->wm.hw;
3532	bool changed = false;
3533
3534	if (dirty & WM_DIRTY_LP(3) && previous->wm_lp[2] & WM1_LP_SR_EN) {
3535		previous->wm_lp[2] &= ~WM1_LP_SR_EN;
3536		I915_WRITE(WM3_LP_ILK, previous->wm_lp[2]);
3537		changed = true;
3538	}
3539	if (dirty & WM_DIRTY_LP(2) && previous->wm_lp[1] & WM1_LP_SR_EN) {
3540		previous->wm_lp[1] &= ~WM1_LP_SR_EN;
3541		I915_WRITE(WM2_LP_ILK, previous->wm_lp[1]);
3542		changed = true;
3543	}
3544	if (dirty & WM_DIRTY_LP(1) && previous->wm_lp[0] & WM1_LP_SR_EN) {
3545		previous->wm_lp[0] &= ~WM1_LP_SR_EN;
3546		I915_WRITE(WM1_LP_ILK, previous->wm_lp[0]);
3547		changed = true;
3548	}
3549
3550	/*
3551	 * Don't touch WM1S_LP_EN here.
3552	 * Doing so could cause underruns.
3553	 */
3554
3555	return changed;
3556}
3557
3558/*
3559 * The spec says we shouldn't write when we don't need, because every write
3560 * causes WMs to be re-evaluated, expending some power.
3561 */
3562static void ilk_write_wm_values(struct drm_i915_private *dev_priv,
3563				struct ilk_wm_values *results)
3564{
3565	struct ilk_wm_values *previous = &dev_priv->wm.hw;
3566	unsigned int dirty;
3567	u32 val;
3568
3569	dirty = ilk_compute_wm_dirty(dev_priv, previous, results);
3570	if (!dirty)
3571		return;
3572
3573	_ilk_disable_lp_wm(dev_priv, dirty);
3574
3575	if (dirty & WM_DIRTY_PIPE(PIPE_A))
3576		I915_WRITE(WM0_PIPEA_ILK, results->wm_pipe[0]);
3577	if (dirty & WM_DIRTY_PIPE(PIPE_B))
3578		I915_WRITE(WM0_PIPEB_ILK, results->wm_pipe[1]);
3579	if (dirty & WM_DIRTY_PIPE(PIPE_C))
3580		I915_WRITE(WM0_PIPEC_IVB, results->wm_pipe[2]);
3581
3582	if (dirty & WM_DIRTY_DDB) {
3583		if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv)) {
3584			val = I915_READ(WM_MISC);
3585			if (results->partitioning == INTEL_DDB_PART_1_2)
3586				val &= ~WM_MISC_DATA_PARTITION_5_6;
3587			else
3588				val |= WM_MISC_DATA_PARTITION_5_6;
3589			I915_WRITE(WM_MISC, val);
3590		} else {
3591			val = I915_READ(DISP_ARB_CTL2);
3592			if (results->partitioning == INTEL_DDB_PART_1_2)
3593				val &= ~DISP_DATA_PARTITION_5_6;
3594			else
3595				val |= DISP_DATA_PARTITION_5_6;
3596			I915_WRITE(DISP_ARB_CTL2, val);
3597		}
3598	}
3599
3600	if (dirty & WM_DIRTY_FBC) {
3601		val = I915_READ(DISP_ARB_CTL);
3602		if (results->enable_fbc_wm)
3603			val &= ~DISP_FBC_WM_DIS;
3604		else
3605			val |= DISP_FBC_WM_DIS;
3606		I915_WRITE(DISP_ARB_CTL, val);
3607	}
3608
3609	if (dirty & WM_DIRTY_LP(1) &&
3610	    previous->wm_lp_spr[0] != results->wm_lp_spr[0])
3611		I915_WRITE(WM1S_LP_ILK, results->wm_lp_spr[0]);
3612
3613	if (INTEL_GEN(dev_priv) >= 7) {
3614		if (dirty & WM_DIRTY_LP(2) && previous->wm_lp_spr[1] != results->wm_lp_spr[1])
3615			I915_WRITE(WM2S_LP_IVB, results->wm_lp_spr[1]);
3616		if (dirty & WM_DIRTY_LP(3) && previous->wm_lp_spr[2] != results->wm_lp_spr[2])
3617			I915_WRITE(WM3S_LP_IVB, results->wm_lp_spr[2]);
3618	}
3619
3620	if (dirty & WM_DIRTY_LP(1) && previous->wm_lp[0] != results->wm_lp[0])
3621		I915_WRITE(WM1_LP_ILK, results->wm_lp[0]);
3622	if (dirty & WM_DIRTY_LP(2) && previous->wm_lp[1] != results->wm_lp[1])
3623		I915_WRITE(WM2_LP_ILK, results->wm_lp[1]);
3624	if (dirty & WM_DIRTY_LP(3) && previous->wm_lp[2] != results->wm_lp[2])
3625		I915_WRITE(WM3_LP_ILK, results->wm_lp[2]);
3626
3627	dev_priv->wm.hw = *results;
3628}
3629
3630bool ilk_disable_lp_wm(struct drm_i915_private *dev_priv)
3631{
3632	return _ilk_disable_lp_wm(dev_priv, WM_DIRTY_LP_ALL);
3633}
3634
3635u8 intel_enabled_dbuf_slices_mask(struct drm_i915_private *dev_priv)
3636{
3637	int i;
3638	int max_slices = INTEL_INFO(dev_priv)->num_supported_dbuf_slices;
3639	u8 enabled_slices_mask = 0;
3640
3641	for (i = 0; i < max_slices; i++) {
3642		if (I915_READ(DBUF_CTL_S(i)) & DBUF_POWER_STATE)
3643			enabled_slices_mask |= BIT(i);
3644	}
3645
3646	return enabled_slices_mask;
3647}
3648
3649/*
3650 * FIXME: We still don't have the proper code detect if we need to apply the WA,
3651 * so assume we'll always need it in order to avoid underruns.
3652 */
3653static bool skl_needs_memory_bw_wa(struct drm_i915_private *dev_priv)
3654{
3655	return IS_GEN9_BC(dev_priv) || IS_BROXTON(dev_priv);
3656}
3657
3658static bool
3659intel_has_sagv(struct drm_i915_private *dev_priv)
3660{
3661	return (IS_GEN9_BC(dev_priv) || INTEL_GEN(dev_priv) >= 10) &&
3662		dev_priv->sagv_status != I915_SAGV_NOT_CONTROLLED;
3663}
3664
3665static void
3666skl_setup_sagv_block_time(struct drm_i915_private *dev_priv)
3667{
3668	if (INTEL_GEN(dev_priv) >= 12) {
3669		u32 val = 0;
3670		int ret;
3671
3672		ret = sandybridge_pcode_read(dev_priv,
3673					     GEN12_PCODE_READ_SAGV_BLOCK_TIME_US,
3674					     &val, NULL);
3675		if (!ret) {
3676			dev_priv->sagv_block_time_us = val;
3677			return;
3678		}
3679
3680		drm_dbg(&dev_priv->drm, "Couldn't read SAGV block time!\n");
3681	} else if (IS_GEN(dev_priv, 11)) {
3682		dev_priv->sagv_block_time_us = 10;
3683		return;
3684	} else if (IS_GEN(dev_priv, 10)) {
3685		dev_priv->sagv_block_time_us = 20;
3686		return;
3687	} else if (IS_GEN(dev_priv, 9)) {
3688		dev_priv->sagv_block_time_us = 30;
3689		return;
3690	} else {
3691		MISSING_CASE(INTEL_GEN(dev_priv));
3692	}
3693
3694	/* Default to an unusable block time */
3695	dev_priv->sagv_block_time_us = -1;
3696}
3697
3698/*
3699 * SAGV dynamically adjusts the system agent voltage and clock frequencies
3700 * depending on power and performance requirements. The display engine access
3701 * to system memory is blocked during the adjustment time. Because of the
3702 * blocking time, having this enabled can cause full system hangs and/or pipe
3703 * underruns if we don't meet all of the following requirements:
3704 *
3705 *  - <= 1 pipe enabled
3706 *  - All planes can enable watermarks for latencies >= SAGV engine block time
3707 *  - We're not using an interlaced display configuration
3708 */
3709int
3710intel_enable_sagv(struct drm_i915_private *dev_priv)
3711{
3712	int ret;
3713
3714	if (!intel_has_sagv(dev_priv))
3715		return 0;
3716
3717	if (dev_priv->sagv_status == I915_SAGV_ENABLED)
3718		return 0;
3719
3720	drm_dbg_kms(&dev_priv->drm, "Enabling SAGV\n");
3721	ret = sandybridge_pcode_write(dev_priv, GEN9_PCODE_SAGV_CONTROL,
3722				      GEN9_SAGV_ENABLE);
3723
3724	/* We don't need to wait for SAGV when enabling */
3725
3726	/*
3727	 * Some skl systems, pre-release machines in particular,
3728	 * don't actually have SAGV.
3729	 */
3730	if (IS_SKYLAKE(dev_priv) && ret == -ENXIO) {
3731		drm_dbg(&dev_priv->drm, "No SAGV found on system, ignoring\n");
3732		dev_priv->sagv_status = I915_SAGV_NOT_CONTROLLED;
3733		return 0;
3734	} else if (ret < 0) {
3735		drm_err(&dev_priv->drm, "Failed to enable SAGV\n");
3736		return ret;
3737	}
3738
3739	dev_priv->sagv_status = I915_SAGV_ENABLED;
3740	return 0;
3741}
3742
3743int
3744intel_disable_sagv(struct drm_i915_private *dev_priv)
3745{
3746	int ret;
3747
3748	if (!intel_has_sagv(dev_priv))
3749		return 0;
3750
3751	if (dev_priv->sagv_status == I915_SAGV_DISABLED)
3752		return 0;
3753
3754	drm_dbg_kms(&dev_priv->drm, "Disabling SAGV\n");
3755	/* bspec says to keep retrying for at least 1 ms */
3756	ret = skl_pcode_request(dev_priv, GEN9_PCODE_SAGV_CONTROL,
3757				GEN9_SAGV_DISABLE,
3758				GEN9_SAGV_IS_DISABLED, GEN9_SAGV_IS_DISABLED,
3759				1);
3760	/*
3761	 * Some skl systems, pre-release machines in particular,
3762	 * don't actually have SAGV.
3763	 */
3764	if (IS_SKYLAKE(dev_priv) && ret == -ENXIO) {
3765		drm_dbg(&dev_priv->drm, "No SAGV found on system, ignoring\n");
3766		dev_priv->sagv_status = I915_SAGV_NOT_CONTROLLED;
3767		return 0;
3768	} else if (ret < 0) {
3769		drm_err(&dev_priv->drm, "Failed to disable SAGV (%d)\n", ret);
3770		return ret;
3771	}
3772
3773	dev_priv->sagv_status = I915_SAGV_DISABLED;
3774	return 0;
3775}
3776
3777void intel_sagv_pre_plane_update(struct intel_atomic_state *state)
3778{
3779	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
3780	const struct intel_bw_state *new_bw_state;
3781	const struct intel_bw_state *old_bw_state;
3782	u32 new_mask = 0;
3783
3784	/*
3785	 * Just return if we can't control SAGV or don't have it.
3786	 * This is different from situation when we have SAGV but just can't
3787	 * afford it due to DBuf limitation - in case if SAGV is completely
3788	 * disabled in a BIOS, we are not even allowed to send a PCode request,
3789	 * as it will throw an error. So have to check it here.
3790	 */
3791	if (!intel_has_sagv(dev_priv))
3792		return;
3793
3794	new_bw_state = intel_atomic_get_new_bw_state(state);
3795	if (!new_bw_state)
3796		return;
3797
3798	if (INTEL_GEN(dev_priv) < 11 && !intel_can_enable_sagv(dev_priv, new_bw_state)) {
3799		intel_disable_sagv(dev_priv);
3800		return;
3801	}
3802
3803	old_bw_state = intel_atomic_get_old_bw_state(state);
3804	/*
3805	 * Nothing to mask
3806	 */
3807	if (new_bw_state->qgv_points_mask == old_bw_state->qgv_points_mask)
3808		return;
3809
3810	new_mask = old_bw_state->qgv_points_mask | new_bw_state->qgv_points_mask;
3811
3812	/*
3813	 * If new mask is zero - means there is nothing to mask,
3814	 * we can only unmask, which should be done in unmask.
3815	 */
3816	if (!new_mask)
3817		return;
3818
3819	/*
3820	 * Restrict required qgv points before updating the configuration.
3821	 * According to BSpec we can't mask and unmask qgv points at the same
3822	 * time. Also masking should be done before updating the configuration
3823	 * and unmasking afterwards.
3824	 */
3825	icl_pcode_restrict_qgv_points(dev_priv, new_mask);
3826}
3827
3828void intel_sagv_post_plane_update(struct intel_atomic_state *state)
3829{
3830	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
3831	const struct intel_bw_state *new_bw_state;
3832	const struct intel_bw_state *old_bw_state;
3833	u32 new_mask = 0;
3834
3835	/*
3836	 * Just return if we can't control SAGV or don't have it.
3837	 * This is different from situation when we have SAGV but just can't
3838	 * afford it due to DBuf limitation - in case if SAGV is completely
3839	 * disabled in a BIOS, we are not even allowed to send a PCode request,
3840	 * as it will throw an error. So have to check it here.
3841	 */
3842	if (!intel_has_sagv(dev_priv))
3843		return;
3844
3845	new_bw_state = intel_atomic_get_new_bw_state(state);
3846	if (!new_bw_state)
3847		return;
3848
3849	if (INTEL_GEN(dev_priv) < 11 && intel_can_enable_sagv(dev_priv, new_bw_state)) {
3850		intel_enable_sagv(dev_priv);
3851		return;
3852	}
3853
3854	old_bw_state = intel_atomic_get_old_bw_state(state);
3855	/*
3856	 * Nothing to unmask
3857	 */
3858	if (new_bw_state->qgv_points_mask == old_bw_state->qgv_points_mask)
3859		return;
3860
3861	new_mask = new_bw_state->qgv_points_mask;
3862
3863	/*
3864	 * Allow required qgv points after updating the configuration.
3865	 * According to BSpec we can't mask and unmask qgv points at the same
3866	 * time. Also masking should be done before updating the configuration
3867	 * and unmasking afterwards.
3868	 */
3869	icl_pcode_restrict_qgv_points(dev_priv, new_mask);
3870}
3871
3872static bool skl_crtc_can_enable_sagv(const struct intel_crtc_state *crtc_state)
3873{
3874	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
3875	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
3876	struct intel_plane *plane;
3877	const struct intel_plane_state *plane_state;
3878	int level, latency;
3879
3880	if (!intel_has_sagv(dev_priv))
3881		return false;
3882
3883	if (!crtc_state->hw.active)
3884		return true;
3885
3886	if (crtc_state->hw.adjusted_mode.flags & DRM_MODE_FLAG_INTERLACE)
3887		return false;
3888
3889	intel_atomic_crtc_state_for_each_plane_state(plane, plane_state, crtc_state) {
3890		const struct skl_plane_wm *wm =
3891			&crtc_state->wm.skl.optimal.planes[plane->id];
3892
3893		/* Skip this plane if it's not enabled */
3894		if (!wm->wm[0].plane_en)
3895			continue;
3896
3897		/* Find the highest enabled wm level for this plane */
3898		for (level = ilk_wm_max_level(dev_priv);
3899		     !wm->wm[level].plane_en; --level)
3900		     { }
3901
3902		latency = dev_priv->wm.skl_latency[level];
3903
3904		if (skl_needs_memory_bw_wa(dev_priv) &&
3905		    plane_state->uapi.fb->modifier ==
3906		    I915_FORMAT_MOD_X_TILED)
3907			latency += 15;
3908
3909		/*
3910		 * If any of the planes on this pipe don't enable wm levels that
3911		 * incur memory latencies higher than sagv_block_time_us we
3912		 * can't enable SAGV.
3913		 */
3914		if (latency < dev_priv->sagv_block_time_us)
3915			return false;
3916	}
3917
3918	return true;
3919}
3920
3921static bool tgl_crtc_can_enable_sagv(const struct intel_crtc_state *crtc_state)
3922{
3923	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
3924	enum plane_id plane_id;
3925
3926	if (!crtc_state->hw.active)
3927		return true;
3928
3929	for_each_plane_id_on_crtc(crtc, plane_id) {
3930		const struct skl_ddb_entry *plane_alloc =
3931			&crtc_state->wm.skl.plane_ddb_y[plane_id];
3932		const struct skl_plane_wm *wm =
3933			&crtc_state->wm.skl.optimal.planes[plane_id];
3934
3935		if (skl_ddb_entry_size(plane_alloc) < wm->sagv_wm0.min_ddb_alloc)
3936			return false;
3937	}
3938
3939	return true;
3940}
3941
3942static bool intel_crtc_can_enable_sagv(const struct intel_crtc_state *crtc_state)
3943{
3944	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
3945	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
3946
3947	if (INTEL_GEN(dev_priv) >= 12)
3948		return tgl_crtc_can_enable_sagv(crtc_state);
3949	else
3950		return skl_crtc_can_enable_sagv(crtc_state);
3951}
3952
3953bool intel_can_enable_sagv(struct drm_i915_private *dev_priv,
3954			   const struct intel_bw_state *bw_state)
3955{
3956	if (INTEL_GEN(dev_priv) < 11 &&
3957	    bw_state->active_pipes && !is_power_of_2(bw_state->active_pipes))
3958		return false;
3959
3960	return bw_state->pipe_sagv_reject == 0;
3961}
3962
3963static int intel_compute_sagv_mask(struct intel_atomic_state *state)
3964{
3965	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
3966	int ret;
3967	struct intel_crtc *crtc;
3968	struct intel_crtc_state *new_crtc_state;
3969	struct intel_bw_state *new_bw_state = NULL;
3970	const struct intel_bw_state *old_bw_state = NULL;
3971	int i;
3972
3973	for_each_new_intel_crtc_in_state(state, crtc,
3974					 new_crtc_state, i) {
3975		new_bw_state = intel_atomic_get_bw_state(state);
3976		if (IS_ERR(new_bw_state))
3977			return PTR_ERR(new_bw_state);
3978
3979		old_bw_state = intel_atomic_get_old_bw_state(state);
3980
3981		if (intel_crtc_can_enable_sagv(new_crtc_state))
3982			new_bw_state->pipe_sagv_reject &= ~BIT(crtc->pipe);
3983		else
3984			new_bw_state->pipe_sagv_reject |= BIT(crtc->pipe);
3985	}
3986
3987	if (!new_bw_state)
3988		return 0;
3989
3990	new_bw_state->active_pipes =
3991		intel_calc_active_pipes(state, old_bw_state->active_pipes);
3992
3993	if (new_bw_state->active_pipes != old_bw_state->active_pipes) {
3994		ret = intel_atomic_lock_global_state(&new_bw_state->base);
3995		if (ret)
3996			return ret;
3997	}
3998
3999	if (intel_can_enable_sagv(dev_priv, new_bw_state) !=
4000	    intel_can_enable_sagv(dev_priv, old_bw_state)) {
4001		ret = intel_atomic_serialize_global_state(&new_bw_state->base);
4002		if (ret)
4003			return ret;
4004	} else if (new_bw_state->pipe_sagv_reject != old_bw_state->pipe_sagv_reject) {
4005		ret = intel_atomic_lock_global_state(&new_bw_state->base);
4006		if (ret)
4007			return ret;
4008	}
4009
4010	for_each_new_intel_crtc_in_state(state, crtc,
4011					 new_crtc_state, i) {
4012		struct skl_pipe_wm *pipe_wm = &new_crtc_state->wm.skl.optimal;
4013
4014		/*
4015		 * We store use_sagv_wm in the crtc state rather than relying on
4016		 * that bw state since we have no convenient way to get at the
4017		 * latter from the plane commit hooks (especially in the legacy
4018		 * cursor case)
4019		 */
4020		pipe_wm->use_sagv_wm = INTEL_GEN(dev_priv) >= 12 &&
4021				       intel_can_enable_sagv(dev_priv, new_bw_state);
4022	}
4023
4024	return 0;
4025}
4026
4027/*
4028 * Calculate initial DBuf slice offset, based on slice size
4029 * and mask(i.e if slice size is 1024 and second slice is enabled
4030 * offset would be 1024)
4031 */
4032static unsigned int
4033icl_get_first_dbuf_slice_offset(u32 dbuf_slice_mask,
4034				u32 slice_size,
4035				u32 ddb_size)
4036{
4037	unsigned int offset = 0;
4038
4039	if (!dbuf_slice_mask)
4040		return 0;
4041
4042	offset = (ffs(dbuf_slice_mask) - 1) * slice_size;
4043
4044	WARN_ON(offset >= ddb_size);
4045	return offset;
4046}
4047
4048u16 intel_get_ddb_size(struct drm_i915_private *dev_priv)
4049{
4050	u16 ddb_size = INTEL_INFO(dev_priv)->ddb_size;
4051	drm_WARN_ON(&dev_priv->drm, ddb_size == 0);
4052
4053	if (INTEL_GEN(dev_priv) < 11)
4054		return ddb_size - 4; /* 4 blocks for bypass path allocation */
4055
4056	return ddb_size;
4057}
4058
4059u32 skl_ddb_dbuf_slice_mask(struct drm_i915_private *dev_priv,
4060			    const struct skl_ddb_entry *entry)
4061{
4062	u32 slice_mask = 0;
4063	u16 ddb_size = intel_get_ddb_size(dev_priv);
4064	u16 num_supported_slices = INTEL_INFO(dev_priv)->num_supported_dbuf_slices;
4065	u16 slice_size = ddb_size / num_supported_slices;
4066	u16 start_slice;
4067	u16 end_slice;
4068
4069	if (!skl_ddb_entry_size(entry))
4070		return 0;
4071
4072	start_slice = entry->start / slice_size;
4073	end_slice = (entry->end - 1) / slice_size;
4074
4075	/*
4076	 * Per plane DDB entry can in a really worst case be on multiple slices
4077	 * but single entry is anyway contigious.
4078	 */
4079	while (start_slice <= end_slice) {
4080		slice_mask |= BIT(start_slice);
4081		start_slice++;
4082	}
4083
4084	return slice_mask;
4085}
4086
4087static u8 skl_compute_dbuf_slices(const struct intel_crtc_state *crtc_state,
4088				  u8 active_pipes);
4089
4090static int
4091skl_ddb_get_pipe_allocation_limits(struct drm_i915_private *dev_priv,
4092				   const struct intel_crtc_state *crtc_state,
4093				   const u64 total_data_rate,
4094				   struct skl_ddb_entry *alloc, /* out */
4095				   int *num_active /* out */)
4096{
4097	struct drm_atomic_state *state = crtc_state->uapi.state;
4098	struct intel_atomic_state *intel_state = to_intel_atomic_state(state);
4099	struct drm_crtc *for_crtc = crtc_state->uapi.crtc;
4100	const struct intel_crtc *crtc;
4101	u32 pipe_width = 0, total_width_in_range = 0, width_before_pipe_in_range = 0;
4102	enum pipe for_pipe = to_intel_crtc(for_crtc)->pipe;
4103	struct intel_dbuf_state *new_dbuf_state =
4104		intel_atomic_get_new_dbuf_state(intel_state);
4105	const struct intel_dbuf_state *old_dbuf_state =
4106		intel_atomic_get_old_dbuf_state(intel_state);
4107	u8 active_pipes = new_dbuf_state->active_pipes;
4108	u16 ddb_size;
4109	u32 ddb_range_size;
4110	u32 i;
4111	u32 dbuf_slice_mask;
4112	u32 offset;
4113	u32 slice_size;
4114	u32 total_slice_mask;
4115	u32 start, end;
4116	int ret;
4117
4118	*num_active = hweight8(active_pipes);
4119
4120	if (!crtc_state->hw.active) {
4121		alloc->start = 0;
4122		alloc->end = 0;
4123		return 0;
4124	}
4125
4126	ddb_size = intel_get_ddb_size(dev_priv);
4127
4128	slice_size = ddb_size / INTEL_INFO(dev_priv)->num_supported_dbuf_slices;
4129
4130	/*
4131	 * If the state doesn't change the active CRTC's or there is no
4132	 * modeset request, then there's no need to recalculate;
4133	 * the existing pipe allocation limits should remain unchanged.
4134	 * Note that we're safe from racing commits since any racing commit
4135	 * that changes the active CRTC list or do modeset would need to
4136	 * grab _all_ crtc locks, including the one we currently hold.
4137	 */
4138	if (old_dbuf_state->active_pipes == new_dbuf_state->active_pipes &&
4139	    !dev_priv->wm.distrust_bios_wm) {
4140		/*
4141		 * alloc may be cleared by clear_intel_crtc_state,
4142		 * copy from old state to be sure
4143		 *
4144		 * FIXME get rid of this mess
4145		 */
4146		*alloc = to_intel_crtc_state(for_crtc->state)->wm.skl.ddb;
4147		return 0;
4148	}
4149
4150	/*
4151	 * Get allowed DBuf slices for correspondent pipe and platform.
4152	 */
4153	dbuf_slice_mask = skl_compute_dbuf_slices(crtc_state, active_pipes);
4154
4155	/*
4156	 * Figure out at which DBuf slice we start, i.e if we start at Dbuf S2
4157	 * and slice size is 1024, the offset would be 1024
4158	 */
4159	offset = icl_get_first_dbuf_slice_offset(dbuf_slice_mask,
4160						 slice_size, ddb_size);
4161
4162	/*
4163	 * Figure out total size of allowed DBuf slices, which is basically
4164	 * a number of allowed slices for that pipe multiplied by slice size.
4165	 * Inside of this
4166	 * range ddb entries are still allocated in proportion to display width.
4167	 */
4168	ddb_range_size = hweight8(dbuf_slice_mask) * slice_size;
4169
4170	/*
4171	 * Watermark/ddb requirement highly depends upon width of the
4172	 * framebuffer, So instead of allocating DDB equally among pipes
4173	 * distribute DDB based on resolution/width of the display.
4174	 */
4175	total_slice_mask = dbuf_slice_mask;
4176	for_each_new_intel_crtc_in_state(intel_state, crtc, crtc_state, i) {
4177		const struct drm_display_mode *adjusted_mode =
4178			&crtc_state->hw.adjusted_mode;
4179		enum pipe pipe = crtc->pipe;
4180		int hdisplay, vdisplay;
4181		u32 pipe_dbuf_slice_mask;
4182
4183		if (!crtc_state->hw.active)
4184			continue;
4185
4186		pipe_dbuf_slice_mask = skl_compute_dbuf_slices(crtc_state,
4187							       active_pipes);
4188
4189		/*
4190		 * According to BSpec pipe can share one dbuf slice with another
4191		 * pipes or pipe can use multiple dbufs, in both cases we
4192		 * account for other pipes only if they have exactly same mask.
4193		 * However we need to account how many slices we should enable
4194		 * in total.
4195		 */
4196		total_slice_mask |= pipe_dbuf_slice_mask;
4197
4198		/*
4199		 * Do not account pipes using other slice sets
4200		 * luckily as of current BSpec slice sets do not partially
4201		 * intersect(pipes share either same one slice or same slice set
4202		 * i.e no partial intersection), so it is enough to check for
4203		 * equality for now.
4204		 */
4205		if (dbuf_slice_mask != pipe_dbuf_slice_mask)
4206			continue;
4207
4208		drm_mode_get_hv_timing(adjusted_mode, &hdisplay, &vdisplay);
4209
4210		total_width_in_range += hdisplay;
4211
4212		if (pipe < for_pipe)
4213			width_before_pipe_in_range += hdisplay;
4214		else if (pipe == for_pipe)
4215			pipe_width = hdisplay;
4216	}
4217
4218	/*
4219	 * FIXME: For now we always enable slice S1 as per
4220	 * the Bspec display initialization sequence.
4221	 */
4222	new_dbuf_state->enabled_slices = total_slice_mask | BIT(DBUF_S1);
4223
4224	if (old_dbuf_state->enabled_slices != new_dbuf_state->enabled_slices) {
4225		ret = intel_atomic_serialize_global_state(&new_dbuf_state->base);
4226		if (ret)
4227			return ret;
4228	}
4229
4230	start = ddb_range_size * width_before_pipe_in_range / total_width_in_range;
4231	end = ddb_range_size *
4232		(width_before_pipe_in_range + pipe_width) / total_width_in_range;
4233
4234	alloc->start = offset + start;
4235	alloc->end = offset + end;
4236
4237	drm_dbg_kms(&dev_priv->drm,
4238		    "[CRTC:%d:%s] dbuf slices 0x%x, ddb (%d - %d), active pipes 0x%x\n",
4239		    for_crtc->base.id, for_crtc->name,
4240		    dbuf_slice_mask, alloc->start, alloc->end, active_pipes);
4241
4242	return 0;
4243}
4244
4245static int skl_compute_wm_params(const struct intel_crtc_state *crtc_state,
4246				 int width, const struct drm_format_info *format,
4247				 u64 modifier, unsigned int rotation,
4248				 u32 plane_pixel_rate, struct skl_wm_params *wp,
4249				 int color_plane);
4250static void skl_compute_plane_wm(const struct intel_crtc_state *crtc_state,
4251				 int level,
4252				 unsigned int latency,
4253				 const struct skl_wm_params *wp,
4254				 const struct skl_wm_level *result_prev,
4255				 struct skl_wm_level *result /* out */);
4256
4257static unsigned int
4258skl_cursor_allocation(const struct intel_crtc_state *crtc_state,
4259		      int num_active)
4260{
4261	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
4262	int level, max_level = ilk_wm_max_level(dev_priv);
4263	struct skl_wm_level wm = {};
4264	int ret, min_ddb_alloc = 0;
4265	struct skl_wm_params wp;
4266
4267	ret = skl_compute_wm_params(crtc_state, 256,
4268				    drm_format_info(DRM_FORMAT_ARGB8888),
4269				    DRM_FORMAT_MOD_LINEAR,
4270				    DRM_MODE_ROTATE_0,
4271				    crtc_state->pixel_rate, &wp, 0);
4272	drm_WARN_ON(&dev_priv->drm, ret);
4273
4274	for (level = 0; level <= max_level; level++) {
4275		unsigned int latency = dev_priv->wm.skl_latency[level];
4276
4277		skl_compute_plane_wm(crtc_state, level, latency, &wp, &wm, &wm);
4278		if (wm.min_ddb_alloc == U16_MAX)
4279			break;
4280
4281		min_ddb_alloc = wm.min_ddb_alloc;
4282	}
4283
4284	return max(num_active == 1 ? 32 : 8, min_ddb_alloc);
4285}
4286
4287static void skl_ddb_entry_init_from_hw(struct drm_i915_private *dev_priv,
4288				       struct skl_ddb_entry *entry, u32 reg)
4289{
4290
4291	entry->start = reg & DDB_ENTRY_MASK;
4292	entry->end = (reg >> DDB_ENTRY_END_SHIFT) & DDB_ENTRY_MASK;
4293
4294	if (entry->end)
4295		entry->end += 1;
4296}
4297
4298static void
4299skl_ddb_get_hw_plane_state(struct drm_i915_private *dev_priv,
4300			   const enum pipe pipe,
4301			   const enum plane_id plane_id,
4302			   struct skl_ddb_entry *ddb_y,
4303			   struct skl_ddb_entry *ddb_uv)
4304{
4305	u32 val, val2;
4306	u32 fourcc = 0;
4307
4308	/* Cursor doesn't support NV12/planar, so no extra calculation needed */
4309	if (plane_id == PLANE_CURSOR) {
4310		val = I915_READ(CUR_BUF_CFG(pipe));
4311		skl_ddb_entry_init_from_hw(dev_priv, ddb_y, val);
4312		return;
4313	}
4314
4315	val = I915_READ(PLANE_CTL(pipe, plane_id));
4316
4317	/* No DDB allocated for disabled planes */
4318	if (val & PLANE_CTL_ENABLE)
4319		fourcc = skl_format_to_fourcc(val & PLANE_CTL_FORMAT_MASK,
4320					      val & PLANE_CTL_ORDER_RGBX,
4321					      val & PLANE_CTL_ALPHA_MASK);
4322
4323	if (INTEL_GEN(dev_priv) >= 11) {
4324		val = I915_READ(PLANE_BUF_CFG(pipe, plane_id));
4325		skl_ddb_entry_init_from_hw(dev_priv, ddb_y, val);
4326	} else {
4327		val = I915_READ(PLANE_BUF_CFG(pipe, plane_id));
4328		val2 = I915_READ(PLANE_NV12_BUF_CFG(pipe, plane_id));
4329
4330		if (fourcc &&
4331		    drm_format_info_is_yuv_semiplanar(drm_format_info(fourcc)))
4332			swap(val, val2);
4333
4334		skl_ddb_entry_init_from_hw(dev_priv, ddb_y, val);
4335		skl_ddb_entry_init_from_hw(dev_priv, ddb_uv, val2);
4336	}
4337}
4338
4339void skl_pipe_ddb_get_hw_state(struct intel_crtc *crtc,
4340			       struct skl_ddb_entry *ddb_y,
4341			       struct skl_ddb_entry *ddb_uv)
4342{
4343	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
4344	enum intel_display_power_domain power_domain;
4345	enum pipe pipe = crtc->pipe;
4346	intel_wakeref_t wakeref;
4347	enum plane_id plane_id;
4348
4349	power_domain = POWER_DOMAIN_PIPE(pipe);
4350	wakeref = intel_display_power_get_if_enabled(dev_priv, power_domain);
4351	if (!wakeref)
4352		return;
4353
4354	for_each_plane_id_on_crtc(crtc, plane_id)
4355		skl_ddb_get_hw_plane_state(dev_priv, pipe,
4356					   plane_id,
4357					   &ddb_y[plane_id],
4358					   &ddb_uv[plane_id]);
4359
4360	intel_display_power_put(dev_priv, power_domain, wakeref);
4361}
4362
4363/*
4364 * Determines the downscale amount of a plane for the purposes of watermark calculations.
4365 * The bspec defines downscale amount as:
4366 *
4367 * """
4368 * Horizontal down scale amount = maximum[1, Horizontal source size /
4369 *                                           Horizontal destination size]
4370 * Vertical down scale amount = maximum[1, Vertical source size /
4371 *                                         Vertical destination size]
4372 * Total down scale amount = Horizontal down scale amount *
4373 *                           Vertical down scale amount
4374 * """
4375 *
4376 * Return value is provided in 16.16 fixed point form to retain fractional part.
4377 * Caller should take care of dividing & rounding off the value.
4378 */
4379static uint_fixed_16_16_t
4380skl_plane_downscale_amount(const struct intel_crtc_state *crtc_state,
4381			   const struct intel_plane_state *plane_state)
4382{
4383	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
4384	u32 src_w, src_h, dst_w, dst_h;
4385	uint_fixed_16_16_t fp_w_ratio, fp_h_ratio;
4386	uint_fixed_16_16_t downscale_h, downscale_w;
4387
4388	if (drm_WARN_ON(&dev_priv->drm,
4389			!intel_wm_plane_visible(crtc_state, plane_state)))
4390		return u32_to_fixed16(0);
4391
4392	/*
4393	 * Src coordinates are already rotated by 270 degrees for
4394	 * the 90/270 degree plane rotation cases (to match the
4395	 * GTT mapping), hence no need to account for rotation here.
4396	 *
4397	 * n.b., src is 16.16 fixed point, dst is whole integer.
4398	 */
4399	src_w = drm_rect_width(&plane_state->uapi.src) >> 16;
4400	src_h = drm_rect_height(&plane_state->uapi.src) >> 16;
4401	dst_w = drm_rect_width(&plane_state->uapi.dst);
4402	dst_h = drm_rect_height(&plane_state->uapi.dst);
4403
4404	fp_w_ratio = div_fixed16(src_w, dst_w);
4405	fp_h_ratio = div_fixed16(src_h, dst_h);
4406	downscale_w = max_fixed16(fp_w_ratio, u32_to_fixed16(1));
4407	downscale_h = max_fixed16(fp_h_ratio, u32_to_fixed16(1));
4408
4409	return mul_fixed16(downscale_w, downscale_h);
4410}
4411
4412struct dbuf_slice_conf_entry {
4413	u8 active_pipes;
4414	u8 dbuf_mask[I915_MAX_PIPES];
4415};
4416
4417/*
4418 * Table taken from Bspec 12716
4419 * Pipes do have some preferred DBuf slice affinity,
4420 * plus there are some hardcoded requirements on how
4421 * those should be distributed for multipipe scenarios.
4422 * For more DBuf slices algorithm can get even more messy
4423 * and less readable, so decided to use a table almost
4424 * as is from BSpec itself - that way it is at least easier
4425 * to compare, change and check.
4426 */
4427static const struct dbuf_slice_conf_entry icl_allowed_dbufs[] =
4428/* Autogenerated with igt/tools/intel_dbuf_map tool: */
4429{
4430	{
4431		.active_pipes = BIT(PIPE_A),
4432		.dbuf_mask = {
4433			[PIPE_A] = BIT(DBUF_S1),
4434		},
4435	},
4436	{
4437		.active_pipes = BIT(PIPE_B),
4438		.dbuf_mask = {
4439			[PIPE_B] = BIT(DBUF_S1),
4440		},
4441	},
4442	{
4443		.active_pipes = BIT(PIPE_A) | BIT(PIPE_B),
4444		.dbuf_mask = {
4445			[PIPE_A] = BIT(DBUF_S1),
4446			[PIPE_B] = BIT(DBUF_S2),
4447		},
4448	},
4449	{
4450		.active_pipes = BIT(PIPE_C),
4451		.dbuf_mask = {
4452			[PIPE_C] = BIT(DBUF_S2),
4453		},
4454	},
4455	{
4456		.active_pipes = BIT(PIPE_A) | BIT(PIPE_C),
4457		.dbuf_mask = {
4458			[PIPE_A] = BIT(DBUF_S1),
4459			[PIPE_C] = BIT(DBUF_S2),
4460		},
4461	},
4462	{
4463		.active_pipes = BIT(PIPE_B) | BIT(PIPE_C),
4464		.dbuf_mask = {
4465			[PIPE_B] = BIT(DBUF_S1),
4466			[PIPE_C] = BIT(DBUF_S2),
4467		},
4468	},
4469	{
4470		.active_pipes = BIT(PIPE_A) | BIT(PIPE_B) | BIT(PIPE_C),
4471		.dbuf_mask = {
4472			[PIPE_A] = BIT(DBUF_S1),
4473			[PIPE_B] = BIT(DBUF_S1),
4474			[PIPE_C] = BIT(DBUF_S2),
4475		},
4476	},
4477	{}
4478};
4479
4480/*
4481 * Table taken from Bspec 49255
4482 * Pipes do have some preferred DBuf slice affinity,
4483 * plus there are some hardcoded requirements on how
4484 * those should be distributed for multipipe scenarios.
4485 * For more DBuf slices algorithm can get even more messy
4486 * and less readable, so decided to use a table almost
4487 * as is from BSpec itself - that way it is at least easier
4488 * to compare, change and check.
4489 */
4490static const struct dbuf_slice_conf_entry tgl_allowed_dbufs[] =
4491/* Autogenerated with igt/tools/intel_dbuf_map tool: */
4492{
4493	{
4494		.active_pipes = BIT(PIPE_A),
4495		.dbuf_mask = {
4496			[PIPE_A] = BIT(DBUF_S1) | BIT(DBUF_S2),
4497		},
4498	},
4499	{
4500		.active_pipes = BIT(PIPE_B),
4501		.dbuf_mask = {
4502			[PIPE_B] = BIT(DBUF_S1) | BIT(DBUF_S2),
4503		},
4504	},
4505	{
4506		.active_pipes = BIT(PIPE_A) | BIT(PIPE_B),
4507		.dbuf_mask = {
4508			[PIPE_A] = BIT(DBUF_S2),
4509			[PIPE_B] = BIT(DBUF_S1),
4510		},
4511	},
4512	{
4513		.active_pipes = BIT(PIPE_C),
4514		.dbuf_mask = {
4515			[PIPE_C] = BIT(DBUF_S2) | BIT(DBUF_S1),
4516		},
4517	},
4518	{
4519		.active_pipes = BIT(PIPE_A) | BIT(PIPE_C),
4520		.dbuf_mask = {
4521			[PIPE_A] = BIT(DBUF_S1),
4522			[PIPE_C] = BIT(DBUF_S2),
4523		},
4524	},
4525	{
4526		.active_pipes = BIT(PIPE_B) | BIT(PIPE_C),
4527		.dbuf_mask = {
4528			[PIPE_B] = BIT(DBUF_S1),
4529			[PIPE_C] = BIT(DBUF_S2),
4530		},
4531	},
4532	{
4533		.active_pipes = BIT(PIPE_A) | BIT(PIPE_B) | BIT(PIPE_C),
4534		.dbuf_mask = {
4535			[PIPE_A] = BIT(DBUF_S1),
4536			[PIPE_B] = BIT(DBUF_S1),
4537			[PIPE_C] = BIT(DBUF_S2),
4538		},
4539	},
4540	{
4541		.active_pipes = BIT(PIPE_D),
4542		.dbuf_mask = {
4543			[PIPE_D] = BIT(DBUF_S2) | BIT(DBUF_S1),
4544		},
4545	},
4546	{
4547		.active_pipes = BIT(PIPE_A) | BIT(PIPE_D),
4548		.dbuf_mask = {
4549			[PIPE_A] = BIT(DBUF_S1),
4550			[PIPE_D] = BIT(DBUF_S2),
4551		},
4552	},
4553	{
4554		.active_pipes = BIT(PIPE_B) | BIT(PIPE_D),
4555		.dbuf_mask = {
4556			[PIPE_B] = BIT(DBUF_S1),
4557			[PIPE_D] = BIT(DBUF_S2),
4558		},
4559	},
4560	{
4561		.active_pipes = BIT(PIPE_A) | BIT(PIPE_B) | BIT(PIPE_D),
4562		.dbuf_mask = {
4563			[PIPE_A] = BIT(DBUF_S1),
4564			[PIPE_B] = BIT(DBUF_S1),
4565			[PIPE_D] = BIT(DBUF_S2),
4566		},
4567	},
4568	{
4569		.active_pipes = BIT(PIPE_C) | BIT(PIPE_D),
4570		.dbuf_mask = {
4571			[PIPE_C] = BIT(DBUF_S1),
4572			[PIPE_D] = BIT(DBUF_S2),
4573		},
4574	},
4575	{
4576		.active_pipes = BIT(PIPE_A) | BIT(PIPE_C) | BIT(PIPE_D),
4577		.dbuf_mask = {
4578			[PIPE_A] = BIT(DBUF_S1),
4579			[PIPE_C] = BIT(DBUF_S2),
4580			[PIPE_D] = BIT(DBUF_S2),
4581		},
4582	},
4583	{
4584		.active_pipes = BIT(PIPE_B) | BIT(PIPE_C) | BIT(PIPE_D),
4585		.dbuf_mask = {
4586			[PIPE_B] = BIT(DBUF_S1),
4587			[PIPE_C] = BIT(DBUF_S2),
4588			[PIPE_D] = BIT(DBUF_S2),
4589		},
4590	},
4591	{
4592		.active_pipes = BIT(PIPE_A) | BIT(PIPE_B) | BIT(PIPE_C) | BIT(PIPE_D),
4593		.dbuf_mask = {
4594			[PIPE_A] = BIT(DBUF_S1),
4595			[PIPE_B] = BIT(DBUF_S1),
4596			[PIPE_C] = BIT(DBUF_S2),
4597			[PIPE_D] = BIT(DBUF_S2),
4598		},
4599	},
4600	{}
4601};
4602
4603static u8 compute_dbuf_slices(enum pipe pipe, u8 active_pipes,
4604			      const struct dbuf_slice_conf_entry *dbuf_slices)
4605{
4606	int i;
4607
4608	for (i = 0; i < dbuf_slices[i].active_pipes; i++) {
4609		if (dbuf_slices[i].active_pipes == active_pipes)
4610			return dbuf_slices[i].dbuf_mask[pipe];
4611	}
4612	return 0;
4613}
4614
4615/*
4616 * This function finds an entry with same enabled pipe configuration and
4617 * returns correspondent DBuf slice mask as stated in BSpec for particular
4618 * platform.
4619 */
4620static u8 icl_compute_dbuf_slices(enum pipe pipe, u8 active_pipes)
4621{
4622	/*
4623	 * FIXME: For ICL this is still a bit unclear as prev BSpec revision
4624	 * required calculating "pipe ratio" in order to determine
4625	 * if one or two slices can be used for single pipe configurations
4626	 * as additional constraint to the existing table.
4627	 * However based on recent info, it should be not "pipe ratio"
4628	 * but rather ratio between pixel_rate and cdclk with additional
4629	 * constants, so for now we are using only table until this is
4630	 * clarified. Also this is the reason why crtc_state param is
4631	 * still here - we will need it once those additional constraints
4632	 * pop up.
4633	 */
4634	return compute_dbuf_slices(pipe, active_pipes, icl_allowed_dbufs);
4635}
4636
4637static u8 tgl_compute_dbuf_slices(enum pipe pipe, u8 active_pipes)
4638{
4639	return compute_dbuf_slices(pipe, active_pipes, tgl_allowed_dbufs);
4640}
4641
4642static u8 skl_compute_dbuf_slices(const struct intel_crtc_state *crtc_state,
4643				  u8 active_pipes)
4644{
4645	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
4646	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
4647	enum pipe pipe = crtc->pipe;
4648
4649	if (IS_GEN(dev_priv, 12))
4650		return tgl_compute_dbuf_slices(pipe, active_pipes);
4651	else if (IS_GEN(dev_priv, 11))
4652		return icl_compute_dbuf_slices(pipe, active_pipes);
4653	/*
4654	 * For anything else just return one slice yet.
4655	 * Should be extended for other platforms.
4656	 */
4657	return active_pipes & BIT(pipe) ? BIT(DBUF_S1) : 0;
4658}
4659
4660static u64
4661skl_plane_relative_data_rate(const struct intel_crtc_state *crtc_state,
4662			     const struct intel_plane_state *plane_state,
4663			     int color_plane)
4664{
4665	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
4666	const struct drm_framebuffer *fb = plane_state->hw.fb;
4667	u32 data_rate;
4668	u32 width = 0, height = 0;
4669	uint_fixed_16_16_t down_scale_amount;
4670	u64 rate;
4671
4672	if (!plane_state->uapi.visible)
4673		return 0;
4674
4675	if (plane->id == PLANE_CURSOR)
4676		return 0;
4677
4678	if (color_plane == 1 &&
4679	    !intel_format_info_is_yuv_semiplanar(fb->format, fb->modifier))
4680		return 0;
4681
4682	/*
4683	 * Src coordinates are already rotated by 270 degrees for
4684	 * the 90/270 degree plane rotation cases (to match the
4685	 * GTT mapping), hence no need to account for rotation here.
4686	 */
4687	width = drm_rect_width(&plane_state->uapi.src) >> 16;
4688	height = drm_rect_height(&plane_state->uapi.src) >> 16;
4689
4690	/* UV plane does 1/2 pixel sub-sampling */
4691	if (color_plane == 1) {
4692		width /= 2;
4693		height /= 2;
4694	}
4695
4696	data_rate = width * height;
4697
4698	down_scale_amount = skl_plane_downscale_amount(crtc_state, plane_state);
4699
4700	rate = mul_round_up_u32_fixed16(data_rate, down_scale_amount);
4701
4702	rate *= fb->format->cpp[color_plane];
4703	return rate;
4704}
4705
4706static u64
4707skl_get_total_relative_data_rate(struct intel_crtc_state *crtc_state,
4708				 u64 *plane_data_rate,
4709				 u64 *uv_plane_data_rate)
4710{
4711	struct intel_plane *plane;
4712	const struct intel_plane_state *plane_state;
4713	u64 total_data_rate = 0;
4714
4715	/* Calculate and cache data rate for each plane */
4716	intel_atomic_crtc_state_for_each_plane_state(plane, plane_state, crtc_state) {
4717		enum plane_id plane_id = plane->id;
4718		u64 rate;
4719
4720		/* packed/y */
4721		rate = skl_plane_relative_data_rate(crtc_state, plane_state, 0);
4722		plane_data_rate[plane_id] = rate;
4723		total_data_rate += rate;
4724
4725		/* uv-plane */
4726		rate = skl_plane_relative_data_rate(crtc_state, plane_state, 1);
4727		uv_plane_data_rate[plane_id] = rate;
4728		total_data_rate += rate;
4729	}
4730
4731	return total_data_rate;
4732}
4733
4734static u64
4735icl_get_total_relative_data_rate(struct intel_crtc_state *crtc_state,
4736				 u64 *plane_data_rate)
4737{
4738	struct intel_plane *plane;
4739	const struct intel_plane_state *plane_state;
4740	u64 total_data_rate = 0;
4741
4742	/* Calculate and cache data rate for each plane */
4743	intel_atomic_crtc_state_for_each_plane_state(plane, plane_state, crtc_state) {
4744		enum plane_id plane_id = plane->id;
4745		u64 rate;
4746
4747		if (!plane_state->planar_linked_plane) {
4748			rate = skl_plane_relative_data_rate(crtc_state, plane_state, 0);
4749			plane_data_rate[plane_id] = rate;
4750			total_data_rate += rate;
4751		} else {
4752			enum plane_id y_plane_id;
4753
4754			/*
4755			 * The slave plane might not iterate in
4756			 * intel_atomic_crtc_state_for_each_plane_state(),
4757			 * and needs the master plane state which may be
4758			 * NULL if we try get_new_plane_state(), so we
4759			 * always calculate from the master.
4760			 */
4761			if (plane_state->planar_slave)
4762				continue;
4763
4764			/* Y plane rate is calculated on the slave */
4765			rate = skl_plane_relative_data_rate(crtc_state, plane_state, 0);
4766			y_plane_id = plane_state->planar_linked_plane->id;
4767			plane_data_rate[y_plane_id] = rate;
4768			total_data_rate += rate;
4769
4770			rate = skl_plane_relative_data_rate(crtc_state, plane_state, 1);
4771			plane_data_rate[plane_id] = rate;
4772			total_data_rate += rate;
4773		}
4774	}
4775
4776	return total_data_rate;
4777}
4778
4779static const struct skl_wm_level *
4780skl_plane_wm_level(const struct intel_crtc_state *crtc_state,
4781		   enum plane_id plane_id,
4782		   int level)
4783{
4784	const struct skl_pipe_wm *pipe_wm = &crtc_state->wm.skl.optimal;
4785	const struct skl_plane_wm *wm = &pipe_wm->planes[plane_id];
4786
4787	if (level == 0 && pipe_wm->use_sagv_wm)
4788		return &wm->sagv_wm0;
4789
4790	return &wm->wm[level];
4791}
4792
4793static int
4794skl_allocate_pipe_ddb(struct intel_crtc_state *crtc_state)
4795{
4796	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
4797	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
4798	struct skl_ddb_entry *alloc = &crtc_state->wm.skl.ddb;
4799	u16 alloc_size, start = 0;
4800	u16 total[I915_MAX_PLANES] = {};
4801	u16 uv_total[I915_MAX_PLANES] = {};
4802	u64 total_data_rate;
4803	enum plane_id plane_id;
4804	int num_active;
4805	u64 plane_data_rate[I915_MAX_PLANES] = {};
4806	u64 uv_plane_data_rate[I915_MAX_PLANES] = {};
4807	u32 blocks;
4808	int level;
4809	int ret;
4810
4811	/* Clear the partitioning for disabled planes. */
4812	memset(crtc_state->wm.skl.plane_ddb_y, 0, sizeof(crtc_state->wm.skl.plane_ddb_y));
4813	memset(crtc_state->wm.skl.plane_ddb_uv, 0, sizeof(crtc_state->wm.skl.plane_ddb_uv));
4814
4815	if (!crtc_state->hw.active) {
4816		struct intel_atomic_state *state =
4817			to_intel_atomic_state(crtc_state->uapi.state);
4818		struct intel_dbuf_state *new_dbuf_state =
4819			intel_atomic_get_new_dbuf_state(state);
4820		const struct intel_dbuf_state *old_dbuf_state =
4821			intel_atomic_get_old_dbuf_state(state);
4822
4823		/*
4824		 * FIXME hack to make sure we compute this sensibly when
4825		 * turning off all the pipes. Otherwise we leave it at
4826		 * whatever we had previously, and then runtime PM will
4827		 * mess it up by turning off all but S1. Remove this
4828		 * once the dbuf state computation flow becomes sane.
4829		 */
4830		if (new_dbuf_state->active_pipes == 0) {
4831			new_dbuf_state->enabled_slices = BIT(DBUF_S1);
4832
4833			if (old_dbuf_state->enabled_slices != new_dbuf_state->enabled_slices) {
4834				ret = intel_atomic_serialize_global_state(&new_dbuf_state->base);
4835				if (ret)
4836					return ret;
4837			}
4838		}
4839
4840		alloc->start = alloc->end = 0;
4841		return 0;
4842	}
4843
4844	if (INTEL_GEN(dev_priv) >= 11)
4845		total_data_rate =
4846			icl_get_total_relative_data_rate(crtc_state,
4847							 plane_data_rate);
4848	else
4849		total_data_rate =
4850			skl_get_total_relative_data_rate(crtc_state,
4851							 plane_data_rate,
4852							 uv_plane_data_rate);
4853
4854	ret = skl_ddb_get_pipe_allocation_limits(dev_priv, crtc_state,
4855						 total_data_rate,
4856						 alloc, &num_active);
4857	if (ret)
4858		return ret;
4859
4860	alloc_size = skl_ddb_entry_size(alloc);
4861	if (alloc_size == 0)
4862		return 0;
4863
4864	/* Allocate fixed number of blocks for cursor. */
4865	total[PLANE_CURSOR] = skl_cursor_allocation(crtc_state, num_active);
4866	alloc_size -= total[PLANE_CURSOR];
4867	crtc_state->wm.skl.plane_ddb_y[PLANE_CURSOR].start =
4868		alloc->end - total[PLANE_CURSOR];
4869	crtc_state->wm.skl.plane_ddb_y[PLANE_CURSOR].end = alloc->end;
4870
4871	if (total_data_rate == 0)
4872		return 0;
4873
4874	/*
4875	 * Find the highest watermark level for which we can satisfy the block
4876	 * requirement of active planes.
4877	 */
4878	for (level = ilk_wm_max_level(dev_priv); level >= 0; level--) {
4879		blocks = 0;
4880		for_each_plane_id_on_crtc(crtc, plane_id) {
4881			const struct skl_plane_wm *wm =
4882				&crtc_state->wm.skl.optimal.planes[plane_id];
4883
4884			if (plane_id == PLANE_CURSOR) {
4885				if (wm->wm[level].min_ddb_alloc > total[PLANE_CURSOR]) {
4886					drm_WARN_ON(&dev_priv->drm,
4887						    wm->wm[level].min_ddb_alloc != U16_MAX);
4888					blocks = U32_MAX;
4889					break;
4890				}
4891				continue;
4892			}
4893
4894			blocks += wm->wm[level].min_ddb_alloc;
4895			blocks += wm->uv_wm[level].min_ddb_alloc;
4896		}
4897
4898		if (blocks <= alloc_size) {
4899			alloc_size -= blocks;
4900			break;
4901		}
4902	}
4903
4904	if (level < 0) {
4905		drm_dbg_kms(&dev_priv->drm,
4906			    "Requested display configuration exceeds system DDB limitations");
4907		drm_dbg_kms(&dev_priv->drm, "minimum required %d/%d\n",
4908			    blocks, alloc_size);
4909		return -EINVAL;
4910	}
4911
4912	/*
4913	 * Grant each plane the blocks it requires at the highest achievable
4914	 * watermark level, plus an extra share of the leftover blocks
4915	 * proportional to its relative data rate.
4916	 */
4917	for_each_plane_id_on_crtc(crtc, plane_id) {
4918		const struct skl_plane_wm *wm =
4919			&crtc_state->wm.skl.optimal.planes[plane_id];
4920		u64 rate;
4921		u16 extra;
4922
4923		if (plane_id == PLANE_CURSOR)
4924			continue;
4925
4926		/*
4927		 * We've accounted for all active planes; remaining planes are
4928		 * all disabled.
4929		 */
4930		if (total_data_rate == 0)
4931			break;
4932
4933		rate = plane_data_rate[plane_id];
4934		extra = min_t(u16, alloc_size,
4935			      DIV64_U64_ROUND_UP(alloc_size * rate,
4936						 total_data_rate));
4937		total[plane_id] = wm->wm[level].min_ddb_alloc + extra;
4938		alloc_size -= extra;
4939		total_data_rate -= rate;
4940
4941		if (total_data_rate == 0)
4942			break;
4943
4944		rate = uv_plane_data_rate[plane_id];
4945		extra = min_t(u16, alloc_size,
4946			      DIV64_U64_ROUND_UP(alloc_size * rate,
4947						 total_data_rate));
4948		uv_total[plane_id] = wm->uv_wm[level].min_ddb_alloc + extra;
4949		alloc_size -= extra;
4950		total_data_rate -= rate;
4951	}
4952	drm_WARN_ON(&dev_priv->drm, alloc_size != 0 || total_data_rate != 0);
4953
4954	/* Set the actual DDB start/end points for each plane */
4955	start = alloc->start;
4956	for_each_plane_id_on_crtc(crtc, plane_id) {
4957		struct skl_ddb_entry *plane_alloc =
4958			&crtc_state->wm.skl.plane_ddb_y[plane_id];
4959		struct skl_ddb_entry *uv_plane_alloc =
4960			&crtc_state->wm.skl.plane_ddb_uv[plane_id];
4961
4962		if (plane_id == PLANE_CURSOR)
4963			continue;
4964
4965		/* Gen11+ uses a separate plane for UV watermarks */
4966		drm_WARN_ON(&dev_priv->drm,
4967			    INTEL_GEN(dev_priv) >= 11 && uv_total[plane_id]);
4968
4969		/* Leave disabled planes at (0,0) */
4970		if (total[plane_id]) {
4971			plane_alloc->start = start;
4972			start += total[plane_id];
4973			plane_alloc->end = start;
4974		}
4975
4976		if (uv_total[plane_id]) {
4977			uv_plane_alloc->start = start;
4978			start += uv_total[plane_id];
4979			uv_plane_alloc->end = start;
4980		}
4981	}
4982
4983	/*
4984	 * When we calculated watermark values we didn't know how high
4985	 * of a level we'd actually be able to hit, so we just marked
4986	 * all levels as "enabled."  Go back now and disable the ones
4987	 * that aren't actually possible.
4988	 */
4989	for (level++; level <= ilk_wm_max_level(dev_priv); level++) {
4990		for_each_plane_id_on_crtc(crtc, plane_id) {
4991			struct skl_plane_wm *wm =
4992				&crtc_state->wm.skl.optimal.planes[plane_id];
4993
4994			/*
4995			 * We only disable the watermarks for each plane if
4996			 * they exceed the ddb allocation of said plane. This
4997			 * is done so that we don't end up touching cursor
4998			 * watermarks needlessly when some other plane reduces
4999			 * our max possible watermark level.
5000			 *
5001			 * Bspec has this to say about the PLANE_WM enable bit:
5002			 * "All the watermarks at this level for all enabled
5003			 *  planes must be enabled before the level will be used."
5004			 * So this is actually safe to do.
5005			 */
5006			if (wm->wm[level].min_ddb_alloc > total[plane_id] ||
5007			    wm->uv_wm[level].min_ddb_alloc > uv_total[plane_id])
5008				memset(&wm->wm[level], 0, sizeof(wm->wm[level]));
5009
5010			/*
5011			 * Wa_1408961008:icl, ehl
5012			 * Underruns with WM1+ disabled
5013			 */
5014			if (IS_GEN(dev_priv, 11) &&
5015			    level == 1 && wm->wm[0].plane_en) {
5016				wm->wm[level].plane_res_b = wm->wm[0].plane_res_b;
5017				wm->wm[level].plane_res_l = wm->wm[0].plane_res_l;
5018				wm->wm[level].ignore_lines = wm->wm[0].ignore_lines;
5019			}
5020		}
5021	}
5022
5023	/*
5024	 * Go back and disable the transition watermark if it turns out we
5025	 * don't have enough DDB blocks for it.
5026	 */
5027	for_each_plane_id_on_crtc(crtc, plane_id) {
5028		struct skl_plane_wm *wm =
5029			&crtc_state->wm.skl.optimal.planes[plane_id];
5030
5031		if (wm->trans_wm.plane_res_b >= total[plane_id])
5032			memset(&wm->trans_wm, 0, sizeof(wm->trans_wm));
5033	}
5034
5035	return 0;
5036}
5037
5038/*
5039 * The max latency should be 257 (max the punit can code is 255 and we add 2us
5040 * for the read latency) and cpp should always be <= 8, so that
5041 * should allow pixel_rate up to ~2 GHz which seems sufficient since max
5042 * 2xcdclk is 1350 MHz and the pixel rate should never exceed that.
5043*/
5044static uint_fixed_16_16_t
5045skl_wm_method1(const struct drm_i915_private *dev_priv, u32 pixel_rate,
5046	       u8 cpp, u32 latency, u32 dbuf_block_size)
5047{
5048	u32 wm_intermediate_val;
5049	uint_fixed_16_16_t ret;
5050
5051	if (latency == 0)
5052		return FP_16_16_MAX;
5053
5054	wm_intermediate_val = latency * pixel_rate * cpp;
5055	ret = div_fixed16(wm_intermediate_val, 1000 * dbuf_block_size);
5056
5057	if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv))
5058		ret = add_fixed16_u32(ret, 1);
5059
5060	return ret;
5061}
5062
5063static uint_fixed_16_16_t
5064skl_wm_method2(u32 pixel_rate, u32 pipe_htotal, u32 latency,
5065	       uint_fixed_16_16_t plane_blocks_per_line)
5066{
5067	u32 wm_intermediate_val;
5068	uint_fixed_16_16_t ret;
5069
5070	if (latency == 0)
5071		return FP_16_16_MAX;
5072
5073	wm_intermediate_val = latency * pixel_rate;
5074	wm_intermediate_val = DIV_ROUND_UP(wm_intermediate_val,
5075					   pipe_htotal * 1000);
5076	ret = mul_u32_fixed16(wm_intermediate_val, plane_blocks_per_line);
5077	return ret;
5078}
5079
5080static uint_fixed_16_16_t
5081intel_get_linetime_us(const struct intel_crtc_state *crtc_state)
5082{
5083	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5084	u32 pixel_rate;
5085	u32 crtc_htotal;
5086	uint_fixed_16_16_t linetime_us;
5087
5088	if (!crtc_state->hw.active)
5089		return u32_to_fixed16(0);
5090
5091	pixel_rate = crtc_state->pixel_rate;
5092
5093	if (drm_WARN_ON(&dev_priv->drm, pixel_rate == 0))
5094		return u32_to_fixed16(0);
5095
5096	crtc_htotal = crtc_state->hw.adjusted_mode.crtc_htotal;
5097	linetime_us = div_fixed16(crtc_htotal * 1000, pixel_rate);
5098
5099	return linetime_us;
5100}
5101
5102static u32
5103skl_adjusted_plane_pixel_rate(const struct intel_crtc_state *crtc_state,
5104			      const struct intel_plane_state *plane_state)
5105{
5106	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5107	u64 adjusted_pixel_rate;
5108	uint_fixed_16_16_t downscale_amount;
5109
5110	/* Shouldn't reach here on disabled planes... */
5111	if (drm_WARN_ON(&dev_priv->drm,
5112			!intel_wm_plane_visible(crtc_state, plane_state)))
5113		return 0;
5114
5115	/*
5116	 * Adjusted plane pixel rate is just the pipe's adjusted pixel rate
5117	 * with additional adjustments for plane-specific scaling.
5118	 */
5119	adjusted_pixel_rate = crtc_state->pixel_rate;
5120	downscale_amount = skl_plane_downscale_amount(crtc_state, plane_state);
5121
5122	return mul_round_up_u32_fixed16(adjusted_pixel_rate,
5123					    downscale_amount);
5124}
5125
5126static int
5127skl_compute_wm_params(const struct intel_crtc_state *crtc_state,
5128		      int width, const struct drm_format_info *format,
5129		      u64 modifier, unsigned int rotation,
5130		      u32 plane_pixel_rate, struct skl_wm_params *wp,
5131		      int color_plane)
5132{
5133	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
5134	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
5135	u32 interm_pbpl;
5136
5137	/* only planar format has two planes */
5138	if (color_plane == 1 &&
5139	    !intel_format_info_is_yuv_semiplanar(format, modifier)) {
5140		drm_dbg_kms(&dev_priv->drm,
5141			    "Non planar format have single plane\n");
5142		return -EINVAL;
5143	}
5144
5145	wp->y_tiled = modifier == I915_FORMAT_MOD_Y_TILED ||
5146		      modifier == I915_FORMAT_MOD_Yf_TILED ||
5147		      modifier == I915_FORMAT_MOD_Y_TILED_CCS ||
5148		      modifier == I915_FORMAT_MOD_Yf_TILED_CCS ||
5149		      modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS ||
5150		      modifier == I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS;
5151	wp->x_tiled = modifier == I915_FORMAT_MOD_X_TILED;
5152	wp->rc_surface = modifier == I915_FORMAT_MOD_Y_TILED_CCS ||
5153			 modifier == I915_FORMAT_MOD_Yf_TILED_CCS ||
5154			 modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS ||
5155			 modifier == I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS;
5156	wp->is_planar = intel_format_info_is_yuv_semiplanar(format, modifier);
5157
5158	wp->width = width;
5159	if (color_plane == 1 && wp->is_planar)
5160		wp->width /= 2;
5161
5162	wp->cpp = format->cpp[color_plane];
5163	wp->plane_pixel_rate = plane_pixel_rate;
5164
5165	if (INTEL_GEN(dev_priv) >= 11 &&
5166	    modifier == I915_FORMAT_MOD_Yf_TILED  && wp->cpp == 1)
5167		wp->dbuf_block_size = 256;
5168	else
5169		wp->dbuf_block_size = 512;
5170
5171	if (drm_rotation_90_or_270(rotation)) {
5172		switch (wp->cpp) {
5173		case 1:
5174			wp->y_min_scanlines = 16;
5175			break;
5176		case 2:
5177			wp->y_min_scanlines = 8;
5178			break;
5179		case 4:
5180			wp->y_min_scanlines = 4;
5181			break;
5182		default:
5183			MISSING_CASE(wp->cpp);
5184			return -EINVAL;
5185		}
5186	} else {
5187		wp->y_min_scanlines = 4;
5188	}
5189
5190	if (skl_needs_memory_bw_wa(dev_priv))
5191		wp->y_min_scanlines *= 2;
5192
5193	wp->plane_bytes_per_line = wp->width * wp->cpp;
5194	if (wp->y_tiled) {
5195		interm_pbpl = DIV_ROUND_UP(wp->plane_bytes_per_line *
5196					   wp->y_min_scanlines,
5197					   wp->dbuf_block_size);
5198
5199		if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv))
5200			interm_pbpl++;
5201
5202		wp->plane_blocks_per_line = div_fixed16(interm_pbpl,
5203							wp->y_min_scanlines);
5204	} else {
5205		interm_pbpl = DIV_ROUND_UP(wp->plane_bytes_per_line,
5206					   wp->dbuf_block_size);
5207
5208		if (!wp->x_tiled ||
5209		    INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv))
5210			interm_pbpl++;
5211
5212		wp->plane_blocks_per_line = u32_to_fixed16(interm_pbpl);
5213	}
5214
5215	wp->y_tile_minimum = mul_u32_fixed16(wp->y_min_scanlines,
5216					     wp->plane_blocks_per_line);
5217
5218	wp->linetime_us = fixed16_to_u32_round_up(
5219					intel_get_linetime_us(crtc_state));
5220
5221	return 0;
5222}
5223
5224static int
5225skl_compute_plane_wm_params(const struct intel_crtc_state *crtc_state,
5226			    const struct intel_plane_state *plane_state,
5227			    struct skl_wm_params *wp, int color_plane)
5228{
5229	const struct drm_framebuffer *fb = plane_state->hw.fb;
5230	int width;
5231
5232	/*
5233	 * Src coordinates are already rotated by 270 degrees for
5234	 * the 90/270 degree plane rotation cases (to match the
5235	 * GTT mapping), hence no need to account for rotation here.
5236	 */
5237	width = drm_rect_width(&plane_state->uapi.src) >> 16;
5238
5239	return skl_compute_wm_params(crtc_state, width,
5240				     fb->format, fb->modifier,
5241				     plane_state->hw.rotation,
5242				     skl_adjusted_plane_pixel_rate(crtc_state, plane_state),
5243				     wp, color_plane);
5244}
5245
5246static bool skl_wm_has_lines(struct drm_i915_private *dev_priv, int level)
5247{
5248	if (INTEL_GEN(dev_priv) >= 10 || IS_GEMINILAKE(dev_priv))
5249		return true;
5250
5251	/* The number of lines are ignored for the level 0 watermark. */
5252	return level > 0;
5253}
5254
5255static void skl_compute_plane_wm(const struct intel_crtc_state *crtc_state,
5256				 int level,
5257				 unsigned int latency,
5258				 const struct skl_wm_params *wp,
5259				 const struct skl_wm_level *result_prev,
5260				 struct skl_wm_level *result /* out */)
5261{
5262	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5263	uint_fixed_16_16_t method1, method2;
5264	uint_fixed_16_16_t selected_result;
5265	u32 res_blocks, res_lines, min_ddb_alloc = 0;
5266
5267	if (latency == 0) {
5268		/* reject it */
5269		result->min_ddb_alloc = U16_MAX;
5270		return;
5271	}
5272
5273	/*
5274	 * WaIncreaseLatencyIPCEnabled: kbl,cfl
5275	 * Display WA #1141: kbl,cfl
5276	 */
5277	if ((IS_KABYLAKE(dev_priv) ||
5278	     IS_COFFEELAKE(dev_priv) ||
5279	     IS_COMETLAKE(dev_priv)) &&
5280	    dev_priv->ipc_enabled)
5281		latency += 4;
5282
5283	if (skl_needs_memory_bw_wa(dev_priv) && wp->x_tiled)
5284		latency += 15;
5285
5286	method1 = skl_wm_method1(dev_priv, wp->plane_pixel_rate,
5287				 wp->cpp, latency, wp->dbuf_block_size);
5288	method2 = skl_wm_method2(wp->plane_pixel_rate,
5289				 crtc_state->hw.adjusted_mode.crtc_htotal,
5290				 latency,
5291				 wp->plane_blocks_per_line);
5292
5293	if (wp->y_tiled) {
5294		selected_result = max_fixed16(method2, wp->y_tile_minimum);
5295	} else {
5296		if ((wp->cpp * crtc_state->hw.adjusted_mode.crtc_htotal /
5297		     wp->dbuf_block_size < 1) &&
5298		     (wp->plane_bytes_per_line / wp->dbuf_block_size < 1)) {
5299			selected_result = method2;
5300		} else if (latency >= wp->linetime_us) {
5301			if (IS_GEN(dev_priv, 9) &&
5302			    !IS_GEMINILAKE(dev_priv))
5303				selected_result = min_fixed16(method1, method2);
5304			else
5305				selected_result = method2;
5306		} else {
5307			selected_result = method1;
5308		}
5309	}
5310
5311	res_blocks = fixed16_to_u32_round_up(selected_result) + 1;
5312	res_lines = div_round_up_fixed16(selected_result,
5313					 wp->plane_blocks_per_line);
5314
5315	if (IS_GEN9_BC(dev_priv) || IS_BROXTON(dev_priv)) {
5316		/* Display WA #1125: skl,bxt,kbl */
5317		if (level == 0 && wp->rc_surface)
5318			res_blocks +=
5319				fixed16_to_u32_round_up(wp->y_tile_minimum);
5320
5321		/* Display WA #1126: skl,bxt,kbl */
5322		if (level >= 1 && level <= 7) {
5323			if (wp->y_tiled) {
5324				res_blocks +=
5325				    fixed16_to_u32_round_up(wp->y_tile_minimum);
5326				res_lines += wp->y_min_scanlines;
5327			} else {
5328				res_blocks++;
5329			}
5330
5331			/*
5332			 * Make sure result blocks for higher latency levels are
5333			 * atleast as high as level below the current level.
5334			 * Assumption in DDB algorithm optimization for special
5335			 * cases. Also covers Display WA #1125 for RC.
5336			 */
5337			if (result_prev->plane_res_b > res_blocks)
5338				res_blocks = result_prev->plane_res_b;
5339		}
5340	}
5341
5342	if (INTEL_GEN(dev_priv) >= 11) {
5343		if (wp->y_tiled) {
5344			int extra_lines;
5345
5346			if (res_lines % wp->y_min_scanlines == 0)
5347				extra_lines = wp->y_min_scanlines;
5348			else
5349				extra_lines = wp->y_min_scanlines * 2 -
5350					res_lines % wp->y_min_scanlines;
5351
5352			min_ddb_alloc = mul_round_up_u32_fixed16(res_lines + extra_lines,
5353								 wp->plane_blocks_per_line);
5354		} else {
5355			min_ddb_alloc = res_blocks +
5356				DIV_ROUND_UP(res_blocks, 10);
5357		}
5358	}
5359
5360	if (!skl_wm_has_lines(dev_priv, level))
5361		res_lines = 0;
5362
5363	if (res_lines > 31) {
5364		/* reject it */
5365		result->min_ddb_alloc = U16_MAX;
5366		return;
5367	}
5368
5369	/*
5370	 * If res_lines is valid, assume we can use this watermark level
5371	 * for now.  We'll come back and disable it after we calculate the
5372	 * DDB allocation if it turns out we don't actually have enough
5373	 * blocks to satisfy it.
5374	 */
5375	result->plane_res_b = res_blocks;
5376	result->plane_res_l = res_lines;
5377	/* Bspec says: value >= plane ddb allocation -> invalid, hence the +1 here */
5378	result->min_ddb_alloc = max(min_ddb_alloc, res_blocks) + 1;
5379	result->plane_en = true;
5380}
5381
5382static void
5383skl_compute_wm_levels(const struct intel_crtc_state *crtc_state,
5384		      const struct skl_wm_params *wm_params,
5385		      struct skl_wm_level *levels)
5386{
5387	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5388	int level, max_level = ilk_wm_max_level(dev_priv);
5389	struct skl_wm_level *result_prev = &levels[0];
5390
5391	for (level = 0; level <= max_level; level++) {
5392		struct skl_wm_level *result = &levels[level];
5393		unsigned int latency = dev_priv->wm.skl_latency[level];
5394
5395		skl_compute_plane_wm(crtc_state, level, latency,
5396				     wm_params, result_prev, result);
5397
5398		result_prev = result;
5399	}
5400}
5401
5402static void tgl_compute_sagv_wm(const struct intel_crtc_state *crtc_state,
5403				const struct skl_wm_params *wm_params,
5404				struct skl_plane_wm *plane_wm)
5405{
5406	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5407	struct skl_wm_level *sagv_wm = &plane_wm->sagv_wm0;
5408	struct skl_wm_level *levels = plane_wm->wm;
5409	unsigned int latency = dev_priv->wm.skl_latency[0] + dev_priv->sagv_block_time_us;
5410
5411	skl_compute_plane_wm(crtc_state, 0, latency,
5412			     wm_params, &levels[0],
5413			     sagv_wm);
5414}
5415
5416static void skl_compute_transition_wm(const struct intel_crtc_state *crtc_state,
5417				      const struct skl_wm_params *wp,
5418				      struct skl_plane_wm *wm)
5419{
5420	struct drm_device *dev = crtc_state->uapi.crtc->dev;
5421	const struct drm_i915_private *dev_priv = to_i915(dev);
5422	u16 trans_min, trans_amount, trans_y_tile_min;
5423	u16 wm0_sel_res_b, trans_offset_b, res_blocks;
5424
5425	/* Transition WM don't make any sense if ipc is disabled */
5426	if (!dev_priv->ipc_enabled)
5427		return;
5428
5429	/*
5430	 * WaDisableTWM:skl,kbl,cfl,bxt
5431	 * Transition WM are not recommended by HW team for GEN9
5432	 */
5433	if (IS_GEN9_BC(dev_priv) || IS_BROXTON(dev_priv))
5434		return;
5435
5436	if (INTEL_GEN(dev_priv) >= 11)
5437		trans_min = 4;
5438	else
5439		trans_min = 14;
5440
5441	/* Display WA #1140: glk,cnl */
5442	if (IS_CANNONLAKE(dev_priv) || IS_GEMINILAKE(dev_priv))
5443		trans_amount = 0;
5444	else
5445		trans_amount = 10; /* This is configurable amount */
5446
5447	trans_offset_b = trans_min + trans_amount;
5448
5449	/*
5450	 * The spec asks for Selected Result Blocks for wm0 (the real value),
5451	 * not Result Blocks (the integer value). Pay attention to the capital
5452	 * letters. The value wm_l0->plane_res_b is actually Result Blocks, but
5453	 * since Result Blocks is the ceiling of Selected Result Blocks plus 1,
5454	 * and since we later will have to get the ceiling of the sum in the
5455	 * transition watermarks calculation, we can just pretend Selected
5456	 * Result Blocks is Result Blocks minus 1 and it should work for the
5457	 * current platforms.
5458	 */
5459	wm0_sel_res_b = wm->wm[0].plane_res_b - 1;
5460
5461	if (wp->y_tiled) {
5462		trans_y_tile_min =
5463			(u16)mul_round_up_u32_fixed16(2, wp->y_tile_minimum);
5464		res_blocks = max(wm0_sel_res_b, trans_y_tile_min) +
5465				trans_offset_b;
5466	} else {
5467		res_blocks = wm0_sel_res_b + trans_offset_b;
5468	}
5469
5470	/*
5471	 * Just assume we can enable the transition watermark.  After
5472	 * computing the DDB we'll come back and disable it if that
5473	 * assumption turns out to be false.
5474	 */
5475	wm->trans_wm.plane_res_b = res_blocks + 1;
5476	wm->trans_wm.plane_en = true;
5477}
5478
5479static int skl_build_plane_wm_single(struct intel_crtc_state *crtc_state,
5480				     const struct intel_plane_state *plane_state,
5481				     enum plane_id plane_id, int color_plane)
5482{
5483	struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc);
5484	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
5485	struct skl_plane_wm *wm = &crtc_state->wm.skl.optimal.planes[plane_id];
5486	struct skl_wm_params wm_params;
5487	int ret;
5488
5489	ret = skl_compute_plane_wm_params(crtc_state, plane_state,
5490					  &wm_params, color_plane);
5491	if (ret)
5492		return ret;
5493
5494	skl_compute_wm_levels(crtc_state, &wm_params, wm->wm);
5495
5496	if (INTEL_GEN(dev_priv) >= 12)
5497		tgl_compute_sagv_wm(crtc_state, &wm_params, wm);
5498
5499	skl_compute_transition_wm(crtc_state, &wm_params, wm);
5500
5501	return 0;
5502}
5503
5504static int skl_build_plane_wm_uv(struct intel_crtc_state *crtc_state,
5505				 const struct intel_plane_state *plane_state,
5506				 enum plane_id plane_id)
5507{
5508	struct skl_plane_wm *wm = &crtc_state->wm.skl.optimal.planes[plane_id];
5509	struct skl_wm_params wm_params;
5510	int ret;
5511
5512	wm->is_planar = true;
5513
5514	/* uv plane watermarks must also be validated for NV12/Planar */
5515	ret = skl_compute_plane_wm_params(crtc_state, plane_state,
5516					  &wm_params, 1);
5517	if (ret)
5518		return ret;
5519
5520	skl_compute_wm_levels(crtc_state, &wm_params, wm->uv_wm);
5521
5522	return 0;
5523}
5524
5525static int skl_build_plane_wm(struct intel_crtc_state *crtc_state,
5526			      const struct intel_plane_state *plane_state)
5527{
5528	struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
5529	const struct drm_framebuffer *fb = plane_state->hw.fb;
5530	enum plane_id plane_id = plane->id;
5531	int ret;
5532
5533	if (!intel_wm_plane_visible(crtc_state, plane_state))
5534		return 0;
5535
5536	ret = skl_build_plane_wm_single(crtc_state, plane_state,
5537					plane_id, 0);
5538	if (ret)
5539		return ret;
5540
5541	if (fb->format->is_yuv && fb->format->num_planes > 1) {
5542		ret = skl_build_plane_wm_uv(crtc_state, plane_state,
5543					    plane_id);
5544		if (ret)
5545			return ret;
5546	}
5547
5548	return 0;
5549}
5550
5551static int icl_build_plane_wm(struct intel_crtc_state *crtc_state,
5552			      const struct intel_plane_state *plane_state)
5553{
5554	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5555	enum plane_id plane_id = to_intel_plane(plane_state->uapi.plane)->id;
5556	int ret;
5557
5558	/* Watermarks calculated in master */
5559	if (plane_state->planar_slave)
5560		return 0;
5561
5562	if (plane_state->planar_linked_plane) {
5563		const struct drm_framebuffer *fb = plane_state->hw.fb;
5564		enum plane_id y_plane_id = plane_state->planar_linked_plane->id;
5565
5566		drm_WARN_ON(&dev_priv->drm,
5567			    !intel_wm_plane_visible(crtc_state, plane_state));
5568		drm_WARN_ON(&dev_priv->drm, !fb->format->is_yuv ||
5569			    fb->format->num_planes == 1);
5570
5571		ret = skl_build_plane_wm_single(crtc_state, plane_state,
5572						y_plane_id, 0);
5573		if (ret)
5574			return ret;
5575
5576		ret = skl_build_plane_wm_single(crtc_state, plane_state,
5577						plane_id, 1);
5578		if (ret)
5579			return ret;
5580	} else if (intel_wm_plane_visible(crtc_state, plane_state)) {
5581		ret = skl_build_plane_wm_single(crtc_state, plane_state,
5582						plane_id, 0);
5583		if (ret)
5584			return ret;
5585	}
5586
5587	return 0;
5588}
5589
5590static int skl_build_pipe_wm(struct intel_crtc_state *crtc_state)
5591{
5592	struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev);
5593	struct skl_pipe_wm *pipe_wm = &crtc_state->wm.skl.optimal;
5594	struct intel_plane *plane;
5595	const struct intel_plane_state *plane_state;
5596	int ret;
5597
5598	/*
5599	 * We'll only calculate watermarks for planes that are actually
5600	 * enabled, so make sure all other planes are set as disabled.
5601	 */
5602	memset(pipe_wm->planes, 0, sizeof(pipe_wm->planes));
5603
5604	intel_atomic_crtc_state_for_each_plane_state(plane, plane_state,
5605						     crtc_state) {
5606
5607		if (INTEL_GEN(dev_priv) >= 11)
5608			ret = icl_build_plane_wm(crtc_state, plane_state);
5609		else
5610			ret = skl_build_plane_wm(crtc_state, plane_state);
5611		if (ret)
5612			return ret;
5613	}
5614
5615	return 0;
5616}
5617
5618static void skl_ddb_entry_write(struct drm_i915_private *dev_priv,
5619				i915_reg_t reg,
5620				const struct skl_ddb_entry *entry)
5621{
5622	if (entry->end)
5623		intel_de_write_fw(dev_priv, reg,
5624				  (entry->end - 1) << 16 | entry->start);
5625	else
5626		intel_de_write_fw(dev_priv, reg, 0);
5627}
5628
5629static void skl_write_wm_level(struct drm_i915_private *dev_priv,
5630			       i915_reg_t reg,
5631			       const struct skl_wm_level *level)
5632{
5633	u32 val = 0;
5634
5635	if (level->plane_en)
5636		val |= PLANE_WM_EN;
5637	if (level->ignore_lines)
5638		val |= PLANE_WM_IGNORE_LINES;
5639	val |= level->plane_res_b;
5640	val |= level->plane_res_l << PLANE_WM_LINES_SHIFT;
5641
5642	intel_de_write_fw(dev_priv, reg, val);
5643}
5644
5645void skl_write_plane_wm(struct intel_plane *plane,
5646			const struct intel_crtc_state *crtc_state)
5647{
5648	struct drm_i915_private *dev_priv = to_i915(plane->base.dev);
5649	int level, max_level = ilk_wm_max_level(dev_priv);
5650	enum plane_id plane_id = plane->id;
5651	enum pipe pipe = plane->pipe;
5652	const struct skl_plane_wm *wm =
5653		&crtc_state->wm.skl.optimal.planes[plane_id];
5654	const struct skl_ddb_entry *ddb_y =
5655		&crtc_state->wm.skl.plane_ddb_y[plane_id];
5656	const struct skl_ddb_entry *ddb_uv =
5657		&crtc_state->wm.skl.plane_ddb_uv[plane_id];
5658
5659	for (level = 0; level <= max_level; level++) {
5660		const struct skl_wm_level *wm_level;
5661
5662		wm_level = skl_plane_wm_level(crtc_state, plane_id, level);
5663
5664		skl_write_wm_level(dev_priv, PLANE_WM(pipe, plane_id, level),
5665				   wm_level);
5666	}
5667	skl_write_wm_level(dev_priv, PLANE_WM_TRANS(pipe, plane_id),
5668			   &wm->trans_wm);
5669
5670	if (INTEL_GEN(dev_priv) >= 11) {
5671		skl_ddb_entry_write(dev_priv,
5672				    PLANE_BUF_CFG(pipe, plane_id), ddb_y);
5673		return;
5674	}
5675
5676	if (wm->is_planar)
5677		swap(ddb_y, ddb_uv);
5678
5679	skl_ddb_entry_write(dev_priv,
5680			    PLANE_BUF_CFG(pipe, plane_id), ddb_y);
5681	skl_ddb_entry_write(dev_priv,
5682			    PLANE_NV12_BUF_CFG(pipe, plane_id), ddb_uv);
5683}
5684
5685void skl_write_cursor_wm(struct intel_plane *plane,
5686			 const struct intel_crtc_state *crtc_state)
5687{
5688	struct drm_i915_private *dev_priv = to_i915(plane->base.dev);
5689	int level, max_level = ilk_wm_max_level(dev_priv);
5690	enum plane_id plane_id = plane->id;
5691	enum pipe pipe = plane->pipe;
5692	const struct skl_plane_wm *wm =
5693		&crtc_state->wm.skl.optimal.planes[plane_id];
5694	const struct skl_ddb_entry *ddb =
5695		&crtc_state->wm.skl.plane_ddb_y[plane_id];
5696
5697	for (level = 0; level <= max_level; level++) {
5698		const struct skl_wm_level *wm_level;
5699
5700		wm_level = skl_plane_wm_level(crtc_state, plane_id, level);
5701
5702		skl_write_wm_level(dev_priv, CUR_WM(pipe, level),
5703				   wm_level);
5704	}
5705	skl_write_wm_level(dev_priv, CUR_WM_TRANS(pipe), &wm->trans_wm);
5706
5707	skl_ddb_entry_write(dev_priv, CUR_BUF_CFG(pipe), ddb);
5708}
5709
5710bool skl_wm_level_equals(const struct skl_wm_level *l1,
5711			 const struct skl_wm_level *l2)
5712{
5713	return l1->plane_en == l2->plane_en &&
5714		l1->ignore_lines == l2->ignore_lines &&
5715		l1->plane_res_l == l2->plane_res_l &&
5716		l1->plane_res_b == l2->plane_res_b;
5717}
5718
5719static bool skl_plane_wm_equals(struct drm_i915_private *dev_priv,
5720				const struct skl_plane_wm *wm1,
5721				const struct skl_plane_wm *wm2)
5722{
5723	int level, max_level = ilk_wm_max_level(dev_priv);
5724
5725	for (level = 0; level <= max_level; level++) {
5726		/*
5727		 * We don't check uv_wm as the hardware doesn't actually
5728		 * use it. It only gets used for calculating the required
5729		 * ddb allocation.
5730		 */
5731		if (!skl_wm_level_equals(&wm1->wm[level], &wm2->wm[level]))
5732			return false;
5733	}
5734
5735	return skl_wm_level_equals(&wm1->trans_wm, &wm2->trans_wm);
5736}
5737
5738static bool skl_ddb_entries_overlap(const struct skl_ddb_entry *a,
5739				    const struct skl_ddb_entry *b)
5740{
5741	return a->start < b->end && b->start < a->end;
5742}
5743
5744bool skl_ddb_allocation_overlaps(const struct skl_ddb_entry *ddb,
5745				 const struct skl_ddb_entry *entries,
5746				 int num_entries, int ignore_idx)
5747{
5748	int i;
5749
5750	for (i = 0; i < num_entries; i++) {
5751		if (i != ignore_idx &&
5752		    skl_ddb_entries_overlap(ddb, &entries[i]))
5753			return true;
5754	}
5755
5756	return false;
5757}
5758
5759static int
5760skl_ddb_add_affected_planes(const struct intel_crtc_state *old_crtc_state,
5761			    struct intel_crtc_state *new_crtc_state)
5762{
5763	struct intel_atomic_state *state = to_intel_atomic_state(new_crtc_state->uapi.state);
5764	struct intel_crtc *crtc = to_intel_crtc(new_crtc_state->uapi.crtc);
5765	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
5766	struct intel_plane *plane;
5767
5768	for_each_intel_plane_on_crtc(&dev_priv->drm, crtc, plane) {
5769		struct intel_plane_state *plane_state;
5770		enum plane_id plane_id = plane->id;
5771
5772		if (skl_ddb_entry_equal(&old_crtc_state->wm.skl.plane_ddb_y[plane_id],
5773					&new_crtc_state->wm.skl.plane_ddb_y[plane_id]) &&
5774		    skl_ddb_entry_equal(&old_crtc_state->wm.skl.plane_ddb_uv[plane_id],
5775					&new_crtc_state->wm.skl.plane_ddb_uv[plane_id]))
5776			continue;
5777
5778		plane_state = intel_atomic_get_plane_state(state, plane);
5779		if (IS_ERR(plane_state))
5780			return PTR_ERR(plane_state);
5781
5782		new_crtc_state->update_planes |= BIT(plane_id);
5783	}
5784
5785	return 0;
5786}
5787
5788static int
5789skl_compute_ddb(struct intel_atomic_state *state)
5790{
5791	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
5792	const struct intel_dbuf_state *old_dbuf_state;
5793	const struct intel_dbuf_state *new_dbuf_state;
5794	const struct intel_crtc_state *old_crtc_state;
5795	struct intel_crtc_state *new_crtc_state;
5796	struct intel_crtc *crtc;
5797	int ret, i;
5798
5799	for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
5800					    new_crtc_state, i) {
5801		ret = skl_allocate_pipe_ddb(new_crtc_state);
5802		if (ret)
5803			return ret;
5804
5805		ret = skl_ddb_add_affected_planes(old_crtc_state,
5806						  new_crtc_state);
5807		if (ret)
5808			return ret;
5809	}
5810
5811	old_dbuf_state = intel_atomic_get_old_dbuf_state(state);
5812	new_dbuf_state = intel_atomic_get_new_dbuf_state(state);
5813
5814	if (new_dbuf_state &&
5815	    new_dbuf_state->enabled_slices != old_dbuf_state->enabled_slices)
5816		drm_dbg_kms(&dev_priv->drm,
5817			    "Enabled dbuf slices 0x%x -> 0x%x (out of %d dbuf slices)\n",
5818			    old_dbuf_state->enabled_slices,
5819			    new_dbuf_state->enabled_slices,
5820			    INTEL_INFO(dev_priv)->num_supported_dbuf_slices);
5821
5822	return 0;
5823}
5824
5825static char enast(bool enable)
5826{
5827	return enable ? '*' : ' ';
5828}
5829
5830static void
5831skl_print_wm_changes(struct intel_atomic_state *state)
5832{
5833	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
5834	const struct intel_crtc_state *old_crtc_state;
5835	const struct intel_crtc_state *new_crtc_state;
5836	struct intel_plane *plane;
5837	struct intel_crtc *crtc;
5838	int i;
5839
5840	if (!drm_debug_enabled(DRM_UT_KMS))
5841		return;
5842
5843	for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
5844					    new_crtc_state, i) {
5845		const struct skl_pipe_wm *old_pipe_wm, *new_pipe_wm;
5846
5847		old_pipe_wm = &old_crtc_state->wm.skl.optimal;
5848		new_pipe_wm = &new_crtc_state->wm.skl.optimal;
5849
5850		for_each_intel_plane_on_crtc(&dev_priv->drm, crtc, plane) {
5851			enum plane_id plane_id = plane->id;
5852			const struct skl_ddb_entry *old, *new;
5853
5854			old = &old_crtc_state->wm.skl.plane_ddb_y[plane_id];
5855			new = &new_crtc_state->wm.skl.plane_ddb_y[plane_id];
5856
5857			if (skl_ddb_entry_equal(old, new))
5858				continue;
5859
5860			drm_dbg_kms(&dev_priv->drm,
5861				    "[PLANE:%d:%s] ddb (%4d - %4d) -> (%4d - %4d), size %4d -> %4d\n",
5862				    plane->base.base.id, plane->base.name,
5863				    old->start, old->end, new->start, new->end,
5864				    skl_ddb_entry_size(old), skl_ddb_entry_size(new));
5865		}
5866
5867		for_each_intel_plane_on_crtc(&dev_priv->drm, crtc, plane) {
5868			enum plane_id plane_id = plane->id;
5869			const struct skl_plane_wm *old_wm, *new_wm;
5870
5871			old_wm = &old_pipe_wm->planes[plane_id];
5872			new_wm = &new_pipe_wm->planes[plane_id];
5873
5874			if (skl_plane_wm_equals(dev_priv, old_wm, new_wm))
5875				continue;
5876
5877			drm_dbg_kms(&dev_priv->drm,
5878				    "[PLANE:%d:%s]   level %cwm0,%cwm1,%cwm2,%cwm3,%cwm4,%cwm5,%cwm6,%cwm7,%ctwm,%cswm"
5879				    " -> %cwm0,%cwm1,%cwm2,%cwm3,%cwm4,%cwm5,%cwm6,%cwm7,%ctwm,%cswm\n",
5880				    plane->base.base.id, plane->base.name,
5881				    enast(old_wm->wm[0].plane_en), enast(old_wm->wm[1].plane_en),
5882				    enast(old_wm->wm[2].plane_en), enast(old_wm->wm[3].plane_en),
5883				    enast(old_wm->wm[4].plane_en), enast(old_wm->wm[5].plane_en),
5884				    enast(old_wm->wm[6].plane_en), enast(old_wm->wm[7].plane_en),
5885				    enast(old_wm->trans_wm.plane_en),
5886				    enast(old_wm->sagv_wm0.plane_en),
5887				    enast(new_wm->wm[0].plane_en), enast(new_wm->wm[1].plane_en),
5888				    enast(new_wm->wm[2].plane_en), enast(new_wm->wm[3].plane_en),
5889				    enast(new_wm->wm[4].plane_en), enast(new_wm->wm[5].plane_en),
5890				    enast(new_wm->wm[6].plane_en), enast(new_wm->wm[7].plane_en),
5891				    enast(new_wm->trans_wm.plane_en),
5892				    enast(new_wm->sagv_wm0.plane_en));
5893
5894			drm_dbg_kms(&dev_priv->drm,
5895				    "[PLANE:%d:%s]   lines %c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d"
5896				      " -> %c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d,%c%3d\n",
5897				    plane->base.base.id, plane->base.name,
5898				    enast(old_wm->wm[0].ignore_lines), old_wm->wm[0].plane_res_l,
5899				    enast(old_wm->wm[1].ignore_lines), old_wm->wm[1].plane_res_l,
5900				    enast(old_wm->wm[2].ignore_lines), old_wm->wm[2].plane_res_l,
5901				    enast(old_wm->wm[3].ignore_lines), old_wm->wm[3].plane_res_l,
5902				    enast(old_wm->wm[4].ignore_lines), old_wm->wm[4].plane_res_l,
5903				    enast(old_wm->wm[5].ignore_lines), old_wm->wm[5].plane_res_l,
5904				    enast(old_wm->wm[6].ignore_lines), old_wm->wm[6].plane_res_l,
5905				    enast(old_wm->wm[7].ignore_lines), old_wm->wm[7].plane_res_l,
5906				    enast(old_wm->trans_wm.ignore_lines), old_wm->trans_wm.plane_res_l,
5907				    enast(old_wm->sagv_wm0.ignore_lines), old_wm->sagv_wm0.plane_res_l,
5908
5909				    enast(new_wm->wm[0].ignore_lines), new_wm->wm[0].plane_res_l,
5910				    enast(new_wm->wm[1].ignore_lines), new_wm->wm[1].plane_res_l,
5911				    enast(new_wm->wm[2].ignore_lines), new_wm->wm[2].plane_res_l,
5912				    enast(new_wm->wm[3].ignore_lines), new_wm->wm[3].plane_res_l,
5913				    enast(new_wm->wm[4].ignore_lines), new_wm->wm[4].plane_res_l,
5914				    enast(new_wm->wm[5].ignore_lines), new_wm->wm[5].plane_res_l,
5915				    enast(new_wm->wm[6].ignore_lines), new_wm->wm[6].plane_res_l,
5916				    enast(new_wm->wm[7].ignore_lines), new_wm->wm[7].plane_res_l,
5917				    enast(new_wm->trans_wm.ignore_lines), new_wm->trans_wm.plane_res_l,
5918				    enast(new_wm->sagv_wm0.ignore_lines), new_wm->sagv_wm0.plane_res_l);
5919
5920			drm_dbg_kms(&dev_priv->drm,
5921				    "[PLANE:%d:%s]  blocks %4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d"
5922				    " -> %4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d\n",
5923				    plane->base.base.id, plane->base.name,
5924				    old_wm->wm[0].plane_res_b, old_wm->wm[1].plane_res_b,
5925				    old_wm->wm[2].plane_res_b, old_wm->wm[3].plane_res_b,
5926				    old_wm->wm[4].plane_res_b, old_wm->wm[5].plane_res_b,
5927				    old_wm->wm[6].plane_res_b, old_wm->wm[7].plane_res_b,
5928				    old_wm->trans_wm.plane_res_b,
5929				    old_wm->sagv_wm0.plane_res_b,
5930				    new_wm->wm[0].plane_res_b, new_wm->wm[1].plane_res_b,
5931				    new_wm->wm[2].plane_res_b, new_wm->wm[3].plane_res_b,
5932				    new_wm->wm[4].plane_res_b, new_wm->wm[5].plane_res_b,
5933				    new_wm->wm[6].plane_res_b, new_wm->wm[7].plane_res_b,
5934				    new_wm->trans_wm.plane_res_b,
5935				    new_wm->sagv_wm0.plane_res_b);
5936
5937			drm_dbg_kms(&dev_priv->drm,
5938				    "[PLANE:%d:%s] min_ddb %4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d"
5939				    " -> %4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d,%4d\n",
5940				    plane->base.base.id, plane->base.name,
5941				    old_wm->wm[0].min_ddb_alloc, old_wm->wm[1].min_ddb_alloc,
5942				    old_wm->wm[2].min_ddb_alloc, old_wm->wm[3].min_ddb_alloc,
5943				    old_wm->wm[4].min_ddb_alloc, old_wm->wm[5].min_ddb_alloc,
5944				    old_wm->wm[6].min_ddb_alloc, old_wm->wm[7].min_ddb_alloc,
5945				    old_wm->trans_wm.min_ddb_alloc,
5946				    old_wm->sagv_wm0.min_ddb_alloc,
5947				    new_wm->wm[0].min_ddb_alloc, new_wm->wm[1].min_ddb_alloc,
5948				    new_wm->wm[2].min_ddb_alloc, new_wm->wm[3].min_ddb_alloc,
5949				    new_wm->wm[4].min_ddb_alloc, new_wm->wm[5].min_ddb_alloc,
5950				    new_wm->wm[6].min_ddb_alloc, new_wm->wm[7].min_ddb_alloc,
5951				    new_wm->trans_wm.min_ddb_alloc,
5952				    new_wm->sagv_wm0.min_ddb_alloc);
5953		}
5954	}
5955}
5956
5957static int intel_add_affected_pipes(struct intel_atomic_state *state,
5958				    u8 pipe_mask)
5959{
5960	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
5961	struct intel_crtc *crtc;
5962
5963	for_each_intel_crtc(&dev_priv->drm, crtc) {
5964		struct intel_crtc_state *crtc_state;
5965
5966		if ((pipe_mask & BIT(crtc->pipe)) == 0)
5967			continue;
5968
5969		crtc_state = intel_atomic_get_crtc_state(&state->base, crtc);
5970		if (IS_ERR(crtc_state))
5971			return PTR_ERR(crtc_state);
5972	}
5973
5974	return 0;
5975}
5976
5977static int
5978skl_ddb_add_affected_pipes(struct intel_atomic_state *state)
5979{
5980	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
5981	struct intel_crtc_state *crtc_state;
5982	struct intel_crtc *crtc;
5983	int i, ret;
5984
5985	if (dev_priv->wm.distrust_bios_wm) {
5986		/*
5987		 * skl_ddb_get_pipe_allocation_limits() currently requires
5988		 * all active pipes to be included in the state so that
5989		 * it can redistribute the dbuf among them, and it really
5990		 * wants to recompute things when distrust_bios_wm is set
5991		 * so we add all the pipes to the state.
5992		 */
5993		ret = intel_add_affected_pipes(state, ~0);
5994		if (ret)
5995			return ret;
5996	}
5997
5998	for_each_new_intel_crtc_in_state(state, crtc, crtc_state, i) {
5999		struct intel_dbuf_state *new_dbuf_state;
6000		const struct intel_dbuf_state *old_dbuf_state;
6001
6002		new_dbuf_state = intel_atomic_get_dbuf_state(state);
6003		if (IS_ERR(new_dbuf_state))
6004			return PTR_ERR(new_dbuf_state);
6005
6006		old_dbuf_state = intel_atomic_get_old_dbuf_state(state);
6007
6008		new_dbuf_state->active_pipes =
6009			intel_calc_active_pipes(state, old_dbuf_state->active_pipes);
6010
6011		if (old_dbuf_state->active_pipes == new_dbuf_state->active_pipes)
6012			break;
6013
6014		ret = intel_atomic_lock_global_state(&new_dbuf_state->base);
6015		if (ret)
6016			return ret;
6017
6018		/*
6019		 * skl_ddb_get_pipe_allocation_limits() currently requires
6020		 * all active pipes to be included in the state so that
6021		 * it can redistribute the dbuf among them.
6022		 */
6023		ret = intel_add_affected_pipes(state,
6024					       new_dbuf_state->active_pipes);
6025		if (ret)
6026			return ret;
6027
6028		break;
6029	}
6030
6031	return 0;
6032}
6033
6034/*
6035 * To make sure the cursor watermark registers are always consistent
6036 * with our computed state the following scenario needs special
6037 * treatment:
6038 *
6039 * 1. enable cursor
6040 * 2. move cursor entirely offscreen
6041 * 3. disable cursor
6042 *
6043 * Step 2. does call .disable_plane() but does not zero the watermarks
6044 * (since we consider an offscreen cursor still active for the purposes
6045 * of watermarks). Step 3. would not normally call .disable_plane()
6046 * because the actual plane visibility isn't changing, and we don't
6047 * deallocate the cursor ddb until the pipe gets disabled. So we must
6048 * force step 3. to call .disable_plane() to update the watermark
6049 * registers properly.
6050 *
6051 * Other planes do not suffer from this issues as their watermarks are
6052 * calculated based on the actual plane visibility. The only time this
6053 * can trigger for the other planes is during the initial readout as the
6054 * default value of the watermarks registers is not zero.
6055 */
6056static int skl_wm_add_affected_planes(struct intel_atomic_state *state,
6057				      struct intel_crtc *crtc)
6058{
6059	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
6060	const struct intel_crtc_state *old_crtc_state =
6061		intel_atomic_get_old_crtc_state(state, crtc);
6062	struct intel_crtc_state *new_crtc_state =
6063		intel_atomic_get_new_crtc_state(state, crtc);
6064	struct intel_plane *plane;
6065
6066	for_each_intel_plane_on_crtc(&dev_priv->drm, crtc, plane) {
6067		struct intel_plane_state *plane_state;
6068		enum plane_id plane_id = plane->id;
6069
6070		/*
6071		 * Force a full wm update for every plane on modeset.
6072		 * Required because the reset value of the wm registers
6073		 * is non-zero, whereas we want all disabled planes to
6074		 * have zero watermarks. So if we turn off the relevant
6075		 * power well the hardware state will go out of sync
6076		 * with the software state.
6077		 */
6078		if (!drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi) &&
6079		    skl_plane_wm_equals(dev_priv,
6080					&old_crtc_state->wm.skl.optimal.planes[plane_id],
6081					&new_crtc_state->wm.skl.optimal.planes[plane_id]))
6082			continue;
6083
6084		plane_state = intel_atomic_get_plane_state(state, plane);
6085		if (IS_ERR(plane_state))
6086			return PTR_ERR(plane_state);
6087
6088		new_crtc_state->update_planes |= BIT(plane_id);
6089	}
6090
6091	return 0;
6092}
6093
6094static int
6095skl_compute_wm(struct intel_atomic_state *state)
6096{
6097	struct intel_crtc *crtc;
6098	struct intel_crtc_state *new_crtc_state;
6099	struct intel_crtc_state *old_crtc_state;
6100	int ret, i;
6101
6102	ret = skl_ddb_add_affected_pipes(state);
6103	if (ret)
6104		return ret;
6105
6106	/*
6107	 * Calculate WM's for all pipes that are part of this transaction.
6108	 * Note that skl_ddb_add_affected_pipes may have added more CRTC's that
6109	 * weren't otherwise being modified if pipe allocations had to change.
6110	 */
6111	for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
6112					    new_crtc_state, i) {
6113		ret = skl_build_pipe_wm(new_crtc_state);
6114		if (ret)
6115			return ret;
6116	}
6117
6118	ret = skl_compute_ddb(state);
6119	if (ret)
6120		return ret;
6121
6122	ret = intel_compute_sagv_mask(state);
6123	if (ret)
6124		return ret;
6125
6126	/*
6127	 * skl_compute_ddb() will have adjusted the final watermarks
6128	 * based on how much ddb is available. Now we can actually
6129	 * check if the final watermarks changed.
6130	 */
6131	for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
6132					    new_crtc_state, i) {
6133		ret = skl_wm_add_affected_planes(state, crtc);
6134		if (ret)
6135			return ret;
6136	}
6137
6138	skl_print_wm_changes(state);
6139
6140	return 0;
6141}
6142
6143static void ilk_compute_wm_config(struct drm_i915_private *dev_priv,
6144				  struct intel_wm_config *config)
6145{
6146	struct intel_crtc *crtc;
6147
6148	/* Compute the currently _active_ config */
6149	for_each_intel_crtc(&dev_priv->drm, crtc) {
6150		const struct intel_pipe_wm *wm = &crtc->wm.active.ilk;
6151
6152		if (!wm->pipe_enabled)
6153			continue;
6154
6155		config->sprites_enabled |= wm->sprites_enabled;
6156		config->sprites_scaled |= wm->sprites_scaled;
6157		config->num_pipes_active++;
6158	}
6159}
6160
6161static void ilk_program_watermarks(struct drm_i915_private *dev_priv)
6162{
6163	struct intel_pipe_wm lp_wm_1_2 = {}, lp_wm_5_6 = {}, *best_lp_wm;
6164	struct ilk_wm_maximums max;
6165	struct intel_wm_config config = {};
6166	struct ilk_wm_values results = {};
6167	enum intel_ddb_partitioning partitioning;
6168
6169	ilk_compute_wm_config(dev_priv, &config);
6170
6171	ilk_compute_wm_maximums(dev_priv, 1, &config, INTEL_DDB_PART_1_2, &max);
6172	ilk_wm_merge(dev_priv, &config, &max, &lp_wm_1_2);
6173
6174	/* 5/6 split only in single pipe config on IVB+ */
6175	if (INTEL_GEN(dev_priv) >= 7 &&
6176	    config.num_pipes_active == 1 && config.sprites_enabled) {
6177		ilk_compute_wm_maximums(dev_priv, 1, &config, INTEL_DDB_PART_5_6, &max);
6178		ilk_wm_merge(dev_priv, &config, &max, &lp_wm_5_6);
6179
6180		best_lp_wm = ilk_find_best_result(dev_priv, &lp_wm_1_2, &lp_wm_5_6);
6181	} else {
6182		best_lp_wm = &lp_wm_1_2;
6183	}
6184
6185	partitioning = (best_lp_wm == &lp_wm_1_2) ?
6186		       INTEL_DDB_PART_1_2 : INTEL_DDB_PART_5_6;
6187
6188	ilk_compute_wm_results(dev_priv, best_lp_wm, partitioning, &results);
6189
6190	ilk_write_wm_values(dev_priv, &results);
6191}
6192
6193static void ilk_initial_watermarks(struct intel_atomic_state *state,
6194				   struct intel_crtc *crtc)
6195{
6196	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
6197	const struct intel_crtc_state *crtc_state =
6198		intel_atomic_get_new_crtc_state(state, crtc);
6199
6200	mutex_lock(&dev_priv->wm.wm_mutex);
6201	crtc->wm.active.ilk = crtc_state->wm.ilk.intermediate;
6202	ilk_program_watermarks(dev_priv);
6203	mutex_unlock(&dev_priv->wm.wm_mutex);
6204}
6205
6206static void ilk_optimize_watermarks(struct intel_atomic_state *state,
6207				    struct intel_crtc *crtc)
6208{
6209	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
6210	const struct intel_crtc_state *crtc_state =
6211		intel_atomic_get_new_crtc_state(state, crtc);
6212
6213	if (!crtc_state->wm.need_postvbl_update)
6214		return;
6215
6216	mutex_lock(&dev_priv->wm.wm_mutex);
6217	crtc->wm.active.ilk = crtc_state->wm.ilk.optimal;
6218	ilk_program_watermarks(dev_priv);
6219	mutex_unlock(&dev_priv->wm.wm_mutex);
6220}
6221
6222static void skl_wm_level_from_reg_val(u32 val, struct skl_wm_level *level)
6223{
6224	level->plane_en = val & PLANE_WM_EN;
6225	level->ignore_lines = val & PLANE_WM_IGNORE_LINES;
6226	level->plane_res_b = val & PLANE_WM_BLOCKS_MASK;
6227	level->plane_res_l = (val >> PLANE_WM_LINES_SHIFT) &
6228		PLANE_WM_LINES_MASK;
6229}
6230
6231void skl_pipe_wm_get_hw_state(struct intel_crtc *crtc,
6232			      struct skl_pipe_wm *out)
6233{
6234	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
6235	enum pipe pipe = crtc->pipe;
6236	int level, max_level;
6237	enum plane_id plane_id;
6238	u32 val;
6239
6240	max_level = ilk_wm_max_level(dev_priv);
6241
6242	for_each_plane_id_on_crtc(crtc, plane_id) {
6243		struct skl_plane_wm *wm = &out->planes[plane_id];
6244
6245		for (level = 0; level <= max_level; level++) {
6246			if (plane_id != PLANE_CURSOR)
6247				val = I915_READ(PLANE_WM(pipe, plane_id, level));
6248			else
6249				val = I915_READ(CUR_WM(pipe, level));
6250
6251			skl_wm_level_from_reg_val(val, &wm->wm[level]);
6252		}
6253
6254		if (INTEL_GEN(dev_priv) >= 12)
6255			wm->sagv_wm0 = wm->wm[0];
6256
6257		if (plane_id != PLANE_CURSOR)
6258			val = I915_READ(PLANE_WM_TRANS(pipe, plane_id));
6259		else
6260			val = I915_READ(CUR_WM_TRANS(pipe));
6261
6262		skl_wm_level_from_reg_val(val, &wm->trans_wm);
6263	}
6264
6265	if (!crtc->active)
6266		return;
6267}
6268
6269void skl_wm_get_hw_state(struct drm_i915_private *dev_priv)
6270{
6271	struct intel_crtc *crtc;
6272	struct intel_crtc_state *crtc_state;
6273
6274	for_each_intel_crtc(&dev_priv->drm, crtc) {
6275		crtc_state = to_intel_crtc_state(crtc->base.state);
6276
6277		skl_pipe_wm_get_hw_state(crtc, &crtc_state->wm.skl.optimal);
6278	}
6279
6280	if (dev_priv->active_pipes) {
6281		/* Fully recompute DDB on first atomic commit */
6282		dev_priv->wm.distrust_bios_wm = true;
6283	}
6284}
6285
6286static void ilk_pipe_wm_get_hw_state(struct intel_crtc *crtc)
6287{
6288	struct drm_device *dev = crtc->base.dev;
6289	struct drm_i915_private *dev_priv = to_i915(dev);
6290	struct ilk_wm_values *hw = &dev_priv->wm.hw;
6291	struct intel_crtc_state *crtc_state = to_intel_crtc_state(crtc->base.state);
6292	struct intel_pipe_wm *active = &crtc_state->wm.ilk.optimal;
6293	enum pipe pipe = crtc->pipe;
6294	static const i915_reg_t wm0_pipe_reg[] = {
6295		[PIPE_A] = WM0_PIPEA_ILK,
6296		[PIPE_B] = WM0_PIPEB_ILK,
6297		[PIPE_C] = WM0_PIPEC_IVB,
6298	};
6299
6300	hw->wm_pipe[pipe] = I915_READ(wm0_pipe_reg[pipe]);
6301
6302	memset(active, 0, sizeof(*active));
6303
6304	active->pipe_enabled = crtc->active;
6305
6306	if (active->pipe_enabled) {
6307		u32 tmp = hw->wm_pipe[pipe];
6308
6309		/*
6310		 * For active pipes LP0 watermark is marked as
6311		 * enabled, and LP1+ watermaks as disabled since
6312		 * we can't really reverse compute them in case
6313		 * multiple pipes are active.
6314		 */
6315		active->wm[0].enable = true;
6316		active->wm[0].pri_val = (tmp & WM0_PIPE_PLANE_MASK) >> WM0_PIPE_PLANE_SHIFT;
6317		active->wm[0].spr_val = (tmp & WM0_PIPE_SPRITE_MASK) >> WM0_PIPE_SPRITE_SHIFT;
6318		active->wm[0].cur_val = tmp & WM0_PIPE_CURSOR_MASK;
6319	} else {
6320		int level, max_level = ilk_wm_max_level(dev_priv);
6321
6322		/*
6323		 * For inactive pipes, all watermark levels
6324		 * should be marked as enabled but zeroed,
6325		 * which is what we'd compute them to.
6326		 */
6327		for (level = 0; level <= max_level; level++)
6328			active->wm[level].enable = true;
6329	}
6330
6331	crtc->wm.active.ilk = *active;
6332}
6333
6334#define _FW_WM(value, plane) \
6335	(((value) & DSPFW_ ## plane ## _MASK) >> DSPFW_ ## plane ## _SHIFT)
6336#define _FW_WM_VLV(value, plane) \
6337	(((value) & DSPFW_ ## plane ## _MASK_VLV) >> DSPFW_ ## plane ## _SHIFT)
6338
6339static void g4x_read_wm_values(struct drm_i915_private *dev_priv,
6340			       struct g4x_wm_values *wm)
6341{
6342	u32 tmp;
6343
6344	tmp = I915_READ(DSPFW1);
6345	wm->sr.plane = _FW_WM(tmp, SR);
6346	wm->pipe[PIPE_B].plane[PLANE_CURSOR] = _FW_WM(tmp, CURSORB);
6347	wm->pipe[PIPE_B].plane[PLANE_PRIMARY] = _FW_WM(tmp, PLANEB);
6348	wm->pipe[PIPE_A].plane[PLANE_PRIMARY] = _FW_WM(tmp, PLANEA);
6349
6350	tmp = I915_READ(DSPFW2);
6351	wm->fbc_en = tmp & DSPFW_FBC_SR_EN;
6352	wm->sr.fbc = _FW_WM(tmp, FBC_SR);
6353	wm->hpll.fbc = _FW_WM(tmp, FBC_HPLL_SR);
6354	wm->pipe[PIPE_B].plane[PLANE_SPRITE0] = _FW_WM(tmp, SPRITEB);
6355	wm->pipe[PIPE_A].plane[PLANE_CURSOR] = _FW_WM(tmp, CURSORA);
6356	wm->pipe[PIPE_A].plane[PLANE_SPRITE0] = _FW_WM(tmp, SPRITEA);
6357
6358	tmp = I915_READ(DSPFW3);
6359	wm->hpll_en = tmp & DSPFW_HPLL_SR_EN;
6360	wm->sr.cursor = _FW_WM(tmp, CURSOR_SR);
6361	wm->hpll.cursor = _FW_WM(tmp, HPLL_CURSOR);
6362	wm->hpll.plane = _FW_WM(tmp, HPLL_SR);
6363}
6364
6365static void vlv_read_wm_values(struct drm_i915_private *dev_priv,
6366			       struct vlv_wm_values *wm)
6367{
6368	enum pipe pipe;
6369	u32 tmp;
6370
6371	for_each_pipe(dev_priv, pipe) {
6372		tmp = I915_READ(VLV_DDL(pipe));
6373
6374		wm->ddl[pipe].plane[PLANE_PRIMARY] =
6375			(tmp >> DDL_PLANE_SHIFT) & (DDL_PRECISION_HIGH | DRAIN_LATENCY_MASK);
6376		wm->ddl[pipe].plane[PLANE_CURSOR] =
6377			(tmp >> DDL_CURSOR_SHIFT) & (DDL_PRECISION_HIGH | DRAIN_LATENCY_MASK);
6378		wm->ddl[pipe].plane[PLANE_SPRITE0] =
6379			(tmp >> DDL_SPRITE_SHIFT(0)) & (DDL_PRECISION_HIGH | DRAIN_LATENCY_MASK);
6380		wm->ddl[pipe].plane[PLANE_SPRITE1] =
6381			(tmp >> DDL_SPRITE_SHIFT(1)) & (DDL_PRECISION_HIGH | DRAIN_LATENCY_MASK);
6382	}
6383
6384	tmp = I915_READ(DSPFW1);
6385	wm->sr.plane = _FW_WM(tmp, SR);
6386	wm->pipe[PIPE_B].plane[PLANE_CURSOR] = _FW_WM(tmp, CURSORB);
6387	wm->pipe[PIPE_B].plane[PLANE_PRIMARY] = _FW_WM_VLV(tmp, PLANEB);
6388	wm->pipe[PIPE_A].plane[PLANE_PRIMARY] = _FW_WM_VLV(tmp, PLANEA);
6389
6390	tmp = I915_READ(DSPFW2);
6391	wm->pipe[PIPE_A].plane[PLANE_SPRITE1] = _FW_WM_VLV(tmp, SPRITEB);
6392	wm->pipe[PIPE_A].plane[PLANE_CURSOR] = _FW_WM(tmp, CURSORA);
6393	wm->pipe[PIPE_A].plane[PLANE_SPRITE0] = _FW_WM_VLV(tmp, SPRITEA);
6394
6395	tmp = I915_READ(DSPFW3);
6396	wm->sr.cursor = _FW_WM(tmp, CURSOR_SR);
6397
6398	if (IS_CHERRYVIEW(dev_priv)) {
6399		tmp = I915_READ(DSPFW7_CHV);
6400		wm->pipe[PIPE_B].plane[PLANE_SPRITE1] = _FW_WM_VLV(tmp, SPRITED);
6401		wm->pipe[PIPE_B].plane[PLANE_SPRITE0] = _FW_WM_VLV(tmp, SPRITEC);
6402
6403		tmp = I915_READ(DSPFW8_CHV);
6404		wm->pipe[PIPE_C].plane[PLANE_SPRITE1] = _FW_WM_VLV(tmp, SPRITEF);
6405		wm->pipe[PIPE_C].plane[PLANE_SPRITE0] = _FW_WM_VLV(tmp, SPRITEE);
6406
6407		tmp = I915_READ(DSPFW9_CHV);
6408		wm->pipe[PIPE_C].plane[PLANE_PRIMARY] = _FW_WM_VLV(tmp, PLANEC);
6409		wm->pipe[PIPE_C].plane[PLANE_CURSOR] = _FW_WM(tmp, CURSORC);
6410
6411		tmp = I915_READ(DSPHOWM);
6412		wm->sr.plane |= _FW_WM(tmp, SR_HI) << 9;
6413		wm->pipe[PIPE_C].plane[PLANE_SPRITE1] |= _FW_WM(tmp, SPRITEF_HI) << 8;
6414		wm->pipe[PIPE_C].plane[PLANE_SPRITE0] |= _FW_WM(tmp, SPRITEE_HI) << 8;
6415		wm->pipe[PIPE_C].plane[PLANE_PRIMARY] |= _FW_WM(tmp, PLANEC_HI) << 8;
6416		wm->pipe[PIPE_B].plane[PLANE_SPRITE1] |= _FW_WM(tmp, SPRITED_HI) << 8;
6417		wm->pipe[PIPE_B].plane[PLANE_SPRITE0] |= _FW_WM(tmp, SPRITEC_HI) << 8;
6418		wm->pipe[PIPE_B].plane[PLANE_PRIMARY] |= _FW_WM(tmp, PLANEB_HI) << 8;
6419		wm->pipe[PIPE_A].plane[PLANE_SPRITE1] |= _FW_WM(tmp, SPRITEB_HI) << 8;
6420		wm->pipe[PIPE_A].plane[PLANE_SPRITE0] |= _FW_WM(tmp, SPRITEA_HI) << 8;
6421		wm->pipe[PIPE_A].plane[PLANE_PRIMARY] |= _FW_WM(tmp, PLANEA_HI) << 8;
6422	} else {
6423		tmp = I915_READ(DSPFW7);
6424		wm->pipe[PIPE_B].plane[PLANE_SPRITE1] = _FW_WM_VLV(tmp, SPRITED);
6425		wm->pipe[PIPE_B].plane[PLANE_SPRITE0] = _FW_WM_VLV(tmp, SPRITEC);
6426
6427		tmp = I915_READ(DSPHOWM);
6428		wm->sr.plane |= _FW_WM(tmp, SR_HI) << 9;
6429		wm->pipe[PIPE_B].plane[PLANE_SPRITE1] |= _FW_WM(tmp, SPRITED_HI) << 8;
6430		wm->pipe[PIPE_B].plane[PLANE_SPRITE0] |= _FW_WM(tmp, SPRITEC_HI) << 8;
6431		wm->pipe[PIPE_B].plane[PLANE_PRIMARY] |= _FW_WM(tmp, PLANEB_HI) << 8;
6432		wm->pipe[PIPE_A].plane[PLANE_SPRITE1] |= _FW_WM(tmp, SPRITEB_HI) << 8;
6433		wm->pipe[PIPE_A].plane[PLANE_SPRITE0] |= _FW_WM(tmp, SPRITEA_HI) << 8;
6434		wm->pipe[PIPE_A].plane[PLANE_PRIMARY] |= _FW_WM(tmp, PLANEA_HI) << 8;
6435	}
6436}
6437
6438#undef _FW_WM
6439#undef _FW_WM_VLV
6440
6441void g4x_wm_get_hw_state(struct drm_i915_private *dev_priv)
6442{
6443	struct g4x_wm_values *wm = &dev_priv->wm.g4x;
6444	struct intel_crtc *crtc;
6445
6446	g4x_read_wm_values(dev_priv, wm);
6447
6448	wm->cxsr = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN;
6449
6450	for_each_intel_crtc(&dev_priv->drm, crtc) {
6451		struct intel_crtc_state *crtc_state =
6452			to_intel_crtc_state(crtc->base.state);
6453		struct g4x_wm_state *active = &crtc->wm.active.g4x;
6454		struct g4x_pipe_wm *raw;
6455		enum pipe pipe = crtc->pipe;
6456		enum plane_id plane_id;
6457		int level, max_level;
6458
6459		active->cxsr = wm->cxsr;
6460		active->hpll_en = wm->hpll_en;
6461		active->fbc_en = wm->fbc_en;
6462
6463		active->sr = wm->sr;
6464		active->hpll = wm->hpll;
6465
6466		for_each_plane_id_on_crtc(crtc, plane_id) {
6467			active->wm.plane[plane_id] =
6468				wm->pipe[pipe].plane[plane_id];
6469		}
6470
6471		if (wm->cxsr && wm->hpll_en)
6472			max_level = G4X_WM_LEVEL_HPLL;
6473		else if (wm->cxsr)
6474			max_level = G4X_WM_LEVEL_SR;
6475		else
6476			max_level = G4X_WM_LEVEL_NORMAL;
6477
6478		level = G4X_WM_LEVEL_NORMAL;
6479		raw = &crtc_state->wm.g4x.raw[level];
6480		for_each_plane_id_on_crtc(crtc, plane_id)
6481			raw->plane[plane_id] = active->wm.plane[plane_id];
6482
6483		if (++level > max_level)
6484			goto out;
6485
6486		raw = &crtc_state->wm.g4x.raw[level];
6487		raw->plane[PLANE_PRIMARY] = active->sr.plane;
6488		raw->plane[PLANE_CURSOR] = active->sr.cursor;
6489		raw->plane[PLANE_SPRITE0] = 0;
6490		raw->fbc = active->sr.fbc;
6491
6492		if (++level > max_level)
6493			goto out;
6494
6495		raw = &crtc_state->wm.g4x.raw[level];
6496		raw->plane[PLANE_PRIMARY] = active->hpll.plane;
6497		raw->plane[PLANE_CURSOR] = active->hpll.cursor;
6498		raw->plane[PLANE_SPRITE0] = 0;
6499		raw->fbc = active->hpll.fbc;
6500
6501	out:
6502		for_each_plane_id_on_crtc(crtc, plane_id)
6503			g4x_raw_plane_wm_set(crtc_state, level,
6504					     plane_id, USHRT_MAX);
6505		g4x_raw_fbc_wm_set(crtc_state, level, USHRT_MAX);
6506
6507		crtc_state->wm.g4x.optimal = *active;
6508		crtc_state->wm.g4x.intermediate = *active;
6509
6510		drm_dbg_kms(&dev_priv->drm,
6511			    "Initial watermarks: pipe %c, plane=%d, cursor=%d, sprite=%d\n",
6512			    pipe_name(pipe),
6513			    wm->pipe[pipe].plane[PLANE_PRIMARY],
6514			    wm->pipe[pipe].plane[PLANE_CURSOR],
6515			    wm->pipe[pipe].plane[PLANE_SPRITE0]);
6516	}
6517
6518	drm_dbg_kms(&dev_priv->drm,
6519		    "Initial SR watermarks: plane=%d, cursor=%d fbc=%d\n",
6520		    wm->sr.plane, wm->sr.cursor, wm->sr.fbc);
6521	drm_dbg_kms(&dev_priv->drm,
6522		    "Initial HPLL watermarks: plane=%d, SR cursor=%d fbc=%d\n",
6523		    wm->hpll.plane, wm->hpll.cursor, wm->hpll.fbc);
6524	drm_dbg_kms(&dev_priv->drm, "Initial SR=%s HPLL=%s FBC=%s\n",
6525		    yesno(wm->cxsr), yesno(wm->hpll_en), yesno(wm->fbc_en));
6526}
6527
6528void g4x_wm_sanitize(struct drm_i915_private *dev_priv)
6529{
6530	struct intel_plane *plane;
6531	struct intel_crtc *crtc;
6532
6533	mutex_lock(&dev_priv->wm.wm_mutex);
6534
6535	for_each_intel_plane(&dev_priv->drm, plane) {
6536		struct intel_crtc *crtc =
6537			intel_get_crtc_for_pipe(dev_priv, plane->pipe);
6538		struct intel_crtc_state *crtc_state =
6539			to_intel_crtc_state(crtc->base.state);
6540		struct intel_plane_state *plane_state =
6541			to_intel_plane_state(plane->base.state);
6542		struct g4x_wm_state *wm_state = &crtc_state->wm.g4x.optimal;
6543		enum plane_id plane_id = plane->id;
6544		int level;
6545
6546		if (plane_state->uapi.visible)
6547			continue;
6548
6549		for (level = 0; level < 3; level++) {
6550			struct g4x_pipe_wm *raw =
6551				&crtc_state->wm.g4x.raw[level];
6552
6553			raw->plane[plane_id] = 0;
6554			wm_state->wm.plane[plane_id] = 0;
6555		}
6556
6557		if (plane_id == PLANE_PRIMARY) {
6558			for (level = 0; level < 3; level++) {
6559				struct g4x_pipe_wm *raw =
6560					&crtc_state->wm.g4x.raw[level];
6561				raw->fbc = 0;
6562			}
6563
6564			wm_state->sr.fbc = 0;
6565			wm_state->hpll.fbc = 0;
6566			wm_state->fbc_en = false;
6567		}
6568	}
6569
6570	for_each_intel_crtc(&dev_priv->drm, crtc) {
6571		struct intel_crtc_state *crtc_state =
6572			to_intel_crtc_state(crtc->base.state);
6573
6574		crtc_state->wm.g4x.intermediate =
6575			crtc_state->wm.g4x.optimal;
6576		crtc->wm.active.g4x = crtc_state->wm.g4x.optimal;
6577	}
6578
6579	g4x_program_watermarks(dev_priv);
6580
6581	mutex_unlock(&dev_priv->wm.wm_mutex);
6582}
6583
6584void vlv_wm_get_hw_state(struct drm_i915_private *dev_priv)
6585{
6586	struct vlv_wm_values *wm = &dev_priv->wm.vlv;
6587	struct intel_crtc *crtc;
6588	u32 val;
6589
6590	vlv_read_wm_values(dev_priv, wm);
6591
6592	wm->cxsr = I915_READ(FW_BLC_SELF_VLV) & FW_CSPWRDWNEN;
6593	wm->level = VLV_WM_LEVEL_PM2;
6594
6595	if (IS_CHERRYVIEW(dev_priv)) {
6596		vlv_punit_get(dev_priv);
6597
6598		val = vlv_punit_read(dev_priv, PUNIT_REG_DSPSSPM);
6599		if (val & DSP_MAXFIFO_PM5_ENABLE)
6600			wm->level = VLV_WM_LEVEL_PM5;
6601
6602		/*
6603		 * If DDR DVFS is disabled in the BIOS, Punit
6604		 * will never ack the request. So if that happens
6605		 * assume we don't have to enable/disable DDR DVFS
6606		 * dynamically. To test that just set the REQ_ACK
6607		 * bit to poke the Punit, but don't change the
6608		 * HIGH/LOW bits so that we don't actually change
6609		 * the current state.
6610		 */
6611		val = vlv_punit_read(dev_priv, PUNIT_REG_DDR_SETUP2);
6612		val |= FORCE_DDR_FREQ_REQ_ACK;
6613		vlv_punit_write(dev_priv, PUNIT_REG_DDR_SETUP2, val);
6614
6615		if (wait_for((vlv_punit_read(dev_priv, PUNIT_REG_DDR_SETUP2) &
6616			      FORCE_DDR_FREQ_REQ_ACK) == 0, 3)) {
6617			drm_dbg_kms(&dev_priv->drm,
6618				    "Punit not acking DDR DVFS request, "
6619				    "assuming DDR DVFS is disabled\n");
6620			dev_priv->wm.max_level = VLV_WM_LEVEL_PM5;
6621		} else {
6622			val = vlv_punit_read(dev_priv, PUNIT_REG_DDR_SETUP2);
6623			if ((val & FORCE_DDR_HIGH_FREQ) == 0)
6624				wm->level = VLV_WM_LEVEL_DDR_DVFS;
6625		}
6626
6627		vlv_punit_put(dev_priv);
6628	}
6629
6630	for_each_intel_crtc(&dev_priv->drm, crtc) {
6631		struct intel_crtc_state *crtc_state =
6632			to_intel_crtc_state(crtc->base.state);
6633		struct vlv_wm_state *active = &crtc->wm.active.vlv;
6634		const struct vlv_fifo_state *fifo_state =
6635			&crtc_state->wm.vlv.fifo_state;
6636		enum pipe pipe = crtc->pipe;
6637		enum plane_id plane_id;
6638		int level;
6639
6640		vlv_get_fifo_size(crtc_state);
6641
6642		active->num_levels = wm->level + 1;
6643		active->cxsr = wm->cxsr;
6644
6645		for (level = 0; level < active->num_levels; level++) {
6646			struct g4x_pipe_wm *raw =
6647				&crtc_state->wm.vlv.raw[level];
6648
6649			active->sr[level].plane = wm->sr.plane;
6650			active->sr[level].cursor = wm->sr.cursor;
6651
6652			for_each_plane_id_on_crtc(crtc, plane_id) {
6653				active->wm[level].plane[plane_id] =
6654					wm->pipe[pipe].plane[plane_id];
6655
6656				raw->plane[plane_id] =
6657					vlv_invert_wm_value(active->wm[level].plane[plane_id],
6658							    fifo_state->plane[plane_id]);
6659			}
6660		}
6661
6662		for_each_plane_id_on_crtc(crtc, plane_id)
6663			vlv_raw_plane_wm_set(crtc_state, level,
6664					     plane_id, USHRT_MAX);
6665		vlv_invalidate_wms(crtc, active, level);
6666
6667		crtc_state->wm.vlv.optimal = *active;
6668		crtc_state->wm.vlv.intermediate = *active;
6669
6670		drm_dbg_kms(&dev_priv->drm,
6671			    "Initial watermarks: pipe %c, plane=%d, cursor=%d, sprite0=%d, sprite1=%d\n",
6672			    pipe_name(pipe),
6673			    wm->pipe[pipe].plane[PLANE_PRIMARY],
6674			    wm->pipe[pipe].plane[PLANE_CURSOR],
6675			    wm->pipe[pipe].plane[PLANE_SPRITE0],
6676			    wm->pipe[pipe].plane[PLANE_SPRITE1]);
6677	}
6678
6679	drm_dbg_kms(&dev_priv->drm,
6680		    "Initial watermarks: SR plane=%d, SR cursor=%d level=%d cxsr=%d\n",
6681		    wm->sr.plane, wm->sr.cursor, wm->level, wm->cxsr);
6682}
6683
6684void vlv_wm_sanitize(struct drm_i915_private *dev_priv)
6685{
6686	struct intel_plane *plane;
6687	struct intel_crtc *crtc;
6688
6689	mutex_lock(&dev_priv->wm.wm_mutex);
6690
6691	for_each_intel_plane(&dev_priv->drm, plane) {
6692		struct intel_crtc *crtc =
6693			intel_get_crtc_for_pipe(dev_priv, plane->pipe);
6694		struct intel_crtc_state *crtc_state =
6695			to_intel_crtc_state(crtc->base.state);
6696		struct intel_plane_state *plane_state =
6697			to_intel_plane_state(plane->base.state);
6698		struct vlv_wm_state *wm_state = &crtc_state->wm.vlv.optimal;
6699		const struct vlv_fifo_state *fifo_state =
6700			&crtc_state->wm.vlv.fifo_state;
6701		enum plane_id plane_id = plane->id;
6702		int level;
6703
6704		if (plane_state->uapi.visible)
6705			continue;
6706
6707		for (level = 0; level < wm_state->num_levels; level++) {
6708			struct g4x_pipe_wm *raw =
6709				&crtc_state->wm.vlv.raw[level];
6710
6711			raw->plane[plane_id] = 0;
6712
6713			wm_state->wm[level].plane[plane_id] =
6714				vlv_invert_wm_value(raw->plane[plane_id],
6715						    fifo_state->plane[plane_id]);
6716		}
6717	}
6718
6719	for_each_intel_crtc(&dev_priv->drm, crtc) {
6720		struct intel_crtc_state *crtc_state =
6721			to_intel_crtc_state(crtc->base.state);
6722
6723		crtc_state->wm.vlv.intermediate =
6724			crtc_state->wm.vlv.optimal;
6725		crtc->wm.active.vlv = crtc_state->wm.vlv.optimal;
6726	}
6727
6728	vlv_program_watermarks(dev_priv);
6729
6730	mutex_unlock(&dev_priv->wm.wm_mutex);
6731}
6732
6733/*
6734 * FIXME should probably kill this and improve
6735 * the real watermark readout/sanitation instead
6736 */
6737static void ilk_init_lp_watermarks(struct drm_i915_private *dev_priv)
6738{
6739	I915_WRITE(WM3_LP_ILK, I915_READ(WM3_LP_ILK) & ~WM1_LP_SR_EN);
6740	I915_WRITE(WM2_LP_ILK, I915_READ(WM2_LP_ILK) & ~WM1_LP_SR_EN);
6741	I915_WRITE(WM1_LP_ILK, I915_READ(WM1_LP_ILK) & ~WM1_LP_SR_EN);
6742
6743	/*
6744	 * Don't touch WM1S_LP_EN here.
6745	 * Doing so could cause underruns.
6746	 */
6747}
6748
6749void ilk_wm_get_hw_state(struct drm_i915_private *dev_priv)
6750{
6751	struct ilk_wm_values *hw = &dev_priv->wm.hw;
6752	struct intel_crtc *crtc;
6753
6754	ilk_init_lp_watermarks(dev_priv);
6755
6756	for_each_intel_crtc(&dev_priv->drm, crtc)
6757		ilk_pipe_wm_get_hw_state(crtc);
6758
6759	hw->wm_lp[0] = I915_READ(WM1_LP_ILK);
6760	hw->wm_lp[1] = I915_READ(WM2_LP_ILK);
6761	hw->wm_lp[2] = I915_READ(WM3_LP_ILK);
6762
6763	hw->wm_lp_spr[0] = I915_READ(WM1S_LP_ILK);
6764	if (INTEL_GEN(dev_priv) >= 7) {
6765		hw->wm_lp_spr[1] = I915_READ(WM2S_LP_IVB);
6766		hw->wm_lp_spr[2] = I915_READ(WM3S_LP_IVB);
6767	}
6768
6769	if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv))
6770		hw->partitioning = (I915_READ(WM_MISC) & WM_MISC_DATA_PARTITION_5_6) ?
6771			INTEL_DDB_PART_5_6 : INTEL_DDB_PART_1_2;
6772	else if (IS_IVYBRIDGE(dev_priv))
6773		hw->partitioning = (I915_READ(DISP_ARB_CTL2) & DISP_DATA_PARTITION_5_6) ?
6774			INTEL_DDB_PART_5_6 : INTEL_DDB_PART_1_2;
6775
6776	hw->enable_fbc_wm =
6777		!(I915_READ(DISP_ARB_CTL) & DISP_FBC_WM_DIS);
6778}
6779
6780/**
6781 * intel_update_watermarks - update FIFO watermark values based on current modes
6782 * @crtc: the #intel_crtc on which to compute the WM
6783 *
6784 * Calculate watermark values for the various WM regs based on current mode
6785 * and plane configuration.
6786 *
6787 * There are several cases to deal with here:
6788 *   - normal (i.e. non-self-refresh)
6789 *   - self-refresh (SR) mode
6790 *   - lines are large relative to FIFO size (buffer can hold up to 2)
6791 *   - lines are small relative to FIFO size (buffer can hold more than 2
6792 *     lines), so need to account for TLB latency
6793 *
6794 *   The normal calculation is:
6795 *     watermark = dotclock * bytes per pixel * latency
6796 *   where latency is platform & configuration dependent (we assume pessimal
6797 *   values here).
6798 *
6799 *   The SR calculation is:
6800 *     watermark = (trunc(latency/line time)+1) * surface width *
6801 *       bytes per pixel
6802 *   where
6803 *     line time = htotal / dotclock
6804 *     surface width = hdisplay for normal plane and 64 for cursor
6805 *   and latency is assumed to be high, as above.
6806 *
6807 * The final value programmed to the register should always be rounded up,
6808 * and include an extra 2 entries to account for clock crossings.
6809 *
6810 * We don't use the sprite, so we can ignore that.  And on Crestline we have
6811 * to set the non-SR watermarks to 8.
6812 */
6813void intel_update_watermarks(struct intel_crtc *crtc)
6814{
6815	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
6816
6817	if (dev_priv->display.update_wm)
6818		dev_priv->display.update_wm(crtc);
6819}
6820
6821void intel_enable_ipc(struct drm_i915_private *dev_priv)
6822{
6823	u32 val;
6824
6825	if (!HAS_IPC(dev_priv))
6826		return;
6827
6828	val = I915_READ(DISP_ARB_CTL2);
6829
6830	if (dev_priv->ipc_enabled)
6831		val |= DISP_IPC_ENABLE;
6832	else
6833		val &= ~DISP_IPC_ENABLE;
6834
6835	I915_WRITE(DISP_ARB_CTL2, val);
6836}
6837
6838static bool intel_can_enable_ipc(struct drm_i915_private *dev_priv)
6839{
6840	/* Display WA #0477 WaDisableIPC: skl */
6841	if (IS_SKYLAKE(dev_priv))
6842		return false;
6843
6844	/* Display WA #1141: SKL:all KBL:all CFL */
6845	if (IS_KABYLAKE(dev_priv) ||
6846	    IS_COFFEELAKE(dev_priv) ||
6847	    IS_COMETLAKE(dev_priv))
6848		return dev_priv->dram_info.symmetric_memory;
6849
6850	return true;
6851}
6852
6853void intel_init_ipc(struct drm_i915_private *dev_priv)
6854{
6855	if (!HAS_IPC(dev_priv))
6856		return;
6857
6858	dev_priv->ipc_enabled = intel_can_enable_ipc(dev_priv);
6859
6860	intel_enable_ipc(dev_priv);
6861}
6862
6863static void ibx_init_clock_gating(struct drm_i915_private *dev_priv)
6864{
6865	/*
6866	 * On Ibex Peak and Cougar Point, we need to disable clock
6867	 * gating for the panel power sequencer or it will fail to
6868	 * start up when no ports are active.
6869	 */
6870	I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE);
6871}
6872
6873static void g4x_disable_trickle_feed(struct drm_i915_private *dev_priv)
6874{
6875	enum pipe pipe;
6876
6877	for_each_pipe(dev_priv, pipe) {
6878		I915_WRITE(DSPCNTR(pipe),
6879			   I915_READ(DSPCNTR(pipe)) |
6880			   DISPPLANE_TRICKLE_FEED_DISABLE);
6881
6882		I915_WRITE(DSPSURF(pipe), I915_READ(DSPSURF(pipe)));
6883		POSTING_READ(DSPSURF(pipe));
6884	}
6885}
6886
6887static void ilk_init_clock_gating(struct drm_i915_private *dev_priv)
6888{
6889	u32 dspclk_gate = ILK_VRHUNIT_CLOCK_GATE_DISABLE;
6890
6891	/*
6892	 * Required for FBC
6893	 * WaFbcDisableDpfcClockGating:ilk
6894	 */
6895	dspclk_gate |= ILK_DPFCRUNIT_CLOCK_GATE_DISABLE |
6896		   ILK_DPFCUNIT_CLOCK_GATE_DISABLE |
6897		   ILK_DPFDUNIT_CLOCK_GATE_ENABLE;
6898
6899	I915_WRITE(PCH_3DCGDIS0,
6900		   MARIUNIT_CLOCK_GATE_DISABLE |
6901		   SVSMUNIT_CLOCK_GATE_DISABLE);
6902	I915_WRITE(PCH_3DCGDIS1,
6903		   VFMUNIT_CLOCK_GATE_DISABLE);
6904
6905	/*
6906	 * According to the spec the following bits should be set in
6907	 * order to enable memory self-refresh
6908	 * The bit 22/21 of 0x42004
6909	 * The bit 5 of 0x42020
6910	 * The bit 15 of 0x45000
6911	 */
6912	I915_WRITE(ILK_DISPLAY_CHICKEN2,
6913		   (I915_READ(ILK_DISPLAY_CHICKEN2) |
6914		    ILK_DPARB_GATE | ILK_VSDPFD_FULL));
6915	dspclk_gate |= ILK_DPARBUNIT_CLOCK_GATE_ENABLE;
6916	I915_WRITE(DISP_ARB_CTL,
6917		   (I915_READ(DISP_ARB_CTL) |
6918		    DISP_FBC_WM_DIS));
6919
6920	/*
6921	 * Based on the document from hardware guys the following bits
6922	 * should be set unconditionally in order to enable FBC.
6923	 * The bit 22 of 0x42000
6924	 * The bit 22 of 0x42004
6925	 * The bit 7,8,9 of 0x42020.
6926	 */
6927	if (IS_IRONLAKE_M(dev_priv)) {
6928		/* WaFbcAsynchFlipDisableFbcQueue:ilk */
6929		I915_WRITE(ILK_DISPLAY_CHICKEN1,
6930			   I915_READ(ILK_DISPLAY_CHICKEN1) |
6931			   ILK_FBCQ_DIS);
6932		I915_WRITE(ILK_DISPLAY_CHICKEN2,
6933			   I915_READ(ILK_DISPLAY_CHICKEN2) |
6934			   ILK_DPARB_GATE);
6935	}
6936
6937	I915_WRITE(ILK_DSPCLK_GATE_D, dspclk_gate);
6938
6939	I915_WRITE(ILK_DISPLAY_CHICKEN2,
6940		   I915_READ(ILK_DISPLAY_CHICKEN2) |
6941		   ILK_ELPIN_409_SELECT);
6942
6943	g4x_disable_trickle_feed(dev_priv);
6944
6945	ibx_init_clock_gating(dev_priv);
6946}
6947
6948static void cpt_init_clock_gating(struct drm_i915_private *dev_priv)
6949{
6950	enum pipe pipe;
6951	u32 val;
6952
6953	/*
6954	 * On Ibex Peak and Cougar Point, we need to disable clock
6955	 * gating for the panel power sequencer or it will fail to
6956	 * start up when no ports are active.
6957	 */
6958	I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE |
6959		   PCH_DPLUNIT_CLOCK_GATE_DISABLE |
6960		   PCH_CPUNIT_CLOCK_GATE_DISABLE);
6961	I915_WRITE(SOUTH_CHICKEN2, I915_READ(SOUTH_CHICKEN2) |
6962		   DPLS_EDP_PPS_FIX_DIS);
6963	/* The below fixes the weird display corruption, a few pixels shifted
6964	 * downward, on (only) LVDS of some HP laptops with IVY.
6965	 */
6966	for_each_pipe(dev_priv, pipe) {
6967		val = I915_READ(TRANS_CHICKEN2(pipe));
6968		val |= TRANS_CHICKEN2_TIMING_OVERRIDE;
6969		val &= ~TRANS_CHICKEN2_FDI_POLARITY_REVERSED;
6970		if (dev_priv->vbt.fdi_rx_polarity_inverted)
6971			val |= TRANS_CHICKEN2_FDI_POLARITY_REVERSED;
6972		val &= ~TRANS_CHICKEN2_DISABLE_DEEP_COLOR_COUNTER;
6973		val &= ~TRANS_CHICKEN2_DISABLE_DEEP_COLOR_MODESWITCH;
6974		I915_WRITE(TRANS_CHICKEN2(pipe), val);
6975	}
6976	/* WADP0ClockGatingDisable */
6977	for_each_pipe(dev_priv, pipe) {
6978		I915_WRITE(TRANS_CHICKEN1(pipe),
6979			   TRANS_CHICKEN1_DP0UNIT_GC_DISABLE);
6980	}
6981}
6982
6983static void gen6_check_mch_setup(struct drm_i915_private *dev_priv)
6984{
6985	u32 tmp;
6986
6987	tmp = I915_READ(MCH_SSKPD);
6988	if ((tmp & MCH_SSKPD_WM0_MASK) != MCH_SSKPD_WM0_VAL)
6989		drm_dbg_kms(&dev_priv->drm,
6990			    "Wrong MCH_SSKPD value: 0x%08x This can cause underruns.\n",
6991			    tmp);
6992}
6993
6994static void gen6_init_clock_gating(struct drm_i915_private *dev_priv)
6995{
6996	u32 dspclk_gate = ILK_VRHUNIT_CLOCK_GATE_DISABLE;
6997
6998	I915_WRITE(ILK_DSPCLK_GATE_D, dspclk_gate);
6999
7000	I915_WRITE(ILK_DISPLAY_CHICKEN2,
7001		   I915_READ(ILK_DISPLAY_CHICKEN2) |
7002		   ILK_ELPIN_409_SELECT);
7003
7004	I915_WRITE(GEN6_UCGCTL1,
7005		   I915_READ(GEN6_UCGCTL1) |
7006		   GEN6_BLBUNIT_CLOCK_GATE_DISABLE |
7007		   GEN6_CSUNIT_CLOCK_GATE_DISABLE);
7008
7009	/* According to the BSpec vol1g, bit 12 (RCPBUNIT) clock
7010	 * gating disable must be set.  Failure to set it results in
7011	 * flickering pixels due to Z write ordering failures after
7012	 * some amount of runtime in the Mesa "fire" demo, and Unigine
7013	 * Sanctuary and Tropics, and apparently anything else with
7014	 * alpha test or pixel discard.
7015	 *
7016	 * According to the spec, bit 11 (RCCUNIT) must also be set,
7017	 * but we didn't debug actual testcases to find it out.
7018	 *
7019	 * WaDisableRCCUnitClockGating:snb
7020	 * WaDisableRCPBUnitClockGating:snb
7021	 */
7022	I915_WRITE(GEN6_UCGCTL2,
7023		   GEN6_RCPBUNIT_CLOCK_GATE_DISABLE |
7024		   GEN6_RCCUNIT_CLOCK_GATE_DISABLE);
7025
7026	/*
7027	 * According to the spec the following bits should be
7028	 * set in order to enable memory self-refresh and fbc:
7029	 * The bit21 and bit22 of 0x42000
7030	 * The bit21 and bit22 of 0x42004
7031	 * The bit5 and bit7 of 0x42020
7032	 * The bit14 of 0x70180
7033	 * The bit14 of 0x71180
7034	 *
7035	 * WaFbcAsynchFlipDisableFbcQueue:snb
7036	 */
7037	I915_WRITE(ILK_DISPLAY_CHICKEN1,
7038		   I915_READ(ILK_DISPLAY_CHICKEN1) |
7039		   ILK_FBCQ_DIS | ILK_PABSTRETCH_DIS);
7040	I915_WRITE(ILK_DISPLAY_CHICKEN2,
7041		   I915_READ(ILK_DISPLAY_CHICKEN2) |
7042		   ILK_DPARB_GATE | ILK_VSDPFD_FULL);
7043	I915_WRITE(ILK_DSPCLK_GATE_D,
7044		   I915_READ(ILK_DSPCLK_GATE_D) |
7045		   ILK_DPARBUNIT_CLOCK_GATE_ENABLE  |
7046		   ILK_DPFDUNIT_CLOCK_GATE_ENABLE);
7047
7048	g4x_disable_trickle_feed(dev_priv);
7049
7050	cpt_init_clock_gating(dev_priv);
7051
7052	gen6_check_mch_setup(dev_priv);
7053}
7054
7055static void lpt_init_clock_gating(struct drm_i915_private *dev_priv)
7056{
7057	/*
7058	 * TODO: this bit should only be enabled when really needed, then
7059	 * disabled when not needed anymore in order to save power.
7060	 */
7061	if (HAS_PCH_LPT_LP(dev_priv))
7062		I915_WRITE(SOUTH_DSPCLK_GATE_D,
7063			   I915_READ(SOUTH_DSPCLK_GATE_D) |
7064			   PCH_LP_PARTITION_LEVEL_DISABLE);
7065
7066	/* WADPOClockGatingDisable:hsw */
7067	I915_WRITE(TRANS_CHICKEN1(PIPE_A),
7068		   I915_READ(TRANS_CHICKEN1(PIPE_A)) |
7069		   TRANS_CHICKEN1_DP0UNIT_GC_DISABLE);
7070}
7071
7072static void lpt_suspend_hw(struct drm_i915_private *dev_priv)
7073{
7074	if (HAS_PCH_LPT_LP(dev_priv)) {
7075		u32 val = I915_READ(SOUTH_DSPCLK_GATE_D);
7076
7077		val &= ~PCH_LP_PARTITION_LEVEL_DISABLE;
7078		I915_WRITE(SOUTH_DSPCLK_GATE_D, val);
7079	}
7080}
7081
7082static void gen8_set_l3sqc_credits(struct drm_i915_private *dev_priv,
7083				   int general_prio_credits,
7084				   int high_prio_credits)
7085{
7086	u32 misccpctl;
7087	u32 val;
7088
7089	/* WaTempDisableDOPClkGating:bdw */
7090	misccpctl = I915_READ(GEN7_MISCCPCTL);
7091	I915_WRITE(GEN7_MISCCPCTL, misccpctl & ~GEN7_DOP_CLOCK_GATE_ENABLE);
7092
7093	val = I915_READ(GEN8_L3SQCREG1);
7094	val &= ~L3_PRIO_CREDITS_MASK;
7095	val |= L3_GENERAL_PRIO_CREDITS(general_prio_credits);
7096	val |= L3_HIGH_PRIO_CREDITS(high_prio_credits);
7097	I915_WRITE(GEN8_L3SQCREG1, val);
7098
7099	/*
7100	 * Wait at least 100 clocks before re-enabling clock gating.
7101	 * See the definition of L3SQCREG1 in BSpec.
7102	 */
7103	POSTING_READ(GEN8_L3SQCREG1);
7104	udelay(1);
7105	I915_WRITE(GEN7_MISCCPCTL, misccpctl);
7106}
7107
7108static void icl_init_clock_gating(struct drm_i915_private *dev_priv)
7109{
7110	/* Wa_1409120013:icl,ehl */
7111	I915_WRITE(ILK_DPFC_CHICKEN,
7112		   ILK_DPFC_CHICKEN_COMP_DUMMY_PIXEL);
7113
7114	/* This is not an Wa. Enable to reduce Sampler power */
7115	I915_WRITE(GEN10_DFR_RATIO_EN_AND_CHICKEN,
7116		   I915_READ(GEN10_DFR_RATIO_EN_AND_CHICKEN) & ~DFR_DISABLE);
7117
7118	/*Wa_14010594013:icl, ehl */
7119	intel_uncore_rmw(&dev_priv->uncore, GEN8_CHICKEN_DCPR_1,
7120			 0, CNL_DELAY_PMRSP);
7121}
7122
7123static void tgl_init_clock_gating(struct drm_i915_private *dev_priv)
7124{
7125	/* Wa_1409120013:tgl */
7126	I915_WRITE(ILK_DPFC_CHICKEN,
7127		   ILK_DPFC_CHICKEN_COMP_DUMMY_PIXEL);
7128
7129	/* Wa_1409825376:tgl (pre-prod)*/
7130	if (IS_TGL_DISP_REVID(dev_priv, TGL_REVID_A0, TGL_REVID_B1))
7131		I915_WRITE(GEN9_CLKGATE_DIS_3, I915_READ(GEN9_CLKGATE_DIS_3) |
7132			   TGL_VRH_GATING_DIS);
7133
7134	/* Wa_14011059788:tgl */
7135	intel_uncore_rmw(&dev_priv->uncore, GEN10_DFR_RATIO_EN_AND_CHICKEN,
7136			 0, DFR_DISABLE);
7137}
7138
7139static void cnp_init_clock_gating(struct drm_i915_private *dev_priv)
7140{
7141	if (!HAS_PCH_CNP(dev_priv))
7142		return;
7143
7144	/* Display WA #1181 WaSouthDisplayDisablePWMCGEGating: cnp */
7145	I915_WRITE(SOUTH_DSPCLK_GATE_D, I915_READ(SOUTH_DSPCLK_GATE_D) |
7146		   CNP_PWM_CGE_GATING_DISABLE);
7147}
7148
7149static void cnl_init_clock_gating(struct drm_i915_private *dev_priv)
7150{
7151	u32 val;
7152	cnp_init_clock_gating(dev_priv);
7153
7154	/* This is not an Wa. Enable for better image quality */
7155	I915_WRITE(_3D_CHICKEN3,
7156		   _MASKED_BIT_ENABLE(_3D_CHICKEN3_AA_LINE_QUALITY_FIX_ENABLE));
7157
7158	/* WaEnableChickenDCPR:cnl */
7159	I915_WRITE(GEN8_CHICKEN_DCPR_1,
7160		   I915_READ(GEN8_CHICKEN_DCPR_1) | MASK_WAKEMEM);
7161
7162	/*
7163	 * WaFbcWakeMemOn:cnl
7164	 * Display WA #0859: cnl
7165	 */
7166	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
7167		   DISP_FBC_MEMORY_WAKE);
7168
7169	val = I915_READ(SLICE_UNIT_LEVEL_CLKGATE);
7170	/* ReadHitWriteOnlyDisable:cnl */
7171	val |= RCCUNIT_CLKGATE_DIS;
7172	I915_WRITE(SLICE_UNIT_LEVEL_CLKGATE, val);
7173
7174	/* Wa_2201832410:cnl */
7175	val = I915_READ(SUBSLICE_UNIT_LEVEL_CLKGATE);
7176	val |= GWUNIT_CLKGATE_DIS;
7177	I915_WRITE(SUBSLICE_UNIT_LEVEL_CLKGATE, val);
7178
7179	/* WaDisableVFclkgate:cnl */
7180	/* WaVFUnitClockGatingDisable:cnl */
7181	val = I915_READ(UNSLICE_UNIT_LEVEL_CLKGATE);
7182	val |= VFUNIT_CLKGATE_DIS;
7183	I915_WRITE(UNSLICE_UNIT_LEVEL_CLKGATE, val);
7184}
7185
7186static void cfl_init_clock_gating(struct drm_i915_private *dev_priv)
7187{
7188	cnp_init_clock_gating(dev_priv);
7189	gen9_init_clock_gating(dev_priv);
7190
7191	/*
7192	 * WaFbcTurnOffFbcWatermark:cfl
7193	 * Display WA #0562: cfl
7194	 */
7195	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
7196		   DISP_FBC_WM_DIS);
7197
7198	/*
7199	 * WaFbcNukeOnHostModify:cfl
7200	 * Display WA #0873: cfl
7201	 */
7202	I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
7203		   ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
7204}
7205
7206static void kbl_init_clock_gating(struct drm_i915_private *dev_priv)
7207{
7208	gen9_init_clock_gating(dev_priv);
7209
7210	/* WaDisableSDEUnitClockGating:kbl */
7211	if (IS_KBL_GT_REVID(dev_priv, 0, KBL_REVID_B0))
7212		I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
7213			   GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
7214
7215	/* WaDisableGamClockGating:kbl */
7216	if (IS_KBL_GT_REVID(dev_priv, 0, KBL_REVID_B0))
7217		I915_WRITE(GEN6_UCGCTL1, I915_READ(GEN6_UCGCTL1) |
7218			   GEN6_GAMUNIT_CLOCK_GATE_DISABLE);
7219
7220	/*
7221	 * WaFbcTurnOffFbcWatermark:kbl
7222	 * Display WA #0562: kbl
7223	 */
7224	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
7225		   DISP_FBC_WM_DIS);
7226
7227	/*
7228	 * WaFbcNukeOnHostModify:kbl
7229	 * Display WA #0873: kbl
7230	 */
7231	I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
7232		   ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
7233}
7234
7235static void skl_init_clock_gating(struct drm_i915_private *dev_priv)
7236{
7237	gen9_init_clock_gating(dev_priv);
7238
7239	/* WaDisableDopClockGating:skl */
7240	I915_WRITE(GEN7_MISCCPCTL, I915_READ(GEN7_MISCCPCTL) &
7241		   ~GEN7_DOP_CLOCK_GATE_ENABLE);
7242
7243	/* WAC6entrylatency:skl */
7244	I915_WRITE(FBC_LLC_READ_CTRL, I915_READ(FBC_LLC_READ_CTRL) |
7245		   FBC_LLC_FULLY_OPEN);
7246
7247	/*
7248	 * WaFbcTurnOffFbcWatermark:skl
7249	 * Display WA #0562: skl
7250	 */
7251	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
7252		   DISP_FBC_WM_DIS);
7253
7254	/*
7255	 * WaFbcNukeOnHostModify:skl
7256	 * Display WA #0873: skl
7257	 */
7258	I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
7259		   ILK_DPFC_NUKE_ON_ANY_MODIFICATION);
7260
7261	/*
7262	 * WaFbcHighMemBwCorruptionAvoidance:skl
7263	 * Display WA #0883: skl
7264	 */
7265	I915_WRITE(ILK_DPFC_CHICKEN, I915_READ(ILK_DPFC_CHICKEN) |
7266		   ILK_DPFC_DISABLE_DUMMY0);
7267}
7268
7269static void bdw_init_clock_gating(struct drm_i915_private *dev_priv)
7270{
7271	enum pipe pipe;
7272
7273	/* WaFbcAsynchFlipDisableFbcQueue:hsw,bdw */
7274	I915_WRITE(CHICKEN_PIPESL_1(PIPE_A),
7275		   I915_READ(CHICKEN_PIPESL_1(PIPE_A)) |
7276		   HSW_FBCQ_DIS);
7277
7278	/* WaSwitchSolVfFArbitrationPriority:bdw */
7279	I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) | HSW_ECOCHK_ARB_PRIO_SOL);
7280
7281	/* WaPsrDPAMaskVBlankInSRD:bdw */
7282	I915_WRITE(CHICKEN_PAR1_1,
7283		   I915_READ(CHICKEN_PAR1_1) | DPA_MASK_VBLANK_SRD);
7284
7285	/* WaPsrDPRSUnmaskVBlankInSRD:bdw */
7286	for_each_pipe(dev_priv, pipe) {
7287		I915_WRITE(CHICKEN_PIPESL_1(pipe),
7288			   I915_READ(CHICKEN_PIPESL_1(pipe)) |
7289			   BDW_DPRS_MASK_VBLANK_SRD);
7290	}
7291
7292	/* WaVSRefCountFullforceMissDisable:bdw */
7293	/* WaDSRefCountFullforceMissDisable:bdw */
7294	I915_WRITE(GEN7_FF_THREAD_MODE,
7295		   I915_READ(GEN7_FF_THREAD_MODE) &
7296		   ~(GEN8_FF_DS_REF_CNT_FFME | GEN7_FF_VS_REF_CNT_FFME));
7297
7298	I915_WRITE(GEN6_RC_SLEEP_PSMI_CONTROL,
7299		   _MASKED_BIT_ENABLE(GEN8_RC_SEMA_IDLE_MSG_DISABLE));
7300
7301	/* WaDisableSDEUnitClockGating:bdw */
7302	I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
7303		   GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
7304
7305	/* WaProgramL3SqcReg1Default:bdw */
7306	gen8_set_l3sqc_credits(dev_priv, 30, 2);
7307
7308	/* WaKVMNotificationOnConfigChange:bdw */
7309	I915_WRITE(CHICKEN_PAR2_1, I915_READ(CHICKEN_PAR2_1)
7310		   | KVM_CONFIG_CHANGE_NOTIFICATION_SELECT);
7311
7312	lpt_init_clock_gating(dev_priv);
7313
7314	/* WaDisableDopClockGating:bdw
7315	 *
7316	 * Also see the CHICKEN2 write in bdw_init_workarounds() to disable DOP
7317	 * clock gating.
7318	 */
7319	I915_WRITE(GEN6_UCGCTL1,
7320		   I915_READ(GEN6_UCGCTL1) | GEN6_EU_TCUNIT_CLOCK_GATE_DISABLE);
7321}
7322
7323static void hsw_init_clock_gating(struct drm_i915_private *dev_priv)
7324{
7325	/* WaFbcAsynchFlipDisableFbcQueue:hsw,bdw */
7326	I915_WRITE(CHICKEN_PIPESL_1(PIPE_A),
7327		   I915_READ(CHICKEN_PIPESL_1(PIPE_A)) |
7328		   HSW_FBCQ_DIS);
7329
7330	/* This is required by WaCatErrorRejectionIssue:hsw */
7331	I915_WRITE(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG,
7332		   I915_READ(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG) |
7333		   GEN7_SQ_CHICKEN_MBCUNIT_SQINTMOB);
7334
7335	/* WaSwitchSolVfFArbitrationPriority:hsw */
7336	I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) | HSW_ECOCHK_ARB_PRIO_SOL);
7337
7338	lpt_init_clock_gating(dev_priv);
7339}
7340
7341static void ivb_init_clock_gating(struct drm_i915_private *dev_priv)
7342{
7343	u32 snpcr;
7344
7345	I915_WRITE(ILK_DSPCLK_GATE_D, ILK_VRHUNIT_CLOCK_GATE_DISABLE);
7346
7347	/* WaFbcAsynchFlipDisableFbcQueue:ivb */
7348	I915_WRITE(ILK_DISPLAY_CHICKEN1,
7349		   I915_READ(ILK_DISPLAY_CHICKEN1) |
7350		   ILK_FBCQ_DIS);
7351
7352	/* WaDisableBackToBackFlipFix:ivb */
7353	I915_WRITE(IVB_CHICKEN3,
7354		   CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE |
7355		   CHICKEN3_DGMG_DONE_FIX_DISABLE);
7356
7357	if (IS_IVB_GT1(dev_priv))
7358		I915_WRITE(GEN7_ROW_CHICKEN2,
7359			   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
7360	else {
7361		/* must write both registers */
7362		I915_WRITE(GEN7_ROW_CHICKEN2,
7363			   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
7364		I915_WRITE(GEN7_ROW_CHICKEN2_GT2,
7365			   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
7366	}
7367
7368	/*
7369	 * According to the spec, bit 13 (RCZUNIT) must be set on IVB.
7370	 * This implements the WaDisableRCZUnitClockGating:ivb workaround.
7371	 */
7372	I915_WRITE(GEN6_UCGCTL2,
7373		   GEN6_RCZUNIT_CLOCK_GATE_DISABLE);
7374
7375	/* This is required by WaCatErrorRejectionIssue:ivb */
7376	I915_WRITE(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG,
7377			I915_READ(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG) |
7378			GEN7_SQ_CHICKEN_MBCUNIT_SQINTMOB);
7379
7380	g4x_disable_trickle_feed(dev_priv);
7381
7382	snpcr = I915_READ(GEN6_MBCUNIT_SNPCR);
7383	snpcr &= ~GEN6_MBC_SNPCR_MASK;
7384	snpcr |= GEN6_MBC_SNPCR_MED;
7385	I915_WRITE(GEN6_MBCUNIT_SNPCR, snpcr);
7386
7387	if (!HAS_PCH_NOP(dev_priv))
7388		cpt_init_clock_gating(dev_priv);
7389
7390	gen6_check_mch_setup(dev_priv);
7391}
7392
7393static void vlv_init_clock_gating(struct drm_i915_private *dev_priv)
7394{
7395	/* WaDisableBackToBackFlipFix:vlv */
7396	I915_WRITE(IVB_CHICKEN3,
7397		   CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE |
7398		   CHICKEN3_DGMG_DONE_FIX_DISABLE);
7399
7400	/* WaDisableDopClockGating:vlv */
7401	I915_WRITE(GEN7_ROW_CHICKEN2,
7402		   _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE));
7403
7404	/* This is required by WaCatErrorRejectionIssue:vlv */
7405	I915_WRITE(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG,
7406		   I915_READ(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG) |
7407		   GEN7_SQ_CHICKEN_MBCUNIT_SQINTMOB);
7408
7409	/*
7410	 * According to the spec, bit 13 (RCZUNIT) must be set on IVB.
7411	 * This implements the WaDisableRCZUnitClockGating:vlv workaround.
7412	 */
7413	I915_WRITE(GEN6_UCGCTL2,
7414		   GEN6_RCZUNIT_CLOCK_GATE_DISABLE);
7415
7416	/* WaDisableL3Bank2xClockGate:vlv
7417	 * Disabling L3 clock gating- MMIO 940c[25] = 1
7418	 * Set bit 25, to disable L3_BANK_2x_CLK_GATING */
7419	I915_WRITE(GEN7_UCGCTL4,
7420		   I915_READ(GEN7_UCGCTL4) | GEN7_L3BANK2X_CLOCK_GATE_DISABLE);
7421
7422	/*
7423	 * WaDisableVLVClockGating_VBIIssue:vlv
7424	 * Disable clock gating on th GCFG unit to prevent a delay
7425	 * in the reporting of vblank events.
7426	 */
7427	I915_WRITE(VLV_GUNIT_CLOCK_GATE, GCFG_DIS);
7428}
7429
7430static void chv_init_clock_gating(struct drm_i915_private *dev_priv)
7431{
7432	/* WaVSRefCountFullforceMissDisable:chv */
7433	/* WaDSRefCountFullforceMissDisable:chv */
7434	I915_WRITE(GEN7_FF_THREAD_MODE,
7435		   I915_READ(GEN7_FF_THREAD_MODE) &
7436		   ~(GEN8_FF_DS_REF_CNT_FFME | GEN7_FF_VS_REF_CNT_FFME));
7437
7438	/* WaDisableSemaphoreAndSyncFlipWait:chv */
7439	I915_WRITE(GEN6_RC_SLEEP_PSMI_CONTROL,
7440		   _MASKED_BIT_ENABLE(GEN8_RC_SEMA_IDLE_MSG_DISABLE));
7441
7442	/* WaDisableCSUnitClockGating:chv */
7443	I915_WRITE(GEN6_UCGCTL1, I915_READ(GEN6_UCGCTL1) |
7444		   GEN6_CSUNIT_CLOCK_GATE_DISABLE);
7445
7446	/* WaDisableSDEUnitClockGating:chv */
7447	I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) |
7448		   GEN8_SDEUNIT_CLOCK_GATE_DISABLE);
7449
7450	/*
7451	 * WaProgramL3SqcReg1Default:chv
7452	 * See gfxspecs/Related Documents/Performance Guide/
7453	 * LSQC Setting Recommendations.
7454	 */
7455	gen8_set_l3sqc_credits(dev_priv, 38, 2);
7456}
7457
7458static void g4x_init_clock_gating(struct drm_i915_private *dev_priv)
7459{
7460	u32 dspclk_gate;
7461
7462	I915_WRITE(RENCLK_GATE_D1, 0);
7463	I915_WRITE(RENCLK_GATE_D2, VF_UNIT_CLOCK_GATE_DISABLE |
7464		   GS_UNIT_CLOCK_GATE_DISABLE |
7465		   CL_UNIT_CLOCK_GATE_DISABLE);
7466	I915_WRITE(RAMCLK_GATE_D, 0);
7467	dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE |
7468		OVRUNIT_CLOCK_GATE_DISABLE |
7469		OVCUNIT_CLOCK_GATE_DISABLE;
7470	if (IS_GM45(dev_priv))
7471		dspclk_gate |= DSSUNIT_CLOCK_GATE_DISABLE;
7472	I915_WRITE(DSPCLK_GATE_D, dspclk_gate);
7473
7474	g4x_disable_trickle_feed(dev_priv);
7475}
7476
7477static void i965gm_init_clock_gating(struct drm_i915_private *dev_priv)
7478{
7479	struct intel_uncore *uncore = &dev_priv->uncore;
7480
7481	intel_uncore_write(uncore, RENCLK_GATE_D1, I965_RCC_CLOCK_GATE_DISABLE);
7482	intel_uncore_write(uncore, RENCLK_GATE_D2, 0);
7483	intel_uncore_write(uncore, DSPCLK_GATE_D, 0);
7484	intel_uncore_write(uncore, RAMCLK_GATE_D, 0);
7485	intel_uncore_write16(uncore, DEUC, 0);
7486	intel_uncore_write(uncore,
7487			   MI_ARB_STATE,
7488			   _MASKED_BIT_ENABLE(MI_ARB_DISPLAY_TRICKLE_FEED_DISABLE));
7489}
7490
7491static void i965g_init_clock_gating(struct drm_i915_private *dev_priv)
7492{
7493	I915_WRITE(RENCLK_GATE_D1, I965_RCZ_CLOCK_GATE_DISABLE |
7494		   I965_RCC_CLOCK_GATE_DISABLE |
7495		   I965_RCPB_CLOCK_GATE_DISABLE |
7496		   I965_ISC_CLOCK_GATE_DISABLE |
7497		   I965_FBC_CLOCK_GATE_DISABLE);
7498	I915_WRITE(RENCLK_GATE_D2, 0);
7499	I915_WRITE(MI_ARB_STATE,
7500		   _MASKED_BIT_ENABLE(MI_ARB_DISPLAY_TRICKLE_FEED_DISABLE));
7501}
7502
7503static void gen3_init_clock_gating(struct drm_i915_private *dev_priv)
7504{
7505	u32 dstate = I915_READ(D_STATE);
7506
7507	dstate |= DSTATE_PLL_D3_OFF | DSTATE_GFX_CLOCK_GATING |
7508		DSTATE_DOT_CLOCK_GATING;
7509	I915_WRITE(D_STATE, dstate);
7510
7511	if (IS_PINEVIEW(dev_priv))
7512		I915_WRITE(ECOSKPD, _MASKED_BIT_ENABLE(ECO_GATING_CX_ONLY));
7513
7514	/* IIR "flip pending" means done if this bit is set */
7515	I915_WRITE(ECOSKPD, _MASKED_BIT_DISABLE(ECO_FLIP_DONE));
7516
7517	/* interrupts should cause a wake up from C3 */
7518	I915_WRITE(INSTPM, _MASKED_BIT_ENABLE(INSTPM_AGPBUSY_INT_EN));
7519
7520	/* On GEN3 we really need to make sure the ARB C3 LP bit is set */
7521	I915_WRITE(MI_ARB_STATE, _MASKED_BIT_ENABLE(MI_ARB_C3_LP_WRITE_ENABLE));
7522
7523	I915_WRITE(MI_ARB_STATE,
7524		   _MASKED_BIT_ENABLE(MI_ARB_DISPLAY_TRICKLE_FEED_DISABLE));
7525}
7526
7527static void i85x_init_clock_gating(struct drm_i915_private *dev_priv)
7528{
7529	I915_WRITE(RENCLK_GATE_D1, SV_CLOCK_GATE_DISABLE);
7530
7531	/* interrupts should cause a wake up from C3 */
7532	I915_WRITE(MI_STATE, _MASKED_BIT_ENABLE(MI_AGPBUSY_INT_EN) |
7533		   _MASKED_BIT_DISABLE(MI_AGPBUSY_830_MODE));
7534
7535	I915_WRITE(MEM_MODE,
7536		   _MASKED_BIT_ENABLE(MEM_DISPLAY_TRICKLE_FEED_DISABLE));
7537
7538	/*
7539	 * Have FBC ignore 3D activity since we use software
7540	 * render tracking, and otherwise a pure 3D workload
7541	 * (even if it just renders a single frame and then does
7542	 * abosultely nothing) would not allow FBC to recompress
7543	 * until a 2D blit occurs.
7544	 */
7545	I915_WRITE(SCPD0,
7546		   _MASKED_BIT_ENABLE(SCPD_FBC_IGNORE_3D));
7547}
7548
7549static void i830_init_clock_gating(struct drm_i915_private *dev_priv)
7550{
7551	I915_WRITE(MEM_MODE,
7552		   _MASKED_BIT_ENABLE(MEM_DISPLAY_A_TRICKLE_FEED_DISABLE) |
7553		   _MASKED_BIT_ENABLE(MEM_DISPLAY_B_TRICKLE_FEED_DISABLE));
7554}
7555
7556void intel_init_clock_gating(struct drm_i915_private *dev_priv)
7557{
7558	dev_priv->display.init_clock_gating(dev_priv);
7559}
7560
7561void intel_suspend_hw(struct drm_i915_private *dev_priv)
7562{
7563	if (HAS_PCH_LPT(dev_priv))
7564		lpt_suspend_hw(dev_priv);
7565}
7566
7567static void nop_init_clock_gating(struct drm_i915_private *dev_priv)
7568{
7569	drm_dbg_kms(&dev_priv->drm,
7570		    "No clock gating settings or workarounds applied.\n");
7571}
7572
7573/**
7574 * intel_init_clock_gating_hooks - setup the clock gating hooks
7575 * @dev_priv: device private
7576 *
7577 * Setup the hooks that configure which clocks of a given platform can be
7578 * gated and also apply various GT and display specific workarounds for these
7579 * platforms. Note that some GT specific workarounds are applied separately
7580 * when GPU contexts or batchbuffers start their execution.
7581 */
7582void intel_init_clock_gating_hooks(struct drm_i915_private *dev_priv)
7583{
7584	if (IS_GEN(dev_priv, 12))
7585		dev_priv->display.init_clock_gating = tgl_init_clock_gating;
7586	else if (IS_GEN(dev_priv, 11))
7587		dev_priv->display.init_clock_gating = icl_init_clock_gating;
7588	else if (IS_CANNONLAKE(dev_priv))
7589		dev_priv->display.init_clock_gating = cnl_init_clock_gating;
7590	else if (IS_COFFEELAKE(dev_priv) || IS_COMETLAKE(dev_priv))
7591		dev_priv->display.init_clock_gating = cfl_init_clock_gating;
7592	else if (IS_SKYLAKE(dev_priv))
7593		dev_priv->display.init_clock_gating = skl_init_clock_gating;
7594	else if (IS_KABYLAKE(dev_priv))
7595		dev_priv->display.init_clock_gating = kbl_init_clock_gating;
7596	else if (IS_BROXTON(dev_priv))
7597		dev_priv->display.init_clock_gating = bxt_init_clock_gating;
7598	else if (IS_GEMINILAKE(dev_priv))
7599		dev_priv->display.init_clock_gating = glk_init_clock_gating;
7600	else if (IS_BROADWELL(dev_priv))
7601		dev_priv->display.init_clock_gating = bdw_init_clock_gating;
7602	else if (IS_CHERRYVIEW(dev_priv))
7603		dev_priv->display.init_clock_gating = chv_init_clock_gating;
7604	else if (IS_HASWELL(dev_priv))
7605		dev_priv->display.init_clock_gating = hsw_init_clock_gating;
7606	else if (IS_IVYBRIDGE(dev_priv))
7607		dev_priv->display.init_clock_gating = ivb_init_clock_gating;
7608	else if (IS_VALLEYVIEW(dev_priv))
7609		dev_priv->display.init_clock_gating = vlv_init_clock_gating;
7610	else if (IS_GEN(dev_priv, 6))
7611		dev_priv->display.init_clock_gating = gen6_init_clock_gating;
7612	else if (IS_GEN(dev_priv, 5))
7613		dev_priv->display.init_clock_gating = ilk_init_clock_gating;
7614	else if (IS_G4X(dev_priv))
7615		dev_priv->display.init_clock_gating = g4x_init_clock_gating;
7616	else if (IS_I965GM(dev_priv))
7617		dev_priv->display.init_clock_gating = i965gm_init_clock_gating;
7618	else if (IS_I965G(dev_priv))
7619		dev_priv->display.init_clock_gating = i965g_init_clock_gating;
7620	else if (IS_GEN(dev_priv, 3))
7621		dev_priv->display.init_clock_gating = gen3_init_clock_gating;
7622	else if (IS_I85X(dev_priv) || IS_I865G(dev_priv))
7623		dev_priv->display.init_clock_gating = i85x_init_clock_gating;
7624	else if (IS_GEN(dev_priv, 2))
7625		dev_priv->display.init_clock_gating = i830_init_clock_gating;
7626	else {
7627		MISSING_CASE(INTEL_DEVID(dev_priv));
7628		dev_priv->display.init_clock_gating = nop_init_clock_gating;
7629	}
7630}
7631
7632/* Set up chip specific power management-related functions */
7633void intel_init_pm(struct drm_i915_private *dev_priv)
7634{
7635	/* For cxsr */
7636	if (IS_PINEVIEW(dev_priv))
7637		pnv_get_mem_freq(dev_priv);
7638	else if (IS_GEN(dev_priv, 5))
7639		ilk_get_mem_freq(dev_priv);
7640
7641	if (intel_has_sagv(dev_priv))
7642		skl_setup_sagv_block_time(dev_priv);
7643
7644	/* For FIFO watermark updates */
7645	if (INTEL_GEN(dev_priv) >= 9) {
7646		skl_setup_wm_latency(dev_priv);
7647		dev_priv->display.compute_global_watermarks = skl_compute_wm;
7648	} else if (HAS_PCH_SPLIT(dev_priv)) {
7649		ilk_setup_wm_latency(dev_priv);
7650
7651		if ((IS_GEN(dev_priv, 5) && dev_priv->wm.pri_latency[1] &&
7652		     dev_priv->wm.spr_latency[1] && dev_priv->wm.cur_latency[1]) ||
7653		    (!IS_GEN(dev_priv, 5) && dev_priv->wm.pri_latency[0] &&
7654		     dev_priv->wm.spr_latency[0] && dev_priv->wm.cur_latency[0])) {
7655			dev_priv->display.compute_pipe_wm = ilk_compute_pipe_wm;
7656			dev_priv->display.compute_intermediate_wm =
7657				ilk_compute_intermediate_wm;
7658			dev_priv->display.initial_watermarks =
7659				ilk_initial_watermarks;
7660			dev_priv->display.optimize_watermarks =
7661				ilk_optimize_watermarks;
7662		} else {
7663			drm_dbg_kms(&dev_priv->drm,
7664				    "Failed to read display plane latency. "
7665				    "Disable CxSR\n");
7666		}
7667	} else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) {
7668		vlv_setup_wm_latency(dev_priv);
7669		dev_priv->display.compute_pipe_wm = vlv_compute_pipe_wm;
7670		dev_priv->display.compute_intermediate_wm = vlv_compute_intermediate_wm;
7671		dev_priv->display.initial_watermarks = vlv_initial_watermarks;
7672		dev_priv->display.optimize_watermarks = vlv_optimize_watermarks;
7673		dev_priv->display.atomic_update_watermarks = vlv_atomic_update_fifo;
7674	} else if (IS_G4X(dev_priv)) {
7675		g4x_setup_wm_latency(dev_priv);
7676		dev_priv->display.compute_pipe_wm = g4x_compute_pipe_wm;
7677		dev_priv->display.compute_intermediate_wm = g4x_compute_intermediate_wm;
7678		dev_priv->display.initial_watermarks = g4x_initial_watermarks;
7679		dev_priv->display.optimize_watermarks = g4x_optimize_watermarks;
7680	} else if (IS_PINEVIEW(dev_priv)) {
7681		if (!intel_get_cxsr_latency(!IS_MOBILE(dev_priv),
7682					    dev_priv->is_ddr3,
7683					    dev_priv->fsb_freq,
7684					    dev_priv->mem_freq)) {
7685			drm_info(&dev_priv->drm,
7686				 "failed to find known CxSR latency "
7687				 "(found ddr%s fsb freq %d, mem freq %d), "
7688				 "disabling CxSR\n",
7689				 (dev_priv->is_ddr3 == 1) ? "3" : "2",
7690				 dev_priv->fsb_freq, dev_priv->mem_freq);
7691			/* Disable CxSR and never update its watermark again */
7692			intel_set_memory_cxsr(dev_priv, false);
7693			dev_priv->display.update_wm = NULL;
7694		} else
7695			dev_priv->display.update_wm = pnv_update_wm;
7696	} else if (IS_GEN(dev_priv, 4)) {
7697		dev_priv->display.update_wm = i965_update_wm;
7698	} else if (IS_GEN(dev_priv, 3)) {
7699		dev_priv->display.update_wm = i9xx_update_wm;
7700		dev_priv->display.get_fifo_size = i9xx_get_fifo_size;
7701	} else if (IS_GEN(dev_priv, 2)) {
7702		if (INTEL_NUM_PIPES(dev_priv) == 1) {
7703			dev_priv->display.update_wm = i845_update_wm;
7704			dev_priv->display.get_fifo_size = i845_get_fifo_size;
7705		} else {
7706			dev_priv->display.update_wm = i9xx_update_wm;
7707			dev_priv->display.get_fifo_size = i830_get_fifo_size;
7708		}
7709	} else {
7710		drm_err(&dev_priv->drm,
7711			"unexpected fall-through in %s\n", __func__);
7712	}
7713}
7714
7715void intel_pm_setup(struct drm_i915_private *dev_priv)
7716{
7717	dev_priv->runtime_pm.suspended = false;
7718	atomic_set(&dev_priv->runtime_pm.wakeref_count, 0);
7719}
7720
7721static struct intel_global_state *intel_dbuf_duplicate_state(struct intel_global_obj *obj)
7722{
7723	struct intel_dbuf_state *dbuf_state;
7724
7725	dbuf_state = kmemdup(obj->state, sizeof(*dbuf_state), GFP_KERNEL);
7726	if (!dbuf_state)
7727		return NULL;
7728
7729	return &dbuf_state->base;
7730}
7731
7732static void intel_dbuf_destroy_state(struct intel_global_obj *obj,
7733				     struct intel_global_state *state)
7734{
7735	kfree(state);
7736}
7737
7738static const struct intel_global_state_funcs intel_dbuf_funcs = {
7739	.atomic_duplicate_state = intel_dbuf_duplicate_state,
7740	.atomic_destroy_state = intel_dbuf_destroy_state,
7741};
7742
7743struct intel_dbuf_state *
7744intel_atomic_get_dbuf_state(struct intel_atomic_state *state)
7745{
7746	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
7747	struct intel_global_state *dbuf_state;
7748
7749	dbuf_state = intel_atomic_get_global_obj_state(state, &dev_priv->dbuf.obj);
7750	if (IS_ERR(dbuf_state))
7751		return ERR_CAST(dbuf_state);
7752
7753	return to_intel_dbuf_state(dbuf_state);
7754}
7755
7756int intel_dbuf_init(struct drm_i915_private *dev_priv)
7757{
7758	struct intel_dbuf_state *dbuf_state;
7759
7760	dbuf_state = kzalloc(sizeof(*dbuf_state), GFP_KERNEL);
7761	if (!dbuf_state)
7762		return -ENOMEM;
7763
7764	intel_atomic_global_obj_init(dev_priv, &dev_priv->dbuf.obj,
7765				     &dbuf_state->base, &intel_dbuf_funcs);
7766
7767	return 0;
7768}
7769
7770void intel_dbuf_pre_plane_update(struct intel_atomic_state *state)
7771{
7772	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
7773	const struct intel_dbuf_state *new_dbuf_state =
7774		intel_atomic_get_new_dbuf_state(state);
7775	const struct intel_dbuf_state *old_dbuf_state =
7776		intel_atomic_get_old_dbuf_state(state);
7777
7778	if (!new_dbuf_state ||
7779	    new_dbuf_state->enabled_slices == old_dbuf_state->enabled_slices)
7780		return;
7781
7782	WARN_ON(!new_dbuf_state->base.changed);
7783
7784	gen9_dbuf_slices_update(dev_priv,
7785				old_dbuf_state->enabled_slices |
7786				new_dbuf_state->enabled_slices);
7787}
7788
7789void intel_dbuf_post_plane_update(struct intel_atomic_state *state)
7790{
7791	struct drm_i915_private *dev_priv = to_i915(state->base.dev);
7792	const struct intel_dbuf_state *new_dbuf_state =
7793		intel_atomic_get_new_dbuf_state(state);
7794	const struct intel_dbuf_state *old_dbuf_state =
7795		intel_atomic_get_old_dbuf_state(state);
7796
7797	if (!new_dbuf_state ||
7798	    new_dbuf_state->enabled_slices == old_dbuf_state->enabled_slices)
7799		return;
7800
7801	WARN_ON(!new_dbuf_state->base.changed);
7802
7803	gen9_dbuf_slices_update(dev_priv,
7804				new_dbuf_state->enabled_slices);
7805}
7806