vc4_crtc.c 28.6 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
2 3 4 5 6 7 8 9 10
/*
 * Copyright (C) 2015 Broadcom
 */

/**
 * DOC: VC4 CRTC module
 *
 * In VC4, the Pixel Valve is what most closely corresponds to the
 * DRM's concept of a CRTC.  The PV generates video timings from the
11
 * encoder's clock plus its configuration.  It pulls scaled pixels from
12 13 14
 * the HVS at that timing, and feeds it to the encoder.
 *
 * However, the DRM CRTC also collects the configuration of all the
15 16 17
 * DRM planes attached to it.  As a result, the CRTC is also
 * responsible for writing the display list for the HVS channel that
 * the CRTC will use.
18 19 20 21 22 23 24 25 26 27 28 29 30 31
 *
 * The 2835 has 3 different pixel valves.  pv0 in the audio power
 * domain feeds DSI0 or DPI, while pv1 feeds DS1 or SMI.  pv2 in the
 * image domain can feed either HDMI or the SDTV controller.  The
 * pixel valve chooses from the CPRMAN clocks (HSM for HDMI, VEC for
 * SDTV, etc.) according to which output type is chosen in the mux.
 *
 * For power management, the pixel valve's registers are all clocked
 * by the AXI clock, while the timings and FIFOs make use of the
 * output-specific clock.  Since the encoders also directly consume
 * the CPRMAN clocks, and know what timings they need, they are the
 * ones that set the clock.
 */

S
Sam Ravnborg 已提交
32 33 34 35
#include <linux/clk.h>
#include <linux/component.h>
#include <linux/of_device.h>

36 37
#include <drm/drm_atomic.h>
#include <drm/drm_atomic_helper.h>
D
Daniel Vetter 已提交
38
#include <drm/drm_atomic_uapi.h>
S
Sam Ravnborg 已提交
39
#include <drm/drm_fb_cma_helper.h>
40
#include <drm/drm_print.h>
41
#include <drm/drm_probe_helper.h>
S
Sam Ravnborg 已提交
42 43
#include <drm/drm_vblank.h>

44 45 46
#include "vc4_drv.h"
#include "vc4_regs.h"

47 48
#define HVS_FIFO_LATENCY_PIX	6

49 50 51
#define CRTC_WRITE(offset, val) writel(val, vc4_crtc->regs + (offset))
#define CRTC_READ(offset) readl(vc4_crtc->regs + (offset))

52 53 54 55 56 57 58 59 60 61 62 63 64 65
static const struct debugfs_reg32 crtc_regs[] = {
	VC4_REG32(PV_CONTROL),
	VC4_REG32(PV_V_CONTROL),
	VC4_REG32(PV_VSYNCD_EVEN),
	VC4_REG32(PV_HORZA),
	VC4_REG32(PV_HORZB),
	VC4_REG32(PV_VERTA),
	VC4_REG32(PV_VERTB),
	VC4_REG32(PV_VERTA_EVEN),
	VC4_REG32(PV_VERTB_EVEN),
	VC4_REG32(PV_INTEN),
	VC4_REG32(PV_INTSTAT),
	VC4_REG32(PV_STAT),
	VC4_REG32(PV_HACT_ACT),
66 67
};

68 69 70 71 72
static bool vc4_crtc_get_scanout_position(struct drm_crtc *crtc,
					  bool in_vblank_irq,
					  int *vpos, int *hpos,
					  ktime_t *stime, ktime_t *etime,
					  const struct drm_display_mode *mode)
73
{
74
	struct drm_device *dev = crtc->dev;
75
	struct vc4_dev *vc4 = to_vc4_dev(dev);
76
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
77 78 79
	u32 val;
	int fifo_lines;
	int vblank_lines;
80
	bool ret = false;
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101

	/* preempt_disable_rt() should go right here in PREEMPT_RT patchset. */

	/* Get optional system timestamp before query. */
	if (stime)
		*stime = ktime_get();

	/*
	 * Read vertical scanline which is currently composed for our
	 * pixelvalve by the HVS, and also the scaler status.
	 */
	val = HVS_READ(SCALER_DISPSTATX(vc4_crtc->channel));

	/* Get optional system timestamp after query. */
	if (etime)
		*etime = ktime_get();

	/* preempt_enable_rt() should go right here in PREEMPT_RT patchset. */

	/* Vertical position of hvs composed scanline. */
	*vpos = VC4_GET_FIELD(val, SCALER_DISPSTATX_LINE);
102 103 104 105
	*hpos = 0;

	if (mode->flags & DRM_MODE_FLAG_INTERLACE) {
		*vpos /= 2;
106

107 108 109 110
		/* Use hpos to correct for field offset in interlaced mode. */
		if (VC4_GET_FIELD(val, SCALER_DISPSTATX_FRAME_COUNT) % 2)
			*hpos += mode->crtc_htotal / 2;
	}
111 112 113 114 115

	/* This is the offset we need for translating hvs -> pv scanout pos. */
	fifo_lines = vc4_crtc->cob_size / mode->crtc_hdisplay;

	if (fifo_lines > 0)
116
		ret = true;
117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145

	/* HVS more than fifo_lines into frame for compositing? */
	if (*vpos > fifo_lines) {
		/*
		 * We are in active scanout and can get some meaningful results
		 * from HVS. The actual PV scanout can not trail behind more
		 * than fifo_lines as that is the fifo's capacity. Assume that
		 * in active scanout the HVS and PV work in lockstep wrt. HVS
		 * refilling the fifo and PV consuming from the fifo, ie.
		 * whenever the PV consumes and frees up a scanline in the
		 * fifo, the HVS will immediately refill it, therefore
		 * incrementing vpos. Therefore we choose HVS read position -
		 * fifo size in scanlines as a estimate of the real scanout
		 * position of the PV.
		 */
		*vpos -= fifo_lines + 1;

		return ret;
	}

	/*
	 * Less: This happens when we are in vblank and the HVS, after getting
	 * the VSTART restart signal from the PV, just started refilling its
	 * fifo with new lines from the top-most lines of the new framebuffers.
	 * The PV does not scan out in vblank, so does not remove lines from
	 * the fifo, so the fifo will be full quickly and the HVS has to pause.
	 * We can't get meaningful readings wrt. scanline position of the PV
	 * and need to make things up in a approximative but consistent way.
	 */
146
	vblank_lines = mode->vtotal - mode->vdisplay;
147

148
	if (in_vblank_irq) {
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169
		/*
		 * Assume the irq handler got called close to first
		 * line of vblank, so PV has about a full vblank
		 * scanlines to go, and as a base timestamp use the
		 * one taken at entry into vblank irq handler, so it
		 * is not affected by random delays due to lock
		 * contention on event_lock or vblank_time lock in
		 * the core.
		 */
		*vpos = -vblank_lines;

		if (stime)
			*stime = vc4_crtc->t_vblank;
		if (etime)
			*etime = vc4_crtc->t_vblank;

		/*
		 * If the HVS fifo is not yet full then we know for certain
		 * we are at the very beginning of vblank, as the hvs just
		 * started refilling, and the stime and etime timestamps
		 * truly correspond to start of vblank.
170 171 172
		 *
		 * Unfortunately there's no way to report this to upper levels
		 * and make it more useful.
173 174 175 176 177 178 179 180 181 182 183 184 185 186
		 */
	} else {
		/*
		 * No clue where we are inside vblank. Return a vpos of zero,
		 * which will cause calling code to just return the etime
		 * timestamp uncorrected. At least this is no worse than the
		 * standard fallback.
		 */
		*vpos = 0;
	}

	return ret;
}

187
void vc4_crtc_destroy(struct drm_crtc *crtc)
188 189 190 191 192 193 194 195 196 197 198
{
	drm_crtc_cleanup(crtc);
}

static u32 vc4_get_fifo_full_level(u32 format)
{
	static const u32 fifo_len_bytes = 64;

	switch (format) {
	case PV_CONTROL_FORMAT_DSIV_16:
	case PV_CONTROL_FORMAT_DSIC_16:
199
		return fifo_len_bytes - 2 * HVS_FIFO_LATENCY_PIX;
200 201 202 203 204
	case PV_CONTROL_FORMAT_DSIV_18:
		return fifo_len_bytes - 14;
	case PV_CONTROL_FORMAT_24:
	case PV_CONTROL_FORMAT_DSIV_24:
	default:
205
		return fifo_len_bytes - 3 * HVS_FIFO_LATENCY_PIX;
206 207 208 209
	}
}

/*
210 211 212 213 214
 * Returns the encoder attached to the CRTC.
 *
 * VC4 can only scan out to one encoder at a time, while the DRM core
 * allows drivers to push pixels to more than one encoder from the
 * same CRTC.
215
 */
216
static struct drm_encoder *vc4_get_crtc_encoder(struct drm_crtc *crtc)
217 218
{
	struct drm_connector *connector;
219
	struct drm_connector_list_iter conn_iter;
220

221 222
	drm_connector_list_iter_begin(crtc->dev, &conn_iter);
	drm_for_each_connector_iter(connector, &conn_iter) {
J
Julia Lawall 已提交
223
		if (connector->state->crtc == crtc) {
224
			drm_connector_list_iter_end(&conn_iter);
225
			return connector->encoder;
226 227
		}
	}
228
	drm_connector_list_iter_end(&conn_iter);
229

230
	return NULL;
231 232
}

233
static void vc4_crtc_config_pv(struct drm_crtc *crtc)
234
{
235 236
	struct drm_encoder *encoder = vc4_get_crtc_encoder(crtc);
	struct vc4_encoder *vc4_encoder = to_vc4_encoder(encoder);
237 238 239 240
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
	struct drm_crtc_state *state = crtc->state;
	struct drm_display_mode *mode = &state->adjusted_mode;
	bool interlace = mode->flags & DRM_MODE_FLAG_INTERLACE;
241
	u32 pixel_rep = (mode->flags & DRM_MODE_FLAG_DBLCLK) ? 2 : 1;
242 243 244
	bool is_dsi = (vc4_encoder->type == VC4_ENCODER_TYPE_DSI0 ||
		       vc4_encoder->type == VC4_ENCODER_TYPE_DSI1);
	u32 format = is_dsi ? PV_CONTROL_FORMAT_DSIV_24 : PV_CONTROL_FORMAT_24;
245 246 247 248 249 250 251

	/* Reset the PV fifo. */
	CRTC_WRITE(PV_CONTROL, 0);
	CRTC_WRITE(PV_CONTROL, PV_CONTROL_FIFO_CLR | PV_CONTROL_EN);
	CRTC_WRITE(PV_CONTROL, 0);

	CRTC_WRITE(PV_HORZA,
252 253
		   VC4_SET_FIELD((mode->htotal -
				  mode->hsync_end) * pixel_rep,
254
				 PV_HORZA_HBP) |
255 256
		   VC4_SET_FIELD((mode->hsync_end -
				  mode->hsync_start) * pixel_rep,
257 258
				 PV_HORZA_HSYNC));
	CRTC_WRITE(PV_HORZB,
259 260
		   VC4_SET_FIELD((mode->hsync_start -
				  mode->hdisplay) * pixel_rep,
261
				 PV_HORZB_HFP) |
262
		   VC4_SET_FIELD(mode->hdisplay * pixel_rep, PV_HORZB_HACTIVE));
263

264
	CRTC_WRITE(PV_VERTA,
265
		   VC4_SET_FIELD(mode->crtc_vtotal - mode->crtc_vsync_end,
266
				 PV_VERTA_VBP) |
267
		   VC4_SET_FIELD(mode->crtc_vsync_end - mode->crtc_vsync_start,
268 269
				 PV_VERTA_VSYNC));
	CRTC_WRITE(PV_VERTB,
270
		   VC4_SET_FIELD(mode->crtc_vsync_start - mode->crtc_vdisplay,
271
				 PV_VERTB_VFP) |
272
		   VC4_SET_FIELD(mode->crtc_vdisplay, PV_VERTB_VACTIVE));
273

274 275
	if (interlace) {
		CRTC_WRITE(PV_VERTA_EVEN,
276 277
			   VC4_SET_FIELD(mode->crtc_vtotal -
					 mode->crtc_vsync_end - 1,
278
					 PV_VERTA_VBP) |
279 280
			   VC4_SET_FIELD(mode->crtc_vsync_end -
					 mode->crtc_vsync_start,
281 282
					 PV_VERTA_VSYNC));
		CRTC_WRITE(PV_VERTB_EVEN,
283 284
			   VC4_SET_FIELD(mode->crtc_vsync_start -
					 mode->crtc_vdisplay,
285
					 PV_VERTB_VFP) |
286 287 288 289 290 291 292 293 294
			   VC4_SET_FIELD(mode->crtc_vdisplay, PV_VERTB_VACTIVE));

		/* We set up first field even mode for HDMI.  VEC's
		 * NTSC mode would want first field odd instead, once
		 * we support it (to do so, set ODD_FIRST and put the
		 * delay in VSYNCD_EVEN instead).
		 */
		CRTC_WRITE(PV_V_CONTROL,
			   PV_VCONTROL_CONTINUOUS |
295
			   (is_dsi ? PV_VCONTROL_DSI : 0) |
296
			   PV_VCONTROL_INTERLACE |
297
			   VC4_SET_FIELD(mode->htotal * pixel_rep / 2,
298 299 300
					 PV_VCONTROL_ODD_DELAY));
		CRTC_WRITE(PV_VSYNCD_EVEN, 0);
	} else {
301 302 303
		CRTC_WRITE(PV_V_CONTROL,
			   PV_VCONTROL_CONTINUOUS |
			   (is_dsi ? PV_VCONTROL_DSI : 0));
304 305
	}

306 307
	if (is_dsi)
		CRTC_WRITE(PV_HACT_ACT, mode->hdisplay * pixel_rep);
308 309 310 311 312

	CRTC_WRITE(PV_CONTROL,
		   VC4_SET_FIELD(format, PV_CONTROL_FORMAT) |
		   VC4_SET_FIELD(vc4_get_fifo_full_level(format),
				 PV_CONTROL_FIFO_LEVEL) |
313
		   VC4_SET_FIELD(pixel_rep - 1, PV_CONTROL_PIXEL_REP) |
314 315 316
		   PV_CONTROL_CLR_AT_START |
		   PV_CONTROL_TRIGGER_UNDERFLOW |
		   PV_CONTROL_WAIT_HSTART |
317 318
		   VC4_SET_FIELD(vc4_encoder->clock_select,
				 PV_CONTROL_CLK_SELECT) |
319 320
		   PV_CONTROL_FIFO_CLR |
		   PV_CONTROL_EN);
321 322 323 324 325 326 327 328 329
}

static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc)
{
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state);
	bool debug_dump_regs = false;

	if (debug_dump_regs) {
330 331 332 333
		struct drm_printer p = drm_info_printer(&vc4_crtc->pdev->dev);
		dev_info(&vc4_crtc->pdev->dev, "CRTC %d regs before:\n",
			 drm_crtc_index(crtc));
		drm_print_regset32(&p, &vc4_crtc->regset);
334 335 336 337
	}

	if (!vc4_state->feed_txp)
		vc4_crtc_config_pv(crtc);
338

339
	vc4_hvs_mode_set_nofb(crtc);
340

341
	if (debug_dump_regs) {
342 343 344 345
		struct drm_printer p = drm_info_printer(&vc4_crtc->pdev->dev);
		dev_info(&vc4_crtc->pdev->dev, "CRTC %d regs after:\n",
			 drm_crtc_index(crtc));
		drm_print_regset32(&p, &vc4_crtc->regset);
346 347 348 349 350 351 352 353 354 355 356
	}
}

static void require_hvs_enabled(struct drm_device *dev)
{
	struct vc4_dev *vc4 = to_vc4_dev(dev);

	WARN_ON_ONCE((HVS_READ(SCALER_DISPCTRL) & SCALER_DISPCTRL_ENABLE) !=
		     SCALER_DISPCTRL_ENABLE);
}

357 358
static void vc4_crtc_atomic_disable(struct drm_crtc *crtc,
				    struct drm_crtc_state *old_state)
359
{
360
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state);
361 362 363
	struct drm_device *dev = crtc->dev;
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
	int ret;
364

365 366
	require_hvs_enabled(dev);

367 368 369
	/* Disable vblank irq handling before crtc is disabled. */
	drm_crtc_vblank_off(crtc);

370 371 372 373 374 375
	if (!vc4_state->feed_txp) {
		CRTC_WRITE(PV_V_CONTROL,
			   CRTC_READ(PV_V_CONTROL) & ~PV_VCONTROL_VIDEN);
		ret = wait_for(!(CRTC_READ(PV_V_CONTROL) & PV_VCONTROL_VIDEN), 1);
		WARN_ONCE(ret, "Timeout waiting for !PV_VCONTROL_VIDEN\n");
	}
376

377
	vc4_hvs_atomic_disable(crtc, old_state);
378 379 380 381 382 383 384 385 386 387 388 389 390

	/*
	 * Make sure we issue a vblank event after disabling the CRTC if
	 * someone was waiting it.
	 */
	if (crtc->state->event) {
		unsigned long flags;

		spin_lock_irqsave(&dev->event_lock, flags);
		drm_crtc_send_vblank_event(crtc, crtc->state->event);
		crtc->state->event = NULL;
		spin_unlock_irqrestore(&dev->event_lock, flags);
	}
391 392
}

393 394
static void vc4_crtc_atomic_enable(struct drm_crtc *crtc,
				   struct drm_crtc_state *old_state)
395 396 397
{
	struct drm_device *dev = crtc->dev;
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
398
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state);
399 400 401

	require_hvs_enabled(dev);

402 403 404 405 406
	/* Enable vblank irq handling before crtc is started otherwise
	 * drm_crtc_get_vblank() fails in vc4_crtc_update_dlist().
	 */
	drm_crtc_vblank_on(crtc);

407
	vc4_hvs_atomic_enable(crtc, old_state);
408

409 410 411 412 413 414
	/* When feeding the transposer block the pixelvalve is unneeded and
	 * should not be enabled.
	 */
	if (!vc4_state->feed_txp)
		CRTC_WRITE(PV_V_CONTROL,
			   CRTC_READ(PV_V_CONTROL) | PV_VCONTROL_VIDEN);
415 416
}

417 418
static enum drm_mode_status vc4_crtc_mode_valid(struct drm_crtc *crtc,
						const struct drm_display_mode *mode)
419
{
420
	/* Do not allow doublescan modes from user space */
421
	if (mode->flags & DRM_MODE_FLAG_DBLSCAN) {
422 423
		DRM_DEBUG_KMS("[CRTC:%d] Doublescan mode rejected.\n",
			      crtc->base.id);
424
		return MODE_NO_DBLESCAN;
425 426
	}

427
	return MODE_OK;
428 429
}

430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460
void vc4_crtc_get_margins(struct drm_crtc_state *state,
			  unsigned int *left, unsigned int *right,
			  unsigned int *top, unsigned int *bottom)
{
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(state);
	struct drm_connector_state *conn_state;
	struct drm_connector *conn;
	int i;

	*left = vc4_state->margins.left;
	*right = vc4_state->margins.right;
	*top = vc4_state->margins.top;
	*bottom = vc4_state->margins.bottom;

	/* We have to interate over all new connector states because
	 * vc4_crtc_get_margins() might be called before
	 * vc4_crtc_atomic_check() which means margins info in vc4_crtc_state
	 * might be outdated.
	 */
	for_each_new_connector_in_state(state->state, conn, conn_state, i) {
		if (conn_state->crtc != state->crtc)
			continue;

		*left = conn_state->tv.margins.left;
		*right = conn_state->tv.margins.right;
		*top = conn_state->tv.margins.top;
		*bottom = conn_state->tv.margins.bottom;
		break;
	}
}

461 462 463
static int vc4_crtc_atomic_check(struct drm_crtc *crtc,
				 struct drm_crtc_state *state)
{
464
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(state);
465 466 467
	struct drm_connector *conn;
	struct drm_connector_state *conn_state;
	int ret, i;
468

469
	ret = vc4_hvs_atomic_check(crtc, state);
470 471
	if (ret)
		return ret;
472

473 474 475 476
	for_each_new_connector_in_state(state->state, conn, conn_state, i) {
		if (conn_state->crtc != crtc)
			continue;

477 478 479 480
		vc4_state->margins.left = conn_state->tv.margins.left;
		vc4_state->margins.right = conn_state->tv.margins.right;
		vc4_state->margins.top = conn_state->tv.margins.top;
		vc4_state->margins.bottom = conn_state->tv.margins.bottom;
481 482 483
		break;
	}

484 485 486
	return 0;
}

487
static int vc4_enable_vblank(struct drm_crtc *crtc)
488
{
489
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
490 491 492 493
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state);

	if (vc4_state->feed_txp)
		return 0;
494 495 496 497 498 499

	CRTC_WRITE(PV_INTEN, PV_INT_VFP_START);

	return 0;
}

500
static void vc4_disable_vblank(struct drm_crtc *crtc)
501
{
502
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
503 504 505 506
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state);

	if (vc4_state->feed_txp)
		return;
507 508 509 510 511 512 513 514

	CRTC_WRITE(PV_INTEN, 0);
}

static void vc4_crtc_handle_page_flip(struct vc4_crtc *vc4_crtc)
{
	struct drm_crtc *crtc = &vc4_crtc->base;
	struct drm_device *dev = crtc->dev;
515 516 517
	struct vc4_dev *vc4 = to_vc4_dev(dev);
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(crtc->state);
	u32 chan = vc4_crtc->channel;
518 519 520
	unsigned long flags;

	spin_lock_irqsave(&dev->event_lock, flags);
521
	if (vc4_crtc->event &&
522 523
	    (vc4_state->mm.start == HVS_READ(SCALER_DISPLACTX(chan)) ||
	     vc4_state->feed_txp)) {
524 525
		drm_crtc_send_vblank_event(crtc, vc4_crtc->event);
		vc4_crtc->event = NULL;
526
		drm_crtc_vblank_put(crtc);
527 528 529 530 531 532 533 534

		/* Wait for the page flip to unmask the underrun to ensure that
		 * the display list was updated by the hardware. Before that
		 * happens, the HVS will be using the previous display list with
		 * the CRTC and encoder already reconfigured, leading to
		 * underruns. This can be seen when reconfiguring the CRTC.
		 */
		vc4_hvs_unmask_underrun(dev, vc4_crtc->channel);
535 536 537 538
	}
	spin_unlock_irqrestore(&dev->event_lock, flags);
}

539 540 541 542 543 544 545
void vc4_crtc_handle_vblank(struct vc4_crtc *crtc)
{
	crtc->t_vblank = ktime_get();
	drm_crtc_handle_vblank(&crtc->base);
	vc4_crtc_handle_page_flip(crtc);
}

546 547 548 549 550 551 552 553
static irqreturn_t vc4_crtc_irq_handler(int irq, void *data)
{
	struct vc4_crtc *vc4_crtc = data;
	u32 stat = CRTC_READ(PV_INTSTAT);
	irqreturn_t ret = IRQ_NONE;

	if (stat & PV_INT_VFP_START) {
		CRTC_WRITE(PV_INTSTAT, PV_INT_VFP_START);
554
		vc4_crtc_handle_vblank(vc4_crtc);
555 556 557 558 559 560
		ret = IRQ_HANDLED;
	}

	return ret;
}

561 562 563
struct vc4_async_flip_state {
	struct drm_crtc *crtc;
	struct drm_framebuffer *fb;
564
	struct drm_framebuffer *old_fb;
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
	struct drm_pending_vblank_event *event;

	struct vc4_seqno_cb cb;
};

/* Called when the V3D execution for the BO being flipped to is done, so that
 * we can actually update the plane's address to point to it.
 */
static void
vc4_async_page_flip_complete(struct vc4_seqno_cb *cb)
{
	struct vc4_async_flip_state *flip_state =
		container_of(cb, struct vc4_async_flip_state, cb);
	struct drm_crtc *crtc = flip_state->crtc;
	struct drm_device *dev = crtc->dev;
	struct vc4_dev *vc4 = to_vc4_dev(dev);
	struct drm_plane *plane = crtc->primary;

	vc4_plane_async_set_fb(plane, flip_state->fb);
	if (flip_state->event) {
		unsigned long flags;

		spin_lock_irqsave(&dev->event_lock, flags);
		drm_crtc_send_vblank_event(crtc, flip_state->event);
		spin_unlock_irqrestore(&dev->event_lock, flags);
	}

592
	drm_crtc_vblank_put(crtc);
593
	drm_framebuffer_put(flip_state->fb);
594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610

	/* Decrement the BO usecnt in order to keep the inc/dec calls balanced
	 * when the planes are updated through the async update path.
	 * FIXME: we should move to generic async-page-flip when it's
	 * available, so that we can get rid of this hand-made cleanup_fb()
	 * logic.
	 */
	if (flip_state->old_fb) {
		struct drm_gem_cma_object *cma_bo;
		struct vc4_bo *bo;

		cma_bo = drm_fb_cma_get_gem_obj(flip_state->old_fb, 0);
		bo = to_vc4_bo(&cma_bo->base);
		vc4_bo_dec_usecnt(bo);
		drm_framebuffer_put(flip_state->old_fb);
	}

611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634
	kfree(flip_state);

	up(&vc4->async_modeset);
}

/* Implements async (non-vblank-synced) page flips.
 *
 * The page flip ioctl needs to return immediately, so we grab the
 * modeset semaphore on the pipe, and queue the address update for
 * when V3D is done with the BO being flipped to.
 */
static int vc4_async_page_flip(struct drm_crtc *crtc,
			       struct drm_framebuffer *fb,
			       struct drm_pending_vblank_event *event,
			       uint32_t flags)
{
	struct drm_device *dev = crtc->dev;
	struct vc4_dev *vc4 = to_vc4_dev(dev);
	struct drm_plane *plane = crtc->primary;
	int ret = 0;
	struct vc4_async_flip_state *flip_state;
	struct drm_gem_cma_object *cma_bo = drm_fb_cma_get_gem_obj(fb, 0);
	struct vc4_bo *bo = to_vc4_bo(&cma_bo->base);

635 636 637 638 639 640 641 642 643 644 645
	/* Increment the BO usecnt here, so that we never end up with an
	 * unbalanced number of vc4_bo_{dec,inc}_usecnt() calls when the
	 * plane is later updated through the non-async path.
	 * FIXME: we should move to generic async-page-flip when it's
	 * available, so that we can get rid of this hand-made prepare_fb()
	 * logic.
	 */
	ret = vc4_bo_inc_usecnt(bo);
	if (ret)
		return ret;

646
	flip_state = kzalloc(sizeof(*flip_state), GFP_KERNEL);
647 648
	if (!flip_state) {
		vc4_bo_dec_usecnt(bo);
649
		return -ENOMEM;
650
	}
651

652
	drm_framebuffer_get(fb);
653 654 655 656 657 658 659
	flip_state->fb = fb;
	flip_state->crtc = crtc;
	flip_state->event = event;

	/* Make sure all other async modesetes have landed. */
	ret = down_interruptible(&vc4->async_modeset);
	if (ret) {
660
		drm_framebuffer_put(fb);
661
		vc4_bo_dec_usecnt(bo);
662 663 664 665
		kfree(flip_state);
		return ret;
	}

666 667 668 669 670 671 672 673 674 675 676 677
	/* Save the current FB before it's replaced by the new one in
	 * drm_atomic_set_fb_for_plane(). We'll need the old FB in
	 * vc4_async_page_flip_complete() to decrement the BO usecnt and keep
	 * it consistent.
	 * FIXME: we should move to generic async-page-flip when it's
	 * available, so that we can get rid of this hand-made cleanup_fb()
	 * logic.
	 */
	flip_state->old_fb = plane->state->fb;
	if (flip_state->old_fb)
		drm_framebuffer_get(flip_state->old_fb);

678 679
	WARN_ON(drm_crtc_vblank_get(crtc) != 0);

680 681 682 683 684 685 686 687 688 689 690 691 692
	/* Immediately update the plane's legacy fb pointer, so that later
	 * modeset prep sees the state that will be present when the semaphore
	 * is released.
	 */
	drm_atomic_set_fb_for_plane(plane->state, fb);

	vc4_queue_seqno_cb(dev, &flip_state->cb, bo->seqno,
			   vc4_async_page_flip_complete);

	/* Driver takes ownership of state on successful async commit. */
	return 0;
}

693 694 695 696 697
int vc4_page_flip(struct drm_crtc *crtc,
		  struct drm_framebuffer *fb,
		  struct drm_pending_vblank_event *event,
		  uint32_t flags,
		  struct drm_modeset_acquire_ctx *ctx)
698 699 700 701
{
	if (flags & DRM_MODE_PAGE_FLIP_ASYNC)
		return vc4_async_page_flip(crtc, fb, event, flags);
	else
702
		return drm_atomic_helper_page_flip(crtc, fb, event, flags, ctx);
703 704
}

705
struct drm_crtc_state *vc4_crtc_duplicate_state(struct drm_crtc *crtc)
706
{
707
	struct vc4_crtc_state *vc4_state, *old_vc4_state;
708 709 710 711 712

	vc4_state = kzalloc(sizeof(*vc4_state), GFP_KERNEL);
	if (!vc4_state)
		return NULL;

713 714
	old_vc4_state = to_vc4_crtc_state(crtc->state);
	vc4_state->feed_txp = old_vc4_state->feed_txp;
715
	vc4_state->margins = old_vc4_state->margins;
716

717 718 719 720
	__drm_atomic_helper_crtc_duplicate_state(crtc, &vc4_state->base);
	return &vc4_state->base;
}

721 722
void vc4_crtc_destroy_state(struct drm_crtc *crtc,
			    struct drm_crtc_state *state)
723 724 725 726
{
	struct vc4_dev *vc4 = to_vc4_dev(crtc->dev);
	struct vc4_crtc_state *vc4_state = to_vc4_crtc_state(state);

727
	if (drm_mm_node_allocated(&vc4_state->mm)) {
728 729 730 731 732 733 734 735
		unsigned long flags;

		spin_lock_irqsave(&vc4->hvs->mm_lock, flags);
		drm_mm_remove_node(&vc4_state->mm);
		spin_unlock_irqrestore(&vc4->hvs->mm_lock, flags);

	}

736
	drm_atomic_helper_crtc_destroy_state(crtc, state);
737 738
}

739
void vc4_crtc_reset(struct drm_crtc *crtc)
740 741
{
	if (crtc->state)
742
		vc4_crtc_destroy_state(crtc, crtc->state);
743 744
	crtc->state = kzalloc(sizeof(struct vc4_crtc_state), GFP_KERNEL);
	if (crtc->state)
745
		__drm_atomic_helper_crtc_reset(crtc, crtc->state);
746 747
}

748 749 750
static const struct drm_crtc_funcs vc4_crtc_funcs = {
	.set_config = drm_atomic_helper_set_config,
	.destroy = vc4_crtc_destroy,
751
	.page_flip = vc4_page_flip,
752 753 754
	.set_property = NULL,
	.cursor_set = NULL, /* handled by drm_mode_cursor_universal */
	.cursor_move = NULL, /* handled by drm_mode_cursor_universal */
755
	.reset = vc4_crtc_reset,
756 757
	.atomic_duplicate_state = vc4_crtc_duplicate_state,
	.atomic_destroy_state = vc4_crtc_destroy_state,
758
	.gamma_set = drm_atomic_helper_legacy_gamma_set,
759 760
	.enable_vblank = vc4_enable_vblank,
	.disable_vblank = vc4_disable_vblank,
761
	.get_vblank_timestamp = drm_crtc_vblank_helper_get_vblank_timestamp,
762 763 764 765
};

static const struct drm_crtc_helper_funcs vc4_crtc_helper_funcs = {
	.mode_set_nofb = vc4_crtc_mode_set_nofb,
766
	.mode_valid = vc4_crtc_mode_valid,
767
	.atomic_check = vc4_crtc_atomic_check,
768
	.atomic_flush = vc4_hvs_atomic_flush,
769
	.atomic_enable = vc4_crtc_atomic_enable,
770
	.atomic_disable = vc4_crtc_atomic_disable,
771
	.get_scanout_position = vc4_crtc_get_scanout_position,
772 773
};

774 775 776 777
static const struct vc4_pv_data bcm2835_pv0_data = {
	.base = {
		.hvs_channel = 0,
	},
778
	.debugfs_name = "crtc0_regs",
779 780 781 782
	.encoder_types = {
		[PV_CONTROL_CLK_SELECT_DSI] = VC4_ENCODER_TYPE_DSI0,
		[PV_CONTROL_CLK_SELECT_DPI_SMI_HDMI] = VC4_ENCODER_TYPE_DPI,
	},
783 784
};

785 786 787 788
static const struct vc4_pv_data bcm2835_pv1_data = {
	.base = {
		.hvs_channel = 2,
	},
789
	.debugfs_name = "crtc1_regs",
790 791 792 793
	.encoder_types = {
		[PV_CONTROL_CLK_SELECT_DSI] = VC4_ENCODER_TYPE_DSI1,
		[PV_CONTROL_CLK_SELECT_DPI_SMI_HDMI] = VC4_ENCODER_TYPE_SMI,
	},
794 795
};

796 797 798 799
static const struct vc4_pv_data bcm2835_pv2_data = {
	.base = {
		.hvs_channel = 1,
	},
800
	.debugfs_name = "crtc2_regs",
801 802 803 804
	.encoder_types = {
		[PV_CONTROL_CLK_SELECT_DPI_SMI_HDMI] = VC4_ENCODER_TYPE_HDMI,
		[PV_CONTROL_CLK_SELECT_VEC] = VC4_ENCODER_TYPE_VEC,
	},
805 806 807
};

static const struct of_device_id vc4_crtc_dt_match[] = {
808 809 810
	{ .compatible = "brcm,bcm2835-pixelvalve0", .data = &bcm2835_pv0_data },
	{ .compatible = "brcm,bcm2835-pixelvalve1", .data = &bcm2835_pv1_data },
	{ .compatible = "brcm,bcm2835-pixelvalve2", .data = &bcm2835_pv2_data },
811 812 813 814 815 816 817
	{}
};

static void vc4_set_crtc_possible_masks(struct drm_device *drm,
					struct drm_crtc *crtc)
{
	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
818 819
	const struct vc4_pv_data *pv_data = vc4_crtc_to_vc4_pv_data(vc4_crtc);
	const enum vc4_encoder_type *encoder_types = pv_data->encoder_types;
820 821 822
	struct drm_encoder *encoder;

	drm_for_each_encoder(encoder, drm) {
823
		struct vc4_encoder *vc4_encoder;
824 825
		int i;

826
		vc4_encoder = to_vc4_encoder(encoder);
827
		for (i = 0; i < ARRAY_SIZE(pv_data->encoder_types); i++) {
828 829 830 831 832
			if (vc4_encoder->type == encoder_types[i]) {
				vc4_encoder->clock_select = i;
				encoder->possible_crtcs |= drm_crtc_mask(crtc);
				break;
			}
833 834 835 836
		}
	}
}

837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852
static void
vc4_crtc_get_cob_allocation(struct vc4_crtc *vc4_crtc)
{
	struct drm_device *drm = vc4_crtc->base.dev;
	struct vc4_dev *vc4 = to_vc4_dev(drm);
	u32 dispbase = HVS_READ(SCALER_DISPBASEX(vc4_crtc->channel));
	/* Top/base are supposed to be 4-pixel aligned, but the
	 * Raspberry Pi firmware fills the low bits (which are
	 * presumably ignored).
	 */
	u32 top = VC4_GET_FIELD(dispbase, SCALER_DISPBASEX_TOP) & ~3;
	u32 base = VC4_GET_FIELD(dispbase, SCALER_DISPBASEX_BASE) & ~3;

	vc4_crtc->cob_size = top - base + 4;
}

853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894
int vc4_crtc_init(struct drm_device *drm, struct vc4_crtc *vc4_crtc,
		  const struct drm_crtc_funcs *crtc_funcs,
		  const struct drm_crtc_helper_funcs *crtc_helper_funcs)
{
	struct drm_crtc *crtc = &vc4_crtc->base;
	struct drm_plane *primary_plane;
	unsigned int i;

	/* For now, we create just the primary and the legacy cursor
	 * planes.  We should be able to stack more planes on easily,
	 * but to do that we would need to compute the bandwidth
	 * requirement of the plane configuration, and reject ones
	 * that will take too much.
	 */
	primary_plane = vc4_plane_init(drm, DRM_PLANE_TYPE_PRIMARY);
	if (IS_ERR(primary_plane)) {
		dev_err(drm->dev, "failed to construct primary plane\n");
		return PTR_ERR(primary_plane);
	}

	drm_crtc_init_with_planes(drm, crtc, primary_plane, NULL,
				  crtc_funcs, NULL);
	drm_crtc_helper_add(crtc, crtc_helper_funcs);
	vc4_crtc->channel = vc4_crtc->data->hvs_channel;
	drm_mode_crtc_set_gamma_size(crtc, ARRAY_SIZE(vc4_crtc->lut_r));
	drm_crtc_enable_color_mgmt(crtc, 0, false, crtc->gamma_size);

	/* We support CTM, but only for one CRTC at a time. It's therefore
	 * implemented as private driver state in vc4_kms, not here.
	 */
	drm_crtc_enable_color_mgmt(crtc, 0, true, crtc->gamma_size);
	vc4_crtc_get_cob_allocation(vc4_crtc);

	for (i = 0; i < crtc->gamma_size; i++) {
		vc4_crtc->lut_r[i] = i;
		vc4_crtc->lut_g[i] = i;
		vc4_crtc->lut_b[i] = i;
	}

	return 0;
}

895 896 897 898
static int vc4_crtc_bind(struct device *dev, struct device *master, void *data)
{
	struct platform_device *pdev = to_platform_device(dev);
	struct drm_device *drm = dev_get_drvdata(master);
899
	const struct vc4_pv_data *pv_data;
900 901
	struct vc4_crtc *vc4_crtc;
	struct drm_crtc *crtc;
902 903
	struct drm_plane *destroy_plane, *temp;
	int ret;
904 905 906 907 908 909

	vc4_crtc = devm_kzalloc(dev, sizeof(*vc4_crtc), GFP_KERNEL);
	if (!vc4_crtc)
		return -ENOMEM;
	crtc = &vc4_crtc->base;

910 911
	pv_data = of_device_get_match_data(dev);
	if (!pv_data)
912
		return -ENODEV;
913
	vc4_crtc->data = &pv_data->base;
914
	vc4_crtc->pdev = pdev;
915 916 917 918 919

	vc4_crtc->regs = vc4_ioremap_regs(pdev, 0);
	if (IS_ERR(vc4_crtc->regs))
		return PTR_ERR(vc4_crtc->regs);

920 921 922 923
	vc4_crtc->regset.base = vc4_crtc->regs;
	vc4_crtc->regset.regs = crtc_regs;
	vc4_crtc->regset.nregs = ARRAY_SIZE(crtc_regs);

924 925 926 927 928
	ret = vc4_crtc_init(drm, vc4_crtc,
			    &vc4_crtc_funcs, &vc4_crtc_helper_funcs);
	if (ret)
		return ret;
	vc4_set_crtc_possible_masks(drm, crtc);
929

930 931 932 933 934
	CRTC_WRITE(PV_INTEN, 0);
	CRTC_WRITE(PV_INTSTAT, PV_INT_VFP_START);
	ret = devm_request_irq(dev, platform_get_irq(pdev, 0),
			       vc4_crtc_irq_handler, 0, "vc4 crtc", vc4_crtc);
	if (ret)
935
		goto err_destroy_planes;
936 937 938

	platform_set_drvdata(pdev, vc4_crtc);

939
	vc4_debugfs_add_regset32(drm, pv_data->debugfs_name,
940 941
				 &vc4_crtc->regset);

942 943
	return 0;

944 945 946
err_destroy_planes:
	list_for_each_entry_safe(destroy_plane, temp,
				 &drm->mode_config.plane_list, head) {
V
Ville Syrjälä 已提交
947
		if (destroy_plane->possible_crtcs == drm_crtc_mask(crtc))
948 949
		    destroy_plane->funcs->destroy(destroy_plane);
	}
950

951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990
	return ret;
}

static void vc4_crtc_unbind(struct device *dev, struct device *master,
			    void *data)
{
	struct platform_device *pdev = to_platform_device(dev);
	struct vc4_crtc *vc4_crtc = dev_get_drvdata(dev);

	vc4_crtc_destroy(&vc4_crtc->base);

	CRTC_WRITE(PV_INTEN, 0);

	platform_set_drvdata(pdev, NULL);
}

static const struct component_ops vc4_crtc_ops = {
	.bind   = vc4_crtc_bind,
	.unbind = vc4_crtc_unbind,
};

static int vc4_crtc_dev_probe(struct platform_device *pdev)
{
	return component_add(&pdev->dev, &vc4_crtc_ops);
}

static int vc4_crtc_dev_remove(struct platform_device *pdev)
{
	component_del(&pdev->dev, &vc4_crtc_ops);
	return 0;
}

struct platform_driver vc4_crtc_driver = {
	.probe = vc4_crtc_dev_probe,
	.remove = vc4_crtc_dev_remove,
	.driver = {
		.name = "vc4_crtc",
		.of_match_table = vc4_crtc_dt_match,
	},
};