1 /*
2  * Copyright 2020 Advanced Micro Devices, Inc.
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 shall be included in
12  * all copies or substantial portions of the Software.
13  *
14  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
17  * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
18  * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
19  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
20  * OTHER DEALINGS IN THE SOFTWARE.
21  *
22  * Authors: AMD
23  *
24  */
25 
26 #ifdef CONFIG_DRM_AMD_DC_DCN
27 
28 #include "../display_mode_lib.h"
29 #include "../display_mode_vba.h"
30 #include "../dml_inline_defs.h"
31 #include "display_rq_dlg_calc_30.h"
32 
33 static bool is_dual_plane(enum source_format_class source_format)
34 {
35 	bool ret_val = 0;
36 
37 	if ((source_format == dm_420_12) || (source_format == dm_420_8) || (source_format == dm_420_10) || (source_format == dm_rgbe_alpha))
38 		ret_val = 1;
39 
40 	return ret_val;
41 }
42 
43 static double get_refcyc_per_delivery(struct display_mode_lib *mode_lib,
44 	double refclk_freq_in_mhz,
45 	double pclk_freq_in_mhz,
46 	unsigned int odm_combine,
47 	unsigned int recout_width,
48 	unsigned int hactive,
49 	double vratio,
50 	double hscale_pixel_rate,
51 	unsigned int delivery_width,
52 	unsigned int req_per_swath_ub)
53 {
54 	double refcyc_per_delivery = 0.0;
55 
56 	if (vratio <= 1.0) {
57 		if (odm_combine)
58 			refcyc_per_delivery = (double)refclk_freq_in_mhz * (double)((unsigned int)odm_combine*2)
59 			* dml_min((double)recout_width, (double)hactive / ((unsigned int)odm_combine*2))
60 			/ pclk_freq_in_mhz / (double)req_per_swath_ub;
61 		else
62 			refcyc_per_delivery = (double)refclk_freq_in_mhz * (double)recout_width
63 			/ pclk_freq_in_mhz / (double)req_per_swath_ub;
64 	} else {
65 		refcyc_per_delivery = (double)refclk_freq_in_mhz * (double)delivery_width
66 			/ (double)hscale_pixel_rate / (double)req_per_swath_ub;
67 	}
68 
69 	dml_print("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, refclk_freq_in_mhz);
70 	dml_print("DML_DLG: %s: pclk_freq_in_mhz   = %3.2f\n", __func__, pclk_freq_in_mhz);
71 	dml_print("DML_DLG: %s: recout_width       = %d\n", __func__, recout_width);
72 	dml_print("DML_DLG: %s: vratio             = %3.2f\n", __func__, vratio);
73 	dml_print("DML_DLG: %s: req_per_swath_ub   = %d\n", __func__, req_per_swath_ub);
74 	dml_print("DML_DLG: %s: refcyc_per_delivery= %3.2f\n", __func__, refcyc_per_delivery);
75 
76 	return refcyc_per_delivery;
77 
78 }
79 
80 static unsigned int get_blk_size_bytes(const enum source_macro_tile_size tile_size)
81 {
82 	if (tile_size == dm_256k_tile)
83 		return (256 * 1024);
84 	else if (tile_size == dm_64k_tile)
85 		return (64 * 1024);
86 	else
87 		return (4 * 1024);
88 }
89 
90 static void extract_rq_sizing_regs(struct display_mode_lib *mode_lib,
91 	display_data_rq_regs_st *rq_regs,
92 	const display_data_rq_sizing_params_st rq_sizing)
93 {
94 	dml_print("DML_DLG: %s: rq_sizing param\n", __func__);
95 	print__data_rq_sizing_params_st(mode_lib, rq_sizing);
96 
97 	rq_regs->chunk_size = dml_log2(rq_sizing.chunk_bytes) - 10;
98 
99 	if (rq_sizing.min_chunk_bytes == 0)
100 		rq_regs->min_chunk_size = 0;
101 	else
102 		rq_regs->min_chunk_size = dml_log2(rq_sizing.min_chunk_bytes) - 8 + 1;
103 
104 	rq_regs->meta_chunk_size = dml_log2(rq_sizing.meta_chunk_bytes) - 10;
105 	if (rq_sizing.min_meta_chunk_bytes == 0)
106 		rq_regs->min_meta_chunk_size = 0;
107 	else
108 		rq_regs->min_meta_chunk_size = dml_log2(rq_sizing.min_meta_chunk_bytes) - 6 + 1;
109 
110 	rq_regs->dpte_group_size = dml_log2(rq_sizing.dpte_group_bytes) - 6;
111 	rq_regs->mpte_group_size = dml_log2(rq_sizing.mpte_group_bytes) - 6;
112 }
113 
114 static void extract_rq_regs(struct display_mode_lib *mode_lib,
115 	display_rq_regs_st *rq_regs,
116 	const display_rq_params_st rq_param)
117 {
118 	unsigned int detile_buf_size_in_bytes = mode_lib->ip.det_buffer_size_kbytes * 1024;
119 	unsigned int detile_buf_plane1_addr = 0;
120 
121 	extract_rq_sizing_regs(mode_lib, &(rq_regs->rq_regs_l), rq_param.sizing.rq_l);
122 
123 	rq_regs->rq_regs_l.pte_row_height_linear = dml_floor(dml_log2(rq_param.dlg.rq_l.dpte_row_height),
124 		1) - 3;
125 
126 	if (rq_param.yuv420) {
127 		extract_rq_sizing_regs(mode_lib, &(rq_regs->rq_regs_c), rq_param.sizing.rq_c);
128 		rq_regs->rq_regs_c.pte_row_height_linear = dml_floor(dml_log2(rq_param.dlg.rq_c.dpte_row_height),
129 			1) - 3;
130 	}
131 
132 	rq_regs->rq_regs_l.swath_height = dml_log2(rq_param.dlg.rq_l.swath_height);
133 	rq_regs->rq_regs_c.swath_height = dml_log2(rq_param.dlg.rq_c.swath_height);
134 
135 	// FIXME: take the max between luma, chroma chunk size?
136 	// okay for now, as we are setting chunk_bytes to 8kb anyways
137 	if (rq_param.sizing.rq_l.chunk_bytes >= 32 * 1024 || (rq_param.yuv420 && rq_param.sizing.rq_c.chunk_bytes >= 32 * 1024)) { //32kb
138 		rq_regs->drq_expansion_mode = 0;
139 	} else {
140 		rq_regs->drq_expansion_mode = 2;
141 	}
142 	rq_regs->prq_expansion_mode = 1;
143 	rq_regs->mrq_expansion_mode = 1;
144 	rq_regs->crq_expansion_mode = 1;
145 
146 	if (rq_param.yuv420) {
147 	if ((double)rq_param.misc.rq_l.stored_swath_bytes
148 			/ (double)rq_param.misc.rq_c.stored_swath_bytes <= 1.5) {
149 			detile_buf_plane1_addr = (detile_buf_size_in_bytes / 2.0 / 64.0); // half to chroma
150 		} else {
151 			detile_buf_plane1_addr = dml_round_to_multiple((unsigned int)((2.0 * detile_buf_size_in_bytes) / 3.0),
152 				256,
153 				0) / 64.0; // 2/3 to chroma
154 		}
155 	}
156 	rq_regs->plane1_base_address = detile_buf_plane1_addr;
157 }
158 
159 static void handle_det_buf_split(struct display_mode_lib *mode_lib,
160 	display_rq_params_st *rq_param,
161 	const display_pipe_source_params_st pipe_src_param)
162 {
163 	unsigned int total_swath_bytes = 0;
164 	unsigned int swath_bytes_l = 0;
165 	unsigned int swath_bytes_c = 0;
166 	unsigned int full_swath_bytes_packed_l = 0;
167 	unsigned int full_swath_bytes_packed_c = 0;
168 	bool req128_l = false;
169 	bool req128_c = false;
170 	bool surf_linear = (pipe_src_param.sw_mode == dm_sw_linear);
171 	bool surf_vert = (pipe_src_param.source_scan == dm_vert);
172 	unsigned int log2_swath_height_l = 0;
173 	unsigned int log2_swath_height_c = 0;
174 	unsigned int detile_buf_size_in_bytes = mode_lib->ip.det_buffer_size_kbytes * 1024;
175 
176 	full_swath_bytes_packed_l = rq_param->misc.rq_l.full_swath_bytes;
177 	full_swath_bytes_packed_c = rq_param->misc.rq_c.full_swath_bytes;
178 
179 	if (rq_param->yuv420_10bpc) {
180 		full_swath_bytes_packed_l = dml_round_to_multiple(rq_param->misc.rq_l.full_swath_bytes * 2.0 / 3.0,
181 			256,
182 			1) + 256;
183 		full_swath_bytes_packed_c = dml_round_to_multiple(rq_param->misc.rq_c.full_swath_bytes * 2.0 / 3.0,
184 			256,
185 			1) + 256;
186 	}
187 
188 	if (rq_param->yuv420)
189 		total_swath_bytes = 2 * full_swath_bytes_packed_l + 2 * full_swath_bytes_packed_c;
190 	else
191 		total_swath_bytes = 2 * full_swath_bytes_packed_l;
192 
193 	if (total_swath_bytes <= detile_buf_size_in_bytes) { //full 256b request
194 		req128_l = false;
195 		req128_c = false;
196 		swath_bytes_l = full_swath_bytes_packed_l;
197 		swath_bytes_c = full_swath_bytes_packed_c;
198 	} else if (!rq_param->yuv420) {
199 		req128_l = true;
200 		req128_c = false;
201 		swath_bytes_c = full_swath_bytes_packed_c;
202 		swath_bytes_l = full_swath_bytes_packed_l / 2;
203 	} else if ((double)full_swath_bytes_packed_l / (double)full_swath_bytes_packed_c < 1.5) {
204 		req128_l = false;
205 		req128_c = true;
206 		swath_bytes_l = full_swath_bytes_packed_l;
207 		swath_bytes_c = full_swath_bytes_packed_c / 2;
208 
209 		total_swath_bytes = 2 * swath_bytes_l + 2 * swath_bytes_c;
210 
211 		if (total_swath_bytes > detile_buf_size_in_bytes) {
212 			req128_l = true;
213 			swath_bytes_l = full_swath_bytes_packed_l / 2;
214 		}
215 	} else {
216 		req128_l = true;
217 		req128_c = false;
218 		swath_bytes_l = full_swath_bytes_packed_l/2;
219 		swath_bytes_c = full_swath_bytes_packed_c;
220 
221 		total_swath_bytes = 2 * swath_bytes_l + 2 * swath_bytes_c;
222 
223 		if (total_swath_bytes > detile_buf_size_in_bytes) {
224 			req128_c = true;
225 			swath_bytes_c = full_swath_bytes_packed_c/2;
226 		}
227 	}
228 
229 	if (rq_param->yuv420)
230 		total_swath_bytes = 2 * swath_bytes_l + 2 * swath_bytes_c;
231 	else
232 		total_swath_bytes = 2 * swath_bytes_l;
233 
234 	rq_param->misc.rq_l.stored_swath_bytes = swath_bytes_l;
235 	rq_param->misc.rq_c.stored_swath_bytes = swath_bytes_c;
236 
237 	if (surf_linear) {
238 		log2_swath_height_l = 0;
239 		log2_swath_height_c = 0;
240 	} else if (!surf_vert) {
241 		log2_swath_height_l = dml_log2(rq_param->misc.rq_l.blk256_height) - req128_l;
242 		log2_swath_height_c = dml_log2(rq_param->misc.rq_c.blk256_height) - req128_c;
243 	} else {
244 		log2_swath_height_l = dml_log2(rq_param->misc.rq_l.blk256_width) - req128_l;
245 		log2_swath_height_c = dml_log2(rq_param->misc.rq_c.blk256_width) - req128_c;
246 	}
247 	rq_param->dlg.rq_l.swath_height = 1 << log2_swath_height_l;
248 	rq_param->dlg.rq_c.swath_height = 1 << log2_swath_height_c;
249 
250 	dml_print("DML_DLG: %s: req128_l = %0d\n", __func__, req128_l);
251 	dml_print("DML_DLG: %s: req128_c = %0d\n", __func__, req128_c);
252 	dml_print("DML_DLG: %s: full_swath_bytes_packed_l = %0d\n",
253 		__func__,
254 		full_swath_bytes_packed_l);
255 	dml_print("DML_DLG: %s: full_swath_bytes_packed_c = %0d\n",
256 		__func__,
257 		full_swath_bytes_packed_c);
258 }
259 
260 static bool CalculateBytePerPixelAnd256BBlockSizes(
261 		enum source_format_class SourcePixelFormat,
262 		enum dm_swizzle_mode SurfaceTiling,
263 		unsigned int *BytePerPixelY,
264 		unsigned int *BytePerPixelC,
265 		double       *BytePerPixelDETY,
266 		double       *BytePerPixelDETC,
267 		unsigned int *BlockHeight256BytesY,
268 		unsigned int *BlockHeight256BytesC,
269 		unsigned int *BlockWidth256BytesY,
270 		unsigned int *BlockWidth256BytesC)
271 {
272 	if (SourcePixelFormat == dm_444_64) {
273 		*BytePerPixelDETY = 8;
274 		*BytePerPixelDETC = 0;
275 		*BytePerPixelY = 8;
276 		*BytePerPixelC = 0;
277 	} else if (SourcePixelFormat == dm_444_32 || SourcePixelFormat == dm_rgbe) {
278 		*BytePerPixelDETY = 4;
279 		*BytePerPixelDETC = 0;
280 		*BytePerPixelY = 4;
281 		*BytePerPixelC = 0;
282 	} else if (SourcePixelFormat == dm_444_16) {
283 		*BytePerPixelDETY = 2;
284 		*BytePerPixelDETC = 0;
285 		*BytePerPixelY = 2;
286 		*BytePerPixelC = 0;
287 	} else if (SourcePixelFormat == dm_444_8) {
288 		*BytePerPixelDETY = 1;
289 		*BytePerPixelDETC = 0;
290 		*BytePerPixelY = 1;
291 		*BytePerPixelC = 0;
292 	} else if (SourcePixelFormat == dm_rgbe_alpha) {
293 		*BytePerPixelDETY = 4;
294 		*BytePerPixelDETC = 1;
295 		*BytePerPixelY = 4;
296 		*BytePerPixelC = 1;
297 	} else if (SourcePixelFormat == dm_420_8) {
298 		*BytePerPixelDETY = 1;
299 		*BytePerPixelDETC = 2;
300 		*BytePerPixelY = 1;
301 		*BytePerPixelC = 2;
302 	} else if (SourcePixelFormat == dm_420_12) {
303 		*BytePerPixelDETY = 2;
304 		*BytePerPixelDETC = 4;
305 		*BytePerPixelY = 2;
306 		*BytePerPixelC = 4;
307 	} else {
308 		*BytePerPixelDETY = 4.0 / 3;
309 		*BytePerPixelDETC = 8.0 / 3;
310 		*BytePerPixelY = 2;
311 		*BytePerPixelC = 4;
312 	}
313 
314 	if ((SourcePixelFormat == dm_444_64 || SourcePixelFormat == dm_444_32
315 			|| SourcePixelFormat == dm_444_16 || SourcePixelFormat == dm_444_8
316 			|| SourcePixelFormat == dm_mono_16 || SourcePixelFormat == dm_mono_8
317 			|| SourcePixelFormat == dm_rgbe)) {
318 		if (SurfaceTiling == dm_sw_linear) {
319 			*BlockHeight256BytesY = 1;
320 		} else if (SourcePixelFormat == dm_444_64) {
321 			*BlockHeight256BytesY = 4;
322 		} else if (SourcePixelFormat == dm_444_8) {
323 			*BlockHeight256BytesY = 16;
324 		} else {
325 			*BlockHeight256BytesY = 8;
326 		}
327 		*BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY;
328 		*BlockHeight256BytesC = 0;
329 		*BlockWidth256BytesC = 0;
330 	} else {
331 		if (SurfaceTiling == dm_sw_linear) {
332 			*BlockHeight256BytesY = 1;
333 			*BlockHeight256BytesC = 1;
334 		} else if (SourcePixelFormat == dm_rgbe_alpha) {
335 			*BlockHeight256BytesY = 8;
336 			*BlockHeight256BytesC = 16;
337 		} else if (SourcePixelFormat == dm_420_8) {
338 			*BlockHeight256BytesY = 16;
339 			*BlockHeight256BytesC = 8;
340 		} else {
341 			*BlockHeight256BytesY = 8;
342 			*BlockHeight256BytesC = 8;
343 		}
344 		*BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY;
345 		*BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC;
346 	}
347 	return true;
348 }
349 
350 static void get_meta_and_pte_attr(struct display_mode_lib *mode_lib,
351 	display_data_rq_dlg_params_st *rq_dlg_param,
352 	display_data_rq_misc_params_st *rq_misc_param,
353 	display_data_rq_sizing_params_st *rq_sizing_param,
354 	unsigned int vp_width,
355 	unsigned int vp_height,
356 	unsigned int data_pitch,
357 	unsigned int meta_pitch,
358 	unsigned int source_format,
359 	unsigned int tiling,
360 	unsigned int macro_tile_size,
361 	unsigned int source_scan,
362 	unsigned int hostvm_enable,
363 	unsigned int is_chroma,
364 	unsigned int surface_height)
365 {
366 	bool surf_linear = (tiling == dm_sw_linear);
367 	bool surf_vert = (source_scan == dm_vert);
368 
369 	unsigned int bytes_per_element = 0;
370 	unsigned int bytes_per_element_y = 0;
371 	unsigned int bytes_per_element_c = 0;
372 
373 	unsigned int blk256_width = 0;
374 	unsigned int blk256_height = 0;
375 
376 	unsigned int blk256_width_y = 0;
377 	unsigned int blk256_height_y = 0;
378 	unsigned int blk256_width_c = 0;
379 	unsigned int blk256_height_c = 0;
380 	unsigned int log2_bytes_per_element = 0;
381 	unsigned int log2_blk256_width = 0;
382 	unsigned int log2_blk256_height = 0;
383 	unsigned int blk_bytes = 0;
384 	unsigned int log2_blk_bytes = 0;
385 	unsigned int log2_blk_height = 0;
386 	unsigned int log2_blk_width = 0;
387 	unsigned int log2_meta_req_bytes = 0;
388 	unsigned int log2_meta_req_height = 0;
389 	unsigned int log2_meta_req_width = 0;
390 	unsigned int meta_req_width = 0;
391 	unsigned int meta_req_height = 0;
392 	unsigned int log2_meta_row_height = 0;
393 	unsigned int meta_row_width_ub = 0;
394 	unsigned int log2_meta_chunk_bytes = 0;
395 	unsigned int log2_meta_chunk_height = 0;
396 
397 	//full sized meta chunk width in unit of data elements
398 	unsigned int log2_meta_chunk_width = 0;
399 	unsigned int log2_min_meta_chunk_bytes = 0;
400 	unsigned int min_meta_chunk_width = 0;
401 	unsigned int meta_chunk_width = 0;
402 	unsigned int meta_chunk_per_row_int = 0;
403 	unsigned int meta_row_remainder = 0;
404 	unsigned int meta_chunk_threshold = 0;
405 	unsigned int meta_blk_bytes = 0;
406 	unsigned int meta_blk_height = 0;
407 	unsigned int meta_blk_width = 0;
408 	unsigned int meta_surface_bytes = 0;
409 	unsigned int vmpg_bytes = 0;
410 	unsigned int meta_pte_req_per_frame_ub = 0;
411 	unsigned int meta_pte_bytes_per_frame_ub = 0;
412 	const unsigned int log2_vmpg_bytes = dml_log2(mode_lib->soc.gpuvm_min_page_size_bytes);
413 	const bool dual_plane_en = is_dual_plane((enum source_format_class)(source_format));
414 	const unsigned int dpte_buf_in_pte_reqs = dual_plane_en ?
415 		(is_chroma ? mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma : mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma)
416 		: (mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma + mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma);
417 
418 	unsigned int log2_vmpg_height = 0;
419 	unsigned int log2_vmpg_width = 0;
420 	unsigned int log2_dpte_req_height_ptes = 0;
421 	unsigned int log2_dpte_req_height = 0;
422 	unsigned int log2_dpte_req_width = 0;
423 	unsigned int log2_dpte_row_height_linear = 0;
424 	unsigned int log2_dpte_row_height = 0;
425 	unsigned int log2_dpte_group_width = 0;
426 	unsigned int dpte_row_width_ub = 0;
427 	unsigned int dpte_req_height = 0;
428 	unsigned int dpte_req_width = 0;
429 	unsigned int dpte_group_width = 0;
430 	unsigned int log2_dpte_group_bytes = 0;
431 	unsigned int log2_dpte_group_length = 0;
432 	double byte_per_pixel_det_y = 0;
433 	double byte_per_pixel_det_c = 0;
434 
435 	CalculateBytePerPixelAnd256BBlockSizes((enum source_format_class)(source_format),
436 		(enum dm_swizzle_mode)(tiling),
437 		&bytes_per_element_y,
438 		&bytes_per_element_c,
439 		&byte_per_pixel_det_y,
440 		&byte_per_pixel_det_c,
441 		&blk256_height_y,
442 		&blk256_height_c,
443 		&blk256_width_y,
444 		&blk256_width_c);
445 
446 	if (!is_chroma) {
447 		blk256_width = blk256_width_y;
448 		blk256_height = blk256_height_y;
449 		bytes_per_element = bytes_per_element_y;
450 	} else {
451 		blk256_width = blk256_width_c;
452 		blk256_height = blk256_height_c;
453 		bytes_per_element = bytes_per_element_c;
454 	}
455 
456 	log2_bytes_per_element = dml_log2(bytes_per_element);
457 
458 	dml_print("DML_DLG: %s: surf_linear        = %d\n", __func__, surf_linear);
459 	dml_print("DML_DLG: %s: surf_vert          = %d\n", __func__, surf_vert);
460 	dml_print("DML_DLG: %s: blk256_width       = %d\n", __func__, blk256_width);
461 	dml_print("DML_DLG: %s: blk256_height      = %d\n", __func__, blk256_height);
462 
463 	log2_blk256_width = dml_log2((double)blk256_width);
464 	log2_blk256_height = dml_log2((double)blk256_height);
465 	blk_bytes = surf_linear ?
466 		256 : get_blk_size_bytes((enum source_macro_tile_size) macro_tile_size);
467 	log2_blk_bytes = dml_log2((double)blk_bytes);
468 	log2_blk_height = 0;
469 	log2_blk_width = 0;
470 
471 	// remember log rule
472 	// "+" in log is multiply
473 	// "-" in log is divide
474 	// "/2" is like square root
475 	// blk is vertical biased
476 	if (tiling != dm_sw_linear)
477 		log2_blk_height = log2_blk256_height
478 		+ dml_ceil((double)(log2_blk_bytes - 8) / 2.0, 1);
479 	else
480 		log2_blk_height = 0;  // blk height of 1
481 
482 	log2_blk_width = log2_blk_bytes - log2_bytes_per_element - log2_blk_height;
483 
484 	if (!surf_vert) {
485 		int unsigned temp = 0;
486 
487 		temp = dml_round_to_multiple(vp_width - 1, blk256_width, 1) + blk256_width;
488 		if (data_pitch < blk256_width) {
489 			dml_print("WARNING: DML_DLG: %s: swath_size calculation ignoring data_pitch=%u < blk256_width=%u\n", __func__, data_pitch, blk256_width);
490 		} else {
491 			if (temp > data_pitch) {
492 				if (data_pitch >= vp_width)
493 					temp = data_pitch;
494 				else
495 					dml_print("WARNING: DML_DLG: %s: swath_size calculation ignoring data_pitch=%u < vp_width=%u\n", __func__, data_pitch, vp_width);
496 			}
497 		}
498 		rq_dlg_param->swath_width_ub = temp;
499 		rq_dlg_param->req_per_swath_ub = temp >> log2_blk256_width;
500 	} else {
501 		int unsigned temp = 0;
502 
503 		temp = dml_round_to_multiple(vp_height - 1, blk256_height, 1) + blk256_height;
504 		if (surface_height < blk256_height) {
505 			dml_print("WARNING: DML_DLG: %s swath_size calculation ignored surface_height=%u < blk256_height=%u\n", __func__, surface_height, blk256_height);
506 		} else {
507 			if (temp > surface_height) {
508 				if (surface_height >= vp_height)
509 					temp = surface_height;
510 				else
511 					dml_print("WARNING: DML_DLG: %s swath_size calculation ignored surface_height=%u < vp_height=%u\n", __func__, surface_height, vp_height);
512 			}
513 		}
514 		rq_dlg_param->swath_width_ub = temp;
515 		rq_dlg_param->req_per_swath_ub = temp >> log2_blk256_height;
516 	}
517 
518 	if (!surf_vert)
519 		rq_misc_param->full_swath_bytes = rq_dlg_param->swath_width_ub * blk256_height
520 		* bytes_per_element;
521 	else
522 		rq_misc_param->full_swath_bytes = rq_dlg_param->swath_width_ub * blk256_width
523 		* bytes_per_element;
524 
525 	rq_misc_param->blk256_height = blk256_height;
526 	rq_misc_param->blk256_width = blk256_width;
527 
528 	// -------
529 	// meta
530 	// -------
531 	log2_meta_req_bytes = 6; // meta request is 64b and is 8x8byte meta element
532 
533 				 // each 64b meta request for dcn is 8x8 meta elements and
534 				 // a meta element covers one 256b block of the the data surface.
535 	log2_meta_req_height = log2_blk256_height + 3; // meta req is 8x8 byte, each byte represent 1 blk256
536 	log2_meta_req_width = log2_meta_req_bytes + 8 - log2_bytes_per_element
537 		- log2_meta_req_height;
538 	meta_req_width = 1 << log2_meta_req_width;
539 	meta_req_height = 1 << log2_meta_req_height;
540 	log2_meta_row_height = 0;
541 	meta_row_width_ub = 0;
542 
543 	// the dimensions of a meta row are meta_row_width x meta_row_height in elements.
544 	// calculate upper bound of the meta_row_width
545 	if (!surf_vert) {
546 		log2_meta_row_height = log2_meta_req_height;
547 		meta_row_width_ub = dml_round_to_multiple(vp_width - 1, meta_req_width, 1)
548 			+ meta_req_width;
549 		rq_dlg_param->meta_req_per_row_ub = meta_row_width_ub / meta_req_width;
550 	} else {
551 		log2_meta_row_height = log2_meta_req_width;
552 		meta_row_width_ub = dml_round_to_multiple(vp_height - 1, meta_req_height, 1)
553 			+ meta_req_height;
554 		rq_dlg_param->meta_req_per_row_ub = meta_row_width_ub / meta_req_height;
555 	}
556 	rq_dlg_param->meta_bytes_per_row_ub = rq_dlg_param->meta_req_per_row_ub * 64;
557 
558 	rq_dlg_param->meta_row_height = 1 << log2_meta_row_height;
559 
560 	log2_meta_chunk_bytes = dml_log2(rq_sizing_param->meta_chunk_bytes);
561 	log2_meta_chunk_height = log2_meta_row_height;
562 
563 	//full sized meta chunk width in unit of data elements
564 	log2_meta_chunk_width = log2_meta_chunk_bytes + 8 - log2_bytes_per_element
565 		- log2_meta_chunk_height;
566 	log2_min_meta_chunk_bytes = dml_log2(rq_sizing_param->min_meta_chunk_bytes);
567 	min_meta_chunk_width = 1
568 		<< (log2_min_meta_chunk_bytes + 8 - log2_bytes_per_element
569 			- log2_meta_chunk_height);
570 	meta_chunk_width = 1 << log2_meta_chunk_width;
571 	meta_chunk_per_row_int = (unsigned int)(meta_row_width_ub / meta_chunk_width);
572 	meta_row_remainder = meta_row_width_ub % meta_chunk_width;
573 	meta_chunk_threshold = 0;
574 	meta_blk_bytes = 4096;
575 	meta_blk_height = blk256_height * 64;
576 	meta_blk_width = meta_blk_bytes * 256 / bytes_per_element / meta_blk_height;
577 	meta_surface_bytes = meta_pitch
578 		* (dml_round_to_multiple(vp_height - 1, meta_blk_height, 1) + meta_blk_height)
579 		* bytes_per_element / 256;
580 	vmpg_bytes = mode_lib->soc.gpuvm_min_page_size_bytes;
581 	meta_pte_req_per_frame_ub = (dml_round_to_multiple(meta_surface_bytes - vmpg_bytes,
582 		8 * vmpg_bytes,
583 		1) + 8 * vmpg_bytes) / (8 * vmpg_bytes);
584 	meta_pte_bytes_per_frame_ub = meta_pte_req_per_frame_ub * 64; //64B mpte request
585 	rq_dlg_param->meta_pte_bytes_per_frame_ub = meta_pte_bytes_per_frame_ub;
586 
587 	dml_print("DML_DLG: %s: meta_blk_height             = %d\n", __func__, meta_blk_height);
588 	dml_print("DML_DLG: %s: meta_blk_width              = %d\n", __func__, meta_blk_width);
589 	dml_print("DML_DLG: %s: meta_surface_bytes          = %d\n", __func__, meta_surface_bytes);
590 	dml_print("DML_DLG: %s: meta_pte_req_per_frame_ub   = %d\n",
591 		__func__,
592 		meta_pte_req_per_frame_ub);
593 	dml_print("DML_DLG: %s: meta_pte_bytes_per_frame_ub = %d\n",
594 		__func__,
595 		meta_pte_bytes_per_frame_ub);
596 
597 	if (!surf_vert)
598 		meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_width;
599 	else
600 		meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_height;
601 
602 	if (meta_row_remainder <= meta_chunk_threshold)
603 		rq_dlg_param->meta_chunks_per_row_ub = meta_chunk_per_row_int + 1;
604 	else
605 		rq_dlg_param->meta_chunks_per_row_ub = meta_chunk_per_row_int + 2;
606 
607 	// ------
608 	// dpte
609 	// ------
610 	if (surf_linear) {
611 		log2_vmpg_height = 0;   // one line high
612 	} else {
613 		log2_vmpg_height = (log2_vmpg_bytes - 8) / 2 + log2_blk256_height;
614 	}
615 	log2_vmpg_width = log2_vmpg_bytes - log2_bytes_per_element - log2_vmpg_height;
616 
617 	// only 3 possible shapes for dpte request in dimensions of ptes: 8x1, 4x2, 2x4.
618 	if (surf_linear) { //one 64B PTE request returns 8 PTEs
619 		log2_dpte_req_height_ptes = 0;
620 		log2_dpte_req_width = log2_vmpg_width + 3;
621 		log2_dpte_req_height = 0;
622 	} else if (log2_blk_bytes == 12) { //4KB tile means 4kB page size
623 					 //one 64B req gives 8x1 PTEs for 4KB tile
624 		log2_dpte_req_height_ptes = 0;
625 		log2_dpte_req_width = log2_blk_width + 3;
626 		log2_dpte_req_height = log2_blk_height + 0;
627 	} else if ((log2_blk_bytes >= 16) && (log2_vmpg_bytes == 12)) { // tile block >= 64KB
628 									  //two 64B reqs of 2x4 PTEs give 16 PTEs to cover 64KB
629 		log2_dpte_req_height_ptes = 4;
630 		log2_dpte_req_width = log2_blk256_width + 4; // log2_64KB_width
631 		log2_dpte_req_height = log2_blk256_height + 4; // log2_64KB_height
632 	} else { //64KB page size and must 64KB tile block
633 		   //one 64B req gives 8x1 PTEs for 64KB tile
634 		log2_dpte_req_height_ptes = 0;
635 		log2_dpte_req_width = log2_blk_width + 3;
636 		log2_dpte_req_height = log2_blk_height + 0;
637 	}
638 
639 	// The dpte request dimensions in data elements is dpte_req_width x dpte_req_height
640 	// log2_vmpg_width is how much 1 pte represent, now calculating how much a 64b pte req represent
641 	// That depends on the pte shape (i.e. 8x1, 4x2, 2x4)
642 	//log2_dpte_req_height	= log2_vmpg_height + log2_dpte_req_height_ptes;
643 	//log2_dpte_req_width	 = log2_vmpg_width + log2_dpte_req_width_ptes;
644 	dpte_req_height = 1 << log2_dpte_req_height;
645 	dpte_req_width = 1 << log2_dpte_req_width;
646 
647 	// calculate pitch dpte row buffer can hold
648 	// round the result down to a power of two.
649 	if (surf_linear) {
650 		unsigned int dpte_row_height = 0;
651 
652 		log2_dpte_row_height_linear = dml_floor(dml_log2(dpte_buf_in_pte_reqs * dpte_req_width / data_pitch), 1);
653 
654 		dml_print("DML_DLG: %s: is_chroma                   = %d\n", __func__, is_chroma);
655 		dml_print("DML_DLG: %s: dpte_buf_in_pte_reqs        = %d\n", __func__, dpte_buf_in_pte_reqs);
656 		dml_print("DML_DLG: %s: log2_dpte_row_height_linear = %d\n", __func__, log2_dpte_row_height_linear);
657 
658 		ASSERT(log2_dpte_row_height_linear >= 3);
659 
660 		if (log2_dpte_row_height_linear > 7)
661 			log2_dpte_row_height_linear = 7;
662 
663 		log2_dpte_row_height = log2_dpte_row_height_linear;
664 		// For linear, the dpte row is pitch dependent and the pte requests wrap at the pitch boundary.
665 		// the dpte_row_width_ub is the upper bound of data_pitch*dpte_row_height in elements with this unique buffering.
666 		dpte_row_height = 1 << log2_dpte_row_height;
667 		dpte_row_width_ub = dml_round_to_multiple(data_pitch * dpte_row_height - 1,
668 			dpte_req_width,
669 			1) + dpte_req_width;
670 		rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_width;
671 	} else {
672 		// the upper bound of the dpte_row_width without dependency on viewport position follows.
673 		// for tiled mode, row height is the same as req height and row store up to vp size upper bound
674 		if (!surf_vert) {
675 			log2_dpte_row_height = log2_dpte_req_height;
676 			dpte_row_width_ub = dml_round_to_multiple(vp_width - 1, dpte_req_width, 1)
677 				+ dpte_req_width;
678 			rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_width;
679 		} else {
680 			log2_dpte_row_height =
681 				(log2_blk_width < log2_dpte_req_width) ?
682 				log2_blk_width : log2_dpte_req_width;
683 			dpte_row_width_ub = dml_round_to_multiple(vp_height - 1, dpte_req_height, 1)
684 				+ dpte_req_height;
685 			rq_dlg_param->dpte_req_per_row_ub = dpte_row_width_ub / dpte_req_height;
686 		}
687 	}
688 	if (log2_blk_bytes >= 16 && log2_vmpg_bytes == 12) // tile block >= 64KB
689 		rq_dlg_param->dpte_bytes_per_row_ub = rq_dlg_param->dpte_req_per_row_ub * 128; //2*64B dpte request
690 	else
691 		rq_dlg_param->dpte_bytes_per_row_ub = rq_dlg_param->dpte_req_per_row_ub * 64; //64B dpte request
692 
693 	rq_dlg_param->dpte_row_height = 1 << log2_dpte_row_height;
694 
695 	// the dpte_group_bytes is reduced for the specific case of vertical
696 	// access of a tile surface that has dpte request of 8x1 ptes.
697 	if (hostvm_enable)
698 		rq_sizing_param->dpte_group_bytes = 512;
699 	else {
700 		if (!surf_linear & (log2_dpte_req_height_ptes == 0) & surf_vert) //reduced, in this case, will have page fault within a group
701 			rq_sizing_param->dpte_group_bytes = 512;
702 		else
703 			rq_sizing_param->dpte_group_bytes = 2048;
704 	}
705 
706 	//since pte request size is 64byte, the number of data pte requests per full sized group is as follows.
707 	log2_dpte_group_bytes = dml_log2(rq_sizing_param->dpte_group_bytes);
708 	log2_dpte_group_length = log2_dpte_group_bytes - 6; //length in 64b requests
709 
710 								// full sized data pte group width in elements
711 	if (!surf_vert)
712 		log2_dpte_group_width = log2_dpte_group_length + log2_dpte_req_width;
713 	else
714 		log2_dpte_group_width = log2_dpte_group_length + log2_dpte_req_height;
715 
716 	//But if the tile block >=64KB and the page size is 4KB, then each dPTE request is 2*64B
717 	if ((log2_blk_bytes >= 16) && (log2_vmpg_bytes == 12)) // tile block >= 64KB
718 		log2_dpte_group_width = log2_dpte_group_width - 1;
719 
720 	dpte_group_width = 1 << log2_dpte_group_width;
721 
722 	// since dpte groups are only aligned to dpte_req_width and not dpte_group_width,
723 	// the upper bound for the dpte groups per row is as follows.
724 	rq_dlg_param->dpte_groups_per_row_ub = dml_ceil((double)dpte_row_width_ub / dpte_group_width,
725 		1);
726 }
727 
728 static void get_surf_rq_param(struct display_mode_lib *mode_lib,
729 	display_data_rq_sizing_params_st *rq_sizing_param,
730 	display_data_rq_dlg_params_st *rq_dlg_param,
731 	display_data_rq_misc_params_st *rq_misc_param,
732 	const display_pipe_params_st pipe_param,
733 	bool is_chroma,
734 	bool is_alpha)
735 {
736 	bool mode_422 = 0;
737 	unsigned int vp_width = 0;
738 	unsigned int vp_height = 0;
739 	unsigned int data_pitch = 0;
740 	unsigned int meta_pitch = 0;
741 	unsigned int surface_height = 0;
742 	unsigned int ppe = mode_422 ? 2 : 1;
743 
744 	// FIXME check if ppe apply for both luma and chroma in 422 case
745 	if (is_chroma | is_alpha) {
746 		vp_width = pipe_param.src.viewport_width_c / ppe;
747 		vp_height = pipe_param.src.viewport_height_c;
748 		data_pitch = pipe_param.src.data_pitch_c;
749 		meta_pitch = pipe_param.src.meta_pitch_c;
750 		surface_height = pipe_param.src.surface_height_y / 2.0;
751 	} else {
752 		vp_width = pipe_param.src.viewport_width / ppe;
753 		vp_height = pipe_param.src.viewport_height;
754 		data_pitch = pipe_param.src.data_pitch;
755 		meta_pitch = pipe_param.src.meta_pitch;
756 		surface_height = pipe_param.src.surface_height_y;
757 	}
758 
759 	if (pipe_param.dest.odm_combine) {
760 		unsigned int access_dir = 0;
761 		unsigned int full_src_vp_width = 0;
762 		unsigned int hactive_odm = 0;
763 		unsigned int src_hactive_odm = 0;
764 		access_dir = (pipe_param.src.source_scan == dm_vert); // vp access direction: horizontal or vertical accessed
765 		hactive_odm  = pipe_param.dest.hactive / ((unsigned int)pipe_param.dest.odm_combine*2);
766 		if (is_chroma) {
767 			full_src_vp_width = pipe_param.scale_ratio_depth.hscl_ratio_c * pipe_param.dest.full_recout_width;
768 			src_hactive_odm  = pipe_param.scale_ratio_depth.hscl_ratio_c * hactive_odm;
769 		} else {
770 			full_src_vp_width = pipe_param.scale_ratio_depth.hscl_ratio * pipe_param.dest.full_recout_width;
771 			src_hactive_odm  = pipe_param.scale_ratio_depth.hscl_ratio * hactive_odm;
772 		}
773 
774 		if (access_dir == 0) {
775 			vp_width = dml_min(full_src_vp_width, src_hactive_odm);
776 			dml_print("DML_DLG: %s: vp_width = %d\n", __func__, vp_width);
777 		} else {
778 			vp_height = dml_min(full_src_vp_width, src_hactive_odm);
779 			dml_print("DML_DLG: %s: vp_height = %d\n", __func__, vp_height);
780 		}
781 		dml_print("DML_DLG: %s: full_src_vp_width = %d\n", __func__, full_src_vp_width);
782 		dml_print("DML_DLG: %s: hactive_odm = %d\n", __func__, hactive_odm);
783 		dml_print("DML_DLG: %s: src_hactive_odm = %d\n", __func__, src_hactive_odm);
784 	}
785 
786 	rq_sizing_param->chunk_bytes = 8192;
787 
788 	if (is_alpha) {
789 		rq_sizing_param->chunk_bytes = 4096;
790 	}
791 
792 	if (rq_sizing_param->chunk_bytes == 64 * 1024)
793 		rq_sizing_param->min_chunk_bytes = 0;
794 	else
795 		rq_sizing_param->min_chunk_bytes = 1024;
796 
797 	rq_sizing_param->meta_chunk_bytes = 2048;
798 	rq_sizing_param->min_meta_chunk_bytes = 256;
799 
800 	if (pipe_param.src.hostvm)
801 		rq_sizing_param->mpte_group_bytes = 512;
802 	else
803 		rq_sizing_param->mpte_group_bytes = 2048;
804 
805 	get_meta_and_pte_attr(mode_lib,
806 		rq_dlg_param,
807 		rq_misc_param,
808 		rq_sizing_param,
809 		vp_width,
810 		vp_height,
811 		data_pitch,
812 		meta_pitch,
813 		pipe_param.src.source_format,
814 		pipe_param.src.sw_mode,
815 		pipe_param.src.macro_tile_size,
816 		pipe_param.src.source_scan,
817 		pipe_param.src.hostvm,
818 		is_chroma,
819 		surface_height);
820 }
821 
822 static void dml_rq_dlg_get_rq_params(struct display_mode_lib *mode_lib,
823 	display_rq_params_st *rq_param,
824 	const display_pipe_params_st pipe_param)
825 {
826 	// get param for luma surface
827 	rq_param->yuv420 = pipe_param.src.source_format == dm_420_8
828 	|| pipe_param.src.source_format == dm_420_10
829 	|| pipe_param.src.source_format == dm_rgbe_alpha
830 	|| pipe_param.src.source_format == dm_420_12;
831 
832 	rq_param->yuv420_10bpc = pipe_param.src.source_format == dm_420_10;
833 
834 	rq_param->rgbe_alpha = (pipe_param.src.source_format == dm_rgbe_alpha)?1:0;
835 
836 	get_surf_rq_param(mode_lib,
837 		&(rq_param->sizing.rq_l),
838 		&(rq_param->dlg.rq_l),
839 		&(rq_param->misc.rq_l),
840 		pipe_param,
841 		0,
842 		0);
843 
844 	if (is_dual_plane((enum source_format_class)(pipe_param.src.source_format))) {
845 		// get param for chroma surface
846 		get_surf_rq_param(mode_lib,
847 			&(rq_param->sizing.rq_c),
848 			&(rq_param->dlg.rq_c),
849 			&(rq_param->misc.rq_c),
850 			pipe_param,
851 			1,
852 			rq_param->rgbe_alpha);
853 	}
854 
855 	// calculate how to split the det buffer space between luma and chroma
856 	handle_det_buf_split(mode_lib, rq_param, pipe_param.src);
857 	print__rq_params_st(mode_lib, *rq_param);
858 }
859 
860 void dml30_rq_dlg_get_rq_reg(struct display_mode_lib *mode_lib,
861 	display_rq_regs_st *rq_regs,
862 	const display_pipe_params_st pipe_param)
863 {
864 	display_rq_params_st rq_param = { 0 };
865 
866 	memset(rq_regs, 0, sizeof(*rq_regs));
867 	dml_rq_dlg_get_rq_params(mode_lib, &rq_param, pipe_param);
868 	extract_rq_regs(mode_lib, rq_regs, rq_param);
869 
870 	print__rq_regs_st(mode_lib, *rq_regs);
871 }
872 
873 static void calculate_ttu_cursor(struct display_mode_lib *mode_lib,
874 	double *refcyc_per_req_delivery_pre_cur,
875 	double *refcyc_per_req_delivery_cur,
876 	double refclk_freq_in_mhz,
877 	double ref_freq_to_pix_freq,
878 	double hscale_pixel_rate_l,
879 	double hscl_ratio,
880 	double vratio_pre_l,
881 	double vratio_l,
882 	unsigned int cur_width,
883 	enum cursor_bpp cur_bpp)
884 {
885 	unsigned int cur_src_width = cur_width;
886 	unsigned int cur_req_size = 0;
887 	unsigned int cur_req_width = 0;
888 	double cur_width_ub = 0.0;
889 	double cur_req_per_width = 0.0;
890 	double hactive_cur = 0.0;
891 
892 	ASSERT(cur_src_width <= 256);
893 
894 	*refcyc_per_req_delivery_pre_cur = 0.0;
895 	*refcyc_per_req_delivery_cur = 0.0;
896 	if (cur_src_width > 0) {
897 		unsigned int cur_bit_per_pixel = 0;
898 
899 		if (cur_bpp == dm_cur_2bit) {
900 			cur_req_size = 64; // byte
901 			cur_bit_per_pixel = 2;
902 		} else { // 32bit
903 			cur_bit_per_pixel = 32;
904 			if (cur_src_width >= 1 && cur_src_width <= 16)
905 				cur_req_size = 64;
906 			else if (cur_src_width >= 17 && cur_src_width <= 31)
907 				cur_req_size = 128;
908 			else
909 				cur_req_size = 256;
910 		}
911 
912 		cur_req_width = (double)cur_req_size / ((double)cur_bit_per_pixel / 8.0);
913 		cur_width_ub = dml_ceil((double)cur_src_width / (double)cur_req_width, 1)
914 			* (double)cur_req_width;
915 		cur_req_per_width = cur_width_ub / (double)cur_req_width;
916 		hactive_cur = (double)cur_src_width / hscl_ratio; // FIXME: oswin to think about what to do for cursor
917 
918 		if (vratio_pre_l <= 1.0) {
919 			*refcyc_per_req_delivery_pre_cur = hactive_cur * ref_freq_to_pix_freq
920 				/ (double)cur_req_per_width;
921 		} else {
922 			*refcyc_per_req_delivery_pre_cur = (double)refclk_freq_in_mhz
923 				* (double)cur_src_width / hscale_pixel_rate_l
924 				/ (double)cur_req_per_width;
925 		}
926 
927 		ASSERT(*refcyc_per_req_delivery_pre_cur < dml_pow(2, 13));
928 
929 		if (vratio_l <= 1.0) {
930 			*refcyc_per_req_delivery_cur = hactive_cur * ref_freq_to_pix_freq
931 				/ (double)cur_req_per_width;
932 		} else {
933 			*refcyc_per_req_delivery_cur = (double)refclk_freq_in_mhz
934 				* (double)cur_src_width / hscale_pixel_rate_l
935 				/ (double)cur_req_per_width;
936 		}
937 
938 		dml_print("DML_DLG: %s: cur_req_width                     = %d\n",
939 			__func__,
940 			cur_req_width);
941 		dml_print("DML_DLG: %s: cur_width_ub                      = %3.2f\n",
942 			__func__,
943 			cur_width_ub);
944 		dml_print("DML_DLG: %s: cur_req_per_width                 = %3.2f\n",
945 			__func__,
946 			cur_req_per_width);
947 		dml_print("DML_DLG: %s: hactive_cur                       = %3.2f\n",
948 			__func__,
949 			hactive_cur);
950 		dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_cur   = %3.2f\n",
951 			__func__,
952 			*refcyc_per_req_delivery_pre_cur);
953 		dml_print("DML_DLG: %s: refcyc_per_req_delivery_cur       = %3.2f\n",
954 			__func__,
955 			*refcyc_per_req_delivery_cur);
956 
957 		ASSERT(*refcyc_per_req_delivery_cur < dml_pow(2, 13));
958 	}
959 }
960 
961 // Note: currently taken in as is.
962 // Nice to decouple code from hw register implement and extract code that are repeated for luma and chroma.
963 static void dml_rq_dlg_get_dlg_params(struct display_mode_lib *mode_lib,
964 	const display_e2e_pipe_params_st *e2e_pipe_param,
965 	const unsigned int num_pipes,
966 	const unsigned int pipe_idx,
967 	display_dlg_regs_st *disp_dlg_regs,
968 	display_ttu_regs_st *disp_ttu_regs,
969 	const display_rq_dlg_params_st rq_dlg_param,
970 	const display_dlg_sys_params_st dlg_sys_param,
971 	const bool cstate_en,
972 	const bool pstate_en,
973 	const bool vm_en,
974 	const bool ignore_viewport_pos,
975 	const bool immediate_flip_support)
976 {
977 	const display_pipe_source_params_st *src = &e2e_pipe_param[pipe_idx].pipe.src;
978 	const display_pipe_dest_params_st *dst = &e2e_pipe_param[pipe_idx].pipe.dest;
979 	const display_output_params_st *dout = &e2e_pipe_param[pipe_idx].dout;
980 	const display_clocks_and_cfg_st *clks = &e2e_pipe_param[pipe_idx].clks_cfg;
981 	const scaler_ratio_depth_st *scl = &e2e_pipe_param[pipe_idx].pipe.scale_ratio_depth;
982 	const scaler_taps_st *taps = &e2e_pipe_param[pipe_idx].pipe.scale_taps;
983 
984 	// -------------------------
985 	// Section 1.15.2.1: OTG dependent Params
986 	// -------------------------
987 	// Timing
988 	unsigned int htotal = dst->htotal;
989 	//	unsigned int hblank_start = dst.hblank_start; // TODO: Remove
990 	unsigned int hblank_end = dst->hblank_end;
991 	unsigned int vblank_start = dst->vblank_start;
992 	unsigned int vblank_end = dst->vblank_end;
993 	unsigned int min_vblank = mode_lib->ip.min_vblank_lines;
994 
995 	double dppclk_freq_in_mhz = clks->dppclk_mhz;
996 	double dispclk_freq_in_mhz = clks->dispclk_mhz;
997 	double refclk_freq_in_mhz = clks->refclk_mhz;
998 	double pclk_freq_in_mhz = dst->pixel_rate_mhz;
999 	bool interlaced = dst->interlaced;
1000 
1001 	double ref_freq_to_pix_freq = refclk_freq_in_mhz / pclk_freq_in_mhz;
1002 
1003 	double min_dcfclk_mhz = 0;
1004 	double t_calc_us = 0;
1005 	double min_ttu_vblank = 0;
1006 
1007 	double min_dst_y_ttu_vblank = 0;
1008 	unsigned int dlg_vblank_start = 0;
1009 	bool dual_plane = false;
1010 	bool mode_422 = false;
1011 	unsigned int access_dir = 0;
1012 	unsigned int vp_height_l = 0;
1013 	unsigned int vp_width_l = 0;
1014 	unsigned int vp_height_c = 0;
1015 	unsigned int vp_width_c = 0;
1016 
1017 	// Scaling
1018 	unsigned int htaps_l = 0;
1019 	unsigned int htaps_c = 0;
1020 	double hratio_l = 0;
1021 	double hratio_c = 0;
1022 	double vratio_l = 0;
1023 	double vratio_c = 0;
1024 	bool scl_enable = false;
1025 
1026 	double line_time_in_us = 0;
1027 	//	double vinit_l;
1028 	//	double vinit_c;
1029 	//	double vinit_bot_l;
1030 	//	double vinit_bot_c;
1031 
1032 	//	unsigned int swath_height_l;
1033 	unsigned int swath_width_ub_l = 0;
1034 	//	unsigned int dpte_bytes_per_row_ub_l;
1035 	unsigned int dpte_groups_per_row_ub_l = 0;
1036 	//	unsigned int meta_pte_bytes_per_frame_ub_l;
1037 	//	unsigned int meta_bytes_per_row_ub_l;
1038 
1039 	//	unsigned int swath_height_c;
1040 	unsigned int swath_width_ub_c = 0;
1041 	//   unsigned int dpte_bytes_per_row_ub_c;
1042 	unsigned int dpte_groups_per_row_ub_c = 0;
1043 
1044 	unsigned int meta_chunks_per_row_ub_l = 0;
1045 	unsigned int meta_chunks_per_row_ub_c = 0;
1046 	unsigned int vupdate_offset = 0;
1047 	unsigned int vupdate_width = 0;
1048 	unsigned int vready_offset = 0;
1049 
1050 	unsigned int dppclk_delay_subtotal = 0;
1051 	unsigned int dispclk_delay_subtotal = 0;
1052 	unsigned int pixel_rate_delay_subtotal = 0;
1053 
1054 	unsigned int vstartup_start = 0;
1055 	unsigned int dst_x_after_scaler = 0;
1056 	unsigned int dst_y_after_scaler = 0;
1057 	double line_wait = 0;
1058 	double dst_y_prefetch = 0;
1059 	double dst_y_per_vm_vblank = 0;
1060 	double dst_y_per_row_vblank = 0;
1061 	double dst_y_per_vm_flip = 0;
1062 	double dst_y_per_row_flip = 0;
1063 	double max_dst_y_per_vm_vblank = 0;
1064 	double max_dst_y_per_row_vblank = 0;
1065 	double lsw = 0;
1066 	double vratio_pre_l = 0;
1067 	double vratio_pre_c = 0;
1068 	unsigned int req_per_swath_ub_l = 0;
1069 	unsigned int req_per_swath_ub_c = 0;
1070 	unsigned int meta_row_height_l = 0;
1071 	unsigned int meta_row_height_c = 0;
1072 	unsigned int swath_width_pixels_ub_l = 0;
1073 	unsigned int swath_width_pixels_ub_c = 0;
1074 	unsigned int scaler_rec_in_width_l = 0;
1075 	unsigned int scaler_rec_in_width_c = 0;
1076 	unsigned int dpte_row_height_l = 0;
1077 	unsigned int dpte_row_height_c = 0;
1078 	double hscale_pixel_rate_l = 0;
1079 	double hscale_pixel_rate_c = 0;
1080 	double min_hratio_fact_l = 0;
1081 	double min_hratio_fact_c = 0;
1082 	double refcyc_per_line_delivery_pre_l = 0;
1083 	double refcyc_per_line_delivery_pre_c = 0;
1084 	double refcyc_per_line_delivery_l = 0;
1085 	double refcyc_per_line_delivery_c = 0;
1086 
1087 	double refcyc_per_req_delivery_pre_l = 0;
1088 	double refcyc_per_req_delivery_pre_c = 0;
1089 	double refcyc_per_req_delivery_l = 0;
1090 	double refcyc_per_req_delivery_c = 0;
1091 
1092 	unsigned int full_recout_width = 0;
1093 	double refcyc_per_req_delivery_pre_cur0 = 0;
1094 	double refcyc_per_req_delivery_cur0 = 0;
1095 	double refcyc_per_req_delivery_pre_cur1 = 0;
1096 	double refcyc_per_req_delivery_cur1 = 0;
1097 
1098 	unsigned int pipe_index_in_combine[DC__NUM_PIPES__MAX] = { 0 };
1099 
1100 	memset(disp_dlg_regs, 0, sizeof(*disp_dlg_regs));
1101 	memset(disp_ttu_regs, 0, sizeof(*disp_ttu_regs));
1102 
1103 	dml_print("DML_DLG: %s:  cstate_en = %d\n", __func__, cstate_en);
1104 	dml_print("DML_DLG: %s:  pstate_en = %d\n", __func__, pstate_en);
1105 	dml_print("DML_DLG: %s:  vm_en     = %d\n", __func__, vm_en);
1106 	dml_print("DML_DLG: %s:  ignore_viewport_pos  = %d\n", __func__, ignore_viewport_pos);
1107 	dml_print("DML_DLG: %s:  immediate_flip_support  = %d\n", __func__, immediate_flip_support);
1108 
1109 	dml_print("DML_DLG: %s: dppclk_freq_in_mhz     = %3.2f\n", __func__, dppclk_freq_in_mhz);
1110 	dml_print("DML_DLG: %s: dispclk_freq_in_mhz    = %3.2f\n", __func__, dispclk_freq_in_mhz);
1111 	dml_print("DML_DLG: %s: refclk_freq_in_mhz     = %3.2f\n", __func__, refclk_freq_in_mhz);
1112 	dml_print("DML_DLG: %s: pclk_freq_in_mhz       = %3.2f\n", __func__, pclk_freq_in_mhz);
1113 	dml_print("DML_DLG: %s: interlaced             = %d\n", __func__, interlaced);
1114 	ASSERT(ref_freq_to_pix_freq < 4.0);
1115 
1116 	disp_dlg_regs->ref_freq_to_pix_freq =
1117 		(unsigned int)(ref_freq_to_pix_freq * dml_pow(2, 19));
1118 	disp_dlg_regs->refcyc_per_htotal = (unsigned int)(ref_freq_to_pix_freq * (double)htotal
1119 		* dml_pow(2, 8));
1120 	disp_dlg_regs->dlg_vblank_end = interlaced ? (vblank_end / 2) : vblank_end; // 15 bits
1121 
1122 	min_dcfclk_mhz = dlg_sys_param.deepsleep_dcfclk_mhz;
1123 	t_calc_us = get_tcalc(mode_lib, e2e_pipe_param, num_pipes);
1124 	min_ttu_vblank = get_min_ttu_vblank(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1125 
1126 	min_dst_y_ttu_vblank = min_ttu_vblank * pclk_freq_in_mhz / (double)htotal;
1127 	dlg_vblank_start = interlaced ? (vblank_start / 2) : vblank_start;
1128 
1129 	disp_dlg_regs->min_dst_y_next_start = (unsigned int)(((double)dlg_vblank_start
1130 		) * dml_pow(2, 2));
1131 	ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int)dml_pow(2, 18));
1132 
1133 	dml_print("DML_DLG: %s: min_dcfclk_mhz                         = %3.2f\n",
1134 		__func__,
1135 		min_dcfclk_mhz);
1136 	dml_print("DML_DLG: %s: min_ttu_vblank                         = %3.2f\n",
1137 		__func__,
1138 		min_ttu_vblank);
1139 	dml_print("DML_DLG: %s: min_dst_y_ttu_vblank                   = %3.2f\n",
1140 		__func__,
1141 		min_dst_y_ttu_vblank);
1142 	dml_print("DML_DLG: %s: t_calc_us                              = %3.2f\n",
1143 		__func__,
1144 		t_calc_us);
1145 	dml_print("DML_DLG: %s: disp_dlg_regs->min_dst_y_next_start    = 0x%0x\n",
1146 		__func__,
1147 		disp_dlg_regs->min_dst_y_next_start);
1148 	dml_print("DML_DLG: %s: ref_freq_to_pix_freq                   = %3.2f\n",
1149 		__func__,
1150 		ref_freq_to_pix_freq);
1151 
1152 	// -------------------------
1153 	// Section 1.15.2.2: Prefetch, Active and TTU
1154 	// -------------------------
1155 	// Prefetch Calc
1156 	// Source
1157 	//			 dcc_en			  = src.dcc;
1158 	dual_plane = is_dual_plane((enum source_format_class)(src->source_format));
1159 	mode_422 = false; // TODO
1160 	access_dir = (src->source_scan == dm_vert); // vp access direction: horizontal or vertical accessed
1161 	vp_height_l = src->viewport_height;
1162 	vp_width_l = src->viewport_width;
1163 	vp_height_c = src->viewport_height_c;
1164 	vp_width_c = src->viewport_width_c;
1165 
1166 	// Scaling
1167 	htaps_l = taps->htaps;
1168 	htaps_c = taps->htaps_c;
1169 	hratio_l = scl->hscl_ratio;
1170 	hratio_c = scl->hscl_ratio_c;
1171 	vratio_l = scl->vscl_ratio;
1172 	vratio_c = scl->vscl_ratio_c;
1173 	scl_enable = scl->scl_enable;
1174 
1175 	line_time_in_us = (htotal / pclk_freq_in_mhz);
1176 	swath_width_ub_l = rq_dlg_param.rq_l.swath_width_ub;
1177 	dpte_groups_per_row_ub_l = rq_dlg_param.rq_l.dpte_groups_per_row_ub;
1178 	swath_width_ub_c = rq_dlg_param.rq_c.swath_width_ub;
1179 	dpte_groups_per_row_ub_c = rq_dlg_param.rq_c.dpte_groups_per_row_ub;
1180 
1181 	meta_chunks_per_row_ub_l = rq_dlg_param.rq_l.meta_chunks_per_row_ub;
1182 	meta_chunks_per_row_ub_c = rq_dlg_param.rq_c.meta_chunks_per_row_ub;
1183 	vupdate_offset = dst->vupdate_offset;
1184 	vupdate_width = dst->vupdate_width;
1185 	vready_offset = dst->vready_offset;
1186 
1187 	dppclk_delay_subtotal = mode_lib->ip.dppclk_delay_subtotal;
1188 	dispclk_delay_subtotal = mode_lib->ip.dispclk_delay_subtotal;
1189 
1190 	if (scl_enable)
1191 		dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_scl;
1192 	else
1193 		dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_scl_lb_only;
1194 
1195 	dppclk_delay_subtotal += mode_lib->ip.dppclk_delay_cnvc_formatter
1196 		+ src->num_cursors * mode_lib->ip.dppclk_delay_cnvc_cursor;
1197 
1198 	if (dout->dsc_enable) {
1199 		double dsc_delay = get_dsc_delay(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1200 
1201 		dispclk_delay_subtotal += dsc_delay;
1202 	}
1203 
1204 	pixel_rate_delay_subtotal = dppclk_delay_subtotal * pclk_freq_in_mhz / dppclk_freq_in_mhz
1205 		+ dispclk_delay_subtotal * pclk_freq_in_mhz / dispclk_freq_in_mhz;
1206 
1207 	vstartup_start = dst->vstartup_start;
1208 	if (interlaced) {
1209 		if (vstartup_start / 2.0
1210 			- (double)(vready_offset + vupdate_width + vupdate_offset) / htotal
1211 			<= vblank_end / 2.0)
1212 			disp_dlg_regs->vready_after_vcount0 = 1;
1213 		else
1214 			disp_dlg_regs->vready_after_vcount0 = 0;
1215 	} else {
1216 		if (vstartup_start
1217 			- (double)(vready_offset + vupdate_width + vupdate_offset) / htotal
1218 			<= vblank_end)
1219 			disp_dlg_regs->vready_after_vcount0 = 1;
1220 		else
1221 			disp_dlg_regs->vready_after_vcount0 = 0;
1222 	}
1223 
1224 	// TODO: Where is this coming from?
1225 	if (interlaced)
1226 		vstartup_start = vstartup_start / 2;
1227 
1228 	// TODO: What if this min_vblank doesn't match the value in the dml_config_settings.cpp?
1229 	if (vstartup_start >= min_vblank) {
1230 		dml_print("WARNING: DML_DLG: %s: vblank_start=%d vblank_end=%d\n",
1231 			__func__,
1232 			vblank_start,
1233 			vblank_end);
1234 		dml_print("WARNING: DML_DLG: %s: vstartup_start=%d should be less than min_vblank=%d\n",
1235 			__func__,
1236 			vstartup_start,
1237 			min_vblank);
1238 		min_vblank = vstartup_start + 1;
1239 		dml_print("WARNING: DML_DLG: %s: vstartup_start=%d should be less than min_vblank=%d\n",
1240 			__func__,
1241 			vstartup_start,
1242 			min_vblank);
1243 	}
1244 
1245 	dst_x_after_scaler = get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1246 	dst_y_after_scaler = get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1247 
1248 	// do some adjustment on the dst_after scaler to account for odm combine mode
1249 	dml_print("DML_DLG: %s: input dst_x_after_scaler                     = %d\n",
1250 		__func__,
1251 		dst_x_after_scaler);
1252 	dml_print("DML_DLG: %s: input dst_y_after_scaler                     = %d\n",
1253 		__func__,
1254 		dst_y_after_scaler);
1255 
1256 	// need to figure out which side of odm combine we're in
1257 	if (dst->odm_combine) {
1258 		// figure out which pipes go together
1259 		bool visited[DC__NUM_PIPES__MAX] = { false };
1260 		unsigned int i, j, k;
1261 
1262 		for (k = 0; k < num_pipes; ++k) {
1263 			visited[k] = false;
1264 			pipe_index_in_combine[k] = 0;
1265 		}
1266 
1267 		for (i = 0; i < num_pipes; i++) {
1268 			if (e2e_pipe_param[i].pipe.src.is_hsplit && !visited[i]) {
1269 
1270 				unsigned int grp = e2e_pipe_param[i].pipe.src.hsplit_grp;
1271 				unsigned int grp_idx = 0;
1272 
1273 				for (j = i; j < num_pipes; j++) {
1274 					if (e2e_pipe_param[j].pipe.src.hsplit_grp == grp
1275 							&& e2e_pipe_param[j].pipe.src.is_hsplit && !visited[j]) {
1276 						pipe_index_in_combine[j] = grp_idx;
1277 						dml_print("DML_DLG: %s: pipe[%d] is in grp %d idx %d\n", __func__, j, grp, grp_idx);
1278 						grp_idx++;
1279 						visited[j] = true;
1280 					}
1281 				}
1282 			}
1283 		}
1284 
1285 	}
1286 
1287 	if (dst->odm_combine == dm_odm_combine_mode_disabled) {
1288 		disp_dlg_regs->refcyc_h_blank_end = (unsigned int)((double) hblank_end * ref_freq_to_pix_freq);
1289 	} else {
1290 		unsigned int   odm_combine_factor = (dst->odm_combine == dm_odm_combine_mode_2to1 ? 2 : 4); // TODO: We should really check that 4to1 is supported before setting it to 4
1291 		unsigned int   odm_pipe_index = pipe_index_in_combine[pipe_idx];
1292 		disp_dlg_regs->refcyc_h_blank_end = (unsigned int)(((double) hblank_end + odm_pipe_index * (double) dst->hactive / odm_combine_factor) * ref_freq_to_pix_freq);
1293 	}
1294 	ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int)dml_pow(2, 13));
1295 
1296 	dml_print("DML_DLG: %s: htotal                                     = %d\n", __func__, htotal);
1297 	dml_print("DML_DLG: %s: pixel_rate_delay_subtotal                  = %d\n",
1298 		__func__,
1299 		pixel_rate_delay_subtotal);
1300 	dml_print("DML_DLG: %s: dst_x_after_scaler[%d]                     = %d\n",
1301 		__func__,
1302 		pipe_idx,
1303 		dst_x_after_scaler);
1304 	dml_print("DML_DLG: %s: dst_y_after_scaler[%d]                     = %d\n",
1305 		__func__,
1306 		pipe_idx,
1307 		dst_y_after_scaler);
1308 
1309 	// Lwait
1310 		// TODO: Should this be urgent_latency_pixel_mixed_with_vm_data_us?
1311 	line_wait = mode_lib->soc.urgent_latency_pixel_data_only_us;
1312 	if (cstate_en)
1313 		line_wait = dml_max(mode_lib->soc.sr_enter_plus_exit_time_us, line_wait);
1314 	if (pstate_en)
1315 		line_wait = dml_max(mode_lib->soc.dram_clock_change_latency_us
1316 			+ mode_lib->soc.urgent_latency_pixel_data_only_us, // TODO: Should this be urgent_latency_pixel_mixed_with_vm_data_us?
1317 			line_wait);
1318 	line_wait = line_wait / line_time_in_us;
1319 
1320 	dst_y_prefetch = get_dst_y_prefetch(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1321 	dml_print("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, dst_y_prefetch);
1322 
1323 	dst_y_per_vm_vblank = get_dst_y_per_vm_vblank(mode_lib,
1324 		e2e_pipe_param,
1325 		num_pipes,
1326 		pipe_idx);
1327 	dst_y_per_row_vblank = get_dst_y_per_row_vblank(mode_lib,
1328 		e2e_pipe_param,
1329 		num_pipes,
1330 		pipe_idx);
1331 	dst_y_per_vm_flip = get_dst_y_per_vm_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1332 	dst_y_per_row_flip = get_dst_y_per_row_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1333 
1334 	max_dst_y_per_vm_vblank = 32.0;	 //U5.2
1335 	max_dst_y_per_row_vblank = 16.0;	//U4.2
1336 
1337 	// magic!
1338 	if (htotal <= 75) {
1339 		min_vblank = 300;
1340 		max_dst_y_per_vm_vblank = 100.0;
1341 		max_dst_y_per_row_vblank = 100.0;
1342 	}
1343 
1344 	dml_print("DML_DLG: %s: dst_y_per_vm_flip    = %3.2f\n", __func__, dst_y_per_vm_flip);
1345 	dml_print("DML_DLG: %s: dst_y_per_row_flip   = %3.2f\n", __func__, dst_y_per_row_flip);
1346 	dml_print("DML_DLG: %s: dst_y_per_vm_vblank  = %3.2f\n", __func__, dst_y_per_vm_vblank);
1347 	dml_print("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, dst_y_per_row_vblank);
1348 
1349 	ASSERT(dst_y_per_vm_vblank < max_dst_y_per_vm_vblank);
1350 	ASSERT(dst_y_per_row_vblank < max_dst_y_per_row_vblank);
1351 
1352 	ASSERT(dst_y_prefetch > (dst_y_per_vm_vblank + dst_y_per_row_vblank));
1353 	lsw = dst_y_prefetch - (dst_y_per_vm_vblank + dst_y_per_row_vblank);
1354 
1355 	dml_print("DML_DLG: %s: lsw = %3.2f\n", __func__, lsw);
1356 
1357 	vratio_pre_l = get_vratio_prefetch_l(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1358 	vratio_pre_c = get_vratio_prefetch_c(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
1359 
1360 	dml_print("DML_DLG: %s: vratio_pre_l=%3.2f\n", __func__, vratio_pre_l);
1361 	dml_print("DML_DLG: %s: vratio_pre_c=%3.2f\n", __func__, vratio_pre_c);
1362 
1363 	// Active
1364 	req_per_swath_ub_l = rq_dlg_param.rq_l.req_per_swath_ub;
1365 	req_per_swath_ub_c = rq_dlg_param.rq_c.req_per_swath_ub;
1366 	meta_row_height_l = rq_dlg_param.rq_l.meta_row_height;
1367 	meta_row_height_c = rq_dlg_param.rq_c.meta_row_height;
1368 	swath_width_pixels_ub_l = 0;
1369 	swath_width_pixels_ub_c = 0;
1370 	scaler_rec_in_width_l = 0;
1371 	scaler_rec_in_width_c = 0;
1372 	dpte_row_height_l = rq_dlg_param.rq_l.dpte_row_height;
1373 	dpte_row_height_c = rq_dlg_param.rq_c.dpte_row_height;
1374 
1375 	if (mode_422) {
1376 		swath_width_pixels_ub_l = swath_width_ub_l * 2;  // *2 for 2 pixel per element
1377 		swath_width_pixels_ub_c = swath_width_ub_c * 2;
1378 	} else {
1379 		swath_width_pixels_ub_l = swath_width_ub_l * 1;
1380 		swath_width_pixels_ub_c = swath_width_ub_c * 1;
1381 	}
1382 
1383 	hscale_pixel_rate_l = 0.;
1384 	hscale_pixel_rate_c = 0.;
1385 	min_hratio_fact_l = 1.0;
1386 	min_hratio_fact_c = 1.0;
1387 
1388 	if (hratio_l <= 1)
1389 		min_hratio_fact_l = 2.0;
1390 	else if (htaps_l <= 6) {
1391 		if ((hratio_l * 2.0) > 4.0)
1392 			min_hratio_fact_l = 4.0;
1393 		else
1394 			min_hratio_fact_l = hratio_l * 2.0;
1395 	} else {
1396 		if (hratio_l > 4.0)
1397 			min_hratio_fact_l = 4.0;
1398 		else
1399 			min_hratio_fact_l = hratio_l;
1400 	}
1401 
1402 	hscale_pixel_rate_l = min_hratio_fact_l * dppclk_freq_in_mhz;
1403 
1404 	if (hratio_c <= 1)
1405 		min_hratio_fact_c = 2.0;
1406 	else if (htaps_c <= 6) {
1407 		if ((hratio_c * 2.0) > 4.0)
1408 			min_hratio_fact_c = 4.0;
1409 		else
1410 			min_hratio_fact_c = hratio_c * 2.0;
1411 	} else {
1412 		if (hratio_c > 4.0)
1413 			min_hratio_fact_c = 4.0;
1414 		else
1415 			min_hratio_fact_c = hratio_c;
1416 	}
1417 
1418 	hscale_pixel_rate_c = min_hratio_fact_c * dppclk_freq_in_mhz;
1419 
1420 	refcyc_per_line_delivery_pre_l = 0.;
1421 	refcyc_per_line_delivery_pre_c = 0.;
1422 	refcyc_per_line_delivery_l = 0.;
1423 	refcyc_per_line_delivery_c = 0.;
1424 
1425 	refcyc_per_req_delivery_pre_l = 0.;
1426 	refcyc_per_req_delivery_pre_c = 0.;
1427 	refcyc_per_req_delivery_l = 0.;
1428 	refcyc_per_req_delivery_c = 0.;
1429 
1430 	full_recout_width = 0;
1431 	// In ODM
1432 	if (src->is_hsplit) {
1433 		// This "hack"  is only allowed (and valid) for MPC combine. In ODM
1434 		// combine, you MUST specify the full_recout_width...according to Oswin
1435 		if (dst->full_recout_width == 0 && !dst->odm_combine) {
1436 			dml_print("DML_DLG: %s: Warning: full_recout_width not set in hsplit mode\n",
1437 				__func__);
1438 			full_recout_width = dst->recout_width * 2; // assume half split for dcn1
1439 		} else
1440 			full_recout_width = dst->full_recout_width;
1441 	} else
1442 		full_recout_width = dst->recout_width;
1443 
1444 	// As of DCN2, mpc_combine and odm_combine are mutually exclusive
1445 	refcyc_per_line_delivery_pre_l = get_refcyc_per_delivery(mode_lib,
1446 		refclk_freq_in_mhz,
1447 		pclk_freq_in_mhz,
1448 		dst->odm_combine,
1449 		full_recout_width,
1450 		dst->hactive,
1451 		vratio_pre_l,
1452 		hscale_pixel_rate_l,
1453 		swath_width_pixels_ub_l,
1454 		1); // per line
1455 
1456 	refcyc_per_line_delivery_l = get_refcyc_per_delivery(mode_lib,
1457 		refclk_freq_in_mhz,
1458 		pclk_freq_in_mhz,
1459 		dst->odm_combine,
1460 		full_recout_width,
1461 		dst->hactive,
1462 		vratio_l,
1463 		hscale_pixel_rate_l,
1464 		swath_width_pixels_ub_l,
1465 		1); // per line
1466 
1467 	dml_print("DML_DLG: %s: full_recout_width              = %d\n",
1468 		__func__,
1469 		full_recout_width);
1470 	dml_print("DML_DLG: %s: hscale_pixel_rate_l            = %3.2f\n",
1471 		__func__,
1472 		hscale_pixel_rate_l);
1473 	dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n",
1474 		__func__,
1475 		refcyc_per_line_delivery_pre_l);
1476 	dml_print("DML_DLG: %s: refcyc_per_line_delivery_l     = %3.2f\n",
1477 		__func__,
1478 		refcyc_per_line_delivery_l);
1479 
1480 	if (dual_plane) {
1481 		refcyc_per_line_delivery_pre_c = get_refcyc_per_delivery(mode_lib,
1482 			refclk_freq_in_mhz,
1483 			pclk_freq_in_mhz,
1484 			dst->odm_combine,
1485 			full_recout_width,
1486 			dst->hactive,
1487 			vratio_pre_c,
1488 			hscale_pixel_rate_c,
1489 			swath_width_pixels_ub_c,
1490 			1); // per line
1491 
1492 		refcyc_per_line_delivery_c = get_refcyc_per_delivery(mode_lib,
1493 			refclk_freq_in_mhz,
1494 			pclk_freq_in_mhz,
1495 			dst->odm_combine,
1496 			full_recout_width,
1497 			dst->hactive,
1498 			vratio_c,
1499 			hscale_pixel_rate_c,
1500 			swath_width_pixels_ub_c,
1501 			1);  // per line
1502 
1503 		dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n",
1504 			__func__,
1505 			refcyc_per_line_delivery_pre_c);
1506 		dml_print("DML_DLG: %s: refcyc_per_line_delivery_c     = %3.2f\n",
1507 			__func__,
1508 			refcyc_per_line_delivery_c);
1509 	}
1510 
1511 	// smehta: this is a hack added until we get the real dml, sorry, need to make progress
1512 	if (src->dynamic_metadata_enable && src->gpuvm) {
1513 		unsigned int levels = mode_lib->ip.gpuvm_max_page_table_levels;
1514 		double ref_cycles;
1515 
1516 		if (src->hostvm)
1517 			levels = levels * (mode_lib->ip.hostvm_max_page_table_levels+1);
1518 
1519 		ref_cycles = (levels * mode_lib->soc.urgent_latency_vm_data_only_us) * refclk_freq_in_mhz;
1520 		dml_print("BENyamin:    dst_y_prefetch                  = %f %d %f %f \n",
1521 			ref_cycles, levels, mode_lib->soc.urgent_latency_vm_data_only_us, refclk_freq_in_mhz);
1522 		disp_dlg_regs->refcyc_per_vm_dmdata = (unsigned int) ref_cycles;
1523 	}
1524 	dml_print("BENyamin:    dmdta_en vm                     = %d %d \n",
1525 		src->dynamic_metadata_enable, src->vm);
1526 	// TTU - Luma / Chroma
1527 	if (access_dir) {  // vertical access
1528 		scaler_rec_in_width_l = vp_height_l;
1529 		scaler_rec_in_width_c = vp_height_c;
1530 	} else {
1531 		scaler_rec_in_width_l = vp_width_l;
1532 		scaler_rec_in_width_c = vp_width_c;
1533 	}
1534 
1535 	refcyc_per_req_delivery_pre_l = get_refcyc_per_delivery(mode_lib,
1536 		refclk_freq_in_mhz,
1537 		pclk_freq_in_mhz,
1538 		dst->odm_combine,
1539 		full_recout_width,
1540 		dst->hactive,
1541 		vratio_pre_l,
1542 		hscale_pixel_rate_l,
1543 		scaler_rec_in_width_l,
1544 		req_per_swath_ub_l);  // per req
1545 	refcyc_per_req_delivery_l = get_refcyc_per_delivery(mode_lib,
1546 		refclk_freq_in_mhz,
1547 		pclk_freq_in_mhz,
1548 		dst->odm_combine,
1549 		full_recout_width,
1550 		dst->hactive,
1551 		vratio_l,
1552 		hscale_pixel_rate_l,
1553 		scaler_rec_in_width_l,
1554 		req_per_swath_ub_l);  // per req
1555 
1556 	dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n",
1557 		__func__,
1558 		refcyc_per_req_delivery_pre_l);
1559 	dml_print("DML_DLG: %s: refcyc_per_req_delivery_l     = %3.2f\n",
1560 		__func__,
1561 		refcyc_per_req_delivery_l);
1562 
1563 	ASSERT(refcyc_per_req_delivery_pre_l < dml_pow(2, 13));
1564 	ASSERT(refcyc_per_req_delivery_l < dml_pow(2, 13));
1565 
1566 	if (dual_plane) {
1567 		refcyc_per_req_delivery_pre_c = get_refcyc_per_delivery(mode_lib,
1568 			refclk_freq_in_mhz,
1569 			pclk_freq_in_mhz,
1570 			dst->odm_combine,
1571 			full_recout_width,
1572 			dst->hactive,
1573 			vratio_pre_c,
1574 			hscale_pixel_rate_c,
1575 			scaler_rec_in_width_c,
1576 			req_per_swath_ub_c);  // per req
1577 		refcyc_per_req_delivery_c = get_refcyc_per_delivery(mode_lib,
1578 			refclk_freq_in_mhz,
1579 			pclk_freq_in_mhz,
1580 			dst->odm_combine,
1581 			full_recout_width,
1582 			dst->hactive,
1583 			vratio_c,
1584 			hscale_pixel_rate_c,
1585 			scaler_rec_in_width_c,
1586 			req_per_swath_ub_c);  // per req
1587 
1588 		dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n",
1589 			__func__,
1590 			refcyc_per_req_delivery_pre_c);
1591 		dml_print("DML_DLG: %s: refcyc_per_req_delivery_c     = %3.2f\n",
1592 			__func__,
1593 			refcyc_per_req_delivery_c);
1594 
1595 		ASSERT(refcyc_per_req_delivery_pre_c < dml_pow(2, 13));
1596 		ASSERT(refcyc_per_req_delivery_c < dml_pow(2, 13));
1597 	}
1598 
1599 	// TTU - Cursor
1600 	refcyc_per_req_delivery_pre_cur0 = 0.0;
1601 	refcyc_per_req_delivery_cur0 = 0.0;
1602 	if (src->num_cursors > 0) {
1603 		calculate_ttu_cursor(mode_lib,
1604 			&refcyc_per_req_delivery_pre_cur0,
1605 			&refcyc_per_req_delivery_cur0,
1606 			refclk_freq_in_mhz,
1607 			ref_freq_to_pix_freq,
1608 			hscale_pixel_rate_l,
1609 			scl->hscl_ratio,
1610 			vratio_pre_l,
1611 			vratio_l,
1612 			src->cur0_src_width,
1613 			(enum cursor_bpp)(src->cur0_bpp));
1614 	}
1615 
1616 	refcyc_per_req_delivery_pre_cur1 = 0.0;
1617 	refcyc_per_req_delivery_cur1 = 0.0;
1618 	if (src->num_cursors > 1) {
1619 		calculate_ttu_cursor(mode_lib,
1620 			&refcyc_per_req_delivery_pre_cur1,
1621 			&refcyc_per_req_delivery_cur1,
1622 			refclk_freq_in_mhz,
1623 			ref_freq_to_pix_freq,
1624 			hscale_pixel_rate_l,
1625 			scl->hscl_ratio,
1626 			vratio_pre_l,
1627 			vratio_l,
1628 			src->cur1_src_width,
1629 			(enum cursor_bpp)(src->cur1_bpp));
1630 	}
1631 
1632 	// TTU - Misc
1633 	// all hard-coded
1634 
1635 	// Assignment to register structures
1636 	disp_dlg_regs->dst_y_after_scaler = dst_y_after_scaler; // in terms of line
1637 	ASSERT(disp_dlg_regs->dst_y_after_scaler < (unsigned int)8);
1638 	disp_dlg_regs->refcyc_x_after_scaler = dst_x_after_scaler * ref_freq_to_pix_freq; // in terms of refclk
1639 	ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int)dml_pow(2, 13));
1640 	disp_dlg_regs->dst_y_prefetch = (unsigned int)(dst_y_prefetch * dml_pow(2, 2));
1641 	disp_dlg_regs->dst_y_per_vm_vblank = (unsigned int)(dst_y_per_vm_vblank * dml_pow(2, 2));
1642 	disp_dlg_regs->dst_y_per_row_vblank = (unsigned int)(dst_y_per_row_vblank * dml_pow(2, 2));
1643 	disp_dlg_regs->dst_y_per_vm_flip = (unsigned int)(dst_y_per_vm_flip * dml_pow(2, 2));
1644 	disp_dlg_regs->dst_y_per_row_flip = (unsigned int)(dst_y_per_row_flip * dml_pow(2, 2));
1645 
1646 	disp_dlg_regs->vratio_prefetch = (unsigned int)(vratio_pre_l * dml_pow(2, 19));
1647 	disp_dlg_regs->vratio_prefetch_c = (unsigned int)(vratio_pre_c * dml_pow(2, 19));
1648 
1649 	dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank  = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank);
1650 	dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank);
1651 	dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip    = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip);
1652 	dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip   = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip);
1653 	disp_dlg_regs->refcyc_per_pte_group_vblank_l =
1654 		(unsigned int)(dst_y_per_row_vblank * (double)htotal
1655 			* ref_freq_to_pix_freq / (double)dpte_groups_per_row_ub_l);
1656 	ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int)dml_pow(2, 13));
1657 
1658 	if (dual_plane) {
1659 		disp_dlg_regs->refcyc_per_pte_group_vblank_c = (unsigned int)(dst_y_per_row_vblank
1660 			* (double)htotal * ref_freq_to_pix_freq
1661 			/ (double)dpte_groups_per_row_ub_c);
1662 		ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c
1663 			< (unsigned int)dml_pow(2, 13));
1664 	}
1665 
1666 	disp_dlg_regs->refcyc_per_meta_chunk_vblank_l =
1667 		(unsigned int)(dst_y_per_row_vblank * (double)htotal
1668 			* ref_freq_to_pix_freq / (double)meta_chunks_per_row_ub_l);
1669 	ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_l < (unsigned int)dml_pow(2, 13));
1670 
1671 	disp_dlg_regs->refcyc_per_meta_chunk_vblank_c =
1672 		disp_dlg_regs->refcyc_per_meta_chunk_vblank_l; // dcc for 4:2:0 is not supported in dcn1.0.  assigned to be the same as _l for now
1673 
1674 	disp_dlg_regs->refcyc_per_pte_group_flip_l = (unsigned int)(dst_y_per_row_flip * htotal
1675 		* ref_freq_to_pix_freq) / dpte_groups_per_row_ub_l;
1676 	disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (unsigned int)(dst_y_per_row_flip * htotal
1677 		* ref_freq_to_pix_freq) / meta_chunks_per_row_ub_l;
1678 
1679 	if (dual_plane) {
1680 		disp_dlg_regs->refcyc_per_pte_group_flip_c = (unsigned int)(dst_y_per_row_flip
1681 			* htotal * ref_freq_to_pix_freq) / dpte_groups_per_row_ub_c;
1682 		disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (unsigned int)(dst_y_per_row_flip
1683 			* htotal * ref_freq_to_pix_freq) / meta_chunks_per_row_ub_c;
1684 	}
1685 
1686 	disp_dlg_regs->refcyc_per_vm_group_vblank   = get_refcyc_per_vm_group_vblank(mode_lib, e2e_pipe_param, num_pipes, pipe_idx) * refclk_freq_in_mhz;
1687 	disp_dlg_regs->refcyc_per_vm_group_flip	 = get_refcyc_per_vm_group_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx) * refclk_freq_in_mhz;
1688 	disp_dlg_regs->refcyc_per_vm_req_vblank	 = get_refcyc_per_vm_req_vblank(mode_lib, e2e_pipe_param, num_pipes, pipe_idx) * refclk_freq_in_mhz * dml_pow(2, 10);
1689 	disp_dlg_regs->refcyc_per_vm_req_flip	   = get_refcyc_per_vm_req_flip(mode_lib, e2e_pipe_param, num_pipes, pipe_idx) * refclk_freq_in_mhz * dml_pow(2, 10);
1690 
1691 	// Clamp to max for now
1692 	if (disp_dlg_regs->refcyc_per_vm_group_vblank >= (unsigned int)dml_pow(2, 23))
1693 		disp_dlg_regs->refcyc_per_vm_group_vblank = dml_pow(2, 23) - 1;
1694 
1695 	if (disp_dlg_regs->refcyc_per_vm_group_flip >= (unsigned int)dml_pow(2, 23))
1696 		disp_dlg_regs->refcyc_per_vm_group_flip = dml_pow(2, 23) - 1;
1697 
1698 	if (disp_dlg_regs->refcyc_per_vm_req_vblank >= (unsigned int)dml_pow(2, 23))
1699 		disp_dlg_regs->refcyc_per_vm_req_vblank = dml_pow(2, 23) - 1;
1700 
1701 	if (disp_dlg_regs->refcyc_per_vm_req_flip >= (unsigned int)dml_pow(2, 23))
1702 		disp_dlg_regs->refcyc_per_vm_req_flip = dml_pow(2, 23) - 1;
1703 
1704 	disp_dlg_regs->dst_y_per_pte_row_nom_l = (unsigned int)((double)dpte_row_height_l
1705 		/ (double)vratio_l * dml_pow(2, 2));
1706 	ASSERT(disp_dlg_regs->dst_y_per_pte_row_nom_l < (unsigned int)dml_pow(2, 17));
1707 
1708 	if (dual_plane) {
1709 		disp_dlg_regs->dst_y_per_pte_row_nom_c = (unsigned int)((double)dpte_row_height_c
1710 			/ (double)vratio_c * dml_pow(2, 2));
1711 		if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (unsigned int)dml_pow(2, 17)) {
1712 			dml_print("DML_DLG: %s: Warning dst_y_per_pte_row_nom_c %u larger than supported by register format U15.2 %u\n",
1713 				__func__,
1714 				disp_dlg_regs->dst_y_per_pte_row_nom_c,
1715 				(unsigned int)dml_pow(2, 17) - 1);
1716 		}
1717 	}
1718 
1719 	disp_dlg_regs->dst_y_per_meta_row_nom_l = (unsigned int)((double)meta_row_height_l
1720 		/ (double)vratio_l * dml_pow(2, 2));
1721 	ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_l < (unsigned int)dml_pow(2, 17));
1722 
1723 	disp_dlg_regs->dst_y_per_meta_row_nom_c = disp_dlg_regs->dst_y_per_meta_row_nom_l; // TODO: dcc for 4:2:0 is not supported in dcn1.0.  assigned to be the same as _l for now
1724 
1725 	dml_print("DML: Trow: %fus\n", line_time_in_us * (double)dpte_row_height_l / (double)vratio_l);
1726 
1727 	disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int)((double)dpte_row_height_l
1728 		/ (double)vratio_l * (double)htotal * ref_freq_to_pix_freq
1729 		/ (double)dpte_groups_per_row_ub_l);
1730 	if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (unsigned int)dml_pow(2, 23))
1731 		disp_dlg_regs->refcyc_per_pte_group_nom_l = dml_pow(2, 23) - 1;
1732 	disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (unsigned int)((double)meta_row_height_l
1733 		/ (double)vratio_l * (double)htotal * ref_freq_to_pix_freq
1734 		/ (double)meta_chunks_per_row_ub_l);
1735 	if (disp_dlg_regs->refcyc_per_meta_chunk_nom_l >= (unsigned int)dml_pow(2, 23))
1736 		disp_dlg_regs->refcyc_per_meta_chunk_nom_l = dml_pow(2, 23) - 1;
1737 
1738 	if (dual_plane) {
1739 		disp_dlg_regs->refcyc_per_pte_group_nom_c =
1740 			(unsigned int)((double)dpte_row_height_c / (double)vratio_c
1741 				* (double)htotal * ref_freq_to_pix_freq
1742 				/ (double)dpte_groups_per_row_ub_c);
1743 		if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (unsigned int)dml_pow(2, 23))
1744 			disp_dlg_regs->refcyc_per_pte_group_nom_c = dml_pow(2, 23) - 1;
1745 
1746 		// TODO: Is this the right calculation? Does htotal need to be halved?
1747 		disp_dlg_regs->refcyc_per_meta_chunk_nom_c =
1748 			(unsigned int)((double)meta_row_height_c / (double)vratio_c
1749 				* (double)htotal * ref_freq_to_pix_freq
1750 				/ (double)meta_chunks_per_row_ub_c);
1751 		if (disp_dlg_regs->refcyc_per_meta_chunk_nom_c >= (unsigned int)dml_pow(2, 23))
1752 			disp_dlg_regs->refcyc_per_meta_chunk_nom_c = dml_pow(2, 23) - 1;
1753 	}
1754 
1755 	disp_dlg_regs->refcyc_per_line_delivery_pre_l = (unsigned int)dml_floor(refcyc_per_line_delivery_pre_l,
1756 		1);
1757 	disp_dlg_regs->refcyc_per_line_delivery_l = (unsigned int)dml_floor(refcyc_per_line_delivery_l,
1758 		1);
1759 	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int)dml_pow(2, 13));
1760 	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int)dml_pow(2, 13));
1761 
1762 	disp_dlg_regs->refcyc_per_line_delivery_pre_c = (unsigned int)dml_floor(refcyc_per_line_delivery_pre_c,
1763 		1);
1764 	disp_dlg_regs->refcyc_per_line_delivery_c = (unsigned int)dml_floor(refcyc_per_line_delivery_c,
1765 		1);
1766 	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int)dml_pow(2, 13));
1767 	ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int)dml_pow(2, 13));
1768 
1769 	disp_dlg_regs->chunk_hdl_adjust_cur0 = 3;
1770 	disp_dlg_regs->dst_y_offset_cur0 = 0;
1771 	disp_dlg_regs->chunk_hdl_adjust_cur1 = 3;
1772 	disp_dlg_regs->dst_y_offset_cur1 = 0;
1773 
1774 	disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off
1775 
1776 	disp_ttu_regs->refcyc_per_req_delivery_pre_l = (unsigned int)(refcyc_per_req_delivery_pre_l
1777 		* dml_pow(2, 10));
1778 	disp_ttu_regs->refcyc_per_req_delivery_l = (unsigned int)(refcyc_per_req_delivery_l
1779 		* dml_pow(2, 10));
1780 	disp_ttu_regs->refcyc_per_req_delivery_pre_c = (unsigned int)(refcyc_per_req_delivery_pre_c
1781 		* dml_pow(2, 10));
1782 	disp_ttu_regs->refcyc_per_req_delivery_c = (unsigned int)(refcyc_per_req_delivery_c
1783 		* dml_pow(2, 10));
1784 	disp_ttu_regs->refcyc_per_req_delivery_pre_cur0 =
1785 		(unsigned int)(refcyc_per_req_delivery_pre_cur0 * dml_pow(2, 10));
1786 	disp_ttu_regs->refcyc_per_req_delivery_cur0 = (unsigned int)(refcyc_per_req_delivery_cur0
1787 		* dml_pow(2, 10));
1788 	disp_ttu_regs->refcyc_per_req_delivery_pre_cur1 =
1789 		(unsigned int)(refcyc_per_req_delivery_pre_cur1 * dml_pow(2, 10));
1790 	disp_ttu_regs->refcyc_per_req_delivery_cur1 = (unsigned int)(refcyc_per_req_delivery_cur1
1791 		* dml_pow(2, 10));
1792 	disp_ttu_regs->qos_level_low_wm = 0;
1793 	ASSERT(disp_ttu_regs->qos_level_low_wm < dml_pow(2, 14));
1794 	disp_ttu_regs->qos_level_high_wm = (unsigned int)(4.0 * (double)htotal
1795 		* ref_freq_to_pix_freq);
1796 	ASSERT(disp_ttu_regs->qos_level_high_wm < dml_pow(2, 14));
1797 
1798 	disp_ttu_regs->qos_level_flip = 14;
1799 	disp_ttu_regs->qos_level_fixed_l = 8;
1800 	disp_ttu_regs->qos_level_fixed_c = 8;
1801 	disp_ttu_regs->qos_level_fixed_cur0 = 8;
1802 	disp_ttu_regs->qos_ramp_disable_l = 0;
1803 	disp_ttu_regs->qos_ramp_disable_c = 0;
1804 	disp_ttu_regs->qos_ramp_disable_cur0 = 0;
1805 
1806 	disp_ttu_regs->min_ttu_vblank = min_ttu_vblank * refclk_freq_in_mhz;
1807 	ASSERT(disp_ttu_regs->min_ttu_vblank < dml_pow(2, 24));
1808 
1809 	print__ttu_regs_st(mode_lib, *disp_ttu_regs);
1810 	print__dlg_regs_st(mode_lib, *disp_dlg_regs);
1811 }
1812 
1813 void dml30_rq_dlg_get_dlg_reg(struct display_mode_lib *mode_lib,
1814 	display_dlg_regs_st *dlg_regs,
1815 	display_ttu_regs_st *ttu_regs,
1816 	display_e2e_pipe_params_st *e2e_pipe_param,
1817 	const unsigned int num_pipes,
1818 	const unsigned int pipe_idx,
1819 	const bool cstate_en,
1820 	const bool pstate_en,
1821 	const bool vm_en,
1822 	const bool ignore_viewport_pos,
1823 	const bool immediate_flip_support)
1824 {
1825 	display_rq_params_st rq_param = { 0 };
1826 	display_dlg_sys_params_st dlg_sys_param = { 0 };
1827 
1828 	// Get watermark and Tex.
1829 	dlg_sys_param.t_urg_wm_us = get_wm_urgent(mode_lib, e2e_pipe_param, num_pipes);
1830 	dlg_sys_param.deepsleep_dcfclk_mhz = get_clk_dcf_deepsleep(mode_lib,
1831 		e2e_pipe_param,
1832 		num_pipes);
1833 	dlg_sys_param.t_extra_us = get_urgent_extra_latency(mode_lib, e2e_pipe_param, num_pipes);
1834 	dlg_sys_param.mem_trip_us = get_wm_memory_trip(mode_lib, e2e_pipe_param, num_pipes);
1835 	dlg_sys_param.t_mclk_wm_us = get_wm_dram_clock_change(mode_lib, e2e_pipe_param, num_pipes);
1836 	dlg_sys_param.t_sr_wm_us = get_wm_stutter_enter_exit(mode_lib, e2e_pipe_param, num_pipes);
1837 	dlg_sys_param.total_flip_bw = get_total_immediate_flip_bw(mode_lib,
1838 		e2e_pipe_param,
1839 		num_pipes);
1840 	dlg_sys_param.total_flip_bytes = get_total_immediate_flip_bytes(mode_lib,
1841 		e2e_pipe_param,
1842 		num_pipes);
1843 	dlg_sys_param.t_srx_delay_us = mode_lib->ip.dcfclk_cstate_latency
1844 		/ dlg_sys_param.deepsleep_dcfclk_mhz; // TODO: Deprecated
1845 
1846 	print__dlg_sys_params_st(mode_lib, dlg_sys_param);
1847 
1848 	// system parameter calculation done
1849 
1850 	dml_print("DML_DLG: Calculation for pipe[%d] start\n\n", pipe_idx);
1851 	dml_rq_dlg_get_rq_params(mode_lib, &rq_param, e2e_pipe_param[pipe_idx].pipe);
1852 	dml_rq_dlg_get_dlg_params(mode_lib,
1853 		e2e_pipe_param,
1854 		num_pipes,
1855 		pipe_idx,
1856 		dlg_regs,
1857 		ttu_regs,
1858 		rq_param.dlg,
1859 		dlg_sys_param,
1860 		cstate_en,
1861 		pstate_en,
1862 		vm_en,
1863 		ignore_viewport_pos,
1864 		immediate_flip_support);
1865 	dml_print("DML_DLG: Calculation for pipe[%d] end\n", pipe_idx);
1866 }
1867 
1868 #endif
1869