xe_fb_pin.c 10.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12
// SPDX-License-Identifier: MIT
/*
 * Copyright © 2021 Intel Corporation
 */

#include "i915_drv.h"
#include "intel_display_types.h"
#include "intel_dpt.h"
#include "intel_fb.h"
#include "intel_fb_pin.h"
#include "xe_ggtt.h"
#include "xe_gt.h"
13
#include "xe_pm.h"
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

#include <drm/ttm/ttm_bo.h>

static void
write_dpt_rotated(struct xe_bo *bo, struct iosys_map *map, u32 *dpt_ofs, u32 bo_ofs,
		  u32 width, u32 height, u32 src_stride, u32 dst_stride)
{
	struct xe_device *xe = xe_bo_device(bo);
	struct xe_ggtt *ggtt = xe_device_get_root_tile(xe)->mem.ggtt;
	u32 column, row;

	/* TODO: Maybe rewrite so we can traverse the bo addresses sequentially,
	 * by writing dpt/ggtt in a different order?
	 */

	for (column = 0; column < width; column++) {
		u32 src_idx = src_stride * (height - 1) + column + bo_ofs;

		for (row = 0; row < height; row++) {
			u64 pte = ggtt->pt_ops->pte_encode_bo(bo, src_idx * XE_PAGE_SIZE,
34
							      xe->pat.idx[XE_CACHE_NONE]);
35 36 37 38 39 40 41 42 43 44 45 46 47 48

			iosys_map_wr(map, *dpt_ofs, u64, pte);
			*dpt_ofs += 8;
			src_idx -= src_stride;
		}

		/* The DE ignores the PTEs for the padding tiles */
		*dpt_ofs += (dst_stride - height) * 8;
	}

	/* Align to next page */
	*dpt_ofs = ALIGN(*dpt_ofs, 4096);
}

49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
static void
write_dpt_remapped(struct xe_bo *bo, struct iosys_map *map, u32 *dpt_ofs,
		   u32 bo_ofs, u32 width, u32 height, u32 src_stride,
		   u32 dst_stride)
{
	struct xe_device *xe = xe_bo_device(bo);
	struct xe_ggtt *ggtt = xe_device_get_root_tile(xe)->mem.ggtt;
	u64 (*pte_encode_bo)(struct xe_bo *bo, u64 bo_offset, u16 pat_index)
		= ggtt->pt_ops->pte_encode_bo;
	u32 column, row;

	for (row = 0; row < height; row++) {
		u32 src_idx = src_stride * row + bo_ofs;

		for (column = 0; column < width; column++) {
			iosys_map_wr(map, *dpt_ofs, u64,
				     pte_encode_bo(bo, src_idx * XE_PAGE_SIZE,
66
				     xe->pat.idx[XE_CACHE_NONE]));
67 68 69 70 71 72 73 74 75 76 77 78 79

			*dpt_ofs += 8;
			src_idx++;
		}

		/* The DE ignores the PTEs for the padding tiles */
		*dpt_ofs += (dst_stride - width) * 8;
	}

	/* Align to next page */
	*dpt_ofs = ALIGN(*dpt_ofs, 4096);
}

80 81 82 83 84 85 86 87 88 89 90 91
static int __xe_pin_fb_vma_dpt(struct intel_framebuffer *fb,
			       const struct i915_gtt_view *view,
			       struct i915_vma *vma)
{
	struct xe_device *xe = to_xe_device(fb->base.dev);
	struct xe_tile *tile0 = xe_device_get_root_tile(xe);
	struct xe_ggtt *ggtt = tile0->mem.ggtt;
	struct xe_bo *bo = intel_fb_obj(&fb->base), *dpt;
	u32 dpt_size, size = bo->ttm.base.size;

	if (view->type == I915_GTT_VIEW_NORMAL)
		dpt_size = ALIGN(size / XE_PAGE_SIZE * 8, XE_PAGE_SIZE);
92 93 94
	else if (view->type == I915_GTT_VIEW_REMAPPED)
		dpt_size = ALIGN(intel_remapped_info_size(&fb->remapped_view.gtt.remapped) * 8,
				 XE_PAGE_SIZE);
95 96 97 98 99
	else
		/* display uses 4K tiles instead of bytes here, convert to entries.. */
		dpt_size = ALIGN(intel_rotation_info_size(&view->rotated) * 8,
				 XE_PAGE_SIZE);

100 101 102
	if (IS_DGFX(xe))
		dpt = xe_bo_create_pin_map(xe, tile0, NULL, dpt_size,
					   ttm_bo_type_kernel,
103 104 105
					   XE_BO_FLAG_VRAM0 |
					   XE_BO_FLAG_GGTT |
					   XE_BO_FLAG_PAGETABLE);
106
	else
107 108
		dpt = xe_bo_create_pin_map(xe, tile0, NULL, dpt_size,
					   ttm_bo_type_kernel,
109 110 111
					   XE_BO_FLAG_STOLEN |
					   XE_BO_FLAG_GGTT |
					   XE_BO_FLAG_PAGETABLE);
112 113 114
	if (IS_ERR(dpt))
		dpt = xe_bo_create_pin_map(xe, tile0, NULL, dpt_size,
					   ttm_bo_type_kernel,
115 116 117
					   XE_BO_FLAG_SYSTEM |
					   XE_BO_FLAG_GGTT |
					   XE_BO_FLAG_PAGETABLE);
118 119 120 121 122 123 124 125
	if (IS_ERR(dpt))
		return PTR_ERR(dpt);

	if (view->type == I915_GTT_VIEW_NORMAL) {
		u32 x;

		for (x = 0; x < size / XE_PAGE_SIZE; x++) {
			u64 pte = ggtt->pt_ops->pte_encode_bo(bo, x * XE_PAGE_SIZE,
126
							      xe->pat.idx[XE_CACHE_NONE]);
127 128 129

			iosys_map_wr(&dpt->vmap, x * 8, u64, pte);
		}
130 131 132 133 134 135 136 137 138 139 140 141
	} else if (view->type == I915_GTT_VIEW_REMAPPED) {
		const struct intel_remapped_info *remap_info = &view->remapped;
		u32 i, dpt_ofs = 0;

		for (i = 0; i < ARRAY_SIZE(remap_info->plane); i++)
			write_dpt_remapped(bo, &dpt->vmap, &dpt_ofs,
					   remap_info->plane[i].offset,
					   remap_info->plane[i].width,
					   remap_info->plane[i].height,
					   remap_info->plane[i].src_stride,
					   remap_info->plane[i].dst_stride);

142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
	} else {
		const struct intel_rotation_info *rot_info = &view->rotated;
		u32 i, dpt_ofs = 0;

		for (i = 0; i < ARRAY_SIZE(rot_info->plane); i++)
			write_dpt_rotated(bo, &dpt->vmap, &dpt_ofs,
					  rot_info->plane[i].offset,
					  rot_info->plane[i].width,
					  rot_info->plane[i].height,
					  rot_info->plane[i].src_stride,
					  rot_info->plane[i].dst_stride);
	}

	vma->dpt = dpt;
	vma->node = dpt->ggtt_node;
	return 0;
}

static void
write_ggtt_rotated(struct xe_bo *bo, struct xe_ggtt *ggtt, u32 *ggtt_ofs, u32 bo_ofs,
		   u32 width, u32 height, u32 src_stride, u32 dst_stride)
{
	struct xe_device *xe = xe_bo_device(bo);
	u32 column, row;

	for (column = 0; column < width; column++) {
		u32 src_idx = src_stride * (height - 1) + column + bo_ofs;

		for (row = 0; row < height; row++) {
			u64 pte = ggtt->pt_ops->pte_encode_bo(bo, src_idx * XE_PAGE_SIZE,
172
							      xe->pat.idx[XE_CACHE_NONE]);
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196

			xe_ggtt_set_pte(ggtt, *ggtt_ofs, pte);
			*ggtt_ofs += XE_PAGE_SIZE;
			src_idx -= src_stride;
		}

		/* The DE ignores the PTEs for the padding tiles */
		*ggtt_ofs += (dst_stride - height) * XE_PAGE_SIZE;
	}
}

static int __xe_pin_fb_vma_ggtt(struct intel_framebuffer *fb,
				const struct i915_gtt_view *view,
				struct i915_vma *vma)
{
	struct xe_bo *bo = intel_fb_obj(&fb->base);
	struct xe_device *xe = to_xe_device(fb->base.dev);
	struct xe_ggtt *ggtt = xe_device_get_root_tile(xe)->mem.ggtt;
	u32 align;
	int ret;

	/* TODO: Consider sharing framebuffer mapping?
	 * embed i915_vma inside intel_framebuffer
	 */
197
	xe_pm_runtime_get_noresume(tile_to_xe(ggtt->tile));
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217
	ret = mutex_lock_interruptible(&ggtt->lock);
	if (ret)
		goto out;

	align = XE_PAGE_SIZE;
	if (xe_bo_is_vram(bo) && ggtt->flags & XE_GGTT_FLAGS_64K)
		align = max_t(u32, align, SZ_64K);

	if (bo->ggtt_node.size && view->type == I915_GTT_VIEW_NORMAL) {
		vma->node = bo->ggtt_node;
	} else if (view->type == I915_GTT_VIEW_NORMAL) {
		u32 x, size = bo->ttm.base.size;

		ret = xe_ggtt_insert_special_node_locked(ggtt, &vma->node, size,
							 align, 0);
		if (ret)
			goto out_unlock;

		for (x = 0; x < size; x += XE_PAGE_SIZE) {
			u64 pte = ggtt->pt_ops->pte_encode_bo(bo, x,
218
							      xe->pat.idx[XE_CACHE_NONE]);
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247

			xe_ggtt_set_pte(ggtt, vma->node.start + x, pte);
		}
	} else {
		u32 i, ggtt_ofs;
		const struct intel_rotation_info *rot_info = &view->rotated;

		/* display seems to use tiles instead of bytes here, so convert it back.. */
		u32 size = intel_rotation_info_size(rot_info) * XE_PAGE_SIZE;

		ret = xe_ggtt_insert_special_node_locked(ggtt, &vma->node, size,
							 align, 0);
		if (ret)
			goto out_unlock;

		ggtt_ofs = vma->node.start;

		for (i = 0; i < ARRAY_SIZE(rot_info->plane); i++)
			write_ggtt_rotated(bo, ggtt, &ggtt_ofs,
					   rot_info->plane[i].offset,
					   rot_info->plane[i].width,
					   rot_info->plane[i].height,
					   rot_info->plane[i].src_stride,
					   rot_info->plane[i].dst_stride);
	}

out_unlock:
	mutex_unlock(&ggtt->lock);
out:
248
	xe_pm_runtime_put(tile_to_xe(ggtt->tile));
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
	return ret;
}

static struct i915_vma *__xe_pin_fb_vma(struct intel_framebuffer *fb,
					const struct i915_gtt_view *view)
{
	struct drm_device *dev = fb->base.dev;
	struct xe_device *xe = to_xe_device(dev);
	struct i915_vma *vma = kzalloc(sizeof(*vma), GFP_KERNEL);
	struct xe_bo *bo = intel_fb_obj(&fb->base);
	int ret;

	if (!vma)
		return ERR_PTR(-ENODEV);

264 265
	if (IS_DGFX(to_xe_device(bo->ttm.base.dev)) &&
	    intel_fb_rc_ccs_cc_plane(&fb->base) >= 0 &&
266
	    !(bo->flags & XE_BO_FLAG_NEEDS_CPU_ACCESS)) {
267 268 269 270 271 272 273 274 275 276 277 278 279 280
		struct xe_tile *tile = xe_device_get_root_tile(xe);

		/*
		 * If we need to able to access the clear-color value stored in
		 * the buffer, then we require that such buffers are also CPU
		 * accessible.  This is important on small-bar systems where
		 * only some subset of VRAM is CPU accessible.
		 */
		if (tile->mem.vram.io_size < tile->mem.vram.usable_size) {
			ret = -EINVAL;
			goto err;
		}
	}

281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326
	/*
	 * Pin the framebuffer, we can't use xe_bo_(un)pin functions as the
	 * assumptions are incorrect for framebuffers
	 */
	ret = ttm_bo_reserve(&bo->ttm, false, false, NULL);
	if (ret)
		goto err;

	if (IS_DGFX(xe))
		ret = xe_bo_migrate(bo, XE_PL_VRAM0);
	else
		ret = xe_bo_validate(bo, NULL, true);
	if (!ret)
		ttm_bo_pin(&bo->ttm);
	ttm_bo_unreserve(&bo->ttm);
	if (ret)
		goto err;

	vma->bo = bo;
	if (intel_fb_uses_dpt(&fb->base))
		ret = __xe_pin_fb_vma_dpt(fb, view, vma);
	else
		ret = __xe_pin_fb_vma_ggtt(fb, view, vma);
	if (ret)
		goto err_unpin;

	return vma;

err_unpin:
	ttm_bo_reserve(&bo->ttm, false, false, NULL);
	ttm_bo_unpin(&bo->ttm);
	ttm_bo_unreserve(&bo->ttm);
err:
	kfree(vma);
	return ERR_PTR(ret);
}

static void __xe_unpin_fb_vma(struct i915_vma *vma)
{
	struct xe_device *xe = to_xe_device(vma->bo->ttm.base.dev);
	struct xe_ggtt *ggtt = xe_device_get_root_tile(xe)->mem.ggtt;

	if (vma->dpt)
		xe_bo_unpin_map_no_vm(vma->dpt);
	else if (!drm_mm_node_allocated(&vma->bo->ggtt_node) ||
		 vma->bo->ggtt_node.start != vma->node.start)
327
		xe_ggtt_remove_node(ggtt, &vma->node, false);
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358

	ttm_bo_reserve(&vma->bo->ttm, false, false, NULL);
	ttm_bo_unpin(&vma->bo->ttm);
	ttm_bo_unreserve(&vma->bo->ttm);
	kfree(vma);
}

struct i915_vma *
intel_pin_and_fence_fb_obj(struct drm_framebuffer *fb,
			   bool phys_cursor,
			   const struct i915_gtt_view *view,
			   bool uses_fence,
			   unsigned long *out_flags)
{
	*out_flags = 0;

	return __xe_pin_fb_vma(to_intel_framebuffer(fb), view);
}

void intel_unpin_fb_vma(struct i915_vma *vma, unsigned long flags)
{
	__xe_unpin_fb_vma(vma);
}

int intel_plane_pin_fb(struct intel_plane_state *plane_state)
{
	struct drm_framebuffer *fb = plane_state->hw.fb;
	struct xe_bo *bo = intel_fb_obj(fb);
	struct i915_vma *vma;

	/* We reject creating !SCANOUT fb's, so this is weird.. */
359
	drm_WARN_ON(bo->ttm.base.dev, !(bo->flags & XE_BO_FLAG_SCANOUT));
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386

	vma = __xe_pin_fb_vma(to_intel_framebuffer(fb), &plane_state->view.gtt);
	if (IS_ERR(vma))
		return PTR_ERR(vma);

	plane_state->ggtt_vma = vma;
	return 0;
}

void intel_plane_unpin_fb(struct intel_plane_state *old_plane_state)
{
	__xe_unpin_fb_vma(old_plane_state->ggtt_vma);
	old_plane_state->ggtt_vma = NULL;
}

/*
 * For Xe introduce dummy intel_dpt_create which just return NULL and
 * intel_dpt_destroy which does nothing.
 */
struct i915_address_space *intel_dpt_create(struct intel_framebuffer *fb)
{
	return NULL;
}

void intel_dpt_destroy(struct i915_address_space *vm)
{
	return;
387
}