2 * Copyright 2007-2008 Maarten Maathuis
3 * Copyright 2008 Stephane Marchesin
5 * Permission is hereby granted, free of charge, to any person obtaining a
6 * copy of this software and associated documentation files (the "Software"),
7 * to deal in the Software without restriction, including without limitation
8 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
9 * and/or sell copies of the Software, and to permit persons to whom the
10 * Software is furnished to do so, subject to the following conditions:
12 * The above copyright notice and this permission notice (including the next
13 * paragraph) shall be included in all copies or substantial portions of the
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
21 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
22 * DEALINGS IN THE SOFTWARE.
30 #include <X11/extensions/Xv.h>
33 #include "dixstruct.h"
36 #include "nv_include.h"
39 #include "nv30_shaders.h"
41 extern Atom xvSyncToVBlank, xvSetDefaults;
44 * The filtering function used for video scaling. We use a cubic filter as defined in
45 * "Reconstruction Filters in Computer Graphics"
46 * Mitchell & Netravali in SIGGRAPH '88
48 static float filter_func(float x)
51 const double C=(1.0-B)/2.0;
57 return ( (12.0-9.0*B-6.0*C)*x3+(-18.0+12.0*B+6.0*C)*x2+(6.0-2.0*B) )/6.0;
59 return ( (-B-6.0*C)*x3+(6.0*B+30.0*C)*x2+(-12.0*B-48.0*C)*x1+(8.0*B+24.0*C) )/6.0;
62 static int8_t f32tosb8(float v)
64 return (int8_t)(v*127.0);
68 * 512 means 2048 bytes of VRAM
70 #define TABLE_SIZE 512
71 static void compute_filter_table(int8_t *t) {
74 for(i=0;i<TABLE_SIZE;i++) {
77 float w0=filter_func(x+1.0);
78 float w1=filter_func(x);
79 float w2=filter_func(x-1.0);
80 float w3=filter_func(x-2.0);
82 t[4*i+2]=f32tosb8(1.0+x-w1/(w0+w1));
83 t[4*i+1]=f32tosb8(1.0-x+w3/(w2+w3));
84 t[4*i+0]=f32tosb8(w0+w1);
85 t[4*i+3]=f32tosb8(0.0);
89 static struct nouveau_bo *table_mem = NULL;
91 NV30_LoadFilterTable(ScrnInfoPtr pScrn)
93 NVPtr pNv = NVPTR(pScrn);
96 if (nouveau_bo_new(pNv->dev, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART,
97 0, TABLE_SIZE*sizeof(float)*4, &table_mem)) {
98 xf86DrvMsg(pScrn->scrnIndex, X_ERROR,
99 "Couldn't alloc filter table!\n");
103 if (nouveau_bo_map(table_mem, NOUVEAU_BO_RDWR)) {
104 xf86DrvMsg(pScrn->scrnIndex, X_ERROR,
105 "Couldn't map filter table!\n");
109 int8_t *t=table_mem->map;
110 compute_filter_table(t);
114 #define SWIZZLE(ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w) \
116 NV34TCL_TX_SWIZZLE_S0_X_##ts0x | NV34TCL_TX_SWIZZLE_S0_Y_##ts0y | \
117 NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w | \
118 NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y | \
119 NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w \
123 * Texture 0 : filter table
125 * Texture 2 : UV data
128 NV30VideoTexture(ScrnInfoPtr pScrn, struct nouveau_bo *src, int offset,
129 uint16_t width, uint16_t height, uint16_t src_pitch, int unit)
131 NVPtr pNv = NVPTR(pScrn);
132 struct nouveau_channel *chan = pNv->chan;
133 struct nouveau_grobj *rankine = pNv->Nv3D;
135 uint32_t card_fmt = 0;
136 uint32_t card_swz = 0;
140 card_fmt = NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8;
141 card_swz = SWIZZLE(S1, S1, S1, S1, X, Y, Z, W);
144 card_fmt = NV34TCL_TX_FORMAT_FORMAT_A8_RECT2;
145 card_swz = SWIZZLE(S1, S1, S1, S1, X, X, X, X);
148 card_fmt = NV34TCL_TX_FORMAT_FORMAT_L8A8_RECT;
149 #if X_BYTE_ORDER == X_BIG_ENDIAN
150 card_swz = SWIZZLE(S1, S1, S1, S1, Z, W, X, Y); /* x = V, y = U */
152 card_swz = SWIZZLE(S1, S1, S1, S1, W, Z, Y, X); /* x = V, y = U */
157 BEGIN_RING(chan, rankine, NV34TCL_TX_OFFSET(unit), 8);
158 OUT_RELOCl(chan, src, offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
160 OUT_RELOCd(chan, pNv->FB, NV34TCL_TX_FORMAT_DIMS_1D | card_fmt |
161 (1 << NV34TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT) |
163 NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT) |
165 NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT) |
167 NOUVEAU_BO_VRAM | NOUVEAU_BO_RD,
168 NV34TCL_TX_FORMAT_DMA0, 0);
169 OUT_RING (chan, NV34TCL_TX_WRAP_S_REPEAT |
170 NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE |
171 NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE);
173 OUT_RELOCd(chan, pNv->FB, NV34TCL_TX_FORMAT_DIMS_2D | card_fmt |
174 (1 << NV34TCL_TX_FORMAT_MIPMAP_LEVELS_SHIFT) |
176 NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT) |
178 NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT) |
180 NOUVEAU_BO_VRAM | NOUVEAU_BO_RD,
181 NV34TCL_TX_FORMAT_DMA0, 0);
182 OUT_RING (chan, NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE |
183 NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE |
184 NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE);
187 OUT_RING (chan, NV34TCL_TX_ENABLE_ENABLE);
188 OUT_RING (chan, (src_pitch << NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT) |
191 OUT_RING (chan, NV34TCL_TX_FILTER_SIGNED_ALPHA |
192 NV34TCL_TX_FILTER_SIGNED_RED |
193 NV34TCL_TX_FILTER_SIGNED_GREEN |
194 NV34TCL_TX_FILTER_SIGNED_BLUE |
195 NV34TCL_TX_FILTER_MINIFY_LINEAR |
196 NV34TCL_TX_FILTER_MAGNIFY_LINEAR | 0x2000);
198 OUT_RING (chan, NV34TCL_TX_FILTER_MINIFY_LINEAR |
199 NV34TCL_TX_FILTER_MAGNIFY_LINEAR | 0x2000);
200 OUT_RING (chan, (width << NV34TCL_TX_NPOT_SIZE_W_SHIFT) | height);
201 OUT_RING (chan, 0); /* border ARGB */
207 NV30GetSurfaceFormat(PixmapPtr ppix, int *fmt_ret)
209 switch (ppix->drawable.bitsPerPixel) {
211 *fmt_ret = NV34TCL_RT_FORMAT_COLOR_A8R8G8B8;
214 *fmt_ret = NV34TCL_RT_FORMAT_COLOR_X8R8G8B8;
217 *fmt_ret = NV34TCL_RT_FORMAT_COLOR_R5G6B5;
220 *fmt_ret = NV34TCL_RT_FORMAT_COLOR_B8;
230 NV30StopTexturedVideo(ScrnInfoPtr pScrn, pointer data, Bool Exit)
234 #define VERTEX_OUT(sx,sy,dx,dy) do { \
235 BEGIN_RING(chan, rankine, NV34TCL_VTX_ATTR_2F_X(8), 4); \
236 OUT_RINGf (chan, (sx)); OUT_RINGf (chan, (sy)); \
237 OUT_RINGf (chan, (sx)/2.0); OUT_RINGf (chan, (sy)/2.0); \
238 BEGIN_RING(chan, rankine, NV34TCL_VTX_ATTR_2I(0), 1); \
239 OUT_RING (chan, ((dy)<<16)|(dx)); \
243 NV30PutTextureImage(ScrnInfoPtr pScrn, struct nouveau_bo *src, int src_offset,
244 int src_offset2, int id, int src_pitch, BoxPtr dstBox,
245 int x1, int y1, int x2, int y2,
246 uint16_t width, uint16_t height,
247 uint16_t src_w, uint16_t src_h,
248 uint16_t drw_w, uint16_t drw_h,
253 NVPtr pNv = NVPTR(pScrn);
254 struct nouveau_channel *chan = pNv->chan;
255 struct nouveau_grobj *rankine = pNv->Nv3D;
256 Bool redirected = FALSE;
257 float X1, X2, Y1, Y2;
262 if (drw_w > 4096 || drw_h > 4096) {
263 xf86DrvMsg(pScrn->scrnIndex, X_ERROR,
264 "XV: Draw size too large.\n");
268 if (!NV30GetSurfaceFormat(ppix, &dst_format)) {
269 ErrorF("No surface format, bad.\n");
273 if (!NVExaPixmapIsOnscreen(ppix))
277 pbox = REGION_RECTS(clipBoxes);
278 nbox = REGION_NUM_RECTS(clipBoxes);
280 /* Disable blending */
281 BEGIN_RING(chan, rankine, NV34TCL_BLEND_FUNC_ENABLE, 1);
285 BEGIN_RING(chan, rankine, NV34TCL_RT_FORMAT, 3);
286 OUT_RING (chan, NV34TCL_RT_FORMAT_TYPE_LINEAR |
287 NV34TCL_RT_FORMAT_ZETA_Z24S8 |
289 OUT_RING (chan, (exaGetPixmapPitch(ppix) << 16) |
290 exaGetPixmapPitch(ppix));
291 OUT_PIXMAPl(chan, ppix, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
293 if (pNv->NVArch == 0x30) {
296 int w = ppix->drawable.x + ppix->drawable.width;
297 int h = ppix->drawable.y + ppix->drawable.height;
299 BEGIN_RING(chan, rankine, NV34TCL_VIEWPORT_HORIZ, 2);
300 OUT_RING (chan, (w<<16)|x);
301 OUT_RING (chan, (h<<16)|y);
302 BEGIN_RING(chan, rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2);
303 OUT_RING (chan, (w-1+x)<<16);
304 OUT_RING (chan, (h-1+y)<<16);
305 BEGIN_RING(chan, rankine, NV34TCL_VIEWPORT_TX_ORIGIN, 1);
306 OUT_RING (chan, (y<<16)|x);
309 NV30_LoadFilterTable(pScrn);
311 BEGIN_RING(chan, rankine, NV34TCL_TX_UNITS_ENABLE, 1);
312 OUT_RING (chan, NV34TCL_TX_UNITS_ENABLE_TX0 |
313 NV34TCL_TX_UNITS_ENABLE_TX1);
315 NV30VideoTexture(pScrn, table_mem, 0, TABLE_SIZE, 1, 0 , 0);
316 NV30VideoTexture(pScrn, src, src_offset, src_w, src_h, src_pitch, 1);
317 /* We've got NV12 format, which means half width and half height texture of chroma channels. */
318 NV30VideoTexture(pScrn, src, src_offset2, src_w/2, src_h/2, src_pitch, 2);
320 BEGIN_RING(chan, rankine, NV34TCL_TX_ENABLE(3), 1);
321 OUT_RING (chan, 0x0);
324 NV30_LoadFragProg(pScrn, &nv30_fp_yv12_bicubic);
326 NV30_LoadFragProg(pScrn, &nv30_fp_yv12_bilinear);
328 /* Just before rendering we wait for vblank in the non-composited case. */
329 if (pPriv->SyncToVBlank && !redirected) {
330 uint8_t crtcs = nv_window_belongs_to_crtc(pScrn, dstBox->x1, dstBox->y1,
331 dstBox->x2 - dstBox->x1, dstBox->y2 - dstBox->y1);
335 NVWaitVSync(pScrn, 0);
336 else if (crtcs & 0x2)
337 NVWaitVSync(pScrn, 1);
340 /* These are fixed point values in the 16.16 format. */
341 X1 = (float)(x1>>16)+(float)(x1&0xFFFF)/(float)0x10000;
342 Y1 = (float)(y1>>16)+(float)(y1&0xFFFF)/(float)0x10000;
343 X2 = (float)(x2>>16)+(float)(x2&0xFFFF)/(float)0x10000;
344 Y2 = (float)(y2>>16)+(float)(y2&0xFFFF)/(float)0x10000;
346 BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1);
347 OUT_RING (chan, NV34TCL_VERTEX_BEGIN_END_TRIANGLES);
350 float tx1=X1+(float)(pbox->x1 - dstBox->x1)*(X2-X1)/(float)(drw_w);
351 float tx2=X1+(float)(pbox->x2 - dstBox->x1)*(src_w)/(float)(drw_w);
352 float ty1=Y1+(float)(pbox->y1 - dstBox->y1)*(Y2-Y1)/(float)(drw_h);
353 float ty2=Y1+(float)(pbox->y2 - dstBox->y1)*(src_h)/(float)(drw_h);
359 BEGIN_RING(chan, rankine, NV34TCL_SCISSOR_HORIZ, 2);
360 OUT_RING (chan, (sx2 << 16) | 0);
361 OUT_RING (chan, (sy2 << 16) | 0);
363 VERTEX_OUT(tx1, ty1, sx1, sy1);
364 VERTEX_OUT(tx2+(tx2-tx1), ty1, sx2+(sx2-sx1), sy1);
365 VERTEX_OUT(tx1, ty2+(ty2-ty1), sx1, sy2+(sy2-sy1));
370 BEGIN_RING(chan, rankine, NV34TCL_VERTEX_BEGIN_END, 1);
371 OUT_RING (chan, NV34TCL_VERTEX_BEGIN_END_STOP);
379 * NV30SetTexturePortAttribute
380 * sets the attribute "attribute" of port "data" to value "value"
381 * supported attributes:
385 * @param attribute attribute to set
386 * @param value value to which attribute is to be set
387 * @param data port from which the attribute is to be set
389 * @return Success, if setting is successful
390 * BadValue/BadMatch, if value/attribute are invalid
393 NV30SetTexturePortAttribute(ScrnInfoPtr pScrn, Atom attribute,
394 INT32 value, pointer data)
396 NVPortPrivPtr pPriv = (NVPortPrivPtr)data;
397 NVPtr pNv = NVPTR(pScrn);
399 if ((attribute == xvSyncToVBlank) && pNv->WaitVSyncPossible) {
400 if ((value < 0) || (value > 1))
402 pPriv->SyncToVBlank = value;
404 if (attribute == xvSetDefaults) {
405 pPriv->SyncToVBlank = pNv->WaitVSyncPossible;
413 * NV30GetTexturePortAttribute
414 * reads the value of attribute "attribute" from port "data" into INT32 "*value"
417 * @param pScrn unused
418 * @param attribute attribute to be read
419 * @param value value of attribute will be stored here
420 * @param data port from which attribute will be read
421 * @return Success, if queried attribute exists
424 NV30GetTexturePortAttribute(ScrnInfoPtr pScrn, Atom attribute,
425 INT32 *value, pointer data)
427 NVPortPrivPtr pPriv = (NVPortPrivPtr)data;
429 if(attribute == xvSyncToVBlank)
430 *value = (pPriv->SyncToVBlank) ? 1 : 0;