1/*
2 * Copyright (C) 2009 Maciej Cencora <m.cencora@gmail.com>
3 *
4 * All Rights Reserved.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining
7 * a copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sublicense, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
13 *
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial
16 * portions of the Software.
17 *
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
19 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
21 * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
22 * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
23 * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
24 * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
25 *
26 */
27
28#include "radeon_common.h"
29#include "r200_context.h"
30#include "r200_blit.h"
31
32static inline uint32_t cmdpacket0(struct radeon_screen *rscrn,
33                                  int reg, int count)
34{
35    if (count)
36	    return CP_PACKET0(reg, count - 1);
37    return CP_PACKET2;
38}
39
40/* common formats supported as both textures and render targets */
41unsigned r200_check_blit(gl_format mesa_format, uint32_t dst_pitch)
42{
43    /* XXX others?  BE/LE? */
44    switch (mesa_format) {
45    case MESA_FORMAT_ARGB8888:
46    case MESA_FORMAT_XRGB8888:
47    case MESA_FORMAT_RGB565:
48    case MESA_FORMAT_ARGB4444:
49    case MESA_FORMAT_ARGB1555:
50    case MESA_FORMAT_A8:
51    case MESA_FORMAT_L8:
52    case MESA_FORMAT_I8:
53    /* swizzled */
54    case MESA_FORMAT_RGBA8888:
55    case MESA_FORMAT_RGBA8888_REV:
56	    break;
57    default:
58	    return 0;
59    }
60
61    /* Rendering to small buffer doesn't work.
62     * Looks like a hw limitation.
63     */
64    if (dst_pitch < 32)
65            return 0;
66
67    /* ??? */
68    if (_mesa_get_format_bits(mesa_format, GL_DEPTH_BITS) > 0)
69	    return 0;
70
71    return 1;
72}
73
74static inline void emit_vtx_state(struct r200_context *r200)
75{
76    BATCH_LOCALS(&r200->radeon);
77
78    BEGIN_BATCH(14);
79    if (r200->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL) {
80	    OUT_BATCH_REGVAL(R200_SE_VAP_CNTL_STATUS, 0);
81    } else {
82	    OUT_BATCH_REGVAL(R200_SE_VAP_CNTL_STATUS, RADEON_TCL_BYPASS);
83    }
84    OUT_BATCH_REGVAL(R200_SE_VAP_CNTL, (R200_VAP_FORCE_W_TO_ONE |
85					(9 << R200_VAP_VF_MAX_VTX_NUM__SHIFT)));
86    OUT_BATCH_REGVAL(R200_SE_VTX_STATE_CNTL, 0);
87    OUT_BATCH_REGVAL(R200_SE_VTE_CNTL, 0);
88    OUT_BATCH_REGVAL(R200_SE_VTX_FMT_0, R200_VTX_XY);
89    OUT_BATCH_REGVAL(R200_SE_VTX_FMT_1, (2 << R200_VTX_TEX0_COMP_CNT_SHIFT));
90    OUT_BATCH_REGVAL(RADEON_SE_CNTL, (RADEON_DIFFUSE_SHADE_GOURAUD |
91				      RADEON_BFACE_SOLID |
92				      RADEON_FFACE_SOLID |
93				      RADEON_VTX_PIX_CENTER_OGL |
94				      RADEON_ROUND_MODE_ROUND |
95				      RADEON_ROUND_PREC_4TH_PIX));
96    END_BATCH();
97}
98
99static void inline emit_tx_setup(struct r200_context *r200,
100				 gl_format src_mesa_format,
101				 gl_format dst_mesa_format,
102				 struct radeon_bo *bo,
103				 intptr_t offset,
104				 unsigned width,
105				 unsigned height,
106				 unsigned pitch)
107{
108    uint32_t txformat = R200_TXFORMAT_NON_POWER2;
109    BATCH_LOCALS(&r200->radeon);
110
111    assert(width <= 2048);
112    assert(height <= 2048);
113    assert(offset % 32 == 0);
114
115    /* XXX others?  BE/LE? */
116    switch (src_mesa_format) {
117    case MESA_FORMAT_ARGB8888:
118	    txformat |= R200_TXFORMAT_ARGB8888 | R200_TXFORMAT_ALPHA_IN_MAP;
119	    break;
120    case MESA_FORMAT_RGBA8888:
121	    txformat |= R200_TXFORMAT_RGBA8888 | R200_TXFORMAT_ALPHA_IN_MAP;
122	    break;
123    case MESA_FORMAT_RGBA8888_REV:
124	    txformat |= R200_TXFORMAT_ABGR8888 | R200_TXFORMAT_ALPHA_IN_MAP;
125	    break;
126    case MESA_FORMAT_XRGB8888:
127	    txformat |= R200_TXFORMAT_ARGB8888;
128	    break;
129    case MESA_FORMAT_RGB565:
130	    txformat |= R200_TXFORMAT_RGB565;
131	    break;
132    case MESA_FORMAT_ARGB4444:
133	    txformat |= R200_TXFORMAT_ARGB4444 | R200_TXFORMAT_ALPHA_IN_MAP;
134	    break;
135    case MESA_FORMAT_ARGB1555:
136	    txformat |= R200_TXFORMAT_ARGB1555 | R200_TXFORMAT_ALPHA_IN_MAP;
137	    break;
138    case MESA_FORMAT_A8:
139    case MESA_FORMAT_I8:
140	    txformat |= R200_TXFORMAT_I8 | R200_TXFORMAT_ALPHA_IN_MAP;
141	    break;
142    case MESA_FORMAT_L8:
143	    txformat |= R200_TXFORMAT_I8;
144	    break;
145    case MESA_FORMAT_AL88:
146	    txformat |= R200_TXFORMAT_AI88 | R200_TXFORMAT_ALPHA_IN_MAP;
147	    break;
148    default:
149	    break;
150    }
151
152    if (bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
153	offset |= R200_TXO_MACRO_TILE;
154    if (bo->flags & RADEON_BO_FLAGS_MICRO_TILE)
155	offset |= R200_TXO_MICRO_TILE;
156
157    switch (dst_mesa_format) {
158    case MESA_FORMAT_ARGB8888:
159    case MESA_FORMAT_XRGB8888:
160    case MESA_FORMAT_RGB565:
161    case MESA_FORMAT_ARGB4444:
162    case MESA_FORMAT_ARGB1555:
163    case MESA_FORMAT_A8:
164    case MESA_FORMAT_L8:
165    case MESA_FORMAT_I8:
166    default:
167	    /* no swizzle required */
168	    BEGIN_BATCH(10);
169	    OUT_BATCH_REGVAL(RADEON_PP_CNTL, (RADEON_TEX_0_ENABLE |
170					      RADEON_TEX_BLEND_0_ENABLE));
171	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO |
172						  R200_TXC_ARG_B_ZERO |
173						  R200_TXC_ARG_C_R0_COLOR |
174						  R200_TXC_OP_MADD));
175	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, (R200_TXC_CLAMP_0_1 |
176						   R200_TXC_OUTPUT_REG_R0));
177	    OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO |
178						  R200_TXA_ARG_B_ZERO |
179						  R200_TXA_ARG_C_R0_ALPHA |
180						  R200_TXA_OP_MADD));
181	    OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, (R200_TXA_CLAMP_0_1 |
182						   R200_TXA_OUTPUT_REG_R0));
183	    END_BATCH();
184	    break;
185    case MESA_FORMAT_RGBA8888:
186	    BEGIN_BATCH(10);
187	    OUT_BATCH_REGVAL(RADEON_PP_CNTL, (RADEON_TEX_0_ENABLE |
188					      RADEON_TEX_BLEND_0_ENABLE));
189	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO |
190						  R200_TXC_ARG_B_ZERO |
191						  R200_TXC_ARG_C_R0_COLOR |
192						  R200_TXC_OP_MADD));
193	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, (R200_TXC_CLAMP_0_1 |
194						   R200_TXC_OUTPUT_ROTATE_GBA |
195						   R200_TXC_OUTPUT_REG_R0));
196	    OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO |
197						  R200_TXA_ARG_B_ZERO |
198						  R200_TXA_ARG_C_R0_ALPHA |
199						  R200_TXA_OP_MADD));
200	    OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, (R200_TXA_CLAMP_0_1 |
201						   (R200_TXA_REPL_RED << R200_TXA_REPL_ARG_C_SHIFT) |
202						   R200_TXA_OUTPUT_REG_R0));
203	    END_BATCH();
204	    break;
205    case MESA_FORMAT_RGBA8888_REV:
206	    BEGIN_BATCH(34);
207	    OUT_BATCH_REGVAL(RADEON_PP_CNTL, (RADEON_TEX_0_ENABLE |
208					      RADEON_TEX_BLEND_0_ENABLE |
209					      RADEON_TEX_BLEND_1_ENABLE |
210					      RADEON_TEX_BLEND_2_ENABLE |
211					      RADEON_TEX_BLEND_3_ENABLE));
212	    /* r1.r = r0.b */
213	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO |
214						  R200_TXC_ARG_B_ZERO |
215						  R200_TXC_ARG_C_R0_COLOR |
216						  R200_TXC_OP_MADD));
217	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, (R200_TXC_CLAMP_0_1 |
218						   R200_TXC_OUTPUT_MASK_R |
219						   (R200_TXC_REPL_BLUE << R200_TXC_REPL_ARG_C_SHIFT) |
220						   R200_TXC_OUTPUT_REG_R1));
221	    /* r1.a = r0.a */
222	    OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO |
223						  R200_TXA_ARG_B_ZERO |
224						  R200_TXA_ARG_C_R0_ALPHA |
225						  R200_TXA_OP_MADD));
226	    OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, (R200_TXA_CLAMP_0_1 |
227						   R200_TXA_OUTPUT_REG_R1));
228	    /* r1.g = r0.g */
229	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND_1, (R200_TXC_ARG_A_ZERO |
230						  R200_TXC_ARG_B_ZERO |
231						  R200_TXC_ARG_C_R0_COLOR |
232						  R200_TXC_OP_MADD));
233	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_1, (R200_TXC_CLAMP_0_1 |
234						   R200_TXC_OUTPUT_MASK_G |
235						   (R200_TXC_REPL_GREEN << R200_TXC_REPL_ARG_C_SHIFT) |
236						   R200_TXC_OUTPUT_REG_R1));
237	    /* r1.a = r0.a */
238	    OUT_BATCH_REGVAL(R200_PP_TXABLEND_1, (R200_TXA_ARG_A_ZERO |
239						  R200_TXA_ARG_B_ZERO |
240						  R200_TXA_ARG_C_R0_ALPHA |
241						  R200_TXA_OP_MADD));
242	    OUT_BATCH_REGVAL(R200_PP_TXABLEND2_1, (R200_TXA_CLAMP_0_1 |
243						   R200_TXA_OUTPUT_REG_R1));
244	    /* r1.b = r0.r */
245	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND_2, (R200_TXC_ARG_A_ZERO |
246						  R200_TXC_ARG_B_ZERO |
247						  R200_TXC_ARG_C_R0_COLOR |
248						  R200_TXC_OP_MADD));
249	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_2, (R200_TXC_CLAMP_0_1 |
250						   R200_TXC_OUTPUT_MASK_B |
251						   (R200_TXC_REPL_RED << R200_TXC_REPL_ARG_C_SHIFT) |
252						   R200_TXC_OUTPUT_REG_R1));
253	    /* r1.a = r0.a */
254	    OUT_BATCH_REGVAL(R200_PP_TXABLEND_2, (R200_TXA_ARG_A_ZERO |
255						  R200_TXA_ARG_B_ZERO |
256						  R200_TXA_ARG_C_R0_ALPHA |
257						  R200_TXA_OP_MADD));
258	    OUT_BATCH_REGVAL(R200_PP_TXABLEND2_2, (R200_TXA_CLAMP_0_1 |
259						   R200_TXA_OUTPUT_REG_R1));
260	    /* r0.rgb = r1.rgb */
261	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND_3, (R200_TXC_ARG_A_ZERO |
262						  R200_TXC_ARG_B_ZERO |
263						  R200_TXC_ARG_C_R1_COLOR |
264						  R200_TXC_OP_MADD));
265	    OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_3, (R200_TXC_CLAMP_0_1 |
266						   R200_TXC_OUTPUT_REG_R0));
267	    /* r0.a = r1.a */
268	    OUT_BATCH_REGVAL(R200_PP_TXABLEND_3, (R200_TXA_ARG_A_ZERO |
269						  R200_TXA_ARG_B_ZERO |
270						  R200_TXA_ARG_C_R1_ALPHA |
271						  R200_TXA_OP_MADD));
272	    OUT_BATCH_REGVAL(R200_PP_TXABLEND2_3, (R200_TXA_CLAMP_0_1 |
273						   R200_TXA_OUTPUT_REG_R0));
274	    END_BATCH();
275	    break;
276    }
277
278    BEGIN_BATCH(18);
279    OUT_BATCH_REGVAL(R200_PP_CNTL_X, 0);
280    OUT_BATCH_REGVAL(R200_PP_TXMULTI_CTL_0, 0);
281    OUT_BATCH_REGVAL(R200_PP_TXFILTER_0, (R200_CLAMP_S_CLAMP_LAST |
282					  R200_CLAMP_T_CLAMP_LAST |
283					  R200_MAG_FILTER_NEAREST |
284					  R200_MIN_FILTER_NEAREST));
285    OUT_BATCH_REGVAL(R200_PP_TXFORMAT_0, txformat);
286    OUT_BATCH_REGVAL(R200_PP_TXFORMAT_X_0, 0);
287    OUT_BATCH_REGVAL(R200_PP_TXSIZE_0, ((width - 1) |
288					((height - 1) << RADEON_TEX_VSIZE_SHIFT)));
289    OUT_BATCH_REGVAL(R200_PP_TXPITCH_0, pitch * _mesa_get_format_bytes(src_mesa_format) - 32);
290
291    OUT_BATCH_REGSEQ(R200_PP_TXOFFSET_0, 1);
292    OUT_BATCH_RELOC(offset, bo, offset, RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
293
294    END_BATCH();
295}
296
297static inline void emit_cb_setup(struct r200_context *r200,
298				 struct radeon_bo *bo,
299				 intptr_t offset,
300				 gl_format mesa_format,
301				 unsigned pitch,
302				 unsigned width,
303				 unsigned height)
304{
305    uint32_t dst_pitch = pitch;
306    uint32_t dst_format = 0;
307    BATCH_LOCALS(&r200->radeon);
308
309    /* XXX others?  BE/LE? */
310    switch (mesa_format) {
311    case MESA_FORMAT_ARGB8888:
312    case MESA_FORMAT_XRGB8888:
313    case MESA_FORMAT_RGBA8888:
314    case MESA_FORMAT_RGBA8888_REV:
315	    dst_format = RADEON_COLOR_FORMAT_ARGB8888;
316	    break;
317    case MESA_FORMAT_RGB565:
318	    dst_format = RADEON_COLOR_FORMAT_RGB565;
319	    break;
320    case MESA_FORMAT_ARGB4444:
321	    dst_format = RADEON_COLOR_FORMAT_ARGB4444;
322	    break;
323    case MESA_FORMAT_ARGB1555:
324	    dst_format = RADEON_COLOR_FORMAT_ARGB1555;
325	    break;
326    case MESA_FORMAT_A8:
327    case MESA_FORMAT_L8:
328    case MESA_FORMAT_I8:
329	    dst_format = RADEON_COLOR_FORMAT_RGB8;
330	    break;
331    default:
332	    break;
333    }
334
335    if (bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
336	dst_pitch |= R200_COLOR_TILE_ENABLE;
337    if (bo->flags & RADEON_BO_FLAGS_MICRO_TILE)
338	dst_pitch |= R200_COLOR_MICROTILE_ENABLE;
339
340    BEGIN_BATCH_NO_AUTOSTATE(22);
341    OUT_BATCH_REGVAL(R200_RE_AUX_SCISSOR_CNTL, 0);
342    OUT_BATCH_REGVAL(R200_RE_CNTL, 0);
343    OUT_BATCH_REGVAL(RADEON_RE_TOP_LEFT, 0);
344    OUT_BATCH_REGVAL(RADEON_RE_WIDTH_HEIGHT, (((width - 1) << RADEON_RE_WIDTH_SHIFT) |
345					      ((height - 1) << RADEON_RE_HEIGHT_SHIFT)));
346    OUT_BATCH_REGVAL(RADEON_RB3D_PLANEMASK, 0xffffffff);
347    OUT_BATCH_REGVAL(RADEON_RB3D_BLENDCNTL, RADEON_SRC_BLEND_GL_ONE | RADEON_DST_BLEND_GL_ZERO);
348    OUT_BATCH_REGVAL(RADEON_RB3D_CNTL, dst_format);
349
350    OUT_BATCH_REGSEQ(RADEON_RB3D_COLOROFFSET, 1);
351    OUT_BATCH_RELOC(offset, bo, offset, 0, RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0);
352    OUT_BATCH_REGSEQ(RADEON_RB3D_COLORPITCH, 1);
353    OUT_BATCH_RELOC(dst_pitch, bo, dst_pitch, 0, RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0);
354
355    END_BATCH();
356}
357
358static GLboolean validate_buffers(struct r200_context *r200,
359                                  struct radeon_bo *src_bo,
360                                  struct radeon_bo *dst_bo)
361{
362    int ret;
363
364    radeon_cs_space_reset_bos(r200->radeon.cmdbuf.cs);
365
366    ret = radeon_cs_space_check_with_bo(r200->radeon.cmdbuf.cs,
367                                        src_bo, RADEON_GEM_DOMAIN_VRAM | RADEON_GEM_DOMAIN_GTT, 0);
368    if (ret)
369        return GL_FALSE;
370
371    ret = radeon_cs_space_check_with_bo(r200->radeon.cmdbuf.cs,
372                                        dst_bo, 0, RADEON_GEM_DOMAIN_VRAM | RADEON_GEM_DOMAIN_GTT);
373    if (ret)
374        return GL_FALSE;
375
376    return GL_TRUE;
377}
378
379/**
380 * Calculate texcoords for given image region.
381 * Output values are [minx, maxx, miny, maxy]
382 */
383static inline void calc_tex_coords(float img_width, float img_height,
384				   float x, float y,
385				   float reg_width, float reg_height,
386				   unsigned flip_y, float *buf)
387{
388    buf[0] = x / img_width;
389    buf[1] = buf[0] + reg_width / img_width;
390    buf[2] = y / img_height;
391    buf[3] = buf[2] + reg_height / img_height;
392    if (flip_y)
393    {
394        buf[2] = 1.0 - buf[2];
395        buf[3] = 1.0 - buf[3];
396    }
397}
398
399static inline void emit_draw_packet(struct r200_context *r200,
400				    unsigned src_width, unsigned src_height,
401				    unsigned src_x_offset, unsigned src_y_offset,
402				    unsigned dst_x_offset, unsigned dst_y_offset,
403				    unsigned reg_width, unsigned reg_height,
404				    unsigned flip_y)
405{
406    float texcoords[4];
407    float verts[12];
408    BATCH_LOCALS(&r200->radeon);
409
410    calc_tex_coords(src_width, src_height,
411                    src_x_offset, src_y_offset,
412                    reg_width, reg_height,
413                    flip_y, texcoords);
414
415    verts[0] = dst_x_offset;
416    verts[1] = dst_y_offset + reg_height;
417    verts[2] = texcoords[0];
418    verts[3] = texcoords[3];
419
420    verts[4] = dst_x_offset + reg_width;
421    verts[5] = dst_y_offset + reg_height;
422    verts[6] = texcoords[1];
423    verts[7] = texcoords[3];
424
425    verts[8] = dst_x_offset + reg_width;
426    verts[9] = dst_y_offset;
427    verts[10] = texcoords[1];
428    verts[11] = texcoords[2];
429
430    BEGIN_BATCH(14);
431    OUT_BATCH(R200_CP_CMD_3D_DRAW_IMMD_2 | (12 << 16));
432    OUT_BATCH(RADEON_CP_VC_CNTL_PRIM_WALK_RING |
433	      RADEON_CP_VC_CNTL_PRIM_TYPE_RECT_LIST |
434              (3 << 16));
435    OUT_BATCH_TABLE(verts, 12);
436    END_BATCH();
437}
438
439/**
440 * Copy a region of [@a width x @a height] pixels from source buffer
441 * to destination buffer.
442 * @param[in] r200 r200 context
443 * @param[in] src_bo source radeon buffer object
444 * @param[in] src_offset offset of the source image in the @a src_bo
445 * @param[in] src_mesaformat source image format
446 * @param[in] src_pitch aligned source image width
447 * @param[in] src_width source image width
448 * @param[in] src_height source image height
449 * @param[in] src_x_offset x offset in the source image
450 * @param[in] src_y_offset y offset in the source image
451 * @param[in] dst_bo destination radeon buffer object
452 * @param[in] dst_offset offset of the destination image in the @a dst_bo
453 * @param[in] dst_mesaformat destination image format
454 * @param[in] dst_pitch aligned destination image width
455 * @param[in] dst_width destination image width
456 * @param[in] dst_height destination image height
457 * @param[in] dst_x_offset x offset in the destination image
458 * @param[in] dst_y_offset y offset in the destination image
459 * @param[in] width region width
460 * @param[in] height region height
461 * @param[in] flip_y set if y coords of the source image need to be flipped
462 */
463unsigned r200_blit(struct gl_context *ctx,
464                   struct radeon_bo *src_bo,
465                   intptr_t src_offset,
466                   gl_format src_mesaformat,
467                   unsigned src_pitch,
468                   unsigned src_width,
469                   unsigned src_height,
470                   unsigned src_x_offset,
471                   unsigned src_y_offset,
472                   struct radeon_bo *dst_bo,
473                   intptr_t dst_offset,
474                   gl_format dst_mesaformat,
475                   unsigned dst_pitch,
476                   unsigned dst_width,
477                   unsigned dst_height,
478                   unsigned dst_x_offset,
479                   unsigned dst_y_offset,
480                   unsigned reg_width,
481                   unsigned reg_height,
482                   unsigned flip_y)
483{
484    struct r200_context *r200 = R200_CONTEXT(ctx);
485
486    if (!r200_check_blit(dst_mesaformat, dst_pitch))
487        return GL_FALSE;
488
489    /* Make sure that colorbuffer has even width - hw limitation */
490    if (dst_pitch % 2 > 0)
491        ++dst_pitch;
492
493    /* Need to clamp the region size to make sure
494     * we don't read outside of the source buffer
495     * or write outside of the destination buffer.
496     */
497    if (reg_width + src_x_offset > src_width)
498        reg_width = src_width - src_x_offset;
499    if (reg_height + src_y_offset > src_height)
500        reg_height = src_height - src_y_offset;
501    if (reg_width + dst_x_offset > dst_width)
502        reg_width = dst_width - dst_x_offset;
503    if (reg_height + dst_y_offset > dst_height)
504        reg_height = dst_height - dst_y_offset;
505
506    if (src_bo == dst_bo) {
507        return GL_FALSE;
508    }
509
510    if (src_offset % 32 || dst_offset % 32) {
511        return GL_FALSE;
512    }
513
514    if (0) {
515        fprintf(stderr, "src: size [%d x %d], pitch %d, "
516                "offset [%d x %d], format %s, bo %p\n",
517                src_width, src_height, src_pitch,
518                src_x_offset, src_y_offset,
519                _mesa_get_format_name(src_mesaformat),
520                src_bo);
521        fprintf(stderr, "dst: pitch %d, offset[%d x %d], format %s, bo %p\n",
522                dst_pitch, dst_x_offset, dst_y_offset,
523                _mesa_get_format_name(dst_mesaformat), dst_bo);
524        fprintf(stderr, "region: %d x %d\n", reg_width, reg_height);
525    }
526
527    /* Flush is needed to make sure that source buffer has correct data */
528    radeonFlush(r200->radeon.glCtx);
529
530    rcommonEnsureCmdBufSpace(&r200->radeon, 102, __FUNCTION__);
531
532    if (!validate_buffers(r200, src_bo, dst_bo))
533        return GL_FALSE;
534
535    /* 14 */
536    emit_vtx_state(r200);
537    /* 52 */
538    emit_tx_setup(r200, src_mesaformat, dst_mesaformat, src_bo, src_offset, src_width, src_height, src_pitch);
539    /* 22 */
540    emit_cb_setup(r200, dst_bo, dst_offset, dst_mesaformat, dst_pitch, dst_width, dst_height);
541    /* 14 */
542    emit_draw_packet(r200, src_width, src_height,
543                     src_x_offset, src_y_offset,
544                     dst_x_offset, dst_y_offset,
545                     reg_width, reg_height,
546                     flip_y);
547
548    radeonFlush(ctx);
549
550    return GL_TRUE;
551}
552