diff options
Diffstat (limited to 'src/mesa/drivers')
79 files changed, 3399 insertions, 1102 deletions
diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c index a9f3c8e727..f09106b77c 100644 --- a/src/mesa/drivers/common/driverfuncs.c +++ b/src/mesa/drivers/common/driverfuncs.c @@ -56,6 +56,7 @@ #include "swrast/swrast.h" #include "driverfuncs.h" +#include "meta.h" @@ -100,11 +101,11 @@ _mesa_init_driver_functions(struct dd_function_table *driver) driver->TexSubImage2D = _mesa_store_texsubimage2d; driver->TexSubImage3D = _mesa_store_texsubimage3d; driver->GetTexImage = _mesa_get_teximage; - driver->CopyTexImage1D = _swrast_copy_teximage1d; - driver->CopyTexImage2D = _swrast_copy_teximage2d; - driver->CopyTexSubImage1D = _swrast_copy_texsubimage1d; - driver->CopyTexSubImage2D = _swrast_copy_texsubimage2d; - driver->CopyTexSubImage3D = _swrast_copy_texsubimage3d; + driver->CopyTexImage1D = _mesa_meta_CopyTexImage1D; + driver->CopyTexImage2D = _mesa_meta_CopyTexImage2D; + driver->CopyTexSubImage1D = _mesa_meta_CopyTexSubImage1D; + driver->CopyTexSubImage2D = _mesa_meta_CopyTexSubImage2D; + driver->CopyTexSubImage3D = _mesa_meta_CopyTexSubImage3D; driver->GenerateMipmap = _mesa_generate_mipmap; driver->TestProxyTexImage = _mesa_test_proxy_teximage; driver->CompressedTexImage1D = _mesa_store_compressed_teximage1d; @@ -129,10 +130,10 @@ _mesa_init_driver_functions(struct dd_function_table *driver) driver->UpdateTexturePalette = NULL; /* imaging */ - driver->CopyColorTable = _swrast_CopyColorTable; - driver->CopyColorSubTable = _swrast_CopyColorSubTable; - driver->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; - driver->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; + driver->CopyColorTable = _mesa_meta_CopyColorTable; + driver->CopyColorSubTable = _mesa_meta_CopyColorSubTable; + driver->CopyConvolutionFilter1D = _mesa_meta_CopyConvolutionFilter1D; + driver->CopyConvolutionFilter2D = _mesa_meta_CopyConvolutionFilter2D; /* Vertex/fragment programs */ driver->BindProgram = NULL; diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c index 0b9781027e..b445323aab 100644 --- a/src/mesa/drivers/common/meta.c +++ b/src/mesa/drivers/common/meta.c @@ -38,6 +38,8 @@ #include "main/blend.h" #include "main/bufferobj.h" #include "main/buffers.h" +#include "main/colortab.h" +#include "main/convolve.h" #include "main/depth.h" #include "main/enable.h" #include "main/fbobject.h" @@ -49,6 +51,7 @@ #include "main/readpix.h" #include "main/scissor.h" #include "main/shaders.h" +#include "main/state.h" #include "main/stencil.h" #include "main/texobj.h" #include "main/texenv.h" @@ -63,6 +66,33 @@ #include "drivers/common/meta.h" +/** Return offset in bytes of the field within a vertex struct */ +#define OFFSET(FIELD) ((void *) offsetof(struct vertex, FIELD)) + + +/** + * Flags passed to _mesa_meta_begin(). + */ +/*@{*/ +#define META_ALL ~0x0 +#define META_ALPHA_TEST 0x1 +#define META_BLEND 0x2 /**< includes logicop */ +#define META_COLOR_MASK 0x4 +#define META_DEPTH_TEST 0x8 +#define META_FOG 0x10 +#define META_PIXEL_STORE 0x20 +#define META_PIXEL_TRANSFER 0x40 +#define META_RASTERIZATION 0x80 +#define META_SCISSOR 0x100 +#define META_SHADER 0x200 +#define META_STENCIL_TEST 0x400 +#define META_TRANSFORM 0x800 /**< modelview, projection, clip planes */ +#define META_TEXTURE 0x1000 +#define META_VERTEX 0x2000 +#define META_VIEWPORT 0x4000 +/*@}*/ + + /** * State which we may save/restore across meta ops. * XXX this may be incomplete... @@ -90,6 +120,17 @@ struct save_state /** META_PIXEL_STORE */ struct gl_pixelstore_attrib Pack, Unpack; + /** META_PIXEL_TRANSFER */ + GLfloat RedBias, RedScale; + GLfloat GreenBias, GreenScale; + GLfloat BlueBias, BlueScale; + GLfloat AlphaBias, AlphaScale; + GLfloat DepthBias, DepthScale; + GLboolean MapColorFlag; + GLboolean Convolution1DEnabled; + GLboolean Convolution2DEnabled; + GLboolean Separable2DEnabled; + /** META_RASTERIZATION */ GLenum FrontPolygonMode, BackPolygonMode; GLboolean PolygonOffset; @@ -195,7 +236,6 @@ struct copypix_state struct drawpix_state { GLuint ArrayObj; - GLuint VBO; GLuint StencilFP; /**< Fragment program for drawing stencil images */ GLuint DepthFP; /**< Fragment program for drawing depth images */ @@ -233,12 +273,12 @@ struct gl_meta_state struct temp_texture TempTex; - struct blit_state Blit; /**< For _mesa_meta_blit_framebuffer() */ - struct clear_state Clear; /**< For _mesa_meta_clear() */ - struct copypix_state CopyPix; /**< For _mesa_meta_copy_pixels() */ - struct drawpix_state DrawPix; /**< For _mesa_meta_draw_pixels() */ - struct bitmap_state Bitmap; /**< For _mesa_meta_bitmap() */ - struct gen_mipmap_state Mipmap; /**< For _mesa_meta_generate_mipmap() */ + struct blit_state Blit; /**< For _mesa_meta_BlitFramebuffer() */ + struct clear_state Clear; /**< For _mesa_meta_Clear() */ + struct copypix_state CopyPix; /**< For _mesa_meta_CopyPixels() */ + struct drawpix_state DrawPix; /**< For _mesa_meta_DrawPixels() */ + struct bitmap_state Bitmap; /**< For _mesa_meta_Bitmap() */ + struct gen_mipmap_state Mipmap; /**< For _mesa_meta_GenerateMipmap() */ }; @@ -286,7 +326,6 @@ _mesa_meta_free(GLcontext *ctx) _mesa_DeleteVertexArraysAPPLE(1, &meta->CopyPix.ArrayObj); /* glDrawPixels */ - _mesa_DeleteBuffersARB(1, & meta->DrawPix.VBO); _mesa_DeleteVertexArraysAPPLE(1, &meta->DrawPix.ArrayObj); _mesa_DeletePrograms(1, &meta->DrawPix.DepthFP); _mesa_DeletePrograms(1, &meta->DrawPix.StencilFP); @@ -359,6 +398,35 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state) ctx->Unpack = ctx->DefaultPacking; } + if (state & META_PIXEL_TRANSFER) { + save->RedScale = ctx->Pixel.RedScale; + save->RedBias = ctx->Pixel.RedBias; + save->GreenScale = ctx->Pixel.GreenScale; + save->GreenBias = ctx->Pixel.GreenBias; + save->BlueScale = ctx->Pixel.BlueScale; + save->BlueBias = ctx->Pixel.BlueBias; + save->AlphaScale = ctx->Pixel.AlphaScale; + save->AlphaBias = ctx->Pixel.AlphaBias; + save->MapColorFlag = ctx->Pixel.MapColorFlag; + save->Convolution1DEnabled = ctx->Pixel.Convolution1DEnabled; + save->Convolution2DEnabled = ctx->Pixel.Convolution2DEnabled; + save->Separable2DEnabled = ctx->Pixel.Separable2DEnabled; + ctx->Pixel.RedScale = 1.0F; + ctx->Pixel.RedBias = 0.0F; + ctx->Pixel.GreenScale = 1.0F; + ctx->Pixel.GreenBias = 0.0F; + ctx->Pixel.BlueScale = 1.0F; + ctx->Pixel.BlueBias = 0.0F; + ctx->Pixel.AlphaScale = 1.0F; + ctx->Pixel.AlphaBias = 0.0F; + ctx->Pixel.MapColorFlag = GL_FALSE; + ctx->Pixel.Convolution1DEnabled = GL_FALSE; + ctx->Pixel.Convolution2DEnabled = GL_FALSE; + ctx->Pixel.Separable2DEnabled = GL_FALSE; + /* XXX more state */ + ctx->NewState |=_NEW_PIXEL; + } + if (state & META_RASTERIZATION) { save->FrontPolygonMode = ctx->Polygon.FrontMode; save->BackPolygonMode = ctx->Polygon.BackMode; @@ -411,11 +479,6 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state) save->ClientActiveUnit = ctx->Array.ActiveTexture; save->EnvMode = ctx->Texture.Unit[0].EnvMode; - if (ctx->Texture._EnabledUnits | - ctx->Texture._EnabledCoordUnits | - ctx->Texture._TexGenEnabled | - ctx->Texture._TexMatEnabled) { - /* Disable all texture units */ for (u = 0; u < ctx->Const.MaxTextureUnits; u++) { save->TexEnabled[u] = ctx->Texture.Unit[u].Enabled; @@ -434,7 +497,6 @@ _mesa_meta_begin(GLcontext *ctx, GLbitfield state) _mesa_set_enable(ctx, GL_TEXTURE_GEN_Q, GL_FALSE); } } - } /* save current texture objects for unit[0] only */ for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) { @@ -560,6 +622,23 @@ _mesa_meta_end(GLcontext *ctx) ctx->Unpack = save->Unpack; } + if (state & META_PIXEL_TRANSFER) { + ctx->Pixel.RedScale = save->RedScale; + ctx->Pixel.RedBias = save->RedBias; + ctx->Pixel.GreenScale = save->GreenScale; + ctx->Pixel.GreenBias = save->GreenBias; + ctx->Pixel.BlueScale = save->BlueScale; + ctx->Pixel.BlueBias = save->BlueBias; + ctx->Pixel.AlphaScale = save->AlphaScale; + ctx->Pixel.AlphaBias = save->AlphaBias; + ctx->Pixel.MapColorFlag = save->MapColorFlag; + ctx->Pixel.Convolution1DEnabled = save->Convolution1DEnabled; + ctx->Pixel.Convolution2DEnabled = save->Convolution2DEnabled; + ctx->Pixel.Separable2DEnabled = save->Separable2DEnabled; + /* XXX more state */ + ctx->NewState |=_NEW_PIXEL; + } + if (state & META_RASTERIZATION) { _mesa_PolygonMode(GL_FRONT, save->FrontPolygonMode); _mesa_PolygonMode(GL_BACK, save->BackPolygonMode); @@ -691,7 +770,6 @@ _mesa_meta_end(GLcontext *ctx) _mesa_MatrixMode(save->MatrixMode); - save->ClipPlanesEnabled = ctx->Transform.ClipPlanesEnabled; if (save->ClipPlanesEnabled) { GLuint i; for (i = 0; i < ctx->Const.MaxClipPlanes; i++) { @@ -727,9 +805,6 @@ _mesa_meta_end(GLcontext *ctx) if (save->Lighting) { _mesa_set_enable(ctx, GL_LIGHTING, GL_TRUE); } - if (save->Fog) { - _mesa_set_enable(ctx, GL_FOG, GL_TRUE); - } } @@ -974,10 +1049,10 @@ init_blit_depth_pixels(GLcontext *ctx) * of texture mapping and polygon rendering. */ void -_mesa_meta_blit_framebuffer(GLcontext *ctx, - GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, - GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLbitfield mask, GLenum filter) +_mesa_meta_BlitFramebuffer(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter) { struct blit_state *blit = &ctx->Meta->Blit; struct temp_texture *tex = get_temp_texture(ctx); @@ -988,7 +1063,10 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, const GLint srcH = abs(srcY1 - srcY0); const GLboolean srcFlipX = srcX1 < srcX0; const GLboolean srcFlipY = srcY1 < srcY0; - GLfloat verts[4][4]; /* four verts of X,Y,S,T */ + struct vertex { + GLfloat x, y, s, t; + }; + struct vertex verts[4]; GLboolean newTex; if (srcW > maxTexSize || srcH > maxTexSize) { @@ -1027,10 +1105,8 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (2 * sizeof(GLfloat))); + _mesa_VertexPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1043,23 +1119,23 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, /* vertex positions/texcoords (after texture allocation!) */ { - verts[0][0] = (GLfloat) dstX0; - verts[0][1] = (GLfloat) dstY0; - verts[1][0] = (GLfloat) dstX1; - verts[1][1] = (GLfloat) dstY0; - verts[2][0] = (GLfloat) dstX1; - verts[2][1] = (GLfloat) dstY1; - verts[3][0] = (GLfloat) dstX0; - verts[3][1] = (GLfloat) dstY1; - - verts[0][2] = 0.0F; - verts[0][3] = 0.0F; - verts[1][2] = tex->Sright; - verts[1][3] = 0.0F; - verts[2][2] = tex->Sright; - verts[2][3] = tex->Ttop; - verts[3][2] = 0.0F; - verts[3][3] = tex->Ttop; + verts[0].x = (GLfloat) dstX0; + verts[0].y = (GLfloat) dstY0; + verts[1].x = (GLfloat) dstX1; + verts[1].y = (GLfloat) dstY0; + verts[2].x = (GLfloat) dstX1; + verts[2].y = (GLfloat) dstY1; + verts[3].x = (GLfloat) dstX0; + verts[3].y = (GLfloat) dstY1; + + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); @@ -1122,10 +1198,13 @@ _mesa_meta_blit_framebuffer(GLcontext *ctx, * Meta implementation of ctx->Driver.Clear() in terms of polygon rendering. */ void -_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) +_mesa_meta_Clear(GLcontext *ctx, GLbitfield buffers) { struct clear_state *clear = &ctx->Meta->Clear; - GLfloat verts[4][7]; /* four verts of X,Y,Z,R,G,B,A */ + struct vertex { + GLfloat x, y, z, r, g, b, a; + }; + struct vertex verts[4]; /* save all state but scissor, pixel pack/unpack */ GLbitfield metaSave = META_ALL - META_SCISSOR - META_PIXEL_STORE; @@ -1150,10 +1229,8 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_ColorPointer(4, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_ColorPointer(4, GL_FLOAT, sizeof(struct vertex), OFFSET(r)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_COLOR_ARRAY); } @@ -1203,22 +1280,25 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) const GLfloat z = 1.0 - 2.0 * ctx->Depth.Clear; GLuint i; - verts[0][0] = x0; - verts[0][1] = y0; - verts[0][2] = z; - verts[1][0] = x1; - verts[1][1] = y0; - verts[1][2] = z; - verts[2][0] = x1; - verts[2][1] = y1; - verts[2][2] = z; - verts[3][0] = x0; - verts[3][1] = y1; - verts[3][2] = z; + verts[0].x = x0; + verts[0].y = y0; + verts[0].z = z; + verts[1].x = x1; + verts[1].y = y0; + verts[1].z = z; + verts[2].x = x1; + verts[2].y = y1; + verts[2].z = z; + verts[3].x = x0; + verts[3].y = y1; + verts[3].z = z; /* vertex colors */ for (i = 0; i < 4; i++) { - COPY_4FV(&verts[i][3], ctx->Color.ClearColor); + verts[i].r = ctx->Color.ClearColor[0]; + verts[i].g = ctx->Color.ClearColor[1]; + verts[i].b = ctx->Color.ClearColor[2]; + verts[i].a = ctx->Color.ClearColor[3]; } /* upload new vertex data */ @@ -1237,13 +1317,16 @@ _mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) * of texture mapping and polygon rendering. */ void -_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, - GLsizei width, GLsizei height, - GLint dstX, GLint dstY, GLenum type) +_mesa_meta_CopyPixels(GLcontext *ctx, GLint srcX, GLint srcY, + GLsizei width, GLsizei height, + GLint dstX, GLint dstY, GLenum type) { struct copypix_state *copypix = &ctx->Meta->CopyPix; struct temp_texture *tex = get_temp_texture(ctx); - GLfloat verts[4][5]; /* four verts of X,Y,Z,S,T */ + struct vertex { + GLfloat x, y, z, s, t; + }; + struct vertex verts[4]; GLboolean newTex; GLenum intFormat = GL_RGBA; @@ -1281,10 +1364,8 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1303,26 +1384,26 @@ _mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, const GLfloat dstY1 = dstY + height * ctx->Pixel.ZoomY; const GLfloat z = ctx->Current.RasterPos[2]; - verts[0][0] = dstX0; - verts[0][1] = dstY0; - verts[0][2] = z; - verts[0][3] = 0.0F; - verts[0][4] = 0.0F; - verts[1][0] = dstX1; - verts[1][1] = dstY0; - verts[1][2] = z; - verts[1][3] = tex->Sright; - verts[1][4] = 0.0F; - verts[2][0] = dstX1; - verts[2][1] = dstY1; - verts[2][2] = z; - verts[2][3] = tex->Sright; - verts[2][4] = tex->Ttop; - verts[3][0] = dstX0; - verts[3][1] = dstY1; - verts[3][2] = z; - verts[3][3] = 0.0F; - verts[3][4] = tex->Ttop; + verts[0].x = dstX0; + verts[0].y = dstY0; + verts[0].z = z; + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].x = dstX1; + verts[1].y = dstY0; + verts[1].z = z; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].x = dstX1; + verts[2].y = dstY1; + verts[2].z = z; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].x = dstX0; + verts[3].y = dstY1; + verts[3].z = z; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); @@ -1375,9 +1456,8 @@ tiled_draw_pixels(GLcontext *ctx, tileUnpack.SkipRows = unpack->SkipRows + j; - _mesa_meta_draw_pixels(ctx, tileX, tileY, - tileWidth, tileHeight, - format, type, &tileUnpack, pixels); + _mesa_meta_DrawPixels(ctx, tileX, tileY, tileWidth, tileHeight, + format, type, &tileUnpack, pixels); } } } @@ -1486,20 +1566,24 @@ init_draw_depth_pixels(GLcontext *ctx) * of texture mapping and polygon rendering. */ void -_mesa_meta_draw_pixels(GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels) +_mesa_meta_DrawPixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels) { struct drawpix_state *drawpix = &ctx->Meta->DrawPix; struct temp_texture *tex = get_temp_texture(ctx); const struct gl_pixelstore_attrib unpackSave = ctx->Unpack; const GLuint origStencilMask = ctx->Stencil.WriteMask[0]; - GLfloat verts[4][5]; /* four verts of X,Y,Z,S,T */ + struct vertex { + GLfloat x, y, z, s, t; + }; + struct vertex verts[4]; GLenum texIntFormat; GLboolean fallback, newTex; GLbitfield metaExtraSave = 0x0; + GLuint vbo; /* * Determine if we can do the glDrawPixels with texture mapping. @@ -1579,32 +1663,6 @@ _mesa_meta_draw_pixels(GLcontext *ctx, META_VIEWPORT | metaExtraSave)); - if (drawpix->ArrayObj == 0) { - /* one-time setup */ - - /* create vertex array object */ - _mesa_GenVertexArrays(1, &drawpix->ArrayObj); - _mesa_BindVertexArray(drawpix->ArrayObj); - - /* create vertex array buffer */ - _mesa_GenBuffersARB(1, &drawpix->VBO); - _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); - _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), - NULL, GL_DYNAMIC_DRAW_ARB); - - /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); - _mesa_EnableClientState(GL_VERTEX_ARRAY); - _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); - } - else { - _mesa_BindVertexArray(drawpix->ArrayObj); - _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); - } - newTex = alloc_texture(tex, width, height, texIntFormat); /* vertex positions, texcoords (after texture allocation!) */ @@ -1615,30 +1673,45 @@ _mesa_meta_draw_pixels(GLcontext *ctx, const GLfloat y1 = y + height * ctx->Pixel.ZoomY; const GLfloat z = ctx->Current.RasterPos[2]; - verts[0][0] = x0; - verts[0][1] = y0; - verts[0][2] = z; - verts[0][3] = 0.0F; - verts[0][4] = 0.0F; - verts[1][0] = x1; - verts[1][1] = y0; - verts[1][2] = z; - verts[1][3] = tex->Sright; - verts[1][4] = 0.0F; - verts[2][0] = x1; - verts[2][1] = y1; - verts[2][2] = z; - verts[2][3] = tex->Sright; - verts[2][4] = tex->Ttop; - verts[3][0] = x0; - verts[3][1] = y1; - verts[3][2] = z; - verts[3][3] = 0.0F; - verts[3][4] = tex->Ttop; + verts[0].x = x0; + verts[0].y = y0; + verts[0].z = z; + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].x = x1; + verts[1].y = y0; + verts[1].z = z; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].x = x1; + verts[2].y = y1; + verts[2].z = z; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].x = x0; + verts[3].y = y1; + verts[3].z = z; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; + } - /* upload new vertex data */ - _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); + if (drawpix->ArrayObj == 0) { + /* one-time setup: create vertex array object */ + _mesa_GenVertexArrays(1, &drawpix->ArrayObj); } + _mesa_BindVertexArray(drawpix->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, &vbo); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, vbo); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), + verts, GL_DYNAMIC_DRAW_ARB); + + /* setup vertex arrays */ + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); + _mesa_EnableClientState(GL_VERTEX_ARRAY); + _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); /* set given unpack params */ ctx->Unpack = *unpack; @@ -1709,6 +1782,8 @@ _mesa_meta_draw_pixels(GLcontext *ctx, _mesa_set_enable(ctx, tex->Target, GL_FALSE); + _mesa_DeleteBuffersARB(1, &vbo); + /* restore unpack params */ ctx->Unpack = unpackSave; @@ -1724,7 +1799,7 @@ _mesa_meta_draw_pixels(GLcontext *ctx, * improve performance a lot. */ void -_mesa_meta_bitmap(GLcontext *ctx, +_mesa_meta_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap1) @@ -1733,7 +1808,10 @@ _mesa_meta_bitmap(GLcontext *ctx, struct temp_texture *tex = get_bitmap_temp_texture(ctx); const GLenum texIntFormat = GL_ALPHA; const struct gl_pixelstore_attrib unpackSave = *unpack; - GLfloat verts[4][9]; /* four verts of X,Y,Z,S,T,R,G,B,A */ + struct vertex { + GLfloat x, y, z, s, t, r, g, b, a; + }; + struct vertex verts[4]; GLboolean newTex; GLubyte *bitmap8; @@ -1776,13 +1854,9 @@ _mesa_meta_bitmap(GLcontext *ctx, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (3 * sizeof(GLfloat))); - _mesa_ColorPointer(4, GL_FLOAT, sizeof(verts[0]), - (void *) (5 * sizeof(GLfloat))); - + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); + _mesa_ColorPointer(4, GL_FLOAT, sizeof(struct vertex), OFFSET(r)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); _mesa_EnableClientState(GL_COLOR_ARRAY); @@ -1803,32 +1877,32 @@ _mesa_meta_bitmap(GLcontext *ctx, const GLfloat z = ctx->Current.RasterPos[2]; GLuint i; - verts[0][0] = x0; - verts[0][1] = y0; - verts[0][2] = z; - verts[0][3] = 0.0F; - verts[0][4] = 0.0F; - verts[1][0] = x1; - verts[1][1] = y0; - verts[1][2] = z; - verts[1][3] = tex->Sright; - verts[1][4] = 0.0F; - verts[2][0] = x1; - verts[2][1] = y1; - verts[2][2] = z; - verts[2][3] = tex->Sright; - verts[2][4] = tex->Ttop; - verts[3][0] = x0; - verts[3][1] = y1; - verts[3][2] = z; - verts[3][3] = 0.0F; - verts[3][4] = tex->Ttop; + verts[0].x = x0; + verts[0].y = y0; + verts[0].z = z; + verts[0].s = 0.0F; + verts[0].t = 0.0F; + verts[1].x = x1; + verts[1].y = y0; + verts[1].z = z; + verts[1].s = tex->Sright; + verts[1].t = 0.0F; + verts[2].x = x1; + verts[2].y = y1; + verts[2].z = z; + verts[2].s = tex->Sright; + verts[2].t = tex->Ttop; + verts[3].x = x0; + verts[3].y = y1; + verts[3].z = z; + verts[3].s = 0.0F; + verts[3].t = tex->Ttop; for (i = 0; i < 4; i++) { - verts[i][5] = ctx->Current.RasterColor[0]; - verts[i][6] = ctx->Current.RasterColor[1]; - verts[i][7] = ctx->Current.RasterColor[2]; - verts[i][8] = ctx->Current.RasterColor[3]; + verts[i].r = ctx->Current.RasterColor[0]; + verts[i].g = ctx->Current.RasterColor[1]; + verts[i].b = ctx->Current.RasterColor[2]; + verts[i].a = ctx->Current.RasterColor[3]; } /* upload new vertex data */ @@ -1865,23 +1939,37 @@ _mesa_meta_bitmap(GLcontext *ctx, } +/** + * Called via ctx->Driver.GenerateMipmap() + * Note: texture borders and 3D texture support not yet complete. + */ void -_mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, - struct gl_texture_object *texObj) +_mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) { struct gen_mipmap_state *mipmap = &ctx->Meta->Mipmap; - struct { GLfloat x, y, s, t, r; } verts[4]; + struct vertex { + GLfloat x, y, s, t, r; + }; + struct vertex verts[4]; const GLuint baseLevel = texObj->BaseLevel; const GLuint maxLevel = texObj->MaxLevel; const GLenum minFilterSave = texObj->MinFilter; const GLenum magFilterSave = texObj->MagFilter; + const GLint baseLevelSave = texObj->BaseLevel; + const GLint maxLevelSave = texObj->MaxLevel; + const GLboolean genMipmapSave = texObj->GenerateMipmap; + const GLenum wrapSSave = texObj->WrapS; + const GLenum wrapTSave = texObj->WrapT; + const GLenum wrapRSave = texObj->WrapR; const GLuint fboSave = ctx->DrawBuffer->Name; GLenum faceTarget; - GLuint level; + GLuint dstLevel; GLuint border = 0; /* check for fallbacks */ - if (!ctx->Extensions.EXT_framebuffer_object) { + if (!ctx->Extensions.EXT_framebuffer_object || + target == GL_TEXTURE_3D) { _mesa_generate_mipmap(ctx, target, texObj); return; } @@ -1911,11 +1999,8 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, NULL, GL_DYNAMIC_DRAW_ARB); /* setup vertex arrays */ - _mesa_VertexPointer(2, GL_FLOAT, sizeof(verts[0]), - (void *) (0 * sizeof(GLfloat))); - _mesa_TexCoordPointer(3, GL_FLOAT, sizeof(verts[0]), - (void *) (2 * sizeof(GLfloat))); - + _mesa_VertexPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); _mesa_EnableClientState(GL_VERTEX_ARRAY); _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); } @@ -1933,12 +2018,16 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, GL_LINEAR); _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, GL_FALSE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE); + _mesa_set_enable(ctx, target, GL_TRUE); /* setup texcoords once (XXX what about border?) */ switch (faceTarget) { - case GL_TEXTURE_CUBE_MAP_POSITIVE_X: - break; + case GL_TEXTURE_1D: case GL_TEXTURE_2D: verts[0].s = 0.0F; verts[0].t = 0.0F; @@ -1953,63 +2042,180 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, verts[3].t = 1.0F; verts[3].r = 0.0F; break; + case GL_TEXTURE_3D: + abort(); + break; + default: + /* cube face */ + { + static const GLfloat st[4][2] = { + {0.0f, 0.0f}, {1.0f, 0.0f}, {1.0f, 1.0f}, {0.0f, 1.0f} + }; + GLuint i; + + /* loop over quad verts */ + for (i = 0; i < 4; i++) { + /* Compute sc = +/-scale and tc = +/-scale. + * Not +/-1 to avoid cube face selection ambiguity near the edges, + * though that can still sometimes happen with this scale factor... + */ + const GLfloat scale = 0.9999f; + const GLfloat sc = (2.0f * st[i][0] - 1.0f) * scale; + const GLfloat tc = (2.0f * st[i][1] - 1.0f) * scale; + + switch (faceTarget) { + case GL_TEXTURE_CUBE_MAP_POSITIVE_X: + verts[i].s = 1.0f; + verts[i].t = -tc; + verts[i].r = -sc; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X: + verts[i].s = -1.0f; + verts[i].t = -tc; + verts[i].r = sc; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y: + verts[i].s = sc; + verts[i].t = 1.0f; + verts[i].r = tc; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y: + verts[i].s = sc; + verts[i].t = -1.0f; + verts[i].r = -tc; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z: + verts[i].s = sc; + verts[i].t = -tc; + verts[i].r = 1.0f; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z: + verts[i].s = -sc; + verts[i].t = -tc; + verts[i].r = -1.0f; + break; + default: + assert(0); + } + } + } } + _mesa_set_enable(ctx, target, GL_TRUE); + + /* texture is already locked, unlock now */ + _mesa_unlock_texture(ctx, texObj); - for (level = baseLevel + 1; level <= maxLevel; level++) { + for (dstLevel = baseLevel + 1; dstLevel <= maxLevel; dstLevel++) { const struct gl_texture_image *srcImage; - const GLuint srcLevel = level - 1; - GLsizei srcWidth, srcHeight; - GLsizei newWidth, newHeight; + const GLuint srcLevel = dstLevel - 1; + GLsizei srcWidth, srcHeight, srcDepth; + GLsizei dstWidth, dstHeight, dstDepth; GLenum status; - srcImage = _mesa_select_tex_image(ctx, texObj, target, srcLevel); + srcImage = _mesa_select_tex_image(ctx, texObj, faceTarget, srcLevel); assert(srcImage->Border == 0); /* XXX we can fix this */ + /* src size w/out border */ srcWidth = srcImage->Width - 2 * border; srcHeight = srcImage->Height - 2 * border; + srcDepth = srcImage->Depth - 2 * border; - newWidth = MAX2(1, srcWidth / 2) + 2 * border; - newHeight = MAX2(1, srcHeight / 2) + 2 * border; + /* new dst size w/ border */ + dstWidth = MAX2(1, srcWidth / 2) + 2 * border; + dstHeight = MAX2(1, srcHeight / 2) + 2 * border; + dstDepth = MAX2(1, srcDepth / 2) + 2 * border; - if (newWidth == srcImage->Width && newHeight == srcImage->Height) { - break; + if (dstWidth == srcImage->Width && + dstHeight == srcImage->Height && + dstDepth == srcImage->Depth) { + /* all done */ + break; } - /* Create empty image */ - _mesa_TexImage2D(GL_TEXTURE_2D, level, srcImage->InternalFormat, - newWidth, newHeight, border, - GL_RGBA, GL_UNSIGNED_BYTE, NULL); + /* Create empty dest image */ + if (target == GL_TEXTURE_1D) { + _mesa_TexImage1D(target, dstLevel, srcImage->InternalFormat, + dstWidth, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + else if (target == GL_TEXTURE_3D) { + _mesa_TexImage3D(target, dstLevel, srcImage->InternalFormat, + dstWidth, dstHeight, dstDepth, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + else { + /* 2D or cube */ + _mesa_TexImage2D(faceTarget, dstLevel, srcImage->InternalFormat, + dstWidth, dstHeight, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + + if (target == GL_TEXTURE_CUBE_MAP) { + /* If texturing from a cube, we need to make sure all src faces + * have been defined (even if we're not sampling from them.) + * Otherwise the texture object will be 'incomplete' and + * texturing from it will not be allowed. + */ + GLuint face; + for (face = 0; face < 6; face++) { + if (!texObj->Image[face][srcLevel] || + texObj->Image[face][srcLevel]->Width != srcWidth) { + _mesa_TexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, + srcLevel, srcImage->InternalFormat, + srcWidth, srcHeight, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + } + } + } - /* vertex positions */ + /* setup vertex positions */ { verts[0].x = 0.0F; verts[0].y = 0.0F; - verts[1].x = (GLfloat) newWidth; + verts[1].x = (GLfloat) dstWidth; verts[1].y = 0.0F; - verts[2].x = (GLfloat) newWidth; - verts[2].y = (GLfloat) newHeight; + verts[2].x = (GLfloat) dstWidth; + verts[2].y = (GLfloat) dstHeight; verts[3].x = 0.0F; - verts[3].y = (GLfloat) newHeight; + verts[3].y = (GLfloat) dstHeight; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); } /* limit sampling to src level */ - _mesa_TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, srcLevel); - _mesa_TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, srcLevel); - - /* Set to draw into the current level */ - _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, - GL_COLOR_ATTACHMENT0_EXT, - target, - texObj->Name, - level); + _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, srcLevel); + _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, srcLevel); + + /* Set to draw into the current dstLevel */ + if (target == GL_TEXTURE_1D) { + _mesa_FramebufferTexture1DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + target, + texObj->Name, + dstLevel); + } + else if (target == GL_TEXTURE_3D) { + GLint zoffset = 0; /* XXX unfinished */ + _mesa_FramebufferTexture3DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + target, + texObj->Name, + dstLevel, zoffset); + } + else { + /* 2D / cube */ + _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + faceTarget, + texObj->Name, + dstLevel); + } - /* Choose to render to the color attachment. */ _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT); + /* sanity check */ status = _mesa_CheckFramebufferStatusEXT (GL_FRAMEBUFFER_EXT); if (status != GL_FRAMEBUFFER_COMPLETE_EXT) { abort(); @@ -2019,11 +2225,386 @@ _mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); } + _mesa_lock_texture(ctx, texObj); /* relock */ + _mesa_meta_end(ctx); _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, minFilterSave); _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, magFilterSave); + _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, baseLevelSave); + _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, maxLevelSave); + _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, genMipmapSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, wrapSSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, wrapTSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, wrapRSave); - /* restore (XXX add to meta_begin/end()? */ _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fboSave); } + + +/** + * Determine the GL data type to use for the temporary image read with + * ReadPixels() and passed to Tex[Sub]Image(). + */ +static GLenum +get_temp_image_type(GLcontext *ctx, GLenum baseFormat) +{ + switch (baseFormat) { + case GL_RGBA: + case GL_RGB: + case GL_ALPHA: + case GL_LUMINANCE: + case GL_LUMINANCE_ALPHA: + case GL_INTENSITY: + if (ctx->DrawBuffer->Visual.redBits <= 8) + return GL_UNSIGNED_BYTE; + else if (ctx->DrawBuffer->Visual.redBits <= 8) + return GL_UNSIGNED_SHORT; + else + return GL_FLOAT; + case GL_DEPTH_COMPONENT: + return GL_UNSIGNED_INT; + case GL_DEPTH_STENCIL: + return GL_UNSIGNED_INT_24_8; + default: + _mesa_problem(ctx, "Unexpected format in get_temp_image_type()"); + return 0; + } +} + + +/** + * Helper for _mesa_meta_CopyTexImage1/2D() functions. + * Have to be careful with locking and meta state for pixel transfer. + */ +static void +copy_tex_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height, GLint border) +{ + struct gl_texture_unit *texUnit; + struct gl_texture_object *texObj; + struct gl_texture_image *texImage; + GLsizei postConvWidth = width, postConvHeight = height; + GLenum format, type; + GLint bpp; + void *buf; + + texUnit = _mesa_get_current_tex_unit(ctx); + texObj = _mesa_select_tex_object(ctx, texUnit, target); + texImage = _mesa_get_tex_image(ctx, texObj, target, level); + + format = _mesa_base_tex_format(ctx, internalFormat); + type = get_temp_image_type(ctx, format); + bpp = _mesa_bytes_per_pixel(format, type); + if (bpp <= 0) { + _mesa_problem(ctx, "Bad bpp in meta copy_tex_image()"); + return; + } + + /* + * Alloc image buffer (XXX could use a PBO) + */ + buf = _mesa_malloc(width * height * bpp); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexImage%uD", dims); + return; + } + + _mesa_unlock_texture(ctx, texObj); /* need to unlock first */ + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, height, + format, type, &ctx->Pack, buf); + _mesa_meta_end(ctx); + + /* + * Prepare for new texture image size/data + */ +#if FEATURE_convolve + if (_mesa_is_color_format(internalFormat)) { + _mesa_adjust_image_for_convolution(ctx, 2, + &postConvWidth, &postConvHeight); + } +#endif + + if (texImage->Data) { + ctx->Driver.FreeTexImageData(ctx, texImage); + } + + _mesa_init_teximage_fields(ctx, target, texImage, + postConvWidth, postConvHeight, 1, + border, internalFormat); + + /* + * Store texture data (with pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE); + + _mesa_update_state(ctx); /* to update pixel transfer state */ + + if (target == GL_TEXTURE_1D) { + ctx->Driver.TexImage1D(ctx, target, level, internalFormat, + width, border, format, type, + buf, &ctx->Unpack, texObj, texImage); + } + else { + ctx->Driver.TexImage2D(ctx, target, level, internalFormat, + width, height, border, format, type, + buf, &ctx->Unpack, texObj, texImage); + } + _mesa_meta_end(ctx); + + _mesa_lock_texture(ctx, texObj); /* re-lock */ + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyTexImage1D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLint border) +{ + copy_tex_image(ctx, 1, target, level, internalFormat, x, y, + width, 1, border); +} + + +void +_mesa_meta_CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height, GLint border) +{ + copy_tex_image(ctx, 2, target, level, internalFormat, x, y, + width, height, border); +} + + + +/** + * Helper for _mesa_meta_CopyTexSubImage1/2/3D() functions. + * Have to be careful with locking and meta state for pixel transfer. + */ +static void +copy_tex_sub_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, + GLsizei width, GLsizei height) +{ + struct gl_texture_unit *texUnit; + struct gl_texture_object *texObj; + struct gl_texture_image *texImage; + GLenum format, type; + GLint bpp; + void *buf; + + texUnit = _mesa_get_current_tex_unit(ctx); + texObj = _mesa_select_tex_object(ctx, texUnit, target); + texImage = _mesa_select_tex_image(ctx, texObj, target, level); + + format = texImage->TexFormat->BaseFormat; + type = get_temp_image_type(ctx, format); + bpp = _mesa_bytes_per_pixel(format, type); + if (bpp <= 0) { + _mesa_problem(ctx, "Bad bpp in meta copy_tex_sub_image()"); + return; + } + + /* + * Alloc image buffer (XXX could use a PBO) + */ + buf = _mesa_malloc(width * height * bpp); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage%uD", dims); + return; + } + + _mesa_unlock_texture(ctx, texObj); /* need to unlock first */ + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, height, + format, type, &ctx->Pack, buf); + _mesa_meta_end(ctx); + + _mesa_update_state(ctx); /* to update pixel transfer state */ + + /* + * Store texture data (with pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE); + if (target == GL_TEXTURE_1D) { + ctx->Driver.TexSubImage1D(ctx, target, level, xoffset, + width, format, type, buf, + &ctx->Unpack, texObj, texImage); + } + else if (target == GL_TEXTURE_3D) { + ctx->Driver.TexSubImage3D(ctx, target, level, xoffset, yoffset, zoffset, + width, height, 1, format, type, buf, + &ctx->Unpack, texObj, texImage); + } + else { + ctx->Driver.TexSubImage2D(ctx, target, level, xoffset, yoffset, + width, height, format, type, buf, + &ctx->Unpack, texObj, texImage); + } + _mesa_meta_end(ctx); + + _mesa_lock_texture(ctx, texObj); /* re-lock */ + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyTexSubImage1D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, + GLint x, GLint y, GLsizei width) +{ + copy_tex_sub_image(ctx, 1, target, level, xoffset, 0, 0, + x, y, width, 1); +} + + +void +_mesa_meta_CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLint x, GLint y, + GLsizei width, GLsizei height) +{ + copy_tex_sub_image(ctx, 2, target, level, xoffset, yoffset, 0, + x, y, width, height); +} + + +void +_mesa_meta_CopyTexSubImage3D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, + GLsizei width, GLsizei height) +{ + copy_tex_sub_image(ctx, 3, target, level, xoffset, yoffset, zoffset, + x, y, width, height); +} + + +void +_mesa_meta_CopyColorTable(GLcontext *ctx, + GLenum target, GLenum internalformat, + GLint x, GLint y, GLsizei width) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyColorTable"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, 1, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ColorTable(target, internalformat, width, GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyColorSubTable(GLcontext *ctx,GLenum target, GLsizei start, + GLint x, GLint y, GLsizei width) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyColorSubTable"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + ctx->Driver.ReadPixels(ctx, x, y, width, 1, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ColorSubTable(target, start, width, GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, + GLenum internalFormat, + GLint x, GLint y, GLsizei width) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyConvolutionFilter2D"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + _mesa_update_state(ctx); + ctx->Driver.ReadPixels(ctx, x, y, width, 1, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ConvolutionFilter1D(target, internalFormat, width, + GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} + + +void +_mesa_meta_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height) +{ + GLfloat *buf; + + buf = (GLfloat *) _mesa_malloc(width * height * 4 * sizeof(GLfloat)); + if (!buf) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyConvolutionFilter2D"); + return; + } + + /* + * Read image from framebuffer (disable pixel transfer ops) + */ + _mesa_meta_begin(ctx, META_PIXEL_STORE | META_PIXEL_TRANSFER); + _mesa_update_state(ctx); + + ctx->Driver.ReadPixels(ctx, x, y, width, height, + GL_RGBA, GL_FLOAT, &ctx->Pack, buf); + + _mesa_ConvolutionFilter2D(target, internalFormat, width, height, + GL_RGBA, GL_FLOAT, buf); + + _mesa_meta_end(ctx); + + _mesa_free(buf); +} diff --git a/src/mesa/drivers/common/meta.h b/src/mesa/drivers/common/meta.h index 171ad27f26..7f659528dc 100644 --- a/src/mesa/drivers/common/meta.h +++ b/src/mesa/drivers/common/meta.h @@ -27,29 +27,6 @@ #define META_H -/** - * Flags passed to _mesa_meta_begin(). - * XXX these flags may evolve... - */ -/*@{*/ -#define META_ALPHA_TEST 0x1 -#define META_BLEND 0x2 /**< includes logicop */ -#define META_COLOR_MASK 0x4 -#define META_DEPTH_TEST 0x8 -#define META_FOG 0x10 -#define META_RASTERIZATION 0x20 -#define META_SCISSOR 0x40 -#define META_SHADER 0x80 -#define META_STENCIL_TEST 0x100 -#define META_TRANSFORM 0x200 /**< modelview, projection */ -#define META_TEXTURE 0x400 -#define META_VERTEX 0x800 -#define META_VIEWPORT 0x1000 -#define META_PIXEL_STORE 0x2000 -#define META_ALL ~0x0 -/*@}*/ - - extern void _mesa_meta_init(GLcontext *ctx); @@ -57,35 +34,81 @@ extern void _mesa_meta_free(GLcontext *ctx); extern void -_mesa_meta_blit_framebuffer(GLcontext *ctx, - GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, - GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLbitfield mask, GLenum filter); +_mesa_meta_BlitFramebuffer(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter); extern void -_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers); +_mesa_meta_Clear(GLcontext *ctx, GLbitfield buffers); extern void -_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcx, GLint srcy, - GLsizei width, GLsizei height, - GLint dstx, GLint dsty, GLenum type); +_mesa_meta_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, + GLsizei width, GLsizei height, + GLint dstx, GLint dsty, GLenum type); extern void -_mesa_meta_draw_pixels(GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels); +_mesa_meta_DrawPixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels); extern void -_mesa_meta_bitmap(GLcontext *ctx, +_mesa_meta_Bitmap(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap); extern void -_mesa_meta_generate_mipmap(GLcontext *ctx, GLenum target, - struct gl_texture_object *texObj); +_mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj); + +extern void +_mesa_meta_CopyTexImage1D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLint border); + +extern void +_mesa_meta_CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height, GLint border); + +extern void +_mesa_meta_CopyTexSubImage1D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLint x, GLint y, + GLsizei width, GLsizei height); + +extern void +_mesa_meta_CopyTexSubImage3D(GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, GLint zoffset, + GLint x, GLint y, + GLsizei width, GLsizei height); + +extern void +_mesa_meta_CopyColorTable(GLcontext *ctx, + GLenum target, GLenum internalformat, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyColorSubTable(GLcontext *ctx,GLenum target, GLsizei start, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyConvolutionFilter1D(GLcontext *ctx, GLenum target, + GLenum internalFormat, + GLint x, GLint y, GLsizei width); + +extern void +_mesa_meta_CopyConvolutionFilter2D(GLcontext *ctx, GLenum target, + GLenum internalFormat, GLint x, GLint y, + GLsizei width, GLsizei height); #endif /* META_H */ diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h index 2c89a9f58a..5e86324eec 100644 --- a/src/mesa/drivers/dri/common/extension_helper.h +++ b/src/mesa/drivers/dri/common/extension_helper.h @@ -4417,10 +4417,11 @@ static const char SpriteParameterivSGIX_names[] = ""; #endif -#if defined(need_GL_EXT_provoking_vertex) +#if defined(need_GL_EXT_provoking_vertex) || defined(need_GL_ARB_provoking_vertex) static const char ProvokingVertexEXT_names[] = "i\0" /* Parameter signature */ "glProvokingVertexEXT\0" + "glProvokingVertex\0" ""; #endif @@ -5194,6 +5195,13 @@ static const struct dri_extension_function GL_ARB_point_parameters_functions[] = }; #endif +#if defined(need_GL_ARB_provoking_vertex) +static const struct dri_extension_function GL_ARB_provoking_vertex_functions[] = { + { ProvokingVertexEXT_names, ProvokingVertexEXT_remap_index, -1 }, + { NULL, 0, 0 } +}; +#endif + #if defined(need_GL_ARB_shader_objects) static const struct dri_extension_function GL_ARB_shader_objects_functions[] = { { UniformMatrix3fvARB_names, UniformMatrix3fvARB_remap_index, -1 }, diff --git a/src/mesa/drivers/dri/ffb/ffb_tex.c b/src/mesa/drivers/dri/ffb/ffb_tex.c index 69d30aedba..95058e9069 100644 --- a/src/mesa/drivers/dri/ffb/ffb_tex.c +++ b/src/mesa/drivers/dri/ffb/ffb_tex.c @@ -30,24 +30,6 @@ #include "ffb_tex.h" /* No texture unit, all software. */ -/* XXX this function isn't needed since _mesa_init_driver_functions() - * will make all these assignments. - */ void ffbDDInitTexFuncs(GLcontext *ctx) { - /* - ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format; - ctx->Driver.TexImage1D = _mesa_store_teximage1d; - ctx->Driver.TexImage2D = _mesa_store_teximage2d; - ctx->Driver.TexImage3D = _mesa_store_teximage3d; - ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d; - ctx->Driver.TexSubImage2D = _mesa_store_texsubimage2d; - ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; - ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; - ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; - ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; - ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; - ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; - ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; - */ } diff --git a/src/mesa/drivers/dri/glcore/Makefile b/src/mesa/drivers/dri/glcore/Makefile deleted file mode 100644 index ac7e1de928..0000000000 --- a/src/mesa/drivers/dri/glcore/Makefile +++ /dev/null @@ -1,84 +0,0 @@ -# src/mesa/drivers/dri/glcore/Makefile - -TOP = ../../../../.. -include $(TOP)/configs/current - -LIBNAME = glcore_dri.so - -DRIVER_SOURCES = glcore_driver.c \ - $(TOP)/src/mesa/drivers/common/driverfuncs.c \ - ../common/dri_util.c - -C_SOURCES = \ - $(DRIVER_SOURCES) \ - $(DRI_SOURCES) - - -# Include directories -INCLUDE_DIRS = \ - -I. \ - -I../common \ - -I../dri_client \ - -I../dri_client/imports \ - -Iserver \ - -I$(TOP)/include \ - -I$(DRM_SOURCE_PATH)/shared-core \ - -I$(TOP)/src/mesa \ - -I$(TOP)/src/mesa/main \ - -I$(TOP)/src/mesa/glapi \ - -I$(TOP)/src/mesa/math \ - -I$(TOP)/src/mesa/transform \ - -I$(TOP)/src/mesa/shader \ - -I$(TOP)/src/mesa/swrast \ - -I$(TOP)/src/mesa/swrast_setup - -# Core Mesa objects -MESA_MODULES = $(TOP)/src/mesa/libmesa.a - -# Libraries that the driver shared lib depends on -LIB_DEPS = -lm -lpthread -lc -# LIB_DEPS = -lGL -lm -lpthread -lc - - -ASM_SOURCES = - -OBJECTS = $(C_SOURCES:.c=.o) \ - $(ASM_SOURCES:.S=.o) - - -##### RULES ##### - -.c.o: - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@ - -.S.o: - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@ - - -##### TARGETS ##### - -default: depend $(TOP)/$(LIB_DIR)/$(LIBNAME) - - -$(TOP)/$(LIB_DIR)/$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile - CC="$(CC)" CXX="$(CXX)" $(TOP)/bin/mklib -o $(LIBNAME) -noprefix -install $(TOP)/$(LIB_DIR) \ - $(OBJECTS) $(WINLIB) $(LIB_DEPS) $(WINOBJ) $(MESA_MODULES) - - -depend: $(C_SOURCES) $(ASM_SOURCES) - rm -f depend - touch depend - $(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(C_SOURCES) $(ASM_SOURCES) \ - > /dev/null - - -# Emacs tags -tags: - etags `find . -name \*.[ch]` `find ../include` - - -clean: - -rm -f *.o server/*.o - - -include depend diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c index 3c5b848319..c300c33adc 100644 --- a/src/mesa/drivers/dri/i965/brw_context.c +++ b/src/mesa/drivers/dri/i965/brw_context.c @@ -125,6 +125,7 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis, /* We want the GLSL compiler to emit code that uses condition codes */ ctx->Shader.EmitCondCodes = GL_TRUE; + ctx->Shader.EmitNVTempInitialization = GL_TRUE; ctx->Const.VertexProgram.MaxNativeInstructions = (16 * 1024); ctx->Const.VertexProgram.MaxAluInstructions = 0; diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c index 0b0e6931a0..4be6c77aa1 100644 --- a/src/mesa/drivers/dri/i965/brw_curbe.c +++ b/src/mesa/drivers/dri/i965/brw_curbe.c @@ -248,6 +248,9 @@ static void prepare_constant_buffer(struct brw_context *brw) GLuint offset = brw->curbe.vs_start * 16; GLuint nr = brw->vs.prog_data->nr_params / 4; + if (brw->vertex_program->IsNVProgram) + _mesa_load_tracked_matrices(ctx); + /* Updates the ParamaterValues[i] pointers for all parameters of the * basic type of PROGRAM_STATE_VAR. */ diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h index 78572356a3..5335eac895 100644 --- a/src/mesa/drivers/dri/i965/brw_state.h +++ b/src/mesa/drivers/dri/i965/brw_state.h @@ -86,9 +86,6 @@ const struct brw_tracked_state brw_psp_urb_cbs; const struct brw_tracked_state brw_pipe_control; -const struct brw_tracked_state brw_clear_surface_cache; -const struct brw_tracked_state brw_clear_batch_cache; - const struct brw_tracked_state brw_drawing_rect; const struct brw_tracked_state brw_indices; const struct brw_tracked_state brw_vertices; @@ -165,7 +162,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, const void *data, GLuint sz ); void brw_destroy_batch_cache( struct brw_context *brw ); -void brw_clear_batch_cache_flush( struct brw_context *brw ); +void brw_clear_batch_cache( struct brw_context *brw ); /* brw_wm_surface_state.c */ dri_bo * diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c index 811940edc0..7821898cf9 100644 --- a/src/mesa/drivers/dri/i965/brw_state_batch.c +++ b/src/mesa/drivers/dri/i965/brw_state_batch.c @@ -79,7 +79,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, return GL_TRUE; } -static void clear_batch_cache( struct brw_context *brw ) +void brw_clear_batch_cache( struct brw_context *brw ) { struct brw_cached_batch_item *item = brw->cached_batch_items; @@ -93,18 +93,7 @@ static void clear_batch_cache( struct brw_context *brw ) brw->cached_batch_items = NULL; } -void brw_clear_batch_cache_flush( struct brw_context *brw ) -{ - clear_batch_cache(brw); - - brw->state.dirty.mesa |= ~0; - brw->state.dirty.brw |= ~0; - brw->state.dirty.cache |= ~0; -} - - - void brw_destroy_batch_cache( struct brw_context *brw ) { - clear_batch_cache(brw); + brw_clear_batch_cache(brw); } diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c index 414620d0b3..b817b741e7 100644 --- a/src/mesa/drivers/dri/i965/brw_state_upload.c +++ b/src/mesa/drivers/dri/i965/brw_state_upload.c @@ -287,6 +287,7 @@ void brw_validate_state( struct brw_context *brw ) if (brw->emit_state_always) { state->mesa |= ~0; state->brw |= ~0; + state->cache |= ~0; } if (brw->fragment_program != ctx->FragmentProgram._Current) { @@ -305,7 +306,7 @@ void brw_validate_state( struct brw_context *brw ) return; if (brw->state.dirty.brw & BRW_NEW_CONTEXT) - brw_clear_batch_cache_flush(brw); + brw_clear_batch_cache(brw); brw->intel.Fallback = 0; diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c index e3111c6680..f0c79efbd9 100644 --- a/src/mesa/drivers/dri/i965/brw_vs.c +++ b/src/mesa/drivers/dri/i965/brw_vs.c @@ -90,8 +90,6 @@ static void brw_upload_vs_prog(struct brw_context *brw) struct brw_vertex_program *vp = (struct brw_vertex_program *)brw->vertex_program; - assert (vp && !vp->program.IsNVProgram); - memset(&key, 0, sizeof(key)); /* Just upload the program verbatim for now. Always send it all diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 584fdbdfc3..1638ef8111 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -1270,9 +1270,27 @@ post_vs_emit( struct brw_vs_compile *c, } static uint32_t -get_predicate(uint32_t swizzle) +get_predicate(const struct prog_instruction *inst) { - switch (swizzle) { + if (inst->DstReg.CondMask == COND_TR) + return BRW_PREDICATE_NONE; + + /* All of GLSL only produces predicates for COND_NE and one channel per + * vector. Fail badly if someone starts doing something else, as it might + * mean infinite looping or something. + * + * We'd like to support all the condition codes, but our hardware doesn't + * quite match the Mesa IR, which is modeled after the NV extensions. For + * those, the instruction may update the condition codes or not, then any + * later instruction may use one of those condition codes. For gen4, the + * instruction may update the flags register based on one of the condition + * codes output by the instruction, and then further instructions may + * predicate on that. We can probably support this, but it won't + * necessarily be easy. + */ + assert(inst->DstReg.CondMask == COND_NE); + + switch (inst->DstReg.CondSwizzle) { case SWIZZLE_XXXX: return BRW_PREDICATE_ALIGN16_REPLICATE_X; case SWIZZLE_YYYY: @@ -1282,7 +1300,8 @@ get_predicate(uint32_t swizzle) case SWIZZLE_WWWW: return BRW_PREDICATE_ALIGN16_REPLICATE_W; default: - _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", swizzle); + _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", + inst->DstReg.CondMask); return BRW_PREDICATE_NORMAL; } } @@ -1294,6 +1313,7 @@ void brw_vs_emit(struct brw_vs_compile *c ) #define MAX_IF_DEPTH 32 #define MAX_LOOP_DEPTH 32 struct brw_compile *p = &c->func; + struct brw_context *brw = p->brw; const GLuint nr_insns = c->vp->program.Base.NumInstructions; GLuint insn, if_depth = 0, loop_depth = 0; GLuint end_offset = 0; @@ -1492,8 +1512,8 @@ void brw_vs_emit(struct brw_vs_compile *c ) case OPCODE_IF: assert(if_depth < MAX_IF_DEPTH); if_inst[if_depth] = brw_IF(p, BRW_EXECUTE_8); - if_inst[if_depth]->header.predicate_control = - get_predicate(inst->DstReg.CondSwizzle); + /* Note that brw_IF smashes the predicate_control field. */ + if_inst[if_depth]->header.predicate_control = get_predicate(inst); if_depth++; break; case OPCODE_ELSE: @@ -1503,45 +1523,48 @@ void brw_vs_emit(struct brw_vs_compile *c ) assert(if_depth > 0); brw_ENDIF(p, if_inst[--if_depth]); break; -#if 0 case OPCODE_BGNLOOP: loop_inst[loop_depth++] = brw_DO(p, BRW_EXECUTE_8); break; case OPCODE_BRK: + brw_set_predicate_control(p, get_predicate(inst)); brw_BREAK(p); - brw_set_predicate_control(p, BRW_PREDICATE_NONE); + brw_set_predicate_control(p, BRW_PREDICATE_NONE); break; case OPCODE_CONT: + brw_set_predicate_control(p, get_predicate(inst)); brw_CONT(p); brw_set_predicate_control(p, BRW_PREDICATE_NONE); break; case OPCODE_ENDLOOP: { struct brw_instruction *inst0, *inst1; + GLuint br = 1; + loop_depth--; + + if (BRW_IS_IGDNG(brw)) + br = 2; + inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]); /* patch all the BREAK/CONT instructions from last BEGINLOOP */ while (inst0 > loop_inst[loop_depth]) { inst0--; if (inst0->header.opcode == BRW_OPCODE_BREAK) { - inst0->bits3.if_else.jump_count = inst1 - inst0 + 1; + inst0->bits3.if_else.jump_count = br * (inst1 - inst0 + 1); inst0->bits3.if_else.pop_count = 0; } else if (inst0->header.opcode == BRW_OPCODE_CONTINUE) { - inst0->bits3.if_else.jump_count = inst1 - inst0; + inst0->bits3.if_else.jump_count = br * (inst1 - inst0); inst0->bits3.if_else.pop_count = 0; } } } break; -#else - (void) loop_inst; - (void) loop_depth; -#endif case OPCODE_BRA: - brw_set_predicate_control(p, BRW_PREDICATE_NORMAL); + brw_set_predicate_control(p, get_predicate(inst)); brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16)); - brw_set_predicate_control_flag_value(p, 0xff); + brw_set_predicate_control(p, BRW_PREDICATE_NONE); break; case OPCODE_CAL: brw_set_access_mode(p, BRW_ALIGN_1); diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h index ae98b5492d..872b1f3ecf 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.h +++ b/src/mesa/drivers/dri/i965/brw_wm.h @@ -38,6 +38,8 @@ #include "brw_context.h" #include "brw_eu.h" +#define SATURATE (1<<5) + /* A big lookup table is used to figure out which and how many * additional regs will inserted before the main payload in the WM * program execution. These mainly relate to depth and stencil @@ -203,7 +205,6 @@ struct brw_wm_compile { GLuint fp_temp; GLuint fp_interp_emitted; GLuint fp_fragcolor_emitted; - GLuint fp_deriv_emitted; struct prog_src_register pixel_xy; struct prog_src_register delta_xy; @@ -299,5 +300,10 @@ void brw_wm_lookup_iz( GLuint line_aa, GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp); void brw_wm_glsl_emit(struct brw_context *brw, struct brw_wm_compile *c); +void emit_ddxy(struct brw_compile *p, + const struct brw_reg *dst, + GLuint mask, + GLboolean is_ddx, + const struct brw_reg *arg0); #endif diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c index 981864323e..bf80a2942a 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_emit.c +++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c @@ -34,8 +34,6 @@ #include "brw_context.h" #include "brw_wm.h" -#define SATURATE (1<<5) - /* Not quite sure how correct this is - need to understand horiz * vs. vertical strides a little better. */ @@ -281,6 +279,79 @@ static void emit_frontfacing( struct brw_compile *p, brw_set_predicate_control_flag_value(p, 0xff); } +/* For OPCODE_DDX and OPCODE_DDY, per channel of output we've got input + * looking like: + * + * arg0: ss0.tl ss0.tr ss0.bl ss0.br ss1.tl ss1.tr ss1.bl ss1.br + * + * and we're trying to produce: + * + * DDX DDY + * dst: (ss0.tr - ss0.tl) (ss0.tl - ss0.bl) + * (ss0.tr - ss0.tl) (ss0.tr - ss0.br) + * (ss0.br - ss0.bl) (ss0.tl - ss0.bl) + * (ss0.br - ss0.bl) (ss0.tr - ss0.br) + * (ss1.tr - ss1.tl) (ss1.tl - ss1.bl) + * (ss1.tr - ss1.tl) (ss1.tr - ss1.br) + * (ss1.br - ss1.bl) (ss1.tl - ss1.bl) + * (ss1.br - ss1.bl) (ss1.tr - ss1.br) + * + * and add another set of two more subspans if in 16-pixel dispatch mode. + * + * For DDX, it ends up being easy: width = 2, horiz=0 gets us the same result + * for each pair, and vertstride = 2 jumps us 2 elements after processing a + * pair. But for DDY, it's harder, as we want to produce the pairs swizzled + * between each other. We could probably do it like ddx and swizzle the right + * order later, but bail for now and just produce + * ((ss0.tl - ss0.bl)x4 (ss1.tl - ss1.bl)x4) + */ +void emit_ddxy(struct brw_compile *p, + const struct brw_reg *dst, + GLuint mask, + GLboolean is_ddx, + const struct brw_reg *arg0) +{ + int i; + struct brw_reg src0, src1; + + if (mask & SATURATE) + brw_set_saturate(p, 1); + for (i = 0; i < 4; i++ ) { + if (mask & (1<<i)) { + if (is_ddx) { + src0 = brw_reg(arg0[i].file, arg0[i].nr, 1, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_2, + BRW_WIDTH_2, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + src1 = brw_reg(arg0[i].file, arg0[i].nr, 0, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_2, + BRW_WIDTH_2, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + } else { + src0 = brw_reg(arg0[i].file, arg0[i].nr, 0, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_4, + BRW_WIDTH_4, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + src1 = brw_reg(arg0[i].file, arg0[i].nr, 2, + BRW_REGISTER_TYPE_F, + BRW_VERTICAL_STRIDE_4, + BRW_WIDTH_4, + BRW_HORIZONTAL_STRIDE_0, + BRW_SWIZZLE_XYZW, WRITEMASK_XYZW); + } + brw_ADD(p, dst[i], src0, negate(src1)); + } + } + if (mask & SATURATE) + brw_set_saturate(p, 0); +} + static void emit_alu1( struct brw_compile *p, struct brw_instruction *(*func)(struct brw_compile *, struct brw_reg, @@ -1272,6 +1343,14 @@ void brw_wm_emit( struct brw_wm_compile *c ) emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]); break; + case OPCODE_DDX: + emit_ddxy(p, dst, dst_flags, GL_TRUE, args[0]); + break; + + case OPCODE_DDY: + emit_ddxy(p, dst, dst_flags, GL_FALSE, args[0]); + break; + case OPCODE_DP3: emit_dp3(p, dst, dst_flags, args[0], args[1]); break; diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c index 123fe841c3..4e3edfbbff 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_fp.c +++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c @@ -494,38 +494,6 @@ static void emit_interp( struct brw_wm_compile *c, c->fp_interp_emitted |= 1<<idx; } -static void emit_ddx( struct brw_wm_compile *c, - const struct prog_instruction *inst ) -{ - GLuint idx = inst->SrcReg[0].Index; - struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx); - - c->fp_deriv_emitted |= 1<<idx; - emit_op(c, - OPCODE_DDX, - inst->DstReg, - 0, - interp, - get_pixel_w(c), - src_undef()); -} - -static void emit_ddy( struct brw_wm_compile *c, - const struct prog_instruction *inst ) -{ - GLuint idx = inst->SrcReg[0].Index; - struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx); - - c->fp_deriv_emitted |= 1<<idx; - emit_op(c, - OPCODE_DDY, - inst->DstReg, - 0, - interp, - get_pixel_w(c), - src_undef()); -} - /*********************************************************************** * Hacks to extend the program parameter and constant lists. */ @@ -1186,12 +1154,6 @@ void brw_wm_pass_fp( struct brw_wm_compile *c ) */ out->DstReg.WriteMask = 0; break; - case OPCODE_DDX: - emit_ddx(c, inst); - break; - case OPCODE_DDY: - emit_ddy(c, inst); - break; case OPCODE_END: emit_fb_write(c); break; diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c index 7c210abbce..c9fe1dd8ad 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c +++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c @@ -22,6 +22,7 @@ static struct brw_reg get_dst_reg(struct brw_wm_compile *c, GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp) { int i; + for (i = 0; i < fp->Base.NumInstructions; i++) { const struct prog_instruction *inst = &fp->Base.Instructions[i]; switch (inst->Opcode) { @@ -31,8 +32,6 @@ GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp) case OPCODE_CAL: case OPCODE_BRK: case OPCODE_RET: - case OPCODE_DDX: - case OPCODE_DDY: case OPCODE_NOISE1: case OPCODE_NOISE2: case OPCODE_NOISE3: @@ -293,7 +292,7 @@ static void prealloc_reg(struct brw_wm_compile *c) int i, j; struct brw_reg reg; int urb_read_length = 0; - GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted | c->fp_deriv_emitted; + GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted; GLuint reg_index = 0; memset(c->used_grf, GL_FALSE, sizeof(c->used_grf)); @@ -1474,61 +1473,6 @@ static void emit_sne(struct brw_wm_compile *c, emit_sop(c, inst, BRW_CONDITIONAL_NEQ); } -static void emit_ddx(struct brw_wm_compile *c, - const struct prog_instruction *inst) -{ - struct brw_compile *p = &c->func; - GLuint mask = inst->DstReg.WriteMask; - struct brw_reg interp[4]; - struct brw_reg dst; - struct brw_reg src0, w; - GLuint nr, i; - src0 = get_src_reg(c, inst, 0, 0); - w = get_src_reg(c, inst, 1, 3); - nr = src0.nr; - interp[0] = brw_vec1_grf(nr, 0); - interp[1] = brw_vec1_grf(nr, 4); - interp[2] = brw_vec1_grf(nr+1, 0); - interp[3] = brw_vec1_grf(nr+1, 4); - brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF); - for(i = 0; i < 4; i++ ) { - if (mask & (1<<i)) { - dst = get_dst_reg(c, inst, i); - brw_MOV(p, dst, interp[i]); - brw_MUL(p, dst, dst, w); - } - } - brw_set_saturate(p, 0); -} - -static void emit_ddy(struct brw_wm_compile *c, - const struct prog_instruction *inst) -{ - struct brw_compile *p = &c->func; - GLuint mask = inst->DstReg.WriteMask; - struct brw_reg interp[4]; - struct brw_reg dst; - struct brw_reg src0, w; - GLuint nr, i; - - src0 = get_src_reg(c, inst, 0, 0); - nr = src0.nr; - w = get_src_reg(c, inst, 1, 3); - interp[0] = brw_vec1_grf(nr, 0); - interp[1] = brw_vec1_grf(nr, 4); - interp[2] = brw_vec1_grf(nr+1, 0); - interp[3] = brw_vec1_grf(nr+1, 4); - brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF); - for(i = 0; i < 4; i++ ) { - if (mask & (1<<i)) { - dst = get_dst_reg(c, inst, i); - brw_MOV(p, dst, suboffset(interp[i], 1)); - brw_MUL(p, dst, dst, w); - } - } - brw_set_saturate(p, 0); -} - static INLINE struct brw_reg high_words( struct brw_reg reg ) { return stride( suboffset( retype( reg, BRW_REGISTER_TYPE_W ), 1 ), @@ -2780,6 +2724,21 @@ static void post_wm_emit( struct brw_wm_compile *c ) brw_resolve_cals(&c->func); } +static void +get_argument_regs(struct brw_wm_compile *c, + const struct prog_instruction *inst, + int index, + struct brw_reg *regs, + int mask) +{ + int i; + + for (i = 0; i < 4; i++) { + if (mask & (1 << i)) + regs[i] = get_src_reg(c, inst, index, i); + } +} + static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) { #define MAX_IF_DEPTH 32 @@ -2797,6 +2756,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) for (i = 0; i < c->nr_fp_insns; i++) { const struct prog_instruction *inst = &c->prog_instructions[i]; + int dst_flags; + struct brw_reg args[3][4], dst[4]; + int j; c->cur_inst = i; @@ -2814,6 +2776,10 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) else brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE); + dst_flags = inst->DstReg.WriteMask; + if (inst->SaturateMode == SATURATE_ZERO_ONE) + dst_flags |= SATURATE; + switch (inst->Opcode) { case WM_PIXELXY: emit_pixel_xy(c, inst); @@ -2899,10 +2865,16 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) emit_min_max(c, inst); break; case OPCODE_DDX: - emit_ddx(c, inst); - break; case OPCODE_DDY: - emit_ddy(c, inst); + for (j = 0; j < 4; j++) { + if (inst->DstReg.WriteMask & (1 << j)) + dst[j] = get_dst_reg(c, inst, j); + else + dst[j] = brw_null_reg(); + } + get_argument_regs(c, inst, 0, args[0], WRITEMASK_XYZW); + emit_ddxy(p, dst, dst_flags, (inst->Opcode == OPCODE_DDX), + args[0]); break; case OPCODE_SLT: emit_slt(c, inst); diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass1.c b/src/mesa/drivers/dri/i965/brw_wm_pass1.c index 9c68bfd78b..b449394029 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_pass1.c +++ b/src/mesa/drivers/dri/i965/brw_wm_pass1.c @@ -178,6 +178,11 @@ void brw_wm_pass1( struct brw_wm_compile *c ) read1 = writemask; break; + case OPCODE_DDX: + case OPCODE_DDY: + read0 = writemask; + break; + case OPCODE_MAD: case OPCODE_CMP: case OPCODE_LRP: diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c index c55c5c426e..7f6fb66d52 100644 --- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c +++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c @@ -207,8 +207,12 @@ intel_bufferobj_subdata(GLcontext * ctx, if (intel_obj->sys_buffer) memcpy((char *)intel_obj->sys_buffer + offset, data, size); - else + else { + /* Flush any existing batchbuffer that might reference this data. */ + intelFlush(ctx); + dri_bo_subdata(intel_obj->buffer, offset, size, data); + } } diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c index bce23724b3..fb62f0f430 100644 --- a/src/mesa/drivers/dri/intel/intel_clear.c +++ b/src/mesa/drivers/dri/intel/intel_clear.c @@ -38,6 +38,7 @@ #include "intel_fbo.h" #include "intel_pixel.h" #include "intel_regions.h" +#include "intel_batchbuffer.h" #define FILE_DEBUG_FLAG DEBUG_BLIT @@ -170,7 +171,8 @@ intelClear(GLcontext *ctx, GLbitfield mask) } DBG("\n"); } - _mesa_meta_clear(&intel->ctx, tri_mask); + + _mesa_meta_Clear(&intel->ctx, tri_mask); } if (swrast_mask) { diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index 7e21b94acc..d49d95768d 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -507,7 +507,8 @@ intel_flush(GLcontext *ctx, GLboolean needs_mi_flush) if (screen->dri2.loader && (screen->dri2.loader->base.version >= 2) - && (screen->dri2.loader->flushFrontBuffer != NULL)) { + && (screen->dri2.loader->flushFrontBuffer != NULL) && + intel->driDrawable && intel->driDrawable->loaderPrivate) { (*screen->dri2.loader->flushFrontBuffer)(intel->driDrawable, intel->driDrawable->loaderPrivate); @@ -587,11 +588,6 @@ intelInitDriverFunctions(struct dd_function_table *functions) functions->GetString = intelGetString; functions->UpdateState = intelInvalidateState; - functions->CopyColorTable = _swrast_CopyColorTable; - functions->CopyColorSubTable = _swrast_CopyColorSubTable; - functions->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; - functions->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; - intelInitTextureFuncs(functions); intelInitTextureImageFuncs(functions); intelInitTextureSubImageFuncs(functions); @@ -922,6 +918,14 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv) GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv) { + struct intel_context *intel = + (struct intel_context *) driContextPriv->driverPrivate; + + /* Deassociate the context with the drawables. + */ + intel->driDrawable = NULL; + intel->driReadDrawable = NULL; + return GL_TRUE; } diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c index 804c034840..8dfb24290d 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.c +++ b/src/mesa/drivers/dri/intel/intel_fbo.c @@ -715,5 +715,5 @@ intel_fbo_init(struct intel_context *intel) intel->ctx.Driver.FinishRenderTexture = intel_finish_render_texture; intel->ctx.Driver.ResizeBuffers = intel_resize_buffers; intel->ctx.Driver.ValidateFramebuffer = intel_validate_framebuffer; - intel->ctx.Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer; + intel->ctx.Driver.BlitFramebuffer = _mesa_meta_BlitFramebuffer; } diff --git a/src/mesa/drivers/dri/intel/intel_generatemipmap.c b/src/mesa/drivers/dri/intel/intel_generatemipmap.c index fe986092db..5958b36c97 100644 --- a/src/mesa/drivers/dri/intel/intel_generatemipmap.c +++ b/src/mesa/drivers/dri/intel/intel_generatemipmap.c @@ -125,6 +125,8 @@ intel_generate_mipmap_2d(GLcontext *ctx, GLuint fb_name; GLboolean success = GL_FALSE; struct gl_framebuffer *saved_fbo = NULL; + struct gl_buffer_object *saved_array_buffer = NULL; + struct gl_buffer_object *saved_element_buffer = NULL; _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT | GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT | @@ -133,6 +135,16 @@ intel_generate_mipmap_2d(GLcontext *ctx, old_active_texture = ctx->Texture.CurrentUnit; _mesa_reference_framebuffer(&saved_fbo, ctx->DrawBuffer); + /* use default array/index buffers */ + _mesa_reference_buffer_object(ctx, &saved_array_buffer, + ctx->Array.ArrayBufferObj); + _mesa_reference_buffer_object(ctx, &ctx->Array.ArrayBufferObj, + ctx->Shared->NullBufferObj); + _mesa_reference_buffer_object(ctx, &saved_element_buffer, + ctx->Array.ElementArrayBufferObj); + _mesa_reference_buffer_object(ctx, &ctx->Array.ElementArrayBufferObj, + ctx->Shared->NullBufferObj); + _mesa_Disable(GL_POLYGON_STIPPLE); _mesa_Disable(GL_DEPTH_TEST); _mesa_Disable(GL_STENCIL_TEST); @@ -205,6 +217,15 @@ fail: meta_restore_fragment_program(&intel->meta); meta_restore_vertex_program(&intel->meta); + /* restore array/index buffers */ + _mesa_reference_buffer_object(ctx, &ctx->Array.ArrayBufferObj, + saved_array_buffer); + _mesa_reference_buffer_object(ctx, &saved_array_buffer, NULL); + _mesa_reference_buffer_object(ctx, &ctx->Array.ElementArrayBufferObj, + saved_element_buffer); + _mesa_reference_buffer_object(ctx, &saved_element_buffer, NULL); + + _mesa_DeleteFramebuffersEXT(1, &fb_name); _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture); if (saved_fbo) @@ -223,10 +244,6 @@ fail: * * The texture object's miptree must be mapped. * - * It would be really nice if this was just called by Mesa whenever mipmaps - * needed to be regenerated, rather than us having to remember to do so in - * each texture image modification path. - * * This function should also include an accelerated path. */ void diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c index 07ca8f7ddb..f058b3c8e4 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c @@ -240,5 +240,5 @@ intelCopyPixels(GLcontext * ctx, return; /* this will use swrast if needed */ - _mesa_meta_copy_pixels(ctx, srcx, srcy, width, height, destx, desty, type); + _mesa_meta_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type); } diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c index 7fbb89fd6a..5ffa847fd4 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c @@ -54,7 +54,7 @@ #include "intel_fbo.h" -/** XXX compare perf of this vs. _mesa_meta_draw_pixels(STENCIL) */ +/** XXX compare perf of this vs. _mesa_meta_DrawPixels(STENCIL) */ static GLboolean intel_stencil_drawpixels(GLcontext * ctx, GLint x, GLint y, @@ -265,7 +265,7 @@ intelDrawPixels(GLcontext * ctx, /* XXX this function doesn't seem to work reliably even when all * the pre-requisite conditions are met. * Note that this function is never hit with conform. - * Fall back to swrast because even the _mesa_meta_draw_pixels() approach + * Fall back to swrast because even the _mesa_meta_DrawPixels() approach * isn't working because of an apparent stencil bug. */ if (intel_stencil_drawpixels(ctx, x, y, width, height, format, type, @@ -280,6 +280,6 @@ intelDrawPixels(GLcontext * ctx, } #endif - _mesa_meta_draw_pixels(ctx, x, y, width, height, format, type, - unpack, pixels); + _mesa_meta_DrawPixels(ctx, x, y, width, height, format, type, + unpack, pixels); } diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 1e73943457..28eabbc005 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -448,24 +448,22 @@ intel_map_unmap_framebuffer(struct intel_context *intel, struct gl_framebuffer *fb, GLboolean map) { - GLcontext *ctx = &intel->ctx; - GLuint i, j; - - /* color buffers */ - if (fb == ctx->DrawBuffer) { - for (j = 0; j < fb->_NumColorDrawBuffers; j++) { - if (map) - intel_renderbuffer_map(intel, fb->_ColorDrawBuffers[j]); - else - intel_renderbuffer_unmap(intel, fb->_ColorDrawBuffers[j]); - } - } else { + GLuint i; + + /* color draw buffers */ + for (i = 0; i < fb->_NumColorDrawBuffers; i++) { if (map) - intel_renderbuffer_map(intel, fb->_ColorReadBuffer); + intel_renderbuffer_map(intel, fb->_ColorDrawBuffers[i]); else - intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer); + intel_renderbuffer_unmap(intel, fb->_ColorDrawBuffers[i]); } + /* color read buffer */ + if (map) + intel_renderbuffer_map(intel, fb->_ColorReadBuffer); + else + intel_renderbuffer_unmap(intel, fb->_ColorReadBuffer); + /* check for render to textures */ for (i = 0; i < BUFFER_COUNT; i++) { struct gl_renderbuffer_attachment *att = @@ -484,20 +482,17 @@ intel_map_unmap_framebuffer(struct intel_context *intel, /* depth buffer (Note wrapper!) */ if (fb->_DepthBuffer) { if (map) - intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped); + intel_renderbuffer_map(intel, fb->_DepthBuffer->Wrapped); else - intel_renderbuffer_unmap(intel, - fb->_DepthBuffer->Wrapped); + intel_renderbuffer_unmap(intel, fb->_DepthBuffer->Wrapped); } /* stencil buffer (Note wrapper!) */ if (fb->_StencilBuffer) { if (map) - intel_renderbuffer_map(intel, - fb->_StencilBuffer->Wrapped); + intel_renderbuffer_map(intel, fb->_StencilBuffer->Wrapped); else - intel_renderbuffer_unmap(intel, - fb->_StencilBuffer->Wrapped); + intel_renderbuffer_unmap(intel, fb->_StencilBuffer->Wrapped); } } @@ -524,7 +519,8 @@ intelSpanRenderStart(GLcontext * ctx) } intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_TRUE); - intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE); + if (ctx->ReadBuffer != ctx->DrawBuffer) + intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_TRUE); } /** @@ -547,7 +543,8 @@ intelSpanRenderFinish(GLcontext * ctx) } intel_map_unmap_framebuffer(intel, ctx->DrawBuffer, GL_FALSE); - intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE); + if (ctx->ReadBuffer != ctx->DrawBuffer) + intel_map_unmap_framebuffer(intel, ctx->ReadBuffer, GL_FALSE); UNLOCK_HARDWARE(intel); } diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c index 028b49c14d..ac557a9200 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_copy.c +++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c @@ -30,7 +30,8 @@ #include "main/image.h" #include "main/teximage.h" #include "main/mipmap.h" -#include "swrast/swrast.h" + +#include "drivers/common/meta.h" #include "intel_screen.h" #include "intel_context.h" @@ -90,7 +91,6 @@ do_copy_texsubimage(struct intel_context *intel, GLint x, GLint y, GLsizei width, GLsizei height) { GLcontext *ctx = &intel->ctx; - struct gl_texture_object *texObj = intelImage->base.TexObject; const struct intel_region *src = get_teximage_source(intel, internalFormat); @@ -170,11 +170,6 @@ do_copy_texsubimage(struct intel_context *intel, UNLOCK_HARDWARE(intel); - /* GL_SGIS_generate_mipmap */ - if (intelImage->level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, texObj); - } - return GL_TRUE; } @@ -221,8 +216,8 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level, return; fail: - _swrast_copy_teximage1d(ctx, target, level, internalFormat, x, y, - width, border); + _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y, + width, border); } @@ -269,8 +264,8 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level, return; fail: - _swrast_copy_teximage2d(ctx, target, level, internalFormat, x, y, - width, height, border); + _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y, + width, height, border); } @@ -294,7 +289,7 @@ intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level, if (!do_copy_texsubimage(intel_context(ctx), target, intel_texture_image(texImage), internalFormat, xoffset, 0, x, y, width, 1)) { - _swrast_copy_texsubimage1d(ctx, target, level, xoffset, x, y, width); + _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width); } } @@ -320,10 +315,10 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level, internalFormat, xoffset, yoffset, x, y, width, height)) { - DBG("%s - fallback to swrast\n", __FUNCTION__); + DBG("%s - fallback to _mesa_meta_CopyTexSubImage2D\n", __FUNCTION__); - _swrast_copy_texsubimage2d(ctx, target, level, - xoffset, yoffset, x, y, width, height); + _mesa_meta_CopyTexSubImage2D(ctx, target, level, + xoffset, yoffset, x, y, width, height); } } diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c index a206fe6805..66201b1f46 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_image.c +++ b/src/mesa/drivers/dri/intel/intel_tex_image.c @@ -546,11 +546,6 @@ intelTexImage(GLcontext * ctx, } UNLOCK_HARDWARE(intel); - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, texObj); - } } diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c index 89037073f8..751ec2c98c 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_subimage.c +++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c @@ -129,11 +129,6 @@ intelTexSubimage(GLcontext * ctx, } UNLOCK_HARDWARE(intel); - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - intel_generate_mipmap(ctx, target, texObj); - } } diff --git a/src/mesa/drivers/dri/r200/Makefile b/src/mesa/drivers/dri/r200/Makefile index 42635bf9d9..fbce70c37b 100644 --- a/src/mesa/drivers/dri/r200/Makefile +++ b/src/mesa/drivers/dri/r200/Makefile @@ -55,8 +55,7 @@ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES) X86_SOURCES = -DRIVER_DEFINES = -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R200 \ - -Wall +DRIVER_DEFINES = -DRADEON_R200 -Wall DRI_LIB_DEPS += $(RADEON_LDFLAGS) diff --git a/src/mesa/drivers/dri/r300/Makefile b/src/mesa/drivers/dri/r300/Makefile index 5d8d6f6658..fe775eac99 100644 --- a/src/mesa/drivers/dri/r300/Makefile +++ b/src/mesa/drivers/dri/r300/Makefile @@ -62,8 +62,7 @@ DRIVER_SOURCES = \ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES) -DRIVER_DEFINES = -DCOMPILE_R300 -DR200_MERGED=0 \ - -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R300 \ +DRIVER_DEFINES = -DRADEON_R300 # -DRADEON_BO_TRACK \ -Wall diff --git a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c index 305dc074ee..c7227bbd15 100644 --- a/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c +++ b/src/mesa/drivers/dri/r300/compiler/r300_fragprog_emit.c @@ -352,7 +352,7 @@ void r300BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi if (emit.current_node < 3) { int shift = 3 - emit.current_node; int i; - for(i = 0; i <= emit.current_node; ++i) + for(i = emit.current_node; i >= 0; --i) code->code_addr[shift + i] = code->code_addr[i]; for(i = 0; i < shift; ++i) code->code_addr[i] = 0; diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c index 93a516105e..dad27fc98e 100644 --- a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c +++ b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c @@ -89,6 +89,7 @@ static unsigned long t_dst_index(struct r300_vertex_program_code *vp, static unsigned long t_src_class(gl_register_file file) { switch (file) { + case PROGRAM_BUILTIN: case PROGRAM_TEMPORARY: return PVS_SRC_REG_TEMPORARY; case PROGRAM_INPUT: diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.c b/src/mesa/drivers/dri/r300/compiler/radeon_program.c index bbbf0dd776..b636f90a96 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_program.c +++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.c @@ -150,14 +150,37 @@ void rc_mesa_to_rc_program(struct radeon_compiler * c, struct gl_program * progr c->Program.InputsRead = program->InputsRead; c->Program.OutputsWritten = program->OutputsWritten; - for(i = 0; i < program->Parameters->NumParameters; ++i) { - struct rc_constant constant; + int isNVProgram = 0; - constant.Type = RC_CONSTANT_EXTERNAL; - constant.Size = 4; - constant.u.External = i; + if (program->Target == GL_VERTEX_PROGRAM_ARB) { + struct gl_vertex_program * vp = (struct gl_vertex_program *) program; + isNVProgram = vp->IsNVProgram; + } + + if (isNVProgram) { + /* NV_vertex_program has a fixed-sized constant environment. + * This could be handled more efficiently for programs that + * do not use relative addressing. + */ + for(i = 0; i < 96; ++i) { + struct rc_constant constant; - rc_constants_add(&c->Program.Constants, &constant); + constant.Type = RC_CONSTANT_EXTERNAL; + constant.Size = 4; + constant.u.External = i; + + rc_constants_add(&c->Program.Constants, &constant); + } + } else { + for(i = 0; i < program->Parameters->NumParameters; ++i) { + struct rc_constant constant; + + constant.Type = RC_CONSTANT_EXTERNAL; + constant.Size = 4; + constant.u.External = i; + + rc_constants_add(&c->Program.Constants, &constant); + } } } diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c index 8071899eaa..f23ce301ca 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c +++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c @@ -511,23 +511,23 @@ static void sincos_constants(struct radeon_compiler* c, GLuint *constants) * MAD dest, tmp.y, weight, tmp.x */ static void sin_approx( - struct radeon_compiler* c, struct rc_instruction * after, + struct radeon_compiler* c, struct rc_instruction * before, struct prog_dst_register dst, struct prog_src_register src, const GLuint* constants) { GLuint tempreg = rc_find_free_temporary(c); - emit2(c, after->Prev, OPCODE_MUL, 0, dstregtmpmask(tempreg, WRITEMASK_XY), + emit2(c, before->Prev, OPCODE_MUL, 0, dstregtmpmask(tempreg, WRITEMASK_XY), swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), srcreg(PROGRAM_CONSTANT, constants[0])); - emit3(c, after->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_X), + emit3(c, before->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_X), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), absolute(swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)); - emit3(c, after->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_Y), + emit3(c, before->Prev, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_Y), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), absolute(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)), negate(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X))); - emit3(c, after->Prev, OPCODE_MAD, 0, dst, + emit3(c, before->Prev, OPCODE_MAD, 0, dst, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)); @@ -567,7 +567,7 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c, swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z))); - sin_approx(c, inst->Prev, inst->I.DstReg, + sin_approx(c, inst, inst->I.DstReg, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), constants); } else if (inst->I.Opcode == OPCODE_SIN) { @@ -582,7 +582,7 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c, swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z))); - sin_approx(c, inst->Prev, inst->I.DstReg, + sin_approx(c, inst, inst->I.DstReg, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W), constants); } else { @@ -600,12 +600,12 @@ GLboolean radeonTransformTrigSimple(struct radeon_compiler* c, struct prog_dst_register dst = inst->I.DstReg; dst.WriteMask = inst->I.DstReg.WriteMask & WRITEMASK_X; - sin_approx(c, inst->Prev, dst, + sin_approx(c, inst, dst, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X), constants); dst.WriteMask = inst->I.DstReg.WriteMask & WRITEMASK_Y; - sin_approx(c, inst->Prev, dst, + sin_approx(c, inst, dst, swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y), constants); } diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_common.c b/src/mesa/drivers/dri/r300/r300_fragprog_common.c index 469c278b51..0bdc90b4a8 100644 --- a/src/mesa/drivers/dri/r300/r300_fragprog_common.c +++ b/src/mesa/drivers/dri/r300/r300_fragprog_common.c @@ -99,8 +99,8 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler, { int i; + fp->wpos_attr = FRAG_ATTRIB_MAX; if (!(compiler->Base.Program.InputsRead & FRAG_BIT_WPOS)) { - fp->wpos_attr = FRAG_ATTRIB_MAX; return; } @@ -112,6 +112,13 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler, } } + /* No free texcoord found, fall-back to software rendering */ + if (fp->wpos_attr == FRAG_ATTRIB_MAX) + { + compiler->Base.Error = 1; + return; + } + rc_transform_fragment_wpos(&compiler->Base, FRAG_ATTRIB_WPOS, fp->wpos_attr); } @@ -127,8 +134,8 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r struct prog_src_register src; int i; + fp->fog_attr = FRAG_ATTRIB_MAX; if (!(compiler->Base.Program.InputsRead & FRAG_BIT_FOGC)) { - fp->fog_attr = FRAG_ATTRIB_MAX; return; } @@ -140,6 +147,13 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r } } + /* No free texcoord found, fall-back to software rendering */ + if (fp->fog_attr == FRAG_ATTRIB_MAX) + { + compiler->Base.Error = 1; + return; + } + memset(&src, 0, sizeof(src)); src.File = PROGRAM_INPUT; src.Index = fp->fog_attr; diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c index b5ddfdc9f8..3cd38753b8 100644 --- a/src/mesa/drivers/dri/r300/r300_render.c +++ b/src/mesa/drivers/dri/r300/r300_render.c @@ -475,7 +475,7 @@ void r300SwitchFallback(GLcontext *ctx, uint32_t bit, GLboolean mode) /* update only if we have disabled all tcl fallbacks */ if (rmesa->options.hw_tcl_enabled) { - if ((old_fallback & R300_RASTER_FALLBACK_MASK) == bit) { + if ((old_fallback & R300_TCL_FALLBACK_MASK) == bit) { R300_STATECHANGE(rmesa, vap_cntl_status); rmesa->hw.vap_cntl_status.cmd[1] &= ~R300_VAP_TCL_BYPASS; } diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c index dd0f27f9cb..2f7b67c143 100644 --- a/src/mesa/drivers/dri/r300/r300_vertprog.c +++ b/src/mesa/drivers/dri/r300/r300_vertprog.c @@ -43,6 +43,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. #include "compiler/radeon_compiler.h" #include "compiler/radeon_nqssadce.h" #include "r300_context.h" +#include "r300_fragprog_common.h" #include "r300_state.h" /** @@ -203,6 +204,34 @@ static void t_inputs_outputs(struct r300_vertex_program_compiler * c) } } +/** + * The NV_vertex_program spec mandates that all registers be + * initialized to zero. We do this here unconditionally. + * + * \note We rely on dead-code elimination in the compiler. + */ +static void initialize_NV_registers(struct radeon_compiler * compiler) +{ + unsigned int reg; + struct rc_instruction * inst; + + for(reg = 0; reg < 12; ++reg) { + inst = rc_insert_new_instruction(compiler, &compiler->Program.Instructions); + inst->I.Opcode = OPCODE_MOV; + inst->I.DstReg.File = PROGRAM_TEMPORARY; + inst->I.DstReg.Index = reg; + inst->I.SrcReg[0].File = PROGRAM_BUILTIN; + inst->I.SrcReg[0].Swizzle = SWIZZLE_0000; + } + + inst = rc_insert_new_instruction(compiler, &compiler->Program.Instructions); + inst->I.Opcode = OPCODE_ARL; + inst->I.DstReg.File = PROGRAM_ADDRESS; + inst->I.DstReg.Index = 0; + inst->I.DstReg.WriteMask = WRITEMASK_X; + inst->I.SrcReg[0].File = PROGRAM_BUILTIN; + inst->I.SrcReg[0].Swizzle = SWIZZLE_0000; +} static struct r300_vertex_program *build_program(GLcontext *ctx, struct r300_vertex_program_key *wanted_key, @@ -234,6 +263,9 @@ static struct r300_vertex_program *build_program(GLcontext *ctx, rc_mesa_to_rc_program(&compiler.Base, &vp->Base->Base); + if (mesa_vp->IsNVProgram) + initialize_NV_registers(&compiler.Base); + rc_move_output(&compiler.Base, VERT_RESULT_PSIZ, VERT_RESULT_PSIZ, WRITEMASK_X); if (vp->key.WPosAttr != FRAG_ATTRIB_MAX) { @@ -267,6 +299,20 @@ struct r300_vertex_program * r300SelectAndTranslateVertexShader(GLcontext *ctx) struct r300_vertex_program *vp; vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current; + + if (!r300->selected_fp) { + /* This can happen when GetProgramiv is called to check + * whether the program runs natively. + * + * To be honest, this is not a very good solution, + * but solving the problem of reporting good values + * for those queries is tough anyway considering that + * we recompile vertex programs based on the precise + * fragment program that is in use. + */ + r300SelectAndTranslateFragmentShader(ctx); + } + wanted_key.FpReads = r300->selected_fp->InputsRead; wanted_key.FogAttr = r300->selected_fp->fog_attr; wanted_key.WPosAttr = r300->selected_fp->wpos_attr; diff --git a/src/mesa/drivers/dri/r300/radeon_context.h b/src/mesa/drivers/dri/r300/radeon_context.h index 250570f6b8..da4812d323 100644 --- a/src/mesa/drivers/dri/r300/radeon_context.h +++ b/src/mesa/drivers/dri/r300/radeon_context.h @@ -51,26 +51,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_screen.h" -#if R200_MERGED -extern void radeonFallback(GLcontext * ctx, GLuint bit, GLboolean mode); - -#define FALLBACK( radeon, bit, mode ) do { \ - if ( 0 ) fprintf( stderr, "FALLBACK in %s: #%d=%d\n", \ - __FUNCTION__, bit, mode ); \ - radeonFallback( (radeon)->glCtx, bit, mode ); \ -} while (0) -#else #define FALLBACK( radeon, bit, mode ) fprintf(stderr, "%s:%s\n", __LINE__, __FILE__); -#endif /* TCL fallbacks */ extern void radeonTclFallback(GLcontext * ctx, GLuint bit, GLboolean mode); -#if R200_MERGED -#define TCL_FALLBACK( ctx, bit, mode ) radeonTclFallback( ctx, bit, mode ) -#else #define TCL_FALLBACK( ctx, bit, mode ) ; -#endif #endif /* __RADEON_CONTEXT_H__ */ diff --git a/src/mesa/drivers/dri/r600/Makefile b/src/mesa/drivers/dri/r600/Makefile index d925a2dfe3..7d5a7b1ab6 100644 --- a/src/mesa/drivers/dri/r600/Makefile +++ b/src/mesa/drivers/dri/r600/Makefile @@ -29,6 +29,7 @@ COMMON_SOURCES = \ RADEON_COMMON_SOURCES = \ radeon_bo_legacy.c \ radeon_common_context.c \ + radeon_buffer_objects.c \ radeon_common.c \ radeon_cs_legacy.c \ radeon_dma.c \ @@ -65,8 +66,7 @@ DRIVER_SOURCES = \ C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES) -DRIVER_DEFINES = -DCOMPILE_R600 -DR200_MERGED=0 \ - -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R600 \ +DRIVER_DEFINES = -DRADEON_R600 \ # -DRADEON_BO_TRACK \ -Wall diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c index 07a7bcf11f..969144ba12 100644 --- a/src/mesa/drivers/dri/r600/r600_context.c +++ b/src/mesa/drivers/dri/r600/r600_context.c @@ -59,6 +59,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_debug.h" #include "r600_context.h" #include "radeon_common_context.h" +#include "radeon_buffer_objects.h" #include "radeon_span.h" #include "r600_cmdbuf.h" #include "r600_emit.h" @@ -257,6 +258,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, r600InitTextureFuncs(&functions); r700InitShaderFuncs(&functions); r700InitIoctlFuncs(&functions); + radeonInitBufferObjectFuncs(&functions); if (!radeonInitContext(&r600->radeon, &functions, glVisual, driContextPriv, @@ -284,8 +286,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, ctx->Const.MaxTextureMaxAnisotropy = 16.0; ctx->Const.MaxTextureLodBias = 16.0; - ctx->Const.MaxTextureLevels = 13; - ctx->Const.MaxTextureRectSize = 4096; + ctx->Const.MaxTextureLevels = 13; /* hw support 14 */ + ctx->Const.MaxTextureRectSize = 4096; /* hw support 8192 */ ctx->Const.MinPointSize = 0x0001 / 8.0; ctx->Const.MinPointSizeAA = 0x0001 / 8.0; @@ -330,26 +332,27 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, _tnl_allow_pixel_fog(ctx, GL_FALSE); _tnl_allow_vertex_fog(ctx, GL_TRUE); - /* currently bogus data */ - ctx->Const.VertexProgram.MaxInstructions = VSF_MAX_FRAGMENT_LENGTH / 4; - ctx->Const.VertexProgram.MaxNativeInstructions = - VSF_MAX_FRAGMENT_LENGTH / 4; - ctx->Const.VertexProgram.MaxNativeAttribs = 16; /* r420 */ - ctx->Const.VertexProgram.MaxTemps = 32; - ctx->Const.VertexProgram.MaxNativeTemps = - /*VSF_MAX_FRAGMENT_TEMPS */ 32; - ctx->Const.VertexProgram.MaxNativeParameters = 256; /* r420 */ - ctx->Const.VertexProgram.MaxNativeAddressRegs = 1; - - ctx->Const.FragmentProgram.MaxNativeTemps = PFS_NUM_TEMP_REGS; - ctx->Const.FragmentProgram.MaxNativeAttribs = 11; /* copy i915... */ - ctx->Const.FragmentProgram.MaxNativeParameters = PFS_NUM_CONST_REGS; - ctx->Const.FragmentProgram.MaxNativeAluInstructions = PFS_MAX_ALU_INST; - ctx->Const.FragmentProgram.MaxNativeTexInstructions = PFS_MAX_TEX_INST; - ctx->Const.FragmentProgram.MaxNativeInstructions = - PFS_MAX_ALU_INST + PFS_MAX_TEX_INST; - ctx->Const.FragmentProgram.MaxNativeTexIndirections = - PFS_MAX_TEX_INDIRECT; + /* 256 for reg-based consts, inline consts also supported */ + ctx->Const.VertexProgram.MaxInstructions = 8192; /* in theory no limit */ + ctx->Const.VertexProgram.MaxNativeInstructions = 8192; + ctx->Const.VertexProgram.MaxNativeAttribs = 160; + ctx->Const.VertexProgram.MaxTemps = 128; + ctx->Const.VertexProgram.MaxNativeTemps = 128; + ctx->Const.VertexProgram.MaxNativeParameters = 256; + ctx->Const.VertexProgram.MaxNativeAddressRegs = 1; /* ??? */ + + ctx->Const.FragmentProgram.MaxNativeTemps = 128; + ctx->Const.FragmentProgram.MaxNativeAttribs = 32; + ctx->Const.FragmentProgram.MaxNativeParameters = 256; + ctx->Const.FragmentProgram.MaxNativeAluInstructions = 8192; + /* 8 per clause on r6xx, 16 on rv670/r7xx */ + if ((screen->chip_family == CHIP_FAMILY_RV670) || + (screen->chip_family >= CHIP_FAMILY_RV770)) + ctx->Const.FragmentProgram.MaxNativeTexInstructions = 16; + else + ctx->Const.FragmentProgram.MaxNativeTexInstructions = 8; + ctx->Const.FragmentProgram.MaxNativeInstructions = 8192; + ctx->Const.FragmentProgram.MaxNativeTexIndirections = 8; /* ??? */ ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0; /* and these are?? */ ctx->VertexProgram._MaintainTnlProgram = GL_TRUE; ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE; @@ -374,6 +377,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc"); } + r700InitDraw(ctx); + radeon_fbo_init(&r600->radeon); radeonInitSpanFuncs( ctx ); @@ -385,9 +390,6 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, if (driQueryOptionb(&r600->radeon.optionCache, "no_rast")) { radeon_warning("disabling 3D acceleration\n"); -#if R200_MERGED - FALLBACK(&r600->radeon, RADEON_FALLBACK_DISABLE, 1); -#endif } return GL_TRUE; diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h index 8ae05a301c..a296ea23fa 100644 --- a/src/mesa/drivers/dri/r600/r600_context.h +++ b/src/mesa/drivers/dri/r600/r600_context.h @@ -51,6 +51,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "r700_chip.h" #include "r600_tex.h" #include "r700_oglprog.h" +#include "r700_vertprog.h" struct r600_context; typedef struct r600_context context_t; @@ -85,29 +86,10 @@ extern int hw_tcl_on; #include "tnl_dd/t_dd_vertex.h" #undef TAG -#define PFS_MAX_ALU_INST 64 -#define PFS_MAX_TEX_INST 64 -#define PFS_MAX_TEX_INDIRECT 4 -#define PFS_NUM_TEMP_REGS 32 -#define PFS_NUM_CONST_REGS 16 - -#define R600_MAX_AOS_ARRAYS 16 - -#define REG_COORDS 0 -#define REG_COLOR0 1 -#define REG_TEX0 2 - #define R600_FALLBACK_NONE 0 #define R600_FALLBACK_TCL 1 #define R600_FALLBACK_RAST 2 -enum -{ - NO_SHIFT = 0, - LEFT_SHIFT = 1, - RIGHT_SHIFT = 2, -}; - struct r600_hw_state { struct radeon_state_atom sq; struct radeon_state_atom db; @@ -144,6 +126,34 @@ struct r600_hw_state { struct radeon_state_atom tx_brdr_clr; }; +typedef struct StreamDesc +{ + GLint size; //number of data element + GLenum type; //data element type + GLsizei stride; + + struct radeon_bo *bo; + GLint bo_offset; + + GLuint dwords; + GLuint dst_loc; + GLuint _signed; + GLboolean normalize; + GLboolean is_named_bo; + GLubyte element; +} StreamDesc; + +typedef struct r700_index_buffer +{ + struct radeon_bo *bo; + int bo_offset; + + GLboolean is_32bit; + GLuint count; + + GLboolean bHostIb; +} r700_index_buffer; + /** * \brief R600 context structure. */ @@ -155,11 +165,16 @@ struct r600_context { struct r600_hw_state atoms; + struct r700_vertex_program *selected_vp; + /* Vertex buffers */ GLvector4f dummy_attrib[_TNL_ATTRIB_MAX]; GLvector4f *temp_attrib[_TNL_ATTRIB_MAX]; + GLint nNumActiveAos; + StreamDesc stream_desc[VERT_ATTRIB_MAX]; + struct r700_index_buffer ind_buf; }; #define R700_CONTEXT(ctx) ((context_t *)(ctx->DriverCtx)) @@ -193,6 +208,7 @@ extern GLboolean r700SyncSurf(context_t *context, extern void r700SetupStreams(GLcontext * ctx); extern void r700Start3D(context_t *context); extern void r600InitAtoms(context_t *context); +extern void r700InitDraw(GLcontext *ctx); #define RADEON_D_CAPTURE 0 #define RADEON_D_PLAYBACK 1 diff --git a/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h b/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h index 9d5aa3c7e4..edd85b0fac 100644 --- a/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h +++ b/src/mesa/drivers/dri/r600/r600_reg_auto_r6xx.h @@ -1366,6 +1366,7 @@ enum { DB_DEPTH_INFO__READ_SIZE_bit = 1 << 3, DB_DEPTH_INFO__ARRAY_MODE_mask = 0x0f << 15, DB_DEPTH_INFO__ARRAY_MODE_shift = 15, + ARRAY_1D_TILED_THIN1 = 0x02, ARRAY_2D_TILED_THIN1 = 0x04, TILE_SURFACE_ENABLE_bit = 1 << 25, TILE_COMPACT_bit = 1 << 26, @@ -1449,6 +1450,7 @@ enum { CB_COLOR0_INFO__ARRAY_MODE_shift = 8, ARRAY_LINEAR_GENERAL = 0x00, ARRAY_LINEAR_ALIGNED = 0x01, +/* ARRAY_1D_TILED_THIN1 = 0x02, */ /* ARRAY_2D_TILED_THIN1 = 0x04, */ NUMBER_TYPE_mask = 0x07 << 12, NUMBER_TYPE_shift = 12, diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c index 2d8480daaf..903b6968be 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.c +++ b/src/mesa/drivers/dri/r600/r700_assembler.c @@ -213,7 +213,7 @@ GLboolean is_reduction_opcode(PVSDWORD* dest) { if (dest->dst.op3 == 0) { - if ( (dest->dst.opcode == SQ_OP2_INST_DOT4 || dest->dst.opcode == SQ_OP2_INST_DOT4_IEEE) ) + if ( (dest->dst.opcode == SQ_OP2_INST_DOT4 || dest->dst.opcode == SQ_OP2_INST_DOT4_IEEE || dest->dst.opcode == SQ_OP2_INST_CUBE) ) { return GL_TRUE; } @@ -350,6 +350,7 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm) case SQ_OP2_INST_PRED_SETNE: case SQ_OP2_INST_DOT4: case SQ_OP2_INST_DOT4_IEEE: + case SQ_OP2_INST_CUBE: return 2; case SQ_OP2_INST_MOV: @@ -469,6 +470,9 @@ int Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700 pAsm->number_of_inputs = 0; + pAsm->is_tex = GL_FALSE; + pAsm->need_tex_barrier = GL_FALSE; + return 0; } @@ -682,7 +686,7 @@ GLboolean add_tex_instruction(r700_AssemblerBase* pAsm, // If this clause constains any TEX instruction that is dependent on a previous instruction, // set the barrier bit - if( pAsm->pInstDeps[pAsm->uiCurInst].nDstDep > (-1) ) + if( pAsm->pInstDeps[pAsm->uiCurInst].nDstDep > (-1) || pAsm->need_tex_barrier == GL_TRUE ) { pAsm->cf_current_tex_clause_ptr->m_Word1.f.barrier = 0x1; } @@ -786,6 +790,133 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm, return GL_TRUE; } +GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm, + GLuint destination_register, + GLenum type, + GLint size, + GLubyte element, + GLuint _signed, + GLboolean normalize, + VTX_FETCH_METHOD * pFetchMethod) +{ + GLuint client_size_inbyte; + GLuint data_format; + GLuint mega_fetch_count; + GLuint is_mega_fetch_flag; + + R700VertexGenericFetch* vfetch_instruction_ptr; + R700VertexGenericFetch* assembled_vfetch_instruction_ptr + = pAsm->vfetch_instruction_ptr_array[element]; + + if (assembled_vfetch_instruction_ptr == NULL) + { + vfetch_instruction_ptr = (R700VertexGenericFetch*) CALLOC_STRUCT(R700VertexGenericFetch); + if (vfetch_instruction_ptr == NULL) + { + return GL_FALSE; + } + Init_R700VertexGenericFetch(vfetch_instruction_ptr); + } + else + { + vfetch_instruction_ptr = assembled_vfetch_instruction_ptr; + } + + data_format = GetSurfaceFormat(type, size, &client_size_inbyte); + + if(GL_TRUE == pFetchMethod->bEnableMini) //More conditions here + { + //TODO : mini fetch + } + else + { + mega_fetch_count = MEGA_FETCH_BYTES - 1; + is_mega_fetch_flag = 0x1; + pFetchMethod->mega_fetch_remainder = MEGA_FETCH_BYTES - client_size_inbyte; + } + + vfetch_instruction_ptr->m_Word0.f.vtx_inst = SQ_VTX_INST_FETCH; + vfetch_instruction_ptr->m_Word0.f.fetch_type = SQ_VTX_FETCH_VERTEX_DATA; + vfetch_instruction_ptr->m_Word0.f.fetch_whole_quad = 0x0; + + vfetch_instruction_ptr->m_Word0.f.buffer_id = element; + vfetch_instruction_ptr->m_Word0.f.src_gpr = 0x0; + vfetch_instruction_ptr->m_Word0.f.src_rel = SQ_ABSOLUTE; + vfetch_instruction_ptr->m_Word0.f.src_sel_x = SQ_SEL_X; + vfetch_instruction_ptr->m_Word0.f.mega_fetch_count = mega_fetch_count; + + vfetch_instruction_ptr->m_Word1.f.dst_sel_x = (size < 1) ? SQ_SEL_0 : SQ_SEL_X; + vfetch_instruction_ptr->m_Word1.f.dst_sel_y = (size < 2) ? SQ_SEL_0 : SQ_SEL_Y; + vfetch_instruction_ptr->m_Word1.f.dst_sel_z = (size < 3) ? SQ_SEL_0 : SQ_SEL_Z; + vfetch_instruction_ptr->m_Word1.f.dst_sel_w = (size < 4) ? SQ_SEL_1 : SQ_SEL_W; + + vfetch_instruction_ptr->m_Word1.f.use_const_fields = 1; + vfetch_instruction_ptr->m_Word1.f.data_format = data_format; + vfetch_instruction_ptr->m_Word2.f.endian_swap = SQ_ENDIAN_NONE; + + if(1 == _signed) + { + vfetch_instruction_ptr->m_Word1.f.format_comp_all = SQ_FORMAT_COMP_SIGNED; + } + else + { + vfetch_instruction_ptr->m_Word1.f.format_comp_all = SQ_FORMAT_COMP_UNSIGNED; + } + + if(GL_TRUE == normalize) + { + vfetch_instruction_ptr->m_Word1.f.num_format_all = SQ_NUM_FORMAT_NORM; + } + else + { + vfetch_instruction_ptr->m_Word1.f.num_format_all = SQ_NUM_FORMAT_INT; + } + + // Destination register + vfetch_instruction_ptr->m_Word1_GPR.f.dst_gpr = destination_register; + vfetch_instruction_ptr->m_Word1_GPR.f.dst_rel = SQ_ABSOLUTE; + + vfetch_instruction_ptr->m_Word2.f.offset = 0; + vfetch_instruction_ptr->m_Word2.f.const_buf_no_stride = 0x0; + + vfetch_instruction_ptr->m_Word2.f.mega_fetch = is_mega_fetch_flag; + + if (assembled_vfetch_instruction_ptr == NULL) + { + if ( GL_FALSE == add_vfetch_instruction(pAsm, (R700VertexInstruction *)vfetch_instruction_ptr) ) + { + return GL_FALSE; + } + + if (pAsm->vfetch_instruction_ptr_array[element] != NULL) + { + return GL_FALSE; + } + else + { + pAsm->vfetch_instruction_ptr_array[element] = vfetch_instruction_ptr; + } + } + + return GL_TRUE; +} + +GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm) +{ + GLint i; + pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE; + pAsm->cf_current_vtx_clause_ptr = NULL; + + for (i=0; i<VERT_ATTRIB_MAX; i++) + { + pAsm->vfetch_instruction_ptr_array[ i ] = NULL; + } + + cleanup_vfetch_shaderinst(pAsm->pR700Shader); + + return GL_TRUE; +} + GLuint gethelpr(r700_AssemblerBase* pAsm) { GLuint r = pAsm->uHelpReg; @@ -1149,41 +1280,55 @@ GLboolean tex_dst(r700_AssemblerBase *pAsm) GLboolean tex_src(r700_AssemblerBase *pAsm) { struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]); - + GLboolean bValidTexCoord = GL_FALSE; - switch (pILInst->SrcReg[0].File) + if(pAsm->aArgSubst[1] >= 0) { - case PROGRAM_TEMPORARY: bValidTexCoord = GL_TRUE; - - pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + pAsm->starting_temp_register_number; + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; - - break; - case PROGRAM_INPUT: - switch (pILInst->SrcReg[0].Index) - { - case FRAG_ATTRIB_COL0: - case FRAG_ATTRIB_COL1: - case FRAG_ATTRIB_TEX0: - case FRAG_ATTRIB_TEX1: - case FRAG_ATTRIB_TEX2: - case FRAG_ATTRIB_TEX3: - case FRAG_ATTRIB_TEX4: - case FRAG_ATTRIB_TEX5: - case FRAG_ATTRIB_TEX6: - case FRAG_ATTRIB_TEX7: + pAsm->S[0].src.reg = pAsm->aArgSubst[1]; + } + else + { + switch (pILInst->SrcReg[0].File) { + case PROGRAM_CONSTANT: + case PROGRAM_LOCAL_PARAM: + case PROGRAM_ENV_PARAM: + case PROGRAM_STATE_VAR: + break; + case PROGRAM_TEMPORARY: bValidTexCoord = GL_TRUE; - - pAsm->S[0].src.reg = pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; - pAsm->S[0].src.rtype = SRC_REG_INPUT; - } + pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + + pAsm->starting_temp_register_number; + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + break; + case PROGRAM_INPUT: + switch (pILInst->SrcReg[0].Index) + { + case FRAG_ATTRIB_COL0: + case FRAG_ATTRIB_COL1: + case FRAG_ATTRIB_TEX0: + case FRAG_ATTRIB_TEX1: + case FRAG_ATTRIB_TEX2: + case FRAG_ATTRIB_TEX3: + case FRAG_ATTRIB_TEX4: + case FRAG_ATTRIB_TEX5: + case FRAG_ATTRIB_TEX6: + case FRAG_ATTRIB_TEX7: + bValidTexCoord = GL_TRUE; + pAsm->S[0].src.reg = + pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; + pAsm->S[0].src.rtype = SRC_REG_INPUT; + break; + } break; + } } if(GL_TRUE == bValidTexCoord) - { + { setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); } else @@ -1201,7 +1346,7 @@ GLboolean tex_src(r700_AssemblerBase *pAsm) pAsm->S[0].src.negy = (pILInst->SrcReg[0].Negate >> 1) & 0x1; pAsm->S[0].src.negz = (pILInst->SrcReg[0].Negate >> 2) & 0x1; pAsm->S[0].src.negw = (pILInst->SrcReg[0].Negate >> 3) & 0x1; - + return GL_TRUE; } @@ -1933,9 +2078,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) GLuint contiguous_slots_needed; GLuint uNumSrc = r700GetNumOperands(pAsm); - GLuint channel_swizzle, j; - GLuint chan_counter[4] = {0, 0, 0, 0}; - PVSSRC * pSource[3]; + //GLuint channel_swizzle, j; + //GLuint chan_counter[4] = {0, 0, 0, 0}; + //PVSSRC * pSource[3]; GLboolean bSplitInst = GL_FALSE; if (1 == pAsm->D.dst.math) @@ -1947,7 +2092,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) { is_single_scalar_operation = GL_FALSE; number_of_scalar_operations = 4; - + +/* current assembler doesn't do more than 1 register per source */ +#if 0 /* check read port, only very preliminary algorithm, not count in src0/1 same comp case and prev slot repeat case; also not count relative addressing. TODO: improve performance. */ @@ -1982,6 +2129,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) { bSplitInst = GL_TRUE; } +#endif } contiguous_slots_needed = 0; @@ -2016,7 +2164,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) return GL_FALSE; } - if (pAsm->D.dst.math == 0) + if (uNumSrc > 1) { // Process source 1 current_source_index = 1; @@ -2202,7 +2350,7 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) { struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]); - if( GL_TRUE == IsTex(pILInst->Opcode) ) + if( GL_TRUE == pAsm->is_tex ) { if (pILInst->TexSrcTarget == TEXTURE_RECT_INDEX) { if( GL_FALSE == assemble_tex_instruction(pAsm, GL_FALSE) ) @@ -2246,7 +2394,8 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) pAsm->S[0].bits = 0; pAsm->S[1].bits = 0; pAsm->S[2].bits = 0; - + pAsm->is_tex = GL_FALSE; + pAsm->need_tex_barrier = GL_FALSE; return GL_TRUE; } @@ -2870,6 +3019,11 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) return GL_FALSE; } + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + /* dst.y = max(src.x, 0.0) */ pAsm->D.dst.opcode = SQ_OP2_INST_MAX; pAsm->D.dst.rtype = dstType; @@ -2881,11 +3035,6 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_X; - pAsm->S[0].src.swizzley = SQ_SEL_X; - pAsm->S[0].src.swizzlez = SQ_SEL_X; - pAsm->S[0].src.swizzlew = SQ_SEL_X; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; pAsm->S[1].src.reg = tmp; setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); @@ -2899,34 +3048,47 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) return GL_FALSE; } - /* before: dst.w = log(src.y) - * after : dst.x = log(src.y) - * why change dest register is that dst.w has been initialized as 1 before - */ + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_Y, SQ_SEL_Y, SQ_SEL_Y, SQ_SEL_Y); + + /* dst.z = log(src.y) */ pAsm->D.dst.opcode = SQ_OP2_INST_LOG_CLAMPED; pAsm->D.dst.math = 1; pAsm->D.dst.rtype = dstType; pAsm->D.dst.reg = dstReg; - pAsm->D.dst.writex = 1; + pAsm->D.dst.writex = 0; pAsm->D.dst.writey = 0; - pAsm->D.dst.writez = 0; + pAsm->D.dst.writez = 1; pAsm->D.dst.writew = 0; pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_Y; - pAsm->S[0].src.swizzley = SQ_SEL_Y; - pAsm->S[0].src.swizzlez = SQ_SEL_Y; - pAsm->S[0].src.swizzlew = SQ_SEL_Y; if( GL_FALSE == next_ins(pAsm) ) { return GL_FALSE; } - /* before: tmp.x = amd MUL_LIT(src.w, dst.w, src.x ) */ - /* after : tmp.x = amd MUL_LIT(src.w, dst.x, src.x ) */ + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + if( GL_FALSE == assemble_src(pAsm, 0, 2) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_W, SQ_SEL_W, SQ_SEL_W, SQ_SEL_W); + + swizzleagain_PVSSRC(&(pAsm->S[2].src), SQ_SEL_X, SQ_SEL_X, SQ_SEL_X, SQ_SEL_X); + + /* tmp.x = amd MUL_LIT(src.w, dst.z, src.x ) */ pAsm->D.dst.opcode = SQ_OP3_INST_MUL_LIT; + pAsm->D.dst.math = 1; pAsm->D.dst.op3 = 1; pAsm->D.dst.rtype = DST_REG_TEMPORARY; pAsm->D.dst.reg = tmp; @@ -2938,29 +3100,19 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_W; - pAsm->S[0].src.swizzley = SQ_SEL_W; - pAsm->S[0].src.swizzlez = SQ_SEL_W; - pAsm->S[0].src.swizzlew = SQ_SEL_W; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; pAsm->S[1].src.reg = dstReg; setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); noneg_PVSSRC(&(pAsm->S[1].src)); - pAsm->S[1].src.swizzlex = SQ_SEL_X; - pAsm->S[1].src.swizzley = SQ_SEL_X; - pAsm->S[1].src.swizzlez = SQ_SEL_X; - pAsm->S[1].src.swizzlew = SQ_SEL_X; + pAsm->S[1].src.swizzlex = SQ_SEL_Z; + pAsm->S[1].src.swizzley = SQ_SEL_Z; + pAsm->S[1].src.swizzlez = SQ_SEL_Z; + pAsm->S[1].src.swizzlew = SQ_SEL_Z; pAsm->S[2].src.rtype = srcType; pAsm->S[2].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[2].src)); - pAsm->S[2].src.swizzlex = SQ_SEL_X; - pAsm->S[2].src.swizzley = SQ_SEL_X; - pAsm->S[2].src.swizzlez = SQ_SEL_X; - pAsm->S[2].src.swizzlew = SQ_SEL_X; if( GL_FALSE == next_ins(pAsm) ) { @@ -3366,7 +3518,10 @@ GLboolean assemble_STP(r700_AssemblerBase *pAsm) GLboolean assemble_TEX(r700_AssemblerBase *pAsm) { GLboolean src_const; + GLboolean need_barrier = GL_FALSE; + checkop1(pAsm); + switch (pAsm->pILInst[pAsm->uiCurInst].SrcReg[0].File) { case PROGRAM_CONSTANT: @@ -3374,29 +3529,30 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) case PROGRAM_ENV_PARAM: case PROGRAM_STATE_VAR: src_const = GL_TRUE; + break; case PROGRAM_TEMPORARY: case PROGRAM_INPUT: + default: src_const = GL_FALSE; + break; } - if (GL_TRUE == src_const) + if (GL_TRUE == src_const) { - radeon_error("TODO: Texture coordinates from a constant register not supported.\n"); - return GL_FALSE; + if ( GL_FALSE == mov_temp(pAsm, 0) ) + return GL_FALSE; + need_barrier = GL_TRUE; } - switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) + switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) { case OPCODE_TEX: - pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; break; - case OPCODE_TXB: + case OPCODE_TXB: radeon_error("do not support TXB yet\n"); return GL_FALSE; break; - case OPCODE_TXP: - /* TODO : tex proj version : divid first 3 components by 4th */ - pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; + case OPCODE_TXP: break; default: radeon_error("Internal error: bad texture op (not TEX)\n"); @@ -3404,6 +3560,190 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) break; } + if (pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP) + { + GLuint tmp = gethelpr(pAsm); + pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE; + pAsm->D.dst.math = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp; + pAsm->D.dst.writew = 1; + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_W, SQ_SEL_W, SQ_SEL_W, SQ_SEL_W); + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + pAsm->D.dst.opcode = SQ_OP2_INST_MUL; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 1; + pAsm->D.dst.writew = 0; + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = tmp; + setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_W); + + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + pAsm->aArgSubst[1] = tmp; + need_barrier = GL_TRUE; + } + + if (pAsm->pILInst[pAsm->uiCurInst].TexSrcTarget == TEXTURE_CUBE_INDEX ) + { + GLuint tmp1 = gethelpr(pAsm); + GLuint tmp2 = gethelpr(pAsm); + + /* tmp1.xyzw = CUBE(R0.zzxy, R0.yxzz) */ + pAsm->D.dst.opcode = SQ_OP2_INST_CUBE; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + nomask_PVSDST(&(pAsm->D.dst)); + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + if( GL_FALSE == assemble_src(pAsm, 0, 1) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_Z, SQ_SEL_Z, SQ_SEL_X, SQ_SEL_Y); + swizzleagain_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Y, SQ_SEL_X, SQ_SEL_Z, SQ_SEL_Z); + + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + /* tmp1.z = ABS(tmp1.z) dont have abs support in assembler currently + * have to do explicit instruction + */ + pAsm->D.dst.opcode = SQ_OP2_INST_MAX; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writez = 1; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + pAsm->S[1].bits = pAsm->S[0].bits; + flipneg_PVSSRC(&(pAsm->S[1].src)); + + next_ins(pAsm); + + /* tmp1.z = RCP_e(|tmp1.z|) */ + pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE; + pAsm->D.dst.math = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writez = 1; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + pAsm->S[0].src.swizzlex = SQ_SEL_Z; + + next_ins(pAsm); + + /* MULADD R0.x, R0.x, PS1, (0x3FC00000, 1.5f).x + * MULADD R0.y, R0.y, PS1, (0x3FC00000, 1.5f).x + * muladd has no writemask, have to use another temp + * also no support for imm constants, so add 1 here + */ + pAsm->D.dst.opcode = SQ_OP3_INST_MULADD; + pAsm->D.dst.op3 = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp2; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = tmp1; + setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Z); + setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE); + pAsm->S[2].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[2].src.reg = tmp1; + setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_1); + + next_ins(pAsm); + + /* ADD the remaining .5 */ + pAsm->D.dst.opcode = SQ_OP2_INST_ADD; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp2; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 0; + pAsm->D.dst.writew = 0; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp2; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = 252; // SQ_ALU_SRC_0_5 + noswizzle_PVSSRC(&(pAsm->S[1].src)); + + next_ins(pAsm); + + /* tmp1.xy = temp2.xy */ + pAsm->D.dst.opcode = SQ_OP2_INST_MOV; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 0; + pAsm->D.dst.writew = 0; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp2; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + + next_ins(pAsm); + pAsm->aArgSubst[1] = tmp1; + need_barrier = GL_TRUE; + + } + + pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; + pAsm->is_tex = GL_TRUE; + if ( GL_TRUE == need_barrier ) + { + pAsm->need_tex_barrier = GL_TRUE; + } // Set src1 to tex unit id pAsm->S[1].src.reg = pAsm->pILInst[pAsm->uiCurInst].TexSrcUnit; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; @@ -3418,16 +3758,31 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) { return GL_FALSE; } - + if( GL_FALSE == tex_src(pAsm) ) { return GL_FALSE; } - if ( GL_FALSE == next_ins(pAsm) ) + if(pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP) { - return GL_FALSE; + /* hopefully did swizzles before */ + noswizzle_PVSSRC(&(pAsm->S[0].src)); } + + if(pAsm->pILInst[pAsm->uiCurInst].TexSrcTarget == TEXTURE_CUBE_INDEX) + { + /* SAMPLE dst, tmp.yxwy, CUBE */ + pAsm->S[0].src.swizzlex = SQ_SEL_Y; + pAsm->S[0].src.swizzley = SQ_SEL_X; + pAsm->S[0].src.swizzlez = SQ_SEL_W; + pAsm->S[0].src.swizzlew = SQ_SEL_Y; + } + + if ( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } return GL_TRUE; } diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h index f9c4d849c6..0d4283e4ba 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.h +++ b/src/mesa/drivers/dri/r600/r700_assembler.h @@ -374,6 +374,10 @@ typedef struct r700_AssemblerBase struct prog_instruction * pILInst; GLuint uiCurInst; GLboolean bR6xx; + /* helper to decide which type of instruction to assemble */ + GLboolean is_tex; + /* we inserted helper intructions and need barrier on next TEX ins */ + GLboolean need_tex_barrier; } r700_AssemblerBase; //Internal use @@ -411,6 +415,15 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm, GLuint number_of_elements, GLenum dataElementType, VTX_FETCH_METHOD* pFetchMethod); +GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm, + GLuint destination_register, + GLenum type, + GLint size, + GLubyte element, + GLuint _signed, + GLboolean normalize, + VTX_FETCH_METHOD * pFetchMethod); +GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm); GLuint gethelpr(r700_AssemblerBase* pAsm); void resethelpr(r700_AssemblerBase* pAsm); void checkop_init(r700_AssemblerBase* pAsm); diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c index 312cacffda..3b7f6fffe0 100644 --- a/src/mesa/drivers/dri/r600/r700_chip.c +++ b/src/mesa/drivers/dri/r600/r700_chip.c @@ -208,11 +208,97 @@ static void r700SetupVTXConstants(GLcontext * ctx, } +extern int getTypeSize(GLenum type); +static void r700SetupVTXConstants2(GLcontext * ctx, + void * pAos, + StreamDesc * pStreamDesc) +{ + context_t *context = R700_CONTEXT(ctx); + struct radeon_aos * paos = (struct radeon_aos *)pAos; + unsigned int nVBsize; + BATCH_LOCALS(&context->radeon); + + unsigned int uSQ_VTX_CONSTANT_WORD0_0; + unsigned int uSQ_VTX_CONSTANT_WORD1_0; + unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0; + unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0; + unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0; + + if (!paos->bo) + return; + + if ((context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV610) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV620) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS780) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS880) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV710)) + r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, TC_ACTION_ENA_bit); + else + r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, VC_ACTION_ENA_bit); + + if(0 == pStreamDesc->stride) + { + nVBsize = paos->count * pStreamDesc->size * getTypeSize(pStreamDesc->type); + } + else + { + nVBsize = paos->count * pStreamDesc->stride; + } + + uSQ_VTX_CONSTANT_WORD0_0 = paos->offset; + uSQ_VTX_CONSTANT_WORD1_0 = nVBsize - 1; + + SETfield(uSQ_VTX_CONSTANT_WORD2_0, 0, BASE_ADDRESS_HI_shift, BASE_ADDRESS_HI_mask); /* TODO */ + SETfield(uSQ_VTX_CONSTANT_WORD2_0, pStreamDesc->stride, SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift, + SQ_VTX_CONSTANT_WORD2_0__STRIDE_mask); + SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(pStreamDesc->type, pStreamDesc->size, NULL), + SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift, + SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_mask); /* TODO : trace back api for initial data type, not only GL_FLOAT */ + + if(GL_TRUE == pStreamDesc->normalize) + { + SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_NORM, + SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask); + } + //else + //{ + // SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_INT, + // SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask); + //} + + if(1 == pStreamDesc->_signed) + { + SETbit(uSQ_VTX_CONSTANT_WORD2_0, SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit); + } + + SETfield(uSQ_VTX_CONSTANT_WORD3_0, 1, MEM_REQUEST_SIZE_shift, MEM_REQUEST_SIZE_mask); + SETfield(uSQ_VTX_CONSTANT_WORD6_0, SQ_TEX_VTX_VALID_BUFFER, + SQ_TEX_RESOURCE_WORD6_0__TYPE_shift, SQ_TEX_RESOURCE_WORD6_0__TYPE_mask); + + BEGIN_BATCH_NO_AUTOSTATE(9 + 2); + + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + R600_OUT_BATCH((pStreamDesc->element + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD0_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0); + R600_OUT_BATCH(0); + R600_OUT_BATCH(0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD6_0); + R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD0_0, + paos->bo, + uSQ_VTX_CONSTANT_WORD0_0, + RADEON_GEM_DOMAIN_GTT, 0, 0); + END_BATCH(); + COMMIT_BATCH(); + +} + void r700SetupStreams(GLcontext *ctx) { context_t *context = R700_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; TNLcontext *tnl = TNL_CONTEXT(ctx); struct vertex_buffer *vb = &tnl->vb; unsigned int i, j = 0; @@ -221,7 +307,7 @@ void r700SetupStreams(GLcontext *ctx) R600_STATECHANGE(context, vtx); for(i=0; i<VERT_ATTRIB_MAX; i++) { - if(vpc->mesa_program.Base.InputsRead & (1 << i)) { + if(vp->mesa_program->Base.InputsRead & (1 << i)) { rcommon_emit_vector(ctx, &context->radeon.tcl.aos[j], vb->AttribPtr[i]->data, @@ -237,8 +323,7 @@ void r700SetupStreams(GLcontext *ctx) static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom) { context_t *context = R700_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; unsigned int i, j = 0; BATCH_LOCALS(&context->radeon); radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__); @@ -258,14 +343,24 @@ static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom) COMMIT_BATCH(); for(i=0; i<VERT_ATTRIB_MAX; i++) { - if(vpc->mesa_program.Base.InputsRead & (1 << i)) { - /* currently aos are packed */ - r700SetupVTXConstants(ctx, - i, - (void*)(&context->radeon.tcl.aos[j]), - (unsigned int)context->radeon.tcl.aos[j].components, - (unsigned int)context->radeon.tcl.aos[j].stride * 4, - (unsigned int)context->radeon.tcl.aos[j].count); + if(vp->mesa_program->Base.InputsRead & (1 << i)) + { + if(1 == context->selected_vp->uiVersion) + { + /* currently aos are packed */ + r700SetupVTXConstants(ctx, + i, + (void*)(&context->radeon.tcl.aos[j]), + (unsigned int)context->radeon.tcl.aos[j].components, + (unsigned int)context->radeon.tcl.aos[j].stride * 4, + (unsigned int)context->radeon.tcl.aos[j].count); + } + else + { /* context->selected_vp->uiVersion == 2 : aos not always packed */ + r700SetupVTXConstants2(ctx, + (void*)(&context->radeon.tcl.aos[j]), + &(context->stream_desc[j])); + } j++; } } @@ -353,7 +448,7 @@ static void r700SetDepthTarget(context_t *context) SETfield(r700->DB_DEPTH_INFO.u32All, DEPTH_16, DB_DEPTH_INFO__FORMAT_shift, DB_DEPTH_INFO__FORMAT_mask); } - SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_2D_TILED_THIN1, + SETfield(r700->DB_DEPTH_INFO.u32All, ARRAY_1D_TILED_THIN1, DB_DEPTH_INFO__ARRAY_MODE_shift, DB_DEPTH_INFO__ARRAY_MODE_mask); /* r700->DB_PREFETCH_LIMIT.bits.DEPTH_HEIGHT_TILE_MAX = (context->currentDraw->h >> 3) - 1; */ /* z buffer sie may much bigger than what need, so use actual used h. */ } diff --git a/src/mesa/drivers/dri/r600/r700_oglprog.c b/src/mesa/drivers/dri/r600/r700_oglprog.c index 3c8c1fd7a3..5290ef31be 100644 --- a/src/mesa/drivers/dri/r600/r700_oglprog.c +++ b/src/mesa/drivers/dri/r600/r700_oglprog.c @@ -46,7 +46,7 @@ static struct gl_program *r700NewProgram(GLcontext * ctx, { struct gl_program *pProgram = NULL; - struct r700_vertex_program *vp; + struct r700_vertex_program_cont *vpc; struct r700_fragment_program *fp; radeon_print(RADEON_SHADER, RADEON_VERBOSE, @@ -56,16 +56,11 @@ static struct gl_program *r700NewProgram(GLcontext * ctx, { case GL_VERTEX_STATE_PROGRAM_NV: case GL_VERTEX_PROGRAM_ARB: - vp = CALLOC_STRUCT(r700_vertex_program); + vpc = CALLOC_STRUCT(r700_vertex_program_cont); pProgram = _mesa_init_vertex_program(ctx, - &vp->mesa_program, + &vpc->mesa_program, target, id); - vp->translated = GL_FALSE; - vp->loaded = GL_FALSE; - - vp->shaderbo = NULL; - break; case GL_FRAGMENT_PROGRAM_NV: case GL_FRAGMENT_PROGRAM_ARB: @@ -89,7 +84,8 @@ static struct gl_program *r700NewProgram(GLcontext * ctx, static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog) { - struct r700_vertex_program * vp; + struct r700_vertex_program_cont * vpc; + struct r700_vertex_program *vp, *tmp; struct r700_fragment_program * fp; radeon_print(RADEON_SHADER, RADEON_VERBOSE, @@ -99,14 +95,20 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog) { case GL_VERTEX_STATE_PROGRAM_NV: case GL_VERTEX_PROGRAM_ARB: - vp = (struct r700_vertex_program*)prog; - /* Release DMA region */ - - r600DeleteShader(ctx, vp->shaderbo); - - /* Clean up */ - Clean_Up_Assembler(&(vp->r700AsmCode)); - Clean_Up_Shader(&(vp->r700Shader)); + vpc = (struct r700_vertex_program_cont*)prog; + vp = vpc->progs; + while (vp) { + tmp = vp->next; + /* Release DMA region */ + + r600DeleteShader(ctx, vp->shaderbo); + + /* Clean up */ + Clean_Up_Assembler(&(vp->r700AsmCode)); + Clean_Up_Shader(&(vp->r700Shader)); + _mesa_free(vp); + vp = tmp; + } break; case GL_FRAGMENT_PROGRAM_NV: case GL_FRAGMENT_PROGRAM_ARB: diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c index 3566bf3ca7..5627984cf9 100644 --- a/src/mesa/drivers/dri/r600/r700_render.c +++ b/src/mesa/drivers/dri/r600/r700_render.c @@ -43,6 +43,7 @@ #include "tnl/t_context.h" #include "tnl/t_vertex.h" #include "tnl/t_pipeline.h" +#include "vbo/vbo_context.h" #include "r600_context.h" #include "r600_cmdbuf.h" @@ -53,6 +54,7 @@ #include "r700_fragprog.h" #include "r700_state.h" +#include "radeon_buffer_objects.h" #include "radeon_common_context.h" void r700WaitForIdle(context_t *context); @@ -260,6 +262,16 @@ static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim TNLcontext *tnl = TNL_CONTEXT(ctx); struct vertex_buffer *vb = &tnl->vb; + GLboolean bUseDrawIndex; + if( (NULL != context->ind_buf.bo) && (GL_TRUE != context->ind_buf.bHostIb) ) + { + bUseDrawIndex = GL_TRUE; + } + else + { + bUseDrawIndex = GL_FALSE; + } + type = r700PrimitiveType(prim); num_indices = r700NumVerts(end - start, prim); @@ -270,67 +282,146 @@ static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim if (type < 0 || num_indices <= 0) return; + if(GL_TRUE == bUseDrawIndex) + { + total_emit = 3 /* VGT_PRIMITIVE_TYPE */ + + 2 /* VGT_INDEX_TYPE */ + + 2 /* NUM_INSTANCES */ + + 5+2; /* DRAW_INDEX */ + } + else + { total_emit = 3 /* VGT_PRIMITIVE_TYPE */ - + 2 /* VGT_INDEX_TYPE */ - + 2 /* NUM_INSTANCES */ - + num_indices + 3; /* DRAW_INDEX_IMMD */ + + 2 /* VGT_INDEX_TYPE */ + + 2 /* NUM_INSTANCES */ + + num_indices + 3; /* DRAW_INDEX_IMMD */ + } - BEGIN_BATCH_NO_AUTOSTATE(total_emit); + BEGIN_BATCH_NO_AUTOSTATE(total_emit); // prim - SETfield(vgt_primitive_type, type, - VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); - R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); - R600_OUT_BATCH(vgt_primitive_type); + SETfield(vgt_primitive_type, type, + VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); + R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); + R600_OUT_BATCH(vgt_primitive_type); // index type - SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); - R600_OUT_BATCH(vgt_index_type); + SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); + + if(GL_TRUE == bUseDrawIndex) + { + if(GL_TRUE != context->ind_buf.is_32bit) + { + SETfield(vgt_index_type, DI_INDEX_SIZE_16_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); + } + } + + R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); + R600_OUT_BATCH(vgt_index_type); // num instances R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0)); R600_OUT_BATCH(1); // draw packet - vgt_num_indices = num_indices; + vgt_num_indices = num_indices; + + if(GL_TRUE == bUseDrawIndex) + { + SETfield(vgt_draw_initiator, DI_SRC_SEL_DMA, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + } + else + { SETfield(vgt_draw_initiator, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + } + SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask); + if(GL_TRUE == bUseDrawIndex) + { + R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX, 3)); + R600_OUT_BATCH(context->ind_buf.bo_offset); + R600_OUT_BATCH(0); + R600_OUT_BATCH(vgt_num_indices); + R600_OUT_BATCH(vgt_draw_initiator); + R600_OUT_BATCH_RELOC(context->ind_buf.bo_offset, + context->ind_buf.bo, + context->ind_buf.bo_offset, + RADEON_GEM_DOMAIN_GTT, 0, 0); + } + else + { R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (num_indices + 1))); R600_OUT_BATCH(vgt_num_indices); R600_OUT_BATCH(vgt_draw_initiator); + } + if(NULL == context->ind_buf.bo) + { for (i = start; i < (start + num_indices); i++) { - if(vb->Elts) - R600_OUT_BATCH(vb->Elts[i]); - else - R600_OUT_BATCH(i); + if(vb->Elts) + { + R600_OUT_BATCH(vb->Elts[i]); + } + else + R600_OUT_BATCH(i); } - END_BATCH(); - COMMIT_BATCH(); + } + else + { + if(GL_TRUE == context->ind_buf.bHostIb) + { + if(GL_TRUE != context->ind_buf.is_32bit) + { + GLushort * pIndex = (GLushort*)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + pIndex += start; + for (i = 0; i < num_indices; i++) + { + R600_OUT_BATCH(*pIndex); + pIndex++; + } + } + else + { + GLuint * pIndex = (GLuint*)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + pIndex += start; + + for (i = 0; i < num_indices; i++) + { + R600_OUT_BATCH(*pIndex); + pIndex++; + } + } + } + } + END_BATCH(); + COMMIT_BATCH(); } /* start 3d, idle, cb/db flush */ #define PRE_EMIT_STATE_BUFSZ 10 + 5 + 14 -static GLuint r700PredictRenderSize(GLcontext* ctx) +static GLuint r700PredictRenderSize(GLcontext* ctx, GLuint nr_prims) { context_t *context = R700_CONTEXT(ctx); - TNLcontext *tnl = TNL_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; - struct vertex_buffer *vb = &tnl->vb; + struct r700_vertex_program *vp = context->selected_vp; GLboolean flushed; GLuint dwords, i; GLuint state_size; /* pre calculate aos count so state prediction works */ - context->radeon.tcl.aos_count = _mesa_bitcount(vpc->mesa_program.Base.InputsRead); + context->radeon.tcl.aos_count = _mesa_bitcount(vp->mesa_program->Base.InputsRead); dwords = PRE_EMIT_STATE_BUFSZ; - for (i = 0; i < vb->PrimitiveCount; i++) - dwords += vb->Primitive[i].count + 10; + if (nr_prims) + dwords += nr_prims * 14; + else { + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + + for (i = 0; i < vb->PrimitiveCount; i++) + dwords += vb->Primitive[i].count + 10; + } state_size = radeonCountStateEmitSize(&context->radeon); flushed = rcommonEnsureCmdBufSpace(&context->radeon, dwords + state_size, __FUNCTION__); @@ -365,13 +456,12 @@ static GLboolean r700RunRender(GLcontext * ctx, /* mark vtx as dirty since it changes per-draw */ R600_STATECHANGE(context, vtx); - r700UpdateShaders(ctx); r700SetScissor(context); r700SetupVertexProgram(ctx); r700SetupFragmentProgram(ctx); r600UpdateTextureState(ctx); - GLuint emit_end = r700PredictRenderSize(ctx) + GLuint emit_end = r700PredictRenderSize(ctx, 0) + context->radeon.cmdbuf.cs->cdw; r700SetupStreams(ctx); @@ -427,7 +517,10 @@ static GLboolean r700RunTCLRender(GLcontext * ctx, /*----------------------*/ /* TODO : sw fallback */ + /* Need shader bo's setup before bo check */ + r700UpdateShaders(ctx); /** + * Ensure all enabled and complete textures are uploaded along with any buffers being used. */ if(!r600ValidateBuffers(ctx)) @@ -476,4 +569,587 @@ const struct tnl_pipeline_stage *r700_pipeline[] = 0, }; +#define CONVERT( TYPE, MACRO ) do { \ + GLuint i, j, sz; \ + sz = input->Size; \ + if (input->Normalized) { \ + for (i = 0; i < count; i++) { \ + const TYPE *in = (TYPE *)src_ptr; \ + for (j = 0; j < sz; j++) { \ + *dst_ptr++ = MACRO(*in); \ + in++; \ + } \ + src_ptr += stride; \ + } \ + } else { \ + for (i = 0; i < count; i++) { \ + const TYPE *in = (TYPE *)src_ptr; \ + for (j = 0; j < sz; j++) { \ + *dst_ptr++ = (GLfloat)(*in); \ + in++; \ + } \ + src_ptr += stride; \ + } \ + } \ +} while (0) + +/** + * Convert attribute data type to float + * If the attribute uses named buffer object replace the bo with newly allocated bo + */ +static void r700ConvertAttrib(GLcontext *ctx, int count, + const struct gl_client_array *input, + struct StreamDesc *attr) +{ + context_t *context = R700_CONTEXT(ctx); + const GLvoid *src_ptr; + GLboolean mapped_named_bo = GL_FALSE; + GLfloat *dst_ptr; + GLuint stride; + + stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size : input->StrideB; + + /* Convert value for first element only */ + if (input->StrideB == 0) + { + count = 1; + } + + if (input->BufferObj->Name) + { + if (!input->BufferObj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj); + mapped_named_bo = GL_TRUE; + } + + src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr); + } + else + { + src_ptr = input->Ptr; + } + + radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, + sizeof(GLfloat) * input->Size * count, 32); + dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset); + + assert(src_ptr != NULL); + + switch (input->Type) + { + case GL_DOUBLE: + CONVERT(GLdouble, (GLfloat)); + break; + case GL_UNSIGNED_INT: + CONVERT(GLuint, UINT_TO_FLOAT); + break; + case GL_INT: + CONVERT(GLint, INT_TO_FLOAT); + break; + case GL_UNSIGNED_SHORT: + CONVERT(GLushort, USHORT_TO_FLOAT); + break; + case GL_SHORT: + CONVERT(GLshort, SHORT_TO_FLOAT); + break; + case GL_UNSIGNED_BYTE: + assert(input->Format != GL_BGRA); + CONVERT(GLubyte, UBYTE_TO_FLOAT); + break; + case GL_BYTE: + CONVERT(GLbyte, BYTE_TO_FLOAT); + break; + default: + assert(0); + break; + } + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj); + } +} + +static void r700AlignDataToDword(GLcontext *ctx, + const struct gl_client_array *input, + int count, + struct StreamDesc *attr) +{ + context_t *context = R700_CONTEXT(ctx); + const int dst_stride = (input->StrideB + 3) & ~3; + const int size = getTypeSize(input->Type) * input->Size * count; + GLboolean mapped_named_bo = GL_FALSE; + + radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, size, 32); + + if (!input->BufferObj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj); + mapped_named_bo = GL_TRUE; + } + + { + GLvoid *src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr); + GLvoid *dst_ptr = ADD_POINTERS(attr->bo->ptr, attr->bo_offset); + int i; + + for (i = 0; i < count; ++i) + { + _mesa_memcpy(dst_ptr, src_ptr, input->StrideB); + src_ptr += input->StrideB; + dst_ptr += dst_stride; + } + } + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj); + } + + attr->stride = dst_stride; +} + +static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *input[], int count) +{ + context_t *context = R700_CONTEXT(ctx); + GLuint stride; + int ret; + int i, index; + + R600_STATECHANGE(context, vtx); + + for(index = 0; index < context->nNumActiveAos; index++) + { + struct radeon_aos *aos = &context->radeon.tcl.aos[index]; + i = context->stream_desc[index].element; + + stride = (input[i]->StrideB == 0) ? getTypeSize(input[i]->Type) * input[i]->Size : input[i]->StrideB; + + if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT || +#if MESA_BIG_ENDIAN + getTypeSize(input[i]->Type) != 4 || +#endif + stride < 4) + { + r700ConvertAttrib(ctx, count, input[i], &context->stream_desc[index]); + } + else + { + if (input[i]->BufferObj->Name) + { + if (stride % 4 != 0) + { + assert(((intptr_t) input[i]->Ptr) % input[i]->StrideB == 0); + r700AlignDataToDword(ctx, input[i], count, &context->stream_desc[index]); + context->stream_desc[index].is_named_bo = GL_FALSE; + } + else + { + context->stream_desc[index].stride = input[i]->StrideB; + context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr; + context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo; + context->stream_desc[index].is_named_bo = GL_TRUE; + } + } + else + { + int size; + int local_count = count; + uint32_t *dst; + + if (input[i]->StrideB == 0) + { + size = getTypeSize(input[i]->Type) * input[i]->Size; + local_count = 1; + } + else + { + size = getTypeSize(input[i]->Type) * input[i]->Size * local_count; + } + + radeonAllocDmaRegion(&context->radeon, &context->stream_desc[index].bo, + &context->stream_desc[index].bo_offset, size, 32); + assert(context->stream_desc[index].bo->ptr != NULL); + dst = (uint32_t *)ADD_POINTERS(context->stream_desc[index].bo->ptr, + context->stream_desc[index].bo_offset); + + switch (context->stream_desc[index].dwords) + { + case 1: + radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 4; + break; + case 2: + radeonEmitVec8(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 8; + break; + case 3: + radeonEmitVec12(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 12; + break; + case 4: + radeonEmitVec16(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 16; + break; + default: + assert(0); + break; + } + } + } + + aos->count = context->stream_desc[index].stride == 0 ? 1 : count; + aos->stride = context->stream_desc[index].stride / sizeof(float); + aos->components = context->stream_desc[index].dwords; + aos->bo = context->stream_desc[index].bo; + aos->offset = context->stream_desc[index].bo_offset; + + if(context->stream_desc[index].is_named_bo) + { + radeon_cs_space_add_persistent_bo(context->radeon.cmdbuf.cs, + context->stream_desc[index].bo, + RADEON_GEM_DOMAIN_GTT, 0); + } + } + + context->radeon.tcl.aos_count = context->nNumActiveAos; + ret = radeon_cs_space_check_with_bo(context->radeon.cmdbuf.cs, + first_elem(&context->radeon.dma.reserved)->bo, + RADEON_GEM_DOMAIN_GTT, 0); +} + +static void r700FreeData(GLcontext *ctx) +{ + /* Need to zero tcl.aos[n].bo and tcl.elt_dma_bo + * to prevent double unref in radeonReleaseArrays + * called during context destroy + */ + context_t *context = R700_CONTEXT(ctx); + + int i; + + for (i = 0; i < context->nNumActiveAos; i++) + { + if (!context->stream_desc[i].is_named_bo) + { + radeon_bo_unref(context->stream_desc[i].bo); + } + context->radeon.tcl.aos[i].bo = NULL; + } + + if (context->ind_buf.bo != NULL) + { + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeon_bo_unref(context->ind_buf.bo); + } + else + { + FREE(context->ind_buf.bo->ptr); + FREE(context->ind_buf.bo); + context->ind_buf.bo = NULL; + } + } +} + +static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf) +{ + context_t *context = R700_CONTEXT(ctx); + GLvoid *src_ptr; + GLuint *out; + int i; + GLboolean mapped_named_bo = GL_FALSE; + + if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj); + mapped_named_bo = GL_TRUE; + assert(mesa_ind_buf->obj->Pointer != NULL); + } + src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr); + + if (mesa_ind_buf->type == GL_UNSIGNED_BYTE) + { + GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1); + GLubyte *in = (GLubyte *)src_ptr; + + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + + assert(context->ind_buf.bo->ptr != NULL); + out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + } + else + { + context->ind_buf.bo = MALLOC_STRUCT(radeon_bo); + context->ind_buf.bo->ptr = ALIGN_MALLOC(size, 4); + context->ind_buf.bo_offset = 0; + out = (GLuint *)context->ind_buf.bo->ptr; + } + + for (i = 0; i + 1 < mesa_ind_buf->count; i += 2) + { + *out++ = in[i] | in[i + 1] << 16; + } + + if (i < mesa_ind_buf->count) + { + *out++ = in[i]; + } + +#if MESA_BIG_ENDIAN + } + else + { /* if (mesa_ind_buf->type == GL_UNSIGNED_SHORT) */ + GLushort *in = (GLushort *)src_ptr; + GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1); + + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + + assert(context->ind_buf.bo->ptr != NULL); + out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + } + else + { + context->ind_buf.bo = MALLOC_STRUCT(radeon_bo); + context->ind_buf.bo->ptr = ALIGN_MALLOC(size, 4); + context->ind_buf.bo_offset = 0; + out = (GLuint *)context->ind_buf.bo->ptr; + } + + for (i = 0; i + 1 < mesa_ind_buf->count; i += 2) + { + *out++ = in[i] | in[i + 1] << 16; + } + + if (i < mesa_ind_buf->count) + { + *out++ = in[i]; + } +#endif + } + + context->ind_buf.is_32bit = GL_FALSE; + context->ind_buf.count = mesa_ind_buf->count; + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj); + } +} + +static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf) +{ + context_t *context = R700_CONTEXT(ctx); + + if (!mesa_ind_buf) { + context->ind_buf.bo = NULL; + return; + } + + context->ind_buf.bHostIb = GL_FALSE; + +#if MESA_BIG_ENDIAN + if (mesa_ind_buf->type == GL_UNSIGNED_INT) + { +#else + if (mesa_ind_buf->type != GL_UNSIGNED_BYTE) + { +#endif + const GLvoid *src_ptr; + GLvoid *dst_ptr; + GLboolean mapped_named_bo = GL_FALSE; + + if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj); + assert(mesa_ind_buf->obj->Pointer != NULL); + mapped_named_bo = GL_TRUE; + } + + src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr); + + const GLuint size = mesa_ind_buf->count * getTypeSize(mesa_ind_buf->type); + + if(context->ind_buf.bHostIb != GL_TRUE) + { + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + assert(context->ind_buf.bo->ptr != NULL); + dst_ptr = ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + } + else + { + context->ind_buf.bo = MALLOC_STRUCT(radeon_bo); + context->ind_buf.bo->ptr = ALIGN_MALLOC(size, 4); + context->ind_buf.bo_offset = 0; + dst_ptr = context->ind_buf.bo->ptr; + } + + _mesa_memcpy(dst_ptr, src_ptr, size); + + context->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT); + context->ind_buf.count = mesa_ind_buf->count; + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj); + } + } + else + { + r700FixupIndexBuffer(ctx, mesa_ind_buf); + } +} + +static GLboolean r700TryDrawPrims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLuint min_index, + GLuint max_index ) +{ + context_t *context = R700_CONTEXT(ctx); + radeonContextPtr radeon = &context->radeon; + GLuint i, id = 0; + struct radeon_renderbuffer *rrb; + + if (ctx->NewState) + { + _mesa_update_state( ctx ); + } + + _tnl_UpdateFixedFunctionProgram(ctx); + r700SetVertexFormat(ctx, arrays, max_index + 1); + r700SetupIndexBuffer(ctx, ib); + /* shaders need to be updated before buffers are validated */ + r700UpdateShaders2(ctx); + if (!r600ValidateBuffers(ctx)) + return GL_FALSE; + + /* always emit CB base to prevent + * lock ups on some chips. + */ + R600_STATECHANGE(context, cb_target); + /* mark vtx as dirty since it changes per-draw */ + R600_STATECHANGE(context, vtx); + + r700SetScissor(context); + r700SetupVertexProgram(ctx); + r700SetupFragmentProgram(ctx); + r600UpdateTextureState(ctx); + + GLuint emit_end = r700PredictRenderSize(ctx, nr_prims) + + context->radeon.cmdbuf.cs->cdw; + + r700SetupStreams2(ctx, arrays, max_index + 1); + + radeonEmitState(radeon); + + radeon_debug_add_indent(); + for (i = 0; i < nr_prims; ++i) + { + r700RunRenderPrimitive(ctx, + prim[i].start, + prim[i].start + prim[i].count, + prim[i].mode); + } + radeon_debug_remove_indent(); + + /* Flush render op cached for last several quads. */ + r700WaitForIdleClean(context); + + rrb = radeon_get_colorbuffer(&context->radeon); + if (rrb && rrb->bo) + r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM, + CB_ACTION_ENA_bit | (1 << (id + 6))); + + rrb = radeon_get_depthbuffer(&context->radeon); + if (rrb && rrb->bo) + r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM, + DB_ACTION_ENA_bit | DB_DEST_BASE_ENA_bit); + + r700FreeData(ctx); + + if (emit_end < context->radeon.cmdbuf.cs->cdw) + { + WARN_ONCE("Rendering was %d commands larger than predicted size." + " We might overflow command buffer.\n", context->radeon.cmdbuf.cs->cdw - emit_end); + } + + return GL_TRUE; +} + +static void r700DrawPrimsRe(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + GLboolean retval = GL_FALSE; + + /* This check should get folded into just the places that + * min/max index are really needed. + */ + if (!index_bounds_valid) { + vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index); + } + + if (min_index) { + vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r700DrawPrimsRe ); + return; + } + + /* Make an attempt at drawing */ + retval = r700TryDrawPrims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); + + /* If failed run tnl pipeline - it should take care of fallbacks */ + if (!retval) + _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); +} + +static void r700DrawPrims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + context_t *context = R700_CONTEXT(ctx); + + /* For non indexed drawing, using tnl pipe. */ + if(!ib) + { + context->ind_buf.bo = NULL; + + _tnl_vbo_draw_prims(ctx, arrays, prim, nr_prims, ib, + index_bounds_valid, min_index, max_index); + return; + } + + r700DrawPrimsRe(ctx, arrays, prim, nr_prims, ib, index_bounds_valid, min_index, max_index); +} + +void r700InitDraw(GLcontext *ctx) +{ + struct vbo_context *vbo = vbo_context(ctx); + + /* to be enabled */ + /* + vbo->draw_prims = r700DrawPrims; + */ +} + diff --git a/src/mesa/drivers/dri/r600/r700_shader.c b/src/mesa/drivers/dri/r600/r700_shader.c index b4fd51c137..955ea4e4e1 100644 --- a/src/mesa/drivers/dri/r600/r700_shader.c +++ b/src/mesa/drivers/dri/r600/r700_shader.c @@ -60,6 +60,55 @@ void AddInstToList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * plstCFInstructions->uNumOfNode++; } +void TakeInstOutFromList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * pInst) +{ + GLuint ulIndex = 0; + GLboolean bFound = GL_FALSE; + R700ShaderInstruction * pPrevInst = NULL; + R700ShaderInstruction * pCurInst = plstCFInstructions->pHead; + + /* Need go thro list to make sure pInst is there. */ + while(NULL != pCurInst) + { + if(pCurInst == pInst) + { + bFound = GL_TRUE; + break; + } + + pPrevInst = pCurInst; + pCurInst = pCurInst->pNextInst; + } + if(GL_TRUE == bFound) + { + plstCFInstructions->uNumOfNode--; + + pCurInst = pInst->pNextInst; + ulIndex = pInst->m_uIndex; + while(NULL != pCurInst) + { + pCurInst->m_uIndex = ulIndex; + ulIndex++; + pCurInst = pCurInst->pNextInst; + } + + if(plstCFInstructions->pHead == pInst) + { + plstCFInstructions->pHead = pInst->pNextInst; + } + if(plstCFInstructions->pTail == pInst) + { + plstCFInstructions->pTail = pPrevInst; + } + if(NULL != pPrevInst) + { + pPrevInst->pNextInst = pInst->pNextInst; + } + + FREE(pInst); + } +} + void Init_R700_Shader(R700_Shader * pShader) { pShader->Type = R700_SHADER_INVALID; @@ -488,6 +537,47 @@ void DebugPrint(void) { } +void cleanup_vfetch_shaderinst(R700_Shader *pShader) +{ + R700ShaderInstruction *pInst; + R700ShaderInstruction *pInstToFree; + R700VertexInstruction *pVTXInst; + R700ControlFlowInstruction *pCFInst; + + pInst = pShader->lstVTXInstructions.pHead; + while(NULL != pInst) + { + pVTXInst = (R700VertexInstruction *)pInst; + pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pVTXInst->m_ShaderInstType); + + if(NULL != pVTXInst->m_pLinkedGenericClause) + { + pCFInst = (R700ControlFlowInstruction*)(pVTXInst->m_pLinkedGenericClause); + + TakeInstOutFromList(&(pShader->lstCFInstructions), + (R700ShaderInstruction*)pCFInst); + + pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pCFInst->m_ShaderInstType); + } + + pInst = pInst->pNextInst; + }; + + //destroy each item in pShader->lstVTXInstructions; + pInst = pShader->lstVTXInstructions.pHead; + while(NULL != pInst) + { + pInstToFree = pInst; + pInst = pInst->pNextInst; + FREE(pInstToFree); + }; + + //set NULL pShader->lstVTXInstructions + pShader->lstVTXInstructions.pHead=NULL; + pShader->lstVTXInstructions.pTail=NULL; + pShader->lstVTXInstructions.uNumOfNode=0; +} + void Clean_Up_Shader(R700_Shader *pShader) { FREE(pShader->pProgram); diff --git a/src/mesa/drivers/dri/r600/r700_shader.h b/src/mesa/drivers/dri/r600/r700_shader.h index bfd01e1a93..997cb05aaf 100644 --- a/src/mesa/drivers/dri/r600/r700_shader.h +++ b/src/mesa/drivers/dri/r600/r700_shader.h @@ -143,6 +143,7 @@ void LoadProgram(R700_Shader *pShader); void UpdateShaderRegisters(R700_Shader *pShader); void DeleteInstructions(R700_Shader *pShader); void DebugPrint(void); +void cleanup_vfetch_shaderinst(R700_Shader *pShader); void Clean_Up_Shader(R700_Shader *pShader); diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c index e91aa43118..fbff109455 100644 --- a/src/mesa/drivers/dri/r600/r700_state.c +++ b/src/mesa/drivers/dri/r600/r700_state.c @@ -92,7 +92,25 @@ void r700UpdateShaders (GLcontext * ctx) //---------------------------------- } } - r700SelectVertexShader(ctx); + r700SelectVertexShader(ctx, 1); + r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS); + context->radeon.NewGLState = 0; +} + +void r700UpdateShaders2(GLcontext * ctx) +{ + context_t *context = R700_CONTEXT(ctx); + + /* should only happenen once, just after context is created */ + /* TODO: shouldn't we fallback to sw here? */ + if (!ctx->FragmentProgram._Current) { + _mesa_fprintf(stderr, "No ctx->FragmentProgram._Current!!\n"); + return; + } + + r700SelectFragmentShader(ctx); + + r700SelectVertexShader(ctx, 2); r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS); context->radeon.NewGLState = 0; } @@ -845,9 +863,9 @@ static void r700PointSize(GLcontext * ctx, GLfloat size) size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize); /* format is 12.4 fixed point */ - SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 16), + SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 8.0), PA_SU_POINT_SIZE__HEIGHT_shift, PA_SU_POINT_SIZE__HEIGHT_mask); - SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 16), + SETfield(r700->PA_SU_POINT_SIZE.u32All, (int)(size * 8.0), PA_SU_POINT_SIZE__WIDTH_shift, PA_SU_POINT_SIZE__WIDTH_mask); } @@ -862,11 +880,11 @@ static void r700PointParameter(GLcontext * ctx, GLenum pname, const GLfloat * pa /* format is 12.4 fixed point */ switch (pname) { case GL_POINT_SIZE_MIN: - SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MinSize * 16.0), + SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MinSize * 8.0), MIN_SIZE_shift, MIN_SIZE_mask); break; case GL_POINT_SIZE_MAX: - SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MaxSize * 16.0), + SETfield(r700->PA_SU_POINT_MINMAX.u32All, (int)(ctx->Point.MaxSize * 8.0), MAX_SIZE_shift, MAX_SIZE_mask); break; case GL_POINT_DISTANCE_ATTENUATION: @@ -1130,20 +1148,25 @@ static void r700PolygonOffset(GLcontext * ctx, GLfloat factor, GLfloat units) // context_t *context = R700_CONTEXT(ctx); R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); GLfloat constant = units; + GLchar depth = 0; + + R600_STATECHANGE(context, poly); switch (ctx->Visual.depthBits) { case 16: constant *= 4.0; + depth = -16; break; case 24: constant *= 2.0; + depth = -24; break; } factor *= 12.0; - - R600_STATECHANGE(context, poly); - + SETfield(r700->PA_SU_POLY_OFFSET_DB_FMT_CNTL.u32All, depth, + POLY_OFFSET_NEG_NUM_DB_BITS_shift, POLY_OFFSET_NEG_NUM_DB_BITS_mask); + //r700->PA_SU_POLY_OFFSET_CLAMP.f32All = constant; //??? r700->PA_SU_POLY_OFFSET_FRONT_SCALE.f32All = factor; r700->PA_SU_POLY_OFFSET_FRONT_OFFSET.f32All = constant; r700->PA_SU_POLY_OFFSET_BACK_SCALE.f32All = factor; @@ -1272,10 +1295,11 @@ void r700SetScissor(context_t *context) //--------------- return; } if (context->radeon.state.scissor.enabled) { + /* r600 has exclusive scissors */ x1 = context->radeon.state.scissor.rect.x1; y1 = context->radeon.state.scissor.rect.y1; - x2 = context->radeon.state.scissor.rect.x2; - y2 = context->radeon.state.scissor.rect.y2; + x2 = context->radeon.state.scissor.rect.x2 + 1; + y2 = context->radeon.state.scissor.rect.y2 + 1; } else { if (context->radeon.radeonScreen->driScreen->dri2.enabled) { x1 = 0; diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h index 0f53d5b4c5..209189d8d7 100644 --- a/src/mesa/drivers/dri/r600/r700_state.h +++ b/src/mesa/drivers/dri/r600/r700_state.h @@ -35,6 +35,7 @@ extern void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state); extern void r700UpdateShaders (GLcontext * ctx); +extern void r700UpdateShaders2(GLcontext * ctx); extern void r700UpdateViewportOffset(GLcontext * ctx); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c index d107f99e7b..e7a209be9d 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.c +++ b/src/mesa/drivers/dri/r600/r700_vertprog.c @@ -35,12 +35,14 @@ #include "main/mtypes.h" #include "tnl/t_context.h" +#include "shader/program.h" #include "shader/prog_parameter.h" #include "shader/prog_statevars.h" #include "radeon_debug.h" #include "r600_context.h" #include "r600_cmdbuf.h" +#include "shader/programopt.c" #include "r700_debug.h" #include "r700_vertprog.h" @@ -157,7 +159,35 @@ GLboolean Process_Vertex_Program_Vfetch_Instructions( return GL_TRUE; } -void Map_Vertex_Program(struct r700_vertex_program *vp, +GLboolean Process_Vertex_Program_Vfetch_Instructions2( + GLcontext *ctx, + struct r700_vertex_program *vp, + struct gl_vertex_program *mesa_vp) +{ + int i; + context_t *context = R700_CONTEXT(ctx); + + VTX_FETCH_METHOD vtxFetchMethod; + vtxFetchMethod.bEnableMini = GL_FALSE; + vtxFetchMethod.mega_fetch_remainder = 0; + + for(i=0; i<context->nNumActiveAos; i++) + { + assemble_vfetch_instruction2(&vp->r700AsmCode, + vp->r700AsmCode.ucVP_AttributeMap[context->stream_desc[i].element], + context->stream_desc[i].type, + context->stream_desc[i].size, + context->stream_desc[i].element, + context->stream_desc[i]._signed, + context->stream_desc[i].normalize, + &vtxFetchMethod); + } + + return GL_TRUE; +} + +void Map_Vertex_Program(GLcontext *ctx, + struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp) { GLuint ui; @@ -173,11 +203,22 @@ void Map_Vertex_Program(struct r700_vertex_program *vp, pAsm->number_used_registers += num_inputs; // Create VFETCH instructions for inputs - if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) - { - radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n"); - return; //error - } + if(1 == vp->uiVersion) + { + if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) + { + radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n"); + return; + } + } + else + { + if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions2(ctx, vp, mesa_vp) ) + { + radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions2 return error. \n"); + return; + } + } // Map Outputs pAsm->number_of_exports = Map_Vertex_Output(pAsm, mesa_vp, pAsm->number_used_registers); @@ -258,28 +299,61 @@ GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp, return GL_TRUE; } -GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, - struct gl_vertex_program *mesa_vp) +struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, + struct gl_vertex_program *mesa_vp, + GLint nVer) { + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vp; + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + unsigned int unBit; + unsigned int i; + + vp = _mesa_calloc(sizeof(*vp)); + vp->uiVersion = nVer; + vp->mesa_program = (struct gl_vertex_program *)_mesa_clone_program(ctx, &mesa_vp->Base); + + if (mesa_vp->IsPositionInvariant) + { + _mesa_insert_mvp_code(ctx, vp->mesa_program); + } + + for(i=0; i<VERT_ATTRIB_MAX; i++) + { + unBit = 1 << i; + if(vp->mesa_program->Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */ + { + vp->aos_desc[i].size = vb->AttribPtr[i]->size; + vp->aos_desc[i].stride = vb->AttribPtr[i]->size * sizeof(GL_FLOAT);/* when emit array, data is packed. vb->AttribPtr[i]->stride;*/ + vp->aos_desc[i].type = GL_FLOAT; + } + } + + if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) + { + vp->r700AsmCode.bR6xx = 1; + } + //Init_Program Init_r700_AssemblerBase(SPT_VP, &(vp->r700AsmCode), &(vp->r700Shader) ); - Map_Vertex_Program( vp, mesa_vp ); + Map_Vertex_Program(ctx, vp, vp->mesa_program ); - if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, mesa_vp)) + if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, vp->mesa_program)) { - return GL_FALSE; + return NULL; } - if(GL_FALSE == AssembleInstr(mesa_vp->Base.NumInstructions, - &(mesa_vp->Base.Instructions[0]), + if(GL_FALSE == AssembleInstr(vp->mesa_program->Base.NumInstructions, + &(vp->mesa_program->Base.Instructions[0]), &(vp->r700AsmCode)) ) { - return GL_FALSE; + return NULL; } - if(GL_FALSE == Process_Vertex_Exports(&(vp->r700AsmCode), mesa_vp->Base.OutputsWritten) ) + if(GL_FALSE == Process_Vertex_Exports(&(vp->r700AsmCode), vp->mesa_program->Base.OutputsWritten) ) { - return GL_FALSE; + return NULL; } vp->r700Shader.nRegs = (vp->r700AsmCode.number_used_registers == 0) ? 0 @@ -289,72 +363,216 @@ GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, vp->translated = GL_TRUE; - return GL_TRUE; + return vp; } -void r700SelectVertexShader(GLcontext *ctx) +void r700SelectVertexShader(GLcontext *ctx, GLint nVersion) { context_t *context = R700_CONTEXT(ctx); - struct r700_vertex_program *vpc - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program_cont *vpc; + struct r700_vertex_program *vp; TNLcontext *tnl = TNL_CONTEXT(ctx); struct vertex_buffer *vb = &tnl->vb; unsigned int unBit; unsigned int i; + GLboolean match; + GLbitfield InputsRead; - if (context->radeon.NewGLState & (_NEW_PROGRAM_CONSTANTS|_NEW_PROGRAM)) - { - vpc->needUpdateVF = 1; - } + vpc = (struct r700_vertex_program_cont *)ctx->VertexProgram._Current; - if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) + InputsRead = vpc->mesa_program.Base.InputsRead; + if (vpc->mesa_program.IsPositionInvariant) { - vpc->r700AsmCode.bR6xx = 1; - } - + InputsRead |= VERT_BIT_POS; + } + + for (vp = vpc->progs; vp; vp = vp->next) + { + match = GL_TRUE; for(i=0; i<VERT_ATTRIB_MAX; i++) { unBit = 1 << i; - if(vpc->mesa_program.Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */ + if(InputsRead & unBit) { - vpc->aos_desc[i].size = vb->AttribPtr[i]->size; - vpc->aos_desc[i].stride = vb->AttribPtr[i]->size * sizeof(GL_FLOAT);/* when emit array, data is packed. vb->AttribPtr[i]->stride;*/ - vpc->aos_desc[i].type = GL_FLOAT; + if (vp->aos_desc[i].size != vb->AttribPtr[i]->size) + match = GL_FALSE; + break; } } + if (match) + { + context->selected_vp = vp; + return; + } + } + + vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program), nVersion); + if(!vp) + { + radeon_error("Failed to translate vertex shader. \n"); + return; + } + vp->next = vpc->progs; + vpc->progs = vp; + context->selected_vp = vp; + return; +} + +int getTypeSize(GLenum type) +{ + switch (type) + { + case GL_DOUBLE: + return sizeof(GLdouble); + case GL_FLOAT: + return sizeof(GLfloat); + case GL_INT: + return sizeof(GLint); + case GL_UNSIGNED_INT: + return sizeof(GLuint); + case GL_SHORT: + return sizeof(GLshort); + case GL_UNSIGNED_SHORT: + return sizeof(GLushort); + case GL_BYTE: + return sizeof(GLbyte); + case GL_UNSIGNED_BYTE: + return sizeof(GLubyte); + default: + assert(0); + return 0; + } +} + +static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const struct gl_client_array *input) +{ + context_t *context = R700_CONTEXT(ctx); + + StreamDesc * pStreamDesc = &(context->stream_desc[context->nNumActiveAos]); + + GLuint stride; + + stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size + : input->StrideB; + + if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT || +#if MESA_BIG_ENDIAN + getTypeSize(input->Type) != 4 || +#endif + stride < 4) + { + pStreamDesc->type = GL_FLOAT; + + if (input->StrideB == 0) + { + pStreamDesc->stride = 0; + } + else + { + pStreamDesc->stride = sizeof(GLfloat) * input->Size; + } + pStreamDesc->dwords = input->Size; + pStreamDesc->is_named_bo = GL_FALSE; + } + else + { + pStreamDesc->type = input->Type; + pStreamDesc->dwords = (getTypeSize(input->Type) * input->Size + 3)/ 4; + if (!input->BufferObj->Name) + { + if (input->StrideB == 0) + { + pStreamDesc->stride = 0; + } + else + { + pStreamDesc->stride = (getTypeSize(pStreamDesc->type) * input->Size + 3) & ~3; + } + + pStreamDesc->is_named_bo = GL_FALSE; + } + } - if(GL_FALSE == vpc->translated) { - r700TranslateVertexShader(vpc, &(vpc->mesa_program) ); + pStreamDesc->size = input->Size; + pStreamDesc->dst_loc = context->nNumActiveAos; + pStreamDesc->element = unLoc; + + switch (pStreamDesc->type) + { //GetSurfaceFormat + case GL_FLOAT: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = GL_FALSE; + break; + case GL_SHORT: + pStreamDesc->_signed = 1; + pStreamDesc->normalize = input->Normalized; + break; + case GL_BYTE: + pStreamDesc->_signed = 1; + pStreamDesc->normalize = input->Normalized; + break; + case GL_UNSIGNED_SHORT: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = input->Normalized; + break; + case GL_UNSIGNED_BYTE: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = input->Normalized; + break; + default: + case GL_INT: + case GL_UNSIGNED_INT: + case GL_DOUBLE: + assert(0); + break; } + context->nNumActiveAos++; +} + +void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count) +{ + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vpc + = (struct r700_vertex_program *)ctx->VertexProgram._Current; + + struct gl_vertex_program * mesa_vp = (struct gl_vertex_program *)&(vpc->mesa_program); + unsigned int unLoc = 0; + unsigned int unBit = mesa_vp->Base.InputsRead; + context->nNumActiveAos = 0; + + while(unBit) + { + if(unBit & 1) + { + r700TranslateAttrib(ctx, unLoc, count, arrays[unLoc]); + } + + unBit >>= 1; + ++unLoc; + } } void * r700GetActiveVpShaderBo(GLcontext * ctx) { - struct r700_vertex_program *vp - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vp = context->selected_vp;; - return vp->shaderbo; + if (vp) + return vp->shaderbo; + else + return NULL; } GLboolean r700SetupVertexProgram(GLcontext * ctx) { context_t *context = R700_CONTEXT(ctx); R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); - struct r700_vertex_program *vp - = (struct r700_vertex_program *)ctx->VertexProgram._Current; + struct r700_vertex_program *vp = context->selected_vp; struct gl_program_parameter_list *paramList; unsigned int unNumParamData; unsigned int ui; - if (vp->needUpdateVF) - { - vp->loaded = GL_FALSE; - vp->r700Shader.bNeedsAssembly = GL_TRUE; - Process_Vertex_Program_Vfetch_Instructions(vp, &(vp->mesa_program)); - r600DeleteShader(ctx, vp->shaderbo); - } - if(GL_FALSE == vp->loaded) { if(vp->r700Shader.bNeedsAssembly == GL_TRUE) @@ -410,7 +628,7 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx) */ /* sent out shader constants. */ - paramList = vp->mesa_program.Base.Parameters; + paramList = vp->mesa_program->Base.Parameters; if(NULL != paramList) { _mesa_load_state_parameters(ctx, paramList); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h index e2e65021fd..f9a3e395ee 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.h +++ b/src/mesa/drivers/dri/r600/r700_vertprog.h @@ -43,7 +43,7 @@ typedef struct ArrayDesc //TEMP struct r700_vertex_program { - struct gl_vertex_program mesa_program; /* Must be first */ + struct gl_vertex_program *mesa_program; /* Must be first */ struct r700_vertex_program *next; @@ -52,13 +52,20 @@ struct r700_vertex_program GLboolean translated; GLboolean loaded; - GLboolean needUpdateVF; + GLint uiVersion; void * shaderbo; ArrayDesc aos_desc[VERT_ATTRIB_MAX]; }; +struct r700_vertex_program_cont +{ + struct gl_vertex_program mesa_program; + + struct r700_vertex_program *progs; +}; + //Internal unsigned int Map_Vertex_Output(r700_AssemblerBase *pAsm, struct gl_vertex_program *mesa_vp, @@ -69,19 +76,28 @@ unsigned int Map_Vertex_Input(r700_AssemblerBase *pAsm, GLboolean Process_Vertex_Program_Vfetch_Instructions( struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); -void Map_Vertex_Program(struct r700_vertex_program *vp, +GLboolean Process_Vertex_Program_Vfetch_Instructions2( + GLcontext *ctx, + struct r700_vertex_program *vp, + struct gl_vertex_program *mesa_vp); +void Map_Vertex_Program(GLcontext *ctx, + struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); -GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp, - struct gl_vertex_program *mesa_vp); +struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, + struct gl_vertex_program *mesa_vp, + GLint nVer); /* Interface */ -extern void r700SelectVertexShader(GLcontext *ctx); +extern void r700SelectVertexShader(GLcontext *ctx, GLint nVersion); +extern void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count); extern GLboolean r700SetupVertexProgram(GLcontext * ctx); extern void * r700GetActiveVpShaderBo(GLcontext * ctx); +extern int getTypeSize(GLenum type); + #endif /* _R700_VERTPROG_H_ */ diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.c b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c new file mode 120000 index 0000000000..f6a5f66470 --- /dev/null +++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c @@ -0,0 +1 @@ +../radeon/radeon_buffer_objects.c
\ No newline at end of file diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.h b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h new file mode 120000 index 0000000000..2f134fd17b --- /dev/null +++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h @@ -0,0 +1 @@ +../radeon/radeon_buffer_objects.h
\ No newline at end of file diff --git a/src/mesa/drivers/dri/radeon/Makefile b/src/mesa/drivers/dri/radeon/Makefile index 1f286776b5..b1efc72872 100644 --- a/src/mesa/drivers/dri/radeon/Makefile +++ b/src/mesa/drivers/dri/radeon/Makefile @@ -47,8 +47,7 @@ C_SOURCES = \ $(DRIVER_SOURCES) \ $(CS_SOURCES) -DRIVER_DEFINES = -DRADEON_COMMON=0 \ - -Wall +DRIVER_DEFINES = -DRADEON_R100 -Wall DRI_LIB_DEPS += $(RADEON_LDFLAGS) diff --git a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c index a24b6dac26..8fac5c6c51 100644 --- a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c +++ b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c @@ -187,7 +187,11 @@ radeonMapBuffer(GLcontext * ctx, radeon_bo_map(radeon_obj->bo, access == GL_WRITE_ONLY_ARB); - return obj->Pointer = radeon_obj->bo->ptr; + obj->Pointer = radeon_obj->bo->ptr; + obj->Length = obj->Size; + obj->Offset = 0; + + return obj->Pointer; } @@ -203,9 +207,12 @@ radeonUnmapBuffer(GLcontext * ctx, if (radeon_obj->bo != NULL) { radeon_bo_unmap(radeon_obj->bo); - obj->Pointer = NULL; } + obj->Pointer = NULL; + obj->Offset = 0; + obj->Length = 0; + return GL_TRUE; } diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c index a4c7b40798..f8a4cdb495 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common.c +++ b/src/mesa/drivers/dri/radeon/radeon_common.c @@ -232,13 +232,13 @@ void radeonUpdateScissor( GLcontext *ctx ) __DRIdrawablePrivate *dPriv = radeon_get_drawable(rmesa); x1 += dPriv->x; - x2 += dPriv->x + 1; + x2 += dPriv->x; min_x += dPriv->x; - max_x += dPriv->x + 1; + max_x += dPriv->x; y1 += dPriv->y; - y2 += dPriv->y + 1; + y2 += dPriv->y; min_y += dPriv->y; - max_y += dPriv->y + 1; + max_y += dPriv->y; } rmesa->state.scissor.rect.x1 = CLAMP(x1, min_x, max_x); @@ -1345,5 +1345,5 @@ void rcommonBeginBatch(radeonContextPtr rmesa, int n, void radeonUserClear(GLcontext *ctx, GLuint mask) { - _mesa_meta_clear(ctx, mask); + _mesa_meta_Clear(ctx, mask); } diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c index 71ee06d9a7..6b9b1e3c5e 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common_context.c +++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c @@ -47,7 +47,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "swrast_setup/swrast_setup.h" #include "tnl/tnl.h" -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) /* +r6/r7 */ +#if defined(RADEON_R600) #include "r600_context.h" #endif @@ -227,11 +227,8 @@ GLboolean radeonInitContext(radeonContextPtr radeon, fthrottle_mode = driQueryOptioni(&radeon->optionCache, "fthrottle_mode"); radeon->iw.irq_seq = -1; radeon->irqsEmitted = 0; - if (IS_R600_CLASS(radeon->radeonScreen)) - radeon->do_irqs = 0; - else - radeon->do_irqs = (fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS && - radeon->radeonScreen->irq); + radeon->do_irqs = (fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS && + radeon->radeonScreen->irq); radeon->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS); diff --git a/src/mesa/drivers/dri/radeon/radeon_debug.c b/src/mesa/drivers/dri/radeon/radeon_debug.c index 3b6f003803..413000b6c0 100644 --- a/src/mesa/drivers/dri/radeon/radeon_debug.c +++ b/src/mesa/drivers/dri/radeon/radeon_debug.c @@ -39,7 +39,7 @@ static const struct dri_debug_control debug_control[] = { {"fall", RADEON_FALLBACKS}, {"tex", RADEON_TEXTURE}, {"ioctl", RADEON_IOCTL}, - {"verts", RADEON_RENDER}, + {"verts", RADEON_VERTS}, {"render", RADEON_RENDER}, {"swrender", RADEON_SWRENDER}, {"state", RADEON_STATE}, diff --git a/src/mesa/drivers/dri/radeon/radeon_debug.h b/src/mesa/drivers/dri/radeon/radeon_debug.h index 2a8302293b..26da31c1c4 100644 --- a/src/mesa/drivers/dri/radeon/radeon_debug.h +++ b/src/mesa/drivers/dri/radeon/radeon_debug.h @@ -149,21 +149,22 @@ static inline void radeon_debug_remove_indent(void) } } + /* From http://gcc. gnu.org/onlinedocs/gcc-3.2.3/gcc/Variadic-Macros.html . I suppose we could inline this and use macro to fetch out __LINE__ and stuff in case we run into trouble with other compilers ... GLUE! */ -#define WARN_ONCE(a, ...) { \ - static int warn##__LINE__=1; \ - if(warn##__LINE__){ \ +#define WARN_ONCE(a, ...) do { \ + static int __warn_once=1; \ + if(__warn_once){ \ radeon_warning("*********************************WARN_ONCE*********************************\n"); \ radeon_warning("File %s function %s line %d\n", \ __FILE__, __FUNCTION__, __LINE__); \ radeon_warning( (a), ## __VA_ARGS__);\ radeon_warning("***************************************************************************\n"); \ - warn##__LINE__=0;\ + __warn_once=0;\ } \ - } + } while(0) #endif diff --git a/src/mesa/drivers/dri/radeon/radeon_dma.c b/src/mesa/drivers/dri/radeon/radeon_dma.c index 2eefa3f2b1..c6edbae9a1 100644 --- a/src/mesa/drivers/dri/radeon/radeon_dma.c +++ b/src/mesa/drivers/dri/radeon/radeon_dma.c @@ -207,7 +207,6 @@ again_alloc: counter on unused buffers for later freeing them from begin of list */ dma_bo = last_elem(&rmesa->dma.free); - assert(dma_bo->bo->cref == 1); remove_from_list(dma_bo); insert_at_head(&rmesa->dma.reserved, dma_bo); } @@ -263,7 +262,7 @@ void radeonAllocDmaRegion(radeonContextPtr rmesa, void radeonFreeDmaRegions(radeonContextPtr rmesa) { - struct radeon_dma_bo *dma_bo = CALLOC_STRUCT(radeon_dma_bo); + struct radeon_dma_bo *dma_bo; struct radeon_dma_bo *temp; if (RADEON_DEBUG & RADEON_DMA) fprintf(stderr, "%s\n", __FUNCTION__); @@ -307,6 +306,10 @@ static int radeon_bo_is_idle(struct radeon_bo* bo) WARN_ONCE("Your libdrm or kernel doesn't have support for busy query.\n" "This may cause small performance drop for you.\n"); } + /* Protect against bug in legacy bo handling that causes bos stay + * referenced even after they should be freed */ + if (bo->cref != 1) + return 0; return ret != -EBUSY; } @@ -343,7 +346,9 @@ void radeonReleaseDmaRegions(radeonContextPtr rmesa) foreach_s(dma_bo, temp, &rmesa->dma.wait) { if (dma_bo->expire_counter == time) { WARN_ONCE("Leaking dma buffer object!\n"); - radeon_bo_unref(dma_bo->bo); + /* force free of buffer so we don't realy start + * leaking stuff now*/ + while ((dma_bo->bo = radeon_bo_unref(dma_bo->bo))) {} remove_from_list(dma_bo); FREE(dma_bo); continue; diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c index d83b166742..7ac53ec0ca 100644 --- a/src/mesa/drivers/dri/radeon/radeon_fbo.c +++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c @@ -583,7 +583,7 @@ void radeon_fbo_init(struct radeon_context *radeon) radeon->glCtx->Driver.FinishRenderTexture = radeon_finish_render_texture; radeon->glCtx->Driver.ResizeBuffers = radeon_resize_buffers; radeon->glCtx->Driver.ValidateFramebuffer = radeon_validate_framebuffer; - radeon->glCtx->Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer; + radeon->glCtx->Driver.BlitFramebuffer = _mesa_meta_BlitFramebuffer; } diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c index 5ffb55db5e..573eb6c9c1 100644 --- a/src/mesa/drivers/dri/radeon/radeon_screen.c +++ b/src/mesa/drivers/dri/radeon/radeon_screen.c @@ -48,17 +48,17 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_screen.h" #include "radeon_common.h" #include "radeon_span.h" -#if !RADEON_COMMON +#if defined(RADEON_R100) #include "radeon_context.h" #include "radeon_tex.h" -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R200) #include "r200_context.h" #include "r200_ioctl.h" #include "r200_tex.h" -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#elif defined(RADEON_R300) #include "r300_context.h" #include "r300_tex.h" -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#elif defined(RADEON_R600) #include "r600_context.h" #include "r700_driconf.h" /* +r6/r7 */ #include "r600_tex.h" /* +r6/r7 */ @@ -82,7 +82,7 @@ DRI_CONF_OPT_BEGIN_V(command_buffer_size,int,def, # min ":" # max ) \ DRI_CONF_DESC(de,"Grösse des Befehlspuffers (in KB)") \ DRI_CONF_OPT_END -#if !RADEON_COMMON /* R100 */ +#if defined(RADEON_R100) /* R100 */ PUBLIC const char __driConfigOptions[] = DRI_CONF_BEGIN DRI_CONF_SECTION_PERFORMANCE @@ -109,7 +109,7 @@ DRI_CONF_BEGIN DRI_CONF_END; static const GLuint __driNConfigOptions = 15; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R200) PUBLIC const char __driConfigOptions[] = DRI_CONF_BEGIN @@ -147,7 +147,7 @@ extern const struct dri_extension NV_vp_extension[]; extern const struct dri_extension ATI_fs_extension[]; extern const struct dri_extension point_extensions[]; -#elif RADEON_COMMON && (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600)) +#elif defined(RADEON_R300) || defined(RADEON_R600) #define DRI_CONF_FP_OPTIMIZATION_SPEED 0 #define DRI_CONF_FP_OPTIMIZATION_QUALITY 1 @@ -220,7 +220,7 @@ static const GLuint __driNConfigOptions = 17; extern const struct dri_extension gl_20_extension[]; -#endif /* RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) */ +#endif extern const struct dri_extension card_extensions[]; extern const struct dri_extension mm_extensions[]; @@ -337,7 +337,7 @@ radeonFillInModes( __DRIscreenPrivate *psp, return (const __DRIconfig **) configs; } -#if !RADEON_COMMON +#if defined(RADEON_R100) static const __DRItexOffsetExtension radeonTexOffsetExtension = { { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION }, radeonSetTexOffset, @@ -350,7 +350,7 @@ static const __DRItexBufferExtension radeonTexBufferExtension = { }; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) static const __DRIallocateExtension r200AllocateExtension = { { __DRI_ALLOCATE, __DRI_ALLOCATE_VERSION }, r200AllocateMemoryMESA, @@ -370,7 +370,7 @@ static const __DRItexBufferExtension r200TexBufferExtension = { }; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#if defined(RADEON_R300) static const __DRItexOffsetExtension r300texOffsetExtension = { { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION }, r300SetTexOffset, @@ -383,7 +383,7 @@ static const __DRItexBufferExtension r300TexBufferExtension = { }; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#if defined(RADEON_R600) static const __DRItexOffsetExtension r600texOffsetExtension = { { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION }, r600SetTexOffset, /* +r6/r7 */ @@ -1222,22 +1222,22 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv ) screen->extensions[i++] = &driMediaStreamCounterExtension.base; } -#if !RADEON_COMMON +#if defined(RADEON_R100) screen->extensions[i++] = &radeonTexOffsetExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) if (IS_R200_CLASS(screen)) screen->extensions[i++] = &r200AllocateExtension.base; screen->extensions[i++] = &r200texOffsetExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#if defined(RADEON_R300) screen->extensions[i++] = &r300texOffsetExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#if defined(RADEON_R600) screen->extensions[i++] = &r600texOffsetExtension.base; #endif @@ -1376,22 +1376,22 @@ radeonCreateScreen2(__DRIscreenPrivate *sPriv) screen->extensions[i++] = &driMediaStreamCounterExtension.base; } -#if !RADEON_COMMON +#if defined(RADEON_R100) screen->extensions[i++] = &radeonTexBufferExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) if (IS_R200_CLASS(screen)) screen->extensions[i++] = &r200AllocateExtension.base; screen->extensions[i++] = &r200TexBufferExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#if defined(RADEON_R300) screen->extensions[i++] = &r300TexBufferExtension.base; #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#if defined(RADEON_R600) screen->extensions[i++] = &r600TexBufferExtension.base; #endif @@ -1589,22 +1589,22 @@ radeonDestroyBuffer(__DRIdrawablePrivate *driDrawPriv) static const __DRIconfig ** radeonInitScreen(__DRIscreenPrivate *psp) { -#if !RADEON_COMMON +#if defined(RADEON_R100) static const char *driver_name = "Radeon"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 1, 6, 0 }; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R200) static const char *driver_name = "R200"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 1, 6, 0 }; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#elif defined(RADEON_R300) static const char *driver_name = "R300"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 1, 24, 0 }; -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#elif defined(RADEON_R600) static const char *driver_name = "R600"; static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; @@ -1630,13 +1630,13 @@ radeonInitScreen(__DRIscreenPrivate *psp) * Hello chicken. Hello egg. How are you two today? */ driInitExtensions( NULL, card_extensions, GL_FALSE ); -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) driInitExtensions( NULL, blend_extensions, GL_FALSE ); driInitSingleExtension( NULL, ARB_vp_extension ); driInitSingleExtension( NULL, NV_vp_extension ); driInitSingleExtension( NULL, ATI_fs_extension ); driInitExtensions( NULL, point_extensions, GL_FALSE ); -#elif (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600)) +#elif (defined(RADEON_R300) || defined(RADEON_R600)) driInitSingleExtension( NULL, gl_20_extension ); #endif @@ -1684,13 +1684,13 @@ __DRIconfig **radeonInitScreen2(__DRIscreenPrivate *psp) */ driInitExtensions( NULL, card_extensions, GL_FALSE ); driInitExtensions( NULL, mm_extensions, GL_FALSE ); -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) driInitExtensions( NULL, blend_extensions, GL_FALSE ); driInitSingleExtension( NULL, ARB_vp_extension ); driInitSingleExtension( NULL, NV_vp_extension ); driInitSingleExtension( NULL, ATI_fs_extension ); driInitExtensions( NULL, point_extensions, GL_FALSE ); -#elif (defined(RADEON_COMMON_FOR_R300) || defined(RADEON_COMMON_FOR_R600)) +#elif (defined(RADEON_R300) || defined(RADEON_R600)) driInitSingleExtension( NULL, gl_20_extension ); #endif @@ -1772,13 +1772,13 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo ) const struct __DriverAPIRec driDriverAPI = { .InitScreen = radeonInitScreen, .DestroyScreen = radeonDestroyScreen, -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) .CreateContext = r200CreateContext, .DestroyContext = r200DestroyContext, -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) +#elif defined(RADEON_R600) .CreateContext = r600CreateContext, .DestroyContext = radeonDestroyContext, -#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) +#elif defined(RADEON_R300) .CreateContext = r300CreateContext, .DestroyContext = radeonDestroyContext, #else diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c index 4e100d854e..0c49c3713a 100644 --- a/src/mesa/drivers/dri/radeon/radeon_span.c +++ b/src/mesa/drivers/dri/radeon/radeon_span.c @@ -55,7 +55,7 @@ static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb); /* r200 depth buffer is always tiled - this is the formula according to the docs unless I typo'ed in it */ -#if defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) static GLubyte *r200_depth_2byte(const struct radeon_renderbuffer * rrb, GLint x, GLint y) { @@ -106,6 +106,141 @@ static GLubyte *r200_depth_4byte(const struct radeon_renderbuffer * rrb, } #endif +/* r600 tiling + * two main types: + * - 1D (akin to macro-linear/micro-tiled on older asics) + * - 2D (akin to macro-tiled/micro-tiled on older asics) + * only 1D tiling is implemented below + */ +#if defined(RADEON_R600) +static inline GLint r600_1d_tile_helper(const struct radeon_renderbuffer * rrb, + GLint x, GLint y, GLint is_depth, GLint is_stencil) +{ + GLint element_bytes = rrb->cpp; + GLint num_samples = 1; + GLint tile_width = 8; + GLint tile_height = 8; + GLint tile_thickness = 1; + GLint pitch_elements = rrb->pitch / element_bytes; + GLint height = rrb->base.Height; + GLint z = 0; + GLint sample_number = 0; + /* */ + GLint tile_bytes; + GLint tiles_per_row; + GLint tiles_per_slice; + GLint slice_offset; + GLint tile_row_index; + GLint tile_column_index; + GLint tile_offset; + GLint pixel_number = 0; + GLint element_offset; + GLint offset = 0; + + tile_bytes = tile_width * tile_height * tile_thickness * element_bytes * num_samples; + tiles_per_row = pitch_elements / tile_width; + tiles_per_slice = tiles_per_row * (height / tile_height); + slice_offset = (z / tile_thickness) * tiles_per_slice * tile_bytes; + tile_row_index = y / tile_height; + tile_column_index = x / tile_width; + tile_offset = ((tile_row_index * tiles_per_row) + tile_column_index) * tile_bytes; + + if (is_depth) { + GLint pixel_offset = 0; + + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((y >> 0) & 1) << 1; // pn[1] = y[0] + pixel_number |= ((x >> 1) & 1) << 2; // pn[2] = x[1] + pixel_number |= ((y >> 1) & 1) << 3; // pn[3] = y[1] + pixel_number |= ((x >> 2) & 1) << 4; // pn[4] = x[2] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + switch (element_bytes) { + case 2: + pixel_offset = pixel_number * element_bytes * num_samples; + break; + case 4: + /* stencil and depth data are stored separately within a tile. + * stencil is stored in a contiguous tile before the depth tile. + * stencil element is 1 byte, depth element is 3 bytes. + * stencil tile is 64 bytes. + */ + if (is_stencil) + pixel_offset = pixel_number * 1 * num_samples; + else + pixel_offset = (pixel_number * 3 * num_samples) + 64; + break; + } + element_offset = pixel_offset + (sample_number * element_bytes); + } else { + GLint sample_offset; + + switch (element_bytes) { + case 1: + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1] + pixel_number |= ((x >> 2) & 1) << 2; // pn[2] = x[2] + pixel_number |= ((y >> 1) & 1) << 3; // pn[3] = y[1] + pixel_number |= ((y >> 0) & 1) << 4; // pn[4] = y[0] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + break; + case 2: + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1] + pixel_number |= ((x >> 2) & 1) << 2; // pn[2] = x[2] + pixel_number |= ((y >> 0) & 1) << 3; // pn[3] = y[0] + pixel_number |= ((y >> 1) & 1) << 4; // pn[4] = y[1] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + break; + case 4: + pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0] + pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1] + pixel_number |= ((y >> 0) & 1) << 2; // pn[2] = y[0] + pixel_number |= ((x >> 2) & 1) << 3; // pn[3] = x[2] + pixel_number |= ((y >> 1) & 1) << 4; // pn[4] = y[1] + pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2] + break; + } + sample_offset = sample_number * (tile_bytes / num_samples); + element_offset = sample_offset + (pixel_number * element_bytes); + } + offset = slice_offset + tile_offset + element_offset; + return offset; +} + +/* depth buffers */ +static GLubyte *r600_ptr_depth(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + GLint offset = r600_1d_tile_helper(rrb, x, y, 1, 0); + return &ptr[offset]; +} + +static GLubyte *r600_ptr_stencil(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + GLint offset = r600_1d_tile_helper(rrb, x, y, 1, 1); + return &ptr[offset]; +} + +static GLubyte *r600_ptr_color(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + uint32_t mask = RADEON_BO_FLAGS_MACRO_TILE | RADEON_BO_FLAGS_MICRO_TILE; + GLint offset; + + if (rrb->has_surface || !(rrb->bo->flags & mask)) { + offset = x * rrb->cpp + y * rrb->pitch; + } else { + offset = r600_1d_tile_helper(rrb, x, y, 0, 0); + } + return &ptr[offset]; +} + +#else + /* radeon tiling on r300-r500 has 4 states, macro-linear/micro-linear macro-linear/micro-tiled @@ -197,7 +332,10 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb, return &ptr[offset]; } -#ifndef COMPILE_R300 +#endif + +#ifndef RADEON_R300 +#ifndef RADEON_R600 static uint32_t z24s8_to_s8z24(uint32_t val) { @@ -210,6 +348,7 @@ s8z24_to_z24s8(uint32_t val) return (val >> 24) | (val << 8); } #endif +#endif /* * Note that all information needed to access pixels in a renderbuffer @@ -270,7 +409,11 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_RGB565 #define TAG2(x,y) radeon##x##_RGB565##y +#if defined(RADEON_R600) +#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off) +#else #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off) +#endif #include "spantmp2.h" /* 16 bit, ARGB1555 color spanline and pixel functions @@ -280,7 +423,11 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_ARGB1555 #define TAG2(x,y) radeon##x##_ARGB1555##y +#if defined(RADEON_R600) +#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off) +#else #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off) +#endif #include "spantmp2.h" /* 16 bit, RGBA4 color spanline and pixel functions @@ -290,7 +437,11 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_ARGB4444 #define TAG2(x,y) radeon##x##_ARGB4444##y +#if defined(RADEON_R600) +#define GET_PTR(X,Y) r600_ptr_color(rrb, (X) + x_off, (Y) + y_off) +#else #define GET_PTR(X,Y) radeon_ptr_2byte_8x2(rrb, (X) + x_off, (Y) + y_off) +#endif #include "spantmp2.h" /* 32 bit, xRGB8888 color spanline and pixel functions @@ -300,11 +451,19 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_xRGB8888 #define TAG2(x,y) radeon##x##_xRGB8888##y +#if defined(RADEON_R600) +#define GET_VALUE(_x, _y) ((*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off)) | 0xff000000)) +#define PUT_VALUE(_x, _y, d) { \ + GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off ); \ + *_ptr = d; \ +} while (0) +#else #define GET_VALUE(_x, _y) ((*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) | 0xff000000)) #define PUT_VALUE(_x, _y, d) { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) +#endif #include "spantmp2.h" /* 32 bit, ARGB8888 color spanline and pixel functions @@ -314,11 +473,19 @@ s8z24_to_z24s8(uint32_t val) #define TAG(x) radeon##x##_ARGB8888 #define TAG2(x,y) radeon##x##_ARGB8888##y +#if defined(RADEON_R600) +#define GET_VALUE(_x, _y) (*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off))) +#define PUT_VALUE(_x, _y, d) { \ + GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off ); \ + *_ptr = d; \ +} while (0) +#else #define GET_VALUE(_x, _y) (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))) #define PUT_VALUE(_x, _y, d) { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) +#endif #include "spantmp2.h" /* ================================================================ @@ -339,17 +506,23 @@ s8z24_to_z24s8(uint32_t val) */ #define VALUE_TYPE GLushort -#if defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) #define WRITE_DEPTH( _x, _y, d ) \ *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) = d +#elif defined(RADEON_R600) +#define WRITE_DEPTH( _x, _y, d ) \ + *(GLushort *)r600_ptr_depth(rrb, _x + x_off, _y + y_off) = d #else #define WRITE_DEPTH( _x, _y, d ) \ *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) = d #endif -#if defined(RADEON_COMMON_FOR_R200) +#if defined(RADEON_R200) #define READ_DEPTH( d, _x, _y ) \ d = *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) +#elif defined(RADEON_R600) +#define READ_DEPTH( d, _x, _y ) \ + d = *(GLushort *)r600_ptr_depth(rrb, _x + x_off, _y + y_off) #else #define READ_DEPTH( d, _x, _y ) \ d = *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) @@ -365,7 +538,7 @@ s8z24_to_z24s8(uint32_t val) */ #define VALUE_TYPE GLuint -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -374,7 +547,16 @@ do { \ tmp |= ((d << 8) & 0xffffff00); \ *_ptr = tmp; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + tmp &= 0xff000000; \ + tmp |= ((d) & 0x00ffffff); \ + *_ptr = tmp; \ +} while (0) +#elif defined(RADEON_R200) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -394,12 +576,17 @@ do { \ } while (0) #endif -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define READ_DEPTH( d, _x, _y ) \ do { \ d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) & 0xffffff00) >> 8; \ }while(0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define READ_DEPTH( d, _x, _y ) \ + do { \ + d = (*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off)) & 0x00ffffff); \ + }while(0) +#elif defined(RADEON_R200) #define READ_DEPTH( d, _x, _y ) \ do { \ d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; \ @@ -420,13 +607,27 @@ do { \ */ #define VALUE_TYPE GLuint -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + tmp &= 0xff000000; \ + tmp |= (((d) >> 8) & 0x00ffffff); \ + *_ptr = tmp; \ + _ptr = (GLuint*)r600_ptr_stencil(rrb, _x + x_off, _y + y_off); \ + tmp = *_ptr; \ + tmp &= 0xffffff00; \ + tmp |= (d) & 0xff; \ + *_ptr = tmp; \ +} while (0) +#elif defined(RADEON_R200) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -442,12 +643,18 @@ do { \ } while (0) #endif -#if defined(COMPILE_R300) +#if defined(RADEON_R300) #define READ_DEPTH( d, _x, _y ) \ do { \ d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))); \ }while(0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define READ_DEPTH( d, _x, _y ) \ + do { \ + d = ((*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off))) << 8) & 0xffffff00; \ + d |= (*(GLuint*)(r600_ptr_stencil(rrb, _x + x_off, _y + y_off))) & 0x000000ff; \ + }while(0) +#elif defined(RADEON_R200) #define READ_DEPTH( d, _x, _y ) \ do { \ d = s8z24_to_z24s8(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))); \ @@ -467,7 +674,7 @@ do { \ /* 24 bit depth, 8 bit stencil depthbuffer functions */ -#ifdef COMPILE_R300 +#ifdef RADEON_R300 #define WRITE_STENCIL( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off); \ @@ -476,7 +683,16 @@ do { \ tmp |= (d) & 0xff; \ *_ptr = tmp; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define WRITE_STENCIL( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_stencil(rrb, _x + x_off, _y + y_off); \ + GLuint tmp = *_ptr; \ + tmp &= 0xffffff00; \ + tmp |= (d) & 0xff; \ + *_ptr = tmp; \ +} while (0) +#elif defined(RADEON_R200) #define WRITE_STENCIL( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte(rrb, _x + x_off, _y + y_off); \ @@ -496,14 +712,21 @@ do { \ } while (0) #endif -#ifdef COMPILE_R300 +#ifdef RADEON_R300 #define READ_STENCIL( d, _x, _y ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ GLuint tmp = *_ptr; \ d = tmp & 0x000000ff; \ } while (0) -#elif defined(RADEON_COMMON_FOR_R200) +#elif defined(RADEON_R600) +#define READ_STENCIL( d, _x, _y ) \ +do { \ + GLuint *_ptr = (GLuint*)r600_ptr_stencil( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + d = tmp & 0x000000ff; \ +} while (0) +#elif defined(RADEON_R200) #define READ_STENCIL( d, _x, _y ) \ do { \ GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c index 9d252aa74c..ae41b90efe 100644 --- a/src/mesa/drivers/dri/radeon/radeon_texstate.c +++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c @@ -833,11 +833,14 @@ static void import_tex_obj_state( r100ContextPtr rmesa, cmd[TEX_PP_TXFORMAT] |= texobj->pp_txformat & TEXOBJ_TXFORMAT_MASK; cmd[TEX_PP_BORDER_COLOR] = texobj->pp_border_color; - if (texobj->base.Target == GL_TEXTURE_RECTANGLE_NV) { - GLuint *txr_cmd = RADEON_DB_STATE( txr[unit] ); + if (texobj->pp_txformat & RADEON_TXFORMAT_NON_POWER2) { + uint32_t *txr_cmd = &rmesa->hw.txr[unit].cmd[TXR_CMD_0]; txr_cmd[TXR_PP_TEX_SIZE] = texobj->pp_txsize; /* NPOT only! */ txr_cmd[TXR_PP_TEX_PITCH] = texobj->pp_txpitch; /* NPOT only! */ - RADEON_DB_STATECHANGE( rmesa, &rmesa->hw.txr[unit] ); + RADEON_STATECHANGE( rmesa, txr[unit] ); + } + + if (texobj->base.Target == GL_TEXTURE_RECTANGLE_NV) { se_coord_fmt |= RADEON_VTX_ST0_NONPARAMETRIC << unit; } else { @@ -1114,7 +1117,6 @@ static GLboolean radeon_validate_texture(GLcontext *ctx, struct gl_texture_objec RADEON_STATECHANGE( rmesa, ctx ); rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= (RADEON_TEX_0_ENABLE | RADEON_TEX_BLEND_0_ENABLE) << unit; - RADEON_STATECHANGE( rmesa, tcl ); rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_ST_BIT(unit); diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c index fad3d1ceda..049284ef8c 100644 --- a/src/mesa/drivers/dri/radeon/radeon_texture.c +++ b/src/mesa/drivers/dri/radeon/radeon_texture.c @@ -659,11 +659,6 @@ static void radeon_teximage( if (dims == 3) _mesa_free(dstImageOffsets); } - - /* SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - radeon_generate_mipmap(ctx, target, texObj); - } } _mesa_unmap_teximage_pbo(ctx, packing); @@ -792,11 +787,6 @@ static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int leve format, type, pixels, packing)) _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage"); } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - radeon_generate_mipmap(ctx, target, texObj); - } } radeon_teximage_unmap(image); diff --git a/src/mesa/drivers/dri/s3v/s3v_tex.c b/src/mesa/drivers/dri/s3v/s3v_tex.c index 9b92519862..ec1182f34f 100644 --- a/src/mesa/drivers/dri/s3v/s3v_tex.c +++ b/src/mesa/drivers/dri/s3v/s3v_tex.c @@ -536,24 +536,13 @@ void s3vInitTextureFuncs( GLcontext *ctx ) #endif ctx->Driver.TexEnv = s3vTexEnv; - ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format; - ctx->Driver.TexImage1D = _mesa_store_teximage1d; ctx->Driver.TexImage2D = s3vTexImage2D; - ctx->Driver.TexImage3D = _mesa_store_teximage3d; - ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d; ctx->Driver.TexSubImage2D = s3vTexSubImage2D; - ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; - ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; - ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; - ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; - ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; - ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; ctx->Driver.BindTexture = s3vBindTexture; ctx->Driver.DeleteTexture = s3vDeleteTexture; ctx->Driver.TexParameter = s3vTexParameter; ctx->Driver.UpdateTexturePalette = 0; ctx->Driver.IsTextureResident = s3vIsTextureResident; - ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; s3vInitTextureObjects( ctx ); } diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c index b36cd0274a..f4947daa06 100644 --- a/src/mesa/drivers/dri/swrast/swrast.c +++ b/src/mesa/drivers/dri/swrast/swrast.c @@ -43,6 +43,7 @@ #include "tnl/t_pipeline.h" #include "vbo/vbo.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "utils.h" #include "swrast_priv.h" @@ -650,6 +651,8 @@ driCreateNewContext(__DRIscreen *screen, const __DRIconfig *config, _mesa_enable_2_0_extensions(mesaCtx); _mesa_enable_2_1_extensions(mesaCtx); + _mesa_meta_init(mesaCtx); + return ctx; } @@ -661,6 +664,7 @@ driDestroyContext(__DRIcontext *ctx) if (ctx) { mesaCtx = &ctx->Base; + _mesa_meta_free(mesaCtx); _swsetup_DestroyContext( mesaCtx ); _swrast_DestroyContext( mesaCtx ); _tnl_DestroyContext( mesaCtx ); diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c index 1f7257eaea..f6a48b3ae1 100644 --- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c +++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c @@ -176,6 +176,54 @@ logbase2(int n) } +static void +tdfxGenerateMipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) +{ + GLint mipWidth, mipHeight; + tdfxMipMapLevel *mip; + struct gl_texture_image *mipImage; /* the new/next image */ + struct gl_texture_image *texImage; + const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target); + GLint level = texObj->BaseLevel; + GLsizei width, height, texelBytes; + const tdfxMipMapLevel *mml; + + texImage = _mesa_get_tex_image(ctx, texObj, target, level); + assert(!texImage->IsCompressed); + + mml = TDFX_TEXIMAGE_DATA(texImage); + + width = texImage->Width; + height = texImage->Height; + while (level < texObj->MaxLevel && level < maxLevels - 1) { + mipWidth = width / 2; + if (!mipWidth) { + mipWidth = 1; + } + mipHeight = height / 2; + if (!mipHeight) { + mipHeight = 1; + } + if ((mipWidth == width) && (mipHeight == height)) { + break; + } + ++level; + mipImage = _mesa_select_tex_image(ctx, texObj, target, level); + mip = TDFX_TEXIMAGE_DATA(mipImage); + _mesa_halve2x2_teximage2d(ctx, + texImage, + texelBytes, + mml->width, mml->height, + texImage->Data, mipImage->Data); + texImage = mipImage; + mml = mip; + width = mipWidth; + height = mipHeight; + } +} + + /* * Compute various texture image parameters. * Input: w, h - source texture width and height @@ -1397,45 +1445,6 @@ tdfxTexImage2D(GLcontext *ctx, GLenum target, GLint level, width, height, 1, format, type, pixels, packing); } - - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - GLint mipWidth, mipHeight; - tdfxMipMapLevel *mip; - struct gl_texture_image *mipImage; - const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target); - - assert(!texImage->IsCompressed); - - while (level < texObj->MaxLevel && level < maxLevels - 1) { - mipWidth = width / 2; - if (!mipWidth) { - mipWidth = 1; - } - mipHeight = height / 2; - if (!mipHeight) { - mipHeight = 1; - } - if ((mipWidth == width) && (mipHeight == height)) { - break; - } - _mesa_TexImage2D(target, ++level, internalFormat, - mipWidth, mipHeight, border, - format, type, - NULL); - mipImage = _mesa_select_tex_image(ctx, texObj, target, level); - mip = TDFX_TEXIMAGE_DATA(mipImage); - _mesa_halve2x2_teximage2d(ctx, - texImage, - texelBytes, - mml->width, mml->height, - texImage->Data, mipImage->Data); - texImage = mipImage; - mml = mip; - width = mipWidth; - height = mipHeight; - } - } } RevalidateTexture(ctx, texObj); @@ -1507,44 +1516,6 @@ tdfxTexSubImage2D(GLcontext *ctx, GLenum target, GLint level, format, type, pixels, packing); } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - GLint mipWidth, mipHeight; - tdfxMipMapLevel *mip; - struct gl_texture_image *mipImage; - const GLint maxLevels = _mesa_max_texture_levels(ctx, texObj->Target); - - assert(!texImage->IsCompressed); - - width = texImage->Width; - height = texImage->Height; - while (level < texObj->MaxLevel && level < maxLevels - 1) { - mipWidth = width / 2; - if (!mipWidth) { - mipWidth = 1; - } - mipHeight = height / 2; - if (!mipHeight) { - mipHeight = 1; - } - if ((mipWidth == width) && (mipHeight == height)) { - break; - } - ++level; - mipImage = _mesa_select_tex_image(ctx, texObj, target, level); - mip = TDFX_TEXIMAGE_DATA(mipImage); - _mesa_halve2x2_teximage2d(ctx, - texImage, - texelBytes, - mml->width, mml->height, - texImage->Data, mipImage->Data); - texImage = mipImage; - mml = mip; - width = mipWidth; - height = mipHeight; - } - } - ti->reloadImages = GL_TRUE; /* signal the image needs to be reloaded */ fxMesa->new_state |= TDFX_NEW_TEXTURE; /* XXX this might be a bit much */ } @@ -1703,11 +1674,6 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target, MEMCPY(texImage->Data, data, texImage->CompressedSize); } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - assert(!texImage->IsCompressed); - } - RevalidateTexture(ctx, texObj); ti->reloadImages = GL_TRUE; @@ -1770,11 +1736,6 @@ tdfxCompressedTexSubImage2D( GLcontext *ctx, GLenum target, texImage->Data); } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - assert(!texImage->IsCompressed); - } - RevalidateTexture(ctx, texObj); ti->reloadImages = GL_TRUE; @@ -1914,4 +1875,5 @@ void tdfxInitTextureFuncs( struct dd_function_table *functions ) functions->CompressedTexImage2D = tdfxCompressedTexImage2D; functions->CompressedTexSubImage2D = tdfxCompressedTexSubImage2D; functions->UpdateTexturePalette = tdfxUpdateTexturePalette; + functions->GenerateMipmap = tdfxGenerateMipmap; } diff --git a/src/mesa/drivers/dri/unichrome/via_tex.c b/src/mesa/drivers/dri/unichrome/via_tex.c index d2010f0907..54073e7691 100644 --- a/src/mesa/drivers/dri/unichrome/via_tex.c +++ b/src/mesa/drivers/dri/unichrome/via_tex.c @@ -818,11 +818,6 @@ static void viaTexImage(GLcontext *ctx, } } - /* GL_SGIS_generate_mipmap */ - if (level == texObj->BaseLevel && texObj->GenerateMipmap) { - _mesa_generate_mipmap(ctx, target, texObj); - } - _mesa_unmap_teximage_pbo(ctx, packing); } diff --git a/src/mesa/drivers/osmesa/osmesa.c b/src/mesa/drivers/osmesa/osmesa.c index 904659e345..692657a5df 100644 --- a/src/mesa/drivers/osmesa/osmesa.c +++ b/src/mesa/drivers/osmesa/osmesa.c @@ -50,6 +50,7 @@ #include "tnl/t_context.h" #include "tnl/t_pipeline.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "vbo/vbo.h" @@ -1258,6 +1259,8 @@ OSMesaCreateContextExt( GLenum format, GLint depthBits, GLint stencilBits, osmesa->bInd = bind; osmesa->aInd = aind; + _mesa_meta_init(&osmesa->mesa); + /* Initialize the software rasterizer and helper modules. */ { GLcontext *ctx = &osmesa->mesa; @@ -1304,6 +1307,8 @@ OSMesaDestroyContext( OSMesaContext osmesa ) if (osmesa->rb) _mesa_reference_renderbuffer(&osmesa->rb, NULL); + _mesa_meta_free( &osmesa->mesa ); + _swsetup_DestroyContext( &osmesa->mesa ); _tnl_DestroyContext( &osmesa->mesa ); _vbo_DestroyContext( &osmesa->mesa ); diff --git a/src/mesa/drivers/windows/gdi/mesa.def b/src/mesa/drivers/windows/gdi/mesa.def index bd3e5b2137..5abcd1d927 100644 --- a/src/mesa/drivers/windows/gdi/mesa.def +++ b/src/mesa/drivers/windows/gdi/mesa.def @@ -943,6 +943,15 @@ EXPORTS _mesa_update_framebuffer_visual _mesa_use_program _mesa_Viewport + _mesa_meta_CopyColorSubTable + _mesa_meta_CopyColorTable + _mesa_meta_CopyConvolutionFilter1D + _mesa_meta_CopyConvolutionFilter2D + _mesa_meta_CopyTexImage1D + _mesa_meta_CopyTexImage2D + _mesa_meta_CopyTexSubImage1D + _mesa_meta_CopyTexSubImage2D + _mesa_meta_CopyTexSubImage3D _mesa_wait_query _swrast_Accum _swrast_Bitmap @@ -953,15 +962,6 @@ EXPORTS _swrast_Clear _swrast_choose_line _swrast_choose_triangle - _swrast_CopyColorSubTable - _swrast_CopyColorTable - _swrast_CopyConvolutionFilter1D - _swrast_CopyConvolutionFilter2D - _swrast_copy_teximage1d - _swrast_copy_teximage2d - _swrast_copy_texsubimage1d - _swrast_copy_texsubimage2d - _swrast_copy_texsubimage3d _swrast_CreateContext _swrast_DestroyContext _swrast_exec_fragment_program diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c index e1971db693..8929b22af1 100644 --- a/src/mesa/drivers/windows/gdi/wmesa.c +++ b/src/mesa/drivers/windows/gdi/wmesa.c @@ -12,6 +12,7 @@ #include "framebuffer.h" #include "renderbuffer.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "vbo/vbo.h" #include "swrast/swrast.h" #include "swrast_setup/swrast_setup.h" @@ -1515,6 +1516,8 @@ WMesaContext WMesaCreateContext(HDC hDC, _mesa_enable_2_0_extensions(ctx); _mesa_enable_2_1_extensions(ctx); + _mesa_meta_init(ctx); + /* Initialize the software rasterizer and helper modules. */ if (!_swrast_CreateContext(ctx) || !_vbo_CreateContext(ctx) || @@ -1558,6 +1561,8 @@ void WMesaDestroyContext( WMesaContext pwc ) DeleteObject(pwc->clearPen); DeleteObject(pwc->clearBrush); + _mesa_meta_free(ctx); + _swsetup_DestroyContext(ctx); _tnl_DestroyContext(ctx); _vbo_DestroyContext(ctx); diff --git a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c index 342a742867..7ac425a109 100644 --- a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c +++ b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c @@ -1346,6 +1346,8 @@ static void wmesa_update_state_first_time( struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference( ctx ); TNLcontext *tnl = TNL_CONTEXT(ctx); + _mesa_init_driver_functions(&ctx->Driver); + /* * XXX these function pointers could be initialized just once during * context creation since they don't depend on any state changes. @@ -1362,8 +1364,6 @@ static void wmesa_update_state_first_time( ctx->Driver.Viewport = wmesa_viewport; - ctx->Driver.Accum = _swrast_Accum; - ctx->Driver.Bitmap = _swrast_Bitmap; ctx->Driver.Clear = clear; ctx->Driver.Flush = flush; @@ -1371,28 +1371,6 @@ static void wmesa_update_state_first_time( ctx->Driver.ClearColor = clear_color; ctx->Driver.Enable = enable; - ctx->Driver.CopyPixels = _swrast_CopyPixels; - ctx->Driver.DrawPixels = _swrast_DrawPixels; - ctx->Driver.ReadPixels = _swrast_ReadPixels; - - ctx->Driver.ChooseTextureFormat = _mesa_choose_tex_format; - ctx->Driver.TexImage1D = _mesa_store_teximage1d; - ctx->Driver.TexImage2D = _mesa_store_teximage2d; - ctx->Driver.TexImage3D = _mesa_store_teximage3d; - ctx->Driver.TexSubImage1D = _mesa_store_texsubimage1d; - ctx->Driver.TexSubImage2D = _mesa_store_texsubimage2d; - ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; - ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; - - ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; - ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; - ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; - ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; - ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; - ctx->Driver.CopyColorTable = _swrast_CopyColorTable; - ctx->Driver.CopyColorSubTable = _swrast_CopyColorSubTable; - ctx->Driver.CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; - ctx->Driver.CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; // Does not apply for Mesa 5.x //ctx->Driver.BaseCompressedTexFormat = _mesa_base_compressed_texformat; diff --git a/src/mesa/drivers/windows/icd/mesa.def b/src/mesa/drivers/windows/icd/mesa.def index 465b380a0c..25ac08a2f0 100644 --- a/src/mesa/drivers/windows/icd/mesa.def +++ b/src/mesa/drivers/windows/icd/mesa.def @@ -75,6 +75,15 @@ EXPORTS _mesa_strcmp _mesa_test_proxy_teximage _mesa_Viewport + _mesa_meta_CopyColorSubTable + _mesa_meta_CopyColorTable + _mesa_meta_CopyConvolutionFilter1D + _mesa_meta_CopyConvolutionFilter2D + _mesa_meta_CopyTexImage1D + _mesa_meta_CopyTexImage2D + _mesa_meta_CopyTexSubImage1D + _mesa_meta_CopyTexSubImage2D + _mesa_meta_CopyTexSubImage3D _swrast_Accum _swrast_Bitmap _swrast_CopyPixels @@ -84,15 +93,6 @@ EXPORTS _swrast_Clear _swrast_choose_line _swrast_choose_triangle - _swrast_CopyColorSubTable - _swrast_CopyColorTable - _swrast_CopyConvolutionFilter1D - _swrast_CopyConvolutionFilter2D - _swrast_copy_teximage1d - _swrast_copy_teximage2d - _swrast_copy_texsubimage1d - _swrast_copy_texsubimage2d - _swrast_copy_texsubimage3d _swrast_CreateContext _swrast_DestroyContext _swrast_InvalidateState diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c index 662c61ae7e..79b058634c 100644 --- a/src/mesa/drivers/x11/xm_api.c +++ b/src/mesa/drivers/x11/xm_api.c @@ -1648,8 +1648,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list ) xmesa_register_swrast_functions( mesaCtx ); _swsetup_Wakeup(mesaCtx); - if (TEST_META_FUNCS) - _mesa_meta_init(mesaCtx); + _mesa_meta_init(mesaCtx); return c; } @@ -1665,8 +1664,7 @@ void XMesaDestroyContext( XMesaContext c ) FXdestroyContext( XMESA_BUFFER(mesaCtx->DrawBuffer) ); #endif - if (TEST_META_FUNCS) - _mesa_meta_free( mesaCtx ); + _mesa_meta_free( mesaCtx ); _swsetup_DestroyContext( mesaCtx ); _swrast_DestroyContext( mesaCtx ); diff --git a/src/mesa/drivers/x11/xm_dd.c b/src/mesa/drivers/x11/xm_dd.c index 4e9c001cc7..e2d4aa9b2d 100644 --- a/src/mesa/drivers/x11/xm_dd.c +++ b/src/mesa/drivers/x11/xm_dd.c @@ -1150,11 +1150,11 @@ xmesa_init_driver_functions( XMesaVisual xmvisual, driver->Enable = enable; driver->Viewport = xmesa_viewport; if (TEST_META_FUNCS) { - driver->Clear = _mesa_meta_clear; - driver->CopyPixels = _mesa_meta_copy_pixels; - driver->BlitFramebuffer = _mesa_meta_blit_framebuffer; - driver->DrawPixels = _mesa_meta_draw_pixels; - driver->Bitmap = _mesa_meta_bitmap; + driver->Clear = _mesa_meta_Clear; + driver->CopyPixels = _mesa_meta_CopyPixels; + driver->BlitFramebuffer = _mesa_meta_BlitFramebuffer; + driver->DrawPixels = _mesa_meta_DrawPixels; + driver->Bitmap = _mesa_meta_Bitmap; } else { driver->Clear = clear_buffers; |