X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fstate_tracker%2Fst_cb_drawpixels.c;h=9a3f22465c6dd5418043185cb8bdd867bc356f07;hb=3e3737e9e2548ae751236067cca2a79ad98dc2dd;hp=d80c068ea813546525eaa40dad34c7533521c352;hpb=b2ddb93ff3b8c88682634ccdef247967e31fab84;p=mesa.git diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index d80c068ea81..9a3f22465c6 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -33,9 +33,15 @@ #include "main/imports.h" #include "main/image.h" #include "main/bufferobj.h" +#include "main/format_pack.h" #include "main/macros.h" +#include "main/mfeatures.h" +#include "main/mtypes.h" #include "main/pack.h" +#include "main/pbo.h" +#include "main/readpix.h" #include "main/texformat.h" +#include "main/teximage.h" #include "main/texstore.h" #include "program/program.h" #include "program/prog_print.h" @@ -60,6 +66,7 @@ #include "util/u_inlines.h" #include "util/u_math.h" #include "util/u_tile.h" +#include "util/u_upload_mgr.h" #include "cso_cache/cso_context.h" @@ -90,92 +97,109 @@ is_passthrough_program(const struct gl_fragment_program *prog) } +/** + * Returns a fragment program which implements the current pixel transfer ops. + */ +static struct gl_fragment_program * +get_glsl_pixel_transfer_program(struct st_context *st, + struct st_fragment_program *orig) +{ + int pixelMaps = 0, scaleAndBias = 0; + struct gl_context *ctx = st->ctx; + struct st_fragment_program *fp = (struct st_fragment_program *) + ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 0); + + if (!fp) + return NULL; + + if (ctx->Pixel.RedBias != 0.0 || ctx->Pixel.RedScale != 1.0 || + ctx->Pixel.GreenBias != 0.0 || ctx->Pixel.GreenScale != 1.0 || + ctx->Pixel.BlueBias != 0.0 || ctx->Pixel.BlueScale != 1.0 || + ctx->Pixel.AlphaBias != 0.0 || ctx->Pixel.AlphaScale != 1.0) { + scaleAndBias = 1; + } + + pixelMaps = ctx->Pixel.MapColorFlag; + + if (pixelMaps) { + /* create the colormap/texture now if not already done */ + if (!st->pixel_xfer.pixelmap_texture) { + st->pixel_xfer.pixelmap_texture = st_create_color_map_texture(ctx); + st->pixel_xfer.pixelmap_sampler_view = + st_create_texture_sampler_view(st->pipe, + st->pixel_xfer.pixelmap_texture); + } + } + + get_pixel_transfer_visitor(fp, orig->glsl_to_tgsi, + scaleAndBias, pixelMaps); + + return &fp->Base; +} + /** * Make fragment shader for glDraw/CopyPixels. This shader is made * by combining the pixel transfer shader with the user-defined shader. - * \return pointer to Gallium driver fragment shader + * \param fpIn the current/incoming fragment program + * \param fpOut returns the combined fragment program */ -static void * -combined_drawpix_fragment_program(struct gl_context *ctx) +void +st_make_drawpix_fragment_program(struct st_context *st, + struct gl_fragment_program *fpIn, + struct gl_fragment_program **fpOut) { - struct st_context *st = st_context(ctx); - struct st_fragment_program *stfp; + struct gl_program *newProg; + struct st_fragment_program *stfp = (struct st_fragment_program *) fpIn; - if (st->pixel_xfer.program->serialNo == st->pixel_xfer.xfer_prog_sn - && st->fp->serialNo == st->pixel_xfer.user_prog_sn) { - /* the pixel tranfer program has not changed and the user-defined - * program has not changed, so re-use the combined program. - */ - stfp = st->pixel_xfer.combined_prog; + if (is_passthrough_program(fpIn)) { + newProg = (struct gl_program *) _mesa_clone_fragment_program(st->ctx, + &st->pixel_xfer.program->Base); + } + else if (stfp->glsl_to_tgsi != NULL) { + newProg = (struct gl_program *) get_glsl_pixel_transfer_program(st, stfp); } else { - /* Concatenate the pixel transfer program with the current user- - * defined program. - */ - if (is_passthrough_program(&st->fp->Base)) { - stfp = (struct st_fragment_program *) - _mesa_clone_fragment_program(ctx, &st->pixel_xfer.program->Base); - } - else { #if 0 - printf("Base program:\n"); - _mesa_print_program(&st->fp->Base.Base); - printf("DrawPix program:\n"); - _mesa_print_program(&st->pixel_xfer.program->Base.Base); + /* debug */ + printf("Base program:\n"); + _mesa_print_program(&fpIn->Base); + printf("DrawPix program:\n"); + _mesa_print_program(&st->pixel_xfer.program->Base.Base); #endif - stfp = (struct st_fragment_program *) - _mesa_combine_programs(ctx, - &st->pixel_xfer.program->Base.Base, - &st->fp->Base.Base); - } + newProg = _mesa_combine_programs(st->ctx, + &st->pixel_xfer.program->Base.Base, + &fpIn->Base); + } #if 0 - { - struct gl_program *p = &stfp->Base.Base; - printf("Combined DrawPixels program:\n"); - _mesa_print_program(p); - printf("InputsRead: 0x%x\n", p->InputsRead); - printf("OutputsWritten: 0x%x\n", p->OutputsWritten); - _mesa_print_parameter_list(p->Parameters); - } + /* debug */ + printf("Combined DrawPixels program:\n"); + _mesa_print_program(newProg); + printf("InputsRead: 0x%x\n", newProg->InputsRead); + printf("OutputsWritten: 0x%x\n", newProg->OutputsWritten); + _mesa_print_parameter_list(newProg->Parameters); #endif - /* translate to TGSI tokens */ - st_translate_fragment_program(st, stfp); - - /* save new program, update serial numbers */ - st->pixel_xfer.xfer_prog_sn = st->pixel_xfer.program->serialNo; - st->pixel_xfer.user_prog_sn = st->fp->serialNo; - st->pixel_xfer.combined_prog_sn = stfp->serialNo; - /* can't reference new program directly, already have a reference on it */ - st_reference_fragprog(st, &st->pixel_xfer.combined_prog, NULL); - st->pixel_xfer.combined_prog = stfp; - } - - /* Ideally we'd have updated the pipe constants during the normal - * st/atom mechanism. But we can't since this is specific to glDrawPixels. - */ - st_upload_constants(st, stfp->Base.Base.Parameters, PIPE_SHADER_FRAGMENT); - - return stfp->driver_shader; + *fpOut = (struct gl_fragment_program *) newProg; } /** - * Create fragment shader that does a TEX() instruction to get a Z and/or + * Create fragment program that does a TEX() instruction to get a Z and/or * stencil value value, then writes to FRAG_RESULT_DEPTH/FRAG_RESULT_STENCIL. * Used for glDrawPixels(GL_DEPTH_COMPONENT / GL_STENCIL_INDEX). * Pass fragment color through as-is. - * \return pointer to the Gallium driver fragment shader + * \return pointer to the gl_fragment program */ -static void * -make_fragment_shader_z_stencil(struct st_context *st, GLboolean write_depth, - GLboolean write_stencil) +struct gl_fragment_program * +st_make_drawpix_z_stencil_program(struct st_context *st, + GLboolean write_depth, + GLboolean write_stencil) { struct gl_context *ctx = st->ctx; struct gl_program *p; - struct st_fragment_program *stp; + struct gl_fragment_program *fp; GLuint ic = 0; const GLuint shaderIndex = write_depth * 2 + write_stencil; @@ -183,7 +207,7 @@ make_fragment_shader_z_stencil(struct st_context *st, GLboolean write_depth, if (st->drawpix.shaders[shaderIndex]) { /* already have the proper shader */ - return st->drawpix.shaders[shaderIndex]->driver_shader; + return st->drawpix.shaders[shaderIndex]; } /* @@ -237,26 +261,23 @@ make_fragment_shader_z_stencil(struct st_context *st, GLboolean write_depth, p->InputsRead = FRAG_BIT_TEX0 | FRAG_BIT_COL0; p->OutputsWritten = 0; if (write_depth) - p->OutputsWritten |= (1 << FRAG_RESULT_DEPTH); + p->OutputsWritten |= BITFIELD64_BIT(FRAG_RESULT_DEPTH); if (write_stencil) - p->OutputsWritten |= (1 << FRAG_RESULT_STENCIL); + p->OutputsWritten |= BITFIELD64_BIT(FRAG_RESULT_STENCIL); p->SamplersUsed = 0x1; /* sampler 0 (bit 0) is used */ if (write_stencil) p->SamplersUsed |= 1 << 1; - stp = st_fragment_program((struct gl_fragment_program *) p); + fp = (struct gl_fragment_program *) p; /* save the new shader */ - st->drawpix.shaders[shaderIndex] = stp; - - st_translate_fragment_program(st, stp); + st->drawpix.shaders[shaderIndex] = fp; - return stp->driver_shader; + return fp; } - /** * Create a simple vertex shader that just passes through the * vertex position and texcoord (and optionally, color). @@ -298,40 +319,45 @@ make_passthrough_vertex_shader(struct st_context *st, } -/** - * Return a texture base format for drawing/copying an image - * of the given format. - */ -static GLenum -base_format(GLenum format) -{ - switch (format) { - case GL_DEPTH_COMPONENT: - return GL_DEPTH_COMPONENT; - case GL_DEPTH_STENCIL: - return GL_DEPTH_STENCIL; - case GL_STENCIL_INDEX: - return GL_STENCIL_INDEX; - default: - return GL_RGBA; - } -} - - /** * Return a texture internalFormat for drawing/copying an image * of the given format and type. */ static GLenum -internal_format(GLenum format, GLenum type) +internal_format(struct gl_context *ctx, GLenum format, GLenum type) { switch (format) { case GL_DEPTH_COMPONENT: - return GL_DEPTH_COMPONENT; + switch (type) { + case GL_UNSIGNED_SHORT: + return GL_DEPTH_COMPONENT16; + + case GL_UNSIGNED_INT: + return GL_DEPTH_COMPONENT32; + + case GL_FLOAT: + if (ctx->Extensions.ARB_depth_buffer_float) + return GL_DEPTH_COMPONENT32F; + else + return GL_DEPTH_COMPONENT; + + default: + return GL_DEPTH_COMPONENT; + } + case GL_DEPTH_STENCIL: - return GL_DEPTH_STENCIL; + switch (type) { + case GL_FLOAT_32_UNSIGNED_INT_24_8_REV: + return GL_DEPTH32F_STENCIL8; + + case GL_UNSIGNED_INT_24_8: + default: + return GL_DEPTH24_STENCIL8; + } + case GL_STENCIL_INDEX: return GL_STENCIL_INDEX; + default: if (_mesa_is_integer_format(format)) { switch (type) { @@ -353,7 +379,61 @@ internal_format(GLenum format, GLenum type) } } else { - return GL_RGBA; + switch (type) { + case GL_UNSIGNED_BYTE: + case GL_UNSIGNED_INT_8_8_8_8: + case GL_UNSIGNED_INT_8_8_8_8_REV: + default: + return GL_RGBA8; + + case GL_UNSIGNED_BYTE_3_3_2: + case GL_UNSIGNED_BYTE_2_3_3_REV: + case GL_UNSIGNED_SHORT_4_4_4_4: + case GL_UNSIGNED_SHORT_4_4_4_4_REV: + return GL_RGBA4; + + case GL_UNSIGNED_SHORT_5_6_5: + case GL_UNSIGNED_SHORT_5_6_5_REV: + case GL_UNSIGNED_SHORT_5_5_5_1: + case GL_UNSIGNED_SHORT_1_5_5_5_REV: + return GL_RGB5_A1; + + case GL_UNSIGNED_INT_10_10_10_2: + case GL_UNSIGNED_INT_2_10_10_10_REV: + return GL_RGB10_A2; + + case GL_UNSIGNED_SHORT: + case GL_UNSIGNED_INT: + return GL_RGBA16; + + case GL_BYTE: + return + ctx->Extensions.EXT_texture_snorm ? GL_RGBA8_SNORM : GL_RGBA8; + + case GL_SHORT: + case GL_INT: + return + ctx->Extensions.EXT_texture_snorm ? GL_RGBA16_SNORM : GL_RGBA16; + + case GL_HALF_FLOAT_ARB: + return + ctx->Extensions.ARB_texture_float ? GL_RGBA16F : + ctx->Extensions.EXT_texture_snorm ? GL_RGBA16_SNORM : GL_RGBA16; + + case GL_FLOAT: + case GL_DOUBLE: + return + ctx->Extensions.ARB_texture_float ? GL_RGBA32F : + ctx->Extensions.EXT_texture_snorm ? GL_RGBA16_SNORM : GL_RGBA16; + + case GL_UNSIGNED_INT_5_9_9_9_REV: + assert(ctx->Extensions.EXT_texture_shared_exponent); + return GL_RGB9_E5; + + case GL_UNSIGNED_INT_10F_11F_11F_REV: + assert(ctx->Extensions.EXT_packed_float); + return GL_R11F_G11F_B10F; + } } } } @@ -371,7 +451,7 @@ alloc_texture(struct st_context *st, GLsizei width, GLsizei height, struct pipe_resource *pt; pt = st_texture_create(st, st->internal_target, texFormat, 0, - width, height, 1, PIPE_BIND_SAMPLER_VIEW); + width, height, 1, 1, PIPE_BIND_SAMPLER_VIEW); return pt; } @@ -392,11 +472,10 @@ make_texture(struct st_context *st, gl_format mformat; struct pipe_resource *pt; enum pipe_format pipeFormat; - GLuint cpp; - GLenum baseFormat, intFormat; + GLenum baseInternalFormat, intFormat; - baseFormat = base_format(format); - intFormat = internal_format(format, type); + intFormat = internal_format(ctx, format, type); + baseInternalFormat = _mesa_base_tex_format(ctx, intFormat); mformat = st_ChooseTextureFormat_renderable(ctx, intFormat, format, type, GL_FALSE); @@ -404,7 +483,6 @@ make_texture(struct st_context *st, pipeFormat = st_mesa_format_to_pipe_format(mformat); assert(pipeFormat); - cpp = util_format_get_blocksize(pipeFormat); pixels = _mesa_map_pbo_source(ctx, unpack, pixels); if (!pixels) @@ -419,7 +497,6 @@ make_texture(struct st_context *st, { struct pipe_transfer *transfer; - static const GLuint dstImageOffsets = 0; GLboolean success; GLubyte *dest; const GLbitfield imageTransferStateSave = ctx->_ImageTransferState; @@ -427,9 +504,9 @@ make_texture(struct st_context *st, /* we'll do pixel transfer in a fragment shader */ ctx->_ImageTransferState = 0x0; - transfer = pipe_get_transfer(st->pipe, pt, 0, 0, 0, - PIPE_TRANSFER_WRITE, 0, 0, - width, height); + transfer = pipe_get_transfer(st->pipe, pt, 0, 0, + PIPE_TRANSFER_WRITE, 0, 0, + width, height); /* map texture transfer */ dest = pipe_transfer_map(pipe, transfer); @@ -440,12 +517,10 @@ make_texture(struct st_context *st, * the texture. We deal with that with texcoords. */ success = _mesa_texstore(ctx, 2, /* dims */ - baseFormat, /* baseInternalFormat */ + baseInternalFormat, /* baseInternalFormat */ mformat, /* gl_format */ - dest, /* dest */ - 0, 0, 0, /* dstX/Y/Zoffset */ transfer->stride, /* dstRowStride, bytes */ - &dstImageOffsets, /* dstImageOffsets */ + &dest, /* destSlices */ width, height, 1, /* size */ format, type, /* src format/type */ pixels, /* data source */ @@ -480,7 +555,15 @@ draw_quad(struct gl_context *ctx, GLfloat x0, GLfloat y0, GLfloat z, { struct st_context *st = st_context(ctx); struct pipe_context *pipe = st->pipe; - GLfloat verts[4][3][4]; /* four verts, three attribs, XYZW */ + GLfloat (*verts)[3][4]; /* four verts, three attribs, XYZW */ + struct pipe_resource *buf = NULL; + unsigned offset; + + u_upload_alloc(st->uploader, 0, 4 * sizeof(verts[0]), &offset, &buf, + (void**)&verts); + if (!buf) { + return; + } /* setup vertex data */ { @@ -544,21 +627,12 @@ draw_quad(struct gl_context *ctx, GLfloat x0, GLfloat y0, GLfloat z, } } - { - struct pipe_resource *buf; - - /* allocate/load buffer object with vertex data */ - buf = pipe_buffer_create(pipe->screen, - PIPE_BIND_VERTEX_BUFFER, - sizeof(verts)); - pipe_buffer_write(st->pipe, buf, 0, sizeof(verts), verts); - - util_draw_vertex_buffer(pipe, buf, 0, - PIPE_PRIM_QUADS, - 4, /* verts */ - 3); /* attribs/vert */ - pipe_resource_reference(&buf, NULL); - } + u_upload_unmap(st->uploader); + util_draw_vertex_buffer(pipe, st->cso_context, buf, offset, + PIPE_PRIM_QUADS, + 4, /* verts */ + 3); /* attribs/vert */ + pipe_resource_reference(&buf, NULL); } @@ -596,8 +670,11 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, cso_save_samplers(cso); cso_save_fragment_sampler_views(cso); cso_save_fragment_shader(cso); + cso_save_stream_outputs(cso); cso_save_vertex_shader(cso); + cso_save_geometry_shader(cso); cso_save_vertex_elements(cso); + cso_save_vertex_buffers(cso); if (write_stencil) { cso_save_depth_stencil_alpha(cso); cso_save_blend(cso); @@ -607,7 +684,10 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, { struct pipe_rasterizer_state rasterizer; memset(&rasterizer, 0, sizeof(rasterizer)); + rasterizer.clamp_fragment_color = !st->clamp_frag_color_in_shader && + ctx->Color._ClampFragmentColor; rasterizer.gl_rasterization_rules = 1; + rasterizer.depth_clip = !ctx->Transform.DepthClamp; rasterizer.scissor = ctx->Scissor.Enabled; cso_set_rasterizer(cso, &rasterizer); } @@ -644,6 +724,8 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, /* vertex shader state: position + texcoord pass-through */ cso_set_vertex_shader_handle(cso, driver_vp); + /* geometry shader state: disabled */ + cso_set_geometry_shader_handle(cso, NULL); /* texture sampling state: */ { @@ -681,6 +763,7 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, } cso_set_vertex_elements(cso, 3, st->velems_util_draw); + cso_set_stream_outputs(st->cso_context, 0, NULL, 0); /* texture state: */ cso_set_fragment_sampler_views(cso, num_sampler_view, sv); @@ -713,7 +796,10 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, cso_restore_fragment_sampler_views(cso); cso_restore_fragment_shader(cso); cso_restore_vertex_shader(cso); + cso_restore_geometry_shader(cso); cso_restore_vertex_elements(cso); + cso_restore_vertex_buffers(cso); + cso_restore_stream_outputs(cso); if (write_stencil) { cso_restore_depth_stencil_alpha(cso); cso_restore_blend(cso); @@ -737,9 +823,10 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y, enum pipe_transfer_usage usage; struct pipe_transfer *pt; const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0; - GLint skipPixels; ubyte *stmap; struct gl_pixelstore_attrib clippedUnpack = *unpack; + GLubyte *sValues; + GLuint *zValues; if (!zoom) { if (!_mesa_clip_drawpixels(ctx, &x, &y, &width, &height, @@ -756,42 +843,47 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y, y = ctx->DrawBuffer->Height - y - height; } - if(format != GL_DEPTH_STENCIL && - util_format_get_component_bits(strb->format, - UTIL_FORMAT_COLORSPACE_ZS, 0) != 0) + if (format == GL_STENCIL_INDEX && + _mesa_is_format_packed_depth_stencil(strb->Base.Format)) { + /* writing stencil to a combined depth+stencil buffer */ usage = PIPE_TRANSFER_READ_WRITE; - else + } + else { usage = PIPE_TRANSFER_WRITE; + } - pt = pipe_get_transfer(st_context(ctx)->pipe, strb->texture, 0, 0, 0, - usage, x, y, - width, height); + pt = pipe_get_transfer(pipe, strb->texture, + strb->rtt_level, strb->rtt_face + strb->rtt_slice, + usage, x, y, + width, height); stmap = pipe_transfer_map(pipe, pt); pixels = _mesa_map_pbo_source(ctx, &clippedUnpack, pixels); assert(pixels); - /* if width > MAX_WIDTH, have to process image in chunks */ - skipPixels = 0; - while (skipPixels < width) { - const GLint spanX = skipPixels; - const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH); + sValues = (GLubyte *) malloc(width * sizeof(GLubyte)); + zValues = (GLuint *) malloc(width * sizeof(GLuint)); + + if (sValues && zValues) { GLint row; for (row = 0; row < height; row++) { - GLubyte sValues[MAX_WIDTH]; - GLuint zValues[MAX_WIDTH]; + GLfloat *zValuesFloat = (GLfloat*)zValues; GLenum destType = GL_UNSIGNED_BYTE; const GLvoid *source = _mesa_image_address2d(&clippedUnpack, pixels, width, height, format, type, - row, skipPixels); - _mesa_unpack_stencil_span(ctx, spanWidth, destType, sValues, + row, 0); + _mesa_unpack_stencil_span(ctx, width, destType, sValues, type, source, &clippedUnpack, ctx->_ImageTransferState); if (format == GL_DEPTH_STENCIL) { - _mesa_unpack_depth_span(ctx, spanWidth, GL_UNSIGNED_INT, zValues, + GLenum ztype = + pt->resource->format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT ? + GL_FLOAT : GL_UNSIGNED_INT; + + _mesa_unpack_depth_span(ctx, width, ztype, zValues, (1 << 24) - 1, type, source, &clippedUnpack); } @@ -813,56 +905,81 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y, /* now pack the stencil (and Z) values in the dest format */ switch (pt->resource->format) { - case PIPE_FORMAT_S8_USCALED: + case PIPE_FORMAT_S8_UINT: { - ubyte *dest = stmap + spanY * pt->stride + spanX; + ubyte *dest = stmap + spanY * pt->stride; assert(usage == PIPE_TRANSFER_WRITE); - memcpy(dest, sValues, spanWidth); + memcpy(dest, sValues, width); } break; - case PIPE_FORMAT_Z24_UNORM_S8_USCALED: + case PIPE_FORMAT_Z24_UNORM_S8_UINT: if (format == GL_DEPTH_STENCIL) { - uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4); + uint *dest = (uint *) (stmap + spanY * pt->stride); GLint k; assert(usage == PIPE_TRANSFER_WRITE); - for (k = 0; k < spanWidth; k++) { + for (k = 0; k < width; k++) { dest[k] = zValues[k] | (sValues[k] << 24); } } else { - uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4); + uint *dest = (uint *) (stmap + spanY * pt->stride); GLint k; assert(usage == PIPE_TRANSFER_READ_WRITE); - for (k = 0; k < spanWidth; k++) { + for (k = 0; k < width; k++) { dest[k] = (dest[k] & 0xffffff) | (sValues[k] << 24); } } break; - case PIPE_FORMAT_S8_USCALED_Z24_UNORM: + case PIPE_FORMAT_S8_UINT_Z24_UNORM: if (format == GL_DEPTH_STENCIL) { - uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4); + uint *dest = (uint *) (stmap + spanY * pt->stride); GLint k; assert(usage == PIPE_TRANSFER_WRITE); - for (k = 0; k < spanWidth; k++) { + for (k = 0; k < width; k++) { dest[k] = (zValues[k] << 8) | (sValues[k] & 0xff); } } else { - uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4); + uint *dest = (uint *) (stmap + spanY * pt->stride); GLint k; assert(usage == PIPE_TRANSFER_READ_WRITE); - for (k = 0; k < spanWidth; k++) { + for (k = 0; k < width; k++) { dest[k] = (dest[k] & 0xffffff00) | (sValues[k] & 0xff); } } break; + case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT: + if (format == GL_DEPTH_STENCIL) { + uint *dest = (uint *) (stmap + spanY * pt->stride); + GLfloat *destf = (GLfloat*)dest; + GLint k; + assert(usage == PIPE_TRANSFER_WRITE); + for (k = 0; k < width; k++) { + destf[k*2] = zValuesFloat[k]; + dest[k*2+1] = sValues[k] & 0xff; + } + } + else { + uint *dest = (uint *) (stmap + spanY * pt->stride); + GLint k; + assert(usage == PIPE_TRANSFER_READ_WRITE); + for (k = 0; k < width; k++) { + dest[k*2+1] = sValues[k] & 0xff; + } + } + break; default: assert(0); } } } - skipPixels += spanWidth; } + else { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels()"); + } + + free(sValues); + free(zValues); _mesa_unmap_pbo_source(ctx, &clippedUnpack); @@ -872,6 +989,63 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y, } +/** + * Get fragment program variant for a glDrawPixels or glCopyPixels + * command for RGBA data. + */ +static struct st_fp_variant * +get_color_fp_variant(struct st_context *st) +{ + struct gl_context *ctx = st->ctx; + struct st_fp_variant_key key; + struct st_fp_variant *fpv; + + memset(&key, 0, sizeof(key)); + + key.st = st; + key.drawpixels = 1; + key.scaleAndBias = (ctx->Pixel.RedBias != 0.0 || + ctx->Pixel.RedScale != 1.0 || + ctx->Pixel.GreenBias != 0.0 || + ctx->Pixel.GreenScale != 1.0 || + ctx->Pixel.BlueBias != 0.0 || + ctx->Pixel.BlueScale != 1.0 || + ctx->Pixel.AlphaBias != 0.0 || + ctx->Pixel.AlphaScale != 1.0); + key.pixelMaps = ctx->Pixel.MapColorFlag; + key.clamp_color = st->clamp_frag_color_in_shader && + st->ctx->Color._ClampFragmentColor; + + fpv = st_get_fp_variant(st, st->fp, &key); + + return fpv; +} + + +/** + * Get fragment program variant for a glDrawPixels or glCopyPixels + * command for depth/stencil data. + */ +static struct st_fp_variant * +get_depth_stencil_fp_variant(struct st_context *st, GLboolean write_depth, + GLboolean write_stencil) +{ + struct st_fp_variant_key key; + struct st_fp_variant *fpv; + + memset(&key, 0, sizeof(key)); + + key.st = st; + key.drawpixels = 1; + key.drawpixels_z = write_depth; + key.drawpixels_stencil = write_stencil; + + fpv = st_get_fp_variant(st, st->fp, &key); + + return fpv; +} + + /** * Called via ctx->Driver.DrawPixels() */ @@ -888,7 +1062,7 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, GLboolean write_stencil = GL_FALSE, write_depth = GL_FALSE; struct pipe_sampler_view *sv[2]; int num_sampler_view = 1; - enum pipe_format stencil_format = PIPE_FORMAT_NONE; + struct st_fp_variant *fpv; if (format == GL_DEPTH_STENCIL) write_stencil = write_depth = GL_TRUE; @@ -897,23 +1071,12 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, else if (format == GL_DEPTH_COMPONENT) write_depth = GL_TRUE; - if (write_stencil) { - enum pipe_format tex_format; - /* can we write to stencil if not fallback */ - if (!pipe->screen->get_param(pipe->screen, PIPE_CAP_SHADER_STENCIL_EXPORT)) - goto stencil_fallback; - - tex_format = st_choose_format(st->pipe->screen, base_format(format), - PIPE_TEXTURE_2D, - 0, PIPE_BIND_SAMPLER_VIEW); - if (tex_format == PIPE_FORMAT_Z24_UNORM_S8_USCALED) - stencil_format = PIPE_FORMAT_X24S8_USCALED; - else if (tex_format == PIPE_FORMAT_S8_USCALED_Z24_UNORM) - stencil_format = PIPE_FORMAT_S8X24_USCALED; - else - stencil_format = PIPE_FORMAT_S8_USCALED; - if (stencil_format == PIPE_FORMAT_NONE) - goto stencil_fallback; + if (write_stencil && + !pipe->screen->get_param(pipe->screen, PIPE_CAP_SHADER_STENCIL_EXPORT)) { + /* software fallback */ + draw_stencil_pixels(ctx, x, y, width, height, format, type, + unpack, pixels); + return; } /* Mesa state should be up to date by now */ @@ -921,14 +1084,25 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, st_validate_state(st); + /* + * Get vertex/fragment shaders + */ if (write_depth || write_stencil) { - driver_fp = make_fragment_shader_z_stencil(st, write_depth, write_stencil); + fpv = get_depth_stencil_fp_variant(st, write_depth, write_stencil); + + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_TRUE); + color = ctx->Current.RasterColor; } else { - driver_fp = combined_drawpix_fragment_program(ctx); + fpv = get_color_fp_variant(st); + + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_FALSE); + color = NULL; if (st->pixel_xfer.pixelmap_enabled) { sv[1] = st->pixel_xfer.pixelmap_sampler_view; @@ -936,6 +1110,9 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, } } + /* update fragment program constants */ + st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT); + /* draw with textured quad */ { struct pipe_resource *pt @@ -944,11 +1121,36 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, sv[0] = st_create_texture_sampler_view(st->pipe, pt); if (sv[0]) { - if (write_stencil) { - sv[1] = st_create_texture_sampler_view_format(st->pipe, pt, + /* Create a second sampler view to read stencil. + * The stencil is written using the shader stencil export + * functionality. */ + if (write_stencil) { + enum pipe_format stencil_format = PIPE_FORMAT_NONE; + + switch (pt->format) { + case PIPE_FORMAT_Z24_UNORM_S8_UINT: + case PIPE_FORMAT_X24S8_UINT: + stencil_format = PIPE_FORMAT_X24S8_UINT; + break; + case PIPE_FORMAT_S8_UINT_Z24_UNORM: + case PIPE_FORMAT_S8X24_UINT: + stencil_format = PIPE_FORMAT_S8X24_UINT; + break; + case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT: + case PIPE_FORMAT_X32_S8X24_UINT: + stencil_format = PIPE_FORMAT_X32_S8X24_UINT; + break; + case PIPE_FORMAT_S8_UINT: + stencil_format = PIPE_FORMAT_S8_UINT; + break; + default: + assert(0); + } + + sv[1] = st_create_texture_sampler_view_format(st->pipe, pt, stencil_format); - num_sampler_view++; - } + num_sampler_view++; + } draw_textured_quad(ctx, x, y, ctx->Current.RasterPos[2], width, height, @@ -965,11 +1167,6 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, pipe_resource_reference(&pt, NULL); } } - return; - -stencil_fallback: - draw_stencil_pixels(ctx, x, y, width, height, format, type, - unpack, pixels); } @@ -996,17 +1193,14 @@ copy_stencil_pixels(struct gl_context *ctx, GLint srcx, GLint srcy, return; } - /* Get the dest renderbuffer. If there's a wrapper, use the - * underlying renderbuffer. - */ - rbDraw = st_renderbuffer(ctx->DrawBuffer->_StencilBuffer); - if (rbDraw->Base.Wrapped) - rbDraw = st_renderbuffer(rbDraw->Base.Wrapped); + /* Get the dest renderbuffer */ + rbDraw = st_renderbuffer(ctx->DrawBuffer-> + Attachment[BUFFER_STENCIL].Renderbuffer); /* this will do stencil pixel transfer ops */ - st_read_stencil_pixels(ctx, srcx, srcy, width, height, - GL_STENCIL_INDEX, GL_UNSIGNED_BYTE, - &ctx->DefaultPacking, buffer); + _mesa_readpixels(ctx, srcx, srcy, width, height, + GL_STENCIL_INDEX, GL_UNSIGNED_BYTE, + &ctx->DefaultPacking, buffer); if (0) { /* debug code: dump stencil values */ @@ -1020,20 +1214,21 @@ copy_stencil_pixels(struct gl_context *ctx, GLint srcx, GLint srcy, } } - if (util_format_get_component_bits(rbDraw->format, - UTIL_FORMAT_COLORSPACE_ZS, 0) != 0) + if (_mesa_is_format_packed_depth_stencil(rbDraw->Base.Format)) usage = PIPE_TRANSFER_READ_WRITE; else usage = PIPE_TRANSFER_WRITE; - + if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) { dsty = rbDraw->Base.Height - dsty - height; } - ptDraw = pipe_get_transfer(st_context(ctx)->pipe, - rbDraw->texture, 0, 0, 0, - usage, dstx, dsty, - width, height); + ptDraw = pipe_get_transfer(pipe, + rbDraw->texture, + rbDraw->rtt_level, + rbDraw->rtt_face + rbDraw->rtt_slice, + usage, dstx, dsty, + width, height); assert(util_format_get_blockwidth(ptDraw->resource->format) == 1); assert(util_format_get_blockheight(ptDraw->resource->format) == 1); @@ -1057,36 +1252,7 @@ copy_stencil_pixels(struct gl_context *ctx, GLint srcx, GLint srcy, dst = drawMap + y * ptDraw->stride; src = buffer + i * width; - switch (ptDraw->resource->format) { - case PIPE_FORMAT_Z24_UNORM_S8_USCALED: - { - uint *dst4 = (uint *) dst; - int j; - assert(usage == PIPE_TRANSFER_READ_WRITE); - for (j = 0; j < width; j++) { - *dst4 = (*dst4 & 0xffffff) | (src[j] << 24); - dst4++; - } - } - break; - case PIPE_FORMAT_S8_USCALED_Z24_UNORM: - { - uint *dst4 = (uint *) dst; - int j; - assert(usage == PIPE_TRANSFER_READ_WRITE); - for (j = 0; j < width; j++) { - *dst4 = (*dst4 & 0xffffff00) | (src[j] & 0xff); - dst4++; - } - } - break; - case PIPE_FORMAT_S8_USCALED: - assert(usage == PIPE_TRANSFER_WRITE); - memcpy(dst, src, width); - break; - default: - assert(0); - } + _mesa_pack_ubyte_stencil_row(rbDraw->Base.Format, width, src, dst); } free(buffer); @@ -1097,6 +1263,127 @@ copy_stencil_pixels(struct gl_context *ctx, GLint srcx, GLint srcy, } +/** + * Return renderbuffer to use for reading color pixels for glCopyPixels + */ +static struct st_renderbuffer * +st_get_color_read_renderbuffer(struct gl_context *ctx) +{ + struct gl_framebuffer *fb = ctx->ReadBuffer; + struct st_renderbuffer *strb = + st_renderbuffer(fb->_ColorReadBuffer); + + return strb; +} + + +/** Do the src/dest regions overlap? */ +static GLboolean +regions_overlap(GLint srcX, GLint srcY, GLint dstX, GLint dstY, + GLsizei width, GLsizei height) +{ + if (srcX + width <= dstX || + dstX + width <= srcX || + srcY + height <= dstY || + dstY + height <= srcY) + return GL_FALSE; + else + return GL_TRUE; +} + + +/** + * Try to do a glCopyPixels for simple cases with a blit by calling + * pipe->resource_copy_region(). + * + * We can do this when we're copying color pixels (depth/stencil + * eventually) with no pixel zoom, no pixel transfer ops, no + * per-fragment ops, the src/dest regions don't overlap and the + * src/dest pixel formats are the same. + */ +static GLboolean +blit_copy_pixels(struct gl_context *ctx, GLint srcx, GLint srcy, + GLsizei width, GLsizei height, + GLint dstx, GLint dsty, GLenum type) +{ + struct st_context *st = st_context(ctx); + struct pipe_context *pipe = st->pipe; + struct gl_pixelstore_attrib pack, unpack; + GLint readX, readY, readW, readH; + + if (type == GL_COLOR && + ctx->Pixel.ZoomX == 1.0 && + ctx->Pixel.ZoomY == 1.0 && + ctx->_ImageTransferState == 0x0 && + !ctx->Color.BlendEnabled && + !ctx->Color.AlphaEnabled && + !ctx->Depth.Test && + !ctx->Fog.Enabled && + !ctx->Stencil.Enabled && + !ctx->FragmentProgram.Enabled && + !ctx->VertexProgram.Enabled && + !ctx->Shader.CurrentFragmentProgram && + st_fb_orientation(ctx->ReadBuffer) == st_fb_orientation(ctx->DrawBuffer) && + ctx->DrawBuffer->_NumColorDrawBuffers == 1 && + !ctx->Query.CondRenderQuery) { + struct st_renderbuffer *rbRead, *rbDraw; + GLint drawX, drawY; + + /* + * Clip the read region against the src buffer bounds. + * We'll still allocate a temporary buffer/texture for the original + * src region size but we'll only read the region which is on-screen. + * This may mean that we draw garbage pixels into the dest region, but + * that's expected. + */ + readX = srcx; + readY = srcy; + readW = width; + readH = height; + pack = ctx->DefaultPacking; + if (!_mesa_clip_readpixels(ctx, &readX, &readY, &readW, &readH, &pack)) + return GL_TRUE; /* all done */ + + /* clip against dest buffer bounds and scissor box */ + drawX = dstx + pack.SkipPixels; + drawY = dsty + pack.SkipRows; + unpack = pack; + if (!_mesa_clip_drawpixels(ctx, &drawX, &drawY, &readW, &readH, &unpack)) + return GL_TRUE; /* all done */ + + readX = readX - pack.SkipPixels + unpack.SkipPixels; + readY = readY - pack.SkipRows + unpack.SkipRows; + + rbRead = st_get_color_read_renderbuffer(ctx); + rbDraw = st_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]); + + if ((rbRead != rbDraw || + !regions_overlap(readX, readY, drawX, drawY, readW, readH)) && + rbRead->Base.Format == rbDraw->Base.Format) { + struct pipe_box srcBox; + + /* flip src/dst position if needed */ + if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + /* both buffers will have the same orientation */ + readY = ctx->ReadBuffer->Height - readY - readH; + drawY = ctx->DrawBuffer->Height - drawY - readH; + } + + u_box_2d(readX, readY, readW, readH, &srcBox); + + pipe->resource_copy_region(pipe, + rbDraw->texture, + rbDraw->rtt_level, drawX, drawY, 0, + rbRead->texture, + rbRead->rtt_level, &srcBox); + return GL_TRUE; + } + } + + return GL_FALSE; +} + + static void st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, GLsizei width, GLsizei height, @@ -1116,20 +1403,46 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, GLint readX, readY, readW, readH; GLuint sample_count; struct gl_pixelstore_attrib pack = ctx->DefaultPacking; + struct st_fp_variant *fpv; st_validate_state(st); + if (type == GL_DEPTH_STENCIL) { + /* XXX make this more efficient */ + st_CopyPixels(ctx, srcx, srcy, width, height, dstx, dsty, GL_STENCIL); + st_CopyPixels(ctx, srcx, srcy, width, height, dstx, dsty, GL_DEPTH); + return; + } + if (type == GL_STENCIL) { /* can't use texturing to do stencil */ copy_stencil_pixels(ctx, srcx, srcy, width, height, dstx, dsty); return; } + if (blit_copy_pixels(ctx, srcx, srcy, width, height, dstx, dsty, type)) + return; + + /* + * The subsequent code implements glCopyPixels by copying the source + * pixels into a temporary texture that's then applied to a textured quad. + * When we draw the textured quad, all the usual per-fragment operations + * are handled. + */ + + + /* + * Get vertex/fragment shaders + */ if (type == GL_COLOR) { rbRead = st_get_color_read_renderbuffer(ctx); color = NULL; - driver_fp = combined_drawpix_fragment_program(ctx); + + fpv = get_color_fp_variant(st); + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_FALSE); + if (st->pixel_xfer.pixelmap_enabled) { sv[1] = st->pixel_xfer.pixelmap_sampler_view; num_sampler_view++; @@ -1137,14 +1450,18 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, } else { assert(type == GL_DEPTH); - rbRead = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer); + rbRead = st_renderbuffer(ctx->ReadBuffer-> + Attachment[BUFFER_DEPTH].Renderbuffer); color = ctx->Current.Attrib[VERT_ATTRIB_COLOR0]; - driver_fp = make_fragment_shader_z_stencil(st, GL_TRUE, GL_FALSE); + + fpv = get_depth_stencil_fp_variant(st, GL_TRUE, GL_FALSE); + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_TRUE); } - if (rbRead->Base.Wrapped) - rbRead = st_renderbuffer(rbRead->Base.Wrapped); + /* update fragment program constants */ + st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT); sample_count = rbRead->texture->nr_samples; /* I believe this would be legal, presumably would need to do a resolve @@ -1156,20 +1473,21 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, if (screen->is_format_supported(screen, srcFormat, st->internal_target, sample_count, - PIPE_BIND_SAMPLER_VIEW, 0)) { + PIPE_BIND_SAMPLER_VIEW)) { texFormat = srcFormat; } else { /* srcFormat can't be used as a texture format */ if (type == GL_DEPTH) { texFormat = st_choose_format(screen, GL_DEPTH_COMPONENT, - st->internal_target, sample_count, - PIPE_BIND_DEPTH_STENCIL); + GL_NONE, GL_NONE, st->internal_target, + sample_count, PIPE_BIND_DEPTH_STENCIL); assert(texFormat != PIPE_FORMAT_NONE); } else { /* default color format */ - texFormat = st_choose_format(screen, GL_RGBA, st->internal_target, + texFormat = st_choose_format(screen, GL_RGBA, + GL_NONE, GL_NONE, st->internal_target, sample_count, PIPE_BIND_SAMPLER_VIEW); assert(texFormat != PIPE_FORMAT_NONE); } @@ -1191,7 +1509,15 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, readY = srcy; readW = width; readH = height; - _mesa_clip_readpixels(ctx, &readX, &readY, &readW, &readH, &pack); + if (!_mesa_clip_readpixels(ctx, &readX, &readY, &readW, &readH, &pack)) { + /* The source region is completely out of bounds. Do nothing. + * The GL spec says "Results of copies from outside the window, + * or from regions of the window that are not exposed, are + * hardware dependent and undefined." + */ + return; + } + readW = MAX2(0, readW); readH = MAX2(0, readH); @@ -1209,27 +1535,26 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, /* Make temporary texture which is a copy of the src region. */ if (srcFormat == texFormat) { - struct pipe_subresource srcsub, dstsub; - srcsub.face = 0; - srcsub.level = 0; - dstsub.face = 0; - dstsub.level = 0; + struct pipe_box src_box; + u_box_2d(readX, readY, readW, readH, &src_box); /* copy source framebuffer surface into mipmap/texture */ pipe->resource_copy_region(pipe, pt, /* dest tex */ - dstsub, + 0, /* dest lvl */ pack.SkipPixels, pack.SkipRows, 0, /* dest pos */ rbRead->texture, /* src tex */ - srcsub, - readX, readY, 0, readW, readH); /* src region */ + rbRead->rtt_level, /* src lvl */ + &src_box); } else { /* CPU-based fallback/conversion */ struct pipe_transfer *ptRead = - pipe_get_transfer(st->pipe, rbRead->texture, 0, 0, 0, - PIPE_TRANSFER_READ, - readX, readY, readW, readH); + pipe_get_transfer(st->pipe, rbRead->texture, + rbRead->rtt_level, + rbRead->rtt_face + rbRead->rtt_slice, + PIPE_TRANSFER_READ, + readX, readY, readW, readH); struct pipe_transfer *ptTex; enum pipe_transfer_usage transfer_usage; @@ -1241,22 +1566,26 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, else transfer_usage = PIPE_TRANSFER_WRITE; - ptTex = pipe_get_transfer(st->pipe, pt, 0, 0, 0, transfer_usage, - 0, 0, width, height); + ptTex = pipe_get_transfer(st->pipe, pt, 0, 0, transfer_usage, + 0, 0, width, height); /* copy image from ptRead surface to ptTex surface */ if (type == GL_COLOR) { /* alternate path using get/put_tile() */ GLfloat *buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - pipe_get_tile_rgba(pipe, ptRead, readX, readY, readW, readH, buf); - pipe_put_tile_rgba(pipe, ptTex, pack.SkipPixels, pack.SkipRows, - readW, readH, buf); + enum pipe_format readFormat, drawFormat; + readFormat = util_format_linear(rbRead->texture->format); + drawFormat = util_format_linear(pt->format); + pipe_get_tile_rgba_format(pipe, ptRead, 0, 0, readW, readH, + readFormat, buf); + pipe_put_tile_rgba_format(pipe, ptTex, pack.SkipPixels, pack.SkipRows, + readW, readH, drawFormat, buf); free(buf); } else { /* GL_DEPTH */ GLuint *buf = (GLuint *) malloc(width * height * sizeof(GLuint)); - pipe_get_tile_z(pipe, ptRead, readX, readY, readW, readH, buf); + pipe_get_tile_z(pipe, ptRead, 0, 0, readW, readH, buf); pipe_put_tile_z(pipe, ptTex, pack.SkipPixels, pack.SkipRows, readW, readH, buf); free(buf); @@ -1297,14 +1626,14 @@ st_destroy_drawpix(struct st_context *st) for (i = 0; i < Elements(st->drawpix.shaders); i++) { if (st->drawpix.shaders[i]) - st_reference_fragprog(st, &st->drawpix.shaders[i], NULL); + _mesa_reference_fragprog(st->ctx, &st->drawpix.shaders[i], NULL); } st_reference_fragprog(st, &st->pixel_xfer.combined_prog, NULL); if (st->drawpix.vert_shaders[0]) - ureg_free_tokens(st->drawpix.vert_shaders[0]); + cso_delete_vertex_shader(st->cso_context, st->drawpix.vert_shaders[0]); if (st->drawpix.vert_shaders[1]) - ureg_free_tokens(st->drawpix.vert_shaders[1]); + cso_delete_vertex_shader(st->cso_context, st->drawpix.vert_shaders[1]); } #endif /* FEATURE_drawpix */