diff options
Diffstat (limited to 'src/mesa/drivers/dri')
44 files changed, 1528 insertions, 897 deletions
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template index 6f2314ee8c..43c0e912bf 100644 --- a/src/mesa/drivers/dri/Makefile.template +++ b/src/mesa/drivers/dri/Makefile.template @@ -85,7 +85,7 @@ $(TOP)/$(LIB_DIR)/$(LIBNAME): $(LIBNAME) depend: $(C_SOURCES) $(ASM_SOURCES) $(SYMLINKS) touch depend $(MKDEP) $(MKDEP_OPTIONS) $(DRIVER_DEFINES) $(INCLUDES) $(C_SOURCES) \ - $(ASM_SOURCES) 2>&1 /dev/null + $(ASM_SOURCES) 2> /dev/null # Emacs tags diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c index d785382f39..6e8a5b5218 100644 --- a/src/mesa/drivers/dri/common/dri_util.c +++ b/src/mesa/drivers/dri/common/dri_util.c @@ -434,10 +434,12 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp) if (pdp->pClipRects) { _mesa_free(pdp->pClipRects); + pdp->pClipRects = NULL; } if (pdp->pBackClipRects) { _mesa_free(pdp->pBackClipRects); + pdp->pBackClipRects = NULL; } DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID); diff --git a/src/mesa/drivers/dri/i810/i810state.c b/src/mesa/drivers/dri/i810/i810state.c index 3ad25282d9..e0d5b2b487 100644 --- a/src/mesa/drivers/dri/i810/i810state.c +++ b/src/mesa/drivers/dri/i810/i810state.c @@ -380,7 +380,10 @@ static void i810CullFaceFrontFace(GLcontext *ctx, GLenum unused) static void i810LineWidth( GLcontext *ctx, GLfloat widthf ) { i810ContextPtr imesa = I810_CONTEXT( ctx ); - int width = (int)ctx->Line._Width; + /* AA, non-AA limits are same */ + const int width = (int) CLAMP(ctx->Line.Width, + ctx->Const.MinLineWidth, + ctx->Const.MaxLineWidth); imesa->LcsLineWidth = 0; if (width & 1) imesa->LcsLineWidth |= LCS_LINEWIDTH_1_0; @@ -396,7 +399,10 @@ static void i810LineWidth( GLcontext *ctx, GLfloat widthf ) static void i810PointSize( GLcontext *ctx, GLfloat sz ) { i810ContextPtr imesa = I810_CONTEXT( ctx ); - int size = (int)ctx->Point._Size; + /* AA, non-AA limits are same */ + const int size = (int) CLAMP(ctx->Point.Size, + ctx->Const.MinPointSize, + ctx->Const.MaxPointSize); imesa->LcsPointSize = 0; if (size & 1) imesa->LcsPointSize |= LCS_LINEWIDTH_1_0; diff --git a/src/mesa/drivers/dri/i810/i810tris.c b/src/mesa/drivers/dri/i810/i810tris.c index 3e09427bb9..2c4ee06633 100644 --- a/src/mesa/drivers/dri/i810/i810tris.c +++ b/src/mesa/drivers/dri/i810/i810tris.c @@ -112,7 +112,9 @@ static __inline__ void i810_draw_quad( i810ContextPtr imesa, static __inline__ void i810_draw_point( i810ContextPtr imesa, i810VertexPtr tmp ) { - GLfloat sz = imesa->glCtx->Point._Size * .5; + GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size, + imesa->glCtx->Const.MinPointSize, + imesa->glCtx->Const.MaxPointSize); int vertsize = imesa->vertex_size; GLuint *vb = i810AllocDmaLow( imesa, 2 * 4 * vertsize ); int j; diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c index b0e5f87fc7..cc8a605e50 100644 --- a/src/mesa/drivers/dri/i915/i915_vtbl.c +++ b/src/mesa/drivers/dri/i915/i915_vtbl.c @@ -64,6 +64,7 @@ static void i915_reduced_primitive_state( intelContextPtr intel, st1 &= ~ST1_ENABLE; switch (rprim) { + case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */ case GL_TRIANGLES: if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple) diff --git a/src/mesa/drivers/dri/i915/intel_pixel.c b/src/mesa/drivers/dri/i915/intel_pixel.c index d175870a0c..c3030d42b0 100644 --- a/src/mesa/drivers/dri/i915/intel_pixel.c +++ b/src/mesa/drivers/dri/i915/intel_pixel.c @@ -450,10 +450,16 @@ intelDrawPixels( GLcontext *ctx, * wise happily run the fragment program on each pixel in the image). */ struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current; + /* can't just set current frag prog to 0 here as on buffer resize + we'll get new state checks which will segfault. Remains a hack. */ ctx->FragmentProgram._Current = NULL; + ctx->FragmentProgram._UseTexEnvProgram = GL_FALSE; + ctx->FragmentProgram._Active = GL_FALSE; _swrast_DrawPixels( ctx, x, y, width, height, format, type, unpack, pixels ); ctx->FragmentProgram._Current = fpSave; + ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE; + ctx->FragmentProgram._Active = GL_TRUE; } else { _swrast_DrawPixels( ctx, x, y, width, height, format, type, diff --git a/src/mesa/drivers/dri/i915tex/Makefile b/src/mesa/drivers/dri/i915tex/Makefile index b218929dce..827acc0c46 100644 --- a/src/mesa/drivers/dri/i915tex/Makefile +++ b/src/mesa/drivers/dri/i915tex/Makefile @@ -48,6 +48,7 @@ DRIVER_SOURCES = \ intel_screen.c \ intel_span.c \ intel_state.c \ + intel_surface.c \ intel_tris.c \ intel_fbo.c \ intel_depthstencil.c \ diff --git a/src/mesa/drivers/dri/i915tex/i915_context.c b/src/mesa/drivers/dri/i915tex/i915_context.c index 49e30141e4..b6d004b258 100644 --- a/src/mesa/drivers/dri/i915tex/i915_context.c +++ b/src/mesa/drivers/dri/i915tex/i915_context.c @@ -71,6 +71,8 @@ i915InvalidateState(GLcontext * ctx, GLuint new_state) _tnl_invalidate_vertex_state(ctx, new_state); intel_context(ctx)->NewGLState |= new_state; + st_invalidate_state(ctx, new_state); + /* Todo: gather state values under which tracked parameters become * invalidated, add callbacks for things like * ProgramLocalParameters, etc. diff --git a/src/mesa/drivers/dri/i915tex/i915_vtbl.c b/src/mesa/drivers/dri/i915tex/i915_vtbl.c index 13ca9aee6b..cc74ceae18 100644 --- a/src/mesa/drivers/dri/i915tex/i915_vtbl.c +++ b/src/mesa/drivers/dri/i915tex/i915_vtbl.c @@ -66,6 +66,7 @@ i915_reduced_primitive_state(struct intel_context *intel, GLenum rprim) st1 &= ~ST1_ENABLE; switch (rprim) { + case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */ case GL_TRIANGLES: if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple) st1 |= ST1_ENABLE; diff --git a/src/mesa/drivers/dri/i915tex/intel_buffers.c b/src/mesa/drivers/dri/i915tex/intel_buffers.c index 4b2a4acaf9..72c7b5ae45 100644 --- a/src/mesa/drivers/dri/i915tex/intel_buffers.c +++ b/src/mesa/drivers/dri/i915tex/intel_buffers.c @@ -250,7 +250,8 @@ intelWindowMoved(struct intel_context *intel) flags = intel_fb->vblank_flags & ~VBLANK_FLAG_SECONDARY; } - if (flags != intel_fb->vblank_flags) { + if (flags != intel_fb->vblank_flags && intel_fb->vblank_flags && + !(intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ)) { drmVBlank vbl; int i; @@ -261,7 +262,9 @@ intelWindowMoved(struct intel_context *intel) } for (i = 0; i < intel_fb->pf_num_pages; i++) { - if (!intel_fb->color_rb[i]) + if (!intel_fb->color_rb[i] || + (intel_fb->vbl_waited - intel_fb->color_rb[i]->vbl_pending) <= + (1<<23)) continue; vbl.request.sequence = intel_fb->color_rb[i]->vbl_pending; @@ -583,8 +586,9 @@ intelScheduleSwap(const __DRIdrawablePrivate * dPriv, GLboolean *missed_target) drm_i915_vblank_swap_t swap; GLboolean ret; - if ((intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ) || - intelScreen->drmMinor < (intel_fb->pf_active ? 9 : 6)) + if (!intel_fb->vblank_flags || + (intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ) || + intelScreen->drmMinor < (intel_fb->pf_active ? 9 : 6)) return GL_FALSE; swap.seqtype = DRM_VBLANK_ABSOLUTE; diff --git a/src/mesa/drivers/dri/i915tex/intel_context.c b/src/mesa/drivers/dri/i915tex/intel_context.c index f032c97dfb..5ab919ecb6 100644 --- a/src/mesa/drivers/dri/i915tex/intel_context.c +++ b/src/mesa/drivers/dri/i915tex/intel_context.c @@ -60,6 +60,10 @@ #include "intel_buffer_objects.h" #include "intel_fbo.h" +#include "pipe/softpipe/sp_context.h" +#include "state_tracker/st_public.h" + + #include "drirenderbuffer.h" #include "vblank.h" #include "utils.h" @@ -257,6 +261,9 @@ intelInvalidateState(GLcontext * ctx, GLuint new_state) _vbo_InvalidateState(ctx, new_state); _tnl_InvalidateState(ctx, new_state); _tnl_invalidate_vertex_state(ctx, new_state); + + st_invalidate_state( ctx, new_state ); + intel_context(ctx)->NewGLState |= new_state; } @@ -504,6 +511,11 @@ intelInitContext(struct intel_context *intel, FALLBACK(intel, INTEL_FALLBACK_USER, 1); } + + st_create_context( &intel->ctx, + softpipe_create() ); + + return GL_TRUE; } @@ -605,12 +617,23 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv, if (intel->ctx.DrawBuffer == &intel_fb->Base) { if (intel->driDrawable != driDrawPriv) { - intel_fb->vblank_flags = (intel->intelScreen->irq_active != 0) - ? driGetDefaultVBlankFlags(&intel->optionCache) - : VBLANK_FLAG_NO_IRQ; - (*dri_interface->getUST) (&intel_fb->swap_ust); - driDrawableInitVBlank(driDrawPriv, intel_fb->vblank_flags, - &intel_fb->vbl_seq); + if (driDrawPriv->pdraw->swap_interval == (unsigned)-1) { + int i; + + intel_fb->vblank_flags = (intel->intelScreen->irq_active != 0) + ? driGetDefaultVBlankFlags(&intel->optionCache) + : VBLANK_FLAG_NO_IRQ; + + (*dri_interface->getUST) (&intel_fb->swap_ust); + driDrawableInitVBlank(driDrawPriv, intel_fb->vblank_flags, + &intel_fb->vbl_seq); + intel_fb->vbl_waited = intel_fb->vbl_seq; + + for (i = 0; i < 2; i++) { + if (intel_fb->color_rb[i]) + intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_seq; + } + } } } @@ -660,37 +683,27 @@ intelContendedLock(struct intel_context *intel, GLuint flags) if (sarea->width != intel->width || sarea->height != intel->height || sarea->rotation != intel->current_rotation) { - - void *batchMap = intel->batch->map; - + int numClipRects = intel->numClipRects; + /* * FIXME: Really only need to do this when drawing to a * common back- or front buffer. */ /* - * This will drop the outstanding batchbuffer on the floor + * This will essentially drop the outstanding batchbuffer on the floor. */ + intel->numClipRects = 0; - if (batchMap != NULL) { - driBOUnmap(intel->batch->buffer); - intel->batch->map = NULL; - } - - intel_batchbuffer_reset(intel->batch); + if (intel->Fallback) + _swrast_flush(&intel->ctx); - if (batchMap == NULL) { - driBOUnmap(intel->batch->buffer); - intel->batch->map = NULL; - } + INTEL_FIREVERTICES(intel); - /* lose all primitives */ - intel->prim.primitive = ~0; - intel->prim.start_ptr = 0; - intel->prim.flush = 0; + if (intel->batch->map != intel->batch->ptr) + intel_batchbuffer_flush(intel->batch); - /* re-emit all state */ - intel->vtbl.lost_hardware(intel); + intel->numClipRects = numClipRects; /* force window update */ intel->lastStamp = 0; @@ -724,7 +737,9 @@ void LOCK_HARDWARE( struct intel_context *intel ) BUFFER_BACK_LEFT); } - if (intel_rb && (intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) { + if (intel_rb && intel_fb->vblank_flags && + !(intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ) && + (intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) { drmVBlank vbl; vbl.request.type = DRM_VBLANK_ABSOLUTE; diff --git a/src/mesa/drivers/dri/i915tex/intel_fbo.c b/src/mesa/drivers/dri/i915tex/intel_fbo.c index 04ca5ae8ba..559788f20a 100644 --- a/src/mesa/drivers/dri/i915tex/intel_fbo.c +++ b/src/mesa/drivers/dri/i915tex/intel_fbo.c @@ -291,6 +291,12 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb, rb->Width = width; rb->Height = height; +#if 1 + /* update the surface's size too */ + rb->surface->width = width; + rb->surface->height = height; +#endif + /* This sets the Get/PutRow/Value functions */ intel_set_span_functions(&irb->Base); @@ -362,7 +368,9 @@ intel_new_renderbuffer_fb(GLuint intFormat) irb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT; break; default: - assert(0); + _mesa_problem(NULL, + "Unexpected intFormat in intel_create_renderbuffer"); + return NULL; } /* intel-specific methods */ @@ -371,7 +379,10 @@ intel_new_renderbuffer_fb(GLuint intFormat) irb->Base.GetPointer = intel_get_pointer; /* span routines set in alloc_storage function */ - return irb; + irb->Base.surface = intel_new_surface(intFormat); + irb->Base.surface->rb = irb; + + return &irb->Base; } /** @@ -399,6 +410,9 @@ intel_new_renderbuffer(GLcontext * ctx, GLuint name) irb->Base.GetPointer = intel_get_pointer; /* span routines set in alloc_storage function */ + irb->Base.surface = intel_new_surface(0 /*unknown format*/); + irb->Base.surface->rb = irb; + return &irb->Base; } diff --git a/src/mesa/drivers/dri/i915tex/intel_fbo.h b/src/mesa/drivers/dri/i915tex/intel_fbo.h index 762aac594e..7dc3967533 100644 --- a/src/mesa/drivers/dri/i915tex/intel_fbo.h +++ b/src/mesa/drivers/dri/i915tex/intel_fbo.h @@ -29,9 +29,14 @@ #define INTEL_FBO_H +#include "pipe/p_state.h" +#include "pipe/softpipe/sp_surface.h" + + struct intel_context; struct intel_region; + /** * Intel framebuffer, derived from gl_framebuffer. */ @@ -82,6 +87,8 @@ struct intel_renderbuffer GLuint pf_pending; /**< sequence number of pending flip */ GLuint vbl_pending; /**< vblank sequence number of pending flip */ + + struct intel_surface *surface; }; #if 0 @@ -113,5 +120,8 @@ extern struct intel_region *intel_get_rb_region(struct gl_framebuffer *fb, +extern struct pipe_surface * +intel_new_surface(GLuint intFormat); + #endif /* INTEL_FBO_H */ diff --git a/src/mesa/drivers/dri/i915tex/intel_surface.c b/src/mesa/drivers/dri/i915tex/intel_surface.c new file mode 100644 index 0000000000..043c5aa5fe --- /dev/null +++ b/src/mesa/drivers/dri/i915tex/intel_surface.c @@ -0,0 +1,259 @@ +#include "glheader.h" +#include "context.h" +#include "framebuffer.h" +#include "renderbuffer.h" +#include "utils.h" +#include "main/macros.h" + + +#include "intel_screen.h" + +#include "intel_context.h" +#include "intel_buffers.h" +#include "intel_regions.h" +#include "intel_span.h" +#include "intel_fbo.h" + +#include "pipe/p_state.h" +#include "pipe/p_defines.h" +#include "pipe/softpipe/sp_surface.h" + + +/* + * XXX a lof of this is a temporary kludge + */ + +/** + * Note: the arithmetic/addressing in these functions is a little + * tricky since we need to invert the Y axis. + */ + + +static void +read_quad_f_swz(struct softpipe_surface *sps, GLint x, GLint y, + GLfloat (*rrrr)[QUAD_SIZE]) +{ + const GLint bytesPerRow = sps->surface.stride * sps->surface.cpp; + const GLint invY = sps->surface.height - y - 1; + const GLubyte *src = sps->surface.ptr + invY * bytesPerRow + x * sps->surface.cpp; + GLfloat *dst = (GLfloat *) rrrr; + GLubyte temp[16]; + GLuint j; + + assert(sps->surface.format == PIPE_FORMAT_U_A8_R8_G8_B8); + + memcpy(temp + 8, src, 8); + memcpy(temp + 0, src + bytesPerRow, 8); + + for (j = 0; j < 4; j++) { + dst[0 * 4 + j] = UBYTE_TO_FLOAT(temp[j * 4 + 2]); /*R*/ + dst[1 * 4 + j] = UBYTE_TO_FLOAT(temp[j * 4 + 1]); /*G*/ + dst[2 * 4 + j] = UBYTE_TO_FLOAT(temp[j * 4 + 0]); /*B*/ + dst[3 * 4 + j] = UBYTE_TO_FLOAT(temp[j * 4 + 3]); /*A*/ + } +} + + +static void +write_quad_f_swz(struct softpipe_surface *sps, GLint x, GLint y, + GLfloat (*rrrr)[QUAD_SIZE]) +{ + const GLfloat *src = (const GLfloat *) rrrr; + const GLint bytesPerRow = sps->surface.stride * sps->surface.cpp; + const GLint invY = sps->surface.height - y - 1; + GLubyte *dst = sps->surface.ptr + invY * bytesPerRow + x * sps->surface.cpp; + GLubyte temp[16]; + GLuint j; + + assert(sps->surface.format == PIPE_FORMAT_U_A8_R8_G8_B8); + + for (j = 0; j < 4; j++) { + UNCLAMPED_FLOAT_TO_UBYTE(temp[j * 4 + 2], src[0 * 4 + j]); /*R*/ + UNCLAMPED_FLOAT_TO_UBYTE(temp[j * 4 + 1], src[1 * 4 + j]); /*G*/ + UNCLAMPED_FLOAT_TO_UBYTE(temp[j * 4 + 0], src[2 * 4 + j]); /*B*/ + UNCLAMPED_FLOAT_TO_UBYTE(temp[j * 4 + 3], src[3 * 4 + j]); /*A*/ + } + + memcpy(dst, temp + 8, 8); + memcpy(dst + bytesPerRow, temp + 0, 8); +} + + + +static void +read_quad_z24(struct softpipe_surface *sps, + GLint x, GLint y, GLuint zzzz[QUAD_SIZE]) +{ + static const GLuint mask = 0xffffff; + const GLint invY = sps->surface.height - y - 1; + const GLuint *src + = (GLuint *) (sps->surface.ptr + + (invY * sps->surface.stride + x) * sps->surface.cpp); + + assert(sps->surface.format == PIPE_FORMAT_Z24_S8); + + /* extract lower three bytes */ + zzzz[0] = src[0] & mask; + zzzz[1] = src[1] & mask; + zzzz[2] = src[-sps->surface.stride] & mask; + zzzz[3] = src[-sps->surface.stride + 1] & mask; +} + +static void +write_quad_z24(struct softpipe_surface *sps, + GLint x, GLint y, const GLuint zzzz[QUAD_SIZE]) +{ + static const GLuint mask = 0xff000000; + const GLint invY = sps->surface.height - y - 1; + GLuint *dst + = (GLuint *) (sps->surface.ptr + + (invY * sps->surface.stride + x) * sps->surface.cpp); + + assert(sps->surface.format == PIPE_FORMAT_Z24_S8); + + /* write lower three bytes */ + dst[0] = (dst[0] & mask) | zzzz[0]; + dst[1] = (dst[1] & mask) | zzzz[1]; + dst -= sps->surface.stride; + dst[0] = (dst[0] & mask) | zzzz[2]; + dst[1] = (dst[1] & mask) | zzzz[3]; +} + + +static void +read_quad_stencil(struct softpipe_surface *sps, + GLint x, GLint y, GLubyte ssss[QUAD_SIZE]) +{ + const GLint invY = sps->surface.height - y - 1; + const GLuint *src = (const GLuint *) (sps->surface.ptr + + (invY * sps->surface.stride + x) * sps->surface.cpp); + + assert(sps->surface.format == PIPE_FORMAT_Z24_S8); + + /* extract high byte */ + ssss[0] = src[0] >> 24; + ssss[1] = src[1] >> 24; + ssss[2] = src[-sps->surface.width] >> 24; + ssss[3] = src[-sps->surface.width + 1] >> 24; +} + +static void +write_quad_stencil(struct softpipe_surface *sps, + GLint x, GLint y, const GLubyte ssss[QUAD_SIZE]) +{ + static const GLuint mask = 0x00ffffff; + const GLint invY = sps->surface.height - y - 1; + GLuint *dst = (GLuint *) (sps->surface.ptr + + (invY * sps->surface.stride + x) * sps->surface.cpp); + + assert(sps->surface.format == PIPE_FORMAT_Z24_S8); + + /* write high byte */ + dst[0] = (dst[0] & mask) | (ssss[0] << 24); + dst[1] = (dst[1] & mask) | (ssss[1] << 24); + dst -= sps->surface.stride; + dst[0] = (dst[0] & mask) | (ssss[2] << 24); + dst[1] = (dst[1] & mask) | (ssss[3] << 24); +} + + +static void * +map_surface_buffer(struct pipe_buffer *pb, GLuint access_mode) +{ + struct softpipe_surface *sps = (struct softpipe_surface *) pb; + struct intel_renderbuffer *irb = (struct intel_renderbuffer *) sps->surface.rb; + assert(access_mode == PIPE_MAP_READ_WRITE); + + /*LOCK_HARDWARE(intel);*/ + + if (irb->region) { + GET_CURRENT_CONTEXT(ctx); + struct intel_context *intel = intel_context(ctx); +#if 0 + intelFinish(&intel->ctx); /* XXX need this? */ +#endif + intel_region_map(intel->intelScreen, irb->region); + } + pb->ptr = irb->region->map; + + sps->surface.stride = irb->region->pitch; + sps->surface.cpp = irb->region->cpp; + sps->surface.ptr = irb->region->map; + + return pb->ptr; +} + + +static void +unmap_surface_buffer(struct pipe_buffer *pb) +{ + struct softpipe_surface *sps = (struct softpipe_surface *) pb; + struct intel_renderbuffer *irb = (struct intel_renderbuffer *) sps->surface.rb; + + if (irb->region) { + GET_CURRENT_CONTEXT(ctx); + struct intel_context *intel = intel_context(ctx); + intel_region_unmap(intel->intelScreen, irb->region); + } + pb->ptr = NULL; + + sps->surface.stride = 0; + sps->surface.cpp = 0; + sps->surface.ptr = NULL; + + /*UNLOCK_HARDWARE(intel);*/ +} + + +struct pipe_surface * +intel_new_surface(GLuint intFormat) +{ + struct softpipe_surface *sps = CALLOC_STRUCT(softpipe_surface); + if (!sps) + return NULL; + + sps->surface.width = 0; /* set in intel_alloc_renderbuffer_storage() */ + sps->surface.height = 0; + + if (intFormat == GL_RGBA8) { + sps->surface.format = PIPE_FORMAT_U_A8_R8_G8_B8; + sps->read_quad_f_swz = read_quad_f_swz; + sps->write_quad_f_swz = write_quad_f_swz; + } + else if (intFormat == GL_RGB5) { + sps->surface.format = PIPE_FORMAT_U_R5_G6_B5; + + } + else if (intFormat == GL_DEPTH_COMPONENT16) { + sps->surface.format = PIPE_FORMAT_U_Z16; + + } + else if (intFormat == GL_DEPTH24_STENCIL8_EXT) { + sps->surface.format = PIPE_FORMAT_Z24_S8; + sps->read_quad_z = read_quad_z24; + sps->write_quad_z = write_quad_z24; + sps->read_quad_stencil = read_quad_stencil; + sps->write_quad_stencil = write_quad_stencil; + } + else { + /* TBD / unknown */ + + } + + sps->surface.buffer.map = map_surface_buffer; + sps->surface.buffer.unmap = unmap_surface_buffer; + + return &sps->surface; +} + + + +struct pipe_surface * +xmesa_get_stencil_surface(GLcontext *ctx) +{ + /* XXX fix */ + return NULL; +} + + + diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c index bfac52d765..9a6e5f5f19 100644 --- a/src/mesa/drivers/dri/i965/brw_sf_state.c +++ b/src/mesa/drivers/dri/i965/brw_sf_state.c @@ -173,7 +173,8 @@ static void upload_sf_unit( struct brw_context *brw ) /* _NEW_LINE */ - sf.sf6.line_width = brw->attribs.Line->_Width * (1<<1); + /* XXX use ctx->Const.Min/MaxLineWidth here */ + sf.sf6.line_width = CLAMP(brw->attribs.Line->Width, 1.0, 5.0) * (1<<1); sf.sf6.line_endcap_aa_region_width = 1; if (brw->attribs.Line->SmoothFlag) @@ -183,7 +184,8 @@ static void upload_sf_unit( struct brw_context *brw ) /* _NEW_POINT */ sf.sf6.point_rast_rule = 1; /* opengl conventions */ - sf.sf7.point_size = brw->attribs.Point->_Size * (1<<3); + /* XXX clamp max depends on AA vs. non-AA */ + sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 3.0) * (1<<3); sf.sf7.use_point_size_state = !brw->attribs.Point->_Attenuated; /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons: diff --git a/src/mesa/drivers/dri/i965/intel_blit.c b/src/mesa/drivers/dri/i965/intel_blit.c index 173d1d5b6c..f88cbb2328 100644 --- a/src/mesa/drivers/dri/i965/intel_blit.c +++ b/src/mesa/drivers/dri/i965/intel_blit.c @@ -532,12 +532,15 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, GLuint dst_offset, GLboolean dst_tiled, GLshort x, GLshort y, - GLshort w, GLshort h) + GLshort w, GLshort h, + GLenum logic_op) { struct xy_setup_blit setup; struct xy_text_immediate_blit text; int dwords = ((src_size + 7) & ~7) / 4; + assert( logic_op - GL_CLEAR >= 0 ); + assert( logic_op - GL_CLEAR < 0x10 ); if (w < 0 || h < 0) return; @@ -561,7 +564,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, setup.br0.length = (sizeof(setup) / sizeof(int)) - 2; setup.br13.dest_pitch = dst_pitch; - setup.br13.rop = 0xcc; + setup.br13.rop = translate_raster_op(logic_op); setup.br13.color_depth = (cpp == 4) ? BR13_8888 : BR13_565; setup.br13.clipping_enable = 0; setup.br13.mono_source_transparency = 1; diff --git a/src/mesa/drivers/dri/i965/intel_blit.h b/src/mesa/drivers/dri/i965/intel_blit.h index 8b0cc65243..e361545c8f 100644 --- a/src/mesa/drivers/dri/i965/intel_blit.h +++ b/src/mesa/drivers/dri/i965/intel_blit.h @@ -72,6 +72,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, GLuint dst_offset, GLboolean dst_tiled, GLshort dst_x, GLshort dst_y, - GLshort w, GLshort h); + GLshort w, GLshort h, + GLenum logic_op ); #endif diff --git a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c index 5841afaa3e..421fcc5e51 100644 --- a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c +++ b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c @@ -260,7 +260,9 @@ do_blit_bitmap( GLcontext *ctx, int h = MIN2(DY, box_h - py); int w = MIN2(DX, box_w - px); GLuint sz = align(align(w,8) * h, 64)/8; - + GLenum logic_op = ctx->Color.ColorLogicOpEnabled ? + ctx->Color.LogicOp : GL_COPY; + assert(sz <= sizeof(stipple)); memset(stipple, 0, sz); @@ -288,7 +290,8 @@ do_blit_bitmap( GLcontext *ctx, dst->tiled, rect.x1 + px, rect.y2 - (py + h), - w, h); + w, h, + logic_op); } } } diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vb.c b/src/mesa/drivers/dri/mach64/mach64_native_vb.c index 75cf0e2ed2..248fa2a9a2 100644 --- a/src/mesa/drivers/dri/mach64/mach64_native_vb.c +++ b/src/mesa/drivers/dri/mach64/mach64_native_vb.c @@ -103,7 +103,7 @@ void TAG(translate_vertex)(GLcontext *ctx, assert( p + 1 - (CARD32 *)src == 10 ); - dst->pointSize = ctx->Point._Size; + dst->pointSize = ctx->Point.Size; } diff --git a/src/mesa/drivers/dri/mach64/mach64_tris.c b/src/mesa/drivers/dri/mach64/mach64_tris.c index 08cc1849a1..369f610442 100644 --- a/src/mesa/drivers/dri/mach64/mach64_tris.c +++ b/src/mesa/drivers/dri/mach64/mach64_tris.c @@ -673,7 +673,10 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa, #if MACH64_NATIVE_VTXFMT GLcontext *ctx = mmesa->glCtx; const GLuint vertsize = mmesa->vertex_size; - GLint width = (GLint)(mmesa->glCtx->Line._Width * 2.0); /* 2 fractional bits for hardware */ + /* 2 fractional bits for hardware: */ + const int width = (int) (2.0 * CLAMP(mmesa->glCtx->Line.Width, + mmesa->glCtx->Const.MinLineWidth, + mmesa->glCtx->Const.MaxLineWidth)); GLfloat ooa; GLuint *pxy0, *pxy1; GLuint xy0old, xy0, xy1old, xy1; @@ -691,9 +694,6 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa, mach64_print_vertex( ctx, v1 ); } - if( !width ) - width = 1; /* round to the nearest supported width */ - pxy0 = &v0->ui[xyoffset]; xy0old = *pxy0; xy0 = LE32_IN( &xy0old ); @@ -961,7 +961,10 @@ static __inline void mach64_draw_point( mach64ContextPtr mmesa, #if MACH64_NATIVE_VTXFMT GLcontext *ctx = mmesa->glCtx; const GLuint vertsize = mmesa->vertex_size; - GLint sz = (GLint)(mmesa->glCtx->Point._Size * 2.0); /* 2 fractional bits for hardware */ + /* 2 fractional bits for hardware: */ + GLint sz = (GLint) (2.0 * CLAMP(mmesa->glCtx->Point.Size, + ctx->Const.MinPointSize, + ctx->Const.MaxPointSize)); GLfloat ooa; GLuint *pxy; GLuint xyold, xy; diff --git a/src/mesa/drivers/dri/mga/mgatris.c b/src/mesa/drivers/dri/mga/mgatris.c index 2b7ea05b14..91b413ae76 100644 --- a/src/mesa/drivers/dri/mga/mgatris.c +++ b/src/mesa/drivers/dri/mga/mgatris.c @@ -104,8 +104,10 @@ static void __inline__ mga_draw_quad( mgaContextPtr mmesa, static __inline__ void mga_draw_point( mgaContextPtr mmesa, mgaVertexPtr tmp ) { - GLfloat sz = mmesa->glCtx->Point._Size * .5; - int vertex_size = mmesa->vertex_size; + const GLfloat sz = 0.5 * CLAMP(mmesa->glCtx->Point.Size, + mmesa->glCtx->Const.MinPointSize, + mmesa->glCtx->Const.MaxPointSize); + const int vertex_size = mmesa->vertex_size; GLuint *vb = mgaAllocDmaLow( mmesa, 6 * 4 * vertex_size ); int j; @@ -165,7 +167,9 @@ static __inline__ void mga_draw_line( mgaContextPtr mmesa, GLuint vertex_size = mmesa->vertex_size; GLuint *vb = mgaAllocDmaLow( mmesa, 6 * 4 * vertex_size ); GLfloat dx, dy, ix, iy; - GLfloat width = mmesa->glCtx->Line._Width; + const GLfloat width = CLAMP(mmesa->glCtx->Line.Width, + mmesa->glCtx->Const.MinLineWidth, + mmesa->glCtx->Const.MaxLineWidth); GLint j; #if 0 diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c index 6a2c2013e3..d498f616c9 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c @@ -69,7 +69,7 @@ void nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem) { nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); - drm_nouveau_mem_free_t memf; + struct drm_nouveau_mem_free memf; if (NOUVEAU_DEBUG & DEBUG_MEM) { fprintf(stderr, "%s: type=0x%x, offset=0x%x, size=0x%x\n", @@ -78,8 +78,8 @@ nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem) if (mem->map) drmUnmap(mem->map, mem->size); - memf.flags = mem->type; - memf.region_offset = mem->offset; + memf.flags = mem->type; + memf.offset = mem->offset; drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_MEM_FREE, &memf, sizeof(memf)); FREE(mem); } @@ -88,7 +88,7 @@ nouveau_mem * nouveau_mem_alloc(GLcontext *ctx, int type, GLuint size, GLuint align) { nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); - drm_nouveau_mem_alloc_t mema; + struct drm_nouveau_mem_alloc mema; nouveau_mem *mem; int ret; @@ -111,7 +111,7 @@ nouveau_mem_alloc(GLcontext *ctx, int type, GLuint size, GLuint align) FREE(mem); return NULL; } - mem->offset = mema.region_offset; + mem->offset = mema.offset; mem->type = mema.flags; if (NOUVEAU_DEBUG & DEBUG_MEM) { @@ -120,7 +120,7 @@ nouveau_mem_alloc(GLcontext *ctx, int type, GLuint size, GLuint align) } if (type & NOUVEAU_MEM_MAPPED) - ret = drmMap(nmesa->driFd, mem->offset, mem->size, &mem->map); + ret = drmMap(nmesa->driFd, mema.map_handle, mem->size, &mem->map); if (ret) { mem->map = NULL; nouveau_mem_free(ctx, mem); @@ -135,12 +135,7 @@ nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem) { nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); - if (mem->type & NOUVEAU_MEM_FB) - return (uint32_t)mem->offset - nmesa->vram_phys; - else if (mem->type & NOUVEAU_MEM_AGP) - return (uint32_t)mem->offset - nmesa->gart_phys; - else - return 0xDEADF00D; + return mem->offset; } static GLboolean diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.c b/src/mesa/drivers/dri/nouveau/nouveau_context.c index d96b00242c..44392c0267 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_context.c @@ -180,7 +180,7 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual, driParseConfigFiles (&nmesa->optionCache, &screen->optionCache, screen->driScreen->myNum, "nouveau"); - nmesa->sarea = (drm_nouveau_sarea_t *)((char *)sPriv->pSAREA + + nmesa->sarea = (struct drm_nouveau_sarea *)((char *)sPriv->pSAREA + screen->sarea_priv_offset); /* Enable any supported extensions */ @@ -224,6 +224,8 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual, nv04TriInitFunctions( ctx ); break; case NV_10: + case NV_11: + case NV_17: case NV_20: case NV_30: case NV_40: diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h index 10d2ed6e17..9a0be2cb2a 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_context.h @@ -109,12 +109,12 @@ typedef struct nouveau_context { uint64_t gart_size; /* Channel synchronisation */ - drm_nouveau_notifier_alloc_t *syncNotifier; + struct drm_nouveau_notifier_alloc *syncNotifier; /* ARB_occlusion_query / EXT_timer_query */ GLuint query_object_max; GLboolean * query_alloc; - drm_nouveau_notifier_alloc_t *queryNotifier; + struct drm_nouveau_notifier_alloc *queryNotifier; /* Additional hw-specific functions */ nouveau_hw_func hw_func; @@ -168,7 +168,7 @@ typedef struct nouveau_context { nouveauShader *passthrough_fp; nouveauScreenRec *screen; - drm_nouveau_sarea_t *sarea; + struct drm_nouveau_sarea *sarea; __DRIcontextPrivate *driContext; /* DRI context */ __DRIscreenPrivate *driScreen; /* DRI screen */ diff --git a/src/mesa/drivers/dri/nouveau/nouveau_driver.c b/src/mesa/drivers/dri/nouveau/nouveau_driver.c index 00956aa8f8..ddc9535624 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_driver.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_driver.c @@ -41,7 +41,7 @@ GLboolean nouveauDRMGetParam(nouveauContextPtr nmesa, unsigned int param, uint64_t* value) { - drm_nouveau_getparam_t getp; + struct drm_nouveau_getparam getp; getp.param = param; if (!value || drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_GETPARAM, @@ -56,7 +56,7 @@ GLboolean nouveauDRMSetParam(nouveauContextPtr nmesa, unsigned int param, uint64_t value) { - drm_nouveau_setparam_t setp; + struct drm_nouveau_setparam setp; setp.param = param; setp.value = value; diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c index e9320918f9..7b5e96b4c2 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c @@ -98,7 +98,7 @@ void nouveauWaitForIdle(nouveauContextPtr nmesa) // here we call the fifo initialization ioctl and fill in stuff accordingly GLboolean nouveauFifoInit(nouveauContextPtr nmesa) { - drm_nouveau_fifo_alloc_t fifo_init; + struct drm_nouveau_fifo_alloc fifo_init; int i, ret; #ifdef NOUVEAU_RING_DEBUG diff --git a/src/mesa/drivers/dri/nouveau/nouveau_lock.c b/src/mesa/drivers/dri/nouveau/nouveau_lock.c index c119d14dd7..aa86c9e783 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_lock.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_lock.c @@ -44,7 +44,7 @@ void nouveauGetLock( nouveauContextPtr nmesa, GLuint flags ) { __DRIdrawablePrivate *dPriv = nmesa->driDrawable; __DRIscreenPrivate *sPriv = nmesa->driScreen; - drm_nouveau_sarea_t *sarea = nmesa->sarea; + struct drm_nouveau_sarea *sarea = nmesa->sarea; drmGetLock( nmesa->driFd, nmesa->hHWContext, flags ); diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.c b/src/mesa/drivers/dri/nouveau/nouveau_object.c index 69f8dbf794..a143488e8d 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_object.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_object.c @@ -7,7 +7,7 @@ GLboolean nouveauCreateContextObject(nouveauContextPtr nmesa, uint32_t handle, int class) { - drm_nouveau_grobj_alloc_t cto; + struct drm_nouveau_grobj_alloc cto; int ret; cto.channel = nmesa->fifo.channel; @@ -34,10 +34,13 @@ void nouveauObjectInit(nouveauContextPtr nmesa) nouveauCreateContextObject(nmesa, Nv3D, nmesa->screen->card->class_3d); if (nmesa->screen->card->type>=NV_10) { nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV10_CONTEXT_SURFACES_2D); - nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT); } else { nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D); nouveauCreateContextObject(nmesa, NvCtxSurf3D, NV04_CONTEXT_SURFACES_3D); + } + if (nmesa->screen->card->type>=NV_11) { + nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT); + } else { nouveauCreateContextObject(nmesa, NvImageBlit, NV_IMAGE_BLIT); } nouveauCreateContextObject(nmesa, NvMemFormat, NV_MEMORY_TO_MEMORY_FORMAT); diff --git a/src/mesa/drivers/dri/nouveau/nouveau_screen.c b/src/mesa/drivers/dri/nouveau/nouveau_screen.c index bc7f39b042..69b0691bb7 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_screen.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_screen.c @@ -328,7 +328,7 @@ void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIsc static const __DRIversion ddx_expected = { 1, 2, 0 }; static const __DRIversion dri_expected = { 4, 0, 0 }; static const __DRIversion drm_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL }; -#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 7 +#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 9 #error nouveau_drm.h version doesn't match expected version #endif dri_interface = interface; diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c index 7cb805902a..f618dcfc99 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_state.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_state.c @@ -162,6 +162,8 @@ void nouveauDDInitState(nouveauContextPtr nmesa) nv04InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver); break; case NV_10: + case NV_11: + case NV_17: nv10InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver); break; case NV_20: diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.c b/src/mesa/drivers/dri/nouveau/nouveau_sync.c index 1d1eeede18..8abc847e1e 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_sync.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.c @@ -39,11 +39,11 @@ nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); \ volatile uint32_t *__v = (void*)nmesa->notifier_block + notifier->offset -drm_nouveau_notifier_alloc_t * +struct drm_nouveau_notifier_alloc * nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count) { nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); - drm_nouveau_notifier_alloc_t *notifier; + struct drm_nouveau_notifier_alloc *notifier; int ret; #ifdef NOUVEAU_RING_DEBUG @@ -69,14 +69,16 @@ nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count) } void -nouveau_notifier_destroy(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier) +nouveau_notifier_destroy(GLcontext *ctx, + struct drm_nouveau_notifier_alloc *notifier) { /*XXX: free notifier object.. */ FREE(notifier); } void -nouveau_notifier_reset(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier, +nouveau_notifier_reset(GLcontext *ctx, + struct drm_nouveau_notifier_alloc *notifier, GLuint id) { NOTIFIER(n); @@ -93,7 +95,8 @@ nouveau_notifier_reset(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier, } GLuint -nouveau_notifier_status(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier, +nouveau_notifier_status(GLcontext *ctx, + struct drm_nouveau_notifier_alloc *notifier, GLuint id) { NOTIFIER(n); @@ -103,7 +106,8 @@ nouveau_notifier_status(GLcontext *ctx, drm_nouveau_notifier_alloc_t *notifier, GLuint nouveau_notifier_return_val(GLcontext *ctx, - drm_nouveau_notifier_alloc_t *notifier, GLuint id) + struct drm_nouveau_notifier_alloc *notifier, + GLuint id) { NOTIFIER(n); @@ -112,8 +116,8 @@ nouveau_notifier_return_val(GLcontext *ctx, GLboolean nouveau_notifier_wait_status(GLcontext *ctx, - drm_nouveau_notifier_alloc_t *notifier, GLuint id, - GLuint status, GLuint timeout) + struct drm_nouveau_notifier_alloc *notifier, + GLuint id, GLuint status, GLuint timeout) { NOTIFIER(n); unsigned int time = 0; @@ -146,7 +150,8 @@ nouveau_notifier_wait_status(GLcontext *ctx, void nouveau_notifier_wait_nop(GLcontext *ctx, - drm_nouveau_notifier_alloc_t *notifier, GLuint subc) + struct drm_nouveau_notifier_alloc *notifier, + GLuint subc) { NOTIFIER(n); GLboolean ret; diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.h b/src/mesa/drivers/dri/nouveau/nouveau_sync.h index b56cc5fb54..b76af17276 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_sync.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.h @@ -47,22 +47,24 @@ #define NV_NOTIFY 0x00000104 #define NV_NOTIFY_STYLE_WRITE_ONLY 0 -extern drm_nouveau_notifier_alloc_t * +extern struct drm_nouveau_notifier_alloc * nouveau_notifier_new(GLcontext *, GLuint handle, GLuint count); extern void -nouveau_notifier_destroy(GLcontext *, drm_nouveau_notifier_alloc_t *); +nouveau_notifier_destroy(GLcontext *, struct drm_nouveau_notifier_alloc *); extern void -nouveau_notifier_reset(GLcontext *, drm_nouveau_notifier_alloc_t *, GLuint id); +nouveau_notifier_reset(GLcontext *, struct drm_nouveau_notifier_alloc *, + GLuint id); extern GLuint -nouveau_notifier_status(GLcontext *, drm_nouveau_notifier_alloc_t *, GLuint id); +nouveau_notifier_status(GLcontext *, struct drm_nouveau_notifier_alloc *, + GLuint id); extern GLuint -nouveau_notifier_return_val(GLcontext *, drm_nouveau_notifier_alloc_t *, +nouveau_notifier_return_val(GLcontext *, struct drm_nouveau_notifier_alloc *, GLuint id); extern GLboolean -nouveau_notifier_wait_status(GLcontext *, drm_nouveau_notifier_alloc_t *, +nouveau_notifier_wait_status(GLcontext *, struct drm_nouveau_notifier_alloc *, GLuint id, GLuint status, GLuint timeout); extern void -nouveau_notifier_wait_nop(GLcontext *ctx, drm_nouveau_notifier_alloc_t *, +nouveau_notifier_wait_nop(GLcontext *ctx, struct drm_nouveau_notifier_alloc *, GLuint subc); extern GLboolean nouveauSyncInitFuncs(GLcontext *ctx); diff --git a/src/mesa/drivers/dri/nouveau/nv10_state.c b/src/mesa/drivers/dri/nouveau/nv10_state.c index 5f304ccab9..47c4b14ba6 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_state.c +++ b/src/mesa/drivers/dri/nouveau/nv10_state.c @@ -697,8 +697,7 @@ static GLboolean nv10InitCard(nouveauContextPtr nmesa) BEGIN_RING_SIZE(NvSub3D, 0x03f4, 1); OUT_RING(0); - /* not for nv10, only for >= nv11 */ - if ((nmesa->screen->card->id>>4) >= 0x11) { + if (nmesa->screen->card->type >= NV_11) { BEGIN_RING_SIZE(NvSub3D, 0x120, 3); OUT_RING(0); OUT_RING(1); @@ -739,11 +738,11 @@ static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color, OUT_RING_CACHE(depth ? depth->offset : color[0]->offset); /* Always set to bottom left of buffer */ - BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4); + /*BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4); OUT_RING_CACHEf (0.0); OUT_RING_CACHEf ((GLfloat) h); OUT_RING_CACHEf (0.0); - OUT_RING_CACHEf (0.0); + OUT_RING_CACHEf (0.0);*/ return GL_TRUE; } diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c index 4576c1ede4..611469b6e4 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c +++ b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c @@ -58,7 +58,7 @@ static void nv10ResetLineStipple( GLcontext *ctx ); static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t primitive,uint32_t size) { - if (nmesa->screen->card->type==NV_10) + if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1); else if (nmesa->screen->card->type==NV_20) BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_BEGIN_END,1); @@ -66,7 +66,7 @@ static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t pri BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_BEGIN_END,1); OUT_RING(primitive); - if (nmesa->screen->card->type==NV_10) + if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA|NONINC_METHOD,size); else if (nmesa->screen->card->type==NV_20) BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_VERTEX_DATA|NONINC_METHOD,size); @@ -76,7 +76,7 @@ static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t pri inline void nv10FinishPrimitive(struct nouveau_context *nmesa) { - if (nmesa->screen->card->type==NV_10) + if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1); else if (nmesa->screen->card->type==NV_20) BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_BEGIN_END,1); @@ -454,7 +454,7 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa) /* * Tell the hardware about the vertex format */ - if (nmesa->screen->card->type==NV_10) { + if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) { int size; #define NV_VERTEX_ATTRIBUTE_TYPE_FLOAT 2 diff --git a/src/mesa/drivers/dri/nouveau/nv20_state.c b/src/mesa/drivers/dri/nouveau/nv20_state.c index 3d8d83a865..ccf2f6148b 100644 --- a/src/mesa/drivers/dri/nouveau/nv20_state.c +++ b/src/mesa/drivers/dri/nouveau/nv20_state.c @@ -568,10 +568,10 @@ static void nv20Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h) y += nmesa->drawY; } - BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1, 1); + /*BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1, 1); OUT_RING_CACHE((w << 16) | x ); BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1, 1); - OUT_RING_CACHE((h << 16) | y ); + OUT_RING_CACHE((h << 16) | y );*/ } @@ -764,11 +764,11 @@ static GLboolean nv20BindBuffers(nouveauContextPtr nmesa, int num_color, } /* Always set to bottom left of buffer */ - BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4); + /*BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4); OUT_RING_CACHEf (0.0); OUT_RING_CACHEf ((GLfloat) h); OUT_RING_CACHEf (0.0); - OUT_RING_CACHEf (0.0); + OUT_RING_CACHEf (0.0);*/ return GL_TRUE; } diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c index 2115799b9b..1d975ecd57 100644 --- a/src/mesa/drivers/dri/r200/r200_state.c +++ b/src/mesa/drivers/dri/r200/r200_state.c @@ -772,9 +772,11 @@ static void r200LineWidth( GLcontext *ctx, GLfloat widthf ) R200_STATECHANGE( rmesa, set ); /* Line width is stored in U6.4 format. + * Same min/max limits for AA, non-AA lines. */ rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] &= ~0xffff; - rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint)(ctx->Line._Width * 16.0); + rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint) + (CLAMP(widthf, ctx->Const.MinLineWidth, ctx->Const.MaxLineWidth) * 16.0); if ( widthf > 1.0 ) { rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_WIDELINE_ENABLE; diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h index 6615bc79fb..be6909724a 100644 --- a/src/mesa/drivers/dri/r300/r300_context.h +++ b/src/mesa/drivers/dri/r300/r300_context.h @@ -568,19 +568,10 @@ struct r300_vertex_shader_fragment { union { GLuint d[VSF_MAX_FRAGMENT_LENGTH]; float f[VSF_MAX_FRAGMENT_LENGTH]; - VERTEX_SHADER_INSTRUCTION i[VSF_MAX_FRAGMENT_LENGTH / 4]; + GLuint i[VSF_MAX_FRAGMENT_LENGTH]; } body; }; -#define VSF_DEST_PROGRAM 0x0 -#define VSF_DEST_MATRIX0 0x200 -#define VSF_DEST_MATRIX1 0x204 -#define VSF_DEST_MATRIX2 0x208 -#define VSF_DEST_VECTOR0 0x20c -#define VSF_DEST_VECTOR1 0x20d -#define VSF_DEST_UNKNOWN1 0x400 -#define VSF_DEST_UNKNOWN2 0x406 - struct r300_vertex_shader_state { struct r300_vertex_shader_fragment program; }; diff --git a/src/mesa/drivers/dri/r300/r300_emit.c b/src/mesa/drivers/dri/r300/r300_emit.c index 229439dfa8..424bf44e59 100644 --- a/src/mesa/drivers/dri/r300/r300_emit.c +++ b/src/mesa/drivers/dri/r300/r300_emit.c @@ -299,13 +299,14 @@ GLuint r300VAPOutputCntl0(GLcontext * ctx, GLuint OutputsWritten) if (OutputsWritten & (1 << VERT_RESULT_COL1)) ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT; -#if 0 - if (OutputsWritten & (1 << VERT_RESULT_BFC0)) - ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT; - - if (OutputsWritten & (1 << VERT_RESULT_BFC1)) - ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT; + if (OutputsWritten & (1 << VERT_RESULT_BFC0) + || OutputsWritten & (1 << VERT_RESULT_BFC1)) + ret |= + R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT | + R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT | + R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT; +#if 0 if (OutputsWritten & (1 << VERT_RESULT_FOGC)) ; #endif diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h index e59919be49..1baa74c526 100644 --- a/src/mesa/drivers/dri/r300/r300_reg.h +++ b/src/mesa/drivers/dri/r300/r300_reg.h @@ -282,9 +282,32 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. */ #define R300_VAP_PVS_UPLOAD_ADDRESS 0x2200 # define R300_PVS_UPLOAD_PROGRAM 0x00000000 +/* gap */ # define R300_PVS_UPLOAD_PARAMETERS 0x00000200 +/* gap */ +# define R300_PVS_UPLOAD_CLIP_PLANE0 0x00000400 +# define R300_PVS_UPLOAD_CLIP_PLANE1 0x00000401 +# define R300_PVS_UPLOAD_CLIP_PLANE2 0x00000402 +# define R300_PVS_UPLOAD_CLIP_PLANE3 0x00000403 +# define R300_PVS_UPLOAD_CLIP_PLANE4 0x00000404 +# define R300_PVS_UPLOAD_CLIP_PLANE5 0x00000405 # define R300_PVS_UPLOAD_POINTSIZE 0x00000406 +/* + * These are obsolete defines form r300_context.h, but they might give some + * clues when investigating the addresses further... + */ +#if 0 +#define VSF_DEST_PROGRAM 0x0 +#define VSF_DEST_MATRIX0 0x200 +#define VSF_DEST_MATRIX1 0x204 +#define VSF_DEST_MATRIX2 0x208 +#define VSF_DEST_VECTOR0 0x20c +#define VSF_DEST_VECTOR1 0x20d +#define VSF_DEST_UNKNOWN1 0x400 +#define VSF_DEST_UNKNOWN2 0x406 +#endif + /* gap */ #define R300_VAP_PVS_UPLOAD_DATA 0x2208 diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c index b5cf21d644..088216c76e 100644 --- a/src/mesa/drivers/dri/r300/r300_state.c +++ b/src/mesa/drivers/dri/r300/r300_state.c @@ -82,6 +82,8 @@ static void r300BlendColor(GLcontext * ctx, const GLfloat cf[4]) rmesa->hw.blend_color.cmd[1] = PACK_COLOR_8888(color[3], color[0], color[1], color[2]); + rmesa->hw.blend_color.cmd[2] = 0; + rmesa->hw.blend_color.cmd[3] = 0; } /** @@ -317,20 +319,34 @@ static void r300UpdateCulling(GLcontext * ctx) r300ContextPtr r300 = R300_CONTEXT(ctx); uint32_t val = 0; - R300_STATECHANGE(r300, cul); if (ctx->Polygon.CullFlag) { - if (ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) - val = R300_CULL_FRONT | R300_CULL_BACK; - else if (ctx->Polygon.CullFaceMode == GL_FRONT) + switch (ctx->Polygon.CullFaceMode) { + case GL_FRONT: val = R300_CULL_FRONT; - else + break; + case GL_BACK: val = R300_CULL_BACK; + break; + case GL_FRONT_AND_BACK: + val = R300_CULL_FRONT | R300_CULL_BACK; + break; + default: + break; + } + } - if (ctx->Polygon.FrontFace == GL_CW) - val |= R300_FRONT_FACE_CW; - else - val |= R300_FRONT_FACE_CCW; + switch (ctx->Polygon.FrontFace) { + case GL_CW: + val |= R300_FRONT_FACE_CW; + break; + case GL_CCW: + val |= R300_FRONT_FACE_CCW; + break; + default: + break; } + + R300_STATECHANGE(r300, cul); r300->hw.cul.cmd[R300_CUL_CULL] = val; } @@ -344,6 +360,20 @@ static void r300SetEarlyZState(GLcontext * ctx) r300ContextPtr r300 = R300_CONTEXT(ctx); R300_STATECHANGE(r300, zstencil_format); + switch (ctx->Visual.depthBits) { + case 16: + r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z; + break; + case 24: + r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z; + break; + default: + fprintf(stderr, "Error: Unsupported depth %d... exiting\n", ctx->Visual.depthBits); + _mesa_exit(-1); + } + + // r300->hw.zstencil_format.cmd[1] |= R300_DEPTH_FORMAT_UNK32; + if (ctx->Color.AlphaEnabled && ctx->Color.AlphaFunc != GL_ALWAYS) /* disable early Z */ r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE; @@ -355,6 +385,9 @@ static void r300SetEarlyZState(GLcontext * ctx) /* disable early Z */ r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE; } + + r300->hw.zstencil_format.cmd[3] = 0x00000003; + r300->hw.zstencil_format.cmd[4] = 0x00000000; } static void r300SetAlphaState(GLcontext * ctx) @@ -403,6 +436,7 @@ static void r300SetAlphaState(GLcontext * ctx) R300_STATECHANGE(r300, at); r300->hw.at.cmd[R300_AT_ALPHA_TEST] = pp_misc; + r300->hw.at.cmd[R300_AT_UNKNOWN] = 0; r300SetEarlyZState(ctx); } @@ -513,6 +547,9 @@ static void r300UpdatePolygonMode(GLcontext * ctx) R300_STATECHANGE(r300, polygon_mode); r300->hw.polygon_mode.cmd[1] = hw_mode; } + + r300->hw.polygon_mode.cmd[2] = 0x00000001; + r300->hw.polygon_mode.cmd[3] = 0x00000000; } /** @@ -696,8 +733,8 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param) static void r300PointSize(GLcontext * ctx, GLfloat size) { r300ContextPtr r300 = R300_CONTEXT(ctx); - - size = ctx->Point._Size; + /* same size limits for AA, non-AA points */ + size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize); R300_STATECHANGE(r300, ps); r300->hw.ps.cmd[R300_PS_POINTSIZE] = @@ -712,8 +749,9 @@ static void r300LineWidth(GLcontext * ctx, GLfloat widthf) { r300ContextPtr r300 = R300_CONTEXT(ctx); - widthf = ctx->Line._Width; - + widthf = CLAMP(widthf, + ctx->Const.MinPointSize, + ctx->Const.MaxPointSize); R300_STATECHANGE(r300, lcntl); r300->hw.lcntl.cmd[1] = R300_LINE_CNT_HO | R300_LINE_CNT_VE | (int)(widthf * 6.0); @@ -762,6 +800,7 @@ static void r300ShadeModel(GLcontext * ctx, GLenum mode) r300ContextPtr rmesa = R300_CONTEXT(ctx); R300_STATECHANGE(rmesa, shade); + rmesa->hw.shade.cmd[1] = 0x00000002; switch (mode) { case GL_FLAT: rmesa->hw.shade.cmd[2] = R300_RE_SHADE_MODEL_FLAT; @@ -772,6 +811,8 @@ static void r300ShadeModel(GLcontext * ctx, GLenum mode) default: return; } + rmesa->hw.shade.cmd[3] = 0x00000000; + rmesa->hw.shade.cmd[4] = 0x00000000; } static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face, @@ -1526,30 +1567,31 @@ static void r300SetupDefaultVertexProgram(r300ContextPtr rmesa) for (i = VERT_ATTRIB_POS; i < VERT_ATTRIB_MAX; i++) { if (rmesa->state.sw_tcl_inputs[i] != -1) { - prog->program.body.i[program_end].op = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT); - prog->program.body.i[program_end].src[0] = VSF_REG(rmesa->state.sw_tcl_inputs[i]); - prog->program.body.i[program_end].src[1] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]); - prog->program.body.i[program_end].src[2] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]); - program_end++; + prog->program.body.i[program_end + 0] = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT); + prog->program.body.i[program_end + 1] = VSF_REG(rmesa->state.sw_tcl_inputs[i]); + prog->program.body.i[program_end + 2] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]); + prog->program.body.i[program_end + 3] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]); + program_end += 4; } } - prog->program.length = program_end * 4; + prog->program.length = program_end; - r300SetupVertexProgramFragment(rmesa, VSF_DEST_PROGRAM, &(prog->program)); + r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM, + &(prog->program)); inst_count = (prog->program.length / 4) - 1; R300_STATECHANGE(rmesa, pvs); rmesa->hw.pvs.cmd[R300_PVS_CNTL_1] = - (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) | - (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) | - (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT); + (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) | + (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) | + (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT); rmesa->hw.pvs.cmd[R300_PVS_CNTL_2] = - (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) | - (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT); + (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) | + (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT); rmesa->hw.pvs.cmd[R300_PVS_CNTL_3] = - (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) | - (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT); + (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) | + (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT); } static void r300SetupRealVertexProgram(r300ContextPtr rmesa) @@ -1570,7 +1612,7 @@ static void r300SetupRealVertexProgram(r300ContextPtr rmesa) bump_vpu_count(rmesa->hw.vpp.cmd, param_count); param_count /= 4; - r300SetupVertexProgramFragment(rmesa, VSF_DEST_PROGRAM, &(prog->program)); + r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM, &(prog->program)); inst_count = (prog->program.length / 4) - 1; R300_STATECHANGE(rmesa, pvs); @@ -1848,15 +1890,10 @@ static void r300ResetHwState(r300ContextPtr r300) r300->hw.unk4260.cmd[2] = r300PackFloat32(0.0); r300->hw.unk4260.cmd[3] = r300PackFloat32(1.0); - r300->hw.shade.cmd[1] = 0x00000002; r300ShadeModel(ctx, ctx->Light.ShadeModel); - r300->hw.shade.cmd[3] = 0x00000000; - r300->hw.shade.cmd[4] = 0x00000000; r300PolygonMode(ctx, GL_FRONT, ctx->Polygon.FrontMode); r300PolygonMode(ctx, GL_BACK, ctx->Polygon.BackMode); - r300->hw.polygon_mode.cmd[2] = 0x00000001; - r300->hw.polygon_mode.cmd[3] = 0x00000000; r300->hw.zbias_cntl.cmd[1] = 0x00000000; r300PolygonOffset(ctx, ctx->Polygon.OffsetFactor, @@ -1887,14 +1924,11 @@ static void r300ResetHwState(r300ContextPtr r300) r300Fogfv(ctx, GL_FOG_COLOR, ctx->Fog.Color); r300Fogfv(ctx, GL_FOG_COORDINATE_SOURCE_EXT, NULL); - r300->hw.at.cmd[R300_AT_UNKNOWN] = 0; r300->hw.unk4BD8.cmd[1] = 0; r300->hw.unk4E00.cmd[1] = 0; r300BlendColor(ctx, ctx->Color.BlendColor); - r300->hw.blend_color.cmd[2] = 0; - r300->hw.blend_color.cmd[3] = 0; /* Again, r300ClearBuffer uses this */ r300->hw.cb.cmd[R300_CB_OFFSET] = @@ -1925,25 +1959,6 @@ static void r300ResetHwState(r300ContextPtr r300) r300->hw.unk4EA0.cmd[1] = 0x00000000; r300->hw.unk4EA0.cmd[2] = 0xffffffff; - switch (ctx->Visual.depthBits) { - case 16: - r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z; - break; - case 24: - r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z; - break; - default: - fprintf(stderr, "Error: Unsupported depth %d... exiting\n", - ctx->Visual.depthBits); - _mesa_exit(-1); - - } - /* z compress? */ - //r300->hw.zstencil_format.cmd[1] |= R300_DEPTH_FORMAT_UNK32; - - r300->hw.zstencil_format.cmd[3] = 0x00000003; - r300->hw.zstencil_format.cmd[4] = 0x00000000; - r300->hw.zb.cmd[R300_ZB_OFFSET] = r300->radeon.radeonScreen->depthOffset + r300->radeon.radeonScreen->fbLocation; diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c index 16dddf6557..7d4e8c9511 100644 --- a/src/mesa/drivers/dri/r300/r300_vertprog.c +++ b/src/mesa/drivers/dri/r300/r300_vertprog.c @@ -29,6 +29,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. * \file * * \author Aapo Tahkola <aet@rasterburn.org> + * + * \author Oliver McFadden <z3ro.geek@gmail.com> + * + * For a description of the vertex program instruction set see r300_reg.h. */ #include "glheader.h" @@ -55,54 +59,58 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. #error Cannot change these! #endif -#define SCALAR_FLAG (1<<31) -#define FLAG_MASK (1<<31) -#define OP_MASK (0xf) /* we are unlikely to have more than 15 */ -#define OPN(operator, ip) {#operator, OPCODE_##operator, ip} - -static struct { - char *name; - int opcode; - unsigned long ip; /* number of input operands and flags */ -} op_names[] = { - /* *INDENT-OFF* */ - OPN(ABS, 1), - OPN(ADD, 2), - OPN(ARL, 1 | SCALAR_FLAG), - OPN(DP3, 2), - OPN(DP4, 2), - OPN(DPH, 2), - OPN(DST, 2), - OPN(EX2, 1 | SCALAR_FLAG), - OPN(EXP, 1 | SCALAR_FLAG), - OPN(FLR, 1), - OPN(FRC, 1), - OPN(LG2, 1 | SCALAR_FLAG), - OPN(LIT, 1), - OPN(LOG, 1 | SCALAR_FLAG), - OPN(MAD, 3), - OPN(MAX, 2), - OPN(MIN, 2), - OPN(MOV, 1), - OPN(MUL, 2), - OPN(POW, 2 | SCALAR_FLAG), - OPN(RCP, 1 | SCALAR_FLAG), - OPN(RSQ, 1 | SCALAR_FLAG), - OPN(SGE, 2), - OPN(SLT, 2), - OPN(SUB, 2), - OPN(SWZ, 1), - OPN(XPD, 2), - OPN(RCC, 0), //extra - OPN(PRINT, 0), - OPN(END, 0) - /* *INDENT-ON* */ -}; - -#undef OPN +/* TODO: Get rid of t_src_class call */ +#define CMP_SRCS(a, b) ((a.RelAddr != b.RelAddr) || (a.Index != b.Index && \ + ((t_src_class(a.File) == VSF_IN_CLASS_PARAM && \ + t_src_class(b.File) == VSF_IN_CLASS_PARAM) || \ + (t_src_class(a.File) == VSF_IN_CLASS_ATTR && \ + t_src_class(b.File) == VSF_IN_CLASS_ATTR)))) \ + +#define ZERO_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \ + SWIZZLE_ZERO, SWIZZLE_ZERO, \ + SWIZZLE_ZERO, SWIZZLE_ZERO, \ + t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4)) + +#define ZERO_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \ + SWIZZLE_ZERO, SWIZZLE_ZERO, \ + SWIZZLE_ZERO, SWIZZLE_ZERO, \ + t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4)) + +#define ZERO_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \ + SWIZZLE_ZERO, SWIZZLE_ZERO, \ + SWIZZLE_ZERO, SWIZZLE_ZERO, \ + t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4)) + +#define ONE_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \ + SWIZZLE_ONE, SWIZZLE_ONE, \ + SWIZZLE_ONE, SWIZZLE_ONE, \ + t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4)) + +#define ONE_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \ + SWIZZLE_ONE, SWIZZLE_ONE, \ + SWIZZLE_ONE, SWIZZLE_ONE, \ + t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4)) + +#define ONE_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \ + SWIZZLE_ONE, SWIZZLE_ONE, \ + SWIZZLE_ONE, SWIZZLE_ONE, \ + t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4)) + +/* DP4 version seems to trigger some hw peculiarity */ +//#define PREFER_DP4 + +#define FREE_TEMPS() \ + do { \ + if(u_temp_i < vp->num_temporaries) { \ + WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_i); \ + vp->native = GL_FALSE; \ + } \ + u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1; \ + } while (0) int r300VertexProgUpdateParams(GLcontext * ctx, - struct r300_vertex_program_cont *vp, float *dst) + struct r300_vertex_program_cont *vp, + float *dst) { int pi; struct gl_vertex_program *mesa_vp = &vp->mesa_program; @@ -234,8 +242,8 @@ static void vp_dump_inputs(struct r300_vertex_program *vp, char *caller) int i; if (vp == NULL) { - fprintf(stderr, "vp null in call to %s from %s\n", __FUNCTION__, - caller); + fprintf(stderr, "vp null in call to %s from %s\n", + __FUNCTION__, caller); return; } @@ -276,6 +284,8 @@ static unsigned long t_src_index(struct r300_vertex_program *vp, } } +/* these two functions should probably be merged... */ + static unsigned long t_src(struct r300_vertex_program *vp, struct prog_src_register *src) { @@ -294,7 +304,9 @@ static unsigned long t_src(struct r300_vertex_program *vp, static unsigned long t_src_scalar(struct r300_vertex_program *vp, struct prog_src_register *src) { - + /* src->NegateBase uses the NEGATE_ flags from program_instruction.h, + * which equal our VSF_FLAGS_ values, so it's safe to just pass it here. + */ return MAKE_VSF_SOURCE(t_src_index(vp, src), t_swizzle(GET_SWZ(src->Swizzle, 0)), t_swizzle(GET_SWZ(src->Swizzle, 0)), @@ -306,128 +318,741 @@ static unsigned long t_src_scalar(struct r300_vertex_program *vp, (src->RelAddr << 4); } -static unsigned long t_opcode(enum prog_opcode opcode) +static GLboolean valid_dst(struct r300_vertex_program *vp, + struct prog_dst_register *dst) { + if (dst->File == PROGRAM_OUTPUT && vp->outputs[dst->Index] == -1) { + return GL_FALSE; + } else if (dst->File == PROGRAM_ADDRESS) { + assert(dst->Index == 0); + } - switch (opcode) { - /* *INDENT-OFF* */ - case OPCODE_ARL: return R300_VPI_OUT_OP_ARL; - case OPCODE_DST: return R300_VPI_OUT_OP_DST; - case OPCODE_EX2: return R300_VPI_OUT_OP_EX2; - case OPCODE_EXP: return R300_VPI_OUT_OP_EXP; - case OPCODE_FRC: return R300_VPI_OUT_OP_FRC; - case OPCODE_LG2: return R300_VPI_OUT_OP_LG2; - case OPCODE_LOG: return R300_VPI_OUT_OP_LOG; - case OPCODE_MAX: return R300_VPI_OUT_OP_MAX; - case OPCODE_MIN: return R300_VPI_OUT_OP_MIN; - case OPCODE_MUL: return R300_VPI_OUT_OP_MUL; - case OPCODE_RCP: return R300_VPI_OUT_OP_RCP; - case OPCODE_RSQ: return R300_VPI_OUT_OP_RSQ; - case OPCODE_SGE: return R300_VPI_OUT_OP_SGE; - case OPCODE_SLT: return R300_VPI_OUT_OP_SLT; - case OPCODE_DP4: return R300_VPI_OUT_OP_DOT; - /* *INDENT-ON* */ + return GL_TRUE; +} - default: - fprintf(stderr, "%s: Should not be called with opcode %d!", - __FUNCTION__, opcode); - } - _mesa_exit(-1); - return 0; +/* + * Instruction Inputs Output Description + * ----------- ------ ------ -------------------------------- + * ABS v v absolute value + * ADD v,v v add + * ARL s a address register load + * DP3 v,v ssss 3-component dot product + * DP4 v,v ssss 4-component dot product + * DPH v,v ssss homogeneous dot product + * DST v,v v distance vector + * EX2 s ssss exponential base 2 + * EXP s v exponential base 2 (approximate) + * FLR v v floor + * FRC v v fraction + * LG2 s ssss logarithm base 2 + * LIT v v compute light coefficients + * LOG s v logarithm base 2 (approximate) + * MAD v,v,v v multiply and add + * MAX v,v v maximum + * MIN v,v v minimum + * MOV v v move + * MUL v,v v multiply + * POW s,s ssss exponentiate + * RCP s ssss reciprocal + * RSQ s ssss reciprocal square root + * SGE v,v v set on greater than or equal + * SLT v,v v set on less than + * SUB v,v v subtract + * SWZ v v extended swizzle + * XPD v,v v cross product + * + * Table X.5: Summary of vertex program instructions. "v" indicates a + * floating-point vector input or output, "s" indicates a floating-point + * scalar input, "ssss" indicates a scalar output replicated across a + * 4-component result vector, and "a" indicates a single address register + * component. + */ + +static GLuint *t_opcode_abs(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + //MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_swizzle(GET_SWZ(src[0].Swizzle, 1)), + t_swizzle(GET_SWZ(src[0].Swizzle, 2)), + t_swizzle(GET_SWZ(src[0].Swizzle, 3)), + t_src_class(src[0].File), + (!src[0]. + NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | + (src[0].RelAddr << 4); + inst[3] = 0; + + return inst; } -static unsigned long op_operands(enum prog_opcode opcode) +static GLuint *t_opcode_add(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) { - int i; + unsigned long hw_op; - /* Can we trust mesas opcodes to be in order ? */ - for (i = 0; i < sizeof(op_names) / sizeof(*op_names); i++) - if (op_names[i].opcode == opcode) - return op_names[i].ip; +#if 1 + hw_op = (src[0].File == PROGRAM_TEMPORARY + && src[1].File == + PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : + R300_VPI_OUT_OP_MAD; + + inst[0] = + MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = ONE_SRC_0; + inst[2] = t_src(vp, &src[0]); + inst[3] = t_src(vp, &src[1]); +#else + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; - fprintf(stderr, "op %d not found in op_names\n", opcode); - _mesa_exit(-1); - return 0; +#endif + + return inst; } -static GLboolean valid_dst(struct r300_vertex_program *vp, - struct prog_dst_register *dst) +static GLuint *t_opcode_arl(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) { - if (dst->File == PROGRAM_OUTPUT && vp->outputs[dst->Index] == -1) { - return GL_FALSE; - } else if (dst->File == PROGRAM_ADDRESS) { - assert(dst->Index == 0); - } + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ARL, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); - return GL_TRUE; + inst[1] = t_src(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; } -/* TODO: Get rid of t_src_class call */ -#define CMP_SRCS(a, b) ((a.RelAddr != b.RelAddr) || (a.Index != b.Index && \ - ((t_src_class(a.File) == VSF_IN_CLASS_PARAM && \ - t_src_class(b.File) == VSF_IN_CLASS_PARAM) || \ - (t_src_class(a.File) == VSF_IN_CLASS_ATTR && \ - t_src_class(b.File) == VSF_IN_CLASS_ATTR)))) \ +static GLuint *t_opcode_dp3(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + //DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO} + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_swizzle(GET_SWZ(src[0].Swizzle, 1)), + t_swizzle(GET_SWZ(src[0].Swizzle, 2)), + SWIZZLE_ZERO, t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) | + (src[0].RelAddr << 4); + + inst[2] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), + t_swizzle(GET_SWZ(src[1].Swizzle, 0)), + t_swizzle(GET_SWZ(src[1].Swizzle, 1)), + t_swizzle(GET_SWZ(src[1].Swizzle, 2)), + SWIZZLE_ZERO, t_src_class(src[1].File), + src[1]. + NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) | + (src[1].RelAddr << 4); + + inst[3] = ZERO_SRC_1; + + return inst; +} -#define ZERO_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \ - SWIZZLE_ZERO, SWIZZLE_ZERO, \ - SWIZZLE_ZERO, SWIZZLE_ZERO, \ - t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4)) +static GLuint *t_opcode_dp4(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); -#define ZERO_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \ - SWIZZLE_ZERO, SWIZZLE_ZERO, \ - SWIZZLE_ZERO, SWIZZLE_ZERO, \ - t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4)) + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; -#define ZERO_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \ - SWIZZLE_ZERO, SWIZZLE_ZERO, \ - SWIZZLE_ZERO, SWIZZLE_ZERO, \ - t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4)) + return inst; +} -#define ONE_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \ - SWIZZLE_ONE, SWIZZLE_ONE, \ - SWIZZLE_ONE, SWIZZLE_ONE, \ - t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4)) +static GLuint *t_opcode_dph(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + //DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W} + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_swizzle(GET_SWZ(src[0].Swizzle, 1)), + t_swizzle(GET_SWZ(src[0].Swizzle, 2)), + VSF_IN_COMPONENT_ONE, t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) | + (src[0].RelAddr << 4); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; + + return inst; +} -#define ONE_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \ - SWIZZLE_ONE, SWIZZLE_ONE, \ - SWIZZLE_ONE, SWIZZLE_ONE, \ - t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4)) +static GLuint *t_opcode_dst(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_DST, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); -#define ONE_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \ - SWIZZLE_ONE, SWIZZLE_ONE, \ - SWIZZLE_ONE, SWIZZLE_ONE, \ - t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4)) + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; -/* DP4 version seems to trigger some hw peculiarity */ -//#define PREFER_DP4 + return inst; +} -#define FREE_TEMPS() \ - do { \ - if(u_temp_i < vp->num_temporaries) { \ - WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_i); \ - vp->native = GL_FALSE; \ - } \ - u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1; \ - } while (0) +static GLuint *t_opcode_ex2(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_EX2, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); -static void r300TranslateVertexShader(struct r300_vertex_program *vp, - struct prog_instruction *vpi) + inst[1] = t_src_scalar(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_exp(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_EXP, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src_scalar(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_flr(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3], int *u_temp_i) +{ + /* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W} + ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */ + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, *u_temp_i, + t_dst_mask(vpi->DstReg.WriteMask), + VSF_OUT_CLASS_TMP); + + inst[1] = t_src(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + inst += 4; + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = + MAKE_VSF_SOURCE(*u_temp_i, VSF_IN_COMPONENT_X, + VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z, + VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP, + /* Not 100% sure about this */ + (!src[0]. + NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE + /*VSF_FLAG_ALL */ ); + + inst[3] = ZERO_SRC_0; + (*u_temp_i)--; + + return inst; +} + +static GLuint *t_opcode_frc(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_lg2(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + // LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X} + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_LG2, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), + t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | + (src[0].RelAddr << 4); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_lit(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + //LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W} + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_LIT, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + /* NOTE: Users swizzling might not work. */ + inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x + t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w + VSF_IN_COMPONENT_ZERO, // z + t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y + t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[0].RelAddr << 4); + inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y + t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w + VSF_IN_COMPONENT_ZERO, // z + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x + t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[0].RelAddr << 4); + inst[3] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x + VSF_IN_COMPONENT_ZERO, // z + t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w + t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[0].RelAddr << 4); + + return inst; +} + +static GLuint *t_opcode_log(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_LOG, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src_scalar(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_mad(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) { - int i, cur_reg = 0; - VERTEX_SHADER_INSTRUCTION *o_inst; - unsigned long operands; - int are_srcs_scalar; unsigned long hw_op; - /* Initial value should be last tmp reg that hw supports. - Strangely enough r300 doesnt mind even though these would be out of range. - Smart enough to realize that it doesnt need it? */ - int u_temp_i = VSF_MAX_FRAGMENT_TEMPS - 1; - struct prog_src_register src[3]; - vp->pos_end = 0; /* Not supported yet */ - vp->program.length = 0; - /*vp->num_temporaries=mesa_vp->Base.NumTemporaries; */ + hw_op = (src[0].File == PROGRAM_TEMPORARY + && src[1].File == PROGRAM_TEMPORARY + && src[2].File == + PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : + R300_VPI_OUT_OP_MAD; + + inst[0] = + MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = t_src(vp, &src[2]); + + return inst; +} + +static GLuint *t_opcode_max(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; + + return inst; +} + +static GLuint *t_opcode_min(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_MIN, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; + + return inst; +} + +static GLuint *t_opcode_mov(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + //ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO} + +#if 1 + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; +#else + hw_op = + (src[0].File == + PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : + R300_VPI_OUT_OP_MAD; + + inst[0] = + MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = ONE_SRC_0; + inst[3] = ZERO_SRC_0; +#endif + + return inst; +} + +static GLuint *t_opcode_mul(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + unsigned long hw_op; + + // HW mul can take third arg but appears to have some other limitations. + + hw_op = (src[0].File == PROGRAM_TEMPORARY + && src[1].File == + PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : + R300_VPI_OUT_OP_MAD; + + inst[0] = + MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + + inst[3] = ZERO_SRC_1; + + return inst; +} + +static GLuint *t_opcode_pow(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_POW, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src_scalar(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = t_src_scalar(vp, &src[1]); + + return inst; +} + +static GLuint *t_opcode_rcp(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_RCP, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src_scalar(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_rsq(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_RSQ, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src_scalar(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + + return inst; +} + +static GLuint *t_opcode_sge(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_SGE, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; + + return inst; +} + +static GLuint *t_opcode_slt(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_SLT, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = t_src(vp, &src[1]); + inst[3] = ZERO_SRC_1; + + return inst; +} + +static GLuint *t_opcode_sub(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + unsigned long hw_op; + + //ADD RESULT 1.X Y Z W TMP 0{} {X Y Z W} PARAM 1{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W + +#if 1 + hw_op = (src[0].File == PROGRAM_TEMPORARY + && src[1].File == + PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : + R300_VPI_OUT_OP_MAD; + + inst[0] = + MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = ONE_SRC_0; + inst[3] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), + t_swizzle(GET_SWZ(src[1].Swizzle, 0)), + t_swizzle(GET_SWZ(src[1].Swizzle, 1)), + t_swizzle(GET_SWZ(src[1].Swizzle, 2)), + t_swizzle(GET_SWZ(src[1].Swizzle, 3)), + t_src_class(src[1].File), + (!src[1]. + NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | + (src[1].RelAddr << 4); +#else + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = t_src(vp, &src[0]); + inst[2] = + MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), + t_swizzle(GET_SWZ(src[1].Swizzle, 0)), + t_swizzle(GET_SWZ(src[1].Swizzle, 1)), + t_swizzle(GET_SWZ(src[1].Swizzle, 2)), + t_swizzle(GET_SWZ(src[1].Swizzle, 3)), + t_src_class(src[1].File), + (!src[1]. + NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | + (src[1].RelAddr << 4); + inst[3] = 0; +#endif + + return inst; +} + +static GLuint *t_opcode_swz(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3]) +{ + //ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO} + +#if 1 + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; +#else + hw_op = + (src[0].File == + PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : + R300_VPI_OUT_OP_MAD; + + inst[0] = + MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + inst[1] = t_src(vp, &src[0]); + inst[2] = ONE_SRC_0; + inst[3] = ZERO_SRC_0; +#endif + + return inst; +} + +static GLuint *t_opcode_xpd(struct r300_vertex_program *vp, + struct prog_instruction *vpi, GLuint * inst, + struct prog_src_register src[3], int *u_temp_i) +{ + /* mul r0, r1.yzxw, r2.zxyw + mad r0, -r2.yzxw, r1.zxyw, r0 + NOTE: might need MAD_2 + */ + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, *u_temp_i, + t_dst_mask(vpi->DstReg.WriteMask), + VSF_OUT_CLASS_TMP); + + inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y + t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x + t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w + t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[0].RelAddr << 4); + + inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z + t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x + t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y + t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w + t_src_class(src[1].File), + src[1]. + NegateBase ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[1].RelAddr << 4); + + inst[3] = ZERO_SRC_1; + inst += 4; + (*u_temp_i)--; + + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, t_dst_index(vp, &vpi->DstReg), + t_dst_mask(vpi->DstReg.WriteMask), + t_dst_class(vpi->DstReg.File)); + + inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y + t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z + t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x + t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w + t_src_class(src[1].File), + (!src[1]. + NegateBase) ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[1].RelAddr << 4); + + inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z + t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x + t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y + t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w + t_src_class(src[0].File), + src[0]. + NegateBase ? VSF_FLAG_ALL : + VSF_FLAG_NONE) | (src[0].RelAddr << 4); + + inst[3] = + MAKE_VSF_SOURCE(*u_temp_i + 1, VSF_IN_COMPONENT_X, + VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z, + VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP, + VSF_FLAG_NONE); + + return inst; +} + +static void t_inputs_outputs(struct r300_vertex_program *vp) +{ + int i; + int cur_reg = 0; for (i = 0; i < VERT_ATTRIB_MAX; i++) vp->inputs[i] = -1; @@ -437,39 +1062,71 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp, assert(vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS)); - /* Assign outputs */ - if (vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS)) + if (vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS)) { vp->outputs[VERT_RESULT_HPOS] = cur_reg++; + } - if (vp->key.OutputsWritten & (1 << VERT_RESULT_PSIZ)) + if (vp->key.OutputsWritten & (1 << VERT_RESULT_PSIZ)) { vp->outputs[VERT_RESULT_PSIZ] = cur_reg++; + } - if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL0)) + if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL0)) { vp->outputs[VERT_RESULT_COL0] = cur_reg++; + } - if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL1)) - vp->outputs[VERT_RESULT_COL1] = cur_reg++; - -#if 0 /* Not supported yet */ - if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC0)) - vp->outputs[VERT_RESULT_BFC0] = cur_reg++; + if (vp->key.OutputsWritten & (1 << VERT_RESULT_COL1)) { + vp->outputs[VERT_RESULT_COL1] = + vp->outputs[VERT_RESULT_COL0] + 1; + cur_reg = vp->outputs[VERT_RESULT_COL1] + 1; + } - if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC1)) - vp->outputs[VERT_RESULT_BFC1] = cur_reg++; + if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC0)) { + vp->outputs[VERT_RESULT_BFC0] = + vp->outputs[VERT_RESULT_COL0] + 2; + cur_reg = vp->outputs[VERT_RESULT_BFC0] + 2; + } - if (vp->key.OutputsWritten & (1 << VERT_RESULT_FOGC)) + if (vp->key.OutputsWritten & (1 << VERT_RESULT_BFC1)) { + vp->outputs[VERT_RESULT_BFC1] = + vp->outputs[VERT_RESULT_COL0] + 3; + cur_reg = vp->outputs[VERT_RESULT_BFC1] + 1; + } +#if 0 + if (vp->key.OutputsWritten & (1 << VERT_RESULT_FOGC)) { vp->outputs[VERT_RESULT_FOGC] = cur_reg++; + } #endif - for (i = VERT_RESULT_TEX0; i <= VERT_RESULT_TEX7; i++) - if (vp->key.OutputsWritten & (1 << i)) + for (i = VERT_RESULT_TEX0; i <= VERT_RESULT_TEX7; i++) { + if (vp->key.OutputsWritten & (1 << i)) { vp->outputs[i] = cur_reg++; + } + } +} + +static void r300TranslateVertexShader(struct r300_vertex_program *vp, + struct prog_instruction *vpi) +{ + int i; + GLuint *inst; + unsigned long num_operands; + /* Initial value should be last tmp reg that hw supports. + Strangely enough r300 doesnt mind even though these would be out of range. + Smart enough to realize that it doesnt need it? */ + int u_temp_i = VSF_MAX_FRAGMENT_TEMPS - 1; + struct prog_src_register src[3]; + vp->pos_end = 0; /* Not supported yet */ + vp->program.length = 0; + /*vp->num_temporaries=mesa_vp->Base.NumTemporaries; */ vp->translated = GL_TRUE; vp->native = GL_TRUE; - o_inst = vp->program.body.i; - for (; vpi->Opcode != OPCODE_END; vpi++, o_inst++) { + t_inputs_outputs(vp); + + for (inst = vp->program.body.i; vpi->Opcode != OPCODE_END; + vpi++, inst += 4) { + FREE_TEMPS(); if (!valid_dst(vp, &vpi->DstReg)) { @@ -478,61 +1135,62 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp, vpi->DstReg.Index = u_temp_i; } - operands = op_operands(vpi->Opcode); - are_srcs_scalar = operands & SCALAR_FLAG; - operands &= OP_MASK; + num_operands = _mesa_num_inst_src_regs(vpi->Opcode); - for (i = 0; i < operands; i++) + /* copy the sources (src) from mesa into a local variable... is this needed? */ + for (i = 0; i < num_operands; i++) { src[i] = vpi->SrcReg[i]; + } - if (operands == 3) { /* TODO: scalars */ + if (num_operands == 3) { /* TODO: scalars */ if (CMP_SRCS(src[1], src[2]) || CMP_SRCS(src[0], src[2])) { - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, u_temp_i, - VSF_FLAG_ALL, + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, + u_temp_i, VSF_FLAG_ALL, VSF_OUT_CLASS_TMP); - o_inst->src[0] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), + inst[1] = + MAKE_VSF_SOURCE(t_src_index + (vp, &src[2]), SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W, - t_src_class(src[2].File), - VSF_FLAG_NONE) | (src[2]. - RelAddr << - 4); + t_src_class(src[2]. + File), + VSF_FLAG_NONE) | + (src[2].RelAddr << 4); - o_inst->src[1] = ZERO_SRC_2; - o_inst->src[2] = ZERO_SRC_2; - o_inst++; + inst[2] = ZERO_SRC_2; + inst[3] = ZERO_SRC_2; + inst += 4; src[2].File = PROGRAM_TEMPORARY; src[2].Index = u_temp_i; src[2].RelAddr = 0; u_temp_i--; } - } - if (operands >= 2) { + if (num_operands >= 2) { if (CMP_SRCS(src[1], src[0])) { - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, u_temp_i, - VSF_FLAG_ALL, + inst[0] = + MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, + u_temp_i, VSF_FLAG_ALL, VSF_OUT_CLASS_TMP); - o_inst->src[0] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), + inst[1] = + MAKE_VSF_SOURCE(t_src_index + (vp, &src[0]), SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W, - t_src_class(src[0].File), - VSF_FLAG_NONE) | (src[0]. - RelAddr << - 4); + t_src_class(src[0]. + File), + VSF_FLAG_NONE) | + (src[0].RelAddr << 4); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = ZERO_SRC_0; - o_inst++; + inst[2] = ZERO_SRC_0; + inst[3] = ZERO_SRC_0; + inst += 4; src[0].File = PROGRAM_TEMPORARY; src[0].Index = u_temp_i; @@ -541,517 +1199,103 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp, } } - /* These ops need special handling. */ switch (vpi->Opcode) { - case OPCODE_POW: - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_POW, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src_scalar(vp, &src[0]); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = t_src_scalar(vp, &src[1]); - goto next; - - case OPCODE_MOV: //ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO} - case OPCODE_SWZ: -#if 1 - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = ZERO_SRC_0; -#else - hw_op = - (src[0].File == - PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : - R300_VPI_OUT_OP_MAD; - - o_inst->op = - MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = ONE_SRC_0; - o_inst->src[2] = ZERO_SRC_0; -#endif - - goto next; - + case OPCODE_ABS: + inst = t_opcode_abs(vp, vpi, inst, src); + break; case OPCODE_ADD: -#if 1 - hw_op = (src[0].File == PROGRAM_TEMPORARY && - src[1].File == - PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : - R300_VPI_OUT_OP_MAD; - - o_inst->op = - MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = ONE_SRC_0; - o_inst->src[1] = t_src(vp, &src[0]); - o_inst->src[2] = t_src(vp, &src[1]); -#else - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = t_src(vp, &src[1]); - o_inst->src[2] = ZERO_SRC_1; - -#endif - goto next; - - case OPCODE_MAD: - hw_op = (src[0].File == PROGRAM_TEMPORARY && - src[1].File == PROGRAM_TEMPORARY && - src[2].File == - PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : - R300_VPI_OUT_OP_MAD; - - o_inst->op = - MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = t_src(vp, &src[1]); - o_inst->src[2] = t_src(vp, &src[2]); - goto next; - - case OPCODE_MUL: /* HW mul can take third arg but appears to have some other limitations. */ - hw_op = (src[0].File == PROGRAM_TEMPORARY && - src[1].File == - PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : - R300_VPI_OUT_OP_MAD; - - o_inst->op = - MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = t_src(vp, &src[1]); - - o_inst->src[2] = ZERO_SRC_1; - goto next; - - case OPCODE_DP3: //DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO} - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 1)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 2)), - SWIZZLE_ZERO, - t_src_class(src[0].File), - src[0]. - NegateBase ? VSF_FLAG_XYZ : - VSF_FLAG_NONE) | (src[0]. - RelAddr << 4); - - o_inst->src[1] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), - t_swizzle(GET_SWZ - (src[1].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 1)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 2)), - SWIZZLE_ZERO, - t_src_class(src[1].File), - src[1]. - NegateBase ? VSF_FLAG_XYZ : - VSF_FLAG_NONE) | (src[1]. - RelAddr << 4); - - o_inst->src[2] = ZERO_SRC_1; - goto next; - - case OPCODE_SUB: //ADD RESULT 1.X Y Z W TMP 0{} {X Y Z W} PARAM 1{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W -#if 1 - hw_op = (src[0].File == PROGRAM_TEMPORARY && - src[1].File == - PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : - R300_VPI_OUT_OP_MAD; - - o_inst->op = - MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = ONE_SRC_0; - o_inst->src[2] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), - t_swizzle(GET_SWZ - (src[1].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 1)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 2)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 3)), - t_src_class(src[1].File), - (!src[1]. - NegateBase) ? VSF_FLAG_ALL : - VSF_FLAG_NONE) | (src[1]. - RelAddr << 4); -#else - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), - t_swizzle(GET_SWZ - (src[1].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 1)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 2)), - t_swizzle(GET_SWZ - (src[1].Swizzle, 3)), - t_src_class(src[1].File), - (!src[1]. - NegateBase) ? VSF_FLAG_ALL : - VSF_FLAG_NONE) | (src[1]. - RelAddr << 4); - o_inst->src[2] = 0; -#endif - goto next; - - case OPCODE_ABS: //MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 1)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 2)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 3)), - t_src_class(src[0].File), - (!src[0]. - NegateBase) ? VSF_FLAG_ALL : - VSF_FLAG_NONE) | (src[0]. - RelAddr << 4); - o_inst->src[2] = 0; - goto next; - + inst = t_opcode_add(vp, vpi, inst, src); + break; + case OPCODE_ARL: + inst = t_opcode_arl(vp, vpi, inst, src); + break; + case OPCODE_DP3: + inst = t_opcode_dp3(vp, vpi, inst, src); + break; + case OPCODE_DP4: + inst = t_opcode_dp4(vp, vpi, inst, src); + break; + case OPCODE_DPH: + inst = t_opcode_dph(vp, vpi, inst, src); + break; + case OPCODE_DST: + inst = t_opcode_dst(vp, vpi, inst, src); + break; + case OPCODE_EX2: + inst = t_opcode_ex2(vp, vpi, inst, src); + break; + case OPCODE_EXP: + inst = t_opcode_exp(vp, vpi, inst, src); + break; case OPCODE_FLR: - /* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W} - ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */ - - o_inst->op = MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, u_temp_i, - t_dst_mask(vpi->DstReg. - WriteMask), - VSF_OUT_CLASS_TMP); - - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = ZERO_SRC_0; - o_inst++; - - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = MAKE_VSF_SOURCE(u_temp_i, - VSF_IN_COMPONENT_X, - VSF_IN_COMPONENT_Y, - VSF_IN_COMPONENT_Z, - VSF_IN_COMPONENT_W, - VSF_IN_CLASS_TMP, - /* Not 100% sure about this */ - (!src[0]. - NegateBase) ? - VSF_FLAG_ALL : - VSF_FLAG_NONE - /*VSF_FLAG_ALL */ ); - - o_inst->src[2] = ZERO_SRC_0; - u_temp_i--; - goto next; - - case OPCODE_LG2: // LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X} - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_LG2, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_src_class(src[0].File), - src[0]. - NegateBase ? VSF_FLAG_ALL : - VSF_FLAG_NONE) | (src[0]. - RelAddr << 4); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = ZERO_SRC_0; - goto next; - - case OPCODE_LIT: //LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W} - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_LIT, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - /* NOTE: Users swizzling might not work. */ - o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x - t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w - VSF_IN_COMPONENT_ZERO, // z - t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y - t_src_class(src[0]. - File), - src[0]. - NegateBase ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[0].RelAddr << 4); - o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y - t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w - VSF_IN_COMPONENT_ZERO, // z - t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x - t_src_class(src[0]. - File), - src[0]. - NegateBase ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[0].RelAddr << 4); - o_inst->src[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y - t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x - VSF_IN_COMPONENT_ZERO, // z - t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w - t_src_class(src[0]. - File), - src[0]. - NegateBase ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[0].RelAddr << 4); - goto next; - - case OPCODE_DPH: //DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W} - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = - MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), - t_swizzle(GET_SWZ - (src[0].Swizzle, 0)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 1)), - t_swizzle(GET_SWZ - (src[0].Swizzle, 2)), - VSF_IN_COMPONENT_ONE, - t_src_class(src[0].File), - src[0]. - NegateBase ? VSF_FLAG_XYZ : - VSF_FLAG_NONE) | (src[0]. - RelAddr << 4); - o_inst->src[1] = t_src(vp, &src[1]); - o_inst->src[2] = ZERO_SRC_1; - goto next; - - case OPCODE_XPD: - /* mul r0, r1.yzxw, r2.zxyw - mad r0, -r2.yzxw, r1.zxyw, r0 - NOTE: might need MAD_2 - */ - - o_inst->op = MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, u_temp_i, - t_dst_mask(vpi->DstReg. - WriteMask), - VSF_OUT_CLASS_TMP); - - o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y - t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z - t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x - t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w - t_src_class(src[0]. - File), - src[0]. - NegateBase ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[0].RelAddr << 4); - - o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z - t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x - t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y - t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w - t_src_class(src[1]. - File), - src[1]. - NegateBase ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[1].RelAddr << 4); - - o_inst->src[2] = ZERO_SRC_1; - o_inst++; - u_temp_i--; - - o_inst->op = - MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - o_inst->src[0] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y - t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z - t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x - t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w - t_src_class(src[1]. - File), - (!src[1]. - NegateBase) ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[1].RelAddr << 4); - - o_inst->src[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z - t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x - t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y - t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w - t_src_class(src[0]. - File), - src[0]. - NegateBase ? - VSF_FLAG_ALL : - VSF_FLAG_NONE) | - (src[0].RelAddr << 4); - - o_inst->src[2] = MAKE_VSF_SOURCE(u_temp_i + 1, - VSF_IN_COMPONENT_X, - VSF_IN_COMPONENT_Y, - VSF_IN_COMPONENT_Z, - VSF_IN_COMPONENT_W, - VSF_IN_CLASS_TMP, - VSF_FLAG_NONE); - - goto next; - - case OPCODE_RCC: - fprintf(stderr, "Dont know how to handle op %d yet\n", - vpi->Opcode); - _mesa_exit(-1); + inst = + t_opcode_flr(vp, vpi, inst, src, /* FIXME */ + &u_temp_i); + break; + case OPCODE_FRC: + inst = t_opcode_frc(vp, vpi, inst, src); + break; + case OPCODE_LG2: + inst = t_opcode_lg2(vp, vpi, inst, src); + break; + case OPCODE_LIT: + inst = t_opcode_lit(vp, vpi, inst, src); + break; + case OPCODE_LOG: + inst = t_opcode_log(vp, vpi, inst, src); + break; + case OPCODE_MAD: + inst = t_opcode_mad(vp, vpi, inst, src); + break; + case OPCODE_MAX: + inst = t_opcode_max(vp, vpi, inst, src); + break; + case OPCODE_MIN: + inst = t_opcode_min(vp, vpi, inst, src); + break; + case OPCODE_MOV: + inst = t_opcode_mov(vp, vpi, inst, src); break; - case OPCODE_END: + case OPCODE_MUL: + inst = t_opcode_mul(vp, vpi, inst, src); + break; + case OPCODE_POW: + inst = t_opcode_pow(vp, vpi, inst, src); + break; + case OPCODE_RCP: + inst = t_opcode_rcp(vp, vpi, inst, src); + break; + case OPCODE_RSQ: + inst = t_opcode_rsq(vp, vpi, inst, src); + break; + case OPCODE_SGE: + inst = t_opcode_sge(vp, vpi, inst, src); + break; + case OPCODE_SLT: + inst = t_opcode_slt(vp, vpi, inst, src); + break; + case OPCODE_SUB: + inst = t_opcode_sub(vp, vpi, inst, src); + break; + case OPCODE_SWZ: + inst = t_opcode_swz(vp, vpi, inst, src); + break; + case OPCODE_XPD: + inst = + t_opcode_xpd(vp, vpi, inst, src, /* FIXME */ + &u_temp_i); break; default: + assert(0); break; } - - o_inst->op = - MAKE_VSF_OP(t_opcode(vpi->Opcode), - t_dst_index(vp, &vpi->DstReg), - t_dst_mask(vpi->DstReg.WriteMask), - t_dst_class(vpi->DstReg.File)); - - if (are_srcs_scalar) { - switch (operands) { - case 1: - o_inst->src[0] = t_src_scalar(vp, &src[0]); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = ZERO_SRC_0; - break; - - case 2: - o_inst->src[0] = t_src_scalar(vp, &src[0]); - o_inst->src[1] = t_src_scalar(vp, &src[1]); - o_inst->src[2] = ZERO_SRC_1; - break; - - case 3: - o_inst->src[0] = t_src_scalar(vp, &src[0]); - o_inst->src[1] = t_src_scalar(vp, &src[1]); - o_inst->src[2] = t_src_scalar(vp, &src[2]); - break; - - default: - fprintf(stderr, - "scalars and op RCC not handled yet"); - _mesa_exit(-1); - break; - } - } else { - switch (operands) { - case 1: - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = ZERO_SRC_0; - o_inst->src[2] = ZERO_SRC_0; - break; - - case 2: - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = t_src(vp, &src[1]); - o_inst->src[2] = ZERO_SRC_1; - break; - - case 3: - o_inst->src[0] = t_src(vp, &src[0]); - o_inst->src[1] = t_src(vp, &src[1]); - o_inst->src[2] = t_src(vp, &src[2]); - break; - - default: - fprintf(stderr, - "scalars and op RCC not handled yet"); - _mesa_exit(-1); - break; - } - } - next:; } - /* Will most likely segfault before we get here... fix later. */ - if (o_inst - vp->program.body.i >= VSF_MAX_FRAGMENT_LENGTH / 4) { + vp->program.length = (inst - vp->program.body.i); + if (vp->program.length >= VSF_MAX_FRAGMENT_LENGTH) { vp->program.length = 0; vp->native = GL_FALSE; - return; } - vp->program.length = (o_inst - vp->program.body.i) * 4; #if 0 fprintf(stderr, "hw program:\n"); for (i = 0; i < vp->program.length; i++) @@ -1065,7 +1309,8 @@ static void position_invariant(struct gl_program *prog) struct gl_program_parameter_list *paramList; int i; - gl_state_index tokens[STATE_LENGTH] = { STATE_MVP_MATRIX, 0, 0, 0, 0 }; + gl_state_index tokens[STATE_LENGTH] = + { STATE_MVP_MATRIX, 0, 0, 0, 0 }; /* tokens[4] = matrix modifier */ #ifdef PREFER_DP4 @@ -1159,8 +1404,8 @@ static void insert_wpos(struct r300_vertex_program *vp, prog->NumInstructions - 1); /* END */ _mesa_copy_instructions(&vpi[prog->NumInstructions + 1], - &prog->Instructions[prog->NumInstructions - 1], - 1); + &prog->Instructions[prog->NumInstructions - + 1], 1); vpi_insert = &vpi[prog->NumInstructions - 1]; vpi_insert[i].Opcode = OPCODE_MOV; @@ -1206,8 +1451,8 @@ static void pos_as_texcoord(struct r300_vertex_program *vp, prog->NumTemporaries++; for (vpi = prog->Instructions; vpi->Opcode != OPCODE_END; vpi++) { - if (vpi->DstReg.File == PROGRAM_OUTPUT && - vpi->DstReg.Index == VERT_RESULT_HPOS) { + if (vpi->DstReg.File == PROGRAM_OUTPUT + && vpi->DstReg.Index == VERT_RESULT_HPOS) { vpi->DstReg.File = PROGRAM_TEMPORARY; vpi->DstReg.Index = tempregi; } @@ -1223,20 +1468,18 @@ static struct r300_vertex_program *build_program(struct r300_vertex_program_key vp = _mesa_calloc(sizeof(*vp)); _mesa_memcpy(&vp->key, wanted_key, sizeof(vp->key)); - vp->wpos_idx = wpos_idx; if (mesa_vp->IsPositionInvariant) { position_invariant(&mesa_vp->Base); } - if (wpos_idx > -1) + if (wpos_idx > -1) { pos_as_texcoord(vp, &mesa_vp->Base); + } assert(mesa_vp->Base.NumInstructions); - vp->num_temporaries = mesa_vp->Base.NumTemporaries; - r300TranslateVertexShader(vp, mesa_vp->Base.Instructions); return vp; @@ -1252,11 +1495,10 @@ void r300SelectVertexShader(r300ContextPtr r300) struct r300_vertex_program *vp; GLint wpos_idx; - vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current; + vpc = + (struct r300_vertex_program_cont *)ctx->VertexProgram._Current; InputsRead = ctx->FragmentProgram._Current->Base.InputsRead; - wanted_key.OutputsWritten |= 1 << VERT_RESULT_HPOS; - wpos_idx = -1; if (InputsRead & FRAG_BIT_WPOS) { for (i = 0; i < ctx->Const.MaxTextureUnits; i++) @@ -1271,28 +1513,35 @@ void r300SelectVertexShader(r300ContextPtr r300) InputsRead |= (FRAG_BIT_TEX0 << i); wpos_idx = i; } + wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead; + wanted_key.OutputsWritten = vpc->mesa_program.Base.OutputsWritten; + + wanted_key.OutputsWritten |= 1 << VERT_RESULT_HPOS; - if (InputsRead & FRAG_BIT_COL0) + if (InputsRead & FRAG_BIT_COL0) { wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL0; + } - if ((InputsRead & FRAG_BIT_COL1) /*|| - (InputsRead & FRAG_BIT_FOGC) */ ) + if ((InputsRead & FRAG_BIT_COL1)) { wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL1; + } - for (i = 0; i < ctx->Const.MaxTextureUnits; i++) - if (InputsRead & (FRAG_BIT_TEX0 << i)) + for (i = 0; i < ctx->Const.MaxTextureUnits; i++) { + if (InputsRead & (FRAG_BIT_TEX0 << i)) { wanted_key.OutputsWritten |= 1 << (VERT_RESULT_TEX0 + i); + } + } - wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead; if (vpc->mesa_program.IsPositionInvariant) { /* we wan't position don't we ? */ wanted_key.InputsRead |= (1 << VERT_ATTRIB_POS); + wanted_key.OutputsWritten |= (1 << VERT_RESULT_HPOS); } for (vp = vpc->progs; vp; vp = vp->next) - if (_mesa_memcmp(&vp->key, &wanted_key, sizeof(wanted_key)) == - 0) { + if (_mesa_memcmp(&vp->key, &wanted_key, sizeof(wanted_key)) + == 0) { r300->selected_vp = vp; return; } diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.h b/src/mesa/drivers/dri/r300/r300_vertprog.h index 252d5a901f..3df0eee799 100644 --- a/src/mesa/drivers/dri/r300/r300_vertprog.h +++ b/src/mesa/drivers/dri/r300/r300_vertprog.h @@ -3,11 +3,6 @@ #include "r300_reg.h" -typedef struct { - GLuint op; - GLuint src[3]; -} VERTEX_SHADER_INSTRUCTION; - #define VSF_FLAG_X 1 #define VSF_FLAG_Y 2 #define VSF_FLAG_Z 4 diff --git a/src/mesa/drivers/dri/savage/savagetris.c b/src/mesa/drivers/dri/savage/savagetris.c index 3dd821a4d3..4ce2f60b4f 100644 --- a/src/mesa/drivers/dri/savage/savagetris.c +++ b/src/mesa/drivers/dri/savage/savagetris.c @@ -131,7 +131,9 @@ static __inline__ void savage_draw_point (savageContextPtr imesa, u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize); const GLfloat x = tmp->v.x; const GLfloat y = tmp->v.y; - const GLfloat sz = imesa->glCtx->Point._Size * .5; + const GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size, + imesa->glCtx->Const.MinPointSize, + imesa->glCtx->Const.MaxPointSize); GLuint j; *(float *)&vb[0] = x - sz; @@ -164,7 +166,9 @@ static __inline__ void savage_draw_line (savageContextPtr imesa, savageVertexPtr v1 ) { GLuint vertsize = imesa->HwVertexSize; u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize); - GLfloat width = imesa->glCtx->Line._Width; + const GLfloat width = CLAMP(imesa->glCtx->Line.Width, + imesa->glCtx->Const.MinLineWidth, + imesa->glCtx->Const.MaxLineWidth); GLfloat dx, dy, ix, iy; GLuint j; @@ -234,7 +238,9 @@ static __inline__ void savage_ptex_line (savageContextPtr imesa, savageVertexPtr v1 ) { GLuint vertsize = imesa->HwVertexSize; u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize); - GLfloat width = imesa->glCtx->Line._Width; + const GLfloat width = CLAMP(imesa->glCtx->Line.Width, + imesa->glCtx->Const.MinLineWidth, + imesa->glCtx->Const.MaxLineWidth); GLfloat dx, dy, ix, iy; savageVertex tmp0, tmp1; GLuint j; @@ -281,7 +287,9 @@ static __inline__ void savage_ptex_point (savageContextPtr imesa, u_int32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize); const GLfloat x = v0->v.x; const GLfloat y = v0->v.y; - const GLfloat sz = imesa->glCtx->Point._Size * .5; + const GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size, + imesa->glCtx->Const.MinPointSize, + imesa->glCtx->Const.MaxPointSize); savageVertex tmp; GLuint j; diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tris.c b/src/mesa/drivers/dri/tdfx/tdfx_tris.c index 96f9ae27fc..7252a7e7dc 100644 --- a/src/mesa/drivers/dri/tdfx/tdfx_tris.c +++ b/src/mesa/drivers/dri/tdfx/tdfx_tris.c @@ -184,7 +184,7 @@ tdfx_translate_vertex( GLcontext *ctx, const tdfxVertex *src, SWvertex *dst) } } - dst->pointSize = ctx->Point._Size; + dst->pointSize = ctx->Point.Size; } |