X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fdrivers%2Fdri%2Fi965%2Fgen7_sf_state.c;h=86809a1b0a31a8febebe5bdf4e15b05e0f7f813e;hb=8371c68a4b4c12f4dd75f82b8b29a624705910a5;hp=e36c44aa3fdb01ec15ea103d4c96db3aa8bf20a6;hpb=147d0102952b7c0b16906c08da2ae447ec68185a;p=mesa.git diff --git a/src/mesa/drivers/dri/i965/gen7_sf_state.c b/src/mesa/drivers/dri/i965/gen7_sf_state.c index e36c44aa3fd..86809a1b0a3 100644 --- a/src/mesa/drivers/dri/i965/gen7_sf_state.c +++ b/src/mesa/drivers/dri/i965/gen7_sf_state.c @@ -26,6 +26,7 @@ #include "brw_defines.h" #include "brw_util.h" #include "main/macros.h" +#include "main/fbobject.h" #include "intel_batchbuffer.h" static void @@ -33,52 +34,65 @@ upload_sbe_state(struct brw_context *brw) { struct intel_context *intel = &brw->intel; struct gl_context *ctx = &intel->ctx; - /* CACHE_NEW_VS_PROG */ - uint32_t num_inputs = brw_count_bits(brw->vs.prog_data->outputs_written); /* BRW_NEW_FRAGMENT_PROGRAM */ - uint32_t num_outputs = brw_count_bits(brw->fragment_program->Base.InputsRead); + uint32_t num_outputs = _mesa_bitcount_64(brw->fragment_program->Base.InputsRead); + /* _NEW_LIGHT */ + bool shade_model_flat = ctx->Light.ShadeModel == GL_FLAT; uint32_t dw1, dw10, dw11; int i; int attr = 0, input_index = 0; - /* _NEW_TRANSFORM */ - int urb_start = ctx->Transform.ClipPlanesEnabled ? 2 : 1; - /* _NEW_LIGHT */ - int two_side_color = (ctx->Light.Enabled && ctx->Light.Model.TwoSide); - uint16_t attr_overrides[FRAG_ATTRIB_MAX]; + int urb_entry_read_offset = 1; + uint16_t attr_overrides[VARYING_SLOT_MAX]; + /* _NEW_BUFFERS */ + bool render_to_fbo = _mesa_is_user_fbo(ctx->DrawBuffer); + uint32_t point_sprite_origin; /* FINISHME: Attribute Swizzle Control Mode? */ - dw1 = - GEN7_SBE_SWIZZLE_ENABLE | - num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT | - (num_inputs + 1) / 2 << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT | - urb_start << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT; + dw1 = GEN7_SBE_SWIZZLE_ENABLE | num_outputs << GEN7_SBE_NUM_OUTPUTS_SHIFT; + + /* _NEW_POINT + * + * Window coordinates in an FBO are inverted, which means point + * sprite origin must be inverted. + */ + if ((ctx->Point.SpriteOrigin == GL_LOWER_LEFT) != render_to_fbo) { + point_sprite_origin = GEN6_SF_POINT_SPRITE_LOWERLEFT; + } else { + point_sprite_origin = GEN6_SF_POINT_SPRITE_UPPERLEFT; + } + dw1 |= point_sprite_origin; - /* _NEW_POINT */ - if (ctx->Point.SpriteOrigin == GL_LOWER_LEFT) - dw1 |= GEN6_SF_POINT_SPRITE_LOWERLEFT; dw10 = 0; - - /* _NEW_LIGHT (flat shading) */ dw11 = 0; - if (ctx->Light.ShadeModel == GL_FLAT) { - dw11 |= ((brw->fragment_program->Base.InputsRead & (FRAG_BIT_COL0 | FRAG_BIT_COL1)) >> - ((brw->fragment_program->Base.InputsRead & FRAG_BIT_WPOS) ? 0 : 1)); - } /* Create the mapping from the FS inputs we produce to the VS outputs * they source from. */ - for (; attr < FRAG_ATTRIB_MAX; attr++) { + uint32_t max_source_attr = 0; + for (; attr < VARYING_SLOT_MAX; attr++) { + enum glsl_interp_qualifier interp_qualifier = + brw->fragment_program->InterpQualifier[attr]; + bool is_gl_Color = attr == VARYING_SLOT_COL0 || attr == VARYING_SLOT_COL1; + if (!(brw->fragment_program->Base.InputsRead & BITFIELD64_BIT(attr))) continue; if (ctx->Point.PointSprite && - attr >= FRAG_ATTRIB_TEX0 && attr <= FRAG_ATTRIB_TEX7 && - ctx->Point.CoordReplace[attr - FRAG_ATTRIB_TEX0]) { + attr >= VARYING_SLOT_TEX0 && attr <= VARYING_SLOT_TEX7 && + ctx->Point.CoordReplace[attr - VARYING_SLOT_TEX0]) { dw10 |= (1 << input_index); } + if (attr == VARYING_SLOT_PNTC) + dw10 |= (1 << input_index); + + /* flat shading */ + if (interp_qualifier == INTERP_QUALIFIER_FLAT || + (shade_model_flat && is_gl_Color && + interp_qualifier == INTERP_QUALIFIER_NONE)) + dw11 |= (1 << input_index); + /* The hardware can only do the overrides on 16 overrides at a * time, and the other up to 16 have to be lined up so that the * input index = the output index. We'll need to do some @@ -86,12 +100,31 @@ upload_sbe_state(struct brw_context *brw) */ assert(input_index < 16 || attr == input_index); - attr_overrides[input_index++] = get_attr_override(brw, attr, - two_side_color); + /* CACHE_NEW_VS_PROG | _NEW_LIGHT | _NEW_PROGRAM */ + attr_overrides[input_index++] = + get_attr_override(&brw->vs.prog_data->vue_map, + urb_entry_read_offset, attr, + ctx->VertexProgram._TwoSideEnabled, + &max_source_attr); } - for (; attr < FRAG_ATTRIB_MAX; attr++) - attr_overrides[input_index++] = 0; + /* From the Ivy Bridge PRM, Volume 2, Part 1, documentation for + * 3DSTATE_SBE DWord 1 bits 15:11, "Vertex URB Entry Read Length": + * + * "This field should be set to the minimum length required to read the + * maximum source attribute. The maximum source attribute is indicated + * by the maximum value of the enabled Attribute # Source Attribute if + * Attribute Swizzle Enable is set, Number of Output Attributes-1 if + * enable is not set. + * + * read_length = ceiling((max_source_attr + 1) / 2)" + */ + uint32_t urb_entry_read_length = ALIGN(max_source_attr + 1, 2) / 2; + dw1 |= urb_entry_read_length << GEN7_SBE_URB_ENTRY_READ_LENGTH_SHIFT | + urb_entry_read_offset << GEN7_SBE_URB_ENTRY_READ_OFFSET_SHIFT; + + for (; input_index < VARYING_SLOT_MAX; input_index++) + attr_overrides[input_index] = 0; BEGIN_BATCH(14); OUT_BATCH(_3DSTATE_SBE << 16 | (14 - 2)); @@ -111,9 +144,10 @@ upload_sbe_state(struct brw_context *brw) const struct brw_tracked_state gen7_sbe_state = { .dirty = { - .mesa = (_NEW_LIGHT | + .mesa = (_NEW_BUFFERS | + _NEW_LIGHT | _NEW_POINT | - _NEW_TRANSFORM), + _NEW_PROGRAM), .brw = (BRW_NEW_CONTEXT | BRW_NEW_FRAGMENT_PROGRAM), .cache = CACHE_NEW_VS_PROG @@ -129,12 +163,14 @@ upload_sf_state(struct brw_context *brw) uint32_t dw1, dw2, dw3; float point_size; /* _NEW_BUFFERS */ - bool render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0; + bool render_to_fbo = _mesa_is_user_fbo(brw->intel.ctx.DrawBuffer); + bool multisampled_fbo = ctx->DrawBuffer->Visual.samples > 1; - dw1 = GEN6_SF_STATISTICS_ENABLE | GEN6_SF_VIEWPORT_TRANSFORM_ENABLE; + dw1 = GEN6_SF_STATISTICS_ENABLE | + GEN6_SF_VIEWPORT_TRANSFORM_ENABLE; /* _NEW_BUFFERS */ - dw1 |= (gen7_depth_format(brw) << GEN7_SF_DEPTH_BUFFER_SURFACE_FORMAT_SHIFT); + dw1 |= (brw_depthbuffer_format(brw) << GEN7_SF_DEPTH_BUFFER_SURFACE_FORMAT_SHIFT); /* _NEW_POLYGON */ if ((ctx->Polygon.FrontFace == GL_CCW) ^ render_to_fbo) @@ -211,21 +247,30 @@ upload_sf_state(struct brw_context *brw) dw2 |= GEN6_SF_SCISSOR_ENABLE; /* _NEW_LINE */ - dw2 |= U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7) << - GEN6_SF_LINE_WIDTH_SHIFT; + { + uint32_t line_width_u3_7 = U_FIXED(CLAMP(ctx->Line.Width, 0.0, 7.99), 7); + /* TODO: line width of 0 is not allowed when MSAA enabled */ + if (line_width_u3_7 == 0) + line_width_u3_7 = 1; + dw2 |= line_width_u3_7 << GEN6_SF_LINE_WIDTH_SHIFT; + } if (ctx->Line.SmoothFlag) { dw2 |= GEN6_SF_LINE_AA_ENABLE; - dw2 |= GEN6_SF_LINE_AA_MODE_TRUE; dw2 |= GEN6_SF_LINE_END_CAP_WIDTH_1_0; } + if (ctx->Line.StippleFlag && intel->is_haswell) { + dw2 |= HSW_SF_LINE_STIPPLE_ENABLE; + } + /* _NEW_MULTISAMPLE */ + if (multisampled_fbo && ctx->Multisample.Enabled) + dw2 |= GEN6_SF_MSRAST_ON_PATTERN; /* FINISHME: Last Pixel Enable? Vertex Sub Pixel Precision Select? - * FINISHME: AA Line Distance Mode? */ - dw3 = 0; + dw3 = GEN6_SF_LINE_AA_MODE_TRUE; - /* _NEW_POINT */ + /* _NEW_PROGRAM | _NEW_POINT */ if (!(ctx->VertexProgram.PointSizeEnabled || ctx->Point._Attenuated)) dw3 |= GEN6_SF_USE_STATE_POINT_WIDTH; @@ -259,12 +304,14 @@ upload_sf_state(struct brw_context *brw) const struct brw_tracked_state gen7_sf_state = { .dirty = { .mesa = (_NEW_LIGHT | + _NEW_PROGRAM | _NEW_POLYGON | _NEW_LINE | _NEW_SCISSOR | _NEW_BUFFERS | - _NEW_POINT), - .brw = (BRW_NEW_CONTEXT), + _NEW_POINT | + _NEW_MULTISAMPLE), + .brw = BRW_NEW_CONTEXT, .cache = CACHE_NEW_VS_PROG }, .emit = upload_sf_state,