/**************************************************************************
- *
- * Copyright 2007 VMware, Inc.
+ *
+ * Copyright 2007-2018 VMware, Inc.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* distribute, sub license, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
- *
+ *
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial portions
* of the Software.
- *
+ *
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
* ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
+ *
**************************************************************************/
/**
- * AA line stage: AA lines are converted to texture mapped triangles.
+ * AA line stage: AA lines are converted triangles (with extra generic)
*
* Authors: Brian Paul
*/
#include "pipe/p_shader_tokens.h"
#include "util/u_inlines.h"
-#include "util/u_format.h"
+#include "util/format/u_format.h"
#include "util/u_math.h"
#include "util/u_memory.h"
-#include "util/u_sampler.h"
#include "tgsi/tgsi_transform.h"
#include "tgsi/tgsi_dump.h"
#include "draw_private.h"
#include "draw_pipe.h"
+#include "nir.h"
+#include "nir/nir_draw_helpers.h"
/** Approx number of new tokens for instructions in aa_transform_inst() */
-#define NUM_NEW_TOKENS 50
-
-
-/**
- * Size for the alpha texture used for antialiasing
- */
-#define TEXTURE_SIZE_LOG2 5 /* 32 x 32 */
-
-/**
- * Max texture level for the alpha texture used for antialiasing
- *
- * Don't use the 1x1 and 2x2 mipmap levels.
- */
-#define MAX_TEXTURE_LEVEL (TEXTURE_SIZE_LOG2 - 2)
+#define NUM_NEW_TOKENS 53
/**
struct pipe_shader_state state;
void *driver_fs;
void *aaline_fs;
- uint sampler_unit;
- int generic_attrib; /**< texcoord/generic used for texture */
+ int generic_attrib; /**< generic used for distance */
};
float half_line_width;
- /** For AA lines, this is the vertex attrib slot for the new texcoords */
- uint tex_slot;
+ /** For AA lines, this is the vertex attrib slot for new generic */
+ uint coord_slot;
/** position, not necessarily output zero */
uint pos_slot;
- void *sampler_cso;
- struct pipe_resource *texture;
- struct pipe_sampler_view *sampler_view;
- uint num_samplers;
- uint num_sampler_views;
-
/*
* Currently bound state
*/
struct aaline_fragment_shader *fs;
- struct {
- void *sampler[PIPE_MAX_SAMPLERS];
- struct pipe_sampler_view *sampler_views[PIPE_MAX_SHADER_SAMPLER_VIEWS];
- } state;
/*
* Driver interface/override functions
const struct pipe_shader_state *);
void (*driver_bind_fs_state)(struct pipe_context *, void *);
void (*driver_delete_fs_state)(struct pipe_context *, void *);
-
- void (*driver_bind_sampler_states)(struct pipe_context *, unsigned, unsigned,
- unsigned, void **);
-
- void (*driver_set_sampler_views)(struct pipe_context *, unsigned shader,
- unsigned start, unsigned count,
- struct pipe_sampler_view **);
};
*/
struct aa_transform_context {
struct tgsi_transform_context base;
- uint tempsUsed; /**< bitmask */
+ uint64_t tempsUsed; /**< bitmask */
int colorOutput; /**< which output is the primary color */
- uint samplersUsed; /**< bitfield of samplers used */
- int freeSampler; /** an available sampler for the pstipple */
int maxInput, maxGeneric; /**< max input index found */
- int colorTemp, texTemp; /**< temp registers */
- boolean firstInstruction;
+ int colorTemp, aaTemp; /**< temp registers */
};
-
/**
* TGSI declaration transform callback.
- * Look for a free sampler, a free input attrib, and two free temp regs.
+ * Look for a free input attrib, and two free temp regs.
*/
static void
aa_transform_decl(struct tgsi_transform_context *ctx,
struct tgsi_full_declaration *decl)
{
- struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
+ struct aa_transform_context *aactx = (struct aa_transform_context *)ctx;
if (decl->Declaration.File == TGSI_FILE_OUTPUT &&
decl->Semantic.Name == TGSI_SEMANTIC_COLOR &&
decl->Semantic.Index == 0) {
aactx->colorOutput = decl->Range.First;
}
- else if (decl->Declaration.File == TGSI_FILE_SAMPLER) {
- uint i;
- for (i = decl->Range.First;
- i <= decl->Range.Last; i++) {
- aactx->samplersUsed |= 1 << i;
- }
- }
else if (decl->Declaration.File == TGSI_FILE_INPUT) {
if ((int) decl->Range.Last > aactx->maxInput)
aactx->maxInput = decl->Range.Last;
uint i;
for (i = decl->Range.First;
i <= decl->Range.Last; i++) {
- aactx->tempsUsed |= (1 << i);
+ /*
+ * XXX this bitfield doesn't really cut it...
+ */
+ aactx->tempsUsed |= UINT64_C(1) << i;
}
}
/**
- * Find the lowest zero bit in the given word, or -1 if bitfield is all ones.
+ * Find the lowest zero bit, or -1 if bitfield is all ones.
*/
static int
-free_bit(uint bitfield)
+free_bit(uint64_t bitfield)
{
- return ffs(~bitfield) - 1;
+ return ffsll(~bitfield) - 1;
}
/**
- * TGSI instruction transform callback.
- * Replace writes to result.color w/ a temp reg.
- * Upon END instruction, insert texture sampling code for antialiasing.
+ * TGSI transform prolog callback.
*/
static void
-aa_transform_inst(struct tgsi_transform_context *ctx,
- struct tgsi_full_instruction *inst)
+aa_transform_prolog(struct tgsi_transform_context *ctx)
{
struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
+ uint64_t usedTemps = aactx->tempsUsed;
+
+ /* find two free temp regs */
+ aactx->colorTemp = free_bit(usedTemps);
+ usedTemps |= UINT64_C(1) << aactx->colorTemp;
+ aactx->aaTemp = free_bit(usedTemps);
+ assert(aactx->colorTemp >= 0);
+ assert(aactx->aaTemp >= 0);
+
+ /* declare new generic input/texcoord */
+ tgsi_transform_input_decl(ctx, aactx->maxInput + 1,
+ TGSI_SEMANTIC_GENERIC, aactx->maxGeneric + 1,
+ TGSI_INTERPOLATE_LINEAR);
+
+ /* declare new temp regs */
+ tgsi_transform_temp_decl(ctx, aactx->aaTemp);
+ tgsi_transform_temp_decl(ctx, aactx->colorTemp);
+}
- if (aactx->firstInstruction) {
- /* emit our new declarations before the first instruction */
- struct tgsi_full_declaration decl;
- uint i;
-
- /* find free sampler */
- aactx->freeSampler = free_bit(aactx->samplersUsed);
- if (aactx->freeSampler >= PIPE_MAX_SAMPLERS)
- aactx->freeSampler = PIPE_MAX_SAMPLERS - 1;
-
- /* find two free temp regs */
- for (i = 0; i < 32; i++) {
- if ((aactx->tempsUsed & (1 << i)) == 0) {
- /* found a free temp */
- if (aactx->colorTemp < 0)
- aactx->colorTemp = i;
- else if (aactx->texTemp < 0)
- aactx->texTemp = i;
- else
- break;
- }
- }
- assert(aactx->colorTemp >= 0);
- assert(aactx->texTemp >= 0);
-
- /* declare new generic input/texcoord */
- decl = tgsi_default_full_declaration();
- decl.Declaration.File = TGSI_FILE_INPUT;
- /* XXX this could be linear... */
- decl.Declaration.Interpolate = 1;
- decl.Declaration.Semantic = 1;
- decl.Semantic.Name = TGSI_SEMANTIC_GENERIC;
- decl.Semantic.Index = aactx->maxGeneric + 1;
- decl.Range.First =
- decl.Range.Last = aactx->maxInput + 1;
- decl.Interp.Interpolate = TGSI_INTERPOLATE_PERSPECTIVE;
- ctx->emit_declaration(ctx, &decl);
-
- /* declare new sampler */
- decl = tgsi_default_full_declaration();
- decl.Declaration.File = TGSI_FILE_SAMPLER;
- decl.Range.First =
- decl.Range.Last = aactx->freeSampler;
- ctx->emit_declaration(ctx, &decl);
-
- /* declare new temp regs */
- decl = tgsi_default_full_declaration();
- decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.Range.First =
- decl.Range.Last = aactx->texTemp;
- ctx->emit_declaration(ctx, &decl);
-
- decl = tgsi_default_full_declaration();
- decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.Range.First =
- decl.Range.Last = aactx->colorTemp;
- ctx->emit_declaration(ctx, &decl);
-
- aactx->firstInstruction = FALSE;
- }
+/**
+ * TGSI transform epilog callback.
+ */
+static void
+aa_transform_epilog(struct tgsi_transform_context *ctx)
+{
+ struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
- if (inst->Instruction.Opcode == TGSI_OPCODE_END &&
- aactx->colorOutput != -1) {
- struct tgsi_full_instruction newInst;
-
- /* TEX */
- newInst = tgsi_default_full_instruction();
- newInst.Instruction.Opcode = TGSI_OPCODE_TEX;
- newInst.Instruction.NumDstRegs = 1;
- newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
- newInst.Dst[0].Register.Index = aactx->texTemp;
- newInst.Instruction.NumSrcRegs = 2;
- newInst.Instruction.Texture = TRUE;
- newInst.Texture.Texture = TGSI_TEXTURE_2D;
- newInst.Src[0].Register.File = TGSI_FILE_INPUT;
- newInst.Src[0].Register.Index = aactx->maxInput + 1;
- newInst.Src[1].Register.File = TGSI_FILE_SAMPLER;
- newInst.Src[1].Register.Index = aactx->freeSampler;
-
- ctx->emit_instruction(ctx, &newInst);
+ if (aactx->colorOutput != -1) {
+ struct tgsi_full_instruction inst;
+ /* insert distance-based coverage code for antialiasing. */
+
+ /* saturate(linewidth - fabs(interpx), linelength - fabs(interpz) */
+ inst = tgsi_default_full_instruction();
+ inst.Instruction.Saturate = true;
+ inst.Instruction.Opcode = TGSI_OPCODE_ADD;
+ inst.Instruction.NumDstRegs = 1;
+ tgsi_transform_dst_reg(&inst.Dst[0], TGSI_FILE_TEMPORARY,
+ aactx->aaTemp, TGSI_WRITEMASK_XZ);
+ inst.Instruction.NumSrcRegs = 2;
+ tgsi_transform_src_reg(&inst.Src[1], TGSI_FILE_INPUT, aactx->maxInput + 1,
+ TGSI_SWIZZLE_X, TGSI_SWIZZLE_X,
+ TGSI_SWIZZLE_Z, TGSI_SWIZZLE_Z);
+ tgsi_transform_src_reg(&inst.Src[0], TGSI_FILE_INPUT, aactx->maxInput + 1,
+ TGSI_SWIZZLE_Y, TGSI_SWIZZLE_Y,
+ TGSI_SWIZZLE_W, TGSI_SWIZZLE_W);
+ inst.Src[1].Register.Absolute = true;
+ inst.Src[1].Register.Negate = true;
+ ctx->emit_instruction(ctx, &inst);
+
+ /* MUL width / height alpha */
+ tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_MUL,
+ TGSI_FILE_TEMPORARY, aactx->aaTemp,
+ TGSI_WRITEMASK_W,
+ TGSI_FILE_TEMPORARY, aactx->aaTemp,
+ TGSI_SWIZZLE_X,
+ TGSI_FILE_TEMPORARY, aactx->aaTemp,
+ TGSI_SWIZZLE_Z, false);
/* MOV rgb */
- newInst = tgsi_default_full_instruction();
- newInst.Instruction.Opcode = TGSI_OPCODE_MOV;
- newInst.Instruction.NumDstRegs = 1;
- newInst.Dst[0].Register.File = TGSI_FILE_OUTPUT;
- newInst.Dst[0].Register.Index = aactx->colorOutput;
- newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZ;
- newInst.Instruction.NumSrcRegs = 1;
- newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
- newInst.Src[0].Register.Index = aactx->colorTemp;
- ctx->emit_instruction(ctx, &newInst);
+ tgsi_transform_op1_inst(ctx, TGSI_OPCODE_MOV,
+ TGSI_FILE_OUTPUT, aactx->colorOutput,
+ TGSI_WRITEMASK_XYZ,
+ TGSI_FILE_TEMPORARY, aactx->colorTemp);
/* MUL alpha */
- newInst = tgsi_default_full_instruction();
- newInst.Instruction.Opcode = TGSI_OPCODE_MUL;
- newInst.Instruction.NumDstRegs = 1;
- newInst.Dst[0].Register.File = TGSI_FILE_OUTPUT;
- newInst.Dst[0].Register.Index = aactx->colorOutput;
- newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_W;
- newInst.Instruction.NumSrcRegs = 2;
- newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
- newInst.Src[0].Register.Index = aactx->colorTemp;
- newInst.Src[1].Register.File = TGSI_FILE_TEMPORARY;
- newInst.Src[1].Register.Index = aactx->texTemp;
- ctx->emit_instruction(ctx, &newInst);
-
- /* END */
- newInst = tgsi_default_full_instruction();
- newInst.Instruction.Opcode = TGSI_OPCODE_END;
- newInst.Instruction.NumDstRegs = 0;
- newInst.Instruction.NumSrcRegs = 0;
- ctx->emit_instruction(ctx, &newInst);
+ tgsi_transform_op2_inst(ctx, TGSI_OPCODE_MUL,
+ TGSI_FILE_OUTPUT, aactx->colorOutput,
+ TGSI_WRITEMASK_W,
+ TGSI_FILE_TEMPORARY, aactx->colorTemp,
+ TGSI_FILE_TEMPORARY, aactx->aaTemp, false);
}
- else {
- /* Not an END instruction.
- * Look for writes to result.color and replace with colorTemp reg.
- */
- uint i;
+}
- for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
- struct tgsi_full_dst_register *dst = &inst->Dst[i];
- if (dst->Register.File == TGSI_FILE_OUTPUT &&
- dst->Register.Index == aactx->colorOutput) {
- dst->Register.File = TGSI_FILE_TEMPORARY;
- dst->Register.Index = aactx->colorTemp;
- }
- }
- ctx->emit_instruction(ctx, inst);
+/**
+ * TGSI instruction transform callback.
+ * Replace writes to result.color w/ a temp reg.
+ */
+static void
+aa_transform_inst(struct tgsi_transform_context *ctx,
+ struct tgsi_full_instruction *inst)
+{
+ struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
+ uint i;
+
+ /*
+ * Look for writes to result.color and replace with colorTemp reg.
+ */
+ for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
+ struct tgsi_full_dst_register *dst = &inst->Dst[i];
+ if (dst->Register.File == TGSI_FILE_OUTPUT &&
+ dst->Register.Index == aactx->colorOutput) {
+ dst->Register.File = TGSI_FILE_TEMPORARY;
+ dst->Register.Index = aactx->colorTemp;
+ }
}
+
+ ctx->emit_instruction(ctx, inst);
}
/**
* Generate the frag shader we'll use for drawing AA lines.
- * This will be the user's shader plus some texture/modulate instructions.
+ * This will be the user's shader plus some arithmetic instructions.
*/
static boolean
generate_aaline_fs(struct aaline_stage *aaline)
transform.maxInput = -1;
transform.maxGeneric = -1;
transform.colorTemp = -1;
- transform.texTemp = -1;
- transform.firstInstruction = TRUE;
+ transform.aaTemp = -1;
+ transform.base.prolog = aa_transform_prolog;
+ transform.base.epilog = aa_transform_epilog;
transform.base.transform_instruction = aa_transform_inst;
transform.base.transform_declaration = aa_transform_decl;
tgsi_dump(aaline_fs.tokens, 0);
#endif
- aaline->fs->sampler_unit = transform.freeSampler;
-
aaline->fs->aaline_fs = aaline->driver_create_fs_state(pipe, &aaline_fs);
if (aaline->fs->aaline_fs == NULL)
goto fail;
return FALSE;
}
-
-/**
- * Create the texture map we'll use for antialiasing the lines.
- */
static boolean
-aaline_create_texture(struct aaline_stage *aaline)
+generate_aaline_fs_nir(struct aaline_stage *aaline)
{
+#ifdef LLVM_AVAILABLE
struct pipe_context *pipe = aaline->stage.draw->pipe;
- struct pipe_screen *screen = pipe->screen;
- struct pipe_resource texTemp;
- struct pipe_sampler_view viewTempl;
- uint level;
-
- memset(&texTemp, 0, sizeof(texTemp));
- texTemp.target = PIPE_TEXTURE_2D;
- texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */
- texTemp.last_level = MAX_TEXTURE_LEVEL;
- texTemp.width0 = 1 << TEXTURE_SIZE_LOG2;
- texTemp.height0 = 1 << TEXTURE_SIZE_LOG2;
- texTemp.depth0 = 1;
- texTemp.array_size = 1;
- texTemp.bind = PIPE_BIND_SAMPLER_VIEW;
-
- aaline->texture = screen->resource_create(screen, &texTemp);
- if (!aaline->texture)
- return FALSE;
+ const struct pipe_shader_state *orig_fs = &aaline->fs->state;
+ struct pipe_shader_state aaline_fs;
- u_sampler_view_default_template(&viewTempl,
- aaline->texture,
- aaline->texture->format);
- aaline->sampler_view = pipe->create_sampler_view(pipe,
- aaline->texture,
- &viewTempl);
- if (!aaline->sampler_view) {
+ aaline_fs = *orig_fs; /* copy to init */
+ aaline_fs.ir.nir = nir_shader_clone(NULL, orig_fs->ir.nir);
+ if (!aaline_fs.ir.nir)
return FALSE;
- }
- /* Fill in mipmap images.
- * Basically each level is solid opaque, except for the outermost
- * texels which are zero. Special case the 1x1 and 2x2 levels
- * (though, those levels shouldn't be used - see the max_lod setting).
- */
- for (level = 0; level <= MAX_TEXTURE_LEVEL; level++) {
- struct pipe_transfer *transfer;
- struct pipe_box box;
- const uint size = u_minify(aaline->texture->width0, level);
- ubyte *data;
- uint i, j;
-
- assert(aaline->texture->width0 == aaline->texture->height0);
-
- u_box_origin_2d( size, size, &box );
-
- /* This texture is new, no need to flush.
- */
- data = pipe->transfer_map(pipe,
- aaline->texture,
- level,
- PIPE_TRANSFER_WRITE,
- &box, &transfer);
-
- if (data == NULL)
- return FALSE;
-
- for (i = 0; i < size; i++) {
- for (j = 0; j < size; j++) {
- ubyte d;
- if (size == 1) {
- d = 255;
- }
- else if (size == 2) {
- d = 200; /* tuneable */
- }
- else if (i == 0 || j == 0 || i == size - 1 || j == size - 1) {
- d = 35; /* edge texel */
- }
- else {
- d = 255;
- }
- data[i * transfer->stride + j] = d;
- }
- }
+ nir_lower_aaline_fs(aaline_fs.ir.nir, &aaline->fs->generic_attrib);
+ aaline->fs->aaline_fs = aaline->driver_create_fs_state(pipe, &aaline_fs);
+ if (aaline->fs->aaline_fs == NULL)
+ goto fail;
- /* unmap */
- pipe->transfer_unmap(pipe, transfer);
- }
return TRUE;
-}
-
-/**
- * Create the sampler CSO that'll be used for antialiasing.
- * By using a mipmapped texture, we don't have to generate a different
- * texture image for each line size.
- */
-static boolean
-aaline_create_sampler(struct aaline_stage *aaline)
-{
- struct pipe_sampler_state sampler;
- struct pipe_context *pipe = aaline->stage.draw->pipe;
-
- memset(&sampler, 0, sizeof(sampler));
- sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
- sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
- sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
- sampler.min_mip_filter = PIPE_TEX_MIPFILTER_LINEAR;
- sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR;
- sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR;
- sampler.normalized_coords = 1;
- sampler.min_lod = 0.0f;
- sampler.max_lod = MAX_TEXTURE_LEVEL;
-
- aaline->sampler_cso = pipe->create_sampler_state(pipe, &sampler);
- if (aaline->sampler_cso == NULL)
- return FALSE;
-
- return TRUE;
+fail:
+#endif
+ return FALSE;
}
-
/**
* When we're about to draw our first AA line in a batch, this function is
* called to tell the driver to bind our modified fragment shader.
struct draw_context *draw = aaline->stage.draw;
struct pipe_context *pipe = draw->pipe;
- if (!aaline->fs->aaline_fs &&
- !generate_aaline_fs(aaline))
- return FALSE;
+ if (!aaline->fs->aaline_fs) {
+ if (aaline->fs->state.type == PIPE_SHADER_IR_NIR) {
+ if (!generate_aaline_fs_nir(aaline))
+ return FALSE;
+ } else
+ if (!generate_aaline_fs(aaline))
+ return FALSE;
+ }
draw->suspend_flushing = TRUE;
aaline->driver_bind_fs_state(pipe, aaline->fs->aaline_fs);
-static INLINE struct aaline_stage *
-aaline_stage( struct draw_stage *stage )
+static inline struct aaline_stage *
+aaline_stage(struct draw_stage *stage)
{
return (struct aaline_stage *) stage;
}
const float half_width = aaline->half_line_width;
struct prim_header tri;
struct vertex_header *v[8];
- uint texPos = aaline->tex_slot;
+ uint coordPos = aaline->coord_slot;
uint posPos = aaline->pos_slot;
float *pos, *tex;
float dx = header->v[1]->data[posPos][0] - header->v[0]->data[posPos][0];
float dy = header->v[1]->data[posPos][1] - header->v[0]->data[posPos][1];
- double a = atan2(dy, dx);
- float c_a = (float) cos(a), s_a = (float) sin(a);
+ float a = atan2f(dy, dx);
+ float c_a = cosf(a), s_a = sinf(a);
+ float half_length;
+ float t_l, t_w;
uint i;
- /* XXX the ends of lines aren't quite perfect yet, but probably passable */
- dx = 0.5F * half_width;
- dy = half_width;
+ half_length = 0.5f * sqrtf(dx * dx + dy * dy);
+
+ if (half_length < 0.5f) {
+ /*
+ * The logic we use for "normal" sized segments is incorrect
+ * for very short segments (basically because we only have
+ * one value to interpolate, not a distance to each endpoint).
+ * Therefore, we calculate half_length differently, so that for
+ * original line length (near) 0, we get alpha 0 - otherwise
+ * max alpha would still be 0.5. This also prevents us from
+ * artifacts due to degenerated lines (the endpoints being
+ * identical, which would still receive anywhere from alpha
+ * 0-0.5 otherwise) (at least the pstipple stage may generate
+ * such lines due to float inaccuracies if line length is very
+ * close to a integer).
+ * Might not be fully accurate neither (because the "strength" of
+ * the line is going to be determined by how close to the pixel
+ * center those 1 or 2 fragments are) but it's probably the best
+ * we can do.
+ */
+ half_length = 2.0f * half_length;
+ } else {
+ half_length = half_length + 0.5f;
+ }
+
+ t_w = half_width;
+ t_l = 0.5f;
/* allocate/dup new verts */
- for (i = 0; i < 8; i++) {
- v[i] = dup_vert(stage, header->v[i/4], i);
+ for (i = 0; i < 4; i++) {
+ v[i] = dup_vert(stage, header->v[i/2], i);
}
/*
* Quad strip for line from v0 to v1 (*=endpoints):
*
- * 1 3 5 7
- * +---+---------------------+---+
+ * 1 3
+ * +-----------------------------+
* | |
* | *v0 v1* |
* | |
- * +---+---------------------+---+
- * 0 2 4 6
+ * +-----------------------------+
+ * 0 2
+ */
+
+ /*
+ * We increase line length by 0.5 pixels (at each endpoint),
+ * and calculate the tri endpoints by moving them half-width
+ * distance away perpendicular to the line.
+ * XXX: since we change line endpoints (by 0.5 pixel), should
+ * actually re-interpolate all other values?
*/
/* new verts */
pos = v[0]->data[posPos];
- pos[0] += (-dx * c_a - dy * s_a);
- pos[1] += (-dx * s_a + dy * c_a);
+ pos[0] += (-t_l * c_a - t_w * s_a);
+ pos[1] += (-t_l * s_a + t_w * c_a);
pos = v[1]->data[posPos];
- pos[0] += (-dx * c_a - -dy * s_a);
- pos[1] += (-dx * s_a + -dy * c_a);
+ pos[0] += (-t_l * c_a - -t_w * s_a);
+ pos[1] += (-t_l * s_a + -t_w * c_a);
pos = v[2]->data[posPos];
- pos[0] += ( dx * c_a - dy * s_a);
- pos[1] += ( dx * s_a + dy * c_a);
+ pos[0] += (t_l * c_a - t_w * s_a);
+ pos[1] += (t_l * s_a + t_w * c_a);
pos = v[3]->data[posPos];
- pos[0] += ( dx * c_a - -dy * s_a);
- pos[1] += ( dx * s_a + -dy * c_a);
-
- pos = v[4]->data[posPos];
- pos[0] += (-dx * c_a - dy * s_a);
- pos[1] += (-dx * s_a + dy * c_a);
-
- pos = v[5]->data[posPos];
- pos[0] += (-dx * c_a - -dy * s_a);
- pos[1] += (-dx * s_a + -dy * c_a);
-
- pos = v[6]->data[posPos];
- pos[0] += ( dx * c_a - dy * s_a);
- pos[1] += ( dx * s_a + dy * c_a);
-
- pos = v[7]->data[posPos];
- pos[0] += ( dx * c_a - -dy * s_a);
- pos[1] += ( dx * s_a + -dy * c_a);
+ pos[0] += (t_l * c_a - -t_w * s_a);
+ pos[1] += (t_l * s_a + -t_w * c_a);
/* new texcoords */
- tex = v[0]->data[texPos];
- ASSIGN_4V(tex, 0, 0, 0, 1);
-
- tex = v[1]->data[texPos];
- ASSIGN_4V(tex, 0, 1, 0, 1);
+ tex = v[0]->data[coordPos];
+ ASSIGN_4V(tex, -half_width, half_width, -half_length, half_length);
- tex = v[2]->data[texPos];
- ASSIGN_4V(tex, .5, 0, 0, 1);
+ tex = v[1]->data[coordPos];
+ ASSIGN_4V(tex, half_width, half_width, -half_length, half_length);
- tex = v[3]->data[texPos];
- ASSIGN_4V(tex, .5, 1, 0, 1);
+ tex = v[2]->data[coordPos];
+ ASSIGN_4V(tex, -half_width, half_width, half_length, half_length);
- tex = v[4]->data[texPos];
- ASSIGN_4V(tex, .5, 0, 0, 1);
+ tex = v[3]->data[coordPos];
+ ASSIGN_4V(tex, half_width, half_width, half_length, half_length);
- tex = v[5]->data[texPos];
- ASSIGN_4V(tex, .5, 1, 0, 1);
-
- tex = v[6]->data[texPos];
- ASSIGN_4V(tex, 1, 0, 0, 1);
-
- tex = v[7]->data[texPos];
- ASSIGN_4V(tex, 1, 1, 0, 1);
-
- /* emit 6 tris for the quad strip */
tri.v[0] = v[2]; tri.v[1] = v[1]; tri.v[2] = v[0];
- stage->next->tri( stage->next, &tri );
+ stage->next->tri(stage->next, &tri);
tri.v[0] = v[3]; tri.v[1] = v[1]; tri.v[2] = v[2];
- stage->next->tri( stage->next, &tri );
-
- tri.v[0] = v[4]; tri.v[1] = v[3]; tri.v[2] = v[2];
- stage->next->tri( stage->next, &tri );
-
- tri.v[0] = v[5]; tri.v[1] = v[3]; tri.v[2] = v[4];
- stage->next->tri( stage->next, &tri );
-
- tri.v[0] = v[6]; tri.v[1] = v[5]; tri.v[2] = v[4];
- stage->next->tri( stage->next, &tri );
-
- tri.v[0] = v[7]; tri.v[1] = v[5]; tri.v[2] = v[6];
- stage->next->tri( stage->next, &tri );
+ stage->next->tri(stage->next, &tri);
}
struct draw_context *draw = stage->draw;
struct pipe_context *pipe = draw->pipe;
const struct pipe_rasterizer_state *rast = draw->rasterizer;
- uint num_samplers;
void *r;
- assert(draw->rasterizer->line_smooth);
+ assert(draw->rasterizer->line_smooth && !draw->rasterizer->multisample);
- if (draw->rasterizer->line_width <= 2.2)
- aaline->half_line_width = 1.1f;
+ if (draw->rasterizer->line_width <= 1.0)
+ aaline->half_line_width = 1.0;
else
- aaline->half_line_width = 0.5f * draw->rasterizer->line_width;
+ aaline->half_line_width = 0.5f * draw->rasterizer->line_width + 0.5f;
+
+ if (!draw->rasterizer->half_pixel_center)
+ /*
+ * The tex coords probably would need adjustments?
+ */
+ debug_printf("aa lines without half pixel center may be wrong\n");
/*
- * Bind (generate) our fragprog, sampler and texture
+ * Bind (generate) our fragprog
*/
if (!bind_aaline_fragment_shader(aaline)) {
stage->line = draw_pipe_passthrough_line;
draw_aaline_prepare_outputs(draw, draw->pipeline.aaline);
- /* how many samplers? */
- /* we'll use sampler/texture[pstip->sampler_unit] for the stipple */
- num_samplers = MAX2(aaline->num_sampler_views, aaline->num_samplers);
- num_samplers = MAX2(num_samplers, aaline->fs->sampler_unit + 1);
-
- aaline->state.sampler[aaline->fs->sampler_unit] = aaline->sampler_cso;
- pipe_sampler_view_reference(&aaline->state.sampler_views[aaline->fs->sampler_unit],
- aaline->sampler_view);
-
draw->suspend_flushing = TRUE;
- aaline->driver_bind_sampler_states(pipe, PIPE_SHADER_FRAGMENT, 0,
- num_samplers, aaline->state.sampler);
-
- aaline->driver_set_sampler_views(pipe, PIPE_SHADER_FRAGMENT, 0,
- num_samplers, aaline->state.sampler_views);
-
/* Disable triangle culling, stippling, unfilled mode etc. */
r = draw_get_rasterizer_no_cull(draw, rast->scissor, rast->flatshade);
pipe->bind_rasterizer_state(pipe, r);
struct pipe_context *pipe = draw->pipe;
stage->line = aaline_first_line;
- stage->next->flush( stage->next, flags );
+ stage->next->flush(stage->next, flags);
- /* restore original frag shader, texture, sampler state */
+ /* restore original frag shader */
draw->suspend_flushing = TRUE;
aaline->driver_bind_fs_state(pipe, aaline->fs ? aaline->fs->driver_fs : NULL);
- aaline->driver_bind_sampler_states(pipe, PIPE_SHADER_FRAGMENT, 0,
- aaline->num_samplers,
- aaline->state.sampler);
-
- aaline->driver_set_sampler_views(pipe, PIPE_SHADER_FRAGMENT, 0,
- aaline->num_samplers,
- aaline->state.sampler_views);
-
/* restore original rasterizer state */
if (draw->rast_handle) {
pipe->bind_rasterizer_state(pipe, draw->rast_handle);
static void
aaline_reset_stipple_counter(struct draw_stage *stage)
{
- stage->next->reset_stipple_counter( stage->next );
+ stage->next->reset_stipple_counter(stage->next);
}
{
struct aaline_stage *aaline = aaline_stage(stage);
struct pipe_context *pipe = stage->draw->pipe;
- uint i;
- for (i = 0; i < PIPE_MAX_SHADER_SAMPLER_VIEWS; i++) {
- pipe_sampler_view_reference(&aaline->state.sampler_views[i], NULL);
- }
-
- if (aaline->sampler_cso)
- pipe->delete_sampler_state(pipe, aaline->sampler_cso);
-
- if (aaline->texture)
- pipe_resource_reference(&aaline->texture, NULL);
-
- if (aaline->sampler_view) {
- pipe_sampler_view_reference(&aaline->sampler_view, NULL);
- }
-
- draw_free_temp_verts( stage );
+ draw_free_temp_verts(stage);
/* restore the old entry points */
pipe->create_fs_state = aaline->driver_create_fs_state;
pipe->bind_fs_state = aaline->driver_bind_fs_state;
pipe->delete_fs_state = aaline->driver_delete_fs_state;
- pipe->bind_sampler_states = aaline->driver_bind_sampler_states;
- pipe->set_sampler_views = aaline->driver_set_sampler_views;
-
- FREE( stage );
+ FREE(stage);
}
draw_aaline_stage(struct draw_context *draw)
{
struct aaline_stage *aaline = CALLOC_STRUCT(aaline_stage);
- if (aaline == NULL)
+ if (!aaline)
return NULL;
aaline->stage.draw = draw;
aaline->stage.reset_stipple_counter = aaline_reset_stipple_counter;
aaline->stage.destroy = aaline_destroy;
- if (!draw_alloc_temp_verts( &aaline->stage, 8 ))
+ if (!draw_alloc_temp_verts(&aaline->stage, 8))
goto fail;
return aaline;
fail:
- if (aaline)
- aaline->stage.destroy(&aaline->stage);
+ aaline->stage.destroy(&aaline->stage);
return NULL;
}
/**
* This function overrides the driver's create_fs_state() function and
- * will typically be called by the state tracker.
+ * will typically be called by the gallium frontend.
*/
static void *
aaline_create_fs_state(struct pipe_context *pipe,
struct aaline_stage *aaline = aaline_stage_from_pipe(pipe);
struct aaline_fragment_shader *aafs = NULL;
- if (aaline == NULL)
+ if (!aaline)
return NULL;
aafs = CALLOC_STRUCT(aaline_fragment_shader);
- if (aafs == NULL)
+ if (!aafs)
return NULL;
- aafs->state.tokens = tgsi_dup_tokens(fs->tokens);
+ aafs->state.type = fs->type;
+ if (fs->type == PIPE_SHADER_IR_TGSI)
+ aafs->state.tokens = tgsi_dup_tokens(fs->tokens);
+#ifdef LLVM_AVAILABLE
+ else
+ aafs->state.ir.nir = nir_shader_clone(NULL, fs->ir.nir);
+#endif
/* pass-through */
aafs->driver_fs = aaline->driver_create_fs_state(pipe, fs);
struct aaline_stage *aaline = aaline_stage_from_pipe(pipe);
struct aaline_fragment_shader *aafs = (struct aaline_fragment_shader *) fs;
- if (aaline == NULL) {
+ if (!aaline) {
return;
}
struct aaline_stage *aaline = aaline_stage_from_pipe(pipe);
struct aaline_fragment_shader *aafs = (struct aaline_fragment_shader *) fs;
- if (aafs == NULL) {
+ if (!aafs) {
return;
}
- if (aaline != NULL) {
+ if (aaline) {
/* pass-through */
aaline->driver_delete_fs_state(pipe, aafs->driver_fs);
aaline->driver_delete_fs_state(pipe, aafs->aaline_fs);
}
- FREE((void*)aafs->state.tokens);
+ if (aafs->state.type == PIPE_SHADER_IR_TGSI)
+ FREE((void*)aafs->state.tokens);
+ else
+ ralloc_free(aafs->state.ir.nir);
FREE(aafs);
}
-static void
-aaline_bind_sampler_states(struct pipe_context *pipe, unsigned shader,
- unsigned start, unsigned num, void **sampler)
-{
- struct aaline_stage *aaline = aaline_stage_from_pipe(pipe);
-
- assert(start == 0);
-
- if (aaline == NULL) {
- return;
- }
-
- if (shader == PIPE_SHADER_FRAGMENT) {
- /* save current */
- memcpy(aaline->state.sampler, sampler, num * sizeof(void *));
- aaline->num_samplers = num;
- }
-
- /* pass-through */
- aaline->driver_bind_sampler_states(pipe, shader, start, num, sampler);
-}
-
-
-static void
-aaline_set_sampler_views(struct pipe_context *pipe, unsigned shader,
- unsigned start, unsigned num,
- struct pipe_sampler_view **views)
-{
- struct aaline_stage *aaline = aaline_stage_from_pipe(pipe);
- uint i;
-
- if (aaline == NULL) {
- return;
- }
-
- if (shader == PIPE_SHADER_FRAGMENT) {
- /* save current */
- for (i = 0; i < num; i++) {
- pipe_sampler_view_reference(&aaline->state.sampler_views[start + i],
- views[i]);
- }
- aaline->num_sampler_views = num;
- }
-
- /* pass-through */
- aaline->driver_set_sampler_views(pipe, shader, start, num, views);
-}
-
-
void
draw_aaline_prepare_outputs(struct draw_context *draw,
struct draw_stage *stage)
const struct pipe_rasterizer_state *rast = draw->rasterizer;
/* update vertex attrib info */
- aaline->pos_slot = draw_current_shader_position_output(draw);;
+ aaline->pos_slot = draw_current_shader_position_output(draw);
- if (!rast->line_smooth)
+ if (!rast->line_smooth || rast->multisample)
return;
/* allocate the extra post-transformed vertex attribute */
- aaline->tex_slot = draw_alloc_extra_vertex_attrib(draw,
- TGSI_SEMANTIC_GENERIC,
- aaline->fs->generic_attrib);
+ if (aaline->fs && aaline->fs->aaline_fs)
+ aaline->coord_slot = draw_alloc_extra_vertex_attrib(draw,
+ TGSI_SEMANTIC_GENERIC,
+ aaline->fs->generic_attrib);
+ else
+ aaline->coord_slot = -1;
}
/**
/*
* Create / install AA line drawing / prim stage
*/
- aaline = draw_aaline_stage( draw );
+ aaline = draw_aaline_stage(draw);
if (!aaline)
goto fail;
aaline->driver_bind_fs_state = pipe->bind_fs_state;
aaline->driver_delete_fs_state = pipe->delete_fs_state;
- aaline->driver_bind_sampler_states = pipe->bind_sampler_states;
- aaline->driver_set_sampler_views = pipe->set_sampler_views;
-
- /* create special texture, sampler state */
- if (!aaline_create_texture(aaline))
- goto fail;
-
- if (!aaline_create_sampler(aaline))
- goto fail;
-
/* override the driver's functions */
pipe->create_fs_state = aaline_create_fs_state;
pipe->bind_fs_state = aaline_bind_fs_state;
pipe->delete_fs_state = aaline_delete_fs_state;
- pipe->bind_sampler_states = aaline_bind_sampler_states;
- pipe->set_sampler_views = aaline_set_sampler_views;
-
/* Install once everything is known to be OK:
*/
draw->pipeline.aaline = &aaline->stage;
return TRUE;
- fail:
+fail:
if (aaline)
- aaline->stage.destroy( &aaline->stage );
-
+ aaline->stage.destroy(&aaline->stage);
+
return FALSE;
}
projects / mesa.git / blobdiff