/**
* TGSI transform prolog callback.
*/
static void
aa_transform_prolog(struct tgsi_transform_context *ctx)
{
struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
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 */
tgsi_transform_input_decl(ctx, aactx->maxInput + 1,
TGSI_SEMANTIC_GENERIC, aactx->maxGeneric + 1,
TGSI_INTERPOLATE_LINEAR);
/* declare new sampler */
tgsi_transform_sampler_decl(ctx, aactx->freeSampler);
/* declare new temp regs */
tgsi_transform_temp_decl(ctx, aactx->texTemp);
tgsi_transform_temp_decl(ctx, aactx->colorTemp);
}
static void
transform_instr(struct tgsi_transform_context *tctx,
struct tgsi_full_instruction *current_inst)
{
struct tgsi_atifs_transform *ctx = tgsi_atifs_transform(tctx);
if (ctx->first_instruction_emitted)
goto transform_inst;
ctx->first_instruction_emitted = true;
if (ctx->key->fog) {
/* add a new temp for the fog factor */
ctx->fog_factor_temp = ctx->info.file_max[TGSI_FILE_TEMPORARY] + 1;
tgsi_transform_temp_decl(tctx, ctx->fog_factor_temp);
/* add immediates for clamp */
ctx->fog_clamp_imm = ctx->info.immediate_count;
tgsi_transform_immediate_decl(tctx, 1.0f, 0.0f, 0.0f, 0.0f);
}
transform_inst:
if (current_inst->Instruction.Opcode == TGSI_OPCODE_TEX) {
/* fix texture target */
unsigned newtarget = ctx->key->texture_targets[current_inst->Src[1].Register.Index];
if (newtarget)
current_inst->Texture.Texture = newtarget;
} else if (ctx->key->fog && current_inst->Instruction.Opcode == TGSI_OPCODE_MOV &&
current_inst->Dst[0].Register.File == TGSI_FILE_OUTPUT) {
struct tgsi_full_instruction inst;
unsigned i;
int fogc_index = -1;
int reg0_index = current_inst->Src[0].Register.Index;
/* find FOGC input */
for (i = 0; i < ctx->info.num_inputs; i++) {
if (ctx->info.input_semantic_name[i] == TGSI_SEMANTIC_FOG) {
fogc_index = i;
break;
}
}
if (fogc_index < 0) {
/* should never be reached, because fog coord input is always declared */
tctx->emit_instruction(tctx, current_inst);
return;
}
/* compute the 1 component fog factor f */
if (ctx->key->fog == 1) {
/* LINEAR formula: f = (end - z) / (end - start)
* with optimized parameters:
* f = MAD(fogcoord, oparams.x, oparams.y)
*/
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_MAD;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 3;
SET_SRC(&inst, 0, TGSI_FILE_INPUT, fogc_index, X, Y, Z, W);
SET_SRC(&inst, 1, TGSI_FILE_CONSTANT, MAX_NUM_FRAGMENT_CONSTANTS_ATI, X, X, X, X);
SET_SRC(&inst, 2, TGSI_FILE_CONSTANT, MAX_NUM_FRAGMENT_CONSTANTS_ATI, Y, Y, Y, Y);
tctx->emit_instruction(tctx, &inst);
} else if (ctx->key->fog == 2) {
/* EXP formula: f = exp(-dens * z)
* with optimized parameters:
* f = MUL(fogcoord, oparams.z); f= EX2(-f)
*/
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_MUL;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 2;
SET_SRC(&inst, 0, TGSI_FILE_INPUT, fogc_index, X, Y, Z, W);
SET_SRC(&inst, 1, TGSI_FILE_CONSTANT, MAX_NUM_FRAGMENT_CONSTANTS_ATI, Z, Z, Z, Z);
tctx->emit_instruction(tctx, &inst);
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_EX2;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 1;
SET_SRC(&inst, 0, TGSI_FILE_TEMPORARY, ctx->fog_factor_temp, X, Y, Z, W);
inst.Src[0].Register.Negate = 1;
tctx->emit_instruction(tctx, &inst);
} else if (ctx->key->fog == 3) {
/* EXP2 formula: f = exp(-(dens * z)^2)
* with optimized parameters:
* f = MUL(fogcoord, oparams.w); f=MUL(f, f); f= EX2(-f)
*/
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_MUL;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 2;
SET_SRC(&inst, 0, TGSI_FILE_INPUT, fogc_index, X, Y, Z, W);
SET_SRC(&inst, 1, TGSI_FILE_CONSTANT, MAX_NUM_FRAGMENT_CONSTANTS_ATI, W, W, W, W);
tctx->emit_instruction(tctx, &inst);
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_MUL;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 2;
SET_SRC(&inst, 0, TGSI_FILE_TEMPORARY, ctx->fog_factor_temp, X, Y, Z, W);
SET_SRC(&inst, 1, TGSI_FILE_TEMPORARY, ctx->fog_factor_temp, X, Y, Z, W);
tctx->emit_instruction(tctx, &inst);
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_EX2;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 1;
SET_SRC(&inst, 0, TGSI_FILE_TEMPORARY, ctx->fog_factor_temp, X, Y, Z, W);
inst.Src[0].Register.Negate ^= 1;
tctx->emit_instruction(tctx, &inst);
}
/* f = CLAMP(f, 0.0, 1.0) */
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_CLAMP;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = ctx->fog_factor_temp;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 3;
SET_SRC(&inst, 0, TGSI_FILE_TEMPORARY, ctx->fog_factor_temp, X, Y, Z, W);
SET_SRC(&inst, 1, TGSI_FILE_IMMEDIATE, ctx->fog_clamp_imm, Y, Y, Y, Y); // 0.0
SET_SRC(&inst, 2, TGSI_FILE_IMMEDIATE, ctx->fog_clamp_imm, X, X, X, X); // 1.0
tctx->emit_instruction(tctx, &inst);
/* REG0 = LRP(f, REG0, fogcolor) */
inst = tgsi_default_full_instruction();
inst.Instruction.Opcode = TGSI_OPCODE_LRP;
inst.Instruction.NumDstRegs = 1;
inst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
inst.Dst[0].Register.Index = reg0_index;
inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZW;
inst.Instruction.NumSrcRegs = 3;
SET_SRC(&inst, 0, TGSI_FILE_TEMPORARY, ctx->fog_factor_temp, X, X, X, Y);
SET_SRC(&inst, 1, TGSI_FILE_TEMPORARY, reg0_index, X, Y, Z, W);
SET_SRC(&inst, 2, TGSI_FILE_CONSTANT, MAX_NUM_FRAGMENT_CONSTANTS_ATI + 1, X, Y, Z, W);
tctx->emit_instruction(tctx, &inst);
}
tctx->emit_instruction(tctx, current_inst);
}
/**
* TGSI transform callback.
* Insert new declarations and instructions before first instruction.
*/
static void
aa_transform_prolog(struct tgsi_transform_context *ctx)
{
/* emit our new declarations before the first instruction */
struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
struct tgsi_full_instruction newInst;
const int texInput = aactx->maxInput + 1;
int tmp0;
uint i;
/* find two free temp regs */
for (i = 0; i < 32; i++) {
if ((aactx->tempsUsed & (1 << i)) == 0) {
/* found a free temp */
if (aactx->tmp0 < 0)
aactx->tmp0 = i;
else if (aactx->colorTemp < 0)
aactx->colorTemp = i;
else
break;
}
}
assert(aactx->colorTemp != aactx->tmp0);
tmp0 = aactx->tmp0;
/* declare new generic input/texcoord */
tgsi_transform_input_decl(ctx, texInput,
TGSI_SEMANTIC_GENERIC, aactx->maxGeneric + 1,
TGSI_INTERPOLATE_LINEAR);
/* declare new temp regs */
tgsi_transform_temp_decl(ctx, tmp0);
tgsi_transform_temp_decl(ctx, aactx->colorTemp);
/*
* Emit code to compute fragment coverage, kill if outside point radius
*
* Temp reg0 usage:
* t0.x = distance of fragment from center point
* t0.y = boolean, is t0.x > 1.0, also misc temp usage
* t0.z = temporary for computing 1/(1-k) value
* t0.w = final coverage value
*/
/* MUL t0.xy, tex, tex; # compute x^2, y^2 */
tgsi_transform_op2_inst(ctx, TGSI_OPCODE_MUL,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_XY,
TGSI_FILE_INPUT, texInput,
TGSI_FILE_INPUT, texInput);
/* ADD t0.x, t0.x, t0.y; # x^2 + y^2 */
tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_ADD,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_X,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_X,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_Y);
#if NORMALIZE /* OPTIONAL normalization of length */
/* RSQ t0.x, t0.x; */
tgsi_transform_op1_inst(ctx, TGSI_OPCODE_RSQ,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_X,
TGSI_FILE_TEMPORARY, tmp0);
/* RCP t0.x, t0.x; */
tgsi_transform_op1_inst(ctx, TGSI_OPCODE_RCP,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_X,
TGSI_FILE_TEMPORARY, tmp0);
#endif
/* SGT t0.y, t0.xxxx, tex.wwww; # bool b = d > 1 (NOTE tex.w == 1) */
tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_SGT,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_Y,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_X,
TGSI_FILE_INPUT, texInput, TGSI_SWIZZLE_W);
/* KILL_IF -tmp0.yyyy; # if -tmp0.y < 0, KILL */
tgsi_transform_kill_inst(ctx, TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_Y);
/* compute coverage factor = (1-d)/(1-k) */
/* SUB t0.z, tex.w, tex.z; # m = 1 - k */
tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_SUB,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_Z,
TGSI_FILE_INPUT, texInput, TGSI_SWIZZLE_W,
TGSI_FILE_INPUT, texInput, TGSI_SWIZZLE_Z);
/* RCP t0.z, t0.z; # t0.z = 1 / m */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_RCP;
newInst.Instruction.NumDstRegs = 1;
newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
newInst.Dst[0].Register.Index = tmp0;
newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_Z;
newInst.Instruction.NumSrcRegs = 1;
newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
newInst.Src[0].Register.Index = tmp0;
newInst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_Z;
ctx->emit_instruction(ctx, &newInst);
/* SUB t0.y, 1, t0.x; # d = 1 - d */
tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_SUB,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_Y,
TGSI_FILE_INPUT, texInput, TGSI_SWIZZLE_W,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_X);
/* MUL t0.w, t0.y, t0.z; # coverage = d * m */
tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_MUL,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_W,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_Y,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_Z);
/* SLE t0.y, t0.x, tex.z; # bool b = distance <= k */
tgsi_transform_op2_swz_inst(ctx, TGSI_OPCODE_SLE,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_Y,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_X,
TGSI_FILE_INPUT, texInput, TGSI_SWIZZLE_Z);
/* CMP t0.w, -t0.y, tex.w, t0.w;
* # if -t0.y < 0 then
* t0.w = 1
* else
* t0.w = t0.w
*/
tgsi_transform_op3_swz_inst(ctx, TGSI_OPCODE_CMP,
TGSI_FILE_TEMPORARY, tmp0, TGSI_WRITEMASK_W,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_Y, 1,
TGSI_FILE_INPUT, texInput, TGSI_SWIZZLE_W,
TGSI_FILE_TEMPORARY, tmp0, TGSI_SWIZZLE_W);
}
