*
* Copyright 2011-2012 Advanced Micro Devices, Inc.
* Copyright 2009 VMware, Inc.
- * Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2007-2008 VMware, Inc.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* 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.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
* 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.
#include "lp_bld_tgsi.h"
#include "lp_bld_arit.h"
+#include "lp_bld_bitarit.h"
#include "lp_bld_const.h"
#include "lp_bld_gather.h"
#include "lp_bld_logic.h"
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = lp_build_emit_llvm_unary(bld_base,
- TGSI_OPCODE_ROUND, emit_data->args[0]);
+ LLVMValueRef tmp = lp_build_emit_llvm_unary(bld_base, TGSI_OPCODE_ROUND, emit_data->args[0]);
+ emit_data->output[emit_data->chan] = LLVMBuildFPToSI(bld_base->base.gallivm->builder, tmp,
+ bld_base->uint_bld.vec_type, "");
}
/* TGSI_OPCODE_CLAMP */
}
static struct lp_build_tgsi_action dp2_action = {
- .fetch_args = dp2_fetch_args,
- .emit = dp2_emit
+ dp2_fetch_args, /* fetch_args */
+ dp2_emit /* emit */
};
/* TGSI_OPCODE_DP2A */
}
static struct lp_build_tgsi_action dp2a_action = {
- .fetch_args = dp2a_fetch_args,
- .emit = dp2a_emit
+ dp2a_fetch_args, /* fetch_args */
+ dp2a_emit /* emit */
};
/* TGSI_OPCODE_DP3 */
}
static struct lp_build_tgsi_action dp3_action = {
- .fetch_args = dp3_fetch_args,
- .emit = dp3_emit
+ dp3_fetch_args, /* fetch_args */
+ dp3_emit /* emit */
};
/* TGSI_OPCODDE_DP4 */
}
static struct lp_build_tgsi_action dp4_action = {
- .fetch_args = dp4_fetch_args,
- .emit = dp4_emit
+ dp4_fetch_args, /* fetch_args */
+ dp4_emit /* emit */
};
/* TGSI_OPCODE_DPH */
}
const struct lp_build_tgsi_action dph_action = {
- .fetch_args = dph_fetch_args,
- .emit = dp4_emit
+ dph_fetch_args, /* fetch_args */
+ dp4_emit /* emit */
};
/* TGSI_OPCODE_DST */
}
static struct lp_build_tgsi_action dst_action = {
- .fetch_args = dst_fetch_args,
- .emit = dst_emit
+ dst_fetch_args, /* fetch_args */
+ dst_emit /* emit */
};
/* TGSI_OPCODE_END */
}
const struct lp_build_tgsi_action exp_action = {
- .fetch_args = scalar_unary_fetch_args,
- .emit = exp_emit
+ scalar_unary_fetch_args, /* fetch_args */
+ exp_emit /* emit */
};
/* TGSI_OPCODE_FRC */
TGSI_OPCODE_SUB, emit_data->args[0], tmp);
}
-/* TGSI_OPCODE_KIL */
+/* TGSI_OPCODE_KILL_IF */
static void
kil_fetch_args(
emit_data->dst_type = LLVMVoidTypeInContext(bld_base->base.gallivm->context);
}
-/* TGSI_OPCODE_KILP */
+/* TGSI_OPCODE_KILL */
static void
kilp_fetch_args(
}
static struct lp_build_tgsi_action lit_action = {
- .fetch_args = lit_fetch_args,
- .emit = lit_emit
+ lit_fetch_args, /* fetch_args */
+ lit_emit /* emit */
};
/* TGSI_OPCODE_LOG */
}
static struct lp_build_tgsi_action log_action = {
- .fetch_args = scalar_unary_fetch_args,
- .emit = log_emit
+ scalar_unary_fetch_args, /* fetch_args */
+ log_emit /* emit */
};
/* TGSI_OPCODE_LRP */
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = lp_build_mul(&bld_base->base,
- emit_data->args[0], emit_data->args[1]);
+ emit_data->output[emit_data->chan] = LLVMBuildFMul(
+ bld_base->base.gallivm->builder,
+ emit_data->args[0], emit_data->args[1], "");
+}
+
+/*.TGSI_OPCODE_DIV.*/
+static void fdiv_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = LLVMBuildFDiv(
+ bld_base->base.gallivm->builder,
+ emit_data->args[0], emit_data->args[1], "");
+}
+
+/*.TGSI_OPCODE_RCP.*/
+static void rcp_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMValueRef one;
+ one = lp_build_const_float(bld_base->base.gallivm, 1.0f);
+ emit_data->output[emit_data->chan] = lp_build_emit_llvm_binary(bld_base,
+ TGSI_OPCODE_DIV, one, emit_data->args[0]);
}
/* TGSI_OPCODE_POW */
}
static struct lp_build_tgsi_action pow_action = {
- .fetch_args = scalar_binary_fetch_args,
- .emit = pow_emit
+ scalar_binary_fetch_args, /* fetch_args */
+ pow_emit /* emit */
};
/* TGSI_OPCODE_RSQ */
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->args[0] = lp_build_emit_llvm_unary(bld_base, TGSI_OPCODE_ABS,
- emit_data->args[0]);
if (bld_base->rsq_action.emit) {
bld_base->rsq_action.emit(&bld_base->rsq_action, bld_base, emit_data);
} else {
}
const struct lp_build_tgsi_action rsq_action = {
- .fetch_args = scalar_unary_fetch_args,
- .emit = rsq_emit
+ scalar_unary_fetch_args, /* fetch_args */
+ rsq_emit /* emit */
};
+/* TGSI_OPCODE_SQRT */
+
+static void
+sqrt_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ if (bld_base->sqrt_action.emit) {
+ bld_base->sqrt_action.emit(&bld_base->sqrt_action, bld_base, emit_data);
+ } else {
+ emit_data->output[emit_data->chan] = bld_base->base.undef;
+ }
+}
+
+const struct lp_build_tgsi_action sqrt_action = {
+ scalar_unary_fetch_args, /* fetch_args */
+ sqrt_emit /* emit */
+};
+
/* TGSI_OPCODE_SCS */
static void
scs_emit(
}
const struct lp_build_tgsi_action scs_action = {
- .fetch_args = scalar_unary_fetch_args,
- .emit = scs_emit
+ scalar_unary_fetch_args, /* fetch_args */
+ scs_emit /* emit */
};
-/* TGSI_OPCODE_SFL */
+/* TGSI_OPCODE_SUB */
+static void
+sub_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] =
+ LLVMBuildFSub(bld_base->base.gallivm->builder,
+ emit_data->args[0],
+ emit_data->args[1], "");
+}
+/* TGSI_OPCODE_F2U */
static void
-sfl_emit(
+f2u_emit(
const struct lp_build_tgsi_action * action,
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = bld_base->base.zero;
+ emit_data->output[emit_data->chan] =
+ LLVMBuildFPToUI(bld_base->base.gallivm->builder,
+ emit_data->args[0],
+ bld_base->base.int_vec_type, "");
}
-/* TGSI_OPCODE_STR */
+/* TGSI_OPCODE_U2F */
+static void
+u2f_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] =
+ LLVMBuildUIToFP(bld_base->base.gallivm->builder,
+ emit_data->args[0],
+ bld_base->base.vec_type, "");
+}
static void
-str_emit(
+umad_emit(
const struct lp_build_tgsi_action * action,
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = bld_base->base.one;
+ LLVMValueRef tmp;
+ tmp = lp_build_emit_llvm_binary(bld_base, TGSI_OPCODE_UMUL,
+ emit_data->args[0],
+ emit_data->args[1]);
+ emit_data->output[emit_data->chan] = lp_build_emit_llvm_binary(bld_base,
+ TGSI_OPCODE_UADD, tmp, emit_data->args[2]);
}
-/* TGSI_OPCODE_SUB */
+/* TGSI_OPCODE_UMUL */
static void
-sub_emit(
+umul_emit(
const struct lp_build_tgsi_action * action,
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = LLVMBuildFSub(
- bld_base->base.gallivm->builder,
- emit_data->args[0],
- emit_data->args[1], "");
+ emit_data->output[emit_data->chan] = lp_build_mul(&bld_base->uint_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_IMUL_HI */
+static void
+imul_hi_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ struct lp_build_context *int_bld = &bld_base->int_bld;
+ struct lp_type type = int_bld->type;
+ LLVMValueRef src0, src1;
+ LLVMValueRef dst64;
+ LLVMTypeRef typeRef;
+
+ assert(type.width == 32);
+ type.width = 64;
+ typeRef = lp_build_vec_type(bld_base->base.gallivm, type);
+ src0 = LLVMBuildSExt(builder, emit_data->args[0], typeRef, "");
+ src1 = LLVMBuildSExt(builder, emit_data->args[1], typeRef, "");
+ dst64 = LLVMBuildMul(builder, src0, src1, "");
+ dst64 = LLVMBuildAShr(
+ builder, dst64,
+ lp_build_const_vec(bld_base->base.gallivm, type, 32), "");
+ type.width = 32;
+ typeRef = lp_build_vec_type(bld_base->base.gallivm, type);
+ emit_data->output[emit_data->chan] =
+ LLVMBuildTrunc(builder, dst64, typeRef, "");
+}
+
+/* TGSI_OPCODE_UMUL_HI */
+static void
+umul_hi_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ struct lp_build_context *uint_bld = &bld_base->uint_bld;
+ struct lp_type type = uint_bld->type;
+ LLVMValueRef src0, src1;
+ LLVMValueRef dst64;
+ LLVMTypeRef typeRef;
+
+ assert(type.width == 32);
+ type.width = 64;
+ typeRef = lp_build_vec_type(bld_base->base.gallivm, type);
+ src0 = LLVMBuildZExt(builder, emit_data->args[0], typeRef, "");
+ src1 = LLVMBuildZExt(builder, emit_data->args[1], typeRef, "");
+ dst64 = LLVMBuildMul(builder, src0, src1, "");
+ dst64 = LLVMBuildLShr(
+ builder, dst64,
+ lp_build_const_vec(bld_base->base.gallivm, type, 32), "");
+ type.width = 32;
+ typeRef = lp_build_vec_type(bld_base->base.gallivm, type);
+ emit_data->output[emit_data->chan] =
+ LLVMBuildTrunc(builder, dst64, typeRef, "");
+}
+
+/* TGSI_OPCODE_MAX */
+static void fmax_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ emit_data->output[emit_data->chan] = LLVMBuildSelect(builder,
+ LLVMBuildFCmp(builder, LLVMRealUGE,
+ emit_data->args[0], emit_data->args[1], ""),
+ emit_data->args[0], emit_data->args[1], "");
+}
+
+/* TGSI_OPCODE_MIN */
+static void fmin_emit(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ emit_data->output[emit_data->chan] = LLVMBuildSelect(builder,
+ LLVMBuildFCmp(builder, LLVMRealUGE,
+ emit_data->args[0], emit_data->args[1], ""),
+ emit_data->args[1], emit_data->args[0], "");
}
/* TGSI_OPCODE_XPD */
}
const struct lp_build_tgsi_action xpd_action = {
- .fetch_args = xpd_fetch_args,
- .emit = xpd_emit
+ xpd_fetch_args, /* fetch_args */
+ xpd_emit /* emit */
};
void
bld_base->op_actions[TGSI_OPCODE_LIT] = lit_action;
bld_base->op_actions[TGSI_OPCODE_LOG] = log_action;
bld_base->op_actions[TGSI_OPCODE_RSQ] = rsq_action;
+ bld_base->op_actions[TGSI_OPCODE_SQRT] = sqrt_action;
bld_base->op_actions[TGSI_OPCODE_POW] = pow_action;
bld_base->op_actions[TGSI_OPCODE_SCS] = scs_action;
bld_base->op_actions[TGSI_OPCODE_XPD] = xpd_action;
+ bld_base->op_actions[TGSI_OPCODE_BREAKC].fetch_args = scalar_unary_fetch_args;
+ bld_base->op_actions[TGSI_OPCODE_SWITCH].fetch_args = scalar_unary_fetch_args;
+ bld_base->op_actions[TGSI_OPCODE_CASE].fetch_args = scalar_unary_fetch_args;
bld_base->op_actions[TGSI_OPCODE_COS].fetch_args = scalar_unary_fetch_args;
bld_base->op_actions[TGSI_OPCODE_EX2].fetch_args = scalar_unary_fetch_args;
bld_base->op_actions[TGSI_OPCODE_IF].fetch_args = scalar_unary_fetch_args;
- bld_base->op_actions[TGSI_OPCODE_KIL].fetch_args = kil_fetch_args;
- bld_base->op_actions[TGSI_OPCODE_KILP].fetch_args = kilp_fetch_args;
+ bld_base->op_actions[TGSI_OPCODE_UIF].fetch_args = scalar_unary_fetch_args;
+ bld_base->op_actions[TGSI_OPCODE_KILL_IF].fetch_args = kil_fetch_args;
+ bld_base->op_actions[TGSI_OPCODE_KILL].fetch_args = kilp_fetch_args;
bld_base->op_actions[TGSI_OPCODE_RCP].fetch_args = scalar_unary_fetch_args;
bld_base->op_actions[TGSI_OPCODE_SIN].fetch_args = scalar_unary_fetch_args;
bld_base->op_actions[TGSI_OPCODE_LG2].fetch_args = scalar_unary_fetch_args;
bld_base->op_actions[TGSI_OPCODE_MAD].emit = mad_emit;
bld_base->op_actions[TGSI_OPCODE_MOV].emit = mov_emit;
bld_base->op_actions[TGSI_OPCODE_MUL].emit = mul_emit;
- bld_base->op_actions[TGSI_OPCODE_SFL].emit = sfl_emit;
- bld_base->op_actions[TGSI_OPCODE_STR].emit = str_emit;
+ bld_base->op_actions[TGSI_OPCODE_DIV].emit = fdiv_emit;
+ bld_base->op_actions[TGSI_OPCODE_RCP].emit = rcp_emit;
bld_base->op_actions[TGSI_OPCODE_SUB].emit = sub_emit;
+
+ bld_base->op_actions[TGSI_OPCODE_UARL].emit = mov_emit;
+ bld_base->op_actions[TGSI_OPCODE_F2U].emit = f2u_emit;
+ bld_base->op_actions[TGSI_OPCODE_U2F].emit = u2f_emit;
+ bld_base->op_actions[TGSI_OPCODE_UMAD].emit = umad_emit;
+ bld_base->op_actions[TGSI_OPCODE_UMUL].emit = umul_emit;
+ bld_base->op_actions[TGSI_OPCODE_IMUL_HI].emit = imul_hi_emit;
+ bld_base->op_actions[TGSI_OPCODE_UMUL_HI].emit = umul_hi_emit;
+
+ bld_base->op_actions[TGSI_OPCODE_MAX].emit = fmax_emit;
+ bld_base->op_actions[TGSI_OPCODE_MIN].emit = fmin_emit;
}
/* CPU Only default actions */
emit_data->args[0], emit_data->args[1]);
}
+/* TGSI_OPCODE_AND (CPU Only) */
+static void
+and_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_and(&bld_base->uint_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_ARL (CPU Only) */
+static void
+arl_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMValueRef tmp;
+ tmp = lp_build_floor(&bld_base->base,
+ emit_data->args[0]);
+ emit_data->output[emit_data->chan] = LLVMBuildFPToSI(bld_base->base.gallivm->builder, tmp,
+ bld_base->uint_bld.vec_type, "");
+}
+
+/* TGSI_OPCODE_ARR (CPU Only) */
+static void
+arr_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_iround(&bld_base->base, emit_data->args[0]);
+}
+
/* TGSI_OPCODE_CEIL (CPU Only) */
static void
ceil_emit_cpu(
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = lp_build_trunc(&bld_base->base,
- emit_data->args[0]);
+ emit_data->output[emit_data->chan] = lp_build_ceil(&bld_base->base,
+ emit_data->args[0]);
}
/* TGSI_OPCODE_CMP (CPU Only) */
cond, emit_data->args[1], emit_data->args[2]);
}
-/* TGSI_OPCODE_CND (CPU Only) */
+/* TGSI_OPCODE_UCMP (CPU Only) */
static void
-cnd_emit_cpu(
+ucmp_emit_cpu(
const struct lp_build_tgsi_action * action,
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- LLVMValueRef half, tmp;
- half = lp_build_const_vec(bld_base->base.gallivm, bld_base->base.type, 0.5);
- tmp = lp_build_cmp(&bld_base->base, PIPE_FUNC_GREATER,
- emit_data->args[2], half);
- emit_data->output[emit_data->chan] = lp_build_select(&bld_base->base,
- tmp,
- emit_data->args[0],
- emit_data->args[1]);
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ struct lp_build_context *uint_bld = &bld_base->uint_bld;
+ LLVMValueRef unsigned_cond =
+ LLVMBuildBitCast(builder, emit_data->args[0], uint_bld->vec_type, "");
+ LLVMValueRef cond = lp_build_cmp(uint_bld, PIPE_FUNC_NOTEQUAL,
+ unsigned_cond,
+ uint_bld->zero);
+ emit_data->output[emit_data->chan] =
+ lp_build_select(&bld_base->base,
+ cond, emit_data->args[1], emit_data->args[2]);
}
/* TGSI_OPCODE_COS (CPU Only) */
emit_data->args[0]);
}
-/* TGSI_OPCODE_EXP (CPU Only) */
+/* TGSI_OPCODE_F2I (CPU Only) */
static void
-exp_emit_cpu(
+f2i_emit_cpu(
const struct lp_build_tgsi_action * action,
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- lp_build_exp2_approx(&bld_base->base, emit_data->args[0],
- &emit_data->output[TGSI_CHAN_X],
- &emit_data->output[TGSI_CHAN_Y],
- &emit_data->output[TGSI_CHAN_Z]);
- emit_data->output[TGSI_CHAN_W] = bld_base->base.one;
+ emit_data->output[emit_data->chan] = lp_build_itrunc(&bld_base->base,
+ emit_data->args[0]);
+}
+
+/* TGSI_OPCODE_FSET Helper (CPU Only) */
+static void
+fset_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data,
+ unsigned pipe_func)
+{
+ LLVMValueRef cond;
+
+ if (pipe_func != PIPE_FUNC_NOTEQUAL) {
+ cond = lp_build_cmp_ordered(&bld_base->base, pipe_func,
+ emit_data->args[0], emit_data->args[1]);
+ }
+ else {
+ cond = lp_build_cmp(&bld_base->base, pipe_func,
+ emit_data->args[0], emit_data->args[1]);
+
+ }
+ emit_data->output[emit_data->chan] = cond;
+}
+
+
+/* TGSI_OPCODE_FSEQ (CPU Only) */
+static void
+fseq_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ fset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_EQUAL);
+}
+
+/* TGSI_OPCODE_ISGE (CPU Only) */
+static void
+fsge_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ fset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_GEQUAL);
+}
+
+/* TGSI_OPCODE_ISLT (CPU Only) */
+static void
+fslt_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ fset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_LESS);
+}
+
+/* TGSI_OPCODE_USNE (CPU Only) */
+
+static void
+fsne_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ fset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_NOTEQUAL);
}
/* TGSI_OPCODE_FLR (CPU Only) */
emit_data->args[0]);
}
+/* TGSI_OPCODE_I2F (CPU Only) */
+static void
+i2f_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_int_to_float(&bld_base->base,
+ emit_data->args[0]);
+}
+
+/* TGSI_OPCODE_IABS (CPU Only) */
+static void
+iabs_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_abs(&bld_base->int_bld,
+ emit_data->args[0]);
+}
+
+/* TGSI_OPCODE_IDIV (CPU Only) */
+static void
+idiv_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ LLVMValueRef div_mask = lp_build_cmp(&bld_base->uint_bld,
+ PIPE_FUNC_EQUAL, emit_data->args[1],
+ bld_base->uint_bld.zero);
+ /* We want to make sure that we never divide/mod by zero to not
+ * generate sigfpe. We don't want to crash just because the
+ * shader is doing something weird. */
+ LLVMValueRef divisor = LLVMBuildOr(builder,
+ div_mask,
+ emit_data->args[1], "");
+ LLVMValueRef result = lp_build_div(&bld_base->int_bld,
+ emit_data->args[0], divisor);
+ LLVMValueRef not_div_mask = LLVMBuildNot(builder,
+ div_mask,"");
+ /* idiv by zero doesn't have a guaranteed return value chose 0 for now. */
+ emit_data->output[emit_data->chan] = LLVMBuildAnd(builder,
+ not_div_mask,
+ result, "");
+}
+
+/* TGSI_OPCODE_INEG (CPU Only) */
+static void
+ineg_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_sub(&bld_base->int_bld,
+ bld_base->int_bld.zero,
+ emit_data->args[0]);
+}
+
+/* TGSI_OPCODE_ISET Helper (CPU Only) */
+static void
+iset_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data,
+ unsigned pipe_func)
+{
+ LLVMValueRef cond = lp_build_cmp(&bld_base->int_bld, pipe_func,
+ emit_data->args[0], emit_data->args[1]);
+ emit_data->output[emit_data->chan] = cond;
+}
+
+/* TGSI_OPCODE_IMAX (CPU Only) */
+static void
+imax_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_max(&bld_base->int_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_IMIN (CPU Only) */
+static void
+imin_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_min(&bld_base->int_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_ISGE (CPU Only) */
+static void
+isge_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ iset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_GEQUAL);
+}
+
+/* TGSI_OPCODE_ISHR (CPU Only) */
+static void
+ishr_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ struct lp_build_context *int_bld = &bld_base->int_bld;
+ LLVMValueRef mask = lp_build_const_vec(int_bld->gallivm, int_bld->type,
+ int_bld->type.width - 1);
+ LLVMValueRef masked_count = lp_build_and(int_bld, emit_data->args[1], mask);
+ emit_data->output[emit_data->chan] = lp_build_shr(int_bld, emit_data->args[0],
+ masked_count);
+}
+
+/* TGSI_OPCODE_ISLT (CPU Only) */
+static void
+islt_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ iset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_LESS);
+}
+
+
+/* TGSI_OPCODE_ISSG (CPU Only) */
+static void
+issg_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_sgn(&bld_base->int_bld,
+ emit_data->args[0]);
+}
+
/* TGSI_OPCODE_LG2 (CPU Only) */
static void
lg2_emit_cpu(
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = lp_build_log2(&bld_base->base,
- emit_data->args[0]);
+ emit_data->output[emit_data->chan] = lp_build_log2_safe(&bld_base->base,
+ emit_data->args[0]);
}
/* TGSI_OPCODE_LOG (CPU Only) */
LLVMValueRef src0 = emit_data->args[0];
lp_build_log2_approx(&bld_base->base, src0,
- &p_exp, &p_floor_log2, &p_log2);
+ &p_exp, &p_floor_log2, &p_log2, FALSE);
emit_data->output[TGSI_CHAN_X] = p_floor_log2;
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = lp_build_max(&bld_base->base,
- emit_data->args[0], emit_data->args[1]);
+ emit_data->output[emit_data->chan] =
+ lp_build_max_ext(&bld_base->base,
+ emit_data->args[0], emit_data->args[1],
+ GALLIVM_NAN_RETURN_OTHER);
}
/* TGSI_OPCODE_MIN (CPU Only) */
struct lp_build_tgsi_context * bld_base,
struct lp_build_emit_data * emit_data)
{
- emit_data->output[emit_data->chan] = lp_build_min(&bld_base->base,
+ emit_data->output[emit_data->chan] =
+ lp_build_min_ext(&bld_base->base,
+ emit_data->args[0], emit_data->args[1],
+ GALLIVM_NAN_RETURN_OTHER);
+}
+
+/* TGSI_OPCODE_MOD (CPU Only) */
+static void
+mod_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_mod(&bld_base->int_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_NOT */
+static void
+not_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_not(&bld_base->uint_bld,
+ emit_data->args[0]);
+}
+
+/* TGSI_OPCODE_OR (CPU Only) */
+static void
+or_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_or(&bld_base->uint_bld,
emit_data->args[0], emit_data->args[1]);
}
}
/* Reciprical squareroot (CPU Only) */
-
-/* This is not the same as TGSI_OPCODE_RSQ, which requres the argument to be
- * greater than or equal to 0 */
static void
recip_sqrt_emit_cpu(
const struct lp_build_tgsi_action * action,
emit_data->args[0]);
}
+static void
+sqrt_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_sqrt(&bld_base->base,
+ emit_data->args[0]);
+}
+
+
/* TGSI_OPCODE_ROUND (CPU Only) */
static void
round_emit_cpu(
struct lp_build_emit_data * emit_data,
unsigned pipe_func)
{
- LLVMValueRef cond = lp_build_cmp(&bld_base->base, pipe_func,
- emit_data->args[0], emit_data->args[1]);
+ LLVMValueRef cond;
+
+ if (pipe_func != PIPE_FUNC_NOTEQUAL) {
+ cond = lp_build_cmp_ordered(&bld_base->base, pipe_func,
+ emit_data->args[0], emit_data->args[1]);
+ }
+ else {
+ cond = lp_build_cmp(&bld_base->base, pipe_func,
+ emit_data->args[0], emit_data->args[1]);
+
+ }
emit_data->output[emit_data->chan] = lp_build_select(&bld_base->base,
cond,
bld_base->base.one,
set_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_GREATER);
}
+/* TGSI_OPCODE_SHL (CPU Only) */
+static void
+shl_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ struct lp_build_context *uint_bld = &bld_base->uint_bld;
+ LLVMValueRef mask = lp_build_const_vec(uint_bld->gallivm, uint_bld->type,
+ uint_bld->type.width - 1);
+ LLVMValueRef masked_count = lp_build_and(uint_bld, emit_data->args[1], mask);
+ emit_data->output[emit_data->chan] = lp_build_shl(uint_bld, emit_data->args[0],
+ masked_count);
+}
+
/* TGSI_OPCODE_SIN (CPU Only) */
static void
sin_emit_cpu(
}
/* TGSI_OPCODE_SLT (CPU Only) */
-
static void
slt_emit_cpu(
const struct lp_build_tgsi_action * action,
emit_data->args[0]);
}
+/* TGSI_OPCODE_UADD (CPU Only) */
+static void
+uadd_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_add(&bld_base->uint_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_UDIV (CPU Only) */
+static void
+udiv_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ LLVMValueRef div_mask = lp_build_cmp(&bld_base->uint_bld,
+ PIPE_FUNC_EQUAL, emit_data->args[1],
+ bld_base->uint_bld.zero);
+ /* We want to make sure that we never divide/mod by zero to not
+ * generate sigfpe. We don't want to crash just because the
+ * shader is doing something weird. */
+ LLVMValueRef divisor = LLVMBuildOr(builder,
+ div_mask,
+ emit_data->args[1], "");
+ LLVMValueRef result = lp_build_div(&bld_base->uint_bld,
+ emit_data->args[0], divisor);
+ /* udiv by zero is guaranteed to return 0xffffffff at least with d3d10 */
+ emit_data->output[emit_data->chan] = LLVMBuildOr(builder,
+ div_mask,
+ result, "");
+}
+
+/* TGSI_OPCODE_UMAX (CPU Only) */
+static void
+umax_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_max(&bld_base->uint_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_UMIN (CPU Only) */
+static void
+umin_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_min(&bld_base->uint_bld,
+ emit_data->args[0], emit_data->args[1]);
+}
+
+/* TGSI_OPCODE_UMOD (CPU Only) */
+static void
+umod_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ LLVMBuilderRef builder = bld_base->base.gallivm->builder;
+ LLVMValueRef div_mask = lp_build_cmp(&bld_base->uint_bld,
+ PIPE_FUNC_EQUAL, emit_data->args[1],
+ bld_base->uint_bld.zero);
+ /* We want to make sure that we never divide/mod by zero to not
+ * generate sigfpe. We don't want to crash just because the
+ * shader is doing something weird. */
+ LLVMValueRef divisor = LLVMBuildOr(builder,
+ div_mask,
+ emit_data->args[1], "");
+ LLVMValueRef result = lp_build_mod(&bld_base->uint_bld,
+ emit_data->args[0], divisor);
+ /* umod by zero is guaranteed to return 0xffffffff */
+ emit_data->output[emit_data->chan] = LLVMBuildOr(builder,
+ div_mask,
+ result, "");
+}
+
+/* TGSI_OPCODE_USET Helper (CPU Only) */
+static void
+uset_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data,
+ unsigned pipe_func)
+{
+ LLVMValueRef cond = lp_build_cmp(&bld_base->uint_bld, pipe_func,
+ emit_data->args[0], emit_data->args[1]);
+ emit_data->output[emit_data->chan] = cond;
+}
+
+
+/* TGSI_OPCODE_USEQ (CPU Only) */
+static void
+useq_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ uset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_EQUAL);
+}
+
+/* TGSI_OPCODE_ISGE (CPU Only) */
+static void
+usge_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ uset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_GEQUAL);
+}
+
+/* TGSI_OPCODE_USHR (CPU Only) */
+static void
+ushr_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ struct lp_build_context *uint_bld = &bld_base->uint_bld;
+ LLVMValueRef mask = lp_build_const_vec(uint_bld->gallivm, uint_bld->type,
+ uint_bld->type.width - 1);
+ LLVMValueRef masked_count = lp_build_and(uint_bld, emit_data->args[1], mask);
+ emit_data->output[emit_data->chan] = lp_build_shr(uint_bld, emit_data->args[0],
+ masked_count);
+}
+
+/* TGSI_OPCODE_ISLT (CPU Only) */
+static void
+uslt_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ uset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_LESS);
+}
+
+/* TGSI_OPCODE_USNE (CPU Only) */
+
+static void
+usne_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ uset_emit_cpu(action, bld_base, emit_data, PIPE_FUNC_NOTEQUAL);
+}
+
+/* TGSI_OPCODE_XOR */
+static void
+xor_emit_cpu(
+ const struct lp_build_tgsi_action * action,
+ struct lp_build_tgsi_context * bld_base,
+ struct lp_build_emit_data * emit_data)
+{
+ emit_data->output[emit_data->chan] = lp_build_xor(&bld_base->uint_bld,
+ emit_data->args[0],
+ emit_data->args[1]);
+}
+
void
lp_set_default_actions_cpu(
struct lp_build_tgsi_context * bld_base)
lp_set_default_actions(bld_base);
bld_base->op_actions[TGSI_OPCODE_ABS].emit = abs_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_ADD].emit = add_emit_cpu;
- bld_base->op_actions[TGSI_OPCODE_ARL].emit = flr_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_AND].emit = and_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_ARL].emit = arl_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_ARR].emit = arr_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_CEIL].emit = ceil_emit_cpu;
- bld_base->op_actions[TGSI_OPCODE_CND].emit = cnd_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_COS].emit = cos_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_CMP].emit = cmp_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_DIV].emit = div_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_EX2].emit = ex2_emit_cpu;
- bld_base->op_actions[TGSI_OPCODE_EXP].emit = exp_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_F2I].emit = f2i_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_FLR].emit = flr_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_FSEQ].emit = fseq_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_FSGE].emit = fsge_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_FSLT].emit = fslt_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_FSNE].emit = fsne_emit_cpu;
+
+ bld_base->op_actions[TGSI_OPCODE_I2F].emit = i2f_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_IABS].emit = iabs_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_IDIV].emit = idiv_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_INEG].emit = ineg_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_IMAX].emit = imax_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_IMIN].emit = imin_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_ISGE].emit = isge_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_ISHR].emit = ishr_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_ISLT].emit = islt_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_ISSG].emit = issg_emit_cpu;
+
bld_base->op_actions[TGSI_OPCODE_LG2].emit = lg2_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_LOG].emit = log_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_MAX].emit = max_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_MIN].emit = min_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_MOD].emit = mod_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_NOT].emit = not_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_OR].emit = or_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_POW].emit = pow_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_RCP].emit = rcp_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_ROUND].emit = round_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_SGE].emit = sge_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_SGT].emit = sgt_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_SIN].emit = sin_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_SHL].emit = shl_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_SLE].emit = sle_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_SLT].emit = slt_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_SNE].emit = sne_emit_cpu;
bld_base->op_actions[TGSI_OPCODE_TRUNC].emit = trunc_emit_cpu;
bld_base->rsq_action.emit = recip_sqrt_emit_cpu;
+ bld_base->sqrt_action.emit = sqrt_emit_cpu;
+
+ bld_base->op_actions[TGSI_OPCODE_UADD].emit = uadd_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_UCMP].emit = ucmp_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_UDIV].emit = udiv_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_UMAX].emit = umax_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_UMIN].emit = umin_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_UMOD].emit = umod_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_USEQ].emit = useq_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_USGE].emit = usge_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_USHR].emit = ushr_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_USLT].emit = uslt_emit_cpu;
+ bld_base->op_actions[TGSI_OPCODE_USNE].emit = usne_emit_cpu;
+
+ bld_base->op_actions[TGSI_OPCODE_XOR].emit = xor_emit_cpu;
+
}