LLVM: llvm::ISD Namespace Reference (original) (raw)

ISD::NodeType enum - This enum defines the target-independent operators for a SelectionDAG.

Targets may also define target-dependent operator codes for SDNodes. For example, on x86, these are the enum values in the X86ISD namespace. Targets should aim to use target-independent operators to model their instruction sets as much as possible, and only use target-dependent operators when they have special requirements.

Finally, during and after selection proper, SNodes may use special operator codes that correspond directly with MachineInstr opcodes. These are used to represent selected instructions. See the isMachineOpcode() and getMachineOpcode() member functions of SDNode.

Enumerator

DELETED_NODE

DELETED_NODE - This is an illegal value that is used to catch errors.

This opcode is not a legal opcode for any node.

EntryToken

EntryToken - This is the marker used to indicate the start of a region.

TokenFactor

TokenFactor - This node takes multiple tokens as input and produces a single token result.

This is used to represent the fact that the operand operators are independent of each other.

AssertSext

AssertSext, AssertZext - These nodes record if a register contains a value that has already been zero or sign extended from a narrower type.

These nodes take two operands. The first is the node that has already been extended, and the second is a value type node indicating the width of the extension. NOTE: In case of the source value (or any vector element value) is poisoned the assertion will not be true for that value.

AssertZext

AssertAlign

AssertAlign - These nodes record if a register contains a value that has a known alignment and the trailing bits are known to be zero.

NOTE: In case of the source value (or any vector element value) is poisoned the assertion will not be true for that value.

AssertNoFPClass

AssertNoFPClass - These nodes record if a register contains a float value that is known to be not some type.

This node takes two operands. The first is the node that is known never to be some float types; the second is a constant value with the value of FPClassTest (casted to uint32_t). NOTE: In case of the source value (or any vector element value) is poisoned the assertion will not be true for that value.

BasicBlock

Various leaf nodes.

VALUETYPE

CONDCODE

Register

RegisterMask

Constant

ConstantFP

GlobalAddress

GlobalTLSAddress

FrameIndex

JumpTable

ConstantPool

ExternalSymbol

BlockAddress

PtrAuthGlobalAddress

A ptrauth constant.

ptr, key, addr-disc, disc Note that the addr-disc can be a non-constant value, to allow representing a constant global address signed using address-diversification, in code.

GLOBAL_OFFSET_TABLE

The address of the GOT.

FRAMEADDR

FRAMEADDR, RETURNADDR - These nodes represent llvm.frameaddress and llvm.returnaddress on the DAG.

These nodes take one operand, the index of the frame or return address to return. An index of zero corresponds to the current function's frame or return address, an index of one to the parent's frame or return address, and so on.

RETURNADDR

ADDROFRETURNADDR

ADDROFRETURNADDR - Represents the llvm.addressofreturnaddress intrinsic.

This node takes no operand, returns a target-specific pointer to the place in the stack frame where the return address of the current function is stored.

SPONENTRY

SPONENTRY - Represents the llvm.sponentry intrinsic.

Takes no argument and returns the stack pointer value at the entry of the current function calling this intrinsic.

LOCAL_RECOVER

LOCAL_RECOVER - Represents the llvm.localrecover intrinsic.

Materializes the offset from the local object pointer of another function to a particular local object passed to llvm.localescape. The operand is the MCSymbol label used to represent this offset, since typically the offset is not known until after code generation of the parent.

READ_REGISTER

READ_REGISTER, WRITE_REGISTER - This node represents llvm.register on the DAG, which implements the named register global variables extension.

WRITE_REGISTER

FRAME_TO_ARGS_OFFSET

FRAME_TO_ARGS_OFFSET - This node represents offset from frame pointer to first (possible) on-stack argument.

This is needed for correct stack adjustment during unwind.

EH_DWARF_CFA

EH_DWARF_CFA - This node represents the pointer to the DWARF Canonical Frame Address (CFA), generally the value of the stack pointer at the call site in the previous frame.

EH_RETURN

OUTCHAIN = EH_RETURN(INCHAIN, OFFSET, HANDLER) - This node represents 'eh_return' gcc dwarf builtin, which is used to return from exception.

The general meaning is: adjust stack by OFFSET and pass execution to HANDLER. Many platform-related details also :)

EH_SJLJ_SETJMP

RESULT, OUTCHAIN = EH_SJLJ_SETJMP(INCHAIN, buffer) This corresponds to the eh.sjlj.setjmp intrinsic.

It takes an input chain and a pointer to the jump buffer as inputs and returns an outchain.

EH_SJLJ_LONGJMP

OUTCHAIN = EH_SJLJ_LONGJMP(INCHAIN, buffer) This corresponds to the eh.sjlj.longjmp intrinsic.

It takes an input chain and a pointer to the jump buffer as inputs and returns an outchain.

EH_SJLJ_SETUP_DISPATCH

OUTCHAIN = EH_SJLJ_SETUP_DISPATCH(INCHAIN) The target initializes the dispatch table here.

TargetConstant

TargetConstant* - Like Constant*, but the DAG does not do any folding, simplification, or lowering of the constant.

They are used for constants which are known to fit in the immediate fields of their users, or for carrying magic numbers which are not values which need to be materialized in registers.

TargetConstantFP

TargetGlobalAddress

TargetGlobalAddress - Like GlobalAddress, but the DAG does no folding or anything else with this node, and this is valid in the target-specific dag, turning into a GlobalAddress operand.

TargetGlobalTLSAddress

TargetFrameIndex

TargetJumpTable

TargetConstantPool

TargetExternalSymbol

TargetBlockAddress

MCSymbol

TargetIndex

TargetIndex - Like a constant pool entry, but with completely target-dependent semantics.

Holds target flags, a 32-bit index, and a 64-bit index. Targets can use this however they like.

INTRINSIC_WO_CHAIN

RESULT = INTRINSIC_WO_CHAIN(INTRINSICID, arg1, arg2, ...) This node represents a target intrinsic function with no side effects.

The first operand is the ID number of the intrinsic from the llvm::Intrinsic namespace. The operands to the intrinsic follow. The node returns the result of the intrinsic.

INTRINSIC_W_CHAIN

RESULT,OUTCHAIN = INTRINSIC_W_CHAIN(INCHAIN, INTRINSICID, arg1, ...) This node represents a target intrinsic function with side effects that returns a result.

The first operand is a chain pointer. The second is the ID number of the intrinsic from the llvm::Intrinsic namespace. The operands to the intrinsic follow. The node has two results, the result of the intrinsic and an output chain.

INTRINSIC_VOID

OUTCHAIN = INTRINSIC_VOID(INCHAIN, INTRINSICID, arg1, arg2, ...) This node represents a target intrinsic function with side effects that does not return a result.

The first operand is a chain pointer. The second is the ID number of the intrinsic from the llvm::Intrinsic namespace. The operands to the intrinsic follow.

CopyToReg

CopyToReg - This node has three operands: a chain, a register number to set to this value, and a value.

CopyFromReg

CopyFromReg - This node indicates that the input value is a virtual or physical register that is defined outside of the scope of this SelectionDAG.

The register is available from the RegisterSDNode object. Note that CopyFromReg is considered as also freezing the value.

UNDEF

UNDEF - An undefined node.

POISON

POISON - A poison node.

FREEZE

FREEZE - FREEZE(VAL) returns an arbitrary value if VAL is UNDEF (or is evaluated to UNDEF), or returns VAL otherwise.

Note that each read of UNDEF can yield different value, but FREEZE(UNDEF) cannot.

EXTRACT_ELEMENT

EXTRACT_ELEMENT - This is used to get the lower or upper (determined by a Constant, which is required to be operand #1) half of the integer or float value specified as operand #0.

This is only for use before legalization, for values that will be broken into multiple registers.

BUILD_PAIR

BUILD_PAIR - This is the opposite of EXTRACT_ELEMENT in some ways.

Given two values of the same integer value type, this produces a value twice as big. Like EXTRACT_ELEMENT, this can only be used before legalization. The lower part of the composite value should be in element 0 and the upper part should be in element 1.

MERGE_VALUES

MERGE_VALUES - This node takes multiple discrete operands and returns them all as its individual results.

This nodes has exactly the same number of inputs and outputs. This node is useful for some pieces of the code generator that want to think about a single node with multiple results, not multiple nodes.

ADD

Simple integer binary arithmetic operators.

SUB

MUL

SDIV

UDIV

SREM

UREM

SMUL_LOHI

SMUL_LOHI/UMUL_LOHI - Multiply two integers of type iN, producing a signed/unsigned value of type i[2*N], and return the full value as two results, each of type iN.

UMUL_LOHI

SDIVREM

SDIVREM/UDIVREM - Divide two integers and produce both a quotient and remainder result.

UDIVREM

CARRY_FALSE

CARRY_FALSE - This node is used when folding other nodes, like ADDC/SUBC, which indicate the carry result is always false.

ADDC

Carry-setting nodes for multiple precision addition and subtraction.

These nodes take two operands of the same value type, and produce two results. The first result is the normal add or sub result, the second result is the carry flag result. FIXME: These nodes are deprecated in favor of UADDO_CARRY and USUBO_CARRY. They are kept around for now to provide a smooth transition path toward the use of UADDO_CARRY/USUBO_CARRY and will eventually be removed.

SUBC

ADDE

Carry-using nodes for multiple precision addition and subtraction.

These nodes take three operands: The first two are the normal lhs and rhs to the add or sub, and the third is the input carry flag. These nodes produce two results; the normal result of the add or sub, and the output carry flag. These nodes both read and write a carry flag to allow them to them to be chained together for add and sub of arbitrarily large values.

SUBE

UADDO_CARRY

Carry-using nodes for multiple precision addition and subtraction.

These nodes take three operands: The first two are the normal lhs and rhs to the add or sub, and the third is a boolean value that is 1 if and only if there is an incoming carry/borrow. These nodes produce two results: the normal result of the add or sub, and a boolean value that is 1 if and only if there is an outgoing carry/borrow.

Care must be taken if these opcodes are lowered to hardware instructions that use the inverse logic – 0 if and only if there is an incoming/outgoing carry/borrow. In such cases, you must preserve the semantics of these opcodes by inverting the incoming carry/borrow, feeding it to the add/sub hardware instruction, and then inverting the outgoing carry/borrow.

The use of these opcodes is preferable to ADDE/SUBE if the target supports it, as the carry is a regular value rather than a glue, which allows further optimisation.

These opcodes are different from [US]{ADD,SUB}O in that U{ADD,SUB}O_CARRY consume and produce a carry/borrow, whereas [US]{ADD,SUB}O produce an overflow.

USUBO_CARRY

SADDO_CARRY

Carry-using overflow-aware nodes for multiple precision addition and subtraction.

These nodes take three operands: The first two are normal lhs and rhs to the add or sub, and the third is a boolean indicating if there is an incoming carry. They produce two results: the normal result of the add or sub, and a boolean that indicates if an overflow occurred (not flag, because it may be a store to memory, etc.). If the type of the boolean is not i1 then the high bits conform to getBooleanContents.

SSUBO_CARRY

SADDO

RESULT, BOOL = [SU]ADDO(LHS, RHS) - Overflow-aware nodes for addition.

These nodes take two operands: the normal LHS and RHS to the add. They produce two results: the normal result of the add, and a boolean that indicates if an overflow occurred (not a flag, because it may be store to memory, etc.). If the type of the boolean is not i1 then the high bits conform to getBooleanContents. These nodes are generated from llvm.[su]add.with.overflow intrinsics.

UADDO

SSUBO

Same for subtraction.

USUBO

SMULO

Same for multiplication.

UMULO

SADDSAT

RESULT = [US]ADDSAT(LHS, RHS) - Perform saturation addition on 2 integers with the same bit width (W).

If the true value of LHS + RHS exceeds the largest value that can be represented by W bits, the resulting value is this maximum value. Otherwise, if this value is less than the smallest value that can be represented by W bits, the resulting value is this minimum value.

UADDSAT

SSUBSAT

RESULT = [US]SUBSAT(LHS, RHS) - Perform saturation subtraction on 2 integers with the same bit width (W).

If the true value of LHS - RHS exceeds the largest value that can be represented by W bits, the resulting value is this maximum value. Otherwise, if this value is less than the smallest value that can be represented by W bits, the resulting value is this minimum value.

USUBSAT

SSHLSAT

RESULT = [US]SHLSAT(LHS, RHS) - Perform saturation left shift.

The first operand is the value to be shifted, and the second argument is the amount to shift by. Both must be integers of the same bit width (W). If the true value of LHS << RHS exceeds the largest value that can be represented by W bits, the resulting value is this maximum value, Otherwise, if this value is less than the smallest value that can be represented by W bits, the resulting value is this minimum value.

USHLSAT

SMULFIX

RESULT = [US]MULFIX(LHS, RHS, SCALE) - Perform fixed point multiplication on 2 integers with the same width and scale.

SCALE represents the scale of both operands as fixed point numbers. This SCALE parameter must be a constant integer. A scale of zero is effectively performing multiplication on 2 integers.

UMULFIX

SMULFIXSAT

Same as the corresponding unsaturated fixed point instructions, but the result is clamped between the min and max values representable by the bits of the first 2 operands.

UMULFIXSAT

SDIVFIX

RESULT = [US]DIVFIX(LHS, RHS, SCALE) - Perform fixed point division on 2 integers with the same width and scale.

SCALE represents the scale of both operands as fixed point numbers. This SCALE parameter must be a constant integer.

UDIVFIX

SDIVFIXSAT

Same as the corresponding unsaturated fixed point instructions, but the result is clamped between the min and max values representable by the bits of the first 2 operands.

UDIVFIXSAT

FADD

Simple binary floating point operators.

FSUB

FMUL

FDIV

FREM

STRICT_FADD

Constrained versions of the binary floating point operators.

These will be lowered to the simple operators before final selection. They are used to limit optimizations while the DAG is being optimized.

STRICT_FSUB

STRICT_FMUL

STRICT_FDIV

STRICT_FREM

STRICT_FMA

STRICT_FSQRT

Constrained versions of libm-equivalent floating point intrinsics.

These will be lowered to the equivalent non-constrained pseudo-op (or expanded to the equivalent library call) before final selection. They are used to limit optimizations while the DAG is being optimized.

STRICT_FPOW

STRICT_FPOWI

STRICT_FLDEXP

STRICT_FSIN

STRICT_FCOS

STRICT_FTAN

STRICT_FASIN

STRICT_FACOS

STRICT_FATAN

STRICT_FATAN2

STRICT_FSINH

STRICT_FCOSH

STRICT_FTANH

STRICT_FEXP

STRICT_FEXP2

STRICT_FLOG

STRICT_FLOG10

STRICT_FLOG2

STRICT_FRINT

STRICT_FNEARBYINT

STRICT_FMAXNUM

STRICT_FMINNUM

STRICT_FCEIL

STRICT_FFLOOR

STRICT_FROUND

STRICT_FROUNDEVEN

STRICT_FTRUNC

STRICT_LROUND

STRICT_LLROUND

STRICT_LRINT

STRICT_LLRINT

STRICT_FMAXIMUM

STRICT_FMINIMUM

STRICT_FP_TO_SINT

STRICT_FP_TO_[US]INT - Convert a floating point value to a signed or unsigned integer.

These have the same semantics as fptosi and fptoui in IR. They are used to limit optimizations while the DAG is being optimized.

STRICT_FP_TO_UINT

STRICT_SINT_TO_FP

STRICT_[US]INT_TO_FP - Convert a signed or unsigned integer to a floating point value.

These have the same semantics as sitofp and uitofp in IR. They are used to limit optimizations while the DAG is being optimized.

STRICT_UINT_TO_FP

STRICT_FP_ROUND

X = STRICT_FP_ROUND(Y, TRUNC) - Rounding 'Y' from a larger floating point type down to the precision of the destination VT.

TRUNC is a flag, which is always an integer that is zero or one. If TRUNC is 0, this is a normal rounding, if it is 1, this FP_ROUND is known to not change the value of Y.

The TRUNC = 1 case is used in cases where we know that the value will not be modified by the node, because Y is not using any of the extra precision of source type. This allows certain transformations like STRICT_FP_EXTEND(STRICT_FP_ROUND(X,1)) -> X which are not safe for STRICT_FP_EXTEND(STRICT_FP_ROUND(X,0)) because the extra bits aren't removed. It is used to limit optimizations while the DAG is being optimized.

STRICT_FP_EXTEND

X = STRICT_FP_EXTEND(Y) - Extend a smaller FP type into a larger FP type.

It is used to limit optimizations while the DAG is being optimized.

STRICT_FSETCC

STRICT_FSETCC/STRICT_FSETCCS - Constrained versions of SETCC, used for floating-point operands only.

STRICT_FSETCC performs a quiet comparison operation, while STRICT_FSETCCS performs a signaling comparison operation.

STRICT_FSETCCS

FPTRUNC_ROUND

FPTRUNC_ROUND - This corresponds to the fptrunc_round intrinsic.

FMA

FMA - Perform a * b + c with no intermediate rounding step.

FMAD

FMAD - Perform a * b + c, while getting the same result as the separately rounded operations.

FMULADD

FMULADD - Performs a * b + c, with, or without, intermediate rounding.

It is expected that this will be illegal for most targets, as it usually makes sense to split this or use an FMA. But some targets, such as WebAssembly, can directly support these semantics.

FCOPYSIGN

FCOPYSIGN(X, Y) - Return the value of X with the sign of Y.

NOTE: This DAG node does not require that X and Y have the same type, just that they are both floating point. X and the result must have the same type. FCOPYSIGN(f32, f64) is allowed.

FGETSIGN

INT = FGETSIGN(FP) - Return the sign bit of the specified floating point value as an integer 0/1 value.

FCANONICALIZE

Returns platform specific canonical encoding of a floating point number.

IS_FPCLASS

Performs a check of floating point class property, defined by IEEE-754.

The first operand is the floating point value to check. The second operand specifies the checked property and is a TargetConstant which specifies test in the same way as intrinsic 'is_fpclass'. Returns boolean value.

BUILD_VECTOR

BUILD_VECTOR(ELT0, ELT1, ELT2, ELT3,...) - Return a fixed-width vector with the specified, possibly variable, elements.

The types of the operands must match the vector element type, except that integer types are allowed to be larger than the element type, in which case the operands are implicitly truncated. The types of the operands must all be the same.

INSERT_VECTOR_ELT

INSERT_VECTOR_ELT(VECTOR, VAL, IDX) - Returns VECTOR with the element at IDX replaced with VAL.

If the type of VAL is larger than the vector element type then VAL is truncated before replacement.

If VECTOR is a scalable vector, then IDX may be larger than the minimum vector width. IDX is not first scaled by the runtime scaling factor of VECTOR.

EXTRACT_VECTOR_ELT

EXTRACT_VECTOR_ELT(VECTOR, IDX) - Returns a single element from VECTOR identified by the (potentially variable) element number IDX.

If the return type is an integer type larger than the element type of the vector, the result is extended to the width of the return type. In that case, the high bits are undefined.

If VECTOR is a scalable vector, then IDX may be larger than the minimum vector width. IDX is not first scaled by the runtime scaling factor of VECTOR.

CONCAT_VECTORS

CONCAT_VECTORS(VECTOR0, VECTOR1, ...) - Given a number of values of vector type with the same length and element type, this produces a concatenated vector result value, with length equal to the sum of the lengths of the input vectors.

If VECTOR0 is a fixed-width vector, then VECTOR1..VECTORN must all be fixed-width vectors. Similarly, if VECTOR0 is a scalable vector, then VECTOR1..VECTORN must all be scalable vectors.

INSERT_SUBVECTOR

INSERT_SUBVECTOR(VECTOR1, VECTOR2, IDX) - Returns a vector with VECTOR2 inserted into VECTOR1.

IDX represents the starting element number at which VECTOR2 will be inserted. IDX must be a constant multiple of T's known minimum vector length. Let the type of VECTOR2 be T, then if T is a scalable vector, IDX is first scaled by the runtime scaling factor of T. The elements of VECTOR1 starting at IDX are overwritten with VECTOR2. Elements IDX through (IDX + num_elements(T) - 1) must be valid VECTOR1 indices. If this condition cannot be determined statically but is false at runtime, then the result vector is undefined. The IDX parameter must be a vector index constant type, which for most targets will be an integer pointer type.

This operation supports inserting a fixed-width vector into a scalable vector, but not the other way around.

EXTRACT_SUBVECTOR

EXTRACT_SUBVECTOR(VECTOR, IDX) - Returns a subvector from VECTOR.

Let the result type be T, then IDX represents the starting element number from which a subvector of type T is extracted. IDX must be a constant multiple of T's known minimum vector length. If T is a scalable vector, IDX is first scaled by the runtime scaling factor of T. Elements IDX through (IDX + num_elements(T) - 1) must be valid VECTOR indices. If this condition cannot be determined statically but is false at runtime, then the result vector is undefined. The IDX parameter must be a vector index constant type, which for most targets will be an integer pointer type.

This operation supports extracting a fixed-width vector from a scalable vector, but not the other way around.

VECTOR_DEINTERLEAVE

VECTOR_DEINTERLEAVE(VEC1, VEC2, ...) - Returns N vectors from N input vectors, where N is the factor to deinterleave.

All input and output vectors must have the same type.

Each output contains the deinterleaved indices for a specific field from CONCAT_VECTORS(VEC1, VEC2, ...):

Result[I][J] = CONCAT_VECTORS(...)[I + N * J]

VECTOR_INTERLEAVE

VECTOR_INTERLEAVE(VEC1, VEC2, ...) - Returns N vectors from N input vectors, where N is the factor to interleave.

All input and output vectors must have the same type.

All input vectors are interleaved into one wide vector, which is then chunked into equal sized parts:

Interleaved[I] = VEC(I % N)[I / N] Result[J] = EXTRACT_SUBVECTOR(Interleaved, J * getVectorMinNumElements())

VECTOR_REVERSE

VECTOR_REVERSE(VECTOR) - Returns a vector, of the same type as VECTOR, whose elements are shuffled using the following algorithm: RESULT[i] = VECTOR[VECTOR.ElementCount - 1 - i].

VECTOR_SHUFFLE

VECTOR_SHUFFLE(VEC1, VEC2) - Returns a vector, of the same type as VEC1/VEC2.

A VECTOR_SHUFFLE node also contains an array of constant int values that indicate which value (or undef) each result element will get. These constant ints are accessible through the ShuffleVectorSDNode class. This is quite similar to the Altivec 'vperm' instruction, except that the indices must be constants and are in terms of the element size of VEC1/VEC2, not in terms of bytes.

VECTOR_SPLICE

VECTOR_SPLICE(VEC1, VEC2, IMM) - Returns a subvector of the same type as VEC1/VEC2 from CONCAT_VECTORS(VEC1, VEC2), based on the IMM in two ways.

Let the result type be T, if IMM is positive it represents the starting element number (an index) from which a subvector of type T is extracted from CONCAT_VECTORS(VEC1, VEC2). If IMM is negative it represents a count specifying the number of trailing elements to extract from VEC1, where the elements of T are selected using the following algorithm: RESULT[i] = CONCAT_VECTORS(VEC1,VEC2)[VEC1.ElementCount - ABS(IMM) + i] If IMM is not in the range [-VL, VL-1] the result vector is undefined. IMM is a constant integer.

SCALAR_TO_VECTOR

SCALAR_TO_VECTOR(VAL) - This represents the operation of loading a scalar value into element 0 of the resultant vector type.

The top elements 1 to N-1 of the N-element vector are poison. The type of the operand must match the vector element type, except when they are integer types. In this case the operand is allowed to be wider than the vector element type, and is implicitly truncated to it.

SPLAT_VECTOR

SPLAT_VECTOR(VAL) - Returns a vector with the scalar value VAL duplicated in all lanes.

The type of the operand must match the vector element type, except when they are integer types. In this case the operand is allowed to be wider than the vector element type, and is implicitly truncated to it.

SPLAT_VECTOR_PARTS

SPLAT_VECTOR_PARTS(SCALAR1, SCALAR2, ...) - Returns a vector with the scalar values joined together and then duplicated in all lanes.

This represents a SPLAT_VECTOR that has had its scalar operand expanded. This allows representing a 64-bit splat on a target with 32-bit integers. The total width of the scalars must cover the element width. SCALAR1 contains the least significant bits of the value regardless of endianness and all scalars should have the same type.

STEP_VECTOR

STEP_VECTOR(IMM) - Returns a scalable vector whose lanes are comprised of a linear sequence of unsigned values starting from 0 with a step of IMM, where IMM must be a TargetConstant with type equal to the vector element type.

The arithmetic is performed modulo the bitwidth of the element.

The operation does not support returning fixed-width vectors or non-constant operands.

VECTOR_COMPRESS

VECTOR_COMPRESS(Vec, Mask, Passthru) consecutively place vector elements based on mask e.g., vec = {A, B, C, D} and mask = {1, 0, 1, 0} --> {A, C, ?

, ?} where ? is undefined If passthru is defined, ?s are replaced with elements from passthru. If passthru is undef, ?s remain undefined.

MULHU

MULHU/MULHS - Multiply high - Multiply two integers of type iN, producing an unsigned/signed value of type i[2*N], then return the top part.

MULHS

AVGFLOORS

AVGFLOORS/AVGFLOORU - Averaging add - Add two integers using an integer of type i[N+1], halving the result by shifting it one bit right.

shr(add(ext(X), ext(Y)), 1)

AVGFLOORU

AVGCEILS

AVGCEILS/AVGCEILU - Rounding averaging add - Add two integers using an integer of type i[N+2], add 1 and halve the result by shifting it one bit right.

shr(add(ext(X), ext(Y), 1), 1)

AVGCEILU

ABDS

ABDS/ABDU - Absolute difference - Return the absolute difference between two numbers interpreted as signed/unsigned.

i.e trunc(abs(sext(Op0) - sext(Op1))) becomes abds(Op0, Op1) or trunc(abs(zext(Op0) - zext(Op1))) becomes abdu(Op0, Op1)

ABDU

SMIN

[US]{MIN/MAX} - Binary minimum or maximum of signed or unsigned integers.

SMAX

UMIN

UMAX

SCMP

[US]CMP - 3-way comparison of signed or unsigned integers.

Returns -1, 0, or 1 depending on whether Op0 <, ==, or > Op1. The operands can have type different to the result.

UCMP

AND

Bitwise operators - logical and, logical or, logical xor.

OR

XOR

ABS

ABS - Determine the unsigned absolute value of a signed integer value of the same bitwidth.

Note: A value of INT_MIN will return INT_MIN, no saturation or overflow is performed.

SHL

Shift and rotation operations.

After legalization, the type of the shift amount is known to be TLI.getShiftAmountTy(). Before legalization the shift amount can be any type, but care must be taken to ensure it is large enough. TLI.getShiftAmountTy() is i8 on some targets, but before legalization, types like i1024 can occur and i8 doesn't have enough bits to represent the shift amount. When the 1st operand is a vector, the shift amount must be in the same type. (TLI.getShiftAmountTy() will return the same type when the input type is a vector.) For rotates and funnel shifts, the shift amount is treated as an unsigned amount modulo the element size of the first operand.

Funnel 'double' shifts take 3 operands, 2 inputs and the shift amount.

fshl(X,Y,Z): (X << (Z % BW)) | (Y >> (BW - (Z % BW))) fshr(X,Y,Z): (X << (BW - (Z % BW))) | (Y >> (Z % BW))

SRA

SRL

ROTL

ROTR

FSHL

FSHR

BSWAP

Byte Swap and Counting operators.

CTTZ

CTLZ

CTPOP

BITREVERSE

PARITY

CTTZ_ZERO_UNDEF

Bit counting operators with an undefined result for zero inputs.

CTLZ_ZERO_UNDEF

SELECT

Select(COND, TRUEVAL, FALSEVAL).

If the type of the boolean COND is not i1 then the high bits must conform to getBooleanContents.

VSELECT

Select with a vector condition (op #0) and two vector operands (ops #1 and #2), returning a vector result.

All vectors have the same length. Much like the scalar select and setcc, each bit in the condition selects whether the corresponding result element is taken from op #1 or op #2. At first, the VSELECT condition is of vXi1 type. Later, targets may change the condition type in order to match the VSELECT node using a pattern. The condition follows the BooleanContent format of the target.

SELECT_CC

Select with condition operator - This selects between a true value and a false value (ops #2 and #3) based on the boolean result of comparing the lhs and rhs (ops #0 and #1) of a conditional expression with the condition code in op #4, a CondCodeSDNode.

SETCC

SetCC operator - This evaluates to a true value iff the condition is true.

If the result value type is not i1 then the high bits conform to getBooleanContents. The operands to this are the left and right operands to compare (ops #0, and #1) and the condition code to compare them with (op #2) as a CondCodeSDNode. If the operands are vector types then the result type must also be a vector type.

SETCCCARRY

Like SetCC, ops #0 and #1 are the LHS and RHS operands to compare, but op #2 is a boolean indicating if there is an incoming carry.

This operator checks the result of "LHS - RHS - Carry", and can be used to compare two wide integers: (setcccarry lhshi rhshi (usubo_carry lhslo rhslo) cc). Only valid for integers.

SHL_PARTS

SHL_PARTS/SRA_PARTS/SRL_PARTS - These operators are used for expanded integer shift operations.

The operation ordering is:

[Lo,Hi] = op [LoLHS,HiLHS], Amt

SRA_PARTS

SRL_PARTS

SIGN_EXTEND

Conversion operators.

These are all single input single output operations. For all of these, the result type must be strictly wider or narrower (depending on the operation) than the source type. SIGN_EXTEND - Used for integer types, replicating the sign bit into new bits.

ZERO_EXTEND

ZERO_EXTEND - Used for integer types, zeroing the new bits.

Can carry the NonNeg SDNodeFlag to indicate that the input is known to be non-negative. If the flag is present and the input is negative, the result is poison.

ANY_EXTEND

ANY_EXTEND - Used for integer types. The high bits are undefined.

TRUNCATE

TRUNCATE - Completely drop the high bits.

TRUNCATE_SSAT_S

TRUNCATE_[SU]SAT_[SU] - Truncate for saturated operand [SU] located in middle, prefix for SAT means indicates whether existing truncate target was a signed operation.

For examples, If truncate(smin(smax(x, C), C)) was saturated then become S. If truncate(umin(x, C)) was saturated then become U. [SU] located in last indicates whether range of truncated values is sign-saturated. For example, if truncate(smin(smax(x, C), C)) is a truncation to i8, then if value of C ranges from -128 to 127, it will be saturated against signed values, resulting in S, which will combine to TRUNCATE_SSAT_S. If the value of C ranges from 0 to 255, it will be saturated against unsigned values, resulting in U, which will combine to TRUNCATE_SSAT_U. Similarly, in truncate(umin(x, C)), if value of C ranges from 0 to 255, it becomes U because it is saturated for unsigned values. As a result, it combines to TRUNCATE_USAT_U.

TRUNCATE_SSAT_U

TRUNCATE_USAT_U

SINT_TO_FP

[SU]INT_TO_FP - These operators convert integers (whose interpreted sign depends on the first letter) to floating point.

UINT_TO_FP

SIGN_EXTEND_INREG

SIGN_EXTEND_INREG - This operator atomically performs a SHL/SRA pair to sign extend a small value in a large integer register (e.g.

sign extending the low 8 bits of a 32-bit register to fill the top 24 bits with the 7th bit). The size of the smaller type is indicated by the 1th operand, a ValueType node.

ANY_EXTEND_VECTOR_INREG

ANY_EXTEND_VECTOR_INREG(Vector) - This operator represents an in-register any-extension of the low lanes of an integer vector.

The result type must have fewer elements than the operand type, and those elements must be larger integer types such that the total size of the operand type is less than or equal to the size of the result type. Each of the low operand elements is any-extended into the corresponding, wider result elements with the high bits becoming undef. NOTE: The type legalizer prefers to make the operand and result size the same to allow expansion to shuffle vector during op legalization.

SIGN_EXTEND_VECTOR_INREG

SIGN_EXTEND_VECTOR_INREG(Vector) - This operator represents an in-register sign-extension of the low lanes of an integer vector.

The result type must have fewer elements than the operand type, and those elements must be larger integer types such that the total size of the operand type is less than or equal to the size of the result type. Each of the low operand elements is sign-extended into the corresponding, wider result elements. NOTE: The type legalizer prefers to make the operand and result size the same to allow expansion to shuffle vector during op legalization.

ZERO_EXTEND_VECTOR_INREG

ZERO_EXTEND_VECTOR_INREG(Vector) - This operator represents an in-register zero-extension of the low lanes of an integer vector.

The result type must have fewer elements than the operand type, and those elements must be larger integer types such that the total size of the operand type is less than or equal to the size of the result type. Each of the low operand elements is zero-extended into the corresponding, wider result elements. NOTE: The type legalizer prefers to make the operand and result size the same to allow expansion to shuffle vector during op legalization.

FP_TO_SINT

FP_TO_[US]INT - Convert a floating point value to a signed or unsigned integer.

These have the same semantics as fptosi and fptoui in IR. If the FP value cannot fit in the integer type, the results are undefined.

FP_TO_UINT

FP_TO_SINT_SAT

FP_TO_[US]INT_SAT - Convert floating point value in operand 0 to a signed or unsigned scalar integer type given in operand 1 with the following semantics:

The scalar width of the type given in operand 1 must be equal to, or smaller than, the scalar result type width. It may end up being smaller than the result width as a result of integer type legalization.

After converting to the scalar integer type in operand 1, the value is extended to the result VT. FP_TO_SINT_SAT sign extends and FP_TO_UINT_SAT zero extends.

FP_TO_UINT_SAT

FP_ROUND

X = FP_ROUND(Y, TRUNC) - Rounding 'Y' from a larger floating point type down to the precision of the destination VT.

TRUNC is a flag, which is always an integer that is zero or one. If TRUNC is 0, this is a normal rounding, if it is 1, this FP_ROUND is known to not change the value of Y.

The TRUNC = 1 case is used in cases where we know that the value will not be modified by the node, because Y is not using any of the extra precision of source type. This allows certain transformations like FP_EXTEND(FP_ROUND(X,1)) -> X which are not safe for FP_EXTEND(FP_ROUND(X,0)) because the extra bits aren't removed.

GET_ROUNDING

Returns current rounding mode: -1 Undefined 0 Round to 0 1 Round to nearest, ties to even 2 Round to +inf 3 Round to -inf 4 Round to nearest, ties to zero Other values are target dependent.

Result is rounding mode and chain. Input is a chain.

LOOP_DEPENDENCE_WAR_MASK

Set rounding mode.

The first operand is a chain pointer. The second specifies the required rounding mode, encoded in the same way as used in 'GET_ROUNDING'. SET_ROUNDING,

/ X = FP_EXTEND(Y) - Extend a smaller FP type into a larger FP type. FP_EXTEND,

/ BITCAST - This operator converts between integer, vector and FP / values, as if the value was stored to memory with one type and loaded / from the same address with the other type (or equivalently for vector / format conversions, etc). The source and result are required to have / the same bit size (e.g. f32 <-> i32). This can also be used for / int-to-int or fp-to-fp conversions, but that is a noop, deleted by / getNode(). / / This operator is subtly different from the bitcast instruction from / LLVM-IR since this node may change the bits in the register. For / example, this occurs on big-endian NEON and big-endian MSA where the / layout of the bits in the register depends on the vector type and this / operator acts as a shuffle operation for some vector type combinations. BITCAST,

/ ADDRSPACECAST - This operator converts between pointers of different / address spaces. ADDRSPACECAST,

/ FP16_TO_FP, FP_TO_FP16 - These operators are used to perform promotions / and truncation for half-precision (16 bit) floating numbers. These nodes / form a semi-softened interface for dealing with f16 (as an i16), which / is often a storage-only type but has native conversions. FP16_TO_FP, FP_TO_FP16, STRICT_FP16_TO_FP, STRICT_FP_TO_FP16,

/ BF16_TO_FP, FP_TO_BF16 - These operators are used to perform promotions / and truncation for bfloat16. These nodes form a semi-softened interface / for dealing with bf16 (as an i16), which is often a storage-only type but / has native conversions. BF16_TO_FP, FP_TO_BF16, STRICT_BF16_TO_FP, STRICT_FP_TO_BF16,

/ Perform various unary floating-point operations inspired by libm. For / FPOWI, the result is undefined if the integer operand doesn't fit into / sizeof(int). FNEG, FABS, FSQRT, FCBRT, FSIN, FCOS, FTAN, FASIN, FACOS, FATAN, FSINH, FCOSH, FTANH, FPOW, FPOWI, / FLDEXP - ldexp, inspired by libm (op0 * 2**op1). FLDEXP, / FATAN2 - atan2, inspired by libm. FATAN2,

/ FFREXP - frexp, extract fractional and exponent component of a / floating-point value. Returns the two components as separate return / values. FFREXP,

FLOG, FLOG2, FLOG10, FEXP, FEXP2, FEXP10, FCEIL, FTRUNC, FRINT, FNEARBYINT, FROUND, FROUNDEVEN, FFLOOR, LROUND, LLROUND, LRINT, LLRINT,

/ FMINNUM/FMAXNUM - Perform floating-point minimum maximum on two values, / following IEEE-754 definitions except for signed zero behavior. / / If one input is a signaling NaN, returns a quiet NaN. This matches / IEEE-754 2008's minNum/maxNum behavior for signaling NaNs (which differs / from 2019). / / These treat -0 as ordered less than +0, matching the behavior of IEEE-754 / 2019's minimumNumber/maximumNumber. / / Note that that arithmetic on an sNaN doesn't consistently produce a qNaN, / so arithmetic feeding into a minnum/maxnum can produce inconsistent / results. FMAXIMUN/FMINIMUM or FMAXIMUMNUM/FMINIMUMNUM may be better choice / for non-distinction of sNaN/qNaN handling. FMINNUM, FMAXNUM,

/ FMINNUM_IEEE/FMAXNUM_IEEE - Perform floating-point minimumNumber or / maximumNumber on two values, following IEEE-754 definitions. This differs / from FMINNUM/FMAXNUM in the handling of signaling NaNs, and signed zero. / / If one input is a signaling NaN, returns a quiet NaN. This matches / IEEE-754 2008's minnum/maxnum behavior for signaling NaNs (which differs / from 2019). / / These treat -0 as ordered less than +0, matching the behavior of IEEE-754 / 2019's minimumNumber/maximumNumber. / / Deprecated, and will be removed soon, as FMINNUM/FMAXNUM have the same / semantics now. FMINNUM_IEEE, FMAXNUM_IEEE,

/ FMINIMUM/FMAXIMUM - NaN-propagating minimum/maximum that also treat -0.0 / as less than 0.0. While FMINNUM_IEEE/FMAXNUM_IEEE follow IEEE 754-2008 / semantics, FMINIMUM/FMAXIMUM follow IEEE 754-2019 semantics. FMINIMUM, FMAXIMUM,

/ FMINIMUMNUM/FMAXIMUMNUM - minimumnum/maximumnum that is same with / FMINNUM_IEEE and FMAXNUM_IEEE besides if either operand is sNaN. FMINIMUMNUM, FMAXIMUMNUM,

/ FSINCOS - Compute both fsin and fcos as a single operation. FSINCOS,

/ FSINCOSPI - Compute both the sine and cosine times pi more accurately / than FSINCOS(pi*x), especially for large x. FSINCOSPI,

/ FMODF - Decomposes the operand into integral and fractional parts, each / having the same type and sign as the operand. FMODF,

/ Gets the current floating-point environment. The first operand is a token / chain. The results are FP environment, represented by an integer value, / and a token chain. GET_FPENV,

/ Sets the current floating-point environment. The first operand is a token / chain, the second is FP environment, represented by an integer value. The / result is a token chain. SET_FPENV,

/ Set floating-point environment to default state. The first operand and the / result are token chains. RESET_FPENV,

/ Gets the current floating-point environment. The first operand is a token / chain, the second is a pointer to memory, where FP environment is stored / to. The result is a token chain. GET_FPENV_MEM,

/ Sets the current floating point environment. The first operand is a token / chain, the second is a pointer to memory, where FP environment is loaded / from. The result is a token chain. SET_FPENV_MEM,

/ Reads the current dynamic floating-point control modes. The operand is / a token chain. GET_FPMODE,

/ Sets the current dynamic floating-point control modes. The first operand / is a token chain, the second is control modes set represented as integer / value. SET_FPMODE,

/ Sets default dynamic floating-point control modes. The operand is a / token chain. RESET_FPMODE,

/ LOAD and STORE have token chains as their first operand, then the same / operands as an LLVM load/store instruction, then an offset node that / is added / subtracted from the base pointer to form the address (for / indexed memory ops). LOAD, STORE,

/ DYNAMIC_STACKALLOC - Allocate some number of bytes on the stack aligned / to a specified boundary. This node always has two return values: a new / stack pointer value and a chain. The first operand is the token chain, / the second is the number of bytes to allocate, and the third is the / alignment boundary. The size is guaranteed to be a multiple of the / stack alignment, and the alignment is guaranteed to be bigger than the / stack alignment (if required) or 0 to get standard stack alignment. DYNAMIC_STACKALLOC,

/ Control flow instructions. These all have token chains.

/ BR - Unconditional branch. The first operand is the chain / operand, the second is the MBB to branch to. BR,

/ BRIND - Indirect branch. The first operand is the chain, the second / is the value to branch to, which must be of the same type as the / target's pointer type. BRIND,

/ BR_JT - Jumptable branch. The first operand is the chain, the second / is the jumptable index, the last one is the jumptable entry index. BR_JT,

/ JUMP_TABLE_DEBUG_INFO - Jumptable debug info. The first operand is the / chain, the second is the jumptable index. JUMP_TABLE_DEBUG_INFO,

/ BRCOND - Conditional branch. The first operand is the chain, the / second is the condition, the third is the block to branch to if the / condition is true. If the type of the condition is not i1, then the / high bits must conform to getBooleanContents. If the condition is undef, / it nondeterministically jumps to the block. / TODO: Its semantics w.r.t undef requires further discussion; we need to / make it sure that it is consistent with optimizations in MIR & the / meaning of IMPLICIT_DEF. See https://reviews.llvm.org/D92015 BRCOND,

/ BR_CC - Conditional branch. The behavior is like that of SELECT_CC, in / that the condition is represented as condition code, and two nodes to / compare, rather than as a combined SetCC node. The operands in order / are chain, cc, lhs, rhs, block to branch to if condition is true. If / condition is undef, it nondeterministically jumps to the block. BR_CC,

/ INLINEASM - Represents an inline asm block. This node always has two / return values: a chain and a flag result. The inputs are as follows: / Operand #0 : Input chain. / Operand #1 : a ExternalSymbolSDNode with a pointer to the asm string. / Operand #2 : a MDNodeSDNode with the !srcloc metadata. / Operand #3 : HasSideEffect, IsAlignStack bits. / After this, it is followed by a list of operands with this format: / ConstantSDNode: Flags that encode whether it is a mem or not, the / of operands that follow, etc. See InlineAsm.h. / ... however many operands ... / Operand #last: Optional, an incoming flag. / / The variable width operands are required to represent target addressing / modes as a single "operand", even though they may have multiple / SDOperands. INLINEASM,

/ INLINEASM_BR - Branching version of inline asm. Used by asm-goto. INLINEASM_BR,

/ EH_LABEL - Represents a label in mid basic block used to track / locations needed for debug and exception handling tables. These nodes / take a chain as input and return a chain. EH_LABEL,

/ ANNOTATION_LABEL - Represents a mid basic block label used by / annotations. This should remain within the basic block and be ordered / with respect to other call instructions, but loads and stores may float / past it. ANNOTATION_LABEL,

/ CATCHRET - Represents a return from a catch block funclet. Used for / MSVC compatible exception handling. Takes a chain operand and a / destination basic block operand. CATCHRET,

/ CLEANUPRET - Represents a return from a cleanup block funclet. Used for / MSVC compatible exception handling. Takes only a chain operand. CLEANUPRET,

/ STACKSAVE - STACKSAVE has one operand, an input chain. It produces a / value, the same type as the pointer type for the system, and an output / chain. STACKSAVE,

/ STACKRESTORE has two operands, an input chain and a pointer to restore / to it returns an output chain. STACKRESTORE,

/ CALLSEQ_START/CALLSEQ_END - These operators mark the beginning and end / of a call sequence, and carry arbitrary information that target might / want to know. The first operand is a chain, the rest are specified by / the target and not touched by the DAG optimizers. / Targets that may use stack to pass call arguments define additional / operands: / - size of the call frame part that must be set up within the / CALLSEQ_START..CALLSEQ_END pair, / - part of the call frame prepared prior to CALLSEQ_START. / Both these parameters must be constants, their sum is the total call / frame size. / CALLSEQ_START..CALLSEQ_END pairs may not be nested. CALLSEQ_START, // Beginning of a call sequence CALLSEQ_END, // End of a call sequence

/ VAARG - VAARG has four operands: an input chain, a pointer, a SRCVALUE, / and the alignment. It returns a pair of values: the vaarg value and a / new chain. VAARG,

/ VACOPY - VACOPY has 5 operands: an input chain, a destination pointer, / a source pointer, a SRCVALUE for the destination, and a SRCVALUE for the / source. VACOPY,

/ VAEND, VASTART - VAEND and VASTART have three operands: an input chain, / pointer, and a SRCVALUE. VAEND, VASTART,

/ PREALLOCATED_SETUP - This has 2 operands: an input chain and a SRCVALUE / with the preallocated call Value. PREALLOCATED_SETUP, / PREALLOCATED_ARG - This has 3 operands: an input chain, a SRCVALUE / with the preallocated call Value, and a constant int. PREALLOCATED_ARG,

/ SRCVALUE - This is a node type that holds a Value* that is used to / make reference to a value in the LLVM IR. SRCVALUE,

/ MDNODE_SDNODE - This is a node that holdes an MDNode*, which is used to / reference metadata in the IR. MDNODE_SDNODE,

/ PCMARKER - This corresponds to the pcmarker intrinsic. PCMARKER,

/ READCYCLECOUNTER - This corresponds to the readcyclecounter intrinsic. / It produces a chain and one i64 value. The only operand is a chain. / If i64 is not legal, the result will be expanded into smaller values. / Still, it returns an i64, so targets should set legality for i64. / The result is the content of the architecture-specific cycle / counter-like register (or other high accuracy low latency clock source). READCYCLECOUNTER,

/ READSTEADYCOUNTER - This corresponds to the readfixedcounter intrinsic. / It has the same semantics as the READCYCLECOUNTER implementation except / that the result is the content of the architecture-specific fixed / frequency counter suitable for measuring elapsed time. READSTEADYCOUNTER,

/ HANDLENODE node - Used as a handle for various purposes. HANDLENODE,

/ INIT_TRAMPOLINE - This corresponds to the init_trampoline intrinsic. It / takes as input a token chain, the pointer to the trampoline, the pointer / to the nested function, the pointer to pass for the 'nest' parameter, a / SRCVALUE for the trampoline and another for the nested function / (allowing targets to access the original Function*). / It produces a token chain as output. INIT_TRAMPOLINE,

/ ADJUST_TRAMPOLINE - This corresponds to the adjust_trampoline intrinsic. / It takes a pointer to the trampoline and produces a (possibly) new / pointer to the same trampoline with platform-specific adjustments / applied. The pointer it returns points to an executable block of code. ADJUST_TRAMPOLINE,

/ TRAP - Trapping instruction TRAP,

/ DEBUGTRAP - Trap intended to get the attention of a debugger. DEBUGTRAP,

/ UBSANTRAP - Trap with an immediate describing the kind of sanitizer / failure. UBSANTRAP,

/ PREFETCH - This corresponds to a prefetch intrinsic. The first operand / is the chain. The other operands are the address to prefetch, / read / write specifier, locality specifier and instruction / data cache / specifier. PREFETCH,

/ ARITH_FENCE - This corresponds to a arithmetic fence intrinsic. Both its / operand and output are the same floating type. ARITH_FENCE,

/ MEMBARRIER - Compiler barrier only; generate a no-op. MEMBARRIER,

/ OUTCHAIN = ATOMIC_FENCE(INCHAIN, ordering, scope) / This corresponds to the fence instruction. It takes an input chain, and / two integer constants: an AtomicOrdering and a SynchronizationScope. ATOMIC_FENCE,

/ Val, OUTCHAIN = ATOMIC_LOAD(INCHAIN, ptr) / This corresponds to "load atomic" instruction. ATOMIC_LOAD,

/ OUTCHAIN = ATOMIC_STORE(INCHAIN, val, ptr) / This corresponds to "store atomic" instruction. ATOMIC_STORE,

/ Val, OUTCHAIN = ATOMIC_CMP_SWAP(INCHAIN, ptr, cmp, swap) / For double-word atomic operations: / ValLo, ValHi, OUTCHAIN = ATOMIC_CMP_SWAP(INCHAIN, ptr, cmpLo, cmpHi, / swapLo, swapHi) / This corresponds to the cmpxchg instruction. ATOMIC_CMP_SWAP,

/ Val, Success, OUTCHAIN / = ATOMIC_CMP_SWAP_WITH_SUCCESS(INCHAIN, ptr, cmp, swap) / N.b. this is still a strong cmpxchg operation, so / Success == "Val == cmp". ATOMIC_CMP_SWAP_WITH_SUCCESS,

/ Val, OUTCHAIN = ATOMIC_SWAP(INCHAIN, ptr, amt) / Val, OUTCHAIN = ATOMIC_LOAD_[OpName](INCHAIN, ptr, amt) / For double-word atomic operations: / ValLo, ValHi, OUTCHAIN = ATOMIC_SWAP(INCHAIN, ptr, amtLo, amtHi) / ValLo, ValHi, OUTCHAIN = ATOMIC_LOAD_[OpName](INCHAIN, ptr, amtLo, amtHi) / These correspond to the atomicrmw instruction. ATOMIC_SWAP, ATOMIC_LOAD_ADD, ATOMIC_LOAD_SUB, ATOMIC_LOAD_AND, ATOMIC_LOAD_CLR, ATOMIC_LOAD_OR, ATOMIC_LOAD_XOR, ATOMIC_LOAD_NAND, ATOMIC_LOAD_MIN, ATOMIC_LOAD_MAX, ATOMIC_LOAD_UMIN, ATOMIC_LOAD_UMAX, ATOMIC_LOAD_FADD, ATOMIC_LOAD_FSUB, ATOMIC_LOAD_FMAX, ATOMIC_LOAD_FMIN, ATOMIC_LOAD_FMAXIMUM, ATOMIC_LOAD_FMINIMUM, ATOMIC_LOAD_UINC_WRAP, ATOMIC_LOAD_UDEC_WRAP, ATOMIC_LOAD_USUB_COND, ATOMIC_LOAD_USUB_SAT,

/ Masked load and store - consecutive vector load and store operations / with additional mask operand that prevents memory accesses to the / masked-off lanes. / / Val, OutChain = MLOAD(BasePtr, Mask, PassThru) / OutChain = MSTORE(Value, BasePtr, Mask) MLOAD, MSTORE,

/ Masked gather and scatter - load and store operations for a vector of / random addresses with additional mask operand that prevents memory / accesses to the masked-off lanes. / / Val, OutChain = GATHER(InChain, PassThru, Mask, BasePtr, Index, Scale) / OutChain = SCATTER(InChain, Value, Mask, BasePtr, Index, Scale) / / The Index operand can have more vector elements than the other operands / due to type legalization. The extra elements are ignored. MGATHER, MSCATTER,

/ This corresponds to the llvm.lifetime.* intrinsics. The first operand / is the chain and the second operand is the alloca pointer. LIFETIME_START, LIFETIME_END,

/ FAKE_USE represents a use of the operand but does not do anything. / Its purpose is the extension of the operand's lifetime mainly for / debugging purposes. FAKE_USE,

/ GC_TRANSITION_START/GC_TRANSITION_END - These operators mark the / beginning and end of GC transition sequence, and carry arbitrary / information that target might need for lowering. The first operand is / a chain, the rest are specified by the target and not touched by the DAG / optimizers. GC_TRANSITION_START..GC_TRANSITION_END pairs may not be / nested. GC_TRANSITION_START, GC_TRANSITION_END,

/ GET_DYNAMIC_AREA_OFFSET - get offset from native SP to the address of / the most recent dynamic alloca. For most targets that would be 0, but / for some others (e.g. PowerPC, PowerPC64) that would be compile-time / known nonzero constant. The only operand here is the chain. GET_DYNAMIC_AREA_OFFSET,

/ Pseudo probe for AutoFDO, as a place holder in a basic block to improve / the sample counts quality. PSEUDO_PROBE,

/ VSCALE(IMM) - Returns the runtime scaling factor used to calculate the / number of elements within a scalable vector. IMM is a constant integer / multiplier that is applied to the runtime value. VSCALE,

/ Generic reduction nodes. These nodes represent horizontal vector / reduction operations, producing a scalar result. / The SEQ variants perform reductions in sequential order. The first / operand is an initial scalar accumulator value, and the second operand / is the vector to reduce. / E.g. RES = VECREDUCE_SEQ_FADD f32 ACC, <4 x f32> SRC_VEC / ... is equivalent to / RES = (((ACC + SRC_VEC[0]) + SRC_VEC[1]) + SRC_VEC[2]) + SRC_VEC[3] VECREDUCE_SEQ_FADD, VECREDUCE_SEQ_FMUL,

/ These reductions have relaxed evaluation order semantics, and have a / single vector operand. The order of evaluation is unspecified. For / pow-of-2 vectors, one valid legalizer expansion is to use a tree / reduction, i.e.: / For RES = VECREDUCE_FADD <8 x f16> SRC_VEC / / PART_RDX = FADD SRC_VEC[0:3], SRC_VEC[4:7] / PART_RDX2 = FADD PART_RDX[0:1], PART_RDX[2:3] / RES = FADD PART_RDX2[0], PART_RDX2[1] / / For non-pow-2 vectors, this can be computed by extracting each element / and performing the operation as if it were scalarized. VECREDUCE_FADD, VECREDUCE_FMUL, / FMIN/FMAX nodes can have flags, for NaN/NoNaN variants. VECREDUCE_FMAX, VECREDUCE_FMIN, / FMINIMUM/FMAXIMUM nodes propatate NaNs and signed zeroes using the / llvm.minimum and llvm.maximum semantics. VECREDUCE_FMAXIMUM, VECREDUCE_FMINIMUM, / Integer reductions may have a result type larger than the vector element / type. However, the reduction is performed using the vector element type / and the value in the top bits is unspecified. VECREDUCE_ADD, VECREDUCE_MUL, VECREDUCE_AND, VECREDUCE_OR, VECREDUCE_XOR, VECREDUCE_SMAX, VECREDUCE_SMIN, VECREDUCE_UMAX, VECREDUCE_UMIN,

PARTIAL_REDUCE_[U|S]MLA(Accumulator, Input1, Input2) The partial reduction nodes sign or zero extend Input1 and Input2 (with the extension kind noted below) to the element type of Accumulator before multiplying their results. This result is concatenated to the Accumulator, and this is then reduced, using addition, to the result type. The output is only expected to either be given to another partial reduction operation or an equivalent vector reduce operation, so the order in which the elements are reduced is deliberately not specified. Input1 and Input2 must be the same type. Accumulator and the output must be the same type. The number of elements in Input1 and Input2 must be a positive integer multiple of the number of elements in the Accumulator / output type. Input1 and Input2 must have an element type which is the same as or smaller than the element type of the Accumulator and output. PARTIAL_REDUCE_SMLA, // sext, sext PARTIAL_REDUCE_UMLA, // zext, zext PARTIAL_REDUCE_SUMLA, // sext, zext PARTIAL_REDUCE_FMLA, // fpext, fpext

The llvm.experimental.stackmap intrinsic. Operands: input chain, glue, , , [live0[, live1...]] Outputs: output chain, glue STACKMAP,

The llvm.experimental.patchpoint.* intrinsic. Operands: input chain, [glue], reg-mask, , , callee, , cc, ... Outputs: [rv], output chain, glue PATCHPOINT,

PTRADD represents pointer arithmetic semantics, for targets that opt in using shouldPreservePtrArith(). ptr = PTRADD ptr, offset PTRADD,

Vector Predication #define BEGIN_REGISTER_VP_SDNODE(VPSDID, ...)

Issue a no-op relocation against a given symbol at the current location. RELOC_NONE,

The llvm.experimental.convergence.* intrinsics. CONVERGENCECTRL_ANCHOR, CONVERGENCECTRL_ENTRY, CONVERGENCECTRL_LOOP, This does not correspond to any convergence control intrinsic. It is used to glue a convergence control token to a convergent operation in the DAG, which is later translated to an implicit use in the MIR. CONVERGENCECTRL_GLUE,

Experimental vector histogram intrinsic Operands: Input Chain, Inc, Mask, Base, Index, Scale, ID Output: Output Chain EXPERIMENTAL_VECTOR_HISTOGRAM,

Finds the index of the last active mask element Operands: Mask VECTOR_FIND_LAST_ACTIVE,

GET_ACTIVE_LANE_MASK - this corrosponds to the llvm.get.active.lane.mask intrinsic. It creates a mask representing active and inactive vector lanes, active while Base + index < Trip Count. As with the intrinsic, the operands Base and Trip Count have the same scalar integer type and the internal addition of Base + index cannot overflow. However, the ISD node supports result types which are wider than i1, where the high bits conform to getBooleanContents similar to the SETCC operator. GET_ACTIVE_LANE_MASK,

The llvm.loop.dependence.{war, raw}.mask intrinsics

LOOP_DEPENDENCE_RAW_MASK

CLEAR_CACHE

DEACTIVATION_SYMBOL

BUILTIN_OP_END

BUILTIN_OP_END - This must be the last enum value in this list.

The target-specific pre-isel opcode values start here.