mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 11:13:28 +01:00
bc658bf60a
This adds support for the new 128-bit vector float instructions of z14. Note that these instructions actually only operate on the f128 type, since only each 128-bit vector register can hold only one 128-bit float value. However, this is still preferable to the legacy 128-bit float instructions, since those operate on pairs of floating-point registers (so we can hold at most 8 values in registers), while the new instructions use single vector registers (so we hold up to 32 value in registers). Adding support includes: - Enabling the instructions for the assembler/disassembler. - CodeGen for the instructions. This includes allocating the f128 type now to the VR128BitRegClass instead of FP128BitRegClass. - Scheduler description support for the instructions. Note that for a small number of operations, we have no new vector instructions (like integer <-> 128-bit float conversions), and so we use the legacy instruction and then reformat the operand (i.e. copy between a pair of floating-point registers and a vector register). llvm-svn: 308196
550 lines
24 KiB
TableGen
550 lines
24 KiB
TableGen
//==- SystemZInstrFP.td - Floating-point SystemZ instructions --*- tblgen-*-==//
|
|
//
|
|
// The LLVM Compiler Infrastructure
|
|
//
|
|
// This file is distributed under the University of Illinois Open Source
|
|
// License. See LICENSE.TXT for details.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Select instructions
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// C's ?: operator for floating-point operands.
|
|
def SelectF32 : SelectWrapper<f32, FP32>;
|
|
def SelectF64 : SelectWrapper<f64, FP64>;
|
|
let Predicates = [FeatureNoVectorEnhancements1] in
|
|
def SelectF128 : SelectWrapper<f128, FP128>;
|
|
let Predicates = [FeatureVectorEnhancements1] in
|
|
def SelectVR128 : SelectWrapper<f128, VR128>;
|
|
|
|
defm CondStoreF32 : CondStores<FP32, nonvolatile_store,
|
|
nonvolatile_load, bdxaddr20only>;
|
|
defm CondStoreF64 : CondStores<FP64, nonvolatile_store,
|
|
nonvolatile_load, bdxaddr20only>;
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Move instructions
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Load zero.
|
|
let hasSideEffects = 0, isAsCheapAsAMove = 1, isMoveImm = 1 in {
|
|
def LZER : InherentRRE<"lzer", 0xB374, FP32, fpimm0>;
|
|
def LZDR : InherentRRE<"lzdr", 0xB375, FP64, fpimm0>;
|
|
def LZXR : InherentRRE<"lzxr", 0xB376, FP128, fpimm0>;
|
|
}
|
|
|
|
// Moves between two floating-point registers.
|
|
let hasSideEffects = 0 in {
|
|
def LER : UnaryRR <"ler", 0x38, null_frag, FP32, FP32>;
|
|
def LDR : UnaryRR <"ldr", 0x28, null_frag, FP64, FP64>;
|
|
def LXR : UnaryRRE<"lxr", 0xB365, null_frag, FP128, FP128>;
|
|
|
|
// For z13 we prefer LDR over LER to avoid partial register dependencies.
|
|
let isCodeGenOnly = 1 in
|
|
def LDR32 : UnaryRR<"ldr", 0x28, null_frag, FP32, FP32>;
|
|
}
|
|
|
|
// Moves between two floating-point registers that also set the condition
|
|
// codes.
|
|
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
|
|
defm LTEBR : LoadAndTestRRE<"ltebr", 0xB302, FP32>;
|
|
defm LTDBR : LoadAndTestRRE<"ltdbr", 0xB312, FP64>;
|
|
defm LTXBR : LoadAndTestRRE<"ltxbr", 0xB342, FP128>;
|
|
}
|
|
// Note that LTxBRCompare is not available if we have vector support,
|
|
// since load-and-test instructions will partially clobber the target
|
|
// (vector) register.
|
|
let Predicates = [FeatureNoVector] in {
|
|
defm : CompareZeroFP<LTEBRCompare, FP32>;
|
|
defm : CompareZeroFP<LTDBRCompare, FP64>;
|
|
defm : CompareZeroFP<LTXBRCompare, FP128>;
|
|
}
|
|
|
|
// Use a normal load-and-test for compare against zero in case of
|
|
// vector support (via a pseudo to simplify instruction selection).
|
|
let Defs = [CC], usesCustomInserter = 1 in {
|
|
def LTEBRCompare_VecPseudo : Pseudo<(outs), (ins FP32:$R1, FP32:$R2), []>;
|
|
def LTDBRCompare_VecPseudo : Pseudo<(outs), (ins FP64:$R1, FP64:$R2), []>;
|
|
def LTXBRCompare_VecPseudo : Pseudo<(outs), (ins FP128:$R1, FP128:$R2), []>;
|
|
}
|
|
let Predicates = [FeatureVector] in {
|
|
defm : CompareZeroFP<LTEBRCompare_VecPseudo, FP32>;
|
|
defm : CompareZeroFP<LTDBRCompare_VecPseudo, FP64>;
|
|
}
|
|
let Predicates = [FeatureVector, FeatureNoVectorEnhancements1] in
|
|
defm : CompareZeroFP<LTXBRCompare_VecPseudo, FP128>;
|
|
|
|
// Moves between 64-bit integer and floating-point registers.
|
|
def LGDR : UnaryRRE<"lgdr", 0xB3CD, bitconvert, GR64, FP64>;
|
|
def LDGR : UnaryRRE<"ldgr", 0xB3C1, bitconvert, FP64, GR64>;
|
|
|
|
// fcopysign with an FP32 result.
|
|
let isCodeGenOnly = 1 in {
|
|
def CPSDRss : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP32, FP32, FP32>;
|
|
def CPSDRsd : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP32, FP32, FP64>;
|
|
}
|
|
|
|
// The sign of an FP128 is in the high register.
|
|
let Predicates = [FeatureNoVectorEnhancements1] in
|
|
def : Pat<(fcopysign FP32:$src1, (f32 (fpround (f128 FP128:$src2)))),
|
|
(CPSDRsd FP32:$src1, (EXTRACT_SUBREG FP128:$src2, subreg_h64))>;
|
|
let Predicates = [FeatureVectorEnhancements1] in
|
|
def : Pat<(fcopysign FP32:$src1, (f32 (fpround (f128 VR128:$src2)))),
|
|
(CPSDRsd FP32:$src1, (EXTRACT_SUBREG VR128:$src2, subreg_r64))>;
|
|
|
|
// fcopysign with an FP64 result.
|
|
let isCodeGenOnly = 1 in
|
|
def CPSDRds : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP64, FP64, FP32>;
|
|
def CPSDRdd : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP64, FP64, FP64>;
|
|
|
|
// The sign of an FP128 is in the high register.
|
|
let Predicates = [FeatureNoVectorEnhancements1] in
|
|
def : Pat<(fcopysign FP64:$src1, (f64 (fpround (f128 FP128:$src2)))),
|
|
(CPSDRdd FP64:$src1, (EXTRACT_SUBREG FP128:$src2, subreg_h64))>;
|
|
let Predicates = [FeatureVectorEnhancements1] in
|
|
def : Pat<(fcopysign FP64:$src1, (f64 (fpround (f128 VR128:$src2)))),
|
|
(CPSDRdd FP64:$src1, (EXTRACT_SUBREG VR128:$src2, subreg_r64))>;
|
|
|
|
// fcopysign with an FP128 result. Use "upper" as the high half and leave
|
|
// the low half as-is.
|
|
class CopySign128<RegisterOperand cls, dag upper>
|
|
: Pat<(fcopysign FP128:$src1, cls:$src2),
|
|
(INSERT_SUBREG FP128:$src1, upper, subreg_h64)>;
|
|
|
|
let Predicates = [FeatureNoVectorEnhancements1] in {
|
|
def : CopySign128<FP32, (CPSDRds (EXTRACT_SUBREG FP128:$src1, subreg_h64),
|
|
FP32:$src2)>;
|
|
def : CopySign128<FP64, (CPSDRdd (EXTRACT_SUBREG FP128:$src1, subreg_h64),
|
|
FP64:$src2)>;
|
|
def : CopySign128<FP128, (CPSDRdd (EXTRACT_SUBREG FP128:$src1, subreg_h64),
|
|
(EXTRACT_SUBREG FP128:$src2, subreg_h64))>;
|
|
}
|
|
|
|
defm LoadStoreF32 : MVCLoadStore<load, f32, MVCSequence, 4>;
|
|
defm LoadStoreF64 : MVCLoadStore<load, f64, MVCSequence, 8>;
|
|
defm LoadStoreF128 : MVCLoadStore<load, f128, MVCSequence, 16>;
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Load instructions
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
let canFoldAsLoad = 1, SimpleBDXLoad = 1 in {
|
|
defm LE : UnaryRXPair<"le", 0x78, 0xED64, load, FP32, 4>;
|
|
defm LD : UnaryRXPair<"ld", 0x68, 0xED65, load, FP64, 8>;
|
|
|
|
// For z13 we prefer LDE over LE to avoid partial register dependencies.
|
|
let isCodeGenOnly = 1 in
|
|
def LDE32 : UnaryRXE<"lde", 0xED24, null_frag, FP32, 4>;
|
|
|
|
// These instructions are split after register allocation, so we don't
|
|
// want a custom inserter.
|
|
let Has20BitOffset = 1, HasIndex = 1, Is128Bit = 1 in {
|
|
def LX : Pseudo<(outs FP128:$dst), (ins bdxaddr20only128:$src),
|
|
[(set FP128:$dst, (load bdxaddr20only128:$src))]>;
|
|
}
|
|
}
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Store instructions
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
let SimpleBDXStore = 1 in {
|
|
defm STE : StoreRXPair<"ste", 0x70, 0xED66, store, FP32, 4>;
|
|
defm STD : StoreRXPair<"std", 0x60, 0xED67, store, FP64, 8>;
|
|
|
|
// These instructions are split after register allocation, so we don't
|
|
// want a custom inserter.
|
|
let Has20BitOffset = 1, HasIndex = 1, Is128Bit = 1 in {
|
|
def STX : Pseudo<(outs), (ins FP128:$src, bdxaddr20only128:$dst),
|
|
[(store FP128:$src, bdxaddr20only128:$dst)]>;
|
|
}
|
|
}
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Conversion instructions
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Convert floating-point values to narrower representations, rounding
|
|
// according to the current mode. The destination of LEXBR and LDXBR
|
|
// is a 128-bit value, but only the first register of the pair is used.
|
|
def LEDBR : UnaryRRE<"ledbr", 0xB344, fpround, FP32, FP64>;
|
|
def LEXBR : UnaryRRE<"lexbr", 0xB346, null_frag, FP128, FP128>;
|
|
def LDXBR : UnaryRRE<"ldxbr", 0xB345, null_frag, FP128, FP128>;
|
|
|
|
def LEDBRA : TernaryRRFe<"ledbra", 0xB344, FP32, FP64>,
|
|
Requires<[FeatureFPExtension]>;
|
|
def LEXBRA : TernaryRRFe<"lexbra", 0xB346, FP128, FP128>,
|
|
Requires<[FeatureFPExtension]>;
|
|
def LDXBRA : TernaryRRFe<"ldxbra", 0xB345, FP128, FP128>,
|
|
Requires<[FeatureFPExtension]>;
|
|
|
|
let Predicates = [FeatureNoVectorEnhancements1] in {
|
|
def : Pat<(f32 (fpround FP128:$src)),
|
|
(EXTRACT_SUBREG (LEXBR FP128:$src), subreg_hr32)>;
|
|
def : Pat<(f64 (fpround FP128:$src)),
|
|
(EXTRACT_SUBREG (LDXBR FP128:$src), subreg_h64)>;
|
|
}
|
|
|
|
// Extend register floating-point values to wider representations.
|
|
def LDEBR : UnaryRRE<"ldebr", 0xB304, fpextend, FP64, FP32>;
|
|
def LXEBR : UnaryRRE<"lxebr", 0xB306, null_frag, FP128, FP32>;
|
|
def LXDBR : UnaryRRE<"lxdbr", 0xB305, null_frag, FP128, FP64>;
|
|
let Predicates = [FeatureNoVectorEnhancements1] in {
|
|
def : Pat<(f128 (fpextend (f32 FP32:$src))), (LXEBR FP32:$src)>;
|
|
def : Pat<(f128 (fpextend (f64 FP64:$src))), (LXDBR FP64:$src)>;
|
|
}
|
|
|
|
// Extend memory floating-point values to wider representations.
|
|
def LDEB : UnaryRXE<"ldeb", 0xED04, extloadf32, FP64, 4>;
|
|
def LXEB : UnaryRXE<"lxeb", 0xED06, null_frag, FP128, 4>;
|
|
def LXDB : UnaryRXE<"lxdb", 0xED05, null_frag, FP128, 8>;
|
|
let Predicates = [FeatureNoVectorEnhancements1] in {
|
|
def : Pat<(f128 (extloadf32 bdxaddr12only:$src)),
|
|
(LXEB bdxaddr12only:$src)>;
|
|
def : Pat<(f128 (extloadf64 bdxaddr12only:$src)),
|
|
(LXDB bdxaddr12only:$src)>;
|
|
}
|
|
|
|
// Convert a signed integer register value to a floating-point one.
|
|
def CEFBR : UnaryRRE<"cefbr", 0xB394, sint_to_fp, FP32, GR32>;
|
|
def CDFBR : UnaryRRE<"cdfbr", 0xB395, sint_to_fp, FP64, GR32>;
|
|
def CXFBR : UnaryRRE<"cxfbr", 0xB396, sint_to_fp, FP128, GR32>;
|
|
|
|
def CEGBR : UnaryRRE<"cegbr", 0xB3A4, sint_to_fp, FP32, GR64>;
|
|
def CDGBR : UnaryRRE<"cdgbr", 0xB3A5, sint_to_fp, FP64, GR64>;
|
|
def CXGBR : UnaryRRE<"cxgbr", 0xB3A6, sint_to_fp, FP128, GR64>;
|
|
|
|
// The FP extension feature provides versions of the above that allow
|
|
// specifying rounding mode and inexact-exception suppression flags.
|
|
let Predicates = [FeatureFPExtension] in {
|
|
def CEFBRA : TernaryRRFe<"cefbra", 0xB394, FP32, GR32>;
|
|
def CDFBRA : TernaryRRFe<"cdfbra", 0xB395, FP64, GR32>;
|
|
def CXFBRA : TernaryRRFe<"cxfbra", 0xB396, FP128, GR32>;
|
|
|
|
def CEGBRA : TernaryRRFe<"cegbra", 0xB3A4, FP32, GR64>;
|
|
def CDGBRA : TernaryRRFe<"cdgbra", 0xB3A5, FP64, GR64>;
|
|
def CXGBRA : TernaryRRFe<"cxgbra", 0xB3A6, FP128, GR64>;
|
|
}
|
|
|
|
// Convert am unsigned integer register value to a floating-point one.
|
|
let Predicates = [FeatureFPExtension] in {
|
|
def CELFBR : TernaryRRFe<"celfbr", 0xB390, FP32, GR32>;
|
|
def CDLFBR : TernaryRRFe<"cdlfbr", 0xB391, FP64, GR32>;
|
|
def CXLFBR : TernaryRRFe<"cxlfbr", 0xB392, FP128, GR32>;
|
|
|
|
def CELGBR : TernaryRRFe<"celgbr", 0xB3A0, FP32, GR64>;
|
|
def CDLGBR : TernaryRRFe<"cdlgbr", 0xB3A1, FP64, GR64>;
|
|
def CXLGBR : TernaryRRFe<"cxlgbr", 0xB3A2, FP128, GR64>;
|
|
|
|
def : Pat<(f32 (uint_to_fp GR32:$src)), (CELFBR 0, GR32:$src, 0)>;
|
|
def : Pat<(f64 (uint_to_fp GR32:$src)), (CDLFBR 0, GR32:$src, 0)>;
|
|
def : Pat<(f128 (uint_to_fp GR32:$src)), (CXLFBR 0, GR32:$src, 0)>;
|
|
|
|
def : Pat<(f32 (uint_to_fp GR64:$src)), (CELGBR 0, GR64:$src, 0)>;
|
|
def : Pat<(f64 (uint_to_fp GR64:$src)), (CDLGBR 0, GR64:$src, 0)>;
|
|
def : Pat<(f128 (uint_to_fp GR64:$src)), (CXLGBR 0, GR64:$src, 0)>;
|
|
}
|
|
|
|
// Convert a floating-point register value to a signed integer value,
|
|
// with the second operand (modifier M3) specifying the rounding mode.
|
|
let Defs = [CC] in {
|
|
def CFEBR : BinaryRRFe<"cfebr", 0xB398, GR32, FP32>;
|
|
def CFDBR : BinaryRRFe<"cfdbr", 0xB399, GR32, FP64>;
|
|
def CFXBR : BinaryRRFe<"cfxbr", 0xB39A, GR32, FP128>;
|
|
|
|
def CGEBR : BinaryRRFe<"cgebr", 0xB3A8, GR64, FP32>;
|
|
def CGDBR : BinaryRRFe<"cgdbr", 0xB3A9, GR64, FP64>;
|
|
def CGXBR : BinaryRRFe<"cgxbr", 0xB3AA, GR64, FP128>;
|
|
}
|
|
|
|
// fp_to_sint always rounds towards zero, which is modifier value 5.
|
|
def : Pat<(i32 (fp_to_sint FP32:$src)), (CFEBR 5, FP32:$src)>;
|
|
def : Pat<(i32 (fp_to_sint FP64:$src)), (CFDBR 5, FP64:$src)>;
|
|
def : Pat<(i32 (fp_to_sint FP128:$src)), (CFXBR 5, FP128:$src)>;
|
|
|
|
def : Pat<(i64 (fp_to_sint FP32:$src)), (CGEBR 5, FP32:$src)>;
|
|
def : Pat<(i64 (fp_to_sint FP64:$src)), (CGDBR 5, FP64:$src)>;
|
|
def : Pat<(i64 (fp_to_sint FP128:$src)), (CGXBR 5, FP128:$src)>;
|
|
|
|
// The FP extension feature provides versions of the above that allow
|
|
// also specifying the inexact-exception suppression flag.
|
|
let Predicates = [FeatureFPExtension], Defs = [CC] in {
|
|
def CFEBRA : TernaryRRFe<"cfebra", 0xB398, GR32, FP32>;
|
|
def CFDBRA : TernaryRRFe<"cfdbra", 0xB399, GR32, FP64>;
|
|
def CFXBRA : TernaryRRFe<"cfxbra", 0xB39A, GR32, FP128>;
|
|
|
|
def CGEBRA : TernaryRRFe<"cgebra", 0xB3A8, GR64, FP32>;
|
|
def CGDBRA : TernaryRRFe<"cgdbra", 0xB3A9, GR64, FP64>;
|
|
def CGXBRA : TernaryRRFe<"cgxbra", 0xB3AA, GR64, FP128>;
|
|
}
|
|
|
|
// Convert a floating-point register value to an unsigned integer value.
|
|
let Predicates = [FeatureFPExtension] in {
|
|
let Defs = [CC] in {
|
|
def CLFEBR : TernaryRRFe<"clfebr", 0xB39C, GR32, FP32>;
|
|
def CLFDBR : TernaryRRFe<"clfdbr", 0xB39D, GR32, FP64>;
|
|
def CLFXBR : TernaryRRFe<"clfxbr", 0xB39E, GR32, FP128>;
|
|
|
|
def CLGEBR : TernaryRRFe<"clgebr", 0xB3AC, GR64, FP32>;
|
|
def CLGDBR : TernaryRRFe<"clgdbr", 0xB3AD, GR64, FP64>;
|
|
def CLGXBR : TernaryRRFe<"clgxbr", 0xB3AE, GR64, FP128>;
|
|
}
|
|
|
|
def : Pat<(i32 (fp_to_uint FP32:$src)), (CLFEBR 5, FP32:$src, 0)>;
|
|
def : Pat<(i32 (fp_to_uint FP64:$src)), (CLFDBR 5, FP64:$src, 0)>;
|
|
def : Pat<(i32 (fp_to_uint FP128:$src)), (CLFXBR 5, FP128:$src, 0)>;
|
|
|
|
def : Pat<(i64 (fp_to_uint FP32:$src)), (CLGEBR 5, FP32:$src, 0)>;
|
|
def : Pat<(i64 (fp_to_uint FP64:$src)), (CLGDBR 5, FP64:$src, 0)>;
|
|
def : Pat<(i64 (fp_to_uint FP128:$src)), (CLGXBR 5, FP128:$src, 0)>;
|
|
}
|
|
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Unary arithmetic
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// We prefer generic instructions during isel, because they do not
|
|
// clobber CC and therefore give the scheduler more freedom. In cases
|
|
// the CC is actually useful, the SystemZElimCompare pass will try to
|
|
// convert generic instructions into opcodes that also set CC. Note
|
|
// that lcdf / lpdf / lndf only affect the sign bit, and can therefore
|
|
// be used with fp32 as well. This could be done for fp128, in which
|
|
// case the operands would have to be tied.
|
|
|
|
// Negation (Load Complement).
|
|
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
|
|
def LCEBR : UnaryRRE<"lcebr", 0xB303, null_frag, FP32, FP32>;
|
|
def LCDBR : UnaryRRE<"lcdbr", 0xB313, null_frag, FP64, FP64>;
|
|
def LCXBR : UnaryRRE<"lcxbr", 0xB343, fneg, FP128, FP128>;
|
|
}
|
|
// Generic form, which does not set CC.
|
|
def LCDFR : UnaryRRE<"lcdfr", 0xB373, fneg, FP64, FP64>;
|
|
let isCodeGenOnly = 1 in
|
|
def LCDFR_32 : UnaryRRE<"lcdfr", 0xB373, fneg, FP32, FP32>;
|
|
|
|
// Absolute value (Load Positive).
|
|
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
|
|
def LPEBR : UnaryRRE<"lpebr", 0xB300, null_frag, FP32, FP32>;
|
|
def LPDBR : UnaryRRE<"lpdbr", 0xB310, null_frag, FP64, FP64>;
|
|
def LPXBR : UnaryRRE<"lpxbr", 0xB340, fabs, FP128, FP128>;
|
|
}
|
|
// Generic form, which does not set CC.
|
|
def LPDFR : UnaryRRE<"lpdfr", 0xB370, fabs, FP64, FP64>;
|
|
let isCodeGenOnly = 1 in
|
|
def LPDFR_32 : UnaryRRE<"lpdfr", 0xB370, fabs, FP32, FP32>;
|
|
|
|
// Negative absolute value (Load Negative).
|
|
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
|
|
def LNEBR : UnaryRRE<"lnebr", 0xB301, null_frag, FP32, FP32>;
|
|
def LNDBR : UnaryRRE<"lndbr", 0xB311, null_frag, FP64, FP64>;
|
|
def LNXBR : UnaryRRE<"lnxbr", 0xB341, fnabs, FP128, FP128>;
|
|
}
|
|
// Generic form, which does not set CC.
|
|
def LNDFR : UnaryRRE<"lndfr", 0xB371, fnabs, FP64, FP64>;
|
|
let isCodeGenOnly = 1 in
|
|
def LNDFR_32 : UnaryRRE<"lndfr", 0xB371, fnabs, FP32, FP32>;
|
|
|
|
// Square root.
|
|
def SQEBR : UnaryRRE<"sqebr", 0xB314, fsqrt, FP32, FP32>;
|
|
def SQDBR : UnaryRRE<"sqdbr", 0xB315, fsqrt, FP64, FP64>;
|
|
def SQXBR : UnaryRRE<"sqxbr", 0xB316, fsqrt, FP128, FP128>;
|
|
|
|
def SQEB : UnaryRXE<"sqeb", 0xED14, loadu<fsqrt>, FP32, 4>;
|
|
def SQDB : UnaryRXE<"sqdb", 0xED15, loadu<fsqrt>, FP64, 8>;
|
|
|
|
// Round to an integer, with the second operand (modifier M3) specifying
|
|
// the rounding mode. These forms always check for inexact conditions.
|
|
def FIEBR : BinaryRRFe<"fiebr", 0xB357, FP32, FP32>;
|
|
def FIDBR : BinaryRRFe<"fidbr", 0xB35F, FP64, FP64>;
|
|
def FIXBR : BinaryRRFe<"fixbr", 0xB347, FP128, FP128>;
|
|
|
|
// frint rounds according to the current mode (modifier 0) and detects
|
|
// inexact conditions.
|
|
def : Pat<(frint FP32:$src), (FIEBR 0, FP32:$src)>;
|
|
def : Pat<(frint FP64:$src), (FIDBR 0, FP64:$src)>;
|
|
def : Pat<(frint FP128:$src), (FIXBR 0, FP128:$src)>;
|
|
|
|
let Predicates = [FeatureFPExtension] in {
|
|
// Extended forms of the FIxBR instructions. M4 can be set to 4
|
|
// to suppress detection of inexact conditions.
|
|
def FIEBRA : TernaryRRFe<"fiebra", 0xB357, FP32, FP32>;
|
|
def FIDBRA : TernaryRRFe<"fidbra", 0xB35F, FP64, FP64>;
|
|
def FIXBRA : TernaryRRFe<"fixbra", 0xB347, FP128, FP128>;
|
|
|
|
// fnearbyint is like frint but does not detect inexact conditions.
|
|
def : Pat<(fnearbyint FP32:$src), (FIEBRA 0, FP32:$src, 4)>;
|
|
def : Pat<(fnearbyint FP64:$src), (FIDBRA 0, FP64:$src, 4)>;
|
|
def : Pat<(fnearbyint FP128:$src), (FIXBRA 0, FP128:$src, 4)>;
|
|
|
|
// floor is no longer allowed to raise an inexact condition,
|
|
// so restrict it to the cases where the condition can be suppressed.
|
|
// Mode 7 is round towards -inf.
|
|
def : Pat<(ffloor FP32:$src), (FIEBRA 7, FP32:$src, 4)>;
|
|
def : Pat<(ffloor FP64:$src), (FIDBRA 7, FP64:$src, 4)>;
|
|
def : Pat<(ffloor FP128:$src), (FIXBRA 7, FP128:$src, 4)>;
|
|
|
|
// Same idea for ceil, where mode 6 is round towards +inf.
|
|
def : Pat<(fceil FP32:$src), (FIEBRA 6, FP32:$src, 4)>;
|
|
def : Pat<(fceil FP64:$src), (FIDBRA 6, FP64:$src, 4)>;
|
|
def : Pat<(fceil FP128:$src), (FIXBRA 6, FP128:$src, 4)>;
|
|
|
|
// Same idea for trunc, where mode 5 is round towards zero.
|
|
def : Pat<(ftrunc FP32:$src), (FIEBRA 5, FP32:$src, 4)>;
|
|
def : Pat<(ftrunc FP64:$src), (FIDBRA 5, FP64:$src, 4)>;
|
|
def : Pat<(ftrunc FP128:$src), (FIXBRA 5, FP128:$src, 4)>;
|
|
|
|
// Same idea for round, where mode 1 is round towards nearest with
|
|
// ties away from zero.
|
|
def : Pat<(fround FP32:$src), (FIEBRA 1, FP32:$src, 4)>;
|
|
def : Pat<(fround FP64:$src), (FIDBRA 1, FP64:$src, 4)>;
|
|
def : Pat<(fround FP128:$src), (FIXBRA 1, FP128:$src, 4)>;
|
|
}
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Binary arithmetic
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
// Addition.
|
|
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
|
|
let isCommutable = 1 in {
|
|
def AEBR : BinaryRRE<"aebr", 0xB30A, fadd, FP32, FP32>;
|
|
def ADBR : BinaryRRE<"adbr", 0xB31A, fadd, FP64, FP64>;
|
|
def AXBR : BinaryRRE<"axbr", 0xB34A, fadd, FP128, FP128>;
|
|
}
|
|
def AEB : BinaryRXE<"aeb", 0xED0A, fadd, FP32, load, 4>;
|
|
def ADB : BinaryRXE<"adb", 0xED1A, fadd, FP64, load, 8>;
|
|
}
|
|
|
|
// Subtraction.
|
|
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
|
|
def SEBR : BinaryRRE<"sebr", 0xB30B, fsub, FP32, FP32>;
|
|
def SDBR : BinaryRRE<"sdbr", 0xB31B, fsub, FP64, FP64>;
|
|
def SXBR : BinaryRRE<"sxbr", 0xB34B, fsub, FP128, FP128>;
|
|
|
|
def SEB : BinaryRXE<"seb", 0xED0B, fsub, FP32, load, 4>;
|
|
def SDB : BinaryRXE<"sdb", 0xED1B, fsub, FP64, load, 8>;
|
|
}
|
|
|
|
// Multiplication.
|
|
let isCommutable = 1 in {
|
|
def MEEBR : BinaryRRE<"meebr", 0xB317, fmul, FP32, FP32>;
|
|
def MDBR : BinaryRRE<"mdbr", 0xB31C, fmul, FP64, FP64>;
|
|
def MXBR : BinaryRRE<"mxbr", 0xB34C, fmul, FP128, FP128>;
|
|
}
|
|
def MEEB : BinaryRXE<"meeb", 0xED17, fmul, FP32, load, 4>;
|
|
def MDB : BinaryRXE<"mdb", 0xED1C, fmul, FP64, load, 8>;
|
|
|
|
// f64 multiplication of two FP32 registers.
|
|
def MDEBR : BinaryRRE<"mdebr", 0xB30C, null_frag, FP64, FP32>;
|
|
def : Pat<(fmul (f64 (fpextend FP32:$src1)), (f64 (fpextend FP32:$src2))),
|
|
(MDEBR (INSERT_SUBREG (f64 (IMPLICIT_DEF)),
|
|
FP32:$src1, subreg_r32), FP32:$src2)>;
|
|
|
|
// f64 multiplication of an FP32 register and an f32 memory.
|
|
def MDEB : BinaryRXE<"mdeb", 0xED0C, null_frag, FP64, load, 4>;
|
|
def : Pat<(fmul (f64 (fpextend FP32:$src1)),
|
|
(f64 (extloadf32 bdxaddr12only:$addr))),
|
|
(MDEB (INSERT_SUBREG (f64 (IMPLICIT_DEF)), FP32:$src1, subreg_r32),
|
|
bdxaddr12only:$addr)>;
|
|
|
|
// f128 multiplication of two FP64 registers.
|
|
def MXDBR : BinaryRRE<"mxdbr", 0xB307, null_frag, FP128, FP64>;
|
|
let Predicates = [FeatureNoVectorEnhancements1] in
|
|
def : Pat<(fmul (f128 (fpextend FP64:$src1)), (f128 (fpextend FP64:$src2))),
|
|
(MXDBR (INSERT_SUBREG (f128 (IMPLICIT_DEF)),
|
|
FP64:$src1, subreg_h64), FP64:$src2)>;
|
|
|
|
// f128 multiplication of an FP64 register and an f64 memory.
|
|
def MXDB : BinaryRXE<"mxdb", 0xED07, null_frag, FP128, load, 8>;
|
|
let Predicates = [FeatureNoVectorEnhancements1] in
|
|
def : Pat<(fmul (f128 (fpextend FP64:$src1)),
|
|
(f128 (extloadf64 bdxaddr12only:$addr))),
|
|
(MXDB (INSERT_SUBREG (f128 (IMPLICIT_DEF)), FP64:$src1, subreg_h64),
|
|
bdxaddr12only:$addr)>;
|
|
|
|
// Fused multiply-add.
|
|
def MAEBR : TernaryRRD<"maebr", 0xB30E, z_fma, FP32, FP32>;
|
|
def MADBR : TernaryRRD<"madbr", 0xB31E, z_fma, FP64, FP64>;
|
|
|
|
def MAEB : TernaryRXF<"maeb", 0xED0E, z_fma, FP32, FP32, load, 4>;
|
|
def MADB : TernaryRXF<"madb", 0xED1E, z_fma, FP64, FP64, load, 8>;
|
|
|
|
// Fused multiply-subtract.
|
|
def MSEBR : TernaryRRD<"msebr", 0xB30F, z_fms, FP32, FP32>;
|
|
def MSDBR : TernaryRRD<"msdbr", 0xB31F, z_fms, FP64, FP64>;
|
|
|
|
def MSEB : TernaryRXF<"mseb", 0xED0F, z_fms, FP32, FP32, load, 4>;
|
|
def MSDB : TernaryRXF<"msdb", 0xED1F, z_fms, FP64, FP64, load, 8>;
|
|
|
|
// Division.
|
|
def DEBR : BinaryRRE<"debr", 0xB30D, fdiv, FP32, FP32>;
|
|
def DDBR : BinaryRRE<"ddbr", 0xB31D, fdiv, FP64, FP64>;
|
|
def DXBR : BinaryRRE<"dxbr", 0xB34D, fdiv, FP128, FP128>;
|
|
|
|
def DEB : BinaryRXE<"deb", 0xED0D, fdiv, FP32, load, 4>;
|
|
def DDB : BinaryRXE<"ddb", 0xED1D, fdiv, FP64, load, 8>;
|
|
|
|
// Divide to integer.
|
|
let Defs = [CC] in {
|
|
def DIEBR : TernaryRRFb<"diebr", 0xB353, FP32, FP32, FP32>;
|
|
def DIDBR : TernaryRRFb<"didbr", 0xB35B, FP64, FP64, FP64>;
|
|
}
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Comparisons
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
let Defs = [CC], CCValues = 0xF in {
|
|
def CEBR : CompareRRE<"cebr", 0xB309, z_fcmp, FP32, FP32>;
|
|
def CDBR : CompareRRE<"cdbr", 0xB319, z_fcmp, FP64, FP64>;
|
|
def CXBR : CompareRRE<"cxbr", 0xB349, z_fcmp, FP128, FP128>;
|
|
|
|
def CEB : CompareRXE<"ceb", 0xED09, z_fcmp, FP32, load, 4>;
|
|
def CDB : CompareRXE<"cdb", 0xED19, z_fcmp, FP64, load, 8>;
|
|
|
|
def KEBR : CompareRRE<"kebr", 0xB308, null_frag, FP32, FP32>;
|
|
def KDBR : CompareRRE<"kdbr", 0xB318, null_frag, FP64, FP64>;
|
|
def KXBR : CompareRRE<"kxbr", 0xB348, null_frag, FP128, FP128>;
|
|
|
|
def KEB : CompareRXE<"keb", 0xED08, null_frag, FP32, load, 4>;
|
|
def KDB : CompareRXE<"kdb", 0xED18, null_frag, FP64, load, 8>;
|
|
}
|
|
|
|
// Test Data Class.
|
|
let Defs = [CC], CCValues = 0xC in {
|
|
def TCEB : TestRXE<"tceb", 0xED10, z_tdc, FP32>;
|
|
def TCDB : TestRXE<"tcdb", 0xED11, z_tdc, FP64>;
|
|
def TCXB : TestRXE<"tcxb", 0xED12, z_tdc, FP128>;
|
|
}
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Floating-point control register instructions
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
let hasSideEffects = 1 in {
|
|
def EFPC : InherentRRE<"efpc", 0xB38C, GR32, int_s390_efpc>;
|
|
def STFPC : StoreInherentS<"stfpc", 0xB29C, storei<int_s390_efpc>, 4>;
|
|
|
|
def SFPC : SideEffectUnaryRRE<"sfpc", 0xB384, GR32, int_s390_sfpc>;
|
|
def LFPC : SideEffectUnaryS<"lfpc", 0xB29D, loadu<int_s390_sfpc>, 4>;
|
|
|
|
def SFASR : SideEffectUnaryRRE<"sfasr", 0xB385, GR32, null_frag>;
|
|
def LFAS : SideEffectUnaryS<"lfas", 0xB2BD, null_frag, 4>;
|
|
|
|
def SRNMB : SideEffectAddressS<"srnmb", 0xB2B8, null_frag, shift12only>,
|
|
Requires<[FeatureFPExtension]>;
|
|
def SRNM : SideEffectAddressS<"srnm", 0xB299, null_frag, shift12only>;
|
|
def SRNMT : SideEffectAddressS<"srnmt", 0xB2B9, null_frag, shift12only>;
|
|
}
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
// Peepholes
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
def : Pat<(f32 fpimmneg0), (LCDFR_32 (LZER))>;
|
|
def : Pat<(f64 fpimmneg0), (LCDFR (LZDR))>;
|
|
def : Pat<(f128 fpimmneg0), (LCXBR (LZXR))>;
|