replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Fix VCVT decoding (between floating-point and fixed-point, Floating-point). Patch by Richard Barton. 

llvm-svn: 152814
This commit is contained in:
Kristof Beyls 2012-03-15 17:50:29 +00:00
parent 7a89e0cc01
commit 327d2f9da5
2 changed files with 64 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
llvm/lib/Target/ARM/ARMInstrVFP.td
View File

@ -818,7 +818,29 @@ let Constraints = "$a = $dst" in {
// FP to Fixed-Point: // FP to Fixed-Point:
def VTOSHS : AVConv1XI<0b11101, 0b11, 0b1110, 0b1010, 0, // Single Precision register
class AVConv1XInsS_Encode<bits<5> op1, bits<2> op2, bits<4> op3, bits<4> op4, bit op5,
dag oops, dag iops, InstrItinClass itin, string opc, string asm,
list<dag> pattern>
: AVConv1XI<op1, op2, op3, op4, op5, oops, iops, itin, opc, asm, pattern> {
bits<5> dst;
// if dp_operation then UInt(D:Vd) else UInt(Vd:D);
let Inst{22} = dst{0};
let Inst{15-12} = dst{4-1};
}
// Double Precision register
class AVConv1XInsD_Encode<bits<5> op1, bits<2> op2, bits<4> op3, bits<4> op4, bit op5,
dag oops, dag iops, InstrItinClass itin, string opc, string asm,
list<dag> pattern>
: AVConv1XI<op1, op2, op3, op4, op5, oops, iops, itin, opc, asm, pattern> {
bits<5> dst;
// if dp_operation then UInt(D:Vd) else UInt(Vd:D);
let Inst{22} = dst{4};
let Inst{15-12} = dst{3-0};
}
def VTOSHS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1110, 0b1010, 0,
(outs SPR:$dst), (ins SPR:$a, fbits16:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits16:$fbits),
IIC_fpCVTSI, "vcvt", ".s16.f32\t$dst, $a, $fbits", []> { IIC_fpCVTSI, "vcvt", ".s16.f32\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -826,7 +848,7 @@ def VTOSHS : AVConv1XI<0b11101, 0b11, 0b1110, 0b1010, 0,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VTOUHS : AVConv1XI<0b11101, 0b11, 0b1111, 0b1010, 0, def VTOUHS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1111, 0b1010, 0,
(outs SPR:$dst), (ins SPR:$a, fbits16:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits16:$fbits),
IIC_fpCVTSI, "vcvt", ".u16.f32\t$dst, $a, $fbits", []> { IIC_fpCVTSI, "vcvt", ".u16.f32\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -834,7 +856,7 @@ def VTOUHS : AVConv1XI<0b11101, 0b11, 0b1111, 0b1010, 0,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VTOSLS : AVConv1XI<0b11101, 0b11, 0b1110, 0b1010, 1, def VTOSLS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1110, 0b1010, 1,
(outs SPR:$dst), (ins SPR:$a, fbits32:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits32:$fbits),
IIC_fpCVTSI, "vcvt", ".s32.f32\t$dst, $a, $fbits", []> { IIC_fpCVTSI, "vcvt", ".s32.f32\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -842,7 +864,7 @@ def VTOSLS : AVConv1XI<0b11101, 0b11, 0b1110, 0b1010, 1,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VTOULS : AVConv1XI<0b11101, 0b11, 0b1111, 0b1010, 1, def VTOULS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1111, 0b1010, 1,
(outs SPR:$dst), (ins SPR:$a, fbits32:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits32:$fbits),
IIC_fpCVTSI, "vcvt", ".u32.f32\t$dst, $a, $fbits", []> { IIC_fpCVTSI, "vcvt", ".u32.f32\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -850,25 +872,25 @@ def VTOULS : AVConv1XI<0b11101, 0b11, 0b1111, 0b1010, 1,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VTOSHD : AVConv1XI<0b11101, 0b11, 0b1110, 0b1011, 0, def VTOSHD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1110, 0b1011, 0,
(outs DPR:$dst), (ins DPR:$a, fbits16:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits16:$fbits),
IIC_fpCVTDI, "vcvt", ".s16.f64\t$dst, $a, $fbits", []>; IIC_fpCVTDI, "vcvt", ".s16.f64\t$dst, $a, $fbits", []>;
def VTOUHD : AVConv1XI<0b11101, 0b11, 0b1111, 0b1011, 0, def VTOUHD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1111, 0b1011, 0,
(outs DPR:$dst), (ins DPR:$a, fbits16:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits16:$fbits),
IIC_fpCVTDI, "vcvt", ".u16.f64\t$dst, $a, $fbits", []>; IIC_fpCVTDI, "vcvt", ".u16.f64\t$dst, $a, $fbits", []>;
def VTOSLD : AVConv1XI<0b11101, 0b11, 0b1110, 0b1011, 1, def VTOSLD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1110, 0b1011, 1,
(outs DPR:$dst), (ins DPR:$a, fbits32:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits32:$fbits),
IIC_fpCVTDI, "vcvt", ".s32.f64\t$dst, $a, $fbits", []>; IIC_fpCVTDI, "vcvt", ".s32.f64\t$dst, $a, $fbits", []>;
def VTOULD : AVConv1XI<0b11101, 0b11, 0b1111, 0b1011, 1, def VTOULD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1111, 0b1011, 1,
(outs DPR:$dst), (ins DPR:$a, fbits32:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits32:$fbits),
IIC_fpCVTDI, "vcvt", ".u32.f64\t$dst, $a, $fbits", []>; IIC_fpCVTDI, "vcvt", ".u32.f64\t$dst, $a, $fbits", []>;
// Fixed-Point to FP: // Fixed-Point to FP:
def VSHTOS : AVConv1XI<0b11101, 0b11, 0b1010, 0b1010, 0, def VSHTOS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1010, 0b1010, 0,
(outs SPR:$dst), (ins SPR:$a, fbits16:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits16:$fbits),
IIC_fpCVTIS, "vcvt", ".f32.s16\t$dst, $a, $fbits", []> { IIC_fpCVTIS, "vcvt", ".f32.s16\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -876,7 +898,7 @@ def VSHTOS : AVConv1XI<0b11101, 0b11, 0b1010, 0b1010, 0,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VUHTOS : AVConv1XI<0b11101, 0b11, 0b1011, 0b1010, 0, def VUHTOS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1011, 0b1010, 0,
(outs SPR:$dst), (ins SPR:$a, fbits16:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits16:$fbits),
IIC_fpCVTIS, "vcvt", ".f32.u16\t$dst, $a, $fbits", []> { IIC_fpCVTIS, "vcvt", ".f32.u16\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -884,7 +906,7 @@ def VUHTOS : AVConv1XI<0b11101, 0b11, 0b1011, 0b1010, 0,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VSLTOS : AVConv1XI<0b11101, 0b11, 0b1010, 0b1010, 1, def VSLTOS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1010, 0b1010, 1,
(outs SPR:$dst), (ins SPR:$a, fbits32:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits32:$fbits),
IIC_fpCVTIS, "vcvt", ".f32.s32\t$dst, $a, $fbits", []> { IIC_fpCVTIS, "vcvt", ".f32.s32\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -892,7 +914,7 @@ def VSLTOS : AVConv1XI<0b11101, 0b11, 0b1010, 0b1010, 1,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VULTOS : AVConv1XI<0b11101, 0b11, 0b1011, 0b1010, 1, def VULTOS : AVConv1XInsS_Encode<0b11101, 0b11, 0b1011, 0b1010, 1,
(outs SPR:$dst), (ins SPR:$a, fbits32:$fbits), (outs SPR:$dst), (ins SPR:$a, fbits32:$fbits),
IIC_fpCVTIS, "vcvt", ".f32.u32\t$dst, $a, $fbits", []> { IIC_fpCVTIS, "vcvt", ".f32.u32\t$dst, $a, $fbits", []> {
// Some single precision VFP instructions may be executed on both NEON and // Some single precision VFP instructions may be executed on both NEON and
@ -900,19 +922,19 @@ def VULTOS : AVConv1XI<0b11101, 0b11, 0b1011, 0b1010, 1,
let D = VFPNeonA8Domain; let D = VFPNeonA8Domain;
} }
def VSHTOD : AVConv1XI<0b11101, 0b11, 0b1010, 0b1011, 0, def VSHTOD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1010, 0b1011, 0,
(outs DPR:$dst), (ins DPR:$a, fbits16:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits16:$fbits),
IIC_fpCVTID, "vcvt", ".f64.s16\t$dst, $a, $fbits", []>; IIC_fpCVTID, "vcvt", ".f64.s16\t$dst, $a, $fbits", []>;
def VUHTOD : AVConv1XI<0b11101, 0b11, 0b1011, 0b1011, 0, def VUHTOD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1011, 0b1011, 0,
(outs DPR:$dst), (ins DPR:$a, fbits16:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits16:$fbits),
IIC_fpCVTID, "vcvt", ".f64.u16\t$dst, $a, $fbits", []>; IIC_fpCVTID, "vcvt", ".f64.u16\t$dst, $a, $fbits", []>;
def VSLTOD : AVConv1XI<0b11101, 0b11, 0b1010, 0b1011, 1, def VSLTOD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1010, 0b1011, 1,
(outs DPR:$dst), (ins DPR:$a, fbits32:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits32:$fbits),
IIC_fpCVTID, "vcvt", ".f64.s32\t$dst, $a, $fbits", []>; IIC_fpCVTID, "vcvt", ".f64.s32\t$dst, $a, $fbits", []>;
def VULTOD : AVConv1XI<0b11101, 0b11, 0b1011, 0b1011, 1, def VULTOD : AVConv1XInsD_Encode<0b11101, 0b11, 0b1011, 0b1011, 1,
(outs DPR:$dst), (ins DPR:$a, fbits32:$fbits), (outs DPR:$dst), (ins DPR:$a, fbits32:$fbits),
IIC_fpCVTID, "vcvt", ".f64.u32\t$dst, $a, $fbits", []>; IIC_fpCVTID, "vcvt", ".f64.u32\t$dst, $a, $fbits", []>;

27
llvm/test/MC/ARM/simple-fp-encoding.s
View File

@ -302,15 +302,40 @@
@ VCVT (between floating-point and fixed-point) @ VCVT (between floating-point and fixed-point)
vcvt.f32.u32 s0, s0, #20 vcvt.f32.u32 s0, s0, #20
vcvt.f64.s32 d0, d0, #32 vcvt.f64.s32 d0, d0, #32
vcvt.f32.u16 s0, s0, #1 vcvt.f32.u16 s0, s0, #1
vcvt.f64.s16 d0, d0, #16 vcvt.f64.s16 d0, d0, #16
vcvt.f32.s32 s1, s1, #20
vcvt.f64.u32 d20, d20, #32
vcvt.f32.s16 s17, s17, #1
vcvt.f64.u16 d23, d23, #16
vcvt.u32.f32 s12, s12, #20
vcvt.s32.f64 d2, d2, #32
vcvt.u16.f32 s28, s28, #1
vcvt.s16.f64 d15, d15, #16
vcvt.s32.f32 s1, s1, #20
vcvt.u32.f64 d20, d20, #32
vcvt.s16.f32 s17, s17, #1
vcvt.u16.f64 d23, d23, #16
@ CHECK: vcvt.f32.u32 s0, s0, #20 @ encoding: [0xc6,0x0a,0xbb,0xee] @ CHECK: vcvt.f32.u32 s0, s0, #20 @ encoding: [0xc6,0x0a,0xbb,0xee]
@ CHECK: vcvt.f64.s32 d0, d0, #32 @ encoding: [0xc0,0x0b,0xba,0xee] @ CHECK: vcvt.f64.s32 d0, d0, #32 @ encoding: [0xc0,0x0b,0xba,0xee]
@ CHECK: vcvt.f32.u16 s0, s0, #1 @ encoding: [0x67,0x0a,0xbb,0xee] @ CHECK: vcvt.f32.u16 s0, s0, #1 @ encoding: [0x67,0x0a,0xbb,0xee]
@ CHECK: vcvt.f64.s16 d0, d0, #16 @ encoding: [0x40,0x0b,0xba,0xee] @ CHECK: vcvt.f64.s16 d0, d0, #16 @ encoding: [0x40,0x0b,0xba,0xee]
@ CHECK: vcvt.f32.s32 s1, s1, #20 @ encoding: [0xc6,0x0a,0xfa,0xee]
@ CHECK: vcvt.f64.u32 d20, d20, #32 @ encoding: [0xc0,0x4b,0xfb,0xee]
@ CHECK: vcvt.f32.s16 s17, s17, #1 @ encoding: [0x67,0x8a,0xfa,0xee]
@ CHECK: vcvt.f64.u16 d23, d23, #16 @ encoding: [0x40,0x7b,0xfb,0xee]
@ CHECK: vcvt.u32.f32 s12, s12, #20 @ encoding: [0xc6,0x6a,0xbf,0xee]
@ CHECK: vcvt.s32.f64 d2, d2, #32 @ encoding: [0xc0,0x2b,0xbe,0xee]
@ CHECK: vcvt.u16.f32 s28, s28, #1 @ encoding: [0x67,0xea,0xbf,0xee]
@ CHECK: vcvt.s16.f64 d15, d15, #16 @ encoding: [0x40,0xfb,0xbe,0xee]
@ CHECK: vcvt.s32.f32 s1, s1, #20 @ encoding: [0xc6,0x0a,0xfe,0xee]
@ CHECK: vcvt.u32.f64 d20, d20, #32 @ encoding: [0xc0,0x4b,0xff,0xee]
@ CHECK: vcvt.s16.f32 s17, s17, #1 @ encoding: [0x67,0x8a,0xfe,0xee]
@ CHECK: vcvt.u16.f64 d23, d23, #16 @ encoding: [0x40,0x7b,0xff,0xee]
@ Use NEON to load some f32 immediates that don't fit the f8 representation. @ Use NEON to load some f32 immediates that don't fit the f8 representation.