1 //===-- X86InstrShiftRotate.td - Shift and Rotate Instrs ---*- tablegen -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file describes the shift and rotate instructions.
12 //===----------------------------------------------------------------------===//
14 // FIXME: Someone needs to smear multipattern goodness all over this file.
16 let Defs = [EFLAGS] in {
18 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
20 def SHL8rCL : I<0xD2, MRM4r, (outs GR8 :$dst), (ins GR8 :$src1),
21 "shl{b}\t{%cl, $dst|$dst, cl}",
22 [(set GR8:$dst, (shl GR8:$src1, CL))], IIC_SR>;
23 def SHL16rCL : I<0xD3, MRM4r, (outs GR16:$dst), (ins GR16:$src1),
24 "shl{w}\t{%cl, $dst|$dst, cl}",
25 [(set GR16:$dst, (shl GR16:$src1, CL))], IIC_SR>, OpSize16;
26 def SHL32rCL : I<0xD3, MRM4r, (outs GR32:$dst), (ins GR32:$src1),
27 "shl{l}\t{%cl, $dst|$dst, cl}",
28 [(set GR32:$dst, (shl GR32:$src1, CL))], IIC_SR>, OpSize32;
29 def SHL64rCL : RI<0xD3, MRM4r, (outs GR64:$dst), (ins GR64:$src1),
30 "shl{q}\t{%cl, $dst|$dst, cl}",
31 [(set GR64:$dst, (shl GR64:$src1, CL))], IIC_SR>;
34 def SHL8ri : Ii8<0xC0, MRM4r, (outs GR8 :$dst), (ins GR8 :$src1, u8imm:$src2),
35 "shl{b}\t{$src2, $dst|$dst, $src2}",
36 [(set GR8:$dst, (shl GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
38 let isConvertibleToThreeAddress = 1 in { // Can transform into LEA.
39 def SHL16ri : Ii8<0xC1, MRM4r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$src2),
40 "shl{w}\t{$src2, $dst|$dst, $src2}",
41 [(set GR16:$dst, (shl GR16:$src1, (i8 imm:$src2)))], IIC_SR>,
43 def SHL32ri : Ii8<0xC1, MRM4r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$src2),
44 "shl{l}\t{$src2, $dst|$dst, $src2}",
45 [(set GR32:$dst, (shl GR32:$src1, (i8 imm:$src2)))], IIC_SR>,
47 def SHL64ri : RIi8<0xC1, MRM4r, (outs GR64:$dst),
48 (ins GR64:$src1, u8imm:$src2),
49 "shl{q}\t{$src2, $dst|$dst, $src2}",
50 [(set GR64:$dst, (shl GR64:$src1, (i8 imm:$src2)))],
52 } // isConvertibleToThreeAddress = 1
54 // NOTE: We don't include patterns for shifts of a register by one, because
55 // 'add reg,reg' is cheaper (and we have a Pat pattern for shift-by-one).
56 let hasSideEffects = 0 in {
57 def SHL8r1 : I<0xD0, MRM4r, (outs GR8:$dst), (ins GR8:$src1),
58 "shl{b}\t$dst", [], IIC_SR>;
59 def SHL16r1 : I<0xD1, MRM4r, (outs GR16:$dst), (ins GR16:$src1),
60 "shl{w}\t$dst", [], IIC_SR>, OpSize16;
61 def SHL32r1 : I<0xD1, MRM4r, (outs GR32:$dst), (ins GR32:$src1),
62 "shl{l}\t$dst", [], IIC_SR>, OpSize32;
63 def SHL64r1 : RI<0xD1, MRM4r, (outs GR64:$dst), (ins GR64:$src1),
64 "shl{q}\t$dst", [], IIC_SR>;
65 } // hasSideEffects = 0
66 } // Constraints = "$src = $dst", SchedRW
69 let SchedRW = [WriteShiftLd, WriteRMW] in {
70 // FIXME: Why do we need an explicit "Uses = [CL]" when the instr has a pattern
73 def SHL8mCL : I<0xD2, MRM4m, (outs), (ins i8mem :$dst),
74 "shl{b}\t{%cl, $dst|$dst, cl}",
75 [(store (shl (loadi8 addr:$dst), CL), addr:$dst)], IIC_SR>;
76 def SHL16mCL : I<0xD3, MRM4m, (outs), (ins i16mem:$dst),
77 "shl{w}\t{%cl, $dst|$dst, cl}",
78 [(store (shl (loadi16 addr:$dst), CL), addr:$dst)], IIC_SR>,
80 def SHL32mCL : I<0xD3, MRM4m, (outs), (ins i32mem:$dst),
81 "shl{l}\t{%cl, $dst|$dst, cl}",
82 [(store (shl (loadi32 addr:$dst), CL), addr:$dst)], IIC_SR>,
84 def SHL64mCL : RI<0xD3, MRM4m, (outs), (ins i64mem:$dst),
85 "shl{q}\t{%cl, $dst|$dst, cl}",
86 [(store (shl (loadi64 addr:$dst), CL), addr:$dst)], IIC_SR>;
88 def SHL8mi : Ii8<0xC0, MRM4m, (outs), (ins i8mem :$dst, u8imm:$src),
89 "shl{b}\t{$src, $dst|$dst, $src}",
90 [(store (shl (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
92 def SHL16mi : Ii8<0xC1, MRM4m, (outs), (ins i16mem:$dst, u8imm:$src),
93 "shl{w}\t{$src, $dst|$dst, $src}",
94 [(store (shl (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
96 def SHL32mi : Ii8<0xC1, MRM4m, (outs), (ins i32mem:$dst, u8imm:$src),
97 "shl{l}\t{$src, $dst|$dst, $src}",
98 [(store (shl (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
100 def SHL64mi : RIi8<0xC1, MRM4m, (outs), (ins i64mem:$dst, u8imm:$src),
101 "shl{q}\t{$src, $dst|$dst, $src}",
102 [(store (shl (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
106 def SHL8m1 : I<0xD0, MRM4m, (outs), (ins i8mem :$dst),
108 [(store (shl (loadi8 addr:$dst), (i8 1)), addr:$dst)],
110 def SHL16m1 : I<0xD1, MRM4m, (outs), (ins i16mem:$dst),
112 [(store (shl (loadi16 addr:$dst), (i8 1)), addr:$dst)],
114 def SHL32m1 : I<0xD1, MRM4m, (outs), (ins i32mem:$dst),
116 [(store (shl (loadi32 addr:$dst), (i8 1)), addr:$dst)],
118 def SHL64m1 : RI<0xD1, MRM4m, (outs), (ins i64mem:$dst),
120 [(store (shl (loadi64 addr:$dst), (i8 1)), addr:$dst)],
124 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
126 def SHR8rCL : I<0xD2, MRM5r, (outs GR8 :$dst), (ins GR8 :$src1),
127 "shr{b}\t{%cl, $dst|$dst, cl}",
128 [(set GR8:$dst, (srl GR8:$src1, CL))], IIC_SR>;
129 def SHR16rCL : I<0xD3, MRM5r, (outs GR16:$dst), (ins GR16:$src1),
130 "shr{w}\t{%cl, $dst|$dst, cl}",
131 [(set GR16:$dst, (srl GR16:$src1, CL))], IIC_SR>, OpSize16;
132 def SHR32rCL : I<0xD3, MRM5r, (outs GR32:$dst), (ins GR32:$src1),
133 "shr{l}\t{%cl, $dst|$dst, cl}",
134 [(set GR32:$dst, (srl GR32:$src1, CL))], IIC_SR>, OpSize32;
135 def SHR64rCL : RI<0xD3, MRM5r, (outs GR64:$dst), (ins GR64:$src1),
136 "shr{q}\t{%cl, $dst|$dst, cl}",
137 [(set GR64:$dst, (srl GR64:$src1, CL))], IIC_SR>;
140 def SHR8ri : Ii8<0xC0, MRM5r, (outs GR8:$dst), (ins GR8:$src1, u8imm:$src2),
141 "shr{b}\t{$src2, $dst|$dst, $src2}",
142 [(set GR8:$dst, (srl GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
143 def SHR16ri : Ii8<0xC1, MRM5r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$src2),
144 "shr{w}\t{$src2, $dst|$dst, $src2}",
145 [(set GR16:$dst, (srl GR16:$src1, (i8 imm:$src2)))],
147 def SHR32ri : Ii8<0xC1, MRM5r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$src2),
148 "shr{l}\t{$src2, $dst|$dst, $src2}",
149 [(set GR32:$dst, (srl GR32:$src1, (i8 imm:$src2)))],
151 def SHR64ri : RIi8<0xC1, MRM5r, (outs GR64:$dst), (ins GR64:$src1, u8imm:$src2),
152 "shr{q}\t{$src2, $dst|$dst, $src2}",
153 [(set GR64:$dst, (srl GR64:$src1, (i8 imm:$src2)))], IIC_SR>;
156 def SHR8r1 : I<0xD0, MRM5r, (outs GR8:$dst), (ins GR8:$src1),
158 [(set GR8:$dst, (srl GR8:$src1, (i8 1)))], IIC_SR>;
159 def SHR16r1 : I<0xD1, MRM5r, (outs GR16:$dst), (ins GR16:$src1),
161 [(set GR16:$dst, (srl GR16:$src1, (i8 1)))], IIC_SR>, OpSize16;
162 def SHR32r1 : I<0xD1, MRM5r, (outs GR32:$dst), (ins GR32:$src1),
164 [(set GR32:$dst, (srl GR32:$src1, (i8 1)))], IIC_SR>, OpSize32;
165 def SHR64r1 : RI<0xD1, MRM5r, (outs GR64:$dst), (ins GR64:$src1),
167 [(set GR64:$dst, (srl GR64:$src1, (i8 1)))], IIC_SR>;
168 } // Constraints = "$src = $dst", SchedRW
171 let SchedRW = [WriteShiftLd, WriteRMW] in {
173 def SHR8mCL : I<0xD2, MRM5m, (outs), (ins i8mem :$dst),
174 "shr{b}\t{%cl, $dst|$dst, cl}",
175 [(store (srl (loadi8 addr:$dst), CL), addr:$dst)], IIC_SR>;
176 def SHR16mCL : I<0xD3, MRM5m, (outs), (ins i16mem:$dst),
177 "shr{w}\t{%cl, $dst|$dst, cl}",
178 [(store (srl (loadi16 addr:$dst), CL), addr:$dst)], IIC_SR>,
180 def SHR32mCL : I<0xD3, MRM5m, (outs), (ins i32mem:$dst),
181 "shr{l}\t{%cl, $dst|$dst, cl}",
182 [(store (srl (loadi32 addr:$dst), CL), addr:$dst)], IIC_SR>,
184 def SHR64mCL : RI<0xD3, MRM5m, (outs), (ins i64mem:$dst),
185 "shr{q}\t{%cl, $dst|$dst, cl}",
186 [(store (srl (loadi64 addr:$dst), CL), addr:$dst)], IIC_SR>;
188 def SHR8mi : Ii8<0xC0, MRM5m, (outs), (ins i8mem :$dst, u8imm:$src),
189 "shr{b}\t{$src, $dst|$dst, $src}",
190 [(store (srl (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
192 def SHR16mi : Ii8<0xC1, MRM5m, (outs), (ins i16mem:$dst, u8imm:$src),
193 "shr{w}\t{$src, $dst|$dst, $src}",
194 [(store (srl (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
196 def SHR32mi : Ii8<0xC1, MRM5m, (outs), (ins i32mem:$dst, u8imm:$src),
197 "shr{l}\t{$src, $dst|$dst, $src}",
198 [(store (srl (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
200 def SHR64mi : RIi8<0xC1, MRM5m, (outs), (ins i64mem:$dst, u8imm:$src),
201 "shr{q}\t{$src, $dst|$dst, $src}",
202 [(store (srl (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
206 def SHR8m1 : I<0xD0, MRM5m, (outs), (ins i8mem :$dst),
208 [(store (srl (loadi8 addr:$dst), (i8 1)), addr:$dst)],
210 def SHR16m1 : I<0xD1, MRM5m, (outs), (ins i16mem:$dst),
212 [(store (srl (loadi16 addr:$dst), (i8 1)), addr:$dst)],
214 def SHR32m1 : I<0xD1, MRM5m, (outs), (ins i32mem:$dst),
216 [(store (srl (loadi32 addr:$dst), (i8 1)), addr:$dst)],
218 def SHR64m1 : RI<0xD1, MRM5m, (outs), (ins i64mem:$dst),
220 [(store (srl (loadi64 addr:$dst), (i8 1)), addr:$dst)],
224 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
226 def SAR8rCL : I<0xD2, MRM7r, (outs GR8 :$dst), (ins GR8 :$src1),
227 "sar{b}\t{%cl, $dst|$dst, cl}",
228 [(set GR8:$dst, (sra GR8:$src1, CL))],
230 def SAR16rCL : I<0xD3, MRM7r, (outs GR16:$dst), (ins GR16:$src1),
231 "sar{w}\t{%cl, $dst|$dst, cl}",
232 [(set GR16:$dst, (sra GR16:$src1, CL))],
234 def SAR32rCL : I<0xD3, MRM7r, (outs GR32:$dst), (ins GR32:$src1),
235 "sar{l}\t{%cl, $dst|$dst, cl}",
236 [(set GR32:$dst, (sra GR32:$src1, CL))],
238 def SAR64rCL : RI<0xD3, MRM7r, (outs GR64:$dst), (ins GR64:$src1),
239 "sar{q}\t{%cl, $dst|$dst, cl}",
240 [(set GR64:$dst, (sra GR64:$src1, CL))],
244 def SAR8ri : Ii8<0xC0, MRM7r, (outs GR8 :$dst), (ins GR8 :$src1, u8imm:$src2),
245 "sar{b}\t{$src2, $dst|$dst, $src2}",
246 [(set GR8:$dst, (sra GR8:$src1, (i8 imm:$src2)))],
248 def SAR16ri : Ii8<0xC1, MRM7r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$src2),
249 "sar{w}\t{$src2, $dst|$dst, $src2}",
250 [(set GR16:$dst, (sra GR16:$src1, (i8 imm:$src2)))],
252 def SAR32ri : Ii8<0xC1, MRM7r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$src2),
253 "sar{l}\t{$src2, $dst|$dst, $src2}",
254 [(set GR32:$dst, (sra GR32:$src1, (i8 imm:$src2)))],
256 def SAR64ri : RIi8<0xC1, MRM7r, (outs GR64:$dst),
257 (ins GR64:$src1, u8imm:$src2),
258 "sar{q}\t{$src2, $dst|$dst, $src2}",
259 [(set GR64:$dst, (sra GR64:$src1, (i8 imm:$src2)))],
263 def SAR8r1 : I<0xD0, MRM7r, (outs GR8 :$dst), (ins GR8 :$src1),
265 [(set GR8:$dst, (sra GR8:$src1, (i8 1)))],
267 def SAR16r1 : I<0xD1, MRM7r, (outs GR16:$dst), (ins GR16:$src1),
269 [(set GR16:$dst, (sra GR16:$src1, (i8 1)))],
271 def SAR32r1 : I<0xD1, MRM7r, (outs GR32:$dst), (ins GR32:$src1),
273 [(set GR32:$dst, (sra GR32:$src1, (i8 1)))],
275 def SAR64r1 : RI<0xD1, MRM7r, (outs GR64:$dst), (ins GR64:$src1),
277 [(set GR64:$dst, (sra GR64:$src1, (i8 1)))],
279 } // Constraints = "$src = $dst", SchedRW
282 let SchedRW = [WriteShiftLd, WriteRMW] in {
284 def SAR8mCL : I<0xD2, MRM7m, (outs), (ins i8mem :$dst),
285 "sar{b}\t{%cl, $dst|$dst, cl}",
286 [(store (sra (loadi8 addr:$dst), CL), addr:$dst)],
288 def SAR16mCL : I<0xD3, MRM7m, (outs), (ins i16mem:$dst),
289 "sar{w}\t{%cl, $dst|$dst, cl}",
290 [(store (sra (loadi16 addr:$dst), CL), addr:$dst)],
292 def SAR32mCL : I<0xD3, MRM7m, (outs), (ins i32mem:$dst),
293 "sar{l}\t{%cl, $dst|$dst, cl}",
294 [(store (sra (loadi32 addr:$dst), CL), addr:$dst)],
296 def SAR64mCL : RI<0xD3, MRM7m, (outs), (ins i64mem:$dst),
297 "sar{q}\t{%cl, $dst|$dst, cl}",
298 [(store (sra (loadi64 addr:$dst), CL), addr:$dst)],
301 def SAR8mi : Ii8<0xC0, MRM7m, (outs), (ins i8mem :$dst, u8imm:$src),
302 "sar{b}\t{$src, $dst|$dst, $src}",
303 [(store (sra (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
305 def SAR16mi : Ii8<0xC1, MRM7m, (outs), (ins i16mem:$dst, u8imm:$src),
306 "sar{w}\t{$src, $dst|$dst, $src}",
307 [(store (sra (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
309 def SAR32mi : Ii8<0xC1, MRM7m, (outs), (ins i32mem:$dst, u8imm:$src),
310 "sar{l}\t{$src, $dst|$dst, $src}",
311 [(store (sra (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
313 def SAR64mi : RIi8<0xC1, MRM7m, (outs), (ins i64mem:$dst, u8imm:$src),
314 "sar{q}\t{$src, $dst|$dst, $src}",
315 [(store (sra (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
319 def SAR8m1 : I<0xD0, MRM7m, (outs), (ins i8mem :$dst),
321 [(store (sra (loadi8 addr:$dst), (i8 1)), addr:$dst)],
323 def SAR16m1 : I<0xD1, MRM7m, (outs), (ins i16mem:$dst),
325 [(store (sra (loadi16 addr:$dst), (i8 1)), addr:$dst)],
327 def SAR32m1 : I<0xD1, MRM7m, (outs), (ins i32mem:$dst),
329 [(store (sra (loadi32 addr:$dst), (i8 1)), addr:$dst)],
331 def SAR64m1 : RI<0xD1, MRM7m, (outs), (ins i64mem:$dst),
333 [(store (sra (loadi64 addr:$dst), (i8 1)), addr:$dst)],
337 //===----------------------------------------------------------------------===//
338 // Rotate instructions
339 //===----------------------------------------------------------------------===//
341 let hasSideEffects = 0 in {
342 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
344 let Uses = [CL, EFLAGS] in {
345 def RCL8rCL : I<0xD2, MRM2r, (outs GR8:$dst), (ins GR8:$src1),
346 "rcl{b}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
347 def RCL16rCL : I<0xD3, MRM2r, (outs GR16:$dst), (ins GR16:$src1),
348 "rcl{w}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize16;
349 def RCL32rCL : I<0xD3, MRM2r, (outs GR32:$dst), (ins GR32:$src1),
350 "rcl{l}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize32;
351 def RCL64rCL : RI<0xD3, MRM2r, (outs GR64:$dst), (ins GR64:$src1),
352 "rcl{q}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
353 } // Uses = [CL, EFLAGS]
355 let Uses = [EFLAGS] in {
356 def RCL8r1 : I<0xD0, MRM2r, (outs GR8:$dst), (ins GR8:$src1),
357 "rcl{b}\t$dst", [], IIC_SR>;
358 def RCL8ri : Ii8<0xC0, MRM2r, (outs GR8:$dst), (ins GR8:$src1, u8imm:$cnt),
359 "rcl{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
360 def RCL16r1 : I<0xD1, MRM2r, (outs GR16:$dst), (ins GR16:$src1),
361 "rcl{w}\t$dst", [], IIC_SR>, OpSize16;
362 def RCL16ri : Ii8<0xC1, MRM2r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$cnt),
363 "rcl{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize16;
364 def RCL32r1 : I<0xD1, MRM2r, (outs GR32:$dst), (ins GR32:$src1),
365 "rcl{l}\t$dst", [], IIC_SR>, OpSize32;
366 def RCL32ri : Ii8<0xC1, MRM2r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$cnt),
367 "rcl{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize32;
368 def RCL64r1 : RI<0xD1, MRM2r, (outs GR64:$dst), (ins GR64:$src1),
369 "rcl{q}\t$dst", [], IIC_SR>;
370 def RCL64ri : RIi8<0xC1, MRM2r, (outs GR64:$dst), (ins GR64:$src1, u8imm:$cnt),
371 "rcl{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
374 let Uses = [CL, EFLAGS] in {
375 def RCR8rCL : I<0xD2, MRM3r, (outs GR8:$dst), (ins GR8:$src1),
376 "rcr{b}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
377 def RCR16rCL : I<0xD3, MRM3r, (outs GR16:$dst), (ins GR16:$src1),
378 "rcr{w}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize16;
379 def RCR32rCL : I<0xD3, MRM3r, (outs GR32:$dst), (ins GR32:$src1),
380 "rcr{l}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize32;
381 def RCR64rCL : RI<0xD3, MRM3r, (outs GR64:$dst), (ins GR64:$src1),
382 "rcr{q}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
383 } // Uses = [CL, EFLAGS]
385 let Uses = [EFLAGS] in {
386 def RCR8r1 : I<0xD0, MRM3r, (outs GR8:$dst), (ins GR8:$src1),
387 "rcr{b}\t$dst", [], IIC_SR>;
388 def RCR8ri : Ii8<0xC0, MRM3r, (outs GR8:$dst), (ins GR8:$src1, u8imm:$cnt),
389 "rcr{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
390 def RCR16r1 : I<0xD1, MRM3r, (outs GR16:$dst), (ins GR16:$src1),
391 "rcr{w}\t$dst", [], IIC_SR>, OpSize16;
392 def RCR16ri : Ii8<0xC1, MRM3r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$cnt),
393 "rcr{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize16;
394 def RCR32r1 : I<0xD1, MRM3r, (outs GR32:$dst), (ins GR32:$src1),
395 "rcr{l}\t$dst", [], IIC_SR>, OpSize32;
396 def RCR32ri : Ii8<0xC1, MRM3r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$cnt),
397 "rcr{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize32;
398 def RCR64r1 : RI<0xD1, MRM3r, (outs GR64:$dst), (ins GR64:$src1),
399 "rcr{q}\t$dst", [], IIC_SR>;
400 def RCR64ri : RIi8<0xC1, MRM3r, (outs GR64:$dst), (ins GR64:$src1, u8imm:$cnt),
401 "rcr{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
404 } // Constraints = "$src = $dst"
406 let SchedRW = [WriteShiftLd, WriteRMW], mayStore = 1 in {
407 let Uses = [EFLAGS] in {
408 def RCL8m1 : I<0xD0, MRM2m, (outs), (ins i8mem:$dst),
409 "rcl{b}\t$dst", [], IIC_SR>;
410 def RCL8mi : Ii8<0xC0, MRM2m, (outs), (ins i8mem:$dst, u8imm:$cnt),
411 "rcl{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
412 def RCL16m1 : I<0xD1, MRM2m, (outs), (ins i16mem:$dst),
413 "rcl{w}\t$dst", [], IIC_SR>, OpSize16;
414 def RCL16mi : Ii8<0xC1, MRM2m, (outs), (ins i16mem:$dst, u8imm:$cnt),
415 "rcl{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize16;
416 def RCL32m1 : I<0xD1, MRM2m, (outs), (ins i32mem:$dst),
417 "rcl{l}\t$dst", [], IIC_SR>, OpSize32;
418 def RCL32mi : Ii8<0xC1, MRM2m, (outs), (ins i32mem:$dst, u8imm:$cnt),
419 "rcl{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize32;
420 def RCL64m1 : RI<0xD1, MRM2m, (outs), (ins i64mem:$dst),
421 "rcl{q}\t$dst", [], IIC_SR>;
422 def RCL64mi : RIi8<0xC1, MRM2m, (outs), (ins i64mem:$dst, u8imm:$cnt),
423 "rcl{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
425 def RCR8m1 : I<0xD0, MRM3m, (outs), (ins i8mem:$dst),
426 "rcr{b}\t$dst", [], IIC_SR>;
427 def RCR8mi : Ii8<0xC0, MRM3m, (outs), (ins i8mem:$dst, u8imm:$cnt),
428 "rcr{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
429 def RCR16m1 : I<0xD1, MRM3m, (outs), (ins i16mem:$dst),
430 "rcr{w}\t$dst", [], IIC_SR>, OpSize16;
431 def RCR16mi : Ii8<0xC1, MRM3m, (outs), (ins i16mem:$dst, u8imm:$cnt),
432 "rcr{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize16;
433 def RCR32m1 : I<0xD1, MRM3m, (outs), (ins i32mem:$dst),
434 "rcr{l}\t$dst", [], IIC_SR>, OpSize32;
435 def RCR32mi : Ii8<0xC1, MRM3m, (outs), (ins i32mem:$dst, u8imm:$cnt),
436 "rcr{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize32;
437 def RCR64m1 : RI<0xD1, MRM3m, (outs), (ins i64mem:$dst),
438 "rcr{q}\t$dst", [], IIC_SR>;
439 def RCR64mi : RIi8<0xC1, MRM3m, (outs), (ins i64mem:$dst, u8imm:$cnt),
440 "rcr{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
443 let Uses = [CL, EFLAGS] in {
444 def RCL8mCL : I<0xD2, MRM2m, (outs), (ins i8mem:$dst),
445 "rcl{b}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
446 def RCL16mCL : I<0xD3, MRM2m, (outs), (ins i16mem:$dst),
447 "rcl{w}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize16;
448 def RCL32mCL : I<0xD3, MRM2m, (outs), (ins i32mem:$dst),
449 "rcl{l}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize32;
450 def RCL64mCL : RI<0xD3, MRM2m, (outs), (ins i64mem:$dst),
451 "rcl{q}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
453 def RCR8mCL : I<0xD2, MRM3m, (outs), (ins i8mem:$dst),
454 "rcr{b}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
455 def RCR16mCL : I<0xD3, MRM3m, (outs), (ins i16mem:$dst),
456 "rcr{w}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize16;
457 def RCR32mCL : I<0xD3, MRM3m, (outs), (ins i32mem:$dst),
458 "rcr{l}\t{%cl, $dst|$dst, cl}", [], IIC_SR>, OpSize32;
459 def RCR64mCL : RI<0xD3, MRM3m, (outs), (ins i64mem:$dst),
460 "rcr{q}\t{%cl, $dst|$dst, cl}", [], IIC_SR>;
461 } // Uses = [CL, EFLAGS]
463 } // hasSideEffects = 0
465 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
466 // FIXME: provide shorter instructions when imm8 == 1
468 def ROL8rCL : I<0xD2, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1),
469 "rol{b}\t{%cl, $dst|$dst, cl}",
470 [(set GR8:$dst, (rotl GR8:$src1, CL))], IIC_SR>;
471 def ROL16rCL : I<0xD3, MRM0r, (outs GR16:$dst), (ins GR16:$src1),
472 "rol{w}\t{%cl, $dst|$dst, cl}",
473 [(set GR16:$dst, (rotl GR16:$src1, CL))], IIC_SR>, OpSize16;
474 def ROL32rCL : I<0xD3, MRM0r, (outs GR32:$dst), (ins GR32:$src1),
475 "rol{l}\t{%cl, $dst|$dst, cl}",
476 [(set GR32:$dst, (rotl GR32:$src1, CL))], IIC_SR>, OpSize32;
477 def ROL64rCL : RI<0xD3, MRM0r, (outs GR64:$dst), (ins GR64:$src1),
478 "rol{q}\t{%cl, $dst|$dst, cl}",
479 [(set GR64:$dst, (rotl GR64:$src1, CL))], IIC_SR>;
482 def ROL8ri : Ii8<0xC0, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1, u8imm:$src2),
483 "rol{b}\t{$src2, $dst|$dst, $src2}",
484 [(set GR8:$dst, (rotl GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
485 def ROL16ri : Ii8<0xC1, MRM0r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$src2),
486 "rol{w}\t{$src2, $dst|$dst, $src2}",
487 [(set GR16:$dst, (rotl GR16:$src1, (i8 imm:$src2)))],
489 def ROL32ri : Ii8<0xC1, MRM0r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$src2),
490 "rol{l}\t{$src2, $dst|$dst, $src2}",
491 [(set GR32:$dst, (rotl GR32:$src1, (i8 imm:$src2)))],
493 def ROL64ri : RIi8<0xC1, MRM0r, (outs GR64:$dst),
494 (ins GR64:$src1, u8imm:$src2),
495 "rol{q}\t{$src2, $dst|$dst, $src2}",
496 [(set GR64:$dst, (rotl GR64:$src1, (i8 imm:$src2)))],
500 def ROL8r1 : I<0xD0, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1),
502 [(set GR8:$dst, (rotl GR8:$src1, (i8 1)))],
504 def ROL16r1 : I<0xD1, MRM0r, (outs GR16:$dst), (ins GR16:$src1),
506 [(set GR16:$dst, (rotl GR16:$src1, (i8 1)))],
508 def ROL32r1 : I<0xD1, MRM0r, (outs GR32:$dst), (ins GR32:$src1),
510 [(set GR32:$dst, (rotl GR32:$src1, (i8 1)))],
512 def ROL64r1 : RI<0xD1, MRM0r, (outs GR64:$dst), (ins GR64:$src1),
514 [(set GR64:$dst, (rotl GR64:$src1, (i8 1)))],
516 } // Constraints = "$src = $dst", SchedRW
518 let SchedRW = [WriteShiftLd, WriteRMW] in {
520 def ROL8mCL : I<0xD2, MRM0m, (outs), (ins i8mem :$dst),
521 "rol{b}\t{%cl, $dst|$dst, cl}",
522 [(store (rotl (loadi8 addr:$dst), CL), addr:$dst)],
524 def ROL16mCL : I<0xD3, MRM0m, (outs), (ins i16mem:$dst),
525 "rol{w}\t{%cl, $dst|$dst, cl}",
526 [(store (rotl (loadi16 addr:$dst), CL), addr:$dst)],
528 def ROL32mCL : I<0xD3, MRM0m, (outs), (ins i32mem:$dst),
529 "rol{l}\t{%cl, $dst|$dst, cl}",
530 [(store (rotl (loadi32 addr:$dst), CL), addr:$dst)],
532 def ROL64mCL : RI<0xD3, MRM0m, (outs), (ins i64mem:$dst),
533 "rol{q}\t{%cl, $dst|$dst, cl}",
534 [(store (rotl (loadi64 addr:$dst), CL), addr:$dst)],
537 def ROL8mi : Ii8<0xC0, MRM0m, (outs), (ins i8mem :$dst, u8imm:$src1),
538 "rol{b}\t{$src1, $dst|$dst, $src1}",
539 [(store (rotl (loadi8 addr:$dst), (i8 imm:$src1)), addr:$dst)],
541 def ROL16mi : Ii8<0xC1, MRM0m, (outs), (ins i16mem:$dst, u8imm:$src1),
542 "rol{w}\t{$src1, $dst|$dst, $src1}",
543 [(store (rotl (loadi16 addr:$dst), (i8 imm:$src1)), addr:$dst)],
545 def ROL32mi : Ii8<0xC1, MRM0m, (outs), (ins i32mem:$dst, u8imm:$src1),
546 "rol{l}\t{$src1, $dst|$dst, $src1}",
547 [(store (rotl (loadi32 addr:$dst), (i8 imm:$src1)), addr:$dst)],
549 def ROL64mi : RIi8<0xC1, MRM0m, (outs), (ins i64mem:$dst, u8imm:$src1),
550 "rol{q}\t{$src1, $dst|$dst, $src1}",
551 [(store (rotl (loadi64 addr:$dst), (i8 imm:$src1)), addr:$dst)],
555 def ROL8m1 : I<0xD0, MRM0m, (outs), (ins i8mem :$dst),
557 [(store (rotl (loadi8 addr:$dst), (i8 1)), addr:$dst)],
559 def ROL16m1 : I<0xD1, MRM0m, (outs), (ins i16mem:$dst),
561 [(store (rotl (loadi16 addr:$dst), (i8 1)), addr:$dst)],
563 def ROL32m1 : I<0xD1, MRM0m, (outs), (ins i32mem:$dst),
565 [(store (rotl (loadi32 addr:$dst), (i8 1)), addr:$dst)],
567 def ROL64m1 : RI<0xD1, MRM0m, (outs), (ins i64mem:$dst),
569 [(store (rotl (loadi64 addr:$dst), (i8 1)), addr:$dst)],
573 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
575 def ROR8rCL : I<0xD2, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1),
576 "ror{b}\t{%cl, $dst|$dst, cl}",
577 [(set GR8:$dst, (rotr GR8:$src1, CL))], IIC_SR>;
578 def ROR16rCL : I<0xD3, MRM1r, (outs GR16:$dst), (ins GR16:$src1),
579 "ror{w}\t{%cl, $dst|$dst, cl}",
580 [(set GR16:$dst, (rotr GR16:$src1, CL))], IIC_SR>, OpSize16;
581 def ROR32rCL : I<0xD3, MRM1r, (outs GR32:$dst), (ins GR32:$src1),
582 "ror{l}\t{%cl, $dst|$dst, cl}",
583 [(set GR32:$dst, (rotr GR32:$src1, CL))], IIC_SR>, OpSize32;
584 def ROR64rCL : RI<0xD3, MRM1r, (outs GR64:$dst), (ins GR64:$src1),
585 "ror{q}\t{%cl, $dst|$dst, cl}",
586 [(set GR64:$dst, (rotr GR64:$src1, CL))], IIC_SR>;
589 def ROR8ri : Ii8<0xC0, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1, u8imm:$src2),
590 "ror{b}\t{$src2, $dst|$dst, $src2}",
591 [(set GR8:$dst, (rotr GR8:$src1, (i8 relocImm:$src2)))],
593 def ROR16ri : Ii8<0xC1, MRM1r, (outs GR16:$dst), (ins GR16:$src1, u8imm:$src2),
594 "ror{w}\t{$src2, $dst|$dst, $src2}",
595 [(set GR16:$dst, (rotr GR16:$src1, (i8 relocImm:$src2)))],
597 def ROR32ri : Ii8<0xC1, MRM1r, (outs GR32:$dst), (ins GR32:$src1, u8imm:$src2),
598 "ror{l}\t{$src2, $dst|$dst, $src2}",
599 [(set GR32:$dst, (rotr GR32:$src1, (i8 relocImm:$src2)))],
601 def ROR64ri : RIi8<0xC1, MRM1r, (outs GR64:$dst),
602 (ins GR64:$src1, u8imm:$src2),
603 "ror{q}\t{$src2, $dst|$dst, $src2}",
604 [(set GR64:$dst, (rotr GR64:$src1, (i8 relocImm:$src2)))],
608 def ROR8r1 : I<0xD0, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1),
610 [(set GR8:$dst, (rotl GR8:$src1, (i8 7)))],
612 def ROR16r1 : I<0xD1, MRM1r, (outs GR16:$dst), (ins GR16:$src1),
614 [(set GR16:$dst, (rotl GR16:$src1, (i8 15)))],
616 def ROR32r1 : I<0xD1, MRM1r, (outs GR32:$dst), (ins GR32:$src1),
618 [(set GR32:$dst, (rotl GR32:$src1, (i8 31)))],
620 def ROR64r1 : RI<0xD1, MRM1r, (outs GR64:$dst), (ins GR64:$src1),
622 [(set GR64:$dst, (rotl GR64:$src1, (i8 63)))],
624 } // Constraints = "$src = $dst", SchedRW
626 let SchedRW = [WriteShiftLd, WriteRMW] in {
628 def ROR8mCL : I<0xD2, MRM1m, (outs), (ins i8mem :$dst),
629 "ror{b}\t{%cl, $dst|$dst, cl}",
630 [(store (rotr (loadi8 addr:$dst), CL), addr:$dst)],
632 def ROR16mCL : I<0xD3, MRM1m, (outs), (ins i16mem:$dst),
633 "ror{w}\t{%cl, $dst|$dst, cl}",
634 [(store (rotr (loadi16 addr:$dst), CL), addr:$dst)],
636 def ROR32mCL : I<0xD3, MRM1m, (outs), (ins i32mem:$dst),
637 "ror{l}\t{%cl, $dst|$dst, cl}",
638 [(store (rotr (loadi32 addr:$dst), CL), addr:$dst)],
640 def ROR64mCL : RI<0xD3, MRM1m, (outs), (ins i64mem:$dst),
641 "ror{q}\t{%cl, $dst|$dst, cl}",
642 [(store (rotr (loadi64 addr:$dst), CL), addr:$dst)],
645 def ROR8mi : Ii8<0xC0, MRM1m, (outs), (ins i8mem :$dst, u8imm:$src),
646 "ror{b}\t{$src, $dst|$dst, $src}",
647 [(store (rotr (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
649 def ROR16mi : Ii8<0xC1, MRM1m, (outs), (ins i16mem:$dst, u8imm:$src),
650 "ror{w}\t{$src, $dst|$dst, $src}",
651 [(store (rotr (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
653 def ROR32mi : Ii8<0xC1, MRM1m, (outs), (ins i32mem:$dst, u8imm:$src),
654 "ror{l}\t{$src, $dst|$dst, $src}",
655 [(store (rotr (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
657 def ROR64mi : RIi8<0xC1, MRM1m, (outs), (ins i64mem:$dst, u8imm:$src),
658 "ror{q}\t{$src, $dst|$dst, $src}",
659 [(store (rotr (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
663 def ROR8m1 : I<0xD0, MRM1m, (outs), (ins i8mem :$dst),
665 [(store (rotl (loadi8 addr:$dst), (i8 7)), addr:$dst)],
667 def ROR16m1 : I<0xD1, MRM1m, (outs), (ins i16mem:$dst),
669 [(store (rotl (loadi16 addr:$dst), (i8 15)), addr:$dst)],
671 def ROR32m1 : I<0xD1, MRM1m, (outs), (ins i32mem:$dst),
673 [(store (rotl (loadi32 addr:$dst), (i8 31)), addr:$dst)],
675 def ROR64m1 : RI<0xD1, MRM1m, (outs), (ins i64mem:$dst),
677 [(store (rotl (loadi64 addr:$dst), (i8 63)), addr:$dst)],
682 //===----------------------------------------------------------------------===//
683 // Double shift instructions (generalizations of rotate)
684 //===----------------------------------------------------------------------===//
686 let Constraints = "$src1 = $dst", SchedRW = [WriteShift] in {
689 def SHLD16rrCL : I<0xA5, MRMDestReg, (outs GR16:$dst),
690 (ins GR16:$src1, GR16:$src2),
691 "shld{w}\t{%cl, $src2, $dst|$dst, $src2, cl}",
692 [(set GR16:$dst, (X86shld GR16:$src1, GR16:$src2, CL))],
695 def SHRD16rrCL : I<0xAD, MRMDestReg, (outs GR16:$dst),
696 (ins GR16:$src1, GR16:$src2),
697 "shrd{w}\t{%cl, $src2, $dst|$dst, $src2, cl}",
698 [(set GR16:$dst, (X86shrd GR16:$src1, GR16:$src2, CL))],
701 def SHLD32rrCL : I<0xA5, MRMDestReg, (outs GR32:$dst),
702 (ins GR32:$src1, GR32:$src2),
703 "shld{l}\t{%cl, $src2, $dst|$dst, $src2, cl}",
704 [(set GR32:$dst, (X86shld GR32:$src1, GR32:$src2, CL))],
705 IIC_SHD32_REG_CL>, TB, OpSize32;
706 def SHRD32rrCL : I<0xAD, MRMDestReg, (outs GR32:$dst),
707 (ins GR32:$src1, GR32:$src2),
708 "shrd{l}\t{%cl, $src2, $dst|$dst, $src2, cl}",
709 [(set GR32:$dst, (X86shrd GR32:$src1, GR32:$src2, CL))],
710 IIC_SHD32_REG_CL>, TB, OpSize32;
711 def SHLD64rrCL : RI<0xA5, MRMDestReg, (outs GR64:$dst),
712 (ins GR64:$src1, GR64:$src2),
713 "shld{q}\t{%cl, $src2, $dst|$dst, $src2, cl}",
714 [(set GR64:$dst, (X86shld GR64:$src1, GR64:$src2, CL))],
717 def SHRD64rrCL : RI<0xAD, MRMDestReg, (outs GR64:$dst),
718 (ins GR64:$src1, GR64:$src2),
719 "shrd{q}\t{%cl, $src2, $dst|$dst, $src2, cl}",
720 [(set GR64:$dst, (X86shrd GR64:$src1, GR64:$src2, CL))],
725 let isCommutable = 1 in { // These instructions commute to each other.
726 def SHLD16rri8 : Ii8<0xA4, MRMDestReg,
728 (ins GR16:$src1, GR16:$src2, u8imm:$src3),
729 "shld{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
730 [(set GR16:$dst, (X86shld GR16:$src1, GR16:$src2,
731 (i8 imm:$src3)))], IIC_SHD16_REG_IM>,
733 def SHRD16rri8 : Ii8<0xAC, MRMDestReg,
735 (ins GR16:$src1, GR16:$src2, u8imm:$src3),
736 "shrd{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
737 [(set GR16:$dst, (X86shrd GR16:$src1, GR16:$src2,
738 (i8 imm:$src3)))], IIC_SHD16_REG_IM>,
740 def SHLD32rri8 : Ii8<0xA4, MRMDestReg,
742 (ins GR32:$src1, GR32:$src2, u8imm:$src3),
743 "shld{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
744 [(set GR32:$dst, (X86shld GR32:$src1, GR32:$src2,
745 (i8 imm:$src3)))], IIC_SHD32_REG_IM>,
747 def SHRD32rri8 : Ii8<0xAC, MRMDestReg,
749 (ins GR32:$src1, GR32:$src2, u8imm:$src3),
750 "shrd{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
751 [(set GR32:$dst, (X86shrd GR32:$src1, GR32:$src2,
752 (i8 imm:$src3)))], IIC_SHD32_REG_IM>,
754 def SHLD64rri8 : RIi8<0xA4, MRMDestReg,
756 (ins GR64:$src1, GR64:$src2, u8imm:$src3),
757 "shld{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
758 [(set GR64:$dst, (X86shld GR64:$src1, GR64:$src2,
759 (i8 imm:$src3)))], IIC_SHD64_REG_IM>,
761 def SHRD64rri8 : RIi8<0xAC, MRMDestReg,
763 (ins GR64:$src1, GR64:$src2, u8imm:$src3),
764 "shrd{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
765 [(set GR64:$dst, (X86shrd GR64:$src1, GR64:$src2,
766 (i8 imm:$src3)))], IIC_SHD64_REG_IM>,
769 } // Constraints = "$src = $dst", SchedRW
771 let SchedRW = [WriteShiftLd, WriteRMW] in {
773 def SHLD16mrCL : I<0xA5, MRMDestMem, (outs), (ins i16mem:$dst, GR16:$src2),
774 "shld{w}\t{%cl, $src2, $dst|$dst, $src2, cl}",
775 [(store (X86shld (loadi16 addr:$dst), GR16:$src2, CL),
776 addr:$dst)], IIC_SHD16_MEM_CL>, TB, OpSize16;
777 def SHRD16mrCL : I<0xAD, MRMDestMem, (outs), (ins i16mem:$dst, GR16:$src2),
778 "shrd{w}\t{%cl, $src2, $dst|$dst, $src2, cl}",
779 [(store (X86shrd (loadi16 addr:$dst), GR16:$src2, CL),
780 addr:$dst)], IIC_SHD16_MEM_CL>, TB, OpSize16;
782 def SHLD32mrCL : I<0xA5, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src2),
783 "shld{l}\t{%cl, $src2, $dst|$dst, $src2, cl}",
784 [(store (X86shld (loadi32 addr:$dst), GR32:$src2, CL),
785 addr:$dst)], IIC_SHD32_MEM_CL>, TB, OpSize32;
786 def SHRD32mrCL : I<0xAD, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src2),
787 "shrd{l}\t{%cl, $src2, $dst|$dst, $src2, cl}",
788 [(store (X86shrd (loadi32 addr:$dst), GR32:$src2, CL),
789 addr:$dst)], IIC_SHD32_MEM_CL>, TB, OpSize32;
791 def SHLD64mrCL : RI<0xA5, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src2),
792 "shld{q}\t{%cl, $src2, $dst|$dst, $src2, cl}",
793 [(store (X86shld (loadi64 addr:$dst), GR64:$src2, CL),
794 addr:$dst)], IIC_SHD64_MEM_CL>, TB;
795 def SHRD64mrCL : RI<0xAD, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src2),
796 "shrd{q}\t{%cl, $src2, $dst|$dst, $src2, cl}",
797 [(store (X86shrd (loadi64 addr:$dst), GR64:$src2, CL),
798 addr:$dst)], IIC_SHD64_MEM_CL>, TB;
801 def SHLD16mri8 : Ii8<0xA4, MRMDestMem,
802 (outs), (ins i16mem:$dst, GR16:$src2, u8imm:$src3),
803 "shld{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
804 [(store (X86shld (loadi16 addr:$dst), GR16:$src2,
805 (i8 imm:$src3)), addr:$dst)],
808 def SHRD16mri8 : Ii8<0xAC, MRMDestMem,
809 (outs), (ins i16mem:$dst, GR16:$src2, u8imm:$src3),
810 "shrd{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
811 [(store (X86shrd (loadi16 addr:$dst), GR16:$src2,
812 (i8 imm:$src3)), addr:$dst)],
816 def SHLD32mri8 : Ii8<0xA4, MRMDestMem,
817 (outs), (ins i32mem:$dst, GR32:$src2, u8imm:$src3),
818 "shld{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
819 [(store (X86shld (loadi32 addr:$dst), GR32:$src2,
820 (i8 imm:$src3)), addr:$dst)],
823 def SHRD32mri8 : Ii8<0xAC, MRMDestMem,
824 (outs), (ins i32mem:$dst, GR32:$src2, u8imm:$src3),
825 "shrd{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
826 [(store (X86shrd (loadi32 addr:$dst), GR32:$src2,
827 (i8 imm:$src3)), addr:$dst)],
831 def SHLD64mri8 : RIi8<0xA4, MRMDestMem,
832 (outs), (ins i64mem:$dst, GR64:$src2, u8imm:$src3),
833 "shld{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
834 [(store (X86shld (loadi64 addr:$dst), GR64:$src2,
835 (i8 imm:$src3)), addr:$dst)],
838 def SHRD64mri8 : RIi8<0xAC, MRMDestMem,
839 (outs), (ins i64mem:$dst, GR64:$src2, u8imm:$src3),
840 "shrd{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
841 [(store (X86shrd (loadi64 addr:$dst), GR64:$src2,
842 (i8 imm:$src3)), addr:$dst)],
849 // Sandy Bridge and newer Intel processors support faster rotates using
850 // SHLD to avoid a partial flag update on the normal rotate instructions.
851 let Predicates = [HasFastSHLDRotate], AddedComplexity = 5 in {
852 def : Pat<(rotl GR32:$src, (i8 imm:$shamt)),
853 (SHLD32rri8 GR32:$src, GR32:$src, imm:$shamt)>;
854 def : Pat<(rotl GR64:$src, (i8 imm:$shamt)),
855 (SHLD64rri8 GR64:$src, GR64:$src, imm:$shamt)>;
858 def ROT32L2R_imm8 : SDNodeXForm<imm, [{
859 // Convert a ROTL shamt to a ROTR shamt on 32-bit integer.
860 return getI8Imm(32 - N->getZExtValue(), SDLoc(N));
863 def ROT64L2R_imm8 : SDNodeXForm<imm, [{
864 // Convert a ROTL shamt to a ROTR shamt on 64-bit integer.
865 return getI8Imm(64 - N->getZExtValue(), SDLoc(N));
868 multiclass bmi_rotate<string asm, RegisterClass RC, X86MemOperand x86memop> {
869 let hasSideEffects = 0 in {
870 def ri : Ii8<0xF0, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, u8imm:$src2),
871 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
872 []>, TAXD, VEX, Sched<[WriteShift]>;
874 def mi : Ii8<0xF0, MRMSrcMem, (outs RC:$dst),
875 (ins x86memop:$src1, u8imm:$src2),
876 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
877 []>, TAXD, VEX, Sched<[WriteShiftLd]>;
881 multiclass bmi_shift<string asm, RegisterClass RC, X86MemOperand x86memop> {
882 let hasSideEffects = 0 in {
883 def rr : I<0xF7, MRMSrcReg4VOp3, (outs RC:$dst), (ins RC:$src1, RC:$src2),
884 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>,
885 VEX, Sched<[WriteShift]>;
887 def rm : I<0xF7, MRMSrcMem4VOp3,
888 (outs RC:$dst), (ins x86memop:$src1, RC:$src2),
889 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>,
890 VEX, Sched<[WriteShiftLd,
892 ReadDefault, ReadDefault, ReadDefault, ReadDefault,
899 let Predicates = [HasBMI2] in {
900 defm RORX32 : bmi_rotate<"rorx{l}", GR32, i32mem>;
901 defm RORX64 : bmi_rotate<"rorx{q}", GR64, i64mem>, VEX_W;
902 defm SARX32 : bmi_shift<"sarx{l}", GR32, i32mem>, T8XS;
903 defm SARX64 : bmi_shift<"sarx{q}", GR64, i64mem>, T8XS, VEX_W;
904 defm SHRX32 : bmi_shift<"shrx{l}", GR32, i32mem>, T8XD;
905 defm SHRX64 : bmi_shift<"shrx{q}", GR64, i64mem>, T8XD, VEX_W;
906 defm SHLX32 : bmi_shift<"shlx{l}", GR32, i32mem>, T8PD;
907 defm SHLX64 : bmi_shift<"shlx{q}", GR64, i64mem>, T8PD, VEX_W;
909 // Prefer RORX which is non-destructive and doesn't update EFLAGS.
910 let AddedComplexity = 10 in {
911 def : Pat<(rotl GR32:$src, (i8 imm:$shamt)),
912 (RORX32ri GR32:$src, (ROT32L2R_imm8 imm:$shamt))>;
913 def : Pat<(rotl GR64:$src, (i8 imm:$shamt)),
914 (RORX64ri GR64:$src, (ROT64L2R_imm8 imm:$shamt))>;
917 def : Pat<(rotl (loadi32 addr:$src), (i8 imm:$shamt)),
918 (RORX32mi addr:$src, (ROT32L2R_imm8 imm:$shamt))>;
919 def : Pat<(rotl (loadi64 addr:$src), (i8 imm:$shamt)),
920 (RORX64mi addr:$src, (ROT64L2R_imm8 imm:$shamt))>;
922 // Prefer SARX/SHRX/SHLX over SAR/SHR/SHL with variable shift BUT not
923 // immedidate shift, i.e. the following code is considered better
932 // shlx %sil, %edi, %esi
935 let AddedComplexity = 1 in {
936 def : Pat<(sra GR32:$src1, GR8:$src2),
937 (SARX32rr GR32:$src1,
939 (i32 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
940 def : Pat<(sra GR64:$src1, GR8:$src2),
941 (SARX64rr GR64:$src1,
943 (i64 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
945 def : Pat<(srl GR32:$src1, GR8:$src2),
946 (SHRX32rr GR32:$src1,
948 (i32 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
949 def : Pat<(srl GR64:$src1, GR8:$src2),
950 (SHRX64rr GR64:$src1,
952 (i64 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
954 def : Pat<(shl GR32:$src1, GR8:$src2),
955 (SHLX32rr GR32:$src1,
957 (i32 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
958 def : Pat<(shl GR64:$src1, GR8:$src2),
959 (SHLX64rr GR64:$src1,
961 (i64 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
964 // Patterns on SARXrm/SHRXrm/SHLXrm are explicitly omitted to favor
972 // shlx %al, (%ecx), %esi
974 // As SARXrr/SHRXrr/SHLXrr is favored on variable shift, the peephole
975 // optimization will fold them into SARXrm/SHRXrm/SHLXrm if possible.