1//===-- X86InstrShiftRotate.td - Shift and Rotate Instrs ---*- tablegen -*-===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file describes the shift and rotate instructions.
11//
12//===----------------------------------------------------------------------===//
13
14// FIXME: Someone needs to smear multipattern goodness all over this file.
15
16let Defs = [EFLAGS] in {
17
18let Constraints = "$src1 = $dst" in {
19let Uses = [CL] in {
20def SHL8rCL  : I<0xD2, MRM4r, (outs GR8 :$dst), (ins GR8 :$src1),
21                 "shl{b}\t{%cl, $dst|$dst, CL}",
22                 [(set GR8:$dst, (shl GR8:$src1, CL))], IIC_SR>;
23def 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>, OpSize;
26def 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>;
29def 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>;
32} // Uses = [CL]
33
34def SHL8ri   : Ii8<0xC0, MRM4r, (outs GR8 :$dst), (ins GR8 :$src1, i8imm:$src2),
35                   "shl{b}\t{$src2, $dst|$dst, $src2}",
36                   [(set GR8:$dst, (shl GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
37                   
38let isConvertibleToThreeAddress = 1 in {   // Can transform into LEA.
39def SHL16ri  : Ii8<0xC1, MRM4r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$src2),
40                   "shl{w}\t{$src2, $dst|$dst, $src2}",
41                   [(set GR16:$dst, (shl GR16:$src1, (i8 imm:$src2)))], IIC_SR>,
42                   OpSize;
43def SHL32ri  : Ii8<0xC1, MRM4r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$src2),
44                   "shl{l}\t{$src2, $dst|$dst, $src2}",
45                   [(set GR32:$dst, (shl GR32:$src1, (i8 imm:$src2)))], IIC_SR>;
46def SHL64ri  : RIi8<0xC1, MRM4r, (outs GR64:$dst), 
47                    (ins GR64:$src1, i8imm:$src2),
48                    "shl{q}\t{$src2, $dst|$dst, $src2}",
49                    [(set GR64:$dst, (shl GR64:$src1, (i8 imm:$src2)))],
50                    IIC_SR>;
51
52// NOTE: We don't include patterns for shifts of a register by one, because
53// 'add reg,reg' is cheaper (and we have a Pat pattern for shift-by-one).
54def SHL8r1   : I<0xD0, MRM4r, (outs GR8:$dst), (ins GR8:$src1),
55                 "shl{b}\t$dst", [], IIC_SR>;
56def SHL16r1  : I<0xD1, MRM4r, (outs GR16:$dst), (ins GR16:$src1),
57                 "shl{w}\t$dst", [], IIC_SR>, OpSize;
58def SHL32r1  : I<0xD1, MRM4r, (outs GR32:$dst), (ins GR32:$src1),
59                 "shl{l}\t$dst", [], IIC_SR>;
60def SHL64r1  : RI<0xD1, MRM4r, (outs GR64:$dst), (ins GR64:$src1),
61                 "shl{q}\t$dst", [], IIC_SR>;
62} // isConvertibleToThreeAddress = 1
63} // Constraints = "$src = $dst" 
64
65
66// FIXME: Why do we need an explicit "Uses = [CL]" when the instr has a pattern
67// using CL?
68let Uses = [CL] in {
69def SHL8mCL  : I<0xD2, MRM4m, (outs), (ins i8mem :$dst),
70                 "shl{b}\t{%cl, $dst|$dst, CL}",
71                 [(store (shl (loadi8 addr:$dst), CL), addr:$dst)], IIC_SR>;
72def SHL16mCL : I<0xD3, MRM4m, (outs), (ins i16mem:$dst),
73                 "shl{w}\t{%cl, $dst|$dst, CL}",
74                 [(store (shl (loadi16 addr:$dst), CL), addr:$dst)], IIC_SR>,
75                 OpSize;
76def SHL32mCL : I<0xD3, MRM4m, (outs), (ins i32mem:$dst),
77                 "shl{l}\t{%cl, $dst|$dst, CL}",
78                 [(store (shl (loadi32 addr:$dst), CL), addr:$dst)], IIC_SR>;
79def SHL64mCL : RI<0xD3, MRM4m, (outs), (ins i64mem:$dst),
80                  "shl{q}\t{%cl, $dst|$dst, CL}",
81                  [(store (shl (loadi64 addr:$dst), CL), addr:$dst)], IIC_SR>;
82}
83def SHL8mi   : Ii8<0xC0, MRM4m, (outs), (ins i8mem :$dst, i8imm:$src),
84                   "shl{b}\t{$src, $dst|$dst, $src}",
85                [(store (shl (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
86                IIC_SR>;
87def SHL16mi  : Ii8<0xC1, MRM4m, (outs), (ins i16mem:$dst, i8imm:$src),
88                   "shl{w}\t{$src, $dst|$dst, $src}",
89               [(store (shl (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
90               IIC_SR>,
91                   OpSize;
92def SHL32mi  : Ii8<0xC1, MRM4m, (outs), (ins i32mem:$dst, i8imm:$src),
93                   "shl{l}\t{$src, $dst|$dst, $src}",
94               [(store (shl (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
95               IIC_SR>;
96def SHL64mi : RIi8<0xC1, MRM4m, (outs), (ins i64mem:$dst, i8imm:$src),
97                  "shl{q}\t{$src, $dst|$dst, $src}",
98                 [(store (shl (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
99                 IIC_SR>;
100
101// Shift by 1
102def SHL8m1   : I<0xD0, MRM4m, (outs), (ins i8mem :$dst),
103                 "shl{b}\t$dst",
104                [(store (shl (loadi8 addr:$dst), (i8 1)), addr:$dst)],
105                IIC_SR>;
106def SHL16m1  : I<0xD1, MRM4m, (outs), (ins i16mem:$dst),
107                 "shl{w}\t$dst",
108               [(store (shl (loadi16 addr:$dst), (i8 1)), addr:$dst)],
109               IIC_SR>,
110                   OpSize;
111def SHL32m1  : I<0xD1, MRM4m, (outs), (ins i32mem:$dst),
112                 "shl{l}\t$dst",
113               [(store (shl (loadi32 addr:$dst), (i8 1)), addr:$dst)],
114               IIC_SR>;
115def SHL64m1 : RI<0xD1, MRM4m, (outs), (ins i64mem:$dst),
116                  "shl{q}\t$dst",
117                 [(store (shl (loadi64 addr:$dst), (i8 1)), addr:$dst)],
118                 IIC_SR>;
119
120let Constraints = "$src1 = $dst" in {
121let Uses = [CL] in {
122def SHR8rCL  : I<0xD2, MRM5r, (outs GR8 :$dst), (ins GR8 :$src1),
123                 "shr{b}\t{%cl, $dst|$dst, CL}",
124                 [(set GR8:$dst, (srl GR8:$src1, CL))], IIC_SR>;
125def SHR16rCL : I<0xD3, MRM5r, (outs GR16:$dst), (ins GR16:$src1),
126                 "shr{w}\t{%cl, $dst|$dst, CL}",
127                 [(set GR16:$dst, (srl GR16:$src1, CL))], IIC_SR>, OpSize;
128def SHR32rCL : I<0xD3, MRM5r, (outs GR32:$dst), (ins GR32:$src1),
129                 "shr{l}\t{%cl, $dst|$dst, CL}",
130                 [(set GR32:$dst, (srl GR32:$src1, CL))], IIC_SR>;
131def SHR64rCL : RI<0xD3, MRM5r, (outs GR64:$dst), (ins GR64:$src1),
132                  "shr{q}\t{%cl, $dst|$dst, CL}",
133                  [(set GR64:$dst, (srl GR64:$src1, CL))], IIC_SR>;
134}
135
136def SHR8ri   : Ii8<0xC0, MRM5r, (outs GR8:$dst), (ins GR8:$src1, i8imm:$src2),
137                   "shr{b}\t{$src2, $dst|$dst, $src2}",
138                   [(set GR8:$dst, (srl GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
139def SHR16ri  : Ii8<0xC1, MRM5r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$src2),
140                   "shr{w}\t{$src2, $dst|$dst, $src2}",
141                   [(set GR16:$dst, (srl GR16:$src1, (i8 imm:$src2)))],
142                   IIC_SR>, OpSize;
143def SHR32ri  : Ii8<0xC1, MRM5r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$src2),
144                   "shr{l}\t{$src2, $dst|$dst, $src2}",
145                   [(set GR32:$dst, (srl GR32:$src1, (i8 imm:$src2)))],
146                   IIC_SR>;
147def SHR64ri : RIi8<0xC1, MRM5r, (outs GR64:$dst), (ins GR64:$src1, i8imm:$src2),
148                  "shr{q}\t{$src2, $dst|$dst, $src2}",
149                  [(set GR64:$dst, (srl GR64:$src1, (i8 imm:$src2)))], IIC_SR>;
150
151// Shift right by 1
152def SHR8r1   : I<0xD0, MRM5r, (outs GR8:$dst), (ins GR8:$src1),
153                 "shr{b}\t$dst",
154                 [(set GR8:$dst, (srl GR8:$src1, (i8 1)))], IIC_SR>;
155def SHR16r1  : I<0xD1, MRM5r, (outs GR16:$dst), (ins GR16:$src1),
156                 "shr{w}\t$dst",
157                 [(set GR16:$dst, (srl GR16:$src1, (i8 1)))], IIC_SR>, OpSize;
158def SHR32r1  : I<0xD1, MRM5r, (outs GR32:$dst), (ins GR32:$src1),
159                 "shr{l}\t$dst",
160                 [(set GR32:$dst, (srl GR32:$src1, (i8 1)))], IIC_SR>;
161def SHR64r1  : RI<0xD1, MRM5r, (outs GR64:$dst), (ins GR64:$src1),
162                 "shr{q}\t$dst",
163                 [(set GR64:$dst, (srl GR64:$src1, (i8 1)))], IIC_SR>;
164} // Constraints = "$src = $dst"
165
166
167let Uses = [CL] in {
168def SHR8mCL  : I<0xD2, MRM5m, (outs), (ins i8mem :$dst),
169                 "shr{b}\t{%cl, $dst|$dst, CL}",
170                 [(store (srl (loadi8 addr:$dst), CL), addr:$dst)], IIC_SR>;
171def SHR16mCL : I<0xD3, MRM5m, (outs), (ins i16mem:$dst),
172                 "shr{w}\t{%cl, $dst|$dst, CL}",
173                 [(store (srl (loadi16 addr:$dst), CL), addr:$dst)], IIC_SR>,
174                 OpSize;
175def SHR32mCL : I<0xD3, MRM5m, (outs), (ins i32mem:$dst),
176                 "shr{l}\t{%cl, $dst|$dst, CL}",
177                 [(store (srl (loadi32 addr:$dst), CL), addr:$dst)], IIC_SR>;
178def SHR64mCL : RI<0xD3, MRM5m, (outs), (ins i64mem:$dst),
179                  "shr{q}\t{%cl, $dst|$dst, CL}",
180                  [(store (srl (loadi64 addr:$dst), CL), addr:$dst)], IIC_SR>;
181}
182def SHR8mi   : Ii8<0xC0, MRM5m, (outs), (ins i8mem :$dst, i8imm:$src),
183                   "shr{b}\t{$src, $dst|$dst, $src}",
184                [(store (srl (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
185                IIC_SR>;
186def SHR16mi  : Ii8<0xC1, MRM5m, (outs), (ins i16mem:$dst, i8imm:$src),
187                   "shr{w}\t{$src, $dst|$dst, $src}",
188               [(store (srl (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
189               IIC_SR>,
190                   OpSize;
191def SHR32mi  : Ii8<0xC1, MRM5m, (outs), (ins i32mem:$dst, i8imm:$src),
192                   "shr{l}\t{$src, $dst|$dst, $src}",
193               [(store (srl (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
194               IIC_SR>;
195def SHR64mi : RIi8<0xC1, MRM5m, (outs), (ins i64mem:$dst, i8imm:$src),
196                  "shr{q}\t{$src, $dst|$dst, $src}",
197                 [(store (srl (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
198                 IIC_SR>;
199
200// Shift by 1
201def SHR8m1   : I<0xD0, MRM5m, (outs), (ins i8mem :$dst),
202                 "shr{b}\t$dst",
203                [(store (srl (loadi8 addr:$dst), (i8 1)), addr:$dst)],
204                IIC_SR>;
205def SHR16m1  : I<0xD1, MRM5m, (outs), (ins i16mem:$dst),
206                 "shr{w}\t$dst",
207               [(store (srl (loadi16 addr:$dst), (i8 1)), addr:$dst)],
208               IIC_SR>,OpSize;
209def SHR32m1  : I<0xD1, MRM5m, (outs), (ins i32mem:$dst),
210                 "shr{l}\t$dst",
211               [(store (srl (loadi32 addr:$dst), (i8 1)), addr:$dst)],
212               IIC_SR>;
213def SHR64m1 : RI<0xD1, MRM5m, (outs), (ins i64mem:$dst),
214                  "shr{q}\t$dst",
215                 [(store (srl (loadi64 addr:$dst), (i8 1)), addr:$dst)],
216                 IIC_SR>;
217
218let Constraints = "$src1 = $dst" in {
219let Uses = [CL] in {
220def SAR8rCL  : I<0xD2, MRM7r, (outs GR8 :$dst), (ins GR8 :$src1),
221                 "sar{b}\t{%cl, $dst|$dst, CL}",
222                 [(set GR8:$dst, (sra GR8:$src1, CL))],
223                 IIC_SR>;
224def SAR16rCL : I<0xD3, MRM7r, (outs GR16:$dst), (ins GR16:$src1),
225                 "sar{w}\t{%cl, $dst|$dst, CL}",
226                 [(set GR16:$dst, (sra GR16:$src1, CL))],
227                 IIC_SR>, OpSize;
228def SAR32rCL : I<0xD3, MRM7r, (outs GR32:$dst), (ins GR32:$src1),
229                 "sar{l}\t{%cl, $dst|$dst, CL}",
230                 [(set GR32:$dst, (sra GR32:$src1, CL))],
231                 IIC_SR>;
232def SAR64rCL : RI<0xD3, MRM7r, (outs GR64:$dst), (ins GR64:$src1),
233                 "sar{q}\t{%cl, $dst|$dst, CL}",
234                 [(set GR64:$dst, (sra GR64:$src1, CL))],
235                 IIC_SR>;
236}
237
238def SAR8ri   : Ii8<0xC0, MRM7r, (outs GR8 :$dst), (ins GR8 :$src1, i8imm:$src2),
239                   "sar{b}\t{$src2, $dst|$dst, $src2}",
240                   [(set GR8:$dst, (sra GR8:$src1, (i8 imm:$src2)))],
241                   IIC_SR>;
242def SAR16ri  : Ii8<0xC1, MRM7r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$src2),
243                   "sar{w}\t{$src2, $dst|$dst, $src2}",
244                   [(set GR16:$dst, (sra GR16:$src1, (i8 imm:$src2)))],
245                   IIC_SR>,
246                   OpSize;
247def SAR32ri  : Ii8<0xC1, MRM7r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$src2),
248                   "sar{l}\t{$src2, $dst|$dst, $src2}",
249                   [(set GR32:$dst, (sra GR32:$src1, (i8 imm:$src2)))],
250                   IIC_SR>;
251def SAR64ri  : RIi8<0xC1, MRM7r, (outs GR64:$dst),
252                    (ins GR64:$src1, i8imm:$src2),
253                    "sar{q}\t{$src2, $dst|$dst, $src2}",
254                    [(set GR64:$dst, (sra GR64:$src1, (i8 imm:$src2)))],
255                    IIC_SR>;
256
257// Shift by 1
258def SAR8r1   : I<0xD0, MRM7r, (outs GR8 :$dst), (ins GR8 :$src1),
259                 "sar{b}\t$dst",
260                 [(set GR8:$dst, (sra GR8:$src1, (i8 1)))],
261                 IIC_SR>;
262def SAR16r1  : I<0xD1, MRM7r, (outs GR16:$dst), (ins GR16:$src1),
263                 "sar{w}\t$dst",
264                 [(set GR16:$dst, (sra GR16:$src1, (i8 1)))],
265                 IIC_SR>, OpSize;
266def SAR32r1  : I<0xD1, MRM7r, (outs GR32:$dst), (ins GR32:$src1),
267                 "sar{l}\t$dst",
268                 [(set GR32:$dst, (sra GR32:$src1, (i8 1)))],
269                 IIC_SR>;
270def SAR64r1  : RI<0xD1, MRM7r, (outs GR64:$dst), (ins GR64:$src1),
271                 "sar{q}\t$dst",
272                 [(set GR64:$dst, (sra GR64:$src1, (i8 1)))],
273                 IIC_SR>;
274} // Constraints = "$src = $dst"
275
276
277let Uses = [CL] in {
278def SAR8mCL  : I<0xD2, MRM7m, (outs), (ins i8mem :$dst),
279                 "sar{b}\t{%cl, $dst|$dst, CL}",
280                 [(store (sra (loadi8 addr:$dst), CL), addr:$dst)],
281                 IIC_SR>;
282def SAR16mCL : I<0xD3, MRM7m, (outs), (ins i16mem:$dst),
283                 "sar{w}\t{%cl, $dst|$dst, CL}",
284                 [(store (sra (loadi16 addr:$dst), CL), addr:$dst)],
285                 IIC_SR>, OpSize;
286def SAR32mCL : I<0xD3, MRM7m, (outs), (ins i32mem:$dst), 
287                 "sar{l}\t{%cl, $dst|$dst, CL}",
288                 [(store (sra (loadi32 addr:$dst), CL), addr:$dst)],
289                 IIC_SR>;
290def SAR64mCL : RI<0xD3, MRM7m, (outs), (ins i64mem:$dst), 
291                 "sar{q}\t{%cl, $dst|$dst, CL}",
292                 [(store (sra (loadi64 addr:$dst), CL), addr:$dst)],
293                 IIC_SR>;
294}
295def SAR8mi   : Ii8<0xC0, MRM7m, (outs), (ins i8mem :$dst, i8imm:$src),
296                   "sar{b}\t{$src, $dst|$dst, $src}",
297                [(store (sra (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
298                IIC_SR>;
299def SAR16mi  : Ii8<0xC1, MRM7m, (outs), (ins i16mem:$dst, i8imm:$src),
300                   "sar{w}\t{$src, $dst|$dst, $src}",
301               [(store (sra (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
302               IIC_SR>,
303                   OpSize;
304def SAR32mi  : Ii8<0xC1, MRM7m, (outs), (ins i32mem:$dst, i8imm:$src),
305                   "sar{l}\t{$src, $dst|$dst, $src}",
306               [(store (sra (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
307               IIC_SR>;
308def SAR64mi  : RIi8<0xC1, MRM7m, (outs), (ins i64mem:$dst, i8imm:$src),
309                    "sar{q}\t{$src, $dst|$dst, $src}",
310                 [(store (sra (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
311                 IIC_SR>;
312
313// Shift by 1
314def SAR8m1   : I<0xD0, MRM7m, (outs), (ins i8mem :$dst),
315                 "sar{b}\t$dst",
316                [(store (sra (loadi8 addr:$dst), (i8 1)), addr:$dst)],
317                IIC_SR>;
318def SAR16m1  : I<0xD1, MRM7m, (outs), (ins i16mem:$dst),
319                 "sar{w}\t$dst",
320               [(store (sra (loadi16 addr:$dst), (i8 1)), addr:$dst)],
321               IIC_SR>,
322                   OpSize;
323def SAR32m1  : I<0xD1, MRM7m, (outs), (ins i32mem:$dst),
324                 "sar{l}\t$dst",
325               [(store (sra (loadi32 addr:$dst), (i8 1)), addr:$dst)],
326               IIC_SR>;
327def SAR64m1 : RI<0xD1, MRM7m, (outs), (ins i64mem:$dst),
328                  "sar{q}\t$dst",
329                 [(store (sra (loadi64 addr:$dst), (i8 1)), addr:$dst)],
330                 IIC_SR>;
331
332//===----------------------------------------------------------------------===//
333// Rotate instructions
334//===----------------------------------------------------------------------===//
335
336let Constraints = "$src1 = $dst" in {
337def RCL8r1 : I<0xD0, MRM2r, (outs GR8:$dst), (ins GR8:$src1),
338               "rcl{b}\t$dst", [], IIC_SR>;
339def RCL8ri : Ii8<0xC0, MRM2r, (outs GR8:$dst), (ins GR8:$src1, i8imm:$cnt),
340                 "rcl{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
341let Uses = [CL] in
342def RCL8rCL : I<0xD2, MRM2r, (outs GR8:$dst), (ins GR8:$src1),
343                "rcl{b}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
344  
345def RCL16r1 : I<0xD1, MRM2r, (outs GR16:$dst), (ins GR16:$src1),
346                "rcl{w}\t$dst", [], IIC_SR>, OpSize;
347def RCL16ri : Ii8<0xC1, MRM2r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$cnt),
348                  "rcl{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize;
349let Uses = [CL] in
350def RCL16rCL : I<0xD3, MRM2r, (outs GR16:$dst), (ins GR16:$src1),
351                 "rcl{w}\t{%cl, $dst|$dst, CL}", [], IIC_SR>, OpSize;
352
353def RCL32r1 : I<0xD1, MRM2r, (outs GR32:$dst), (ins GR32:$src1),
354                "rcl{l}\t$dst", [], IIC_SR>;
355def RCL32ri : Ii8<0xC1, MRM2r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$cnt),
356                  "rcl{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
357let Uses = [CL] in
358def RCL32rCL : I<0xD3, MRM2r, (outs GR32:$dst), (ins GR32:$src1),
359                 "rcl{l}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
360
361
362def RCL64r1 : RI<0xD1, MRM2r, (outs GR64:$dst), (ins GR64:$src1),
363                 "rcl{q}\t$dst", [], IIC_SR>;
364def RCL64ri : RIi8<0xC1, MRM2r, (outs GR64:$dst), (ins GR64:$src1, i8imm:$cnt),
365                   "rcl{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
366let Uses = [CL] in
367def RCL64rCL : RI<0xD3, MRM2r, (outs GR64:$dst), (ins GR64:$src1),
368                  "rcl{q}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
369
370
371def RCR8r1 : I<0xD0, MRM3r, (outs GR8:$dst), (ins GR8:$src1),
372               "rcr{b}\t$dst", [], IIC_SR>;
373def RCR8ri : Ii8<0xC0, MRM3r, (outs GR8:$dst), (ins GR8:$src1, i8imm:$cnt),
374                 "rcr{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
375let Uses = [CL] in
376def RCR8rCL : I<0xD2, MRM3r, (outs GR8:$dst), (ins GR8:$src1),
377                "rcr{b}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
378  
379def RCR16r1 : I<0xD1, MRM3r, (outs GR16:$dst), (ins GR16:$src1),
380                "rcr{w}\t$dst", [], IIC_SR>, OpSize;
381def RCR16ri : Ii8<0xC1, MRM3r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$cnt),
382                  "rcr{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize;
383let Uses = [CL] in
384def RCR16rCL : I<0xD3, MRM3r, (outs GR16:$dst), (ins GR16:$src1),
385                 "rcr{w}\t{%cl, $dst|$dst, CL}", [], IIC_SR>, OpSize;
386
387def RCR32r1 : I<0xD1, MRM3r, (outs GR32:$dst), (ins GR32:$src1),
388                "rcr{l}\t$dst", [], IIC_SR>;
389def RCR32ri : Ii8<0xC1, MRM3r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$cnt),
390                  "rcr{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
391let Uses = [CL] in
392def RCR32rCL : I<0xD3, MRM3r, (outs GR32:$dst), (ins GR32:$src1),
393                 "rcr{l}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
394                 
395def RCR64r1 : RI<0xD1, MRM3r, (outs GR64:$dst), (ins GR64:$src1),
396                 "rcr{q}\t$dst", [], IIC_SR>;
397def RCR64ri : RIi8<0xC1, MRM3r, (outs GR64:$dst), (ins GR64:$src1, i8imm:$cnt),
398                   "rcr{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
399let Uses = [CL] in
400def RCR64rCL : RI<0xD3, MRM3r, (outs GR64:$dst), (ins GR64:$src1),
401                  "rcr{q}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
402
403} // Constraints = "$src = $dst"
404
405def RCL8m1 : I<0xD0, MRM2m, (outs), (ins i8mem:$dst),
406               "rcl{b}\t$dst", [], IIC_SR>;
407def RCL8mi : Ii8<0xC0, MRM2m, (outs), (ins i8mem:$dst, i8imm:$cnt),
408                 "rcl{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
409def RCL16m1 : I<0xD1, MRM2m, (outs), (ins i16mem:$dst),
410                "rcl{w}\t$dst", [], IIC_SR>, OpSize;
411def RCL16mi : Ii8<0xC1, MRM2m, (outs), (ins i16mem:$dst, i8imm:$cnt),
412                  "rcl{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize;
413def RCL32m1 : I<0xD1, MRM2m, (outs), (ins i32mem:$dst),
414                "rcl{l}\t$dst", [], IIC_SR>;
415def RCL32mi : Ii8<0xC1, MRM2m, (outs), (ins i32mem:$dst, i8imm:$cnt),
416                  "rcl{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
417def RCL64m1 : RI<0xD1, MRM2m, (outs), (ins i64mem:$dst),
418                 "rcl{q}\t$dst", [], IIC_SR>;
419def RCL64mi : RIi8<0xC1, MRM2m, (outs), (ins i64mem:$dst, i8imm:$cnt),
420                   "rcl{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
421
422def RCR8m1 : I<0xD0, MRM3m, (outs), (ins i8mem:$dst),
423               "rcr{b}\t$dst", [], IIC_SR>;
424def RCR8mi : Ii8<0xC0, MRM3m, (outs), (ins i8mem:$dst, i8imm:$cnt),
425                 "rcr{b}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
426def RCR16m1 : I<0xD1, MRM3m, (outs), (ins i16mem:$dst),
427                "rcr{w}\t$dst", [], IIC_SR>, OpSize;
428def RCR16mi : Ii8<0xC1, MRM3m, (outs), (ins i16mem:$dst, i8imm:$cnt),
429                  "rcr{w}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>, OpSize;
430def RCR32m1 : I<0xD1, MRM3m, (outs), (ins i32mem:$dst),
431                "rcr{l}\t$dst", [], IIC_SR>;
432def RCR32mi : Ii8<0xC1, MRM3m, (outs), (ins i32mem:$dst, i8imm:$cnt),
433                  "rcr{l}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
434def RCR64m1 : RI<0xD1, MRM3m, (outs), (ins i64mem:$dst),
435                 "rcr{q}\t$dst", [], IIC_SR>;
436def RCR64mi : RIi8<0xC1, MRM3m, (outs), (ins i64mem:$dst, i8imm:$cnt),
437                   "rcr{q}\t{$cnt, $dst|$dst, $cnt}", [], IIC_SR>;
438
439let Uses = [CL] in {
440def RCL8mCL : I<0xD2, MRM2m, (outs), (ins i8mem:$dst),
441                "rcl{b}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
442def RCL16mCL : I<0xD3, MRM2m, (outs), (ins i16mem:$dst),
443                 "rcl{w}\t{%cl, $dst|$dst, CL}", [], IIC_SR>, OpSize;
444def RCL32mCL : I<0xD3, MRM2m, (outs), (ins i32mem:$dst),
445                 "rcl{l}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
446def RCL64mCL : RI<0xD3, MRM2m, (outs), (ins i64mem:$dst),
447                  "rcl{q}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
448
449def RCR8mCL : I<0xD2, MRM3m, (outs), (ins i8mem:$dst),
450                "rcr{b}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
451def RCR16mCL : I<0xD3, MRM3m, (outs), (ins i16mem:$dst),
452                 "rcr{w}\t{%cl, $dst|$dst, CL}", [], IIC_SR>, OpSize;
453def RCR32mCL : I<0xD3, MRM3m, (outs), (ins i32mem:$dst),
454                 "rcr{l}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
455def RCR64mCL : RI<0xD3, MRM3m, (outs), (ins i64mem:$dst),
456                  "rcr{q}\t{%cl, $dst|$dst, CL}", [], IIC_SR>;
457}
458
459let Constraints = "$src1 = $dst" in {
460// FIXME: provide shorter instructions when imm8 == 1
461let Uses = [CL] in {
462def ROL8rCL  : I<0xD2, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1),
463                 "rol{b}\t{%cl, $dst|$dst, CL}",
464                 [(set GR8:$dst, (rotl GR8:$src1, CL))], IIC_SR>;
465def ROL16rCL : I<0xD3, MRM0r, (outs GR16:$dst), (ins GR16:$src1),
466                 "rol{w}\t{%cl, $dst|$dst, CL}",
467                 [(set GR16:$dst, (rotl GR16:$src1, CL))], IIC_SR>, OpSize;
468def ROL32rCL : I<0xD3, MRM0r, (outs GR32:$dst), (ins GR32:$src1),
469                 "rol{l}\t{%cl, $dst|$dst, CL}",
470                 [(set GR32:$dst, (rotl GR32:$src1, CL))], IIC_SR>;
471def ROL64rCL : RI<0xD3, MRM0r, (outs GR64:$dst), (ins GR64:$src1),
472                  "rol{q}\t{%cl, $dst|$dst, CL}",
473                  [(set GR64:$dst, (rotl GR64:$src1, CL))], IIC_SR>;
474}
475
476def ROL8ri   : Ii8<0xC0, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1, i8imm:$src2),
477                   "rol{b}\t{$src2, $dst|$dst, $src2}",
478                   [(set GR8:$dst, (rotl GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
479def ROL16ri  : Ii8<0xC1, MRM0r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$src2),
480                   "rol{w}\t{$src2, $dst|$dst, $src2}",
481                   [(set GR16:$dst, (rotl GR16:$src1, (i8 imm:$src2)))],
482                   IIC_SR>, 
483                   OpSize;
484def ROL32ri  : Ii8<0xC1, MRM0r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$src2),
485                   "rol{l}\t{$src2, $dst|$dst, $src2}",
486                   [(set GR32:$dst, (rotl GR32:$src1, (i8 imm:$src2)))],
487                   IIC_SR>;
488def ROL64ri  : RIi8<0xC1, MRM0r, (outs GR64:$dst), 
489                    (ins GR64:$src1, i8imm:$src2),
490                    "rol{q}\t{$src2, $dst|$dst, $src2}",
491                    [(set GR64:$dst, (rotl GR64:$src1, (i8 imm:$src2)))],
492                    IIC_SR>;
493
494// Rotate by 1
495def ROL8r1   : I<0xD0, MRM0r, (outs GR8 :$dst), (ins GR8 :$src1),
496                 "rol{b}\t$dst",
497                 [(set GR8:$dst, (rotl GR8:$src1, (i8 1)))],
498                 IIC_SR>;
499def ROL16r1  : I<0xD1, MRM0r, (outs GR16:$dst), (ins GR16:$src1),
500                 "rol{w}\t$dst",
501                 [(set GR16:$dst, (rotl GR16:$src1, (i8 1)))],
502                 IIC_SR>, OpSize;
503def ROL32r1  : I<0xD1, MRM0r, (outs GR32:$dst), (ins GR32:$src1),
504                 "rol{l}\t$dst",
505                 [(set GR32:$dst, (rotl GR32:$src1, (i8 1)))],
506                 IIC_SR>;
507def ROL64r1  : RI<0xD1, MRM0r, (outs GR64:$dst), (ins GR64:$src1),
508                  "rol{q}\t$dst",
509                  [(set GR64:$dst, (rotl GR64:$src1, (i8 1)))],
510                  IIC_SR>;
511} // Constraints = "$src = $dst"
512
513let Uses = [CL] in {
514def ROL8mCL  : I<0xD2, MRM0m, (outs), (ins i8mem :$dst),
515                 "rol{b}\t{%cl, $dst|$dst, CL}",
516                 [(store (rotl (loadi8 addr:$dst), CL), addr:$dst)],
517                 IIC_SR>;
518def ROL16mCL : I<0xD3, MRM0m, (outs), (ins i16mem:$dst),
519                 "rol{w}\t{%cl, $dst|$dst, CL}",
520                 [(store (rotl (loadi16 addr:$dst), CL), addr:$dst)],
521                 IIC_SR>, OpSize;
522def ROL32mCL : I<0xD3, MRM0m, (outs), (ins i32mem:$dst),
523                 "rol{l}\t{%cl, $dst|$dst, CL}",
524                 [(store (rotl (loadi32 addr:$dst), CL), addr:$dst)],
525                 IIC_SR>;
526def ROL64mCL :  RI<0xD3, MRM0m, (outs), (ins i64mem:$dst),
527                   "rol{q}\t{%cl, $dst|$dst, %cl}",
528                   [(store (rotl (loadi64 addr:$dst), CL), addr:$dst)],
529                   IIC_SR>;
530}
531def ROL8mi   : Ii8<0xC0, MRM0m, (outs), (ins i8mem :$dst, i8imm:$src1),
532                   "rol{b}\t{$src1, $dst|$dst, $src1}",
533               [(store (rotl (loadi8 addr:$dst), (i8 imm:$src1)), addr:$dst)],
534               IIC_SR>;
535def ROL16mi  : Ii8<0xC1, MRM0m, (outs), (ins i16mem:$dst, i8imm:$src1),
536                   "rol{w}\t{$src1, $dst|$dst, $src1}",
537              [(store (rotl (loadi16 addr:$dst), (i8 imm:$src1)), addr:$dst)],
538              IIC_SR>,
539                   OpSize;
540def ROL32mi  : Ii8<0xC1, MRM0m, (outs), (ins i32mem:$dst, i8imm:$src1),
541                   "rol{l}\t{$src1, $dst|$dst, $src1}",
542              [(store (rotl (loadi32 addr:$dst), (i8 imm:$src1)), addr:$dst)],
543              IIC_SR>;
544def ROL64mi  : RIi8<0xC1, MRM0m, (outs), (ins i64mem:$dst, i8imm:$src1),
545                    "rol{q}\t{$src1, $dst|$dst, $src1}",
546                [(store (rotl (loadi64 addr:$dst), (i8 imm:$src1)), addr:$dst)],
547                IIC_SR>;
548
549// Rotate by 1
550def ROL8m1   : I<0xD0, MRM0m, (outs), (ins i8mem :$dst),
551                 "rol{b}\t$dst",
552               [(store (rotl (loadi8 addr:$dst), (i8 1)), addr:$dst)],
553               IIC_SR>;
554def ROL16m1  : I<0xD1, MRM0m, (outs), (ins i16mem:$dst),
555                 "rol{w}\t$dst",
556              [(store (rotl (loadi16 addr:$dst), (i8 1)), addr:$dst)],
557              IIC_SR>,
558                   OpSize;
559def ROL32m1  : I<0xD1, MRM0m, (outs), (ins i32mem:$dst),
560                 "rol{l}\t$dst",
561              [(store (rotl (loadi32 addr:$dst), (i8 1)), addr:$dst)],
562              IIC_SR>;
563def ROL64m1  : RI<0xD1, MRM0m, (outs), (ins i64mem:$dst),
564                 "rol{q}\t$dst",
565               [(store (rotl (loadi64 addr:$dst), (i8 1)), addr:$dst)],
566               IIC_SR>;
567
568let Constraints = "$src1 = $dst" in {
569let Uses = [CL] in {
570def ROR8rCL  : I<0xD2, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1),
571                 "ror{b}\t{%cl, $dst|$dst, CL}",
572                 [(set GR8:$dst, (rotr GR8:$src1, CL))], IIC_SR>;
573def ROR16rCL : I<0xD3, MRM1r, (outs GR16:$dst), (ins GR16:$src1),
574                 "ror{w}\t{%cl, $dst|$dst, CL}",
575                 [(set GR16:$dst, (rotr GR16:$src1, CL))], IIC_SR>, OpSize;
576def ROR32rCL : I<0xD3, MRM1r, (outs GR32:$dst), (ins GR32:$src1),
577                 "ror{l}\t{%cl, $dst|$dst, CL}",
578                 [(set GR32:$dst, (rotr GR32:$src1, CL))], IIC_SR>;
579def ROR64rCL : RI<0xD3, MRM1r, (outs GR64:$dst), (ins GR64:$src1),
580                  "ror{q}\t{%cl, $dst|$dst, CL}",
581                  [(set GR64:$dst, (rotr GR64:$src1, CL))], IIC_SR>;
582}
583
584def ROR8ri   : Ii8<0xC0, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1, i8imm:$src2),
585                   "ror{b}\t{$src2, $dst|$dst, $src2}",
586                   [(set GR8:$dst, (rotr GR8:$src1, (i8 imm:$src2)))], IIC_SR>;
587def ROR16ri  : Ii8<0xC1, MRM1r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$src2),
588                   "ror{w}\t{$src2, $dst|$dst, $src2}",
589                   [(set GR16:$dst, (rotr GR16:$src1, (i8 imm:$src2)))],
590                   IIC_SR>, 
591                   OpSize;
592def ROR32ri  : Ii8<0xC1, MRM1r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$src2),
593                   "ror{l}\t{$src2, $dst|$dst, $src2}",
594                   [(set GR32:$dst, (rotr GR32:$src1, (i8 imm:$src2)))],
595                   IIC_SR>;
596def ROR64ri  : RIi8<0xC1, MRM1r, (outs GR64:$dst), 
597                    (ins GR64:$src1, i8imm:$src2),
598                    "ror{q}\t{$src2, $dst|$dst, $src2}",
599                    [(set GR64:$dst, (rotr GR64:$src1, (i8 imm:$src2)))],
600                    IIC_SR>;
601
602// Rotate by 1
603def ROR8r1   : I<0xD0, MRM1r, (outs GR8 :$dst), (ins GR8 :$src1),
604                 "ror{b}\t$dst",
605                 [(set GR8:$dst, (rotr GR8:$src1, (i8 1)))],
606                 IIC_SR>;
607def ROR16r1  : I<0xD1, MRM1r, (outs GR16:$dst), (ins GR16:$src1),
608                 "ror{w}\t$dst",
609                 [(set GR16:$dst, (rotr GR16:$src1, (i8 1)))],
610                 IIC_SR>, OpSize;
611def ROR32r1  : I<0xD1, MRM1r, (outs GR32:$dst), (ins GR32:$src1),
612                 "ror{l}\t$dst",
613                 [(set GR32:$dst, (rotr GR32:$src1, (i8 1)))],
614                 IIC_SR>;
615def ROR64r1  : RI<0xD1, MRM1r, (outs GR64:$dst), (ins GR64:$src1),
616                  "ror{q}\t$dst",
617                  [(set GR64:$dst, (rotr GR64:$src1, (i8 1)))],
618                  IIC_SR>;
619} // Constraints = "$src = $dst"
620
621let Uses = [CL] in {
622def ROR8mCL  : I<0xD2, MRM1m, (outs), (ins i8mem :$dst),
623                 "ror{b}\t{%cl, $dst|$dst, CL}",
624                 [(store (rotr (loadi8 addr:$dst), CL), addr:$dst)],
625                 IIC_SR>;
626def ROR16mCL : I<0xD3, MRM1m, (outs), (ins i16mem:$dst),
627                 "ror{w}\t{%cl, $dst|$dst, CL}",
628                 [(store (rotr (loadi16 addr:$dst), CL), addr:$dst)],
629                 IIC_SR>, OpSize;
630def ROR32mCL : I<0xD3, MRM1m, (outs), (ins i32mem:$dst), 
631                 "ror{l}\t{%cl, $dst|$dst, CL}",
632                 [(store (rotr (loadi32 addr:$dst), CL), addr:$dst)],
633                 IIC_SR>;
634def ROR64mCL : RI<0xD3, MRM1m, (outs), (ins i64mem:$dst), 
635                  "ror{q}\t{%cl, $dst|$dst, CL}",
636                  [(store (rotr (loadi64 addr:$dst), CL), addr:$dst)],
637                  IIC_SR>;
638}
639def ROR8mi   : Ii8<0xC0, MRM1m, (outs), (ins i8mem :$dst, i8imm:$src),
640                   "ror{b}\t{$src, $dst|$dst, $src}",
641               [(store (rotr (loadi8 addr:$dst), (i8 imm:$src)), addr:$dst)],
642               IIC_SR>;
643def ROR16mi  : Ii8<0xC1, MRM1m, (outs), (ins i16mem:$dst, i8imm:$src),
644                   "ror{w}\t{$src, $dst|$dst, $src}",
645              [(store (rotr (loadi16 addr:$dst), (i8 imm:$src)), addr:$dst)],
646              IIC_SR>,
647                   OpSize;
648def ROR32mi  : Ii8<0xC1, MRM1m, (outs), (ins i32mem:$dst, i8imm:$src),
649                   "ror{l}\t{$src, $dst|$dst, $src}",
650              [(store (rotr (loadi32 addr:$dst), (i8 imm:$src)), addr:$dst)],
651              IIC_SR>;
652def ROR64mi  : RIi8<0xC1, MRM1m, (outs), (ins i64mem:$dst, i8imm:$src),
653                    "ror{q}\t{$src, $dst|$dst, $src}",
654                [(store (rotr (loadi64 addr:$dst), (i8 imm:$src)), addr:$dst)],
655                IIC_SR>;
656
657// Rotate by 1
658def ROR8m1   : I<0xD0, MRM1m, (outs), (ins i8mem :$dst),
659                 "ror{b}\t$dst",
660               [(store (rotr (loadi8 addr:$dst), (i8 1)), addr:$dst)],
661               IIC_SR>;
662def ROR16m1  : I<0xD1, MRM1m, (outs), (ins i16mem:$dst),
663                 "ror{w}\t$dst",
664              [(store (rotr (loadi16 addr:$dst), (i8 1)), addr:$dst)],
665              IIC_SR>,
666                   OpSize;
667def ROR32m1  : I<0xD1, MRM1m, (outs), (ins i32mem:$dst),
668                 "ror{l}\t$dst",
669              [(store (rotr (loadi32 addr:$dst), (i8 1)), addr:$dst)],
670              IIC_SR>;
671def ROR64m1  : RI<0xD1, MRM1m, (outs), (ins i64mem:$dst),
672                 "ror{q}\t$dst",
673               [(store (rotr (loadi64 addr:$dst), (i8 1)), addr:$dst)],
674               IIC_SR>;
675
676
677//===----------------------------------------------------------------------===//
678// Double shift instructions (generalizations of rotate)
679//===----------------------------------------------------------------------===//
680
681let Constraints = "$src1 = $dst" in {
682
683let Uses = [CL] in {
684def SHLD16rrCL : I<0xA5, MRMDestReg, (outs GR16:$dst), 
685                   (ins GR16:$src1, GR16:$src2),
686                   "shld{w}\t{%cl, $src2, $dst|$dst, $src2, CL}",
687                   [(set GR16:$dst, (X86shld GR16:$src1, GR16:$src2, CL))],
688                    IIC_SHD16_REG_CL>,
689                   TB, OpSize;
690def SHRD16rrCL : I<0xAD, MRMDestReg, (outs GR16:$dst), 
691                   (ins GR16:$src1, GR16:$src2),
692                   "shrd{w}\t{%cl, $src2, $dst|$dst, $src2, CL}",
693                   [(set GR16:$dst, (X86shrd GR16:$src1, GR16:$src2, CL))],
694                    IIC_SHD16_REG_CL>,
695                   TB, OpSize;
696def SHLD32rrCL : I<0xA5, MRMDestReg, (outs GR32:$dst), 
697                   (ins GR32:$src1, GR32:$src2),
698                   "shld{l}\t{%cl, $src2, $dst|$dst, $src2, CL}",
699                   [(set GR32:$dst, (X86shld GR32:$src1, GR32:$src2, CL))],
700                    IIC_SHD32_REG_CL>, TB;
701def SHRD32rrCL : I<0xAD, MRMDestReg, (outs GR32:$dst),
702                   (ins GR32:$src1, GR32:$src2),
703                   "shrd{l}\t{%cl, $src2, $dst|$dst, $src2, CL}",
704                   [(set GR32:$dst, (X86shrd GR32:$src1, GR32:$src2, CL))],
705                   IIC_SHD32_REG_CL>, TB;
706def SHLD64rrCL : RI<0xA5, MRMDestReg, (outs GR64:$dst), 
707                    (ins GR64:$src1, GR64:$src2),
708                    "shld{q}\t{%cl, $src2, $dst|$dst, $src2, CL}",
709                    [(set GR64:$dst, (X86shld GR64:$src1, GR64:$src2, CL))],
710                    IIC_SHD64_REG_CL>, 
711                    TB;
712def SHRD64rrCL : RI<0xAD, MRMDestReg, (outs GR64:$dst), 
713                    (ins GR64:$src1, GR64:$src2),
714                    "shrd{q}\t{%cl, $src2, $dst|$dst, $src2, CL}",
715                    [(set GR64:$dst, (X86shrd GR64:$src1, GR64:$src2, CL))],
716                    IIC_SHD64_REG_CL>, 
717                    TB;
718}
719
720let isCommutable = 1 in {  // These instructions commute to each other.
721def SHLD16rri8 : Ii8<0xA4, MRMDestReg,
722                     (outs GR16:$dst), 
723                     (ins GR16:$src1, GR16:$src2, i8imm:$src3),
724                     "shld{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
725                     [(set GR16:$dst, (X86shld GR16:$src1, GR16:$src2,
726                                      (i8 imm:$src3)))], IIC_SHD16_REG_IM>,
727                     TB, OpSize;
728def SHRD16rri8 : Ii8<0xAC, MRMDestReg,
729                     (outs GR16:$dst), 
730                     (ins GR16:$src1, GR16:$src2, i8imm:$src3),
731                     "shrd{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
732                     [(set GR16:$dst, (X86shrd GR16:$src1, GR16:$src2,
733                                      (i8 imm:$src3)))], IIC_SHD16_REG_IM>,
734                     TB, OpSize;
735def SHLD32rri8 : Ii8<0xA4, MRMDestReg,
736                     (outs GR32:$dst), 
737                     (ins GR32:$src1, GR32:$src2, i8imm:$src3),
738                     "shld{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
739                     [(set GR32:$dst, (X86shld GR32:$src1, GR32:$src2,
740                                      (i8 imm:$src3)))], IIC_SHD32_REG_IM>,
741                 TB;
742def SHRD32rri8 : Ii8<0xAC, MRMDestReg,
743                     (outs GR32:$dst), 
744                     (ins GR32:$src1, GR32:$src2, i8imm:$src3),
745                     "shrd{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
746                     [(set GR32:$dst, (X86shrd GR32:$src1, GR32:$src2,
747                                      (i8 imm:$src3)))], IIC_SHD32_REG_IM>,
748                 TB;
749def SHLD64rri8 : RIi8<0xA4, MRMDestReg,
750                      (outs GR64:$dst), 
751                      (ins GR64:$src1, GR64:$src2, i8imm:$src3),
752                      "shld{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
753                      [(set GR64:$dst, (X86shld GR64:$src1, GR64:$src2,
754                                       (i8 imm:$src3)))], IIC_SHD64_REG_IM>,
755                 TB;
756def SHRD64rri8 : RIi8<0xAC, MRMDestReg,
757                      (outs GR64:$dst), 
758                      (ins GR64:$src1, GR64:$src2, i8imm:$src3),
759                      "shrd{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
760                      [(set GR64:$dst, (X86shrd GR64:$src1, GR64:$src2,
761                                       (i8 imm:$src3)))], IIC_SHD64_REG_IM>,
762                 TB;
763}
764} // Constraints = "$src = $dst"
765
766let Uses = [CL] in {
767def SHLD16mrCL : I<0xA5, MRMDestMem, (outs), (ins i16mem:$dst, GR16:$src2),
768                   "shld{w}\t{%cl, $src2, $dst|$dst, $src2, CL}",
769                   [(store (X86shld (loadi16 addr:$dst), GR16:$src2, CL),
770                     addr:$dst)], IIC_SHD16_MEM_CL>, TB, OpSize;
771def SHRD16mrCL : I<0xAD, MRMDestMem, (outs), (ins i16mem:$dst, GR16:$src2),
772                  "shrd{w}\t{%cl, $src2, $dst|$dst, $src2, CL}",
773                  [(store (X86shrd (loadi16 addr:$dst), GR16:$src2, CL),
774                    addr:$dst)], IIC_SHD16_MEM_CL>, TB, OpSize;
775
776def SHLD32mrCL : I<0xA5, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src2),
777                   "shld{l}\t{%cl, $src2, $dst|$dst, $src2, CL}",
778                   [(store (X86shld (loadi32 addr:$dst), GR32:$src2, CL),
779                     addr:$dst)], IIC_SHD32_MEM_CL>, TB;
780def SHRD32mrCL : I<0xAD, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src2),
781                  "shrd{l}\t{%cl, $src2, $dst|$dst, $src2, CL}",
782                  [(store (X86shrd (loadi32 addr:$dst), GR32:$src2, CL),
783                    addr:$dst)], IIC_SHD32_MEM_CL>, TB;
784                    
785def SHLD64mrCL : RI<0xA5, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src2),
786                    "shld{q}\t{%cl, $src2, $dst|$dst, $src2, CL}",
787                    [(store (X86shld (loadi64 addr:$dst), GR64:$src2, CL),
788                      addr:$dst)], IIC_SHD64_MEM_CL>, TB;
789def SHRD64mrCL : RI<0xAD, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src2),
790                    "shrd{q}\t{%cl, $src2, $dst|$dst, $src2, CL}",
791                    [(store (X86shrd (loadi64 addr:$dst), GR64:$src2, CL),
792                      addr:$dst)], IIC_SHD64_MEM_CL>, TB;
793}
794
795def SHLD16mri8 : Ii8<0xA4, MRMDestMem,
796                    (outs), (ins i16mem:$dst, GR16:$src2, i8imm:$src3),
797                    "shld{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
798                    [(store (X86shld (loadi16 addr:$dst), GR16:$src2,
799                                      (i8 imm:$src3)), addr:$dst)],
800                                      IIC_SHD16_MEM_IM>,
801                    TB, OpSize;
802def SHRD16mri8 : Ii8<0xAC, MRMDestMem, 
803                     (outs), (ins i16mem:$dst, GR16:$src2, i8imm:$src3),
804                     "shrd{w}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
805                    [(store (X86shrd (loadi16 addr:$dst), GR16:$src2,
806                                      (i8 imm:$src3)), addr:$dst)],
807                                      IIC_SHD16_MEM_IM>,
808                     TB, OpSize;
809
810def SHLD32mri8 : Ii8<0xA4, MRMDestMem,
811                    (outs), (ins i32mem:$dst, GR32:$src2, i8imm:$src3),
812                    "shld{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
813                    [(store (X86shld (loadi32 addr:$dst), GR32:$src2,
814                                      (i8 imm:$src3)), addr:$dst)],
815                                      IIC_SHD32_MEM_IM>,
816                    TB;
817def SHRD32mri8 : Ii8<0xAC, MRMDestMem, 
818                     (outs), (ins i32mem:$dst, GR32:$src2, i8imm:$src3),
819                     "shrd{l}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
820                     [(store (X86shrd (loadi32 addr:$dst), GR32:$src2,
821                                       (i8 imm:$src3)), addr:$dst)],
822                                       IIC_SHD32_MEM_IM>,
823                     TB;
824
825def SHLD64mri8 : RIi8<0xA4, MRMDestMem,
826                      (outs), (ins i64mem:$dst, GR64:$src2, i8imm:$src3),
827                      "shld{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
828                      [(store (X86shld (loadi64 addr:$dst), GR64:$src2,
829                                       (i8 imm:$src3)), addr:$dst)],
830                                       IIC_SHD64_MEM_IM>,
831                 TB;
832def SHRD64mri8 : RIi8<0xAC, MRMDestMem, 
833                      (outs), (ins i64mem:$dst, GR64:$src2, i8imm:$src3),
834                      "shrd{q}\t{$src3, $src2, $dst|$dst, $src2, $src3}",
835                      [(store (X86shrd (loadi64 addr:$dst), GR64:$src2,
836                                       (i8 imm:$src3)), addr:$dst)],
837                                       IIC_SHD64_MEM_IM>,
838                 TB;
839
840} // Defs = [EFLAGS]
841
842def ROT32L2R_imm8  : SDNodeXForm<imm, [{
843  // Convert a ROTL shamt to a ROTR shamt on 32-bit integer.
844  return getI8Imm(32 - N->getZExtValue());
845}]>;
846
847def ROT64L2R_imm8  : SDNodeXForm<imm, [{
848  // Convert a ROTL shamt to a ROTR shamt on 64-bit integer.
849  return getI8Imm(64 - N->getZExtValue());
850}]>;
851
852multiclass bmi_rotate<string asm, RegisterClass RC, X86MemOperand x86memop> {
853let neverHasSideEffects = 1 in {
854  def ri : Ii8<0xF0, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, i8imm:$src2),
855               !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
856               []>, TAXD, VEX;
857  let mayLoad = 1 in
858  def mi : Ii8<0xF0, MRMSrcMem, (outs RC:$dst),
859               (ins x86memop:$src1, i8imm:$src2),
860               !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
861               []>, TAXD, VEX;
862}
863}
864
865multiclass bmi_shift<string asm, RegisterClass RC, X86MemOperand x86memop> {
866let neverHasSideEffects = 1 in {
867  def rr : I<0xF7, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
868             !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>,
869             VEX_4VOp3;
870  let mayLoad = 1 in
871  def rm : I<0xF7, MRMSrcMem, (outs RC:$dst), (ins x86memop:$src1, RC:$src2),
872             !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>,
873             VEX_4VOp3;
874}
875}
876
877let Predicates = [HasBMI2] in {
878  defm RORX32 : bmi_rotate<"rorx{l}", GR32, i32mem>;
879  defm RORX64 : bmi_rotate<"rorx{q}", GR64, i64mem>, VEX_W;
880  defm SARX32 : bmi_shift<"sarx{l}", GR32, i32mem>, T8XS;
881  defm SARX64 : bmi_shift<"sarx{q}", GR64, i64mem>, T8XS, VEX_W;
882  defm SHRX32 : bmi_shift<"shrx{l}", GR32, i32mem>, T8XD;
883  defm SHRX64 : bmi_shift<"shrx{q}", GR64, i64mem>, T8XD, VEX_W;
884  defm SHLX32 : bmi_shift<"shlx{l}", GR32, i32mem>, T8, OpSize;
885  defm SHLX64 : bmi_shift<"shlx{q}", GR64, i64mem>, T8, OpSize, VEX_W;
886
887  // Prefer RORX which is non-destructive and doesn't update EFLAGS.
888  let AddedComplexity = 10 in {
889    def : Pat<(rotl GR32:$src, (i8 imm:$shamt)),
890              (RORX32ri GR32:$src, (ROT32L2R_imm8 imm:$shamt))>;
891    def : Pat<(rotl GR64:$src, (i8 imm:$shamt)),
892              (RORX64ri GR64:$src, (ROT64L2R_imm8 imm:$shamt))>;
893  }
894
895  def : Pat<(rotl (loadi32 addr:$src), (i8 imm:$shamt)),
896            (RORX32mi addr:$src, (ROT32L2R_imm8 imm:$shamt))>;
897  def : Pat<(rotl (loadi64 addr:$src), (i8 imm:$shamt)),
898            (RORX64mi addr:$src, (ROT64L2R_imm8 imm:$shamt))>;
899
900  // Prefer SARX/SHRX/SHLX over SAR/SHR/SHL with variable shift BUT not
901  // immedidate shift, i.e. the following code is considered better
902  //
903  //  mov %edi, %esi
904  //  shl $imm, %esi
905  //  ... %edi, ...
906  //
907  // than
908  //
909  //  movb $imm, %sil
910  //  shlx %sil, %edi, %esi
911  //  ... %edi, ...
912  //
913  let AddedComplexity = 1 in {
914    def : Pat<(sra GR32:$src1, GR8:$src2),
915              (SARX32rr GR32:$src1,
916                        (INSERT_SUBREG
917                          (i32 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
918    def : Pat<(sra GR64:$src1, GR8:$src2),
919              (SARX64rr GR64:$src1,
920                        (INSERT_SUBREG
921                          (i64 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
922
923    def : Pat<(srl GR32:$src1, GR8:$src2),
924              (SHRX32rr GR32:$src1,
925                        (INSERT_SUBREG
926                          (i32 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
927    def : Pat<(srl GR64:$src1, GR8:$src2),
928              (SHRX64rr GR64:$src1,
929                        (INSERT_SUBREG
930                          (i64 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
931
932    def : Pat<(shl GR32:$src1, GR8:$src2),
933              (SHLX32rr GR32:$src1,
934                        (INSERT_SUBREG
935                          (i32 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
936    def : Pat<(shl GR64:$src1, GR8:$src2),
937              (SHLX64rr GR64:$src1,
938                        (INSERT_SUBREG
939                          (i64 (IMPLICIT_DEF)), GR8:$src2, sub_8bit))>;
940  }
941
942  // Patterns on SARXrm/SHRXrm/SHLXrm are explicitly omitted to favor
943  //
944  //  mov (%ecx), %esi
945  //  shl $imm, $esi
946  //
947  // over
948  //
949  //  movb $imm %al
950  //  shlx %al, (%ecx), %esi
951  //
952  // As SARXrr/SHRXrr/SHLXrr is favored on variable shift, the peephole
953  // optimization will fold them into SARXrm/SHRXrm/SHLXrm if possible.
954}
955