Skip to content

Commit 3bb0c73

Browse files
aemersonaemerson
committed
[AArch64][GlobalISel] Add more test coverage for bitreverse.
The IR tests show there's some missing optimizations to form rbit.
1 parent cac6777 commit 3bb0c73

File tree

2 files changed

+391
-38
lines changed

2 files changed

+391
-38
lines changed

‎llvm/test/CodeGen/AArch64/GlobalISel/legalize-bitreverse.mir‎

Lines changed: 180 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -186,3 +186,183 @@ body: |
186186
$d0 = COPY %bitreverse
187187
RET_ReallyLR implicit $q0
188188
...
189+
---
190+
name: v2s32
191+
tracksRegLiveness: true
192+
body: |
193+
bb.0:
194+
liveins: $d0
195+
; CHECK-LABEL: name: v2s32
196+
; CHECK: liveins: $d0
197+
; CHECK-NEXT: {{ $}}
198+
; CHECK-NEXT: %vec:_(<2 x s32>) = COPY $d0
199+
; CHECK-NEXT: [[BSWAP:%[0-9]+]]:_(<2 x s32>) = G_BSWAP %vec
200+
; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 4
201+
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[C]](s32), [[C]](s32)
202+
; CHECK-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 -252645136
203+
; CHECK-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[C1]](s32), [[C1]](s32)
204+
; CHECK-NEXT: [[AND:%[0-9]+]]:_(<2 x s32>) = G_AND [[BSWAP]], [[BUILD_VECTOR1]]
205+
; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(<2 x s32>) = G_LSHR [[AND]], [[BUILD_VECTOR]](<2 x s32>)
206+
; CHECK-NEXT: [[SHL:%[0-9]+]]:_(<2 x s32>) = G_SHL [[BSWAP]], [[BUILD_VECTOR]](<2 x s32>)
207+
; CHECK-NEXT: [[AND1:%[0-9]+]]:_(<2 x s32>) = G_AND [[SHL]], [[BUILD_VECTOR1]]
208+
; CHECK-NEXT: [[OR:%[0-9]+]]:_(<2 x s32>) = G_OR [[LSHR]], [[AND1]]
209+
; CHECK-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 2
210+
; CHECK-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[C2]](s32), [[C2]](s32)
211+
; CHECK-NEXT: [[C3:%[0-9]+]]:_(s32) = G_CONSTANT i32 -858993460
212+
; CHECK-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[C3]](s32), [[C3]](s32)
213+
; CHECK-NEXT: [[AND2:%[0-9]+]]:_(<2 x s32>) = G_AND [[OR]], [[BUILD_VECTOR3]]
214+
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:_(<2 x s32>) = G_LSHR [[AND2]], [[BUILD_VECTOR2]](<2 x s32>)
215+
; CHECK-NEXT: [[SHL1:%[0-9]+]]:_(<2 x s32>) = G_SHL [[OR]], [[BUILD_VECTOR2]](<2 x s32>)
216+
; CHECK-NEXT: [[AND3:%[0-9]+]]:_(<2 x s32>) = G_AND [[SHL1]], [[BUILD_VECTOR3]]
217+
; CHECK-NEXT: [[OR1:%[0-9]+]]:_(<2 x s32>) = G_OR [[LSHR1]], [[AND3]]
218+
; CHECK-NEXT: [[C4:%[0-9]+]]:_(s32) = G_CONSTANT i32 1
219+
; CHECK-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[C4]](s32), [[C4]](s32)
220+
; CHECK-NEXT: [[C5:%[0-9]+]]:_(s32) = G_CONSTANT i32 -1431655766
221+
; CHECK-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[C5]](s32), [[C5]](s32)
222+
; CHECK-NEXT: [[AND4:%[0-9]+]]:_(<2 x s32>) = G_AND [[OR1]], [[BUILD_VECTOR5]]
223+
; CHECK-NEXT: [[LSHR2:%[0-9]+]]:_(<2 x s32>) = G_LSHR [[AND4]], [[BUILD_VECTOR4]](<2 x s32>)
224+
; CHECK-NEXT: [[SHL2:%[0-9]+]]:_(<2 x s32>) = G_SHL [[OR1]], [[BUILD_VECTOR4]](<2 x s32>)
225+
; CHECK-NEXT: [[AND5:%[0-9]+]]:_(<2 x s32>) = G_AND [[SHL2]], [[BUILD_VECTOR5]]
226+
; CHECK-NEXT: %bitreverse:_(<2 x s32>) = G_OR [[LSHR2]], [[AND5]]
227+
; CHECK-NEXT: $d0 = COPY %bitreverse(<2 x s32>)
228+
; CHECK-NEXT: RET_ReallyLR implicit $d0
229+
%vec:_(<2 x s32>) = COPY $d0
230+
%bitreverse:_(<2 x s32>) = G_BITREVERSE %vec
231+
$d0 = COPY %bitreverse
232+
RET_ReallyLR implicit $d0
233+
...
234+
---
235+
name: v2s64
236+
tracksRegLiveness: true
237+
body: |
238+
bb.0:
239+
liveins: $q0
240+
; CHECK-LABEL: name: v2s64
241+
; CHECK: liveins: $q0
242+
; CHECK-NEXT: {{ $}}
243+
; CHECK-NEXT: %vec:_(<2 x s64>) = COPY $q0
244+
; CHECK-NEXT: [[BSWAP:%[0-9]+]]:_(<2 x s64>) = G_BSWAP %vec
245+
; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 4
246+
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C]](s64), [[C]](s64)
247+
; CHECK-NEXT: [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 -1085102592571150096
248+
; CHECK-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C1]](s64), [[C1]](s64)
249+
; CHECK-NEXT: [[AND:%[0-9]+]]:_(<2 x s64>) = G_AND [[BSWAP]], [[BUILD_VECTOR1]]
250+
; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(<2 x s64>) = G_LSHR [[AND]], [[BUILD_VECTOR]](<2 x s64>)
251+
; CHECK-NEXT: [[SHL:%[0-9]+]]:_(<2 x s64>) = G_SHL [[BSWAP]], [[BUILD_VECTOR]](<2 x s64>)
252+
; CHECK-NEXT: [[AND1:%[0-9]+]]:_(<2 x s64>) = G_AND [[SHL]], [[BUILD_VECTOR1]]
253+
; CHECK-NEXT: [[OR:%[0-9]+]]:_(<2 x s64>) = G_OR [[LSHR]], [[AND1]]
254+
; CHECK-NEXT: [[C2:%[0-9]+]]:_(s64) = G_CONSTANT i64 2
255+
; CHECK-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C2]](s64), [[C2]](s64)
256+
; CHECK-NEXT: [[C3:%[0-9]+]]:_(s64) = G_CONSTANT i64 -3689348814741910324
257+
; CHECK-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C3]](s64), [[C3]](s64)
258+
; CHECK-NEXT: [[AND2:%[0-9]+]]:_(<2 x s64>) = G_AND [[OR]], [[BUILD_VECTOR3]]
259+
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:_(<2 x s64>) = G_LSHR [[AND2]], [[BUILD_VECTOR2]](<2 x s64>)
260+
; CHECK-NEXT: [[SHL1:%[0-9]+]]:_(<2 x s64>) = G_SHL [[OR]], [[BUILD_VECTOR2]](<2 x s64>)
261+
; CHECK-NEXT: [[AND3:%[0-9]+]]:_(<2 x s64>) = G_AND [[SHL1]], [[BUILD_VECTOR3]]
262+
; CHECK-NEXT: [[OR1:%[0-9]+]]:_(<2 x s64>) = G_OR [[LSHR1]], [[AND3]]
263+
; CHECK-NEXT: [[C4:%[0-9]+]]:_(s64) = G_CONSTANT i64 1
264+
; CHECK-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C4]](s64), [[C4]](s64)
265+
; CHECK-NEXT: [[C5:%[0-9]+]]:_(s64) = G_CONSTANT i64 -6148914691236517206
266+
; CHECK-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C5]](s64), [[C5]](s64)
267+
; CHECK-NEXT: [[AND4:%[0-9]+]]:_(<2 x s64>) = G_AND [[OR1]], [[BUILD_VECTOR5]]
268+
; CHECK-NEXT: [[LSHR2:%[0-9]+]]:_(<2 x s64>) = G_LSHR [[AND4]], [[BUILD_VECTOR4]](<2 x s64>)
269+
; CHECK-NEXT: [[SHL2:%[0-9]+]]:_(<2 x s64>) = G_SHL [[OR1]], [[BUILD_VECTOR4]](<2 x s64>)
270+
; CHECK-NEXT: [[AND5:%[0-9]+]]:_(<2 x s64>) = G_AND [[SHL2]], [[BUILD_VECTOR5]]
271+
; CHECK-NEXT: %bitreverse:_(<2 x s64>) = G_OR [[LSHR2]], [[AND5]]
272+
; CHECK-NEXT: $q0 = COPY %bitreverse(<2 x s64>)
273+
; CHECK-NEXT: RET_ReallyLR implicit $q0
274+
%vec:_(<2 x s64>) = COPY $q0
275+
%bitreverse:_(<2 x s64>) = G_BITREVERSE %vec
276+
$q0 = COPY %bitreverse
277+
RET_ReallyLR implicit $q0
278+
...
279+
---
280+
name: v4s32
281+
tracksRegLiveness: true
282+
body: |
283+
bb.0:
284+
liveins: $q0
285+
; CHECK-LABEL: name: v4s32
286+
; CHECK: liveins: $q0
287+
; CHECK-NEXT: {{ $}}
288+
; CHECK-NEXT: %vec:_(<4 x s32>) = COPY $q0
289+
; CHECK-NEXT: [[BSWAP:%[0-9]+]]:_(<4 x s32>) = G_BSWAP %vec
290+
; CHECK-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 4
291+
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[C]](s32), [[C]](s32), [[C]](s32), [[C]](s32)
292+
; CHECK-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 -252645136
293+
; CHECK-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[C1]](s32), [[C1]](s32), [[C1]](s32), [[C1]](s32)
294+
; CHECK-NEXT: [[AND:%[0-9]+]]:_(<4 x s32>) = G_AND [[BSWAP]], [[BUILD_VECTOR1]]
295+
; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(<4 x s32>) = G_LSHR [[AND]], [[BUILD_VECTOR]](<4 x s32>)
296+
; CHECK-NEXT: [[SHL:%[0-9]+]]:_(<4 x s32>) = G_SHL [[BSWAP]], [[BUILD_VECTOR]](<4 x s32>)
297+
; CHECK-NEXT: [[AND1:%[0-9]+]]:_(<4 x s32>) = G_AND [[SHL]], [[BUILD_VECTOR1]]
298+
; CHECK-NEXT: [[OR:%[0-9]+]]:_(<4 x s32>) = G_OR [[LSHR]], [[AND1]]
299+
; CHECK-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 2
300+
; CHECK-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[C2]](s32), [[C2]](s32), [[C2]](s32), [[C2]](s32)
301+
; CHECK-NEXT: [[C3:%[0-9]+]]:_(s32) = G_CONSTANT i32 -858993460
302+
; CHECK-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[C3]](s32), [[C3]](s32), [[C3]](s32), [[C3]](s32)
303+
; CHECK-NEXT: [[AND2:%[0-9]+]]:_(<4 x s32>) = G_AND [[OR]], [[BUILD_VECTOR3]]
304+
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:_(<4 x s32>) = G_LSHR [[AND2]], [[BUILD_VECTOR2]](<4 x s32>)
305+
; CHECK-NEXT: [[SHL1:%[0-9]+]]:_(<4 x s32>) = G_SHL [[OR]], [[BUILD_VECTOR2]](<4 x s32>)
306+
; CHECK-NEXT: [[AND3:%[0-9]+]]:_(<4 x s32>) = G_AND [[SHL1]], [[BUILD_VECTOR3]]
307+
; CHECK-NEXT: [[OR1:%[0-9]+]]:_(<4 x s32>) = G_OR [[LSHR1]], [[AND3]]
308+
; CHECK-NEXT: [[C4:%[0-9]+]]:_(s32) = G_CONSTANT i32 1
309+
; CHECK-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[C4]](s32), [[C4]](s32), [[C4]](s32), [[C4]](s32)
310+
; CHECK-NEXT: [[C5:%[0-9]+]]:_(s32) = G_CONSTANT i32 -1431655766
311+
; CHECK-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[C5]](s32), [[C5]](s32), [[C5]](s32), [[C5]](s32)
312+
; CHECK-NEXT: [[AND4:%[0-9]+]]:_(<4 x s32>) = G_AND [[OR1]], [[BUILD_VECTOR5]]
313+
; CHECK-NEXT: [[LSHR2:%[0-9]+]]:_(<4 x s32>) = G_LSHR [[AND4]], [[BUILD_VECTOR4]](<4 x s32>)
314+
; CHECK-NEXT: [[SHL2:%[0-9]+]]:_(<4 x s32>) = G_SHL [[OR1]], [[BUILD_VECTOR4]](<4 x s32>)
315+
; CHECK-NEXT: [[AND5:%[0-9]+]]:_(<4 x s32>) = G_AND [[SHL2]], [[BUILD_VECTOR5]]
316+
; CHECK-NEXT: %bitreverse:_(<4 x s32>) = G_OR [[LSHR2]], [[AND5]]
317+
; CHECK-NEXT: $q0 = COPY %bitreverse(<4 x s32>)
318+
; CHECK-NEXT: RET_ReallyLR implicit $q0
319+
%vec:_(<4 x s32>) = COPY $q0
320+
%bitreverse:_(<4 x s32>) = G_BITREVERSE %vec
321+
$q0 = COPY %bitreverse
322+
RET_ReallyLR implicit $q0
323+
...
324+
---
325+
name: v8s16
326+
tracksRegLiveness: true
327+
body: |
328+
bb.0:
329+
liveins: $q0
330+
; CHECK-LABEL: name: v8s16
331+
; CHECK: liveins: $q0
332+
; CHECK-NEXT: {{ $}}
333+
; CHECK-NEXT: %vec:_(<8 x s16>) = COPY $q0
334+
; CHECK-NEXT: [[BSWAP:%[0-9]+]]:_(<8 x s16>) = G_BSWAP %vec
335+
; CHECK-NEXT: [[C:%[0-9]+]]:_(s16) = G_CONSTANT i16 4
336+
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s16>) = G_BUILD_VECTOR [[C]](s16), [[C]](s16), [[C]](s16), [[C]](s16), [[C]](s16), [[C]](s16), [[C]](s16), [[C]](s16)
337+
; CHECK-NEXT: [[C1:%[0-9]+]]:_(s16) = G_CONSTANT i16 -3856
338+
; CHECK-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<8 x s16>) = G_BUILD_VECTOR [[C1]](s16), [[C1]](s16), [[C1]](s16), [[C1]](s16), [[C1]](s16), [[C1]](s16), [[C1]](s16), [[C1]](s16)
339+
; CHECK-NEXT: [[AND:%[0-9]+]]:_(<8 x s16>) = G_AND [[BSWAP]], [[BUILD_VECTOR1]]
340+
; CHECK-NEXT: [[LSHR:%[0-9]+]]:_(<8 x s16>) = G_LSHR [[AND]], [[BUILD_VECTOR]](<8 x s16>)
341+
; CHECK-NEXT: [[SHL:%[0-9]+]]:_(<8 x s16>) = G_SHL [[BSWAP]], [[BUILD_VECTOR]](<8 x s16>)
342+
; CHECK-NEXT: [[AND1:%[0-9]+]]:_(<8 x s16>) = G_AND [[SHL]], [[BUILD_VECTOR1]]
343+
; CHECK-NEXT: [[OR:%[0-9]+]]:_(<8 x s16>) = G_OR [[LSHR]], [[AND1]]
344+
; CHECK-NEXT: [[C2:%[0-9]+]]:_(s16) = G_CONSTANT i16 2
345+
; CHECK-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<8 x s16>) = G_BUILD_VECTOR [[C2]](s16), [[C2]](s16), [[C2]](s16), [[C2]](s16), [[C2]](s16), [[C2]](s16), [[C2]](s16), [[C2]](s16)
346+
; CHECK-NEXT: [[C3:%[0-9]+]]:_(s16) = G_CONSTANT i16 -13108
347+
; CHECK-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<8 x s16>) = G_BUILD_VECTOR [[C3]](s16), [[C3]](s16), [[C3]](s16), [[C3]](s16), [[C3]](s16), [[C3]](s16), [[C3]](s16), [[C3]](s16)
348+
; CHECK-NEXT: [[AND2:%[0-9]+]]:_(<8 x s16>) = G_AND [[OR]], [[BUILD_VECTOR3]]
349+
; CHECK-NEXT: [[LSHR1:%[0-9]+]]:_(<8 x s16>) = G_LSHR [[AND2]], [[BUILD_VECTOR2]](<8 x s16>)
350+
; CHECK-NEXT: [[SHL1:%[0-9]+]]:_(<8 x s16>) = G_SHL [[OR]], [[BUILD_VECTOR2]](<8 x s16>)
351+
; CHECK-NEXT: [[AND3:%[0-9]+]]:_(<8 x s16>) = G_AND [[SHL1]], [[BUILD_VECTOR3]]
352+
; CHECK-NEXT: [[OR1:%[0-9]+]]:_(<8 x s16>) = G_OR [[LSHR1]], [[AND3]]
353+
; CHECK-NEXT: [[C4:%[0-9]+]]:_(s16) = G_CONSTANT i16 1
354+
; CHECK-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<8 x s16>) = G_BUILD_VECTOR [[C4]](s16), [[C4]](s16), [[C4]](s16), [[C4]](s16), [[C4]](s16), [[C4]](s16), [[C4]](s16), [[C4]](s16)
355+
; CHECK-NEXT: [[C5:%[0-9]+]]:_(s16) = G_CONSTANT i16 -21846
356+
; CHECK-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<8 x s16>) = G_BUILD_VECTOR [[C5]](s16), [[C5]](s16), [[C5]](s16), [[C5]](s16), [[C5]](s16), [[C5]](s16), [[C5]](s16), [[C5]](s16)
357+
; CHECK-NEXT: [[AND4:%[0-9]+]]:_(<8 x s16>) = G_AND [[OR1]], [[BUILD_VECTOR5]]
358+
; CHECK-NEXT: [[LSHR2:%[0-9]+]]:_(<8 x s16>) = G_LSHR [[AND4]], [[BUILD_VECTOR4]](<8 x s16>)
359+
; CHECK-NEXT: [[SHL2:%[0-9]+]]:_(<8 x s16>) = G_SHL [[OR1]], [[BUILD_VECTOR4]](<8 x s16>)
360+
; CHECK-NEXT: [[AND5:%[0-9]+]]:_(<8 x s16>) = G_AND [[SHL2]], [[BUILD_VECTOR5]]
361+
; CHECK-NEXT: %bitreverse:_(<8 x s16>) = G_OR [[LSHR2]], [[AND5]]
362+
; CHECK-NEXT: $q0 = COPY %bitreverse(<8 x s16>)
363+
; CHECK-NEXT: RET_ReallyLR implicit $q0
364+
%vec:_(<8 x s16>) = COPY $q0
365+
%bitreverse:_(<8 x s16>) = G_BITREVERSE %vec
366+
$q0 = COPY %bitreverse
367+
RET_ReallyLR implicit $q0
368+
...

0 commit comments

Comments
 (0)