Skip to content

Commit 81e3dd7

Browse files
zhaoqi5zhaoqi5
committed
update tests
1 parent 07260f2 commit 81e3dd7

File tree

1 file changed

+38
-50
lines changed

1 file changed

+38
-50
lines changed

‎llvm/test/CodeGen/LoongArch/sink-fold-addi.ll‎

Lines changed: 38 additions & 50 deletions
Original file line numberDiff line numberDiff line change
@@ -25,14 +25,13 @@ define void @sink_fold_i64(i64 %k, i64 %n, ptr %a) nounwind {
2525
; LA32-NEXT: move $s1, $a2
2626
; LA32-NEXT: slli.w $a1, $a0, 4
2727
; LA32-NEXT: alsl.w $a0, $a0, $a1, 3
28-
; LA32-NEXT: add.w $a0, $a4, $a0
2928
; LA32-NEXT: sltui $a1, $a3, 1
3029
; LA32-NEXT: slti $a2, $a3, 0
3130
; LA32-NEXT: masknez $a2, $a2, $a1
3231
; LA32-NEXT: sltui $a3, $s1, 1
3332
; LA32-NEXT: maskeqz $a1, $a3, $a1
3433
; LA32-NEXT: or $a1, $a1, $a2
35-
; LA32-NEXT: addi.w $s2, $a0, 8
34+
; LA32-NEXT: add.w $s2, $a4, $a0
3635
; LA32-NEXT: bnez $a1, .LBB0_3
3736
; LA32-NEXT: # %bb.1: # %for.body.preheader
3837
; LA32-NEXT: move $fp, $a4
@@ -45,8 +44,8 @@ define void @sink_fold_i64(i64 %k, i64 %n, ptr %a) nounwind {
4544
; LA32-NEXT: # =>This Inner Loop Header: Depth=1
4645
; LA32-NEXT: move $a0, $fp
4746
; LA32-NEXT: bl f
48-
; LA32-NEXT: ld.w $a0, $s2, 4
49-
; LA32-NEXT: ld.w $a1, $s2, 0
47+
; LA32-NEXT: ld.w $a0, $s2, 12
48+
; LA32-NEXT: ld.w $a1, $s2, 8
5049
; LA32-NEXT: add.w $a0, $a0, $s6
5150
; LA32-NEXT: add.w $s3, $a1, $s3
5251
; LA32-NEXT: sltu $a1, $s3, $a1
@@ -63,8 +62,8 @@ define void @sink_fold_i64(i64 %k, i64 %n, ptr %a) nounwind {
6362
; LA32-NEXT: move $s3, $zero
6463
; LA32-NEXT: move $s6, $zero
6564
; LA32-NEXT: .LBB0_4: # %for.cond.cleanup
66-
; LA32-NEXT: st.w $s3, $s2, 0
67-
; LA32-NEXT: st.w $s6, $s2, 4
65+
; LA32-NEXT: st.w $s3, $s2, 8
66+
; LA32-NEXT: st.w $s6, $s2, 12
6867
; LA32-NEXT: ld.w $s6, $sp, 12 # 4-byte Folded Reload
6968
; LA32-NEXT: ld.w $s5, $sp, 16 # 4-byte Folded Reload
7069
; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload
@@ -88,8 +87,7 @@ define void @sink_fold_i64(i64 %k, i64 %n, ptr %a) nounwind {
8887
; LA64-NEXT: move $s0, $a1
8988
; LA64-NEXT: slli.d $a1, $a0, 4
9089
; LA64-NEXT: alsl.d $a0, $a0, $a1, 3
91-
; LA64-NEXT: add.d $a0, $a2, $a0
92-
; LA64-NEXT: addi.d $s1, $a0, 8
90+
; LA64-NEXT: add.d $s1, $a2, $a0
9391
; LA64-NEXT: blez $s0, .LBB0_3
9492
; LA64-NEXT: # %bb.1: # %for.body.preheader
9593
; LA64-NEXT: move $fp, $a2
@@ -100,15 +98,15 @@ define void @sink_fold_i64(i64 %k, i64 %n, ptr %a) nounwind {
10098
; LA64-NEXT: move $a0, $fp
10199
; LA64-NEXT: pcaddu18i $ra, %call36(f)
102100
; LA64-NEXT: jirl $ra, $ra, 0
103-
; LA64-NEXT: ld.d $a0, $s1, 0
101+
; LA64-NEXT: ld.d $a0, $s1, 8
104102
; LA64-NEXT: addi.d $s0, $s0, -1
105103
; LA64-NEXT: add.d $s2, $a0, $s2
106104
; LA64-NEXT: bnez $s0, .LBB0_2
107105
; LA64-NEXT: b .LBB0_4
108106
; LA64-NEXT: .LBB0_3:
109107
; LA64-NEXT: move $s2, $zero
110108
; LA64-NEXT: .LBB0_4: # %for.cond.cleanup
111-
; LA64-NEXT: st.d $s2, $s1, 0
109+
; LA64-NEXT: st.d $s2, $s1, 8
112110
; LA64-NEXT: ld.d $s2, $sp, 8 # 8-byte Folded Reload
113111
; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload
114112
; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload
@@ -153,14 +151,13 @@ define void @sink_fold_f32(i64 %k, i64 %n, ptr %a) nounwind {
153151
; LA32-NEXT: move $s1, $a2
154152
; LA32-NEXT: slli.w $a1, $a0, 4
155153
; LA32-NEXT: alsl.w $a0, $a0, $a1, 3
156-
; LA32-NEXT: add.w $a0, $a4, $a0
157154
; LA32-NEXT: sltui $a1, $a3, 1
158155
; LA32-NEXT: slti $a2, $a3, 0
159156
; LA32-NEXT: masknez $a2, $a2, $a1
160157
; LA32-NEXT: sltui $a3, $s1, 1
161158
; LA32-NEXT: maskeqz $a1, $a3, $a1
162159
; LA32-NEXT: or $a1, $a1, $a2
163-
; LA32-NEXT: addi.w $s2, $a0, 16
160+
; LA32-NEXT: add.w $s2, $a4, $a0
164161
; LA32-NEXT: bnez $a1, .LBB1_3
165162
; LA32-NEXT: # %bb.1: # %for.body.preheader
166163
; LA32-NEXT: move $fp, $a4
@@ -172,7 +169,7 @@ define void @sink_fold_f32(i64 %k, i64 %n, ptr %a) nounwind {
172169
; LA32-NEXT: # =>This Inner Loop Header: Depth=1
173170
; LA32-NEXT: move $a0, $fp
174171
; LA32-NEXT: bl f
175-
; LA32-NEXT: fld.s $fa0, $s2, 0
172+
; LA32-NEXT: fld.s $fa0, $s2, 16
176173
; LA32-NEXT: addi.w $s3, $s3, 1
177174
; LA32-NEXT: sltui $a0, $s3, 1
178175
; LA32-NEXT: add.w $s4, $s4, $a0
@@ -185,7 +182,7 @@ define void @sink_fold_f32(i64 %k, i64 %n, ptr %a) nounwind {
185182
; LA32-NEXT: .LBB1_3:
186183
; LA32-NEXT: movgr2fr.w $fs0, $zero
187184
; LA32-NEXT: .LBB1_4: # %for.cond.cleanup
188-
; LA32-NEXT: fst.s $fs0, $s2, 0
185+
; LA32-NEXT: fst.s $fs0, $s2, 16
189186
; LA32-NEXT: fld.d $fs0, $sp, 8 # 8-byte Folded Reload
190187
; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload
191188
; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload
@@ -208,8 +205,7 @@ define void @sink_fold_f32(i64 %k, i64 %n, ptr %a) nounwind {
208205
; LA64-NEXT: move $s0, $a1
209206
; LA64-NEXT: slli.d $a1, $a0, 4
210207
; LA64-NEXT: alsl.d $a0, $a0, $a1, 3
211-
; LA64-NEXT: add.d $a0, $a2, $a0
212-
; LA64-NEXT: addi.d $s1, $a0, 16
208+
; LA64-NEXT: add.d $s1, $a2, $a0
213209
; LA64-NEXT: blez $s0, .LBB1_3
214210
; LA64-NEXT: # %bb.1: # %for.body.preheader
215211
; LA64-NEXT: move $fp, $a2
@@ -220,15 +216,15 @@ define void @sink_fold_f32(i64 %k, i64 %n, ptr %a) nounwind {
220216
; LA64-NEXT: move $a0, $fp
221217
; LA64-NEXT: pcaddu18i $ra, %call36(f)
222218
; LA64-NEXT: jirl $ra, $ra, 0
223-
; LA64-NEXT: fld.s $fa0, $s1, 0
219+
; LA64-NEXT: fld.s $fa0, $s1, 16
224220
; LA64-NEXT: addi.d $s0, $s0, -1
225221
; LA64-NEXT: fadd.s $fs0, $fa0, $fs0
226222
; LA64-NEXT: bnez $s0, .LBB1_2
227223
; LA64-NEXT: b .LBB1_4
228224
; LA64-NEXT: .LBB1_3:
229225
; LA64-NEXT: movgr2fr.w $fs0, $zero
230226
; LA64-NEXT: .LBB1_4: # %for.cond.cleanup
231-
; LA64-NEXT: fst.s $fs0, $s1, 0
227+
; LA64-NEXT: fst.s $fs0, $s1, 16
232228
; LA64-NEXT: fld.d $fs0, $sp, 8 # 8-byte Folded Reload
233229
; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload
234230
; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload
@@ -271,14 +267,13 @@ define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind {
271267
; LA32-NEXT: move $s0, $a3
272268
; LA32-NEXT: move $s1, $a2
273269
; LA32-NEXT: slli.w $a0, $a0, 6
274-
; LA32-NEXT: add.w $a0, $a4, $a0
275270
; LA32-NEXT: sltui $a1, $a3, 1
276271
; LA32-NEXT: slti $a2, $a3, 0
277272
; LA32-NEXT: masknez $a2, $a2, $a1
278273
; LA32-NEXT: sltui $a3, $s1, 1
279274
; LA32-NEXT: maskeqz $a1, $a3, $a1
280275
; LA32-NEXT: or $a1, $a1, $a2
281-
; LA32-NEXT: addi.w $s2, $a0, 16
276+
; LA32-NEXT: add.w $s2, $a4, $a0
282277
; LA32-NEXT: bnez $a1, .LBB2_3
283278
; LA32-NEXT: # %bb.1: # %for.body.preheader
284279
; LA32-NEXT: move $fp, $a4
@@ -291,7 +286,7 @@ define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind {
291286
; LA32-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill
292287
; LA32-NEXT: move $a0, $fp
293288
; LA32-NEXT: bl f
294-
; LA32-NEXT: vld $vr0, $s2, 0
289+
; LA32-NEXT: vld $vr0, $s2, 16
295290
; LA32-NEXT: addi.w $s3, $s3, 1
296291
; LA32-NEXT: sltui $a0, $s3, 1
297292
; LA32-NEXT: add.w $s4, $s4, $a0
@@ -307,7 +302,7 @@ define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind {
307302
; LA32-NEXT: .LBB2_3:
308303
; LA32-NEXT: vrepli.b $vr0, 0
309304
; LA32-NEXT: .LBB2_4: # %for.cond.cleanup
310-
; LA32-NEXT: vst $vr0, $s2, 0
305+
; LA32-NEXT: vst $vr0, $s2, 16
311306
; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload
312307
; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload
313308
; LA32-NEXT: ld.w $s2, $sp, 28 # 4-byte Folded Reload
@@ -326,8 +321,7 @@ define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind {
326321
; LA64-NEXT: st.d $s0, $sp, 24 # 8-byte Folded Spill
327322
; LA64-NEXT: st.d $s1, $sp, 16 # 8-byte Folded Spill
328323
; LA64-NEXT: slli.d $a0, $a0, 6
329-
; LA64-NEXT: add.d $a0, $a2, $a0
330-
; LA64-NEXT: addi.d $s1, $a0, 16
324+
; LA64-NEXT: add.d $s1, $a2, $a0
331325
; LA64-NEXT: blez $a1, .LBB2_3
332326
; LA64-NEXT: # %bb.1: # %for.body.preheader
333327
; LA64-NEXT: move $fp, $a2
@@ -340,7 +334,7 @@ define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind {
340334
; LA64-NEXT: move $a0, $fp
341335
; LA64-NEXT: pcaddu18i $ra, %call36(f)
342336
; LA64-NEXT: jirl $ra, $ra, 0
343-
; LA64-NEXT: vld $vr0, $s1, 0
337+
; LA64-NEXT: vld $vr0, $s1, 16
344338
; LA64-NEXT: addi.d $s0, $s0, -1
345339
; LA64-NEXT: vld $vr1, $sp, 0 # 16-byte Folded Reload
346340
; LA64-NEXT: vadd.w $vr1, $vr0, $vr1
@@ -351,7 +345,7 @@ define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind {
351345
; LA64-NEXT: .LBB2_3:
352346
; LA64-NEXT: vrepli.b $vr0, 0
353347
; LA64-NEXT: .LBB2_4: # %for.cond.cleanup
354-
; LA64-NEXT: vst $vr0, $s1, 0
348+
; LA64-NEXT: vst $vr0, $s1, 16
355349
; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload
356350
; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload
357351
; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload
@@ -393,14 +387,13 @@ define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind {
393387
; LA32-NEXT: move $s0, $a3
394388
; LA32-NEXT: move $s1, $a2
395389
; LA32-NEXT: slli.w $a0, $a0, 6
396-
; LA32-NEXT: add.w $a0, $a4, $a0
397390
; LA32-NEXT: sltui $a1, $a3, 1
398391
; LA32-NEXT: slti $a2, $a3, 0
399392
; LA32-NEXT: masknez $a2, $a2, $a1
400393
; LA32-NEXT: sltui $a3, $s1, 1
401394
; LA32-NEXT: maskeqz $a1, $a3, $a1
402395
; LA32-NEXT: or $a1, $a1, $a2
403-
; LA32-NEXT: addi.w $s2, $a0, 32
396+
; LA32-NEXT: add.w $s2, $a4, $a0
404397
; LA32-NEXT: bnez $a1, .LBB3_3
405398
; LA32-NEXT: # %bb.1: # %for.body.preheader
406399
; LA32-NEXT: move $fp, $a4
@@ -413,7 +406,7 @@ define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind {
413406
; LA32-NEXT: xvst $xr0, $sp, 16 # 32-byte Folded Spill
414407
; LA32-NEXT: move $a0, $fp
415408
; LA32-NEXT: bl f
416-
; LA32-NEXT: xvld $xr0, $s2, 0
409+
; LA32-NEXT: xvld $xr0, $s2, 32
417410
; LA32-NEXT: addi.w $s3, $s3, 1
418411
; LA32-NEXT: sltui $a0, $s3, 1
419412
; LA32-NEXT: add.w $s4, $s4, $a0
@@ -429,7 +422,7 @@ define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind {
429422
; LA32-NEXT: .LBB3_3:
430423
; LA32-NEXT: xvrepli.b $xr0, 0
431424
; LA32-NEXT: .LBB3_4: # %for.cond.cleanup
432-
; LA32-NEXT: xvst $xr0, $s2, 0
425+
; LA32-NEXT: xvst $xr0, $s2, 32
433426
; LA32-NEXT: ld.w $s4, $sp, 52 # 4-byte Folded Reload
434427
; LA32-NEXT: ld.w $s3, $sp, 56 # 4-byte Folded Reload
435428
; LA32-NEXT: ld.w $s2, $sp, 60 # 4-byte Folded Reload
@@ -448,8 +441,7 @@ define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind {
448441
; LA64-NEXT: st.d $s0, $sp, 56 # 8-byte Folded Spill
449442
; LA64-NEXT: st.d $s1, $sp, 48 # 8-byte Folded Spill
450443
; LA64-NEXT: slli.d $a0, $a0, 6
451-
; LA64-NEXT: add.d $a0, $a2, $a0
452-
; LA64-NEXT: addi.d $s1, $a0, 32
444+
; LA64-NEXT: add.d $s1, $a2, $a0
453445
; LA64-NEXT: blez $a1, .LBB3_3
454446
; LA64-NEXT: # %bb.1: # %for.body.preheader
455447
; LA64-NEXT: move $fp, $a2
@@ -462,7 +454,7 @@ define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind {
462454
; LA64-NEXT: move $a0, $fp
463455
; LA64-NEXT: pcaddu18i $ra, %call36(f)
464456
; LA64-NEXT: jirl $ra, $ra, 0
465-
; LA64-NEXT: xvld $xr0, $s1, 0
457+
; LA64-NEXT: xvld $xr0, $s1, 32
466458
; LA64-NEXT: addi.d $s0, $s0, -1
467459
; LA64-NEXT: xvld $xr1, $sp, 16 # 32-byte Folded Reload
468460
; LA64-NEXT: xvadd.h $xr1, $xr0, $xr1
@@ -473,7 +465,7 @@ define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind {
473465
; LA64-NEXT: .LBB3_3:
474466
; LA64-NEXT: xvrepli.b $xr0, 0
475467
; LA64-NEXT: .LBB3_4: # %for.cond.cleanup
476-
; LA64-NEXT: xvst $xr0, $s1, 0
468+
; LA64-NEXT: xvst $xr0, $s1, 32
477469
; LA64-NEXT: ld.d $s1, $sp, 48 # 8-byte Folded Reload
478470
; LA64-NEXT: ld.d $s0, $sp, 56 # 8-byte Folded Reload
479471
; LA64-NEXT: ld.d $fp, $sp, 64 # 8-byte Folded Reload
@@ -516,14 +508,13 @@ define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind {
516508
; LA32-NEXT: move $s1, $a2
517509
; LA32-NEXT: slli.w $a1, $a0, 4
518510
; LA32-NEXT: alsl.w $a0, $a0, $a1, 3
519-
; LA32-NEXT: add.w $a0, $a4, $a0
520511
; LA32-NEXT: sltui $a1, $a3, 1
521512
; LA32-NEXT: slti $a2, $a3, 0
522513
; LA32-NEXT: masknez $a2, $a2, $a1
523514
; LA32-NEXT: sltui $a3, $s1, 1
524515
; LA32-NEXT: maskeqz $a1, $a3, $a1
525516
; LA32-NEXT: or $a1, $a1, $a2
526-
; LA32-NEXT: addi.w $s2, $a0, 16
517+
; LA32-NEXT: add.w $s2, $a4, $a0
527518
; LA32-NEXT: bnez $a1, .LBB4_3
528519
; LA32-NEXT: # %bb.1: # %for.body.preheader
529520
; LA32-NEXT: move $fp, $a4
@@ -536,7 +527,7 @@ define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind {
536527
; LA32-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill
537528
; LA32-NEXT: move $a0, $fp
538529
; LA32-NEXT: bl f
539-
; LA32-NEXT: vldrepl.b $vr0, $s2, 0
530+
; LA32-NEXT: vldrepl.b $vr0, $s2, 16
540531
; LA32-NEXT: addi.w $s3, $s3, 1
541532
; LA32-NEXT: sltui $a0, $s3, 1
542533
; LA32-NEXT: add.w $s4, $s4, $a0
@@ -552,7 +543,7 @@ define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind {
552543
; LA32-NEXT: .LBB4_3:
553544
; LA32-NEXT: vrepli.b $vr0, 0
554545
; LA32-NEXT: .LBB4_4: # %for.cond.cleanup
555-
; LA32-NEXT: vstelm.b $vr0, $s2, 0, 1
546+
; LA32-NEXT: vstelm.b $vr0, $s2, 16, 1
556547
; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload
557548
; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload
558549
; LA32-NEXT: ld.w $s2, $sp, 28 # 4-byte Folded Reload
@@ -573,8 +564,7 @@ define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind {
573564
; LA64-NEXT: move $s0, $a1
574565
; LA64-NEXT: slli.d $a1, $a0, 4
575566
; LA64-NEXT: alsl.d $a0, $a0, $a1, 3
576-
; LA64-NEXT: add.d $a0, $a2, $a0
577-
; LA64-NEXT: addi.d $s1, $a0, 16
567+
; LA64-NEXT: add.d $s1, $a2, $a0
578568
; LA64-NEXT: blez $s0, .LBB4_3
579569
; LA64-NEXT: # %bb.1: # %for.body.preheader
580570
; LA64-NEXT: move $fp, $a2
@@ -586,7 +576,7 @@ define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind {
586576
; LA64-NEXT: move $a0, $fp
587577
; LA64-NEXT: pcaddu18i $ra, %call36(f)
588578
; LA64-NEXT: jirl $ra, $ra, 0
589-
; LA64-NEXT: vldrepl.b $vr0, $s1, 0
579+
; LA64-NEXT: vldrepl.b $vr0, $s1, 16
590580
; LA64-NEXT: addi.d $s0, $s0, -1
591581
; LA64-NEXT: vld $vr1, $sp, 0 # 16-byte Folded Reload
592582
; LA64-NEXT: vadd.b $vr1, $vr0, $vr1
@@ -597,7 +587,7 @@ define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind {
597587
; LA64-NEXT: .LBB4_3:
598588
; LA64-NEXT: vrepli.b $vr0, 0
599589
; LA64-NEXT: .LBB4_4: # %for.cond.cleanup
600-
; LA64-NEXT: vstelm.b $vr0, $s1, 0, 1
590+
; LA64-NEXT: vstelm.b $vr0, $s1, 16, 1
601591
; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload
602592
; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload
603593
; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload
@@ -643,14 +633,13 @@ define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind {
643633
; LA32-NEXT: move $s1, $a2
644634
; LA32-NEXT: slli.w $a1, $a0, 4
645635
; LA32-NEXT: alsl.w $a0, $a0, $a1, 3
646-
; LA32-NEXT: add.w $a0, $a4, $a0
647636
; LA32-NEXT: sltui $a1, $a3, 1
648637
; LA32-NEXT: slti $a2, $a3, 0
649638
; LA32-NEXT: masknez $a2, $a2, $a1
650639
; LA32-NEXT: sltui $a3, $s1, 1
651640
; LA32-NEXT: maskeqz $a1, $a3, $a1
652641
; LA32-NEXT: or $a1, $a1, $a2
653-
; LA32-NEXT: addi.w $s2, $a0, 8
642+
; LA32-NEXT: add.w $s2, $a4, $a0
654643
; LA32-NEXT: bnez $a1, .LBB5_3
655644
; LA32-NEXT: # %bb.1: # %for.body.preheader
656645
; LA32-NEXT: move $fp, $a4
@@ -663,7 +652,7 @@ define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind {
663652
; LA32-NEXT: xvst $xr0, $sp, 16 # 32-byte Folded Spill
664653
; LA32-NEXT: move $a0, $fp
665654
; LA32-NEXT: bl f
666-
; LA32-NEXT: xvldrepl.d $xr0, $s2, 0
655+
; LA32-NEXT: xvldrepl.d $xr0, $s2, 8
667656
; LA32-NEXT: addi.w $s3, $s3, 1
668657
; LA32-NEXT: sltui $a0, $s3, 1
669658
; LA32-NEXT: add.w $s4, $s4, $a0
@@ -679,7 +668,7 @@ define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind {
679668
; LA32-NEXT: .LBB5_3:
680669
; LA32-NEXT: xvrepli.b $xr0, 0
681670
; LA32-NEXT: .LBB5_4: # %for.cond.cleanup
682-
; LA32-NEXT: xvstelm.d $xr0, $s2, 0, 1
671+
; LA32-NEXT: xvstelm.d $xr0, $s2, 8, 1
683672
; LA32-NEXT: ld.w $s4, $sp, 52 # 4-byte Folded Reload
684673
; LA32-NEXT: ld.w $s3, $sp, 56 # 4-byte Folded Reload
685674
; LA32-NEXT: ld.w $s2, $sp, 60 # 4-byte Folded Reload
@@ -700,8 +689,7 @@ define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind {
700689
; LA64-NEXT: move $s0, $a1
701690
; LA64-NEXT: slli.d $a1, $a0, 4
702691
; LA64-NEXT: alsl.d $a0, $a0, $a1, 3
703-
; LA64-NEXT: add.d $a0, $a2, $a0
704-
; LA64-NEXT: addi.d $s1, $a0, 8
692+
; LA64-NEXT: add.d $s1, $a2, $a0
705693
; LA64-NEXT: blez $s0, .LBB5_3
706694
; LA64-NEXT: # %bb.1: # %for.body.preheader
707695
; LA64-NEXT: move $fp, $a2
@@ -713,7 +701,7 @@ define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind {
713701
; LA64-NEXT: move $a0, $fp
714702
; LA64-NEXT: pcaddu18i $ra, %call36(f)
715703
; LA64-NEXT: jirl $ra, $ra, 0
716-
; LA64-NEXT: xvldrepl.d $xr0, $s1, 0
704+
; LA64-NEXT: xvldrepl.d $xr0, $s1, 8
717705
; LA64-NEXT: addi.d $s0, $s0, -1
718706
; LA64-NEXT: xvld $xr1, $sp, 16 # 32-byte Folded Reload
719707
; LA64-NEXT: xvfadd.d $xr1, $xr0, $xr1
@@ -724,7 +712,7 @@ define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind {
724712
; LA64-NEXT: .LBB5_3:
725713
; LA64-NEXT: xvrepli.b $xr0, 0
726714
; LA64-NEXT: .LBB5_4: # %for.cond.cleanup
727-
; LA64-NEXT: xvstelm.d $xr0, $s1, 0, 1
715+
; LA64-NEXT: xvstelm.d $xr0, $s1, 8, 1
728716
; LA64-NEXT: ld.d $s1, $sp, 48 # 8-byte Folded Reload
729717
; LA64-NEXT: ld.d $s0, $sp, 56 # 8-byte Folded Reload
730718
; LA64-NEXT: ld.d $fp, $sp, 64 # 8-byte Folded Reload

0 commit comments

Comments
 (0)