Skip to content

Commit 12d6039

Browse files
davemgreenchencha3
davemgreen
authored and
chencha3
committed
[AArch64] Extra tests for v2i8 concat loads. NFC
1 parent cb9e65d commit 12d6039

File tree

1 file changed

+127
-2
lines changed

1 file changed

+127
-2
lines changed

‎llvm/test/CodeGen/AArch64/insert-subvector.ll

Lines changed: 127 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -374,6 +374,131 @@ define <16 x i8> @load_v16i8_8_2(float %tmp, <16 x i8> %b, ptr %a) {
374374
ret <16 x i8> %s2
375375
}
376376

377+
define <8 x i8> @load_v8i8_2_1(float %tmp, <8 x i8> %b, ptr %a) {
378+
; CHECK-LABEL: load_v8i8_2_1:
379+
; CHECK: // %bb.0:
380+
; CHECK-NEXT: ld1 { v2.b }[0], [x0]
381+
; CHECK-NEXT: add x8, x0, #1
382+
; CHECK-NEXT: mov v0.16b, v2.16b
383+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
384+
; CHECK-NEXT: mov v2.b[1], v0.b[4]
385+
; CHECK-NEXT: fmov d0, d1
386+
; CHECK-NEXT: mov v0.h[0], v2.h[0]
387+
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
388+
; CHECK-NEXT: ret
389+
%l = load <2 x i8>, ptr %a
390+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
391+
%s2 = shufflevector <8 x i8> %s1, <8 x i8> %b, <8 x i32> <i32 0, i32 1, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
392+
ret <8 x i8> %s2
393+
}
394+
395+
define <8 x i8> @load_v8i8_2_15(float %tmp, <8 x i8> %b, ptr %a) {
396+
; CHECK-LABEL: load_v8i8_2_15:
397+
; CHECK: // %bb.0:
398+
; CHECK-NEXT: ld1 { v0.b }[0], [x0]
399+
; CHECK-NEXT: add x8, x0, #1
400+
; CHECK-NEXT: // kill: def $d1 killed $d1 def $q1
401+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
402+
; CHECK-NEXT: adrp x8, .LCPI33_0
403+
; CHECK-NEXT: mov v0.b[1], v0.b[4]
404+
; CHECK-NEXT: mov v0.d[1], v1.d[0]
405+
; CHECK-NEXT: ldr d1, [x8, :lo12:.LCPI33_0]
406+
; CHECK-NEXT: tbl v0.8b, { v0.16b }, v1.8b
407+
; CHECK-NEXT: ret
408+
%l = load <2 x i8>, ptr %a
409+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
410+
%s2 = shufflevector <8 x i8> %s1, <8 x i8> %b, <8 x i32> <i32 8, i32 0, i32 1, i32 11, i32 12, i32 13, i32 14, i32 15>
411+
ret <8 x i8> %s2
412+
}
413+
414+
define <8 x i8> @load_v8i8_2_2(float %tmp, <8 x i8> %b, ptr %a) {
415+
; CHECK-LABEL: load_v8i8_2_2:
416+
; CHECK: // %bb.0:
417+
; CHECK-NEXT: ld1 { v2.b }[0], [x0]
418+
; CHECK-NEXT: add x8, x0, #1
419+
; CHECK-NEXT: mov v0.16b, v2.16b
420+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
421+
; CHECK-NEXT: mov v2.b[1], v0.b[4]
422+
; CHECK-NEXT: fmov d0, d1
423+
; CHECK-NEXT: mov v0.h[1], v2.h[0]
424+
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
425+
; CHECK-NEXT: ret
426+
%l = load <2 x i8>, ptr %a
427+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
428+
%s2 = shufflevector <8 x i8> %s1, <8 x i8> %b, <8 x i32> <i32 8, i32 9, i32 0, i32 1, i32 12, i32 13, i32 14, i32 15>
429+
ret <8 x i8> %s2
430+
}
431+
432+
define <8 x i8> @load_v8i8_2_3(float %tmp, <8 x i8> %b, ptr %a) {
433+
; CHECK-LABEL: load_v8i8_2_3:
434+
; CHECK: // %bb.0:
435+
; CHECK-NEXT: ld1 { v2.b }[0], [x0]
436+
; CHECK-NEXT: add x8, x0, #1
437+
; CHECK-NEXT: mov v0.16b, v2.16b
438+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
439+
; CHECK-NEXT: mov v2.b[1], v0.b[4]
440+
; CHECK-NEXT: fmov d0, d1
441+
; CHECK-NEXT: mov v0.h[2], v2.h[0]
442+
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
443+
; CHECK-NEXT: ret
444+
%l = load <2 x i8>, ptr %a
445+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
446+
%s2 = shufflevector <8 x i8> %s1, <8 x i8> %b, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 0, i32 1, i32 14, i32 15>
447+
ret <8 x i8> %s2
448+
}
449+
450+
define <8 x i8> @load_v8i8_2_4(float %tmp, <8 x i8> %b, ptr %a) {
451+
; CHECK-LABEL: load_v8i8_2_4:
452+
; CHECK: // %bb.0:
453+
; CHECK-NEXT: ld1 { v2.b }[0], [x0]
454+
; CHECK-NEXT: add x8, x0, #1
455+
; CHECK-NEXT: mov v0.16b, v2.16b
456+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
457+
; CHECK-NEXT: mov v2.b[1], v0.b[4]
458+
; CHECK-NEXT: fmov d0, d1
459+
; CHECK-NEXT: mov v0.h[3], v2.h[0]
460+
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
461+
; CHECK-NEXT: ret
462+
%l = load <2 x i8>, ptr %a
463+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
464+
%s2 = shufflevector <8 x i8> %s1, <8 x i8> %b, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 0, i32 1>
465+
ret <8 x i8> %s2
466+
}
467+
468+
define <4 x i8> @load_v4i8_2_1(float %tmp, <4 x i8> %b, ptr %a) {
469+
; CHECK-LABEL: load_v4i8_2_1:
470+
; CHECK: // %bb.0:
471+
; CHECK-NEXT: ld1 { v0.b }[0], [x0]
472+
; CHECK-NEXT: add x8, x0, #1
473+
; CHECK-NEXT: // kill: def $d1 killed $d1 def $q1
474+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
475+
; CHECK-NEXT: uzp1 v0.4h, v0.4h, v0.4h
476+
; CHECK-NEXT: mov v0.s[1], v1.s[1]
477+
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
478+
; CHECK-NEXT: ret
479+
%l = load <2 x i8>, ptr %a
480+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
481+
%s2 = shufflevector <4 x i8> %s1, <4 x i8> %b, <4 x i32> <i32 0, i32 1, i32 6, i32 7>
482+
ret <4 x i8> %s2
483+
}
484+
485+
define <4 x i8> @load_v4i8_2_2(float %tmp, <4 x i8> %b, ptr %a) {
486+
; CHECK-LABEL: load_v4i8_2_2:
487+
; CHECK: // %bb.0:
488+
; CHECK-NEXT: ld1 { v0.b }[0], [x0]
489+
; CHECK-NEXT: add x8, x0, #1
490+
; CHECK-NEXT: ld1 { v0.b }[4], [x8]
491+
; CHECK-NEXT: uzp1 v2.4h, v0.4h, v0.4h
492+
; CHECK-NEXT: fmov d0, d1
493+
; CHECK-NEXT: mov v0.s[1], v2.s[0]
494+
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
495+
; CHECK-NEXT: ret
496+
%l = load <2 x i8>, ptr %a
497+
%s1 = shufflevector <2 x i8> %l, <2 x i8> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
498+
%s2 = shufflevector <4 x i8> %s1, <4 x i8> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
499+
ret <4 x i8> %s2
500+
}
501+
377502
; i16
378503

379504
define <8 x i16> @load_v8i16_2_1(float %tmp, <8 x i16> %b, ptr %a) {
@@ -400,10 +525,10 @@ define <8 x i16> @load_v8i16_2_15(float %tmp, <8 x i16> %b, ptr %a) {
400525
; CHECK-NEXT: add x9, x0, #2
401526
; CHECK-NEXT: // kill: def $q1 killed $q1 def $q0_q1
402527
; CHECK-NEXT: fmov s2, w8
403-
; CHECK-NEXT: adrp x8, .LCPI33_0
528+
; CHECK-NEXT: adrp x8, .LCPI40_0
404529
; CHECK-NEXT: ld1 { v2.h }[2], [x9]
405530
; CHECK-NEXT: xtn v0.4h, v2.4s
406-
; CHECK-NEXT: ldr q2, [x8, :lo12:.LCPI33_0]
531+
; CHECK-NEXT: ldr q2, [x8, :lo12:.LCPI40_0]
407532
; CHECK-NEXT: tbl v0.16b, { v0.16b, v1.16b }, v2.16b
408533
; CHECK-NEXT: ret
409534
%l = load <2 x i16>, ptr %a

0 commit comments

Comments
 (0)