1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mattr=+sve -force-streaming-compatible < %s | FileCheck %s
; RUN: llc -mattr=+sme -force-streaming < %s | FileCheck %s
; RUN: llc -force-streaming-compatible < %s | FileCheck %s --check-prefix=NONEON-NOSVE
target triple = "aarch64-unknown-linux-gnu"
; Ensure we don't crash when trying to combine fp<->int conversions
define void @fp_convert_combine_crash(ptr %a, ptr %b) {
; CHECK-LABEL: fp_convert_combine_crash:
; CHECK: // %bb.0:
; CHECK-NEXT: fmov z0.s, #8.00000000
; CHECK-NEXT: ldp q1, q2, [x0]
; CHECK-NEXT: ptrue p0.s, vl4
; CHECK-NEXT: fmul z1.s, p0/m, z1.s, z0.s
; CHECK-NEXT: fmul z0.s, p0/m, z0.s, z2.s
; CHECK-NEXT: fcvtzs z1.s, p0/m, z1.s
; CHECK-NEXT: fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT: stp q1, q0, [x1]
; CHECK-NEXT: ret
;
; NONEON-NOSVE-LABEL: fp_convert_combine_crash:
; NONEON-NOSVE: // %bb.0:
; NONEON-NOSVE-NEXT: ldp q0, q1, [x0]
; NONEON-NOSVE-NEXT: stp q0, q1, [sp, #-64]!
; NONEON-NOSVE-NEXT: .cfi_def_cfa_offset 64
; NONEON-NOSVE-NEXT: ldp s1, s0, [sp, #24]
; NONEON-NOSVE-NEXT: fcvtzs w8, s0, #3
; NONEON-NOSVE-NEXT: ldp s0, s2, [sp, #16]
; NONEON-NOSVE-NEXT: fcvtzs w9, s1, #3
; NONEON-NOSVE-NEXT: fcvtzs w10, s2, #3
; NONEON-NOSVE-NEXT: fcvtzs w11, s0, #3
; NONEON-NOSVE-NEXT: ldp s2, s1, [sp, #8]
; NONEON-NOSVE-NEXT: ldp s0, s3, [sp]
; NONEON-NOSVE-NEXT: stp w9, w8, [sp, #56]
; NONEON-NOSVE-NEXT: fcvtzs w12, s1, #3
; NONEON-NOSVE-NEXT: fcvtzs w8, s2, #3
; NONEON-NOSVE-NEXT: stp w11, w10, [sp, #48]
; NONEON-NOSVE-NEXT: fcvtzs w9, s3, #3
; NONEON-NOSVE-NEXT: fcvtzs w10, s0, #3
; NONEON-NOSVE-NEXT: stp w8, w12, [sp, #40]
; NONEON-NOSVE-NEXT: stp w10, w9, [sp, #32]
; NONEON-NOSVE-NEXT: ldp q0, q1, [sp, #32]
; NONEON-NOSVE-NEXT: stp q0, q1, [x1]
; NONEON-NOSVE-NEXT: add sp, sp, #64
; NONEON-NOSVE-NEXT: ret
%f = load <8 x float>, ptr %a
%mul.i = fmul <8 x float> %f, <float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00,
float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00>
%vcvt.i = fptosi <8 x float> %mul.i to <8 x i32>
store <8 x i32> %vcvt.i, ptr %b
ret void
}
|
