| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-march=rv32gcv -mabi=ilp32 -mpreferred-stack-boundary=3 -O3 -fno-schedule-insns -fno-schedule-insns2" } */ |
| 3 |
/* { dg-final { check-function-bodies "**" "" } } */ |
| 4 |
|
| 5 |
#include "riscv_vector.h" |
| 6 |
#include "macro.h" |
| 7 |
|
| 8 |
/* |
| 9 |
** spill: |
| 10 |
** csrr\t[a-x0-9]+,vlenb |
| 11 |
** slli\t[a-x0-9]+,[a-x0-9]+,4 |
| 12 |
** sub\tsp,[a-x0-9]+,[a-x0-9]+ |
| 13 |
** vsetvli\t[a-x0-9]+,zero,e8,mf8,ta,ma |
| 14 |
** vle8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 15 |
** csrr\t[a-x0-9]+,vlenb |
| 16 |
** srli\t[a-x0-9]+,[a-x0-9]+,3 |
| 17 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 18 |
** vse8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 19 |
** addi\t[a-x0-9]+,[a-x0-9]+,1 |
| 20 |
** vsetvli\t[a-x0-9]+,zero,e8,mf4,ta,ma |
| 21 |
** vle8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 22 |
** csrr\t[a-x0-9]+,vlenb |
| 23 |
** srli\t[a-x0-9]+,[a-x0-9]+,2 |
| 24 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 25 |
** vse8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 26 |
** addi\t[a-x0-9]+,[a-x0-9]+,2 |
| 27 |
** vsetvli\t[a-x0-9]+,zero,e8,mf2,ta,ma |
| 28 |
** vle8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 29 |
** csrr\t[a-x0-9]+,vlenb |
| 30 |
** srli\t[a-x0-9]+,[a-x0-9]+,1 |
| 31 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 32 |
** vse8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 33 |
** addi\t[a-x0-9]+,[a-x0-9]+,3 |
| 34 |
** vl1re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 35 |
** csrr\t[a-x0-9]+,vlenb |
| 36 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 37 |
** vs1r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 38 |
** addi\t[a-x0-9]+,[a-x0-9]+,4 |
| 39 |
** vl2re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 40 |
** csrr\t[a-x0-9]+,vlenb |
| 41 |
** slli\t[a-x0-9]+,[a-x0-9]+,1 |
| 42 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 43 |
** vs2r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 44 |
** addi\t[a-x0-9]+,[a-x0-9]+,5 |
| 45 |
** vl4re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 46 |
** mv\t[a-x0-9]+,[a-x0-9]+ |
| 47 |
** slli\t[a-x0-9]+,[a-x0-9]+,2 |
| 48 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 49 |
** vs4r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 50 |
** addi\t[a-x0-9]+,[a-x0-9]+,6 |
| 51 |
** vl8re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 52 |
** ... |
| 53 |
** slli\t[a-x0-9]+,[a-x0-9]+,3 |
| 54 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 55 |
** vs8r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 56 |
** ... |
| 57 |
** srli\t[a-x0-9]+,[a-x0-9]+,3 |
| 58 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 59 |
** ... |
| 60 |
** vle8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 61 |
** vse8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 62 |
** addi\t[a-x0-9]+,[a-x0-9]+,1 |
| 63 |
** srli\t[a-x0-9]+,[a-x0-9]+,2 |
| 64 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 65 |
** ... |
| 66 |
** vle8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 67 |
** vse8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 68 |
** addi\t[a-x0-9]+,[a-x0-9]+,2 |
| 69 |
** srli\t[a-x0-9]+,[a-x0-9]+,1 |
| 70 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 71 |
** ... |
| 72 |
** vle8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 73 |
** vse8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 74 |
** addi\t[a-x0-9]+,[a-x0-9]+,3 |
| 75 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 76 |
** vl1re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 77 |
** vs1r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 78 |
** addi\t[a-x0-9]+,[a-x0-9]+,4 |
| 79 |
** slli\t[a-x0-9]+,[a-x0-9]+,1 |
| 80 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 81 |
** vl2re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 82 |
** vs2r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 83 |
** addi\t[a-x0-9]+,[a-x0-9]+,5 |
| 84 |
** slli\t[a-x0-9]+,[a-x0-9]+,2 |
| 85 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 86 |
** vl4re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 87 |
** vs4r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 88 |
** addi\t[a-x0-9]+,[a-x0-9]+,6 |
| 89 |
** slli\t[a-x0-9]+,[a-x0-9]+,3 |
| 90 |
** add\t[a-x0-9]+,[a-x0-9]+,[a-x0-9]+ |
| 91 |
** vl8re8.v\tv[0-9]+,0\([a-x0-9]+\) |
| 92 |
** vs8r.v\tv[0-9]+,0\([a-x0-9]+\) |
| 93 |
** csrr\t[a-x0-9]+,vlenb |
| 94 |
** slli\t[a-x0-9]+,[a-x0-9]+,4 |
| 95 |
** add\tsp,[a-x0-9]+,[a-x0-9]+ |
| 96 |
** ... |
| 97 |
** jr\tra |
| 98 |
*/ |
| 99 |
void |
| 100 |
spill (int8_t *in, int8_t *out) |
| 101 |
{ |
| 102 |
vint8mf8_t v0 = *(vint8mf8_t*)in; |
| 103 |
vint8mf4_t v1 = *(vint8mf4_t*)(in + 1); |
| 104 |
vint8mf2_t v2 = *(vint8mf2_t*)(in + 2); |
| 105 |
vint8m1_t v3 = *(vint8m1_t*)(in + 3); |
| 106 |
vint8m2_t v4 = *(vint8m2_t*)(in + 4); |
| 107 |
vint8m4_t v8 = *(vint8m4_t*)(in + 5); |
| 108 |
vint8m8_t v16 = *(vint8m8_t*)(in + 6); |
| 109 |
exhaust_vector_regs (); |
| 110 |
*(vint8mf8_t*)out = v0; |
| 111 |
*(vint8mf4_t*)(out + 1) = v1; |
| 112 |
*(vint8mf2_t*)(out + 2) = v2; |
| 113 |
*(vint8m1_t*)(out + 3) = v3; |
| 114 |
*(vint8m2_t*)(out + 4) = v4; |
| 115 |
*(vint8m4_t*)(out + 5) = v8; |
| 116 |
*(vint8m8_t*)(out + 6) = v16; |
| 117 |
} |