Debian Codegraph @

gcc/testsuite/gcc.target/riscv/memcpy-nonoverlapping.c

raw
1 /* { dg-do compile } */
2 /* { dg-options "-mcpu=sifive-u74 -march=rv64gc -mabi=lp64" } */
3 /* { dg-skip-if "" { *-*-* } { "-O0" "-Os" "-Oz" "-Og" } } */
4
5
6 #define COPY_N(N) \
7 void copy##N (char *src, char *dst) \
8 { \
9 dst = __builtin_assume_aligned (dst, 4096); \
10 src = __builtin_assume_aligned (src, 4096); \
11 __builtin_memcpy (dst, src, N); \
12 }
13
14 /* Emits 1x {ld,sd} and 1x {lhu,lbu,sh,sb}. */
15 COPY_N(11)
16
17 /* Emits 1x {ld,sd} and 1x {lw,lbu,sw,sb}. */
18 COPY_N(13)
19
20 /* Emits 1x {ld,sd} and 1x {lw,lhu,sw,sh}. */
21 COPY_N(14)
22
23 /* Emits 1x {ld,sd} and 1x {lw,lhu,lbu,sw,sh,sb}. */
24 COPY_N(15)
25
26 /* Emits 2x {ld,sd} and 1x {lhu,lbu,sh,sb}. */
27 COPY_N(19)
28
29 /* Emits 2x {ld,sd} and 1x {lw,lhu,lbu,sw,sh,sb}. */
30 COPY_N(23)
31
32 /* The by-pieces infrastructure handles up to 24 bytes.
33 So the code below is emitted via cpymemsi/block_move_straight. */
34
35 /* Emits 3x {ld,sd} and 1x {lhu,lbu,sh,sb}. */
36 COPY_N(27)
37
38 /* Emits 3x {ld,sd} and 1x {lw,lbu,sw,sb}. */
39 COPY_N(29)
40
41 /* Emits 3x {ld,sd} and 1x {lw,lhu,lbu,sw,sh,sb}. */
42 COPY_N(31)
43
44 /* { dg-final { scan-assembler-times "ld\t" 17 } } */
45 /* { dg-final { scan-assembler-times "sd\t" 17 } } */
46
47 /* { dg-final { scan-assembler-times "lw\t" 6 } } */
48 /* { dg-final { scan-assembler-times "sw\t" 6 } } */
49
50 /* { dg-final { scan-assembler-times "lhu\t" 7 } } */
51 /* { dg-final { scan-assembler-times "sh\t" 7 } } */
52
53 /* { dg-final { scan-assembler-times "lbu\t" 8 } } */
54 /* { dg-final { scan-assembler-times "sb\t" 8 } } */