Fcalva/fxlibc
1
0
Fork 0
mirror of https://git.planet-casio.com/Vhex-Kernel-Core/fxlibc.git synced 2024-12-28 20:43:38 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions

string: use gint's optimized memset (DONE)

Browse source
This commit is contained in:
Lephenixnoir 2021-05-23 16:16:45 +02:00
parent bd344d5bb2
commit b69e0fd299
No known key found for this signature in database
GPG key ID: 1BBA026E13FC0495
3 changed files with 75 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
CMakeLists.txt
View file

@ -172,6 +172,7 @@ if(sh-generic IN_LIST TARGET_FOLDERS)
src/libc/setjmp/target/sh-generic/setjmp.S src/libc/setjmp/target/sh-generic/setjmp.S
src/libc/setjmp/target/sh-generic/longjmp.S src/libc/setjmp/target/sh-generic/longjmp.S
src/libc/string/target/sh-generic/memchr.S src/libc/string/target/sh-generic/memchr.S
src/libc/string/target/sh-generic/memset.S
src/libc/string/target/sh-generic/strlen.S src/libc/string/target/sh-generic/strlen.S
src/target/sh-generic/cpucap.c) src/target/sh-generic/cpucap.c)
endif() endif()

11
src/libc/string/memset.c
View file

@ -1,16 +1,13 @@
#include <string.h> #include <string.h>
#include <stdint.h> #include <stdint.h>
/* #ifndef __SUPPORT_ARCH_SH
** The memset() function fills the first n bytes of the memory area pointed to
** by s with the constant byte c.
**
** TODO: use DMA !
** TODO: use DSP ?
*/
void *memset(void *s, int c, size_t n) void *memset(void *s, int c, size_t n)
{ {
while ((int)--n >= 0) while ((int)--n >= 0)
((uint8_t *) s)[n] = c; ((uint8_t *) s)[n] = c;
return (s); return (s);
} }
#endif /*__SUPPORT_ARCH_SH*/

70
src/libc/string/target/sh-generic/memset.S Normal file
View file

@ -0,0 +1,70 @@
.global _memset
.text
_memset:
tst r6, r6
bt .zero
/* We'll fill from the end */
mov r4, r3
add r6, r4
/* When setting less than 64 bytes, use the naive method */
mov #64, r0
cmp/ge r6, r0
bt _naive_memset
mov #3, r2
/* Make a 4-byte filler */
mov r5, r0
shll8 r5
or r5, r0
mov r0, r5
shll16 r5
or r5, r0
_memset_align:
/* 4-align the destination */
mov.b r0, @-r4
tst r2, r4
bf/s _memset_align
dt r6
mov #40, r2
.aligned4_32:
add #-32, r4
add #-32, r6
mov.l r0, @(28,r4)
mov.l r0, @(24,r4)
mov.l r0, @(20,r4)
mov.l r0, @(16,r4)
mov.l r0, @(12,r4)
mov.l r0, @(8,r4)
mov.l r0, @(4,r4)
cmp/ge r6, r2
bf/s .aligned4_32
mov.l r0, @r4
mov #8, r2
.aligned4_4:
mov.l r0, @-r4
cmp/ge r6, r2
bf/s .aligned4_4
add #-4, r6
_naive_memset:
/* Tight loop copy one byte */
dt r6
bf/s _naive_memset
mov.b r5, @-r4
.end:
rts
mov r3, r0
.zero:
rts
mov r4, r0