d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Optimize RISC-V copyin(9)/copyout(9) routines.

Browse Source 
The existing copyin(9) and copyout(9) routines on RISC-V perform only a
simple byte-by-byte copy.  Improve their performance by performing
word-sized copies where possible.

Submitted by:	Mitchell Horne <mhorne063@gmail.com>
MFC after:	1 week
Differential Revision:	https://reviews.freebsd.org/D18851
This commit is contained in:
markj 2019-01-21 19:38:53 +00:00
parent d33ac4c04b
commit d0ae1da806
2 changed files with 46 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
sys/riscv/include/riscvreg.h
View File

@ -155,7 +155,8 @@
#define SATP_MODE_SV39 (8ULL << SATP_MODE_S)
#define SATP_MODE_SV48 (9ULL << SATP_MODE_S)
#define XLEN 8
#define XLEN __riscv_xlen
#define XLEN_BYTES (XLEN / 8)
#define INSN_SIZE 4
#define INSN_C_SIZE 2

54
sys/riscv/riscv/copyinout.S
View File

@ -1,5 +1,6 @@
/*-
* Copyright (c) 2015-2018 Ruslan Bukin <br@bsdpad.com>
* Copyright (c) 2019 Mitchell Horne
* All rights reserved.
*
* Portions of this software were developed by SRI International and the
@ -59,19 +60,52 @@ END(copyio_fault)
* a2 - Size of copy
*/
.macro copycommon
la a6, copyio_fault /* Get the handler address */
SET_FAULT_HANDLER(a6, a7) /* Set the handler */
la a6, copyio_fault /* Get the handler address */
SET_FAULT_HANDLER(a6, a7) /* Set the handler */
ENTER_USER_ACCESS(a7)
1: lb a4, 0(a0) /* Load from src */
addi a0, a0, 1
sb a4, 0(a1) /* Store in dest */
addi a1, a1, 1
addi a2, a2, -1 /* len-- */
bnez a2, 1b
li t2, XLEN_BYTES
blt a2, t2, 3f /* Byte-copy if len < XLEN_BYTES */
EXIT_USER_ACCESS(a7)
SET_FAULT_HANDLER(x0, a7) /* Clear the handler */
/*
* Compare lower bits of src and dest.
* If they are aligned with each other, we can do word copy.
*/
andi t0, a0, (XLEN_BYTES-1) /* Low bits of src */
andi t1, a1, (XLEN_BYTES-1) /* Low bits of dest */
bne t0, t1, 3f /* Misaligned. Go to byte copy */
beqz t0, 2f /* Already word-aligned, skip ahead */
/* Byte copy until the first word-aligned address */
1: lb a4, 0(a0) /* Load byte from src */
addi a0, a0, 1
sb a4, 0(a1) /* Store byte in dest */
addi a1, a1, 1
addi a2, a2, -1 /* len-- */
andi t0, a0, (XLEN_BYTES-1)
bnez t0, 1b
/* Copy words */
2: ld a4, 0(a0) /* Load word from src */
addi a0, a0, XLEN_BYTES
sd a4, 0(a1) /* Store word in dest */
addi a1, a1, XLEN_BYTES
addi a2, a2, -XLEN_BYTES /* len -= XLEN_BYTES */
bgeu a2, t2, 2b /* Again if len >= XLEN_BYTES */
/* Check if we're finished */
beqz a2, 4f
/* Copy any remaining bytes */
3: lb a4, 0(a0) /* Load byte from src */
addi a0, a0, 1
sb a4, 0(a1) /* Store byte in dest */
addi a1, a1, 1
addi a2, a2, -1 /* len-- */
bnez a2, 3b
4: EXIT_USER_ACCESS(a7)
SET_FAULT_HANDLER(x0, a7) /* Clear the handler */
.endm
/*