URL
https://opencores.org/ocsvn/or1k_soc_on_altera_embedded_dev_kit/or1k_soc_on_altera_embedded_dev_kit/trunk
Subversion Repositories or1k_soc_on_altera_embedded_dev_kit
[/] [or1k_soc_on_altera_embedded_dev_kit/] [trunk/] [linux-2.6/] [linux-2.6.24/] [arch/] [m32r/] [lib/] [checksum.S] - Rev 3
Compare with Previous | Blame | View Log
/** INET An implementation of the TCP/IP protocol suite for the LINUX* operating system. INET is implemented using the BSD Socket* interface as the means of communication with the user level.** IP/TCP/UDP checksumming routines** Authors: Jorge Cwik, <jorge@laser.satlink.net>* Arnt Gulbrandsen, <agulbra@nvg.unit.no>* Tom May, <ftom@netcom.com>* Pentium Pro/II routines:* Alexander Kjeldaas <astor@guardian.no>* Finn Arne Gangstad <finnag@guardian.no>* Lots of code moved from tcp.c and ip.c; see those files* for more names.** Changes: Ingo Molnar, converted csum_partial_copy() to 2.1 exception* handling.* Andi Kleen, add zeroing on error* converted to pure assembler* Hirokazu Takata,Hiroyuki Kondo rewrite for the m32r architecture.** This program is free software; you can redistribute it and/or* modify it under the terms of the GNU General Public License* as published by the Free Software Foundation; either version* 2 of the License, or (at your option) any later version.*/#include <linux/linkage.h>#include <asm/assembler.h>#include <asm/errno.h>/** computes a partial checksum, e.g. for TCP/UDP fragments*//*unsigned int csum_partial(const unsigned char * buff, int len, unsigned int sum)*/#ifdef CONFIG_ISA_DUAL_ISSUE/** Experiments with Ethernet and SLIP connections show that buff* is aligned on either a 2-byte or 4-byte boundary. We get at* least a twofold speedup on 486 and Pentium if it is 4-byte aligned.* Fortunately, it is easy to convert 2-byte alignment to 4-byte* alignment for the unrolled loop.*/.textENTRY(csum_partial); Function args; r0: unsigned char *buff; r1: int len; r2: unsigned int sumpush r2 || ldi r2, #0and3 r7, r0, #1 ; Check alignment.beqz r7, 1f ; Jump if alignment is ok.; 1-byte mis alignedldub r4, @r0 || addi r0, #1; clear c-bit || Alignment uses up bytes.cmp r0, r0 || addi r1, #-1ldi r3, #0 || addx r2, r4addx r2, r3.fillinsn1:and3 r4, r0, #2 ; Check alignment.beqz r4, 2f ; Jump if alignment is ok.; clear c-bit || Alignment uses up two bytes.cmp r0, r0 || addi r1, #-2bgtz r1, 1f ; Jump if we had at least two bytes.bra 4f || addi r1, #2.fillinsn ; len(r1) was < 2. Deal with it.1:; 2-byte alignedlduh r4, @r0 || ldi r3, #0addx r2, r4 || addi r0, #2addx r2, r3.fillinsn2:; 4-byte alignedcmp r0, r0 ; clear c-bitsrl3 r6, r1, #5beqz r6, 2f.fillinsn1: ld r3, @r0+ld r4, @r0+ ; +4ld r5, @r0+ ; +8ld r3, @r0+ || addx r2, r3 ; +12ld r4, @r0+ || addx r2, r4 ; +16ld r5, @r0+ || addx r2, r5 ; +20ld r3, @r0+ || addx r2, r3 ; +24ld r4, @r0+ || addx r2, r4 ; +28addx r2, r5 || addi r6, #-1addx r2, r3addx r2, r4bnez r6, 1baddx r2, r6 ; r6=0cmp r0, r0 ; This clears c-bit.fillinsn2: and3 r6, r1, #0x1c ; withdraw lenbeqz r6, 4fsrli r6, #2.fillinsn3: ld r4, @r0+ || addi r6, #-1addx r2, r4bnez r6, 3baddx r2, r6 ; r6=0cmp r0, r0 ; This clears c-bit.fillinsn4: and3 r1, r1, #3beqz r1, 7f ; if len == 0 goto endand3 r6, r1, #2beqz r6, 5f ; if len < 2 goto 5f(1byte)lduh r4, @r0 || addi r0, #2addi r1, #-2 || slli r4, #16addx r2, r4beqz r1, 6f.fillinsn5: ldub r4, @r0 || ldi r1, #0#ifndef __LITTLE_ENDIAN__slli r4, #8#endifaddx r2, r4.fillinsn6: addx r2, r1.fillinsn7:and3 r0, r2, #0xffffsrli r2, #16add r0, r2srl3 r2, r0, #16beqz r2, 1faddi r0, #1and3 r0, r0, #0xffff.fillinsn1:beqz r7, 1f ; swap the upper byte for the lowerand3 r2, r0, #0xffsrl3 r0, r0, #8slli r2, #8or r0, r2.fillinsn1:pop r2 || cmp r0, r0addx r0, r2 || ldi r2, #0addx r0, r2jmp r14#else /* not CONFIG_ISA_DUAL_ISSUE *//** Experiments with Ethernet and SLIP connections show that buff* is aligned on either a 2-byte or 4-byte boundary. We get at* least a twofold speedup on 486 and Pentium if it is 4-byte aligned.* Fortunately, it is easy to convert 2-byte alignment to 4-byte* alignment for the unrolled loop.*/.textENTRY(csum_partial); Function args; r0: unsigned char *buff; r1: int len; r2: unsigned int sumpush r2ldi r2, #0and3 r7, r0, #1 ; Check alignment.beqz r7, 1f ; Jump if alignment is ok.; 1-byte mis alignedldub r4, @r0addi r0, #1addi r1, #-1 ; Alignment uses up bytes.cmp r0, r0 ; clear c-bitldi r3, #0addx r2, r4addx r2, r3.fillinsn1:and3 r4, r0, #2 ; Check alignment.beqz r4, 2f ; Jump if alignment is ok.addi r1, #-2 ; Alignment uses up two bytes.cmp r0, r0 ; clear c-bitbgtz r1, 1f ; Jump if we had at least two bytes.addi r1, #2 ; len(r1) was < 2. Deal with it.bra 4f.fillinsn1:; 2-byte alignedlduh r4, @r0addi r0, #2ldi r3, #0addx r2, r4addx r2, r3.fillinsn2:; 4-byte alignedcmp r0, r0 ; clear c-bitsrl3 r6, r1, #5beqz r6, 2f.fillinsn1: ld r3, @r0+ld r4, @r0+ ; +4ld r5, @r0+ ; +8addx r2, r3addx r2, r4addx r2, r5ld r3, @r0+ ; +12ld r4, @r0+ ; +16ld r5, @r0+ ; +20addx r2, r3addx r2, r4addx r2, r5ld r3, @r0+ ; +24ld r4, @r0+ ; +28addi r6, #-1addx r2, r3addx r2, r4bnez r6, 1baddx r2, r6 ; r6=0cmp r0, r0 ; This clears c-bit.fillinsn2: and3 r6, r1, #0x1c ; withdraw lenbeqz r6, 4fsrli r6, #2.fillinsn3: ld r4, @r0+addi r6, #-1addx r2, r4bnez r6, 3baddx r2, r6 ; r6=0cmp r0, r0 ; This clears c-bit.fillinsn4: and3 r1, r1, #3beqz r1, 7f ; if len == 0 goto endand3 r6, r1, #2beqz r6, 5f ; if len < 2 goto 5f(1byte)lduh r4, @r0addi r0, #2addi r1, #-2slli r4, #16addx r2, r4beqz r1, 6f.fillinsn5: ldub r4, @r0#ifndef __LITTLE_ENDIAN__slli r4, #8#endifaddx r2, r4.fillinsn6: ldi r5, #0addx r2, r5.fillinsn7:and3 r0, r2, #0xffffsrli r2, #16add r0, r2srl3 r2, r0, #16beqz r2, 1faddi r0, #1and3 r0, r0, #0xffff.fillinsn1:beqz r7, 1fmv r2, r0srl3 r0, r2, #8and3 r2, r2, #0xffslli r2, #8or r0, r2.fillinsn1:pop r2cmp r0, r0addx r0, r2ldi r2, #0addx r0, r2jmp r14#endif /* not CONFIG_ISA_DUAL_ISSUE *//*unsigned int csum_partial_copy_generic (const char *src, char *dst,int len, int sum, int *src_err_ptr, int *dst_err_ptr)*//** Copy from ds while checksumming, otherwise like csum_partial** The macros SRC and DST specify the type of access for the instruction.* thus we can call a custom exception handler for all access types.** FIXME: could someone double-check whether I haven't mixed up some SRC and* DST definitions? It's damn hard to trigger all cases. I hope I got* them all but there's no guarantee.*/ENTRY(csum_partial_copy_generic)nopnopnopnopjmp r14nopnopnop.end