1/*-
2 * Copyright (c) 1990 The Regents of the University of California.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 */
29/*-
30 * Copyright (c) 2001 by Thomas Moestl <tmm@FreeBSD.org>.
31 * All rights reserved.
32 *
33 * Redistribution and use in source and binary forms, with or without
34 * modification, are permitted provided that the following conditions
35 * are met:
36 * 1. Redistributions of source code must retain the above copyright
37 *    notice, this list of conditions and the following disclaimer.
38 * 2. Redistributions in binary form must reproduce the above copyright
39 *    notice, this list of conditions and the following disclaimer in the
40 *    documentation and/or other materials provided with the distribution.
41 *
42 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
43 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
44 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
45 * IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT,
46 * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
47 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
48 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
49 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
50 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
51 * USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
52 *
53 *	from tahoe:	in_cksum.c	1.2	86/01/05
54 *	from:		@(#)in_cksum.c	1.3 (Berkeley) 1/19/91
55 * 	from: FreeBSD: src/sys/i386/i386/in_cksum.c,v 1.22 2000/11/25
56 */
57
58#include <sys/cdefs.h>
59__FBSDID("$FreeBSD$");
60
61#include <sys/param.h>
62#include <sys/systm.h>
63#include <sys/mbuf.h>
64
65#include <netinet/in.h>
66#include <netinet/in_systm.h>
67#include <netinet/ip.h>
68
69#include <machine/in_cksum.h>
70
71/*
72 * Checksum routine for Internet Protocol family headers.
73 *
74 * This routine is very heavily used in the network
75 * code and should be modified for each CPU to be as fast as possible.
76 *
77 * This implementation is a sparc64 version.  Most code was taken over
78 * and adapted from the i386.  Some optimizations were changed to achieve
79 * (hopefully) better performance.
80 * This uses 64-bit loads, but 32-bit additions due to the lack of a 64-bit
81 * add-with-carry operation.
82 */
83
84/*
85 * REDUCE() is actually not used that frequently... maybe a C implementation
86 * would suffice.
87 */
88#define	REDUCE(sum, tmp) __asm(						\
89	"sll %2, 16, %1\n"						\
90	"addcc %2, %1, %0\n"						\
91	"srl %0, 16, %0\n"						\
92	"addc %0, 0, %0" : "=r" (sum), "=&r" (tmp) : "0" (sum) : "cc")
93
94/*
95 * Note that some of these macros depend on the flags being preserved
96 * between calls, thus they have to be used within a single __asm().
97 */
98#define	LD64_ADD32(n, mod)						\
99	"ldx [%3 + " #n "], %1\n"					\
100	"add" #mod " %2, %1, %0\n"					\
101	"srlx %1, 32, %1\n"						\
102	"addccc %0, %1, %0\n"
103
104#define	LD32_ADD32(n, mod)						\
105	"lduw [%3 + " #n "], %1\n"					\
106	"add" #mod " %2, %1, %0\n"
107
108#define	MOP(sum, tmp, addr)						\
109	"addc %2, 0, %0"						\
110	: "=r" (sum), "=&r" (tmp) : "0" (sum), "r" (addr) : "cc"
111
112u_short
113in_cksum_skip(struct mbuf *m, int len, int skip)
114{
115	u_short *w;
116	unsigned long tmp, sum = 0;
117	int mlen = 0;
118	int byte_swapped = 0;
119	u_short	su = 0;
120
121	len -= skip;
122	for (; skip > 0 && m != NULL; m = m->m_next) {
123		if (m->m_len > skip) {
124			mlen = m->m_len - skip;
125			w = (u_short *)(mtod(m, u_char *) + skip);
126			goto skip_start;
127		} else
128			skip -= m->m_len;
129	}
130
131	for (; m != NULL && len > 0; m = m->m_next) {
132		if (m->m_len == 0)
133			continue;
134		w = mtod(m, u_short *);
135		if (mlen == -1) {
136			/*
137			 * The first byte of this mbuf is the continuation
138			 * of a word spanning between this mbuf and the
139			 * last mbuf.
140			 *
141			 * The high order byte of su is already saved when
142			 * scanning previous mbuf.  sum was REDUCEd when we
143			 * found mlen == -1
144			 */
145			sum += su | *(u_char *)w;
146			w = (u_short *)((u_char *)w + 1);
147			mlen = m->m_len - 1;
148			len--;
149		} else
150			mlen = m->m_len;
151skip_start:
152		if (len < mlen)
153			mlen = len;
154		len -= mlen;
155		/*
156		 * Force to a 8-byte boundary first so that we can use
157		 * LD64_ADD32.
158		 */
159		if (((u_long)w & 7) != 0) {
160			REDUCE(sum, tmp);
161			if (((u_long)w & 1) != 0 && mlen >= 1) {
162				sum <<= 8;
163				su = *(u_char *)w << 8;
164				w = (u_short *)((u_char *)w + 1);
165				mlen--;
166				byte_swapped = 1;
167			}
168			if (((u_long)w & 2) != 0 && mlen >= 2) {
169				sum += *w++;
170				mlen -= 2;
171			}
172			if (((u_long)w & 4) != 0 && mlen >= 4) {
173				__asm(
174				    LD32_ADD32(0, cc)
175				    MOP(sum, tmp, w)
176				);
177				w += 2;
178				mlen -= 4;
179			}
180		}
181		/*
182		 * Do as much of the checksum as possible 64 bits at at time.
183		 * In fact, this loop is unrolled to make overhead from
184		 * branches &c small.
185		 */
186		for (; mlen >= 64; mlen -= 64) {
187			__asm(
188			    LD64_ADD32(0, cc)
189			    LD64_ADD32(8, ccc)
190			    LD64_ADD32(16, ccc)
191			    LD64_ADD32(24, ccc)
192			    LD64_ADD32(32, ccc)
193			    LD64_ADD32(40, ccc)
194			    LD64_ADD32(48, ccc)
195			    LD64_ADD32(56, ccc)
196			    MOP(sum, tmp, w)
197			);
198			w += 32;
199		}
200		if (mlen >= 32) {
201			__asm(
202			    LD64_ADD32(0, cc)
203			    LD64_ADD32(8, ccc)
204			    LD64_ADD32(16, ccc)
205			    LD64_ADD32(24, ccc)
206			    MOP(sum, tmp, w)
207			);
208			w += 16;
209			mlen -= 32;
210		}
211		if (mlen >= 16) {
212			__asm(
213			    LD64_ADD32(0, cc)
214			    LD64_ADD32(8, ccc)
215			    MOP(sum, tmp, w)
216			);
217			w += 8;
218			mlen -= 16;
219		}
220		if (mlen >= 8) {
221			__asm(
222			    LD64_ADD32(0, cc)
223			    MOP(sum, tmp, w)
224			);
225			w += 4;
226			mlen -= 8;
227		}
228		REDUCE(sum, tmp);
229		while ((mlen -= 2) >= 0)
230			sum += *w++;
231		if (byte_swapped) {
232			sum <<= 8;
233			byte_swapped = 0;
234			if (mlen == -1) {
235				su |= *(u_char *)w;
236				sum += su;
237				mlen = 0;
238			} else
239				mlen = -1;
240		} else if (mlen == -1) {
241			/*
242			 * This mbuf has odd number of bytes.
243			 * There could be a word split between
244			 * this mbuf and the next mbuf.
245			 * Save the last byte (to prepend to next mbuf).
246			 */
247			su = *(u_char *)w << 8;
248		}
249	}
250
251	if (len)
252		printf("%s: out of data by %d\n", __func__, len);
253	if (mlen == -1) {
254		/*
255		 * The last mbuf has odd # of bytes.  Follow the
256		 * standard (the odd byte is shifted left by 8 bits).
257		 */
258		sum += su & 0xff00;
259	}
260	REDUCE(sum, tmp);
261	return (~sum & 0xffff);
262}
263