diff options
Diffstat (limited to 'src/VBox/Devices/Network/slirp/bsd')
17 files changed, 6763 insertions, 0 deletions
diff --git a/src/VBox/Devices/Network/slirp/bsd/Makefile.kup b/src/VBox/Devices/Network/slirp/bsd/Makefile.kup new file mode 100644 index 00000000..e69de29b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/Makefile.kup diff --git a/src/VBox/Devices/Network/slirp/bsd/amd64/Makefile.kup b/src/VBox/Devices/Network/slirp/bsd/amd64/Makefile.kup new file mode 100644 index 00000000..e69de29b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/amd64/Makefile.kup diff --git a/src/VBox/Devices/Network/slirp/bsd/amd64/in_cksum.c b/src/VBox/Devices/Network/slirp/bsd/amd64/in_cksum.c new file mode 100644 index 00000000..351003e3 --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/amd64/in_cksum.c @@ -0,0 +1,242 @@ +/* $NetBSD: in_cksum.c,v 1.7 1997/09/02 13:18:15 thorpej Exp $ */ + +/*- + * Copyright (c) 1988, 1992, 1993 + * The Regents of the University of California. All rights reserved. + * Copyright (c) 1996 + * Matt Thomas <matt@3am-software.com> + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)in_cksum.c 8.1 (Berkeley) 6/10/93 + */ + +#ifndef VBOX +#include <sys/cdefs.h> /* RCS ID & Copyright macro defns */ +__FBSDID("$FreeBSD: src/sys/amd64/amd64/in_cksum.c,v 1.5.20.1 2009/04/15 03:14:26 kensmith Exp $"); + +#include <sys/param.h> +#include <sys/mbuf.h> +#include <sys/systm.h> +#include <netinet/in_systm.h> +#include <netinet/in.h> +#include <netinet/ip.h> +#include <machine/in_cksum.h> +#else +# include "in_cksum.h" +# include "slirp.h" +#endif + +/* + * Checksum routine for Internet Protocol family headers + * (Portable Alpha version). + * + * This routine is very heavily used in the network + * code and should be modified for each CPU to be as fast as possible. + */ + +#define ADDCARRY(x) (x > 65535 ? x -= 65535 : x) +#define REDUCE32 \ + { \ + q_util.q = sum; \ + sum = q_util.s[0] + q_util.s[1] + q_util.s[2] + q_util.s[3]; \ + } +#define REDUCE16 \ + { \ + q_util.q = sum; \ + l_util.l = q_util.s[0] + q_util.s[1] + q_util.s[2] + q_util.s[3]; \ + sum = l_util.s[0] + l_util.s[1]; \ + ADDCARRY(sum); \ + } + +static const u_int32_t in_masks[] = { + /*0 bytes*/ /*1 byte*/ /*2 bytes*/ /*3 bytes*/ + 0x00000000, 0x000000FF, 0x0000FFFF, 0x00FFFFFF, /* offset 0 */ + 0x00000000, 0x0000FF00, 0x00FFFF00, 0xFFFFFF00, /* offset 1 */ + 0x00000000, 0x00FF0000, 0xFFFF0000, 0xFFFF0000, /* offset 2 */ + 0x00000000, 0xFF000000, 0xFF000000, 0xFF000000, /* offset 3 */ +}; + +union l_util { + u_int16_t s[2]; + u_int32_t l; +}; +union q_util { + u_int16_t s[4]; + u_int32_t l[2]; + u_int64_t q; +}; + +static u_int64_t +in_cksumdata(const void *buf, int len) +{ + const u_int32_t *lw = (const u_int32_t *) buf; + u_int64_t sum = 0; + u_int64_t prefilled; + int offset; + union q_util q_util; + + if ((3 & (intptr_t) lw) == 0 && len == 20) { + sum = (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3] + lw[4]; + REDUCE32; + return sum; + } + + if ((offset = 3 & (intptr_t) lw) != 0) { + const u_int32_t *masks = in_masks + (offset << 2); + lw = (u_int32_t *) (((RTHCUINTPTR) lw) - offset); + sum = *lw++ & masks[len >= 3 ? 3 : len]; + len -= 4 - offset; + if (len <= 0) { + REDUCE32; + return sum; + } + } +#if 0 + /* + * Force to cache line boundary. + */ + offset = 32 - (0x1f & (long) lw); + if (offset < 32 && len > offset) { + len -= offset; + if (4 & offset) { + sum += (u_int64_t) lw[0]; + lw += 1; + } + if (8 & offset) { + sum += (u_int64_t) lw[0] + lw[1]; + lw += 2; + } + if (16 & offset) { + sum += (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3]; + lw += 4; + } + } +#endif + /* + * access prefilling to start load of next cache line. + * then add current cache line + * save result of prefilling for loop iteration. + */ + prefilled = lw[0]; + while ((len -= 32) >= 4) { + u_int64_t prefilling = lw[8]; + sum += prefilled + lw[1] + lw[2] + lw[3] + + lw[4] + lw[5] + lw[6] + lw[7]; + lw += 8; + prefilled = prefilling; + } + if (len >= 0) { + sum += prefilled + lw[1] + lw[2] + lw[3] + + lw[4] + lw[5] + lw[6] + lw[7]; + lw += 8; + } else { + len += 32; + } + while ((len -= 16) >= 0) { + sum += (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3]; + lw += 4; + } + len += 16; + while ((len -= 4) >= 0) { + sum += (u_int64_t) *lw++; + } + len += 4; + if (len > 0) + sum += (u_int64_t) (in_masks[len] & *lw); + REDUCE32; + return sum; +} + +u_short +in_addword(u_short a, u_short b) +{ + u_int64_t sum = a + b; + + ADDCARRY(sum); + return (sum); +} + +u_short +in_pseudo(u_int32_t a, u_int32_t b, u_int32_t c) +{ + u_int64_t sum; + union q_util q_util; + union l_util l_util; + + sum = (u_int64_t) a + b + c; + REDUCE16; + return (sum); +} + +u_short +in_cksum_skip(struct mbuf *m, int len, int skip) +{ + u_int64_t sum = 0; + int mlen = 0; + int clen = 0; + caddr_t addr; + union q_util q_util; + union l_util l_util; + + len -= skip; + for (; skip && m; m = m->m_next) { + if (m->m_len > skip) { + mlen = m->m_len - skip; + addr = mtod(m, caddr_t) + skip; + goto skip_start; + } else { + skip -= m->m_len; + } + } + + for (; m && len; m = m->m_next) { + if (m->m_len == 0) + continue; + mlen = m->m_len; + addr = mtod(m, caddr_t); +skip_start: + if (len < mlen) + mlen = len; + if ((clen ^ (intptr_t) addr) & 1) + sum += in_cksumdata(addr, mlen) << 8; + else + sum += in_cksumdata(addr, mlen); + + clen += mlen; + len -= mlen; + } + REDUCE16; + return (~sum & 0xffff); +} + +u_int in_cksum_hdr(const struct ip *ip) +{ + u_int64_t sum = in_cksumdata(ip, sizeof(struct ip)); + union q_util q_util; + union l_util l_util; + REDUCE16; + return (~sum & 0xffff); +} diff --git a/src/VBox/Devices/Network/slirp/bsd/amd64/include/in_cksum.h b/src/VBox/Devices/Network/slirp/bsd/amd64/include/in_cksum.h new file mode 100644 index 00000000..47a4565e --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/amd64/include/in_cksum.h @@ -0,0 +1,84 @@ +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from tahoe: in_cksum.c 1.2 86/01/05 + * from: @(#)in_cksum.c 1.3 (Berkeley) 1/19/91 + * from: Id: in_cksum.c,v 1.8 1995/12/03 18:35:19 bde Exp + * $FreeBSD: src/sys/amd64/include/in_cksum.h,v 1.5.20.1 2009/04/15 03:14:26 kensmith Exp $ + */ + +#ifndef _MACHINE_IN_CKSUM_H_ +#define _MACHINE_IN_CKSUM_H_ 1 + +#ifndef VBOX +#ifndef _SYS_CDEFS_H_ +#error this file needs sys/cdefs.h as a prerequisite +#endif + +#include <sys/cdefs.h> +#else +# include "slirp.h" +#endif + +#define in_cksum(m, len) in_cksum_skip(m, len, 0) + +/* + * It it useful to have an Internet checksum routine which is inlineable + * and optimized specifically for the task of computing IP header checksums + * in the normal case (where there are no options and the header length is + * therefore always exactly five 32-bit words. + */ +#ifdef __CC_SUPPORTS___INLINE + +static __inline void +in_cksum_update(struct ip *ip) +{ + int __tmpsum; + __tmpsum = (int)ntohs(ip->ip_sum) + 256; + ip->ip_sum = htons(__tmpsum + (__tmpsum >> 16)); +} + +#else + +#define in_cksum_update(ip) \ + do { \ + int __tmpsum; \ + __tmpsum = (int)ntohs(ip->ip_sum) + 256; \ + ip->ip_sum = htons(__tmpsum + (__tmpsum >> 16)); \ + } while(0) + +#endif + +#if defined(_KERNEL) || defined(VBOX) +u_int in_cksum_hdr(const struct ip *ip); +u_short in_addword(u_short sum, u_short b); +u_short in_pseudo(u_int sum, u_int b, u_int c); +u_short in_cksum_skip(struct mbuf *m, int len, int skip); +#endif + +#endif /* _MACHINE_IN_CKSUM_H_ */ diff --git a/src/VBox/Devices/Network/slirp/bsd/arm64/Makefile.kup b/src/VBox/Devices/Network/slirp/bsd/arm64/Makefile.kup new file mode 100644 index 00000000..e69de29b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/arm64/Makefile.kup diff --git a/src/VBox/Devices/Network/slirp/bsd/arm64/in_cksum.c b/src/VBox/Devices/Network/slirp/bsd/arm64/in_cksum.c new file mode 100644 index 00000000..351003e3 --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/arm64/in_cksum.c @@ -0,0 +1,242 @@ +/* $NetBSD: in_cksum.c,v 1.7 1997/09/02 13:18:15 thorpej Exp $ */ + +/*- + * Copyright (c) 1988, 1992, 1993 + * The Regents of the University of California. All rights reserved. + * Copyright (c) 1996 + * Matt Thomas <matt@3am-software.com> + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)in_cksum.c 8.1 (Berkeley) 6/10/93 + */ + +#ifndef VBOX +#include <sys/cdefs.h> /* RCS ID & Copyright macro defns */ +__FBSDID("$FreeBSD: src/sys/amd64/amd64/in_cksum.c,v 1.5.20.1 2009/04/15 03:14:26 kensmith Exp $"); + +#include <sys/param.h> +#include <sys/mbuf.h> +#include <sys/systm.h> +#include <netinet/in_systm.h> +#include <netinet/in.h> +#include <netinet/ip.h> +#include <machine/in_cksum.h> +#else +# include "in_cksum.h" +# include "slirp.h" +#endif + +/* + * Checksum routine for Internet Protocol family headers + * (Portable Alpha version). + * + * This routine is very heavily used in the network + * code and should be modified for each CPU to be as fast as possible. + */ + +#define ADDCARRY(x) (x > 65535 ? x -= 65535 : x) +#define REDUCE32 \ + { \ + q_util.q = sum; \ + sum = q_util.s[0] + q_util.s[1] + q_util.s[2] + q_util.s[3]; \ + } +#define REDUCE16 \ + { \ + q_util.q = sum; \ + l_util.l = q_util.s[0] + q_util.s[1] + q_util.s[2] + q_util.s[3]; \ + sum = l_util.s[0] + l_util.s[1]; \ + ADDCARRY(sum); \ + } + +static const u_int32_t in_masks[] = { + /*0 bytes*/ /*1 byte*/ /*2 bytes*/ /*3 bytes*/ + 0x00000000, 0x000000FF, 0x0000FFFF, 0x00FFFFFF, /* offset 0 */ + 0x00000000, 0x0000FF00, 0x00FFFF00, 0xFFFFFF00, /* offset 1 */ + 0x00000000, 0x00FF0000, 0xFFFF0000, 0xFFFF0000, /* offset 2 */ + 0x00000000, 0xFF000000, 0xFF000000, 0xFF000000, /* offset 3 */ +}; + +union l_util { + u_int16_t s[2]; + u_int32_t l; +}; +union q_util { + u_int16_t s[4]; + u_int32_t l[2]; + u_int64_t q; +}; + +static u_int64_t +in_cksumdata(const void *buf, int len) +{ + const u_int32_t *lw = (const u_int32_t *) buf; + u_int64_t sum = 0; + u_int64_t prefilled; + int offset; + union q_util q_util; + + if ((3 & (intptr_t) lw) == 0 && len == 20) { + sum = (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3] + lw[4]; + REDUCE32; + return sum; + } + + if ((offset = 3 & (intptr_t) lw) != 0) { + const u_int32_t *masks = in_masks + (offset << 2); + lw = (u_int32_t *) (((RTHCUINTPTR) lw) - offset); + sum = *lw++ & masks[len >= 3 ? 3 : len]; + len -= 4 - offset; + if (len <= 0) { + REDUCE32; + return sum; + } + } +#if 0 + /* + * Force to cache line boundary. + */ + offset = 32 - (0x1f & (long) lw); + if (offset < 32 && len > offset) { + len -= offset; + if (4 & offset) { + sum += (u_int64_t) lw[0]; + lw += 1; + } + if (8 & offset) { + sum += (u_int64_t) lw[0] + lw[1]; + lw += 2; + } + if (16 & offset) { + sum += (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3]; + lw += 4; + } + } +#endif + /* + * access prefilling to start load of next cache line. + * then add current cache line + * save result of prefilling for loop iteration. + */ + prefilled = lw[0]; + while ((len -= 32) >= 4) { + u_int64_t prefilling = lw[8]; + sum += prefilled + lw[1] + lw[2] + lw[3] + + lw[4] + lw[5] + lw[6] + lw[7]; + lw += 8; + prefilled = prefilling; + } + if (len >= 0) { + sum += prefilled + lw[1] + lw[2] + lw[3] + + lw[4] + lw[5] + lw[6] + lw[7]; + lw += 8; + } else { + len += 32; + } + while ((len -= 16) >= 0) { + sum += (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3]; + lw += 4; + } + len += 16; + while ((len -= 4) >= 0) { + sum += (u_int64_t) *lw++; + } + len += 4; + if (len > 0) + sum += (u_int64_t) (in_masks[len] & *lw); + REDUCE32; + return sum; +} + +u_short +in_addword(u_short a, u_short b) +{ + u_int64_t sum = a + b; + + ADDCARRY(sum); + return (sum); +} + +u_short +in_pseudo(u_int32_t a, u_int32_t b, u_int32_t c) +{ + u_int64_t sum; + union q_util q_util; + union l_util l_util; + + sum = (u_int64_t) a + b + c; + REDUCE16; + return (sum); +} + +u_short +in_cksum_skip(struct mbuf *m, int len, int skip) +{ + u_int64_t sum = 0; + int mlen = 0; + int clen = 0; + caddr_t addr; + union q_util q_util; + union l_util l_util; + + len -= skip; + for (; skip && m; m = m->m_next) { + if (m->m_len > skip) { + mlen = m->m_len - skip; + addr = mtod(m, caddr_t) + skip; + goto skip_start; + } else { + skip -= m->m_len; + } + } + + for (; m && len; m = m->m_next) { + if (m->m_len == 0) + continue; + mlen = m->m_len; + addr = mtod(m, caddr_t); +skip_start: + if (len < mlen) + mlen = len; + if ((clen ^ (intptr_t) addr) & 1) + sum += in_cksumdata(addr, mlen) << 8; + else + sum += in_cksumdata(addr, mlen); + + clen += mlen; + len -= mlen; + } + REDUCE16; + return (~sum & 0xffff); +} + +u_int in_cksum_hdr(const struct ip *ip) +{ + u_int64_t sum = in_cksumdata(ip, sizeof(struct ip)); + union q_util q_util; + union l_util l_util; + REDUCE16; + return (~sum & 0xffff); +} diff --git a/src/VBox/Devices/Network/slirp/bsd/arm64/include/in_cksum.h b/src/VBox/Devices/Network/slirp/bsd/arm64/include/in_cksum.h new file mode 100644 index 00000000..47a4565e --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/arm64/include/in_cksum.h @@ -0,0 +1,84 @@ +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from tahoe: in_cksum.c 1.2 86/01/05 + * from: @(#)in_cksum.c 1.3 (Berkeley) 1/19/91 + * from: Id: in_cksum.c,v 1.8 1995/12/03 18:35:19 bde Exp + * $FreeBSD: src/sys/amd64/include/in_cksum.h,v 1.5.20.1 2009/04/15 03:14:26 kensmith Exp $ + */ + +#ifndef _MACHINE_IN_CKSUM_H_ +#define _MACHINE_IN_CKSUM_H_ 1 + +#ifndef VBOX +#ifndef _SYS_CDEFS_H_ +#error this file needs sys/cdefs.h as a prerequisite +#endif + +#include <sys/cdefs.h> +#else +# include "slirp.h" +#endif + +#define in_cksum(m, len) in_cksum_skip(m, len, 0) + +/* + * It it useful to have an Internet checksum routine which is inlineable + * and optimized specifically for the task of computing IP header checksums + * in the normal case (where there are no options and the header length is + * therefore always exactly five 32-bit words. + */ +#ifdef __CC_SUPPORTS___INLINE + +static __inline void +in_cksum_update(struct ip *ip) +{ + int __tmpsum; + __tmpsum = (int)ntohs(ip->ip_sum) + 256; + ip->ip_sum = htons(__tmpsum + (__tmpsum >> 16)); +} + +#else + +#define in_cksum_update(ip) \ + do { \ + int __tmpsum; \ + __tmpsum = (int)ntohs(ip->ip_sum) + 256; \ + ip->ip_sum = htons(__tmpsum + (__tmpsum >> 16)); \ + } while(0) + +#endif + +#if defined(_KERNEL) || defined(VBOX) +u_int in_cksum_hdr(const struct ip *ip); +u_short in_addword(u_short sum, u_short b); +u_short in_pseudo(u_int sum, u_int b, u_int c); +u_short in_cksum_skip(struct mbuf *m, int len, int skip); +#endif + +#endif /* _MACHINE_IN_CKSUM_H_ */ diff --git a/src/VBox/Devices/Network/slirp/bsd/i386/Makefile.kup b/src/VBox/Devices/Network/slirp/bsd/i386/Makefile.kup new file mode 100644 index 00000000..e69de29b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/i386/Makefile.kup diff --git a/src/VBox/Devices/Network/slirp/bsd/i386/in_cksum.c b/src/VBox/Devices/Network/slirp/bsd/i386/in_cksum.c new file mode 100644 index 00000000..2be70636 --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/i386/in_cksum.c @@ -0,0 +1,499 @@ +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from tahoe: in_cksum.c 1.2 86/01/05 + * from: @(#)in_cksum.c 1.3 (Berkeley) 1/19/91 + */ + +#ifndef VBOX +#include <sys/cdefs.h> +__FBSDID("$FreeBSD: src/sys/i386/i386/in_cksum.c,v 1.28.10.1.6.1 2009/04/15 03:14:26 kensmith Exp $"); + +/* + * MPsafe: alfred + */ +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/mbuf.h> + +#include <netinet/in.h> +#include <netinet/in_systm.h> +#include <netinet/ip.h> + +#include <machine/in_cksum.h> +#else +# include "in_cksum.h" +# include "slirp.h" +#endif + +/* + * Checksum routine for Internet Protocol family headers. + * + * This routine is very heavily used in the network + * code and should be modified for each CPU to be as fast as possible. + * + * This implementation is 386 version. + */ + +#undef ADDCARRY +#define ADDCARRY(x) if ((x) > 0xffff) (x) -= 0xffff +/* + * icc needs to be special cased here, as the asm code below results + * in broken code if compiled with icc. + */ +#if !defined(__GNUCLIKE_ASM) || defined(__INTEL_COMPILER) +/* non gcc parts stolen from sys/alpha/alpha/in_cksum.c */ +#define REDUCE32 \ + { \ + q_util.q = sum; \ + sum = q_util.s[0] + q_util.s[1] + q_util.s[2] + q_util.s[3]; \ + } +#define REDUCE16 \ + { \ + q_util.q = sum; \ + l_util.l = q_util.s[0] + q_util.s[1] + q_util.s[2] + q_util.s[3]; \ + sum = l_util.s[0] + l_util.s[1]; \ + ADDCARRY(sum); \ + } +#endif +#define REDUCE {sum = (sum & 0xffff) + (sum >> 16); ADDCARRY(sum);} + +#if !defined(__GNUCLIKE_ASM) || defined(__INTEL_COMPILER) +static const u_int32_t in_masks[] = { + /*0 bytes*/ /*1 byte*/ /*2 bytes*/ /*3 bytes*/ + 0x00000000, 0x000000FF, 0x0000FFFF, 0x00FFFFFF, /* offset 0 */ + 0x00000000, 0x0000FF00, 0x00FFFF00, 0xFFFFFF00, /* offset 1 */ + 0x00000000, 0x00FF0000, 0xFFFF0000, 0xFFFF0000, /* offset 2 */ + 0x00000000, 0xFF000000, 0xFF000000, 0xFF000000, /* offset 3 */ +}; + +union l_util { + u_int16_t s[2]; + u_int32_t l; +}; +union q_util { + u_int16_t s[4]; + u_int32_t l[2]; + u_int64_t q; +}; + +static u_int64_t +in_cksumdata(const u_int32_t *lw, int len) +{ + u_int64_t sum = 0; + u_int64_t prefilled; + int offset; + union q_util q_util; + + if ((3 & (long) lw) == 0 && len == 20) { + sum = (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3] + lw[4]; + REDUCE32; + return sum; + } + + if ((offset = 3 & (long) lw) != 0) { + const u_int32_t *masks = in_masks + (offset << 2); + lw = (u_int32_t *) (((RTHCUINTPTR) lw) - offset); + sum = *lw++ & masks[len >= 3 ? 3 : len]; + len -= 4 - offset; + if (len <= 0) { + REDUCE32; + return sum; + } + } +#if 0 + /* + * Force to cache line boundary. + */ + offset = 32 - (0x1f & (long) lw); + if (offset < 32 && len > offset) { + len -= offset; + if (4 & offset) { + sum += (u_int64_t) lw[0]; + lw += 1; + } + if (8 & offset) { + sum += (u_int64_t) lw[0] + lw[1]; + lw += 2; + } + if (16 & offset) { + sum += (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3]; + lw += 4; + } + } +#endif + /* + * access prefilling to start load of next cache line. + * then add current cache line + * save result of prefilling for loop iteration. + */ + prefilled = lw[0]; + while ((len -= 32) >= 4) { + u_int64_t prefilling = lw[8]; + sum += prefilled + lw[1] + lw[2] + lw[3] + + lw[4] + lw[5] + lw[6] + lw[7]; + lw += 8; + prefilled = prefilling; + } + if (len >= 0) { + sum += prefilled + lw[1] + lw[2] + lw[3] + + lw[4] + lw[5] + lw[6] + lw[7]; + lw += 8; + } else { + len += 32; + } + while ((len -= 16) >= 0) { + sum += (u_int64_t) lw[0] + lw[1] + lw[2] + lw[3]; + lw += 4; + } + len += 16; + while ((len -= 4) >= 0) { + sum += (u_int64_t) *lw++; + } + len += 4; + if (len > 0) + sum += (u_int64_t) (in_masks[len] & *lw); + REDUCE32; + return sum; +} + +u_short +in_addword(u_short a, u_short b) +{ + u_int64_t sum = a + b; + + ADDCARRY(sum); + return (sum); +} + +u_short +in_pseudo(u_int32_t a, u_int32_t b, u_int32_t c) +{ + u_int64_t sum; + union q_util q_util; + union l_util l_util; + + sum = (u_int64_t) a + b + c; + REDUCE16; + return (sum); +} + +u_short +in_cksum_skip(struct mbuf *m, int len, int skip) +{ + u_int64_t sum = 0; + int mlen = 0; + int clen = 0; + caddr_t addr; + union q_util q_util; + union l_util l_util; + + len -= skip; + for (; skip && m; m = m->m_next) { + if (m->m_len > skip) { + mlen = m->m_len - skip; + addr = mtod(m, caddr_t) + skip; + goto skip_start; + } else { + skip -= m->m_len; + } + } + + for (; m && len; m = m->m_next) { + if (m->m_len == 0) + continue; + mlen = m->m_len; + addr = mtod(m, caddr_t); +skip_start: + if (len < mlen) + mlen = len; + if ((clen ^ (long) addr) & 1) + sum += in_cksumdata((const u_int32_t *)addr, mlen) << 8; + else + sum += in_cksumdata((const u_int32_t *)addr, mlen); + + clen += mlen; + len -= mlen; + } + REDUCE16; + return (~sum & 0xffff); +} + +u_int in_cksum_hdr(const struct ip *ip) +{ + u_int64_t sum = in_cksumdata((const u_int32_t *)ip, sizeof(struct ip)); + union q_util q_util; + union l_util l_util; + + REDUCE16; + return (~sum & 0xffff); +} +#else + +/* + * These asm statements require __volatile because they pass information + * via the condition codes. GCC does not currently provide a way to specify + * the condition codes as an input or output operand. + * + * The LOAD macro below is effectively a prefetch into cache. GCC will + * load the value into a register but will not use it. Since modern CPUs + * reorder operations, this will generally take place in parallel with + * other calculations. + */ +u_short +in_cksum_skip(m, len, skip) + struct mbuf *m; + int len; + int skip; +{ + register u_short *w; + register unsigned sum = 0; + register int mlen = 0; + int byte_swapped = 0; + union { char c[2]; u_short s; } su; + + len -= skip; + for (; skip && m; m = m->m_next) { + if (m->m_len > skip) { + mlen = m->m_len - skip; + w = (u_short *)(mtod(m, u_char *) + skip); + goto skip_start; + } else { + skip -= m->m_len; + } + } + + for (;m && len; m = m->m_next) { + if (m->m_len == 0) + continue; + w = mtod(m, u_short *); + if (mlen == -1) { + /* + * The first byte of this mbuf is the continuation + * of a word spanning between this mbuf and the + * last mbuf. + */ + + /* su.c[0] is already saved when scanning previous + * mbuf. sum was REDUCEd when we found mlen == -1 + */ + su.c[1] = *(u_char *)w; + sum += su.s; + w = (u_short *)((char *)w + 1); + mlen = m->m_len - 1; + len--; + } else + mlen = m->m_len; +skip_start: + if (len < mlen) + mlen = len; + len -= mlen; + /* + * Force to long boundary so we do longword aligned + * memory operations + */ + if (3 & (int) w) { + REDUCE; + if ((1 & (int) w) && (mlen > 0)) { + sum <<= 8; + su.c[0] = *(char *)w; + w = (u_short *)((char *)w + 1); + mlen--; + byte_swapped = 1; + } + if ((2 & (int) w) && (mlen >= 2)) { + sum += *w++; + mlen -= 2; + } + } + /* + * Advance to a 486 cache line boundary. + */ + if (4 & (int) w && mlen >= 4) { + __asm __volatile ( + "addl %1, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)w)[0]) + ); + w += 2; + mlen -= 4; + } + if (8 & (int) w && mlen >= 8) { + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)w)[0]), + "g" (((const u_int32_t *)w)[1]) + ); + w += 4; + mlen -= 8; + } + /* + * Do as much of the checksum as possible 32 bits at at time. + * In fact, this loop is unrolled to make overhead from + * branches &c small. + */ + mlen -= 1; + while ((mlen -= 32) >= 0) { + /* + * Add with carry 16 words and fold in the last + * carry by adding a 0 with carry. + * + * The early ADD(16) and the LOAD(32) are to load + * the next 2 cache lines in advance on 486's. The + * 486 has a penalty of 2 clock cycles for loading + * a cache line, plus whatever time the external + * memory takes to load the first word(s) addressed. + * These penalties are unavoidable. Subsequent + * accesses to a cache line being loaded (and to + * other external memory?) are delayed until the + * whole load finishes. These penalties are mostly + * avoided by not accessing external memory for + * 8 cycles after the ADD(16) and 12 cycles after + * the LOAD(32). The loop terminates when mlen + * is initially 33 (not 32) to guaranteed that + * the LOAD(32) is within bounds. + */ + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl %3, %0\n" + "adcl %4, %0\n" + "adcl %5, %0\n" + "mov %6, %%eax\n" + "adcl %7, %0\n" + "adcl %8, %0\n" + "adcl %9, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)w)[4]), + "g" (((const u_int32_t *)w)[0]), + "g" (((const u_int32_t *)w)[1]), + "g" (((const u_int32_t *)w)[2]), + "g" (((const u_int32_t *)w)[3]), + "g" (((const u_int32_t *)w)[8]), + "g" (((const u_int32_t *)w)[5]), + "g" (((const u_int32_t *)w)[6]), + "g" (((const u_int32_t *)w)[7]) + : "eax" + ); + w += 16; + } + mlen += 32 + 1; + if (mlen >= 32) { + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl %3, %0\n" + "adcl %4, %0\n" + "adcl %5, %0\n" + "adcl %6, %0\n" + "adcl %7, %0\n" + "adcl %8, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)w)[4]), + "g" (((const u_int32_t *)w)[0]), + "g" (((const u_int32_t *)w)[1]), + "g" (((const u_int32_t *)w)[2]), + "g" (((const u_int32_t *)w)[3]), + "g" (((const u_int32_t *)w)[5]), + "g" (((const u_int32_t *)w)[6]), + "g" (((const u_int32_t *)w)[7]) + ); + w += 16; + mlen -= 32; + } + if (mlen >= 16) { + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl %3, %0\n" + "adcl %4, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)w)[0]), + "g" (((const u_int32_t *)w)[1]), + "g" (((const u_int32_t *)w)[2]), + "g" (((const u_int32_t *)w)[3]) + ); + w += 8; + mlen -= 16; + } + if (mlen >= 8) { + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)w)[0]), + "g" (((const u_int32_t *)w)[1]) + ); + w += 4; + mlen -= 8; + } + if (mlen == 0 && byte_swapped == 0) + continue; /* worth 1% maybe ?? */ + REDUCE; + while ((mlen -= 2) >= 0) { + sum += *w++; + } + if (byte_swapped) { + sum <<= 8; + byte_swapped = 0; + if (mlen == -1) { + su.c[1] = *(char *)w; + sum += su.s; + mlen = 0; + } else + mlen = -1; + } else if (mlen == -1) + /* + * This mbuf has odd number of bytes. + * There could be a word split betwen + * this mbuf and the next mbuf. + * Save the last byte (to prepend to next mbuf). + */ + su.c[0] = *(char *)w; + } + + if (len) + printf("%s: out of data by %d\n", __func__, len); + if (mlen == -1) { + /* The last mbuf has odd # of bytes. Follow the + standard (the odd byte is shifted left by 8 bits) */ + su.c[1] = 0; + sum += su.s; + } + REDUCE; + return (~sum & 0xffff); +} +#endif diff --git a/src/VBox/Devices/Network/slirp/bsd/i386/include/in_cksum.h b/src/VBox/Devices/Network/slirp/bsd/i386/include/in_cksum.h new file mode 100644 index 00000000..0d20d193 --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/i386/include/in_cksum.h @@ -0,0 +1,145 @@ +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from tahoe: in_cksum.c 1.2 86/01/05 + * from: @(#)in_cksum.c 1.3 (Berkeley) 1/19/91 + * from: Id: in_cksum.c,v 1.8 1995/12/03 18:35:19 bde Exp + * $FreeBSD: src/sys/i386/include/in_cksum.h,v 1.17.10.1.6.1 2009/04/15 03:14:26 kensmith Exp $ + */ + +#ifndef _MACHINE_IN_CKSUM_H_ +#define _MACHINE_IN_CKSUM_H_ 1 + +#ifndef VBOX +#ifndef _SYS_CDEFS_H_ +#error this file needs sys/cdefs.h as a prerequisite +#endif + +/* + * MP safe (alfred) + */ + +#include <sys/cdefs.h> +#else +# include "slirp.h" +#endif + +#define in_cksum(m, len) in_cksum_skip(m, len, 0) + +/* + * It it useful to have an Internet checksum routine which is inlineable + * and optimized specifically for the task of computing IP header checksums + * in the normal case (where there are no options and the header length is + * therefore always exactly five 32-bit words. + */ +#if defined(__GNUCLIKE_ASM) && !defined(__INTEL_COMPILER) +static __inline u_int +in_cksum_hdr(const struct ip *ip) +{ + register u_int sum = 0; + + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl %3, %0\n" + "adcl %4, %0\n" + "adcl %5, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (((const u_int32_t *)ip)[0]), + "g" (((const u_int32_t *)ip)[1]), + "g" (((const u_int32_t *)ip)[2]), + "g" (((const u_int32_t *)ip)[3]), + "g" (((const u_int32_t *)ip)[4]) + ); + sum = (sum & 0xffff) + (sum >> 16); + if (sum > 0xffff) + sum -= 0xffff; + + return ~sum & 0xffff; +} + +static __inline void +in_cksum_update(struct ip *ip) +{ + int __tmpsum; + __tmpsum = (int)ntohs(ip->ip_sum) + 256; + ip->ip_sum = htons(__tmpsum + (__tmpsum >> 16)); +} + +static __inline u_short +in_addword(u_short sum, u_short b) +{ + /* __volatile is necessary because the condition codes are used. */ + __asm __volatile ( + "addw %1, %0\n" + "adcw $0, %0" + : "+r" (sum) + : "r" (b) + ); + return (sum); +} + +static __inline u_short +in_pseudo(u_int sum, u_int b, u_int c) +{ + /* __volatile is necessary because the condition codes are used. */ + __asm __volatile ( + "addl %1, %0\n" + "adcl %2, %0\n" + "adcl $0, %0" + : "+r" (sum) + : "g" (b), + "g" (c) + ); + sum = (sum & 0xffff) + (sum >> 16); + if (sum > 0xffff) + sum -= 0xffff; + return (sum); +} + +#else +#define in_cksum_update(ip) \ + do { \ + int __tmpsum; \ + __tmpsum = (int)ntohs(ip->ip_sum) + 256; \ + ip->ip_sum = htons(__tmpsum + (__tmpsum >> 16)); \ + } while(0) + +#endif + +#if defined(_KERNEL) || defined(VBOX) +#if !defined(__GNUCLIKE_ASM) || defined(__INTEL_COMPILER) +u_int in_cksum_hdr(const struct ip *ip); +u_short in_addword(u_short sum, u_short b); +u_short in_pseudo(u_int sum, u_int b, u_int c); +#endif +u_short in_cksum_skip(struct mbuf *m, int len, int skip); +#endif /* _KERNEL */ + +#endif /* _MACHINE_IN_CKSUM_H_ */ diff --git a/src/VBox/Devices/Network/slirp/bsd/kern/Makefile.kup b/src/VBox/Devices/Network/slirp/bsd/kern/Makefile.kup new file mode 100644 index 00000000..e69de29b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/kern/Makefile.kup diff --git a/src/VBox/Devices/Network/slirp/bsd/kern/kern_mbuf.c b/src/VBox/Devices/Network/slirp/bsd/kern/kern_mbuf.c new file mode 100644 index 00000000..c908e2fd --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/kern/kern_mbuf.c @@ -0,0 +1,824 @@ +/*- + * Copyright (c) 2004, 2005, + * Bosko Milekic <bmilekic@FreeBSD.org>. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice unmodified, this list of conditions and the following + * disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#ifndef VBOX +#include <sys/cdefs.h> +__FBSDID("$FreeBSD: src/sys/kern/kern_mbuf.c,v 1.32.2.5.2.1 2009/04/15 03:14:26 kensmith Exp $"); + +#include "opt_mac.h" +#include "opt_param.h" + +#include <sys/param.h> +#include <sys/malloc.h> +#include <sys/systm.h> +#include <sys/mbuf.h> +#include <sys/domain.h> +#include <sys/eventhandler.h> +#include <sys/kernel.h> +#include <sys/protosw.h> +#include <sys/smp.h> +#include <sys/sysctl.h> + +#include <security/mac/mac_framework.h> + +#include <vm/vm.h> +#include <vm/vm_page.h> +#include <vm/uma.h> +#include <vm/uma_int.h> +#include <vm/uma_dbg.h> +#else +# include <iprt/param.h> +# include <slirp.h> +# define IN_BSD +# include "ext.h" +#endif + +/* + * In FreeBSD, Mbufs and Mbuf Clusters are allocated from UMA + * Zones. + * + * Mbuf Clusters (2K, contiguous) are allocated from the Cluster + * Zone. The Zone can be capped at kern.ipc.nmbclusters, if the + * administrator so desires. + * + * Mbufs are allocated from a UMA Master Zone called the Mbuf + * Zone. + * + * Additionally, FreeBSD provides a Packet Zone, which it + * configures as a Secondary Zone to the Mbuf Master Zone, + * thus sharing backend Slab kegs with the Mbuf Master Zone. + * + * Thus common-case allocations and locking are simplified: + * + * m_clget() m_getcl() + * | | + * | .------------>[(Packet Cache)] m_get(), m_gethdr() + * | | [ Packet ] | + * [(Cluster Cache)] [ Secondary ] [ (Mbuf Cache) ] + * [ Cluster Zone ] [ Zone ] [ Mbuf Master Zone ] + * | \________ | + * [ Cluster Keg ] \ / + * | [ Mbuf Keg ] + * [ Cluster Slabs ] | + * | [ Mbuf Slabs ] + * \____________(VM)_________________/ + * + * + * Whenever an object is allocated with uma_zalloc() out of + * one of the Zones its _ctor_ function is executed. The same + * for any deallocation through uma_zfree() the _dtor_ function + * is executed. + * + * Caches are per-CPU and are filled from the Master Zone. + * + * Whenever an object is allocated from the underlying global + * memory pool it gets pre-initialized with the _zinit_ functions. + * When the Keg's are overfull objects get decomissioned with + * _zfini_ functions and free'd back to the global memory pool. + * + */ + +#ifndef VBOX +int nmbclusters; /* limits number of mbuf clusters */ +int nmbjumbop; /* limits number of page size jumbo clusters */ +int nmbjumbo9; /* limits number of 9k jumbo clusters */ +int nmbjumbo16; /* limits number of 16k jumbo clusters */ +struct mbstat mbstat; +#endif + +/* + * tunable_mbinit() has to be run before init_maxsockets() thus + * the SYSINIT order below is SI_ORDER_MIDDLE while init_maxsockets() + * runs at SI_ORDER_ANY. + */ +static void +tunable_mbinit(void *dummy) +{ +#ifdef VBOX + PNATState pData = (PNATState)dummy; +#endif + /* This has to be done before VM init. */ + nmbclusters = 1024 + maxusers * 64; + nmbjumbop = nmbclusters / 2; + nmbjumbo9 = nmbjumbop / 2; + nmbjumbo16 = nmbjumbo9 / 2; + TUNABLE_INT_FETCH("kern.ipc.nmbclusters", &nmbclusters); +} +SYSINIT(tunable_mbinit, SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_mbinit, NULL); + +#ifndef VBOX +/* XXX: These should be tuneables. Can't change UMA limits on the fly. */ +static int +sysctl_nmbclusters(SYSCTL_HANDLER_ARGS) +{ + int error, newnmbclusters; + + newnmbclusters = nmbclusters; + error = sysctl_handle_int(oidp, &newnmbclusters, 0, req); + if (error == 0 && req->newptr) { + if (newnmbclusters > nmbclusters) { + nmbclusters = newnmbclusters; + uma_zone_set_max(zone_clust, nmbclusters); + EVENTHANDLER_INVOKE(nmbclusters_change); + } else + error = EINVAL; + } + return (error); +} +SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbclusters, CTLTYPE_INT|CTLFLAG_RW, +&nmbclusters, 0, sysctl_nmbclusters, "IU", + "Maximum number of mbuf clusters allowed"); + +static int +sysctl_nmbjumbop(SYSCTL_HANDLER_ARGS) +{ + int error, newnmbjumbop; + + newnmbjumbop = nmbjumbop; + error = sysctl_handle_int(oidp, &newnmbjumbop, 0, req); + if (error == 0 && req->newptr) { + if (newnmbjumbop> nmbjumbop) { + nmbjumbop = newnmbjumbop; + uma_zone_set_max(zone_jumbop, nmbjumbop); + } else + error = EINVAL; + } + return (error); +} +SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbjumbop, CTLTYPE_INT|CTLFLAG_RW, +&nmbjumbop, 0, sysctl_nmbjumbop, "IU", + "Maximum number of mbuf page size jumbo clusters allowed"); + + +static int +sysctl_nmbjumbo9(SYSCTL_HANDLER_ARGS) +{ + int error, newnmbjumbo9; + + newnmbjumbo9 = nmbjumbo9; + error = sysctl_handle_int(oidp, &newnmbjumbo9, 0, req); + if (error == 0 && req->newptr) { + if (newnmbjumbo9> nmbjumbo9) { + nmbjumbo9 = newnmbjumbo9; + uma_zone_set_max(zone_jumbo9, nmbjumbo9); + } else + error = EINVAL; + } + return (error); +} +SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbjumbo9, CTLTYPE_INT|CTLFLAG_RW, +&nmbjumbo9, 0, sysctl_nmbjumbo9, "IU", + "Maximum number of mbuf 9k jumbo clusters allowed"); + +static int +sysctl_nmbjumbo16(SYSCTL_HANDLER_ARGS) +{ + int error, newnmbjumbo16; + + newnmbjumbo16 = nmbjumbo16; + error = sysctl_handle_int(oidp, &newnmbjumbo16, 0, req); + if (error == 0 && req->newptr) { + if (newnmbjumbo16> nmbjumbo16) { + nmbjumbo16 = newnmbjumbo16; + uma_zone_set_max(zone_jumbo16, nmbjumbo16); + } else + error = EINVAL; + } + return (error); +} +SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbjumbo16, CTLTYPE_INT|CTLFLAG_RW, +&nmbjumbo16, 0, sysctl_nmbjumbo16, "IU", + "Maximum number of mbuf 16k jumbo clusters allowed"); + + + +SYSCTL_STRUCT(_kern_ipc, OID_AUTO, mbstat, CTLFLAG_RD, &mbstat, mbstat, + "Mbuf general information and statistics"); + +/* + * Zones from which we allocate. + */ +uma_zone_t zone_mbuf; +uma_zone_t zone_clust; +uma_zone_t zone_pack; +uma_zone_t zone_jumbop; +uma_zone_t zone_jumbo9; +uma_zone_t zone_jumbo16; +uma_zone_t zone_ext_refcnt; + +/* + * Local prototypes. + */ +static int mb_ctor_mbuf(void *, int, void *, int); +static int mb_ctor_clust(void *, int, void *, int); +static int mb_ctor_pack(void *, int, void *, int); +static void mb_dtor_mbuf(void *, int, void *); +static void mb_dtor_clust(void *, int, void *); +static void mb_dtor_pack(void *, int, void *); +static int mb_zinit_pack(void *, int, int); +static void mb_zfini_pack(void *, int); +#else +/* + * Local prototypes. + */ +static int mb_ctor_mbuf(PNATState, void *, int, void *, int); +static int mb_ctor_clust(PNATState, void *, int, void *, int); +static int mb_ctor_pack(PNATState, void *, int, void *, int); +static void mb_dtor_mbuf(PNATState, void *, int, void *); +static void mb_dtor_clust(PNATState, void *, int, void *); +static void mb_dtor_pack(PNATState, void *, int, void *); +static int mb_zinit_pack(PNATState, void *, int, int); +static void mb_zfini_pack(PNATState, void *, int); +#endif + +/*static void mb_reclaim(void *); - unused */ +#ifndef VBOX +static void mbuf_init(void *); +static void *mbuf_jumbo_alloc(uma_zone_t, int, u_int8_t *, int); +static void mbuf_jumbo_free(void *, int, u_int8_t); +#endif + +#ifndef VBOX +static MALLOC_DEFINE(M_JUMBOFRAME, "jumboframes", "mbuf jumbo frame buffers"); + +/* Ensure that MSIZE doesn't break dtom() - it must be a power of 2 */ +CTASSERT((((MSIZE - 1) ^ MSIZE) + 1) >> 1 == MSIZE); +#else +#define uma_zcreate(a0, a1, a2, a3, a4, a5, a6, a7) \ + uma_zcreate(pData, a0, a1, a2, a3, a4, a5, a6, a7) +#endif + +/* + * Initialize FreeBSD Network buffer allocation. + */ +SYSINIT(mbuf, SI_SUB_MBUF, SI_ORDER_FIRST, mbuf_init, NULL); +#ifndef VBOX +static void +#else +void +#endif +mbuf_init(void *dummy) +{ + + /* + * Configure UMA zones for Mbufs, Clusters, and Packets. + */ +#ifndef VBOX + zone_mbuf = uma_zcreate(MBUF_MEM_NAME, MSIZE, + mb_ctor_mbuf, mb_dtor_mbuf, +#ifdef INVARIANTS + trash_init, trash_fini, +#else + NULL, NULL, +#endif + MSIZE - 1, UMA_ZONE_MAXBUCKET); + + zone_clust = uma_zcreate(MBUF_CLUSTER_MEM_NAME, MCLBYTES, + mb_ctor_clust, mb_dtor_clust, +#ifdef INVARIANTS + trash_init, trash_fini, +#else + NULL, NULL, +#endif + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); +#else /*!VBOX*/ + PNATState pData = (PNATState)dummy; + tunable_mbinit(pData); + zone_mbuf = uma_zcreate(MBUF_MEM_NAME, MSIZE, + mb_ctor_mbuf, mb_dtor_mbuf, + NULL, NULL, + MSIZE - 1, UMA_ZONE_MAXBUCKET); + if (nmbclusters > 0) + uma_zone_set_max(zone_mbuf, nmbclusters); + + zone_clust = uma_zcreate(MBUF_CLUSTER_MEM_NAME, MCLBYTES, + mb_ctor_clust, mb_dtor_clust, + NULL, NULL, + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); +#endif /*VBOX*/ + if (nmbclusters > 0) + uma_zone_set_max(zone_clust, nmbclusters); + + zone_pack = uma_zsecond_create(MBUF_PACKET_MEM_NAME, mb_ctor_pack, + mb_dtor_pack, mb_zinit_pack, mb_zfini_pack, zone_mbuf); + + /* Make jumbo frame zone too. Page size, 9k and 16k. */ +#ifndef VBOX + zone_jumbop = uma_zcreate(MBUF_JUMBOP_MEM_NAME, MJUMPAGESIZE, + mb_ctor_clust, mb_dtor_clust, +#ifdef INVARIANTS + trash_init, trash_fini, +#else + NULL, NULL, +#endif + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); + if (nmbjumbop > 0) + uma_zone_set_max(zone_jumbop, nmbjumbop); + + zone_jumbo9 = uma_zcreate(MBUF_JUMBO9_MEM_NAME, MJUM9BYTES, + mb_ctor_clust, mb_dtor_clust, +#ifdef INVARIANTS + trash_init, trash_fini, +#else + NULL, NULL, +#endif + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); + if (nmbjumbo9 > 0) + uma_zone_set_max(zone_jumbo9, nmbjumbo9); + uma_zone_set_allocf(zone_jumbo9, mbuf_jumbo_alloc); + uma_zone_set_freef(zone_jumbo9, mbuf_jumbo_free); + + zone_jumbo16 = uma_zcreate(MBUF_JUMBO16_MEM_NAME, MJUM16BYTES, + mb_ctor_clust, mb_dtor_clust, +#ifdef INVARIANTS + trash_init, trash_fini, +#else + NULL, NULL, +#endif + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); + if (nmbjumbo16 > 0) + uma_zone_set_max(zone_jumbo16, nmbjumbo16); +#else /*!VBOX*/ + zone_jumbop = uma_zcreate(MBUF_JUMBOP_MEM_NAME, MJUMPAGESIZE, + mb_ctor_clust, mb_dtor_clust, + NULL, NULL, + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); + if (nmbjumbop > 0) + uma_zone_set_max(zone_jumbop, nmbjumbop); + + zone_jumbo9 = uma_zcreate(MBUF_JUMBO9_MEM_NAME, MJUM9BYTES, + mb_ctor_clust, mb_dtor_clust, + NULL, NULL, + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); + if (nmbjumbo9 > 0) + uma_zone_set_max(zone_jumbo9, nmbjumbo9); + + zone_jumbo16 = uma_zcreate(MBUF_JUMBO16_MEM_NAME, MJUM16BYTES, + mb_ctor_clust, mb_dtor_clust, + NULL, NULL, + UMA_ALIGN_PTR, UMA_ZONE_REFCNT); + if (nmbjumbo16 > 0) + uma_zone_set_max(zone_jumbo16, nmbjumbo16); +#endif /*VBOX*/ + + zone_ext_refcnt = uma_zcreate(MBUF_EXTREFCNT_MEM_NAME, sizeof(u_int), + NULL, NULL, + NULL, NULL, + UMA_ALIGN_PTR, UMA_ZONE_ZINIT); + + /* uma_prealloc() goes here... */ + + /* + * Hook event handler for low-memory situation, used to + * drain protocols and push data back to the caches (UMA + * later pushes it back to VM). + */ + EVENTHANDLER_REGISTER(vm_lowmem, mb_reclaim, NULL, + EVENTHANDLER_PRI_FIRST); + + /* + * [Re]set counters and local statistics knobs. + * XXX Some of these should go and be replaced, but UMA stat + * gathering needs to be revised. + */ + mbstat.m_mbufs = 0; + mbstat.m_mclusts = 0; + mbstat.m_drain = 0; + mbstat.m_msize = MSIZE; + mbstat.m_mclbytes = MCLBYTES; + mbstat.m_minclsize = MINCLSIZE; + mbstat.m_mlen = MLEN; + mbstat.m_mhlen = MHLEN; + mbstat.m_numtypes = MT_NTYPES; + + mbstat.m_mcfail = mbstat.m_mpfail = 0; + mbstat.sf_iocnt = 0; + mbstat.sf_allocwait = mbstat.sf_allocfail = 0; +} + +#ifndef VBOX +/* + * UMA backend page allocator for the jumbo frame zones. + * + * Allocates kernel virtual memory that is backed by contiguous physical + * pages. + */ +static void * +mbuf_jumbo_alloc(uma_zone_t zone, int bytes, u_int8_t *flags, int fWait) +{ + + /* Inform UMA that this allocator uses kernel_map/object. */ + *flags = UMA_SLAB_KERNEL; + return (contigmalloc(bytes, M_JUMBOFRAME, fWait, (vm_paddr_t)0, + ~(vm_paddr_t)0, 1, 0)); +} + +/* + * UMA backend page deallocator for the jumbo frame zones. + */ +static void +mbuf_jumbo_free(void *mem, int size, u_int8_t flags) +{ + + contigfree(mem, size, M_JUMBOFRAME); +} +#endif + +/* + * Constructor for Mbuf master zone. + * + * The 'arg' pointer points to a mb_args structure which + * contains call-specific information required to support the + * mbuf allocation API. See mbuf.h. + */ +static int +#ifndef VBOX +mb_ctor_mbuf(void *mem, int size, void *arg, int how) +#else +mb_ctor_mbuf(PNATState pData, void *mem, int size, void *arg, int how) +#endif +{ + struct mbuf *m; + struct mb_args *args; +#ifdef MAC + int error; +#endif + int flags; + short type; +#ifdef VBOX + NOREF(pData); +#endif + +#ifdef INVARIANTS + trash_ctor(mem, size, arg, how); +#elif defined(VBOX) + NOREF(size); + NOREF(how); +#endif + m = (struct mbuf *)mem; + args = (struct mb_args *)arg; + flags = args->flags; + type = args->type; + + /* + * The mbuf is initialized later. The caller has the + * responsibility to set up any MAC labels too. + */ + if (type == MT_NOINIT) + return (0); + + m->m_next = NULL; + m->m_nextpkt = NULL; + m->m_len = 0; + m->m_flags = flags; + m->m_type = type; + if (flags & M_PKTHDR) { + m->m_data = m->m_pktdat; + m->m_pkthdr.rcvif = NULL; + m->m_pkthdr.len = 0; + m->m_pkthdr.header = NULL; + m->m_pkthdr.csum_flags = 0; + m->m_pkthdr.csum_data = 0; + m->m_pkthdr.tso_segsz = 0; + m->m_pkthdr.ether_vtag = 0; + SLIST_INIT(&m->m_pkthdr.tags); +#ifdef MAC + /* If the label init fails, fail the alloc */ + error = mac_init_mbuf(m, how); + if (error) + return (error); +#endif + } else + m->m_data = m->m_dat; + return (0); +} + +/* + * The Mbuf master zone destructor. + */ +static void +#ifndef VBOX +mb_dtor_mbuf(void *mem, int size, void *arg) +#else +mb_dtor_mbuf(PNATState pData, void *mem, int size, void *arg) +#endif +{ + struct mbuf *m; + uintptr_t flags; +#ifdef VBOX + NOREF(pData); +#endif + + m = (struct mbuf *)mem; + flags = (uintptr_t)arg; + + if ((flags & MB_NOTAGS) == 0 && (m->m_flags & M_PKTHDR) != 0) + m_tag_delete_chain(m, NULL); + KASSERT((m->m_flags & M_EXT) == 0, ("%s: M_EXT set", __func__)); + KASSERT((m->m_flags & M_NOFREE) == 0, ("%s: M_NOFREE set", __func__)); +#ifdef INVARIANTS + trash_dtor(mem, size, arg); +#elif defined(VBOX) + NOREF(size); + NOREF(arg); +#endif +} + +/* + * The Mbuf Packet zone destructor. + */ +static void +#ifndef VBOX +mb_dtor_pack(void *mem, int size, void *arg) +#else +mb_dtor_pack(PNATState pData, void *mem, int size, void *arg) +#endif +{ + struct mbuf *m; + + m = (struct mbuf *)mem; + if ((m->m_flags & M_PKTHDR) != 0) + m_tag_delete_chain(m, NULL); + + /* Make sure we've got a clean cluster back. */ + KASSERT((m->m_flags & M_EXT) == M_EXT, ("%s: M_EXT not set", __func__)); + KASSERT(m->m_ext.ext_buf != NULL, ("%s: ext_buf == NULL", __func__)); + KASSERT(m->m_ext.ext_free == NULL, ("%s: ext_free != NULL", __func__)); + KASSERT(m->m_ext.ext_args == NULL, ("%s: ext_args != NULL", __func__)); + KASSERT(m->m_ext.ext_size == MCLBYTES, ("%s: ext_size != MCLBYTES", __func__)); + KASSERT(m->m_ext.ext_type == EXT_PACKET, ("%s: ext_type != EXT_PACKET", __func__)); + KASSERT(*m->m_ext.ref_cnt == 1, ("%s: ref_cnt != 1", __func__)); +#ifdef INVARIANTS + trash_dtor(m->m_ext.ext_buf, MCLBYTES, arg); +#elif defined(VBOX) + NOREF(size); + NOREF(arg); +#endif + /* + * If there are processes blocked on zone_clust, waiting for pages to be freed up, + * cause them to be woken up by draining the packet zone. We are exposed to a race here + * (in the check for the UMA_ZFLAG_FULL) where we might miss the flag set, but that is + * deliberate. We don't want to acquire the zone lock for every mbuf free. + */ + if (uma_zone_exhausted_nolock(zone_clust)) + zone_drain(zone_pack); +} + +/* + * The Cluster and Jumbo[PAGESIZE|9|16] zone constructor. + * + * Here the 'arg' pointer points to the Mbuf which we + * are configuring cluster storage for. If 'arg' is + * empty we allocate just the cluster without setting + * the mbuf to it. See mbuf.h. + */ +static int +#ifndef VBOX +mb_ctor_clust(void *mem, int size, void *arg, int how) +#else +mb_ctor_clust(PNATState pData, void *mem, int size, void *arg, int how) +#endif +{ + struct mbuf *m; + u_int *refcnt; + int type; + uma_zone_t zone; +#ifdef VBOX + NOREF(how); +#endif + +#ifdef INVARIANTS + trash_ctor(mem, size, arg, how); +#elif defined(VBOX) + NOREF(how); +#endif + switch (size) { + case MCLBYTES: + type = EXT_CLUSTER; + zone = zone_clust; + break; +#if MJUMPAGESIZE != MCLBYTES + case MJUMPAGESIZE: + type = EXT_JUMBOP; + zone = zone_jumbop; + break; +#endif + case MJUM9BYTES: + type = EXT_JUMBO9; + zone = zone_jumbo9; + break; + case MJUM16BYTES: + type = EXT_JUMBO16; + zone = zone_jumbo16; + break; + default: + panic("unknown cluster size"); + break; + } + + m = (struct mbuf *)arg; + refcnt = uma_find_refcnt(zone, mem); + *refcnt = 1; + if (m != NULL) { + m->m_ext.ext_buf = (caddr_t)mem; + m->m_data = m->m_ext.ext_buf; + m->m_flags |= M_EXT; + m->m_ext.ext_free = NULL; + m->m_ext.ext_args = NULL; + m->m_ext.ext_size = size; + m->m_ext.ext_type = type; + m->m_ext.ref_cnt = refcnt; + } + + return (0); +} + +/* + * The Mbuf Cluster zone destructor. + */ +static void +#ifndef VBOX +mb_dtor_clust(void *mem, int size, void *arg) +#else +mb_dtor_clust(PNATState pData, void *mem, int size, void *arg) +#endif +{ +#ifdef INVARIANTS + uma_zone_t zone; + + zone = m_getzone(size); + KASSERT(*(uma_find_refcnt(zone, mem)) <= 1, + ("%s: refcnt incorrect %u", __func__, + *(uma_find_refcnt(zone, mem))) ); + + trash_dtor(mem, size, arg); +#elif defined(VBOX) + NOREF(pData); + NOREF(mem); + NOREF(size); + NOREF(arg); +#endif +} + +/* + * The Packet secondary zone's init routine, executed on the + * object's transition from mbuf keg slab to zone cache. + */ +static int +#ifndef VBOX +mb_zinit_pack(void *mem, int size, int how) +#else +mb_zinit_pack(PNATState pData, void *mem, int size, int how) +#endif +{ + struct mbuf *m; + + m = (struct mbuf *)mem; /* m is virgin. */ + if (uma_zalloc_arg(zone_clust, m, how) == NULL || + m->m_ext.ext_buf == NULL) + return (ENOMEM); + m->m_ext.ext_type = EXT_PACKET; /* Override. */ +#ifdef INVARIANTS + trash_init(m->m_ext.ext_buf, MCLBYTES, how); +#elif defined(VBOX) + NOREF(size); +#endif + return (0); +} + +/* + * The Packet secondary zone's fini routine, executed on the + * object's transition from zone cache to keg slab. + */ +static void +#ifndef VBOX +mb_zfini_pack(void *mem, int size) +#else +mb_zfini_pack(PNATState pData, void *mem, int size) +#endif +{ + struct mbuf *m; + + m = (struct mbuf *)mem; +#ifdef INVARIANTS + trash_fini(m->m_ext.ext_buf, MCLBYTES); +#endif + uma_zfree_arg(zone_clust, m->m_ext.ext_buf, NULL); +#ifdef INVARIANTS + trash_dtor(mem, size, NULL); +#elif defined(VBOX) + NOREF(size); +#endif +} + +/* + * The "packet" keg constructor. + */ +static int +#ifndef VBOX +mb_ctor_pack(void *mem, int size, void *arg, int how) +#else +mb_ctor_pack(PNATState pData, void *mem, int size, void *arg, int how) +#endif +{ + struct mbuf *m; + struct mb_args *args; +#ifdef MAC + int error; +#endif + int flags; + short type; +#ifdef VBOX + NOREF(pData); + NOREF(size); +#endif + + m = (struct mbuf *)mem; + args = (struct mb_args *)arg; + flags = args->flags; + type = args->type; + +#ifdef INVARIANTS + trash_ctor(m->m_ext.ext_buf, MCLBYTES, arg, how); +#elif defined(VBOX) + NOREF(how); +#endif + m->m_next = NULL; + m->m_nextpkt = NULL; + m->m_data = m->m_ext.ext_buf; + m->m_len = 0; + m->m_flags = (flags | M_EXT); + m->m_type = type; + + if (flags & M_PKTHDR) { + m->m_pkthdr.rcvif = NULL; + m->m_pkthdr.len = 0; + m->m_pkthdr.header = NULL; + m->m_pkthdr.csum_flags = 0; + m->m_pkthdr.csum_data = 0; + m->m_pkthdr.tso_segsz = 0; + m->m_pkthdr.ether_vtag = 0; + SLIST_INIT(&m->m_pkthdr.tags); +#ifdef MAC + /* If the label init fails, fail the alloc */ + error = mac_init_mbuf(m, how); + if (error) + return (error); +#endif + } + /* m_ext is already initialized. */ + + return (0); +} + +#if 0 /* unused */ +/* + * This is the protocol drain routine. + * + * No locks should be held when this is called. The drain routines have to + * presently acquire some locks which raises the possibility of lock order + * reversal. + */ +static void +mb_reclaim(void *junk) +{ +#ifndef VBOX + struct domain *dp; + struct protosw *pr; + + WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK | WARN_PANIC, NULL, + "mb_reclaim()"); + + for (dp = domains; dp != NULL; dp = dp->dom_next) + for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) + if (pr->pr_drain != NULL) + (*pr->pr_drain)(); +#else + NOREF(junk); +#endif +} +#endif /* unused */ diff --git a/src/VBox/Devices/Network/slirp/bsd/kern/subr_sbuf.c b/src/VBox/Devices/Network/slirp/bsd/kern/subr_sbuf.c new file mode 100644 index 00000000..c2021f5b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/kern/subr_sbuf.c @@ -0,0 +1,594 @@ +/*- + * Copyright (c) 2000 Poul-Henning Kamp and Dag-Erling Coïdan Smørgrav + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef VBOX +#include <sys/cdefs.h> +__FBSDID("$FreeBSD: src/sys/kern/subr_sbuf.c,v 1.30.8.1 2009/04/15 03:14:26 kensmith Exp $"); + +#include <sys/param.h> + +#ifdef _KERNEL +#include <sys/ctype.h> +#include <sys/kernel.h> +#include <sys/malloc.h> +#include <sys/systm.h> +#include <sys/uio.h> +#include <machine/stdarg.h> +#else /* _KERNEL */ +#include <ctype.h> +#include <stdarg.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#endif /* _KERNEL */ + +#include <sys/sbuf.h> + +#ifdef _KERNEL +static MALLOC_DEFINE(M_SBUF, "sbuf", "string buffers"); +#define SBMALLOC(size) malloc(size, M_SBUF, M_WAITOK) +#define SBFREE(buf) free(buf, M_SBUF) +#else /* _KERNEL */ +#define KASSERT(e, m) +#define SBMALLOC(size) malloc(size) +#define SBFREE(buf) free(buf) +#define min(x,y) MIN(x,y) +#endif /* _KERNEL */ +#else /* VBOX */ +# include <iprt/param.h> +# include <iprt/ctype.h> +# include <slirp.h> +# define SBMALLOC(size) RTMemAlloc((size)) +# define SBFREE(buf) RTMemFree((buf)) +#endif + +/* + * Predicates + */ +#define SBUF_ISDYNAMIC(s) ((s)->s_flags & SBUF_DYNAMIC) +#define SBUF_ISDYNSTRUCT(s) ((s)->s_flags & SBUF_DYNSTRUCT) +#define SBUF_ISFINISHED(s) ((s)->s_flags & SBUF_FINISHED) +#define SBUF_HASOVERFLOWED(s) ((s)->s_flags & SBUF_OVERFLOWED) +#define SBUF_HASROOM(s) ((s)->s_len < (s)->s_size - 1) +#define SBUF_FREESPACE(s) ((s)->s_size - (s)->s_len - 1) +#define SBUF_CANEXTEND(s) ((s)->s_flags & SBUF_AUTOEXTEND) + +/* + * Set / clear flags + */ +#define SBUF_SETFLAG(s, f) do { (s)->s_flags |= (f); } while (0) +#define SBUF_CLEARFLAG(s, f) do { (s)->s_flags &= ~(f); } while (0) + +#define SBUF_MINEXTENDSIZE 16 /* Should be power of 2. */ +#define SBUF_MAXEXTENDSIZE PAGE_SIZE +#define SBUF_MAXEXTENDINCR PAGE_SIZE + +/* + * Debugging support + */ +#if defined(_KERNEL) && defined(INVARIANTS) +static void +_assert_sbuf_integrity(const char *fun, struct sbuf *s) +{ + KASSERT(s != NULL, + ("%s called with a NULL sbuf pointer", fun)); + KASSERT(s->s_buf != NULL, + ("%s called with uninitialized or corrupt sbuf", fun)); + KASSERT(s->s_len < s->s_size, + ("wrote past end of sbuf (%d >= %d)", s->s_len, s->s_size)); +} + +static void +_assert_sbuf_state(const char *fun, struct sbuf *s, int state) +{ + KASSERT((s->s_flags & SBUF_FINISHED) == state, + ("%s called with %sfinished or corrupt sbuf", fun, + (state ? "un" : ""))); +} +#define assert_sbuf_integrity(s) _assert_sbuf_integrity(__func__, (s)) +#define assert_sbuf_state(s, i) _assert_sbuf_state(__func__, (s), (i)) +#else /* _KERNEL && INVARIANTS */ +#define assert_sbuf_integrity(s) do { } while (0) +#define assert_sbuf_state(s, i) do { } while (0) +#endif /* _KERNEL && INVARIANTS */ + +static int +sbuf_extendsize(int size) +{ + int newsize; + + newsize = SBUF_MINEXTENDSIZE; + while (newsize < size) { + if (newsize < (int)SBUF_MAXEXTENDSIZE) + newsize *= 2; + else + newsize += SBUF_MAXEXTENDINCR; + } + + return (newsize); +} + + +/* + * Extend an sbuf. + */ +static int +sbuf_extend(struct sbuf *s, int addlen) +{ + char *newbuf; + int newsize; + + if (!SBUF_CANEXTEND(s)) + return (-1); + + newsize = sbuf_extendsize(s->s_size + addlen); + newbuf = (char *)SBMALLOC(newsize); + if (newbuf == NULL) + return (-1); + bcopy(s->s_buf, newbuf, s->s_size); + if (SBUF_ISDYNAMIC(s)) + SBFREE(s->s_buf); + else + SBUF_SETFLAG(s, SBUF_DYNAMIC); + s->s_buf = newbuf; + s->s_size = newsize; + return (0); +} + +/* + * Initialize an sbuf. + * If buf is non-NULL, it points to a static or already-allocated string + * big enough to hold at least length characters. + */ +struct sbuf * +sbuf_new(struct sbuf *s, char *buf, int length, int flags) +{ + KASSERT(length >= 0, + ("attempt to create an sbuf of negative length (%d)", length)); + KASSERT((flags & ~SBUF_USRFLAGMSK) == 0, + ("%s called with invalid flags", __func__)); + + flags &= SBUF_USRFLAGMSK; + if (s == NULL) { + s = (struct sbuf *)SBMALLOC(sizeof *s); + if (s == NULL) + return (NULL); + bzero(s, sizeof *s); + s->s_flags = flags; + SBUF_SETFLAG(s, SBUF_DYNSTRUCT); + } else { + bzero(s, sizeof *s); + s->s_flags = flags; + } + s->s_size = length; + if (buf) { + s->s_buf = buf; + return (s); + } + if (flags & SBUF_AUTOEXTEND) + s->s_size = sbuf_extendsize(s->s_size); + s->s_buf = (char *)SBMALLOC(s->s_size); + if (s->s_buf == NULL) { + if (SBUF_ISDYNSTRUCT(s)) + SBFREE(s); + return (NULL); + } + SBUF_SETFLAG(s, SBUF_DYNAMIC); + return (s); +} + +#ifdef _KERNEL +/* + * Create an sbuf with uio data + */ +struct sbuf * +sbuf_uionew(struct sbuf *s, struct uio *uio, int *error) +{ + KASSERT(uio != NULL, + ("%s called with NULL uio pointer", __func__)); + KASSERT(error != NULL, + ("%s called with NULL error pointer", __func__)); + + s = sbuf_new(s, NULL, uio->uio_resid + 1, 0); + if (s == NULL) { + *error = ENOMEM; + return (NULL); + } + *error = uiomove(s->s_buf, uio->uio_resid, uio); + if (*error != 0) { + sbuf_delete(s); + return (NULL); + } + s->s_len = s->s_size - 1; + *error = 0; + return (s); +} +#endif + +/* + * Clear an sbuf and reset its position. + */ +void +sbuf_clear(struct sbuf *s) +{ + assert_sbuf_integrity(s); + /* don't care if it's finished or not */ + + SBUF_CLEARFLAG(s, SBUF_FINISHED); + SBUF_CLEARFLAG(s, SBUF_OVERFLOWED); + s->s_len = 0; +} + +/* + * Set the sbuf's end position to an arbitrary value. + * Effectively truncates the sbuf at the new position. + */ +int +sbuf_setpos(struct sbuf *s, int pos) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + KASSERT(pos >= 0, + ("attempt to seek to a negative position (%d)", pos)); + KASSERT(pos < s->s_size, + ("attempt to seek past end of sbuf (%d >= %d)", pos, s->s_size)); + + if (pos < 0 || pos > s->s_len) + return (-1); + s->s_len = pos; + return (0); +} + +/* + * Append a byte string to an sbuf. + */ +int +sbuf_bcat(struct sbuf *s, const void *buf, size_t len) +{ + const char *str = buf; + + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + + for (; len; len--) { + if (!SBUF_HASROOM(s) && sbuf_extend(s, len) < 0) + break; + s->s_buf[s->s_len++] = *str++; + } + if (len) { + SBUF_SETFLAG(s, SBUF_OVERFLOWED); + return (-1); + } + return (0); +} + +#ifdef _KERNEL +/* + * Copy a byte string from userland into an sbuf. + */ +int +sbuf_bcopyin(struct sbuf *s, const void *uaddr, size_t len) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + + if (len == 0) + return (0); + if (len > SBUF_FREESPACE(s)) { + sbuf_extend(s, len - SBUF_FREESPACE(s)); + len = min(len, SBUF_FREESPACE(s)); + } + if (copyin(uaddr, s->s_buf + s->s_len, len) != 0) + return (-1); + s->s_len += len; + + return (0); +} +#endif + +/* + * Copy a byte string into an sbuf. + */ +int +sbuf_bcpy(struct sbuf *s, const void *buf, size_t len) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + sbuf_clear(s); + return (sbuf_bcat(s, buf, len)); +} + +/* + * Append a string to an sbuf. + */ +int +sbuf_cat(struct sbuf *s, const char *str) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + + while (*str) { + if (!SBUF_HASROOM(s) && sbuf_extend(s, strlen(str)) < 0) + break; + s->s_buf[s->s_len++] = *str++; + } + if (*str) { + SBUF_SETFLAG(s, SBUF_OVERFLOWED); + return (-1); + } + return (0); +} + +#ifdef _KERNEL +/* + * Append a string from userland to an sbuf. + */ +int +sbuf_copyin(struct sbuf *s, const void *uaddr, size_t len) +{ + size_t done; + + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + + if (len == 0) + len = SBUF_FREESPACE(s); /* XXX return 0? */ + if (len > SBUF_FREESPACE(s)) { + sbuf_extend(s, len); + len = min(len, SBUF_FREESPACE(s)); + } + switch (copyinstr(uaddr, s->s_buf + s->s_len, len + 1, &done)) { + case ENAMETOOLONG: + SBUF_SETFLAG(s, SBUF_OVERFLOWED); + RT_FALL_THRU(); + case 0: + s->s_len += done - 1; + break; + default: + return (-1); /* XXX */ + } + + return (done); +} +#endif + +/* + * Copy a string into an sbuf. + */ +int +sbuf_cpy(struct sbuf *s, const char *str) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + sbuf_clear(s); + return (sbuf_cat(s, str)); +} + +/* + * Format the given argument list and append the resulting string to an sbuf. + */ +int +sbuf_vprintf(struct sbuf *s, const char *fmt, va_list ap) +{ + va_list ap_copy; + int len; + + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + KASSERT(fmt != NULL, + ("%s called with a NULL format string", __func__)); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + + do { + va_copy(ap_copy, ap); +#ifndef VBOX + len = vsnprintf(&s->s_buf[s->s_len], SBUF_FREESPACE(s) + 1, + fmt, ap_copy); +#else + len = RTStrPrintfV(&s->s_buf[s->s_len], SBUF_FREESPACE(s) + 1, + fmt, ap_copy); +#endif + va_end(ap_copy); + } while (len > SBUF_FREESPACE(s) && + sbuf_extend(s, len - SBUF_FREESPACE(s)) == 0); + + /* + * s->s_len is the length of the string, without the terminating nul. + * When updating s->s_len, we must subtract 1 from the length that + * we passed into vsnprintf() because that length includes the + * terminating nul. + * + * vsnprintf() returns the amount that would have been copied, + * given sufficient space, hence the min() calculation below. + */ + s->s_len += min(len, SBUF_FREESPACE(s)); + if (!SBUF_HASROOM(s) && !SBUF_CANEXTEND(s)) + SBUF_SETFLAG(s, SBUF_OVERFLOWED); + + KASSERT(s->s_len < s->s_size, + ("wrote past end of sbuf (%d >= %d)", s->s_len, s->s_size)); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + return (0); +} + +/* + * Format the given arguments and append the resulting string to an sbuf. + */ +int +sbuf_printf(struct sbuf *s, const char *fmt, ...) +{ + va_list ap; + int result; + + va_start(ap, fmt); + result = sbuf_vprintf(s, fmt, ap); + va_end(ap); + return(result); +} + +/* + * Append a character to an sbuf. + */ +int +sbuf_putc(struct sbuf *s, int c) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + + if (!SBUF_HASROOM(s) && sbuf_extend(s, 1) < 0) { + SBUF_SETFLAG(s, SBUF_OVERFLOWED); + return (-1); + } + if (c != '\0') + s->s_buf[s->s_len++] = c; + return (0); +} + +/* + * Trim whitespace characters from end of an sbuf. + */ +int +sbuf_trim(struct sbuf *s) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + +#ifndef VBOX + while (s->s_len && isspace(s->s_buf[s->s_len-1])) + --s->s_len; +#else + while (s->s_len && RT_C_IS_SPACE(s->s_buf[s->s_len-1])) + --s->s_len; +#endif + + return (0); +} + +/* + * Check if an sbuf overflowed + */ +int +sbuf_overflowed(struct sbuf *s) +{ + return SBUF_HASOVERFLOWED(s); +} + +/* + * Finish off an sbuf. + */ +void +sbuf_finish(struct sbuf *s) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, 0); + + s->s_buf[s->s_len] = '\0'; + SBUF_CLEARFLAG(s, SBUF_OVERFLOWED); + SBUF_SETFLAG(s, SBUF_FINISHED); +} + +/* + * Return a pointer to the sbuf data. + */ +char * +sbuf_data(struct sbuf *s) +{ + assert_sbuf_integrity(s); + assert_sbuf_state(s, SBUF_FINISHED); + + return s->s_buf; +} + +/* + * Return the length of the sbuf data. + */ +int +sbuf_len(struct sbuf *s) +{ + assert_sbuf_integrity(s); + /* don't care if it's finished or not */ + + if (SBUF_HASOVERFLOWED(s)) + return (-1); + return s->s_len; +} + +/* + * Clear an sbuf, free its buffer if necessary. + */ +void +sbuf_delete(struct sbuf *s) +{ + int isdyn; + + assert_sbuf_integrity(s); + /* don't care if it's finished or not */ + + if (SBUF_ISDYNAMIC(s)) + SBFREE(s->s_buf); + isdyn = SBUF_ISDYNSTRUCT(s); + bzero(s, sizeof *s); + if (isdyn) + SBFREE(s); +} + +/* + * Check if an sbuf has been finished. + */ +int +sbuf_done(struct sbuf *s) +{ + + return(SBUF_ISFINISHED(s)); +} diff --git a/src/VBox/Devices/Network/slirp/bsd/kern/uipc_mbuf.c b/src/VBox/Devices/Network/slirp/bsd/kern/uipc_mbuf.c new file mode 100644 index 00000000..0af49faa --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/kern/uipc_mbuf.c @@ -0,0 +1,2238 @@ +/*- + * Copyright (c) 1982, 1986, 1988, 1991, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)uipc_mbuf.c 8.2 (Berkeley) 1/4/94 + */ + +#ifndef VBOX +#include <sys/cdefs.h> +__FBSDID("$FreeBSD: src/sys/kern/uipc_mbuf.c,v 1.174.2.3.2.1 2009/04/15 03:14:26 kensmith Exp $"); + +#include "opt_mac.h" +#include "opt_param.h" +#include "opt_mbuf_stress_test.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/kernel.h> +#include <sys/limits.h> +#include <sys/lock.h> +#include <sys/malloc.h> +#include <sys/mbuf.h> +#include <sys/sysctl.h> +#include <sys/domain.h> +#include <sys/protosw.h> +#include <sys/uio.h> + +#include <security/mac/mac_framework.h> + +int max_linkhdr; +#ifndef VBOX +int max_protohdr; +#endif +int max_hdr; +int max_datalen; +#ifdef MBUF_STRESS_TEST +int m_defragpackets; +int m_defragbytes; +int m_defraguseless; +int m_defragfailure; +int m_defragrandomfailures; +#endif + +/* + * sysctl(8) exported objects + */ +SYSCTL_INT(_kern_ipc, KIPC_MAX_LINKHDR, max_linkhdr, CTLFLAG_RD, + &max_linkhdr, 0, "Size of largest link layer header"); +SYSCTL_INT(_kern_ipc, KIPC_MAX_PROTOHDR, max_protohdr, CTLFLAG_RD, + &max_protohdr, 0, "Size of largest protocol layer header"); +SYSCTL_INT(_kern_ipc, KIPC_MAX_HDR, max_hdr, CTLFLAG_RD, + &max_hdr, 0, "Size of largest link plus protocol header"); +SYSCTL_INT(_kern_ipc, KIPC_MAX_DATALEN, max_datalen, CTLFLAG_RD, + &max_datalen, 0, "Minimum space left in mbuf after max_hdr"); +#ifdef MBUF_STRESS_TEST +SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragpackets, CTLFLAG_RD, + &m_defragpackets, 0, ""); +SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragbytes, CTLFLAG_RD, + &m_defragbytes, 0, ""); +SYSCTL_INT(_kern_ipc, OID_AUTO, m_defraguseless, CTLFLAG_RD, + &m_defraguseless, 0, ""); +SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragfailure, CTLFLAG_RD, + &m_defragfailure, 0, ""); +SYSCTL_INT(_kern_ipc, OID_AUTO, m_defragrandomfailures, CTLFLAG_RW, + &m_defragrandomfailures, 0, ""); +#endif +#else /* VBOX */ +# include <iprt/asm.h> +# include "slirp.h" +# define atomic_fetchadd_int(var, val) (ASMAtomicAddU32((var), (val))) +# define atomic_add_int(var, val) (ASMAtomicAddU32((var), (val))) +#endif /* VBOX */ + +/* + * Allocate a given length worth of mbufs and/or clusters (whatever fits + * best) and return a pointer to the top of the allocated chain. If an + * existing mbuf chain is provided, then we will append the new chain + * to the existing one but still return the top of the newly allocated + * chain. + */ +struct mbuf * +#ifndef VBOX +m_getm2(struct mbuf *m, int len, int how, short type, int flags) +#else +m_getm2(PNATState pData, struct mbuf *m, int len, int how, short type, int flags) +#endif +{ + struct mbuf *mb, *nm = NULL, *mtail = NULL; + + KASSERT(len >= 0, ("%s: len is < 0", __func__)); + + /* Validate flags. */ + flags &= (M_PKTHDR | M_EOR); + + /* Packet header mbuf must be first in chain. */ + if ((flags & M_PKTHDR) && m != NULL) + flags &= ~M_PKTHDR; + + /* Loop and append maximum sized mbufs to the chain tail. */ + while (len > 0) { +#ifndef VBOX + if (len > MCLBYTES) + mb = m_getjcl(how, type, (flags & M_PKTHDR), + MJUMPAGESIZE); + else if (len >= MINCLSIZE) + mb = m_getcl(how, type, (flags & M_PKTHDR)); + else if (flags & M_PKTHDR) + mb = m_gethdr(how, type); + else + mb = m_get(how, type); + + /* Fail the whole operation if one mbuf can't be allocated. */ + if (mb == NULL) { + if (nm != NULL) + m_freem(nm); + return (NULL); + } +#else + if (len > MCLBYTES) + mb = m_getjcl(pData, how, type, (flags & M_PKTHDR), + MJUMPAGESIZE); + else if (len >= MINCLSIZE) + mb = m_getcl(pData, how, type, (flags & M_PKTHDR)); + else if (flags & M_PKTHDR) + mb = m_gethdr(pData, how, type); + else + mb = m_get(pData, how, type); + /* Fail the whole operation if one mbuf can't be allocated. */ + if (mb == NULL) { + if (nm != NULL) + m_freem(pData, nm); + return (NULL); + } +#endif + + /* Book keeping. */ + len -= (mb->m_flags & M_EXT) ? mb->m_ext.ext_size : + ((mb->m_flags & M_PKTHDR) ? MHLEN : MLEN); + if (mtail != NULL) + mtail->m_next = mb; + else + nm = mb; + mtail = mb; + flags &= ~M_PKTHDR; /* Only valid on the first mbuf. */ + } + if (flags & M_EOR) + mtail->m_flags |= M_EOR; /* Only valid on the last mbuf. */ + + /* If mbuf was supplied, append new chain to the end of it. */ + if (m != NULL) { + for (mtail = m; mtail->m_next != NULL; mtail = mtail->m_next) + ; + mtail->m_next = nm; + mtail->m_flags &= ~M_EOR; + } else + m = nm; + + return (m); +} + +/* + * Free an entire chain of mbufs and associated external buffers, if + * applicable. + */ +void +#ifndef VBOX +m_freem(struct mbuf *mb) +#else +m_freem(PNATState pData, struct mbuf *mb) +#endif +{ + + while (mb != NULL) + mb = m_free(pData, mb); +} + +/*- + * Configure a provided mbuf to refer to the provided external storage + * buffer and setup a reference count for said buffer. If the setting + * up of the reference count fails, the M_EXT bit will not be set. If + * successfull, the M_EXT bit is set in the mbuf's flags. + * + * Arguments: + * mb The existing mbuf to which to attach the provided buffer. + * buf The address of the provided external storage buffer. + * size The size of the provided buffer. + * freef A pointer to a routine that is responsible for freeing the + * provided external storage buffer. + * args A pointer to an argument structure (of any type) to be passed + * to the provided freef routine (may be NULL). + * flags Any other flags to be passed to the provided mbuf. + * type The type that the external storage buffer should be + * labeled with. + * + * Returns: + * Nothing. + */ +void +#ifndef VBOX +m_extadd(struct mbuf *mb, caddr_t buf, u_int size, + void (*freef)(void *, void *), void *args, int flags, int type) +#else +m_extadd(PNATState pData, struct mbuf *mb, caddr_t buf, u_int size, + void (*freef)(void *, void *), void *args, int flags, int type) +#endif +{ + KASSERT(type != EXT_CLUSTER, ("%s: EXT_CLUSTER not allowed", __func__)); + + if (type != EXT_EXTREF) + mb->m_ext.ref_cnt = (u_int *)uma_zalloc(zone_ext_refcnt, M_NOWAIT); + if (mb->m_ext.ref_cnt != NULL) { + *(mb->m_ext.ref_cnt) = 1; + mb->m_flags |= (M_EXT | flags); + mb->m_ext.ext_buf = buf; + mb->m_data = mb->m_ext.ext_buf; + mb->m_ext.ext_size = size; + mb->m_ext.ext_free = freef; + mb->m_ext.ext_args = args; + mb->m_ext.ext_type = type; + } +} + +/* + * Non-directly-exported function to clean up after mbufs with M_EXT + * storage attached to them if the reference count hits 1. + */ +void +#ifndef VBOX +mb_free_ext(struct mbuf *m) +#else +mb_free_ext(PNATState pData, struct mbuf *m) +#endif +{ + int skipmbuf; + + KASSERT((m->m_flags & M_EXT) == M_EXT, ("%s: M_EXT not set", __func__)); + KASSERT(m->m_ext.ref_cnt != NULL, ("%s: ref_cnt not set", __func__)); + + + /* + * check if the header is embedded in the cluster + */ + skipmbuf = (m->m_flags & M_NOFREE); + + /* Free attached storage if this mbuf is the only reference to it. */ + if (*(m->m_ext.ref_cnt) == 1 || + atomic_fetchadd_int(m->m_ext.ref_cnt, (uint32_t)-1) == 1) { + switch (m->m_ext.ext_type) { + case EXT_PACKET: /* The packet zone is special. */ + if (*(m->m_ext.ref_cnt) == 0) + *(m->m_ext.ref_cnt) = 1; + uma_zfree(zone_pack, m); + return; /* Job done. */ + case EXT_CLUSTER: + uma_zfree(zone_clust, m->m_ext.ext_buf); + break; + case EXT_JUMBOP: + uma_zfree(zone_jumbop, m->m_ext.ext_buf); + break; + case EXT_JUMBO9: + uma_zfree(zone_jumbo9, m->m_ext.ext_buf); + break; + case EXT_JUMBO16: + uma_zfree(zone_jumbo16, m->m_ext.ext_buf); + break; + case EXT_SFBUF: + case EXT_NET_DRV: + case EXT_MOD_TYPE: + case EXT_DISPOSABLE: +#ifndef VBOX + /* This code is dead in VBOX port of BSD mbufs (probably will be used for EXT_SBUFS some day) + * @todo once bsd sbufs will be on trunk think about this code. + */ + *(m->m_ext.ref_cnt) = 0; + uma_zfree(zone_ext_refcnt, __DEVOLATILE(u_int *, + m->m_ext.ref_cnt)); +#else + AssertMsgFailed(("unimplemented")); +#endif + RT_FALL_THRU(); + case EXT_EXTREF: + KASSERT(m->m_ext.ext_free != NULL, + ("%s: ext_free not set", __func__)); + (*(m->m_ext.ext_free))(m->m_ext.ext_buf, + m->m_ext.ext_args); + break; + default: + KASSERT(m->m_ext.ext_type == 0, + ("%s: unknown ext_type", __func__)); + } + } + if (skipmbuf) + return; + + /* + * Free this mbuf back to the mbuf zone with all m_ext + * information purged. + */ + m->m_ext.ext_buf = NULL; + m->m_ext.ext_free = NULL; + m->m_ext.ext_args = NULL; + m->m_ext.ref_cnt = NULL; + m->m_ext.ext_size = 0; + m->m_ext.ext_type = 0; + m->m_flags &= ~M_EXT; + uma_zfree(zone_mbuf, m); +} + +/* + * Attach the cluster from *m to *n, set up m_ext in *n + * and bump the refcount of the cluster. + */ +static void +mb_dupcl(struct mbuf *n, struct mbuf *m) +{ + KASSERT((m->m_flags & M_EXT) == M_EXT, ("%s: M_EXT not set", __func__)); + KASSERT(m->m_ext.ref_cnt != NULL, ("%s: ref_cnt not set", __func__)); + KASSERT((n->m_flags & M_EXT) == 0, ("%s: M_EXT set", __func__)); + + if (*(m->m_ext.ref_cnt) == 1) + *(m->m_ext.ref_cnt) += 1; + else + atomic_add_int(m->m_ext.ref_cnt, 1); + n->m_ext.ext_buf = m->m_ext.ext_buf; + n->m_ext.ext_free = m->m_ext.ext_free; + n->m_ext.ext_args = m->m_ext.ext_args; + n->m_ext.ext_size = m->m_ext.ext_size; + n->m_ext.ref_cnt = m->m_ext.ref_cnt; + n->m_ext.ext_type = m->m_ext.ext_type; + n->m_flags |= M_EXT; +} + +/* + * Clean up mbuf (chain) from any tags and packet headers. + * If "all" is set then the first mbuf in the chain will be + * cleaned too. + */ +void +m_demote(struct mbuf *m0, int all) +{ + struct mbuf *m; + + for (m = all ? m0 : m0->m_next; m != NULL; m = m->m_next) { + if (m->m_flags & M_PKTHDR) { + m_tag_delete_chain(m, NULL); + m->m_flags &= ~M_PKTHDR; + bzero(&m->m_pkthdr, sizeof(struct pkthdr)); + } + if (m->m_type == MT_HEADER) + m->m_type = MT_DATA; + if (m != m0 && m->m_nextpkt != NULL) + m->m_nextpkt = NULL; + m->m_flags = m->m_flags & (M_EXT|M_EOR|M_RDONLY|M_FREELIST); + } +} + +/* + * Sanity checks on mbuf (chain) for use in KASSERT() and general + * debugging. + * Returns 0 or panics when bad and 1 on all tests passed. + * Sanitize, 0 to run M_SANITY_ACTION, 1 to garble things so they + * blow up later. + */ +int +#ifndef VBOX +m_sanity(struct mbuf *m0, int sanitize) +#else +m_sanity(PNATState pData, struct mbuf *m0, int sanitize) +#endif +{ + struct mbuf *m; + caddr_t a, b; + int pktlen = 0; + +#ifdef INVARIANTS +#define M_SANITY_ACTION(s) panic("mbuf %p: " s, m) +#else +#define M_SANITY_ACTION(s) printf("mbuf %p: " s, m) +#endif + + for (m = m0; m != NULL; m = m->m_next) { + /* + * Basic pointer checks. If any of these fails then some + * unrelated kernel memory before or after us is trashed. + * No way to recover from that. + */ + a = ((m->m_flags & M_EXT) ? m->m_ext.ext_buf : + ((m->m_flags & M_PKTHDR) ? (caddr_t)(&m->m_pktdat) : + (caddr_t)(&m->m_dat)) ); + b = (caddr_t)(a + (m->m_flags & M_EXT ? m->m_ext.ext_size : + ((m->m_flags & M_PKTHDR) ? MHLEN : MLEN))); + if ((caddr_t)m->m_data < a) + M_SANITY_ACTION("m_data outside mbuf data range left"); + if ((caddr_t)m->m_data > b) + M_SANITY_ACTION("m_data outside mbuf data range right"); + if ((caddr_t)m->m_data + m->m_len > b) + M_SANITY_ACTION("m_data + m_len exeeds mbuf space"); + if ((m->m_flags & M_PKTHDR) && m->m_pkthdr.header) { + if ((caddr_t)m->m_pkthdr.header < a || + (caddr_t)m->m_pkthdr.header > b) + M_SANITY_ACTION("m_pkthdr.header outside mbuf data range"); + } + + /* m->m_nextpkt may only be set on first mbuf in chain. */ + if (m != m0 && m->m_nextpkt != NULL) { + if (sanitize) { +#ifndef VBOX + m_freem(m->m_nextpkt); +#else + m_freem(pData, m->m_nextpkt); +#endif + m->m_nextpkt = (struct mbuf *)(uintptr_t)UINT32_C(0xDEADC0DE); + } else + M_SANITY_ACTION("m->m_nextpkt on in-chain mbuf"); + } + + /* packet length (not mbuf length!) calculation */ + if (m0->m_flags & M_PKTHDR) + pktlen += m->m_len; + + /* m_tags may only be attached to first mbuf in chain. */ + if (m != m0 && m->m_flags & M_PKTHDR && + !SLIST_EMPTY(&m->m_pkthdr.tags)) { + if (sanitize) { + m_tag_delete_chain(m, NULL); + /* put in 0xDEADC0DE perhaps? */ + } else + M_SANITY_ACTION("m_tags on in-chain mbuf"); + } + + /* M_PKTHDR may only be set on first mbuf in chain */ + if (m != m0 && m->m_flags & M_PKTHDR) { + if (sanitize) { + bzero(&m->m_pkthdr, sizeof(m->m_pkthdr)); + m->m_flags &= ~M_PKTHDR; + /* put in 0xDEADCODE and leave hdr flag in */ + } else + M_SANITY_ACTION("M_PKTHDR on in-chain mbuf"); + } + } + m = m0; + if (pktlen && pktlen != m->m_pkthdr.len) { + if (sanitize) + m->m_pkthdr.len = 0; + else + M_SANITY_ACTION("m_pkthdr.len != mbuf chain length"); + } + return 1; + +#undef M_SANITY_ACTION +} + + +/* + * "Move" mbuf pkthdr from "from" to "to". + * "from" must have M_PKTHDR set, and "to" must be empty. + */ +void +m_move_pkthdr(struct mbuf *to, struct mbuf *from) +{ + +#if 0 + /* see below for why these are not enabled */ + M_ASSERTPKTHDR(to); + /* Note: with MAC, this may not be a good assertion. */ + KASSERT(SLIST_EMPTY(&to->m_pkthdr.tags), + ("m_move_pkthdr: to has tags")); +#endif +#ifdef MAC + /* + * XXXMAC: It could be this should also occur for non-MAC? + */ + if (to->m_flags & M_PKTHDR) + m_tag_delete_chain(to, NULL); +#endif + to->m_flags = (from->m_flags & M_COPYFLAGS) | (to->m_flags & M_EXT); + if ((to->m_flags & M_EXT) == 0) + to->m_data = to->m_pktdat; + to->m_pkthdr = from->m_pkthdr; /* especially tags */ + SLIST_INIT(&from->m_pkthdr.tags); /* purge tags from src */ + from->m_flags &= ~M_PKTHDR; +} + +/* + * Duplicate "from"'s mbuf pkthdr in "to". + * "from" must have M_PKTHDR set, and "to" must be empty. + * In particular, this does a deep copy of the packet tags. + */ +int +m_dup_pkthdr(struct mbuf *to, struct mbuf *from, int how) +{ + +#if 0 + /* + * The mbuf allocator only initializes the pkthdr + * when the mbuf is allocated with MGETHDR. Many users + * (e.g. m_copy*, m_prepend) use MGET and then + * smash the pkthdr as needed causing these + * assertions to trip. For now just disable them. + */ + M_ASSERTPKTHDR(to); + /* Note: with MAC, this may not be a good assertion. */ + KASSERT(SLIST_EMPTY(&to->m_pkthdr.tags), ("m_dup_pkthdr: to has tags")); +#endif + MBUF_CHECKSLEEP(how); +#ifdef MAC + if (to->m_flags & M_PKTHDR) + m_tag_delete_chain(to, NULL); +#endif + to->m_flags = (from->m_flags & M_COPYFLAGS) | (to->m_flags & M_EXT); + if ((to->m_flags & M_EXT) == 0) + to->m_data = to->m_pktdat; + to->m_pkthdr = from->m_pkthdr; + SLIST_INIT(&to->m_pkthdr.tags); + return (m_tag_copy_chain(to, from, MBTOM(how))); +} + +/* + * Lesser-used path for M_PREPEND: + * allocate new mbuf to prepend to chain, + * copy junk along. + */ +struct mbuf * +#ifndef VBOX +m_prepend(struct mbuf *m, int len, int how) +#else +m_prepend(PNATState pData, struct mbuf *m, int len, int how) +#endif +{ + struct mbuf *mn; + + if (m->m_flags & M_PKTHDR) + MGETHDR(mn, how, m->m_type); + else + MGET(mn, how, m->m_type); + if (mn == NULL) { +#ifndef VBOX + m_freem(m); +#else + m_freem(pData, m); +#endif + return (NULL); + } + if (m->m_flags & M_PKTHDR) + M_MOVE_PKTHDR(mn, m); + mn->m_next = m; + m = mn; + if(m->m_flags & M_PKTHDR) { + if (len < MHLEN) + MH_ALIGN(m, len); + } else { + if (len < MLEN) + M_ALIGN(m, len); + } + m->m_len = len; + return (m); +} + +/* + * Make a copy of an mbuf chain starting "off0" bytes from the beginning, + * continuing for "len" bytes. If len is M_COPYALL, copy to end of mbuf. + * The wait parameter is a choice of M_TRYWAIT/M_DONTWAIT from caller. + * Note that the copy is read-only, because clusters are not copied, + * only their reference counts are incremented. + */ +struct mbuf * +#ifndef VBOX +m_copym(struct mbuf *m, int off0, int len, int fWait) +#else +m_copym(PNATState pData, struct mbuf *m, int off0, int len, int fWait) +#endif +{ + struct mbuf *n, **np; + int off = off0; + struct mbuf *top; + int copyhdr = 0; + + KASSERT(off >= 0, ("m_copym, negative off %d", off)); + KASSERT(len >= 0, ("m_copym, negative len %d", len)); + MBUF_CHECKSLEEP(fWait); + if (off == 0 && m->m_flags & M_PKTHDR) + copyhdr = 1; + while (off > 0) { + KASSERT(m != NULL, ("m_copym, offset > size of mbuf chain")); + if (off < m->m_len) + break; + off -= m->m_len; + m = m->m_next; + } + np = ⊤ + top = 0; + while (len > 0) { + if (m == NULL) { + KASSERT(len == M_COPYALL, + ("m_copym, length > size of mbuf chain")); + break; + } + if (copyhdr) + MGETHDR(n, fWait, m->m_type); + else + MGET(n, fWait, m->m_type); + *np = n; + if (n == NULL) + goto nospace; + if (copyhdr) { + if (!m_dup_pkthdr(n, m, fWait)) + goto nospace; + if (len == M_COPYALL) + n->m_pkthdr.len -= off0; + else + n->m_pkthdr.len = len; + copyhdr = 0; + } + n->m_len = min(len, m->m_len - off); + if (m->m_flags & M_EXT) { + n->m_data = m->m_data + off; + mb_dupcl(n, m); + } else + bcopy(mtod(m, caddr_t)+off, mtod(n, caddr_t), + (u_int)n->m_len); + if (len != M_COPYALL) + len -= n->m_len; + off = 0; + m = m->m_next; + np = &n->m_next; + } + if (top == NULL) + mbstat.m_mcfail++; /* XXX: No consistency. */ + + return (top); +nospace: +#ifndef VBOX + m_freem(top); +#else + m_freem(pData, top); +#endif + mbstat.m_mcfail++; /* XXX: No consistency. */ + return (NULL); +} + +/* + * Returns mbuf chain with new head for the prepending case. + * Copies from mbuf (chain) n from off for len to mbuf (chain) m + * either prepending or appending the data. + * The resulting mbuf (chain) m is fully writeable. + * m is destination (is made writeable) + * n is source, off is offset in source, len is len from offset + * dir, 0 append, 1 prepend + * how, wait or nowait + */ + +static int +m_bcopyxxx(void *s, void *t, u_int len) +{ + bcopy(s, t, (size_t)len); + return 0; +} + +struct mbuf * +#ifndef VBOX +m_copymdata(struct mbuf *m, struct mbuf *n, int off, int len, + int prep, int how) +#else +m_copymdata(PNATState pData, struct mbuf *m, struct mbuf *n, int off, int len, + int prep, int how) +#endif +{ + struct mbuf *mm, *x, *z, *prev = NULL; + caddr_t p; + int i, nlen = 0; + caddr_t buf[MLEN]; + + KASSERT(m != NULL && n != NULL, ("m_copymdata, no target or source")); + KASSERT(off >= 0, ("m_copymdata, negative off %d", off)); + KASSERT(len >= 0, ("m_copymdata, negative len %d", len)); + KASSERT(prep == 0 || prep == 1, ("m_copymdata, unknown direction %d", prep)); + + mm = m; + if (!prep) { + while(mm->m_next) { + prev = mm; + mm = mm->m_next; + } + } + for (z = n; z != NULL; z = z->m_next) + nlen += z->m_len; + if (len == M_COPYALL) + len = nlen - off; + if (off + len > nlen || len < 1) + return NULL; + + if (!M_WRITABLE(mm)) { + /* XXX: Use proper m_xxx function instead. */ +#ifndef VBOX + x = m_getcl(how, MT_DATA, mm->m_flags); +#else + x = m_getcl(pData, how, MT_DATA, mm->m_flags); +#endif + if (x == NULL) + return NULL; + bcopy(mm->m_ext.ext_buf, x->m_ext.ext_buf, x->m_ext.ext_size); + p = x->m_ext.ext_buf + (mm->m_data - mm->m_ext.ext_buf); + x->m_data = p; + mm->m_next = NULL; + if (mm != m) + prev->m_next = x; +#ifndef VBOX + m_free(mm); +#else + m_free(pData, mm); +#endif + mm = x; + } + + /* + * Append/prepend the data. Allocating mbufs as necessary. + */ + /* Shortcut if enough free space in first/last mbuf. */ + if (!prep && M_TRAILINGSPACE(mm) >= len) { + m_apply(n, off, len, m_bcopyxxx, mtod(mm, caddr_t) + + mm->m_len); + mm->m_len += len; + mm->m_pkthdr.len += len; + return m; + } + if (prep && M_LEADINGSPACE(mm) >= len) { + mm->m_data = mtod(mm, caddr_t) - len; + m_apply(n, off, len, m_bcopyxxx, mtod(mm, caddr_t)); + mm->m_len += len; + mm->m_pkthdr.len += len; + return mm; + } + + /* Expand first/last mbuf to cluster if possible. */ + if (!prep && !(mm->m_flags & M_EXT) && len > M_TRAILINGSPACE(mm)) { + bcopy(mm->m_data, &buf, mm->m_len); +#ifndef VBOX + m_clget(mm, how); +#else + m_clget(pData, mm, how); +#endif + if (!(mm->m_flags & M_EXT)) + return NULL; + bcopy(&buf, mm->m_ext.ext_buf, mm->m_len); + mm->m_data = mm->m_ext.ext_buf; + mm->m_pkthdr.header = NULL; + } + if (prep && !(mm->m_flags & M_EXT) && len > M_LEADINGSPACE(mm)) { + bcopy(mm->m_data, &buf, mm->m_len); +#ifndef VBOX + m_clget(mm, how); +#else + m_clget(pData, mm, how); +#endif + if (!(mm->m_flags & M_EXT)) + return NULL; + bcopy(&buf, (caddr_t *)mm->m_ext.ext_buf + + mm->m_ext.ext_size - mm->m_len, mm->m_len); + mm->m_data = (caddr_t)mm->m_ext.ext_buf + + mm->m_ext.ext_size - mm->m_len; + mm->m_pkthdr.header = NULL; + } + + /* Append/prepend as many mbuf (clusters) as necessary to fit len. */ + if (!prep && len > M_TRAILINGSPACE(mm)) { + if (!m_getm(mm, len - M_TRAILINGSPACE(mm), how, MT_DATA)) + return NULL; + } + if (prep && len > M_LEADINGSPACE(mm)) { + if (!(z = m_getm(NULL, len - M_LEADINGSPACE(mm), how, MT_DATA))) + return NULL; + i = 0; + for (x = z; x != NULL; x = x->m_next) { + i += x->m_flags & M_EXT ? x->m_ext.ext_size : + (x->m_flags & M_PKTHDR ? MHLEN : MLEN); + if (!x->m_next) + break; + } + z->m_data += i - len; + m_move_pkthdr(mm, z); + x->m_next = mm; + mm = z; + } + + /* Seek to start position in source mbuf. Optimization for long chains. */ + while (off > 0) { + if (off < n->m_len) + break; + off -= n->m_len; + n = n->m_next; + } + + /* Copy data into target mbuf. */ + z = mm; + while (len > 0) { + KASSERT(z != NULL, ("m_copymdata, falling off target edge")); + i = M_TRAILINGSPACE(z); + m_apply(n, off, i, m_bcopyxxx, mtod(z, caddr_t) + z->m_len); + z->m_len += i; + /* fixup pkthdr.len if necessary */ + if ((prep ? mm : m)->m_flags & M_PKTHDR) + (prep ? mm : m)->m_pkthdr.len += i; + off += i; + len -= i; + z = z->m_next; + } + return (prep ? mm : m); +} + +/* + * Copy an entire packet, including header (which must be present). + * An optimization of the common case `m_copym(m, 0, M_COPYALL, how)'. + * Note that the copy is read-only, because clusters are not copied, + * only their reference counts are incremented. + * Preserve alignment of the first mbuf so if the creator has left + * some room at the beginning (e.g. for inserting protocol headers) + * the copies still have the room available. + */ +struct mbuf * +#ifndef VBOX +m_copypacket(struct mbuf *m, int how) +#else +m_copypacket(PNATState pData, struct mbuf *m, int how) +#endif +{ + struct mbuf *top, *n, *o; + + MBUF_CHECKSLEEP(how); + MGET(n, how, m->m_type); + top = n; + if (n == NULL) + goto nospace; + + if (!m_dup_pkthdr(n, m, how)) + goto nospace; + n->m_len = m->m_len; + if (m->m_flags & M_EXT) { + n->m_data = m->m_data; + mb_dupcl(n, m); + } else { + n->m_data = n->m_pktdat + (m->m_data - m->m_pktdat ); + bcopy(mtod(m, char *), mtod(n, char *), n->m_len); + } + + m = m->m_next; + while (m) { + MGET(o, how, m->m_type); + if (o == NULL) + goto nospace; + + n->m_next = o; + n = n->m_next; + + n->m_len = m->m_len; + if (m->m_flags & M_EXT) { + n->m_data = m->m_data; + mb_dupcl(n, m); + } else { + bcopy(mtod(m, char *), mtod(n, char *), n->m_len); + } + + m = m->m_next; + } + return top; +nospace: +#ifndef VBOX + m_freem(top); +#else + m_freem(pData, top); +#endif + mbstat.m_mcfail++; /* XXX: No consistency. */ + return (NULL); +} + +/* + * Copy data from an mbuf chain starting "off" bytes from the beginning, + * continuing for "len" bytes, into the indicated buffer. + */ +void +m_copydata(const struct mbuf *m, int off, int len, caddr_t cp) +{ + u_int count; + + KASSERT(off >= 0, ("m_copydata, negative off %d", off)); + KASSERT(len >= 0, ("m_copydata, negative len %d", len)); + while (off > 0) { + KASSERT(m != NULL, ("m_copydata, offset > size of mbuf chain")); + if (off < m->m_len) + break; + off -= m->m_len; + m = m->m_next; + } + while (len > 0) { + KASSERT(m != NULL, ("m_copydata, length > size of mbuf chain")); + count = min(m->m_len - off, len); + bcopy(mtod(m, caddr_t) + off, cp, count); + len -= count; + cp += count; + off = 0; + m = m->m_next; + } +} + +/* + * Copy a packet header mbuf chain into a completely new chain, including + * copying any mbuf clusters. Use this instead of m_copypacket() when + * you need a writable copy of an mbuf chain. + */ +struct mbuf * +#ifndef VBOX +m_dup(struct mbuf *m, int how) +#else +m_dup(PNATState pData, struct mbuf *m, int how) +#endif +{ + struct mbuf **p, *top = NULL; + int remain, moff, nsize; + + MBUF_CHECKSLEEP(how); + /* Sanity check */ + if (m == NULL) + return (NULL); + M_ASSERTPKTHDR(m); + + /* While there's more data, get a new mbuf, tack it on, and fill it */ + remain = m->m_pkthdr.len; + moff = 0; + p = ⊤ + while (remain > 0 || top == NULL) { /* allow m->m_pkthdr.len == 0 */ + struct mbuf *n; + + /* Get the next new mbuf */ + if (remain >= MINCLSIZE) { +#ifndef VBOX + n = m_getcl(how, m->m_type, 0); +#else + n = m_getcl(pData, how, m->m_type, 0); +#endif + nsize = MCLBYTES; + } else { +#ifndef VBOX + n = m_get(how, m->m_type); +#else + n = m_get(pData, how, m->m_type); +#endif + nsize = MLEN; + } + if (n == NULL) + goto nospace; + + if (top == NULL) { /* First one, must be PKTHDR */ + if (!m_dup_pkthdr(n, m, how)) { +#ifndef VBOX + m_free(n); +#else + m_free(pData, n); +#endif + goto nospace; + } + if ((n->m_flags & M_EXT) == 0) + nsize = MHLEN; + } + n->m_len = 0; + + /* Link it into the new chain */ + *p = n; + p = &n->m_next; + + /* Copy data from original mbuf(s) into new mbuf */ + while (n->m_len < nsize && m != NULL) { + int chunk = min(nsize - n->m_len, m->m_len - moff); + + bcopy(m->m_data + moff, n->m_data + n->m_len, chunk); + moff += chunk; + n->m_len += chunk; + remain -= chunk; + if (moff == m->m_len) { + m = m->m_next; + moff = 0; + } + } + + /* Check correct total mbuf length */ + KASSERT((remain > 0 && m != NULL) || (remain == 0 && m == NULL), + ("%s: bogus m_pkthdr.len", __func__)); + } + return (top); + +nospace: +#ifndef VBOX + m_freem(top); +#else + m_freem(pData, top); +#endif + mbstat.m_mcfail++; /* XXX: No consistency. */ + return (NULL); +} + +/* + * Concatenate mbuf chain n to m. + * Both chains must be of the same type (e.g. MT_DATA). + * Any m_pkthdr is not updated. + */ +void +#ifndef VBOX +m_cat(struct mbuf *m, struct mbuf *n) +#else +m_cat(PNATState pData, struct mbuf *m, struct mbuf *n) +#endif +{ + while (m->m_next) + m = m->m_next; + while (n) { + if (m->m_flags & M_EXT || + m->m_data + m->m_len + n->m_len >= &m->m_dat[MLEN]) { + /* just join the two chains */ + m->m_next = n; + return; + } + /* splat the data from one into the other */ + bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len, + (u_int)n->m_len); + m->m_len += n->m_len; +#ifndef VBOX + n = m_free(n); +#else + n = m_free(pData, n); +#endif + } +} + +void +#ifndef VBOX +m_adj(struct mbuf *mp, int req_len) +#else +m_adj(PNATState pData, struct mbuf *mp, int req_len) +#endif +{ + int len = req_len; + struct mbuf *m; + int count; + + if ((m = mp) == NULL) + return; + if (len >= 0) { + /* + * Trim from head. + */ + while (m != NULL && len > 0) { + if (m->m_len <= len) { + len -= m->m_len; + m->m_len = 0; + m = m->m_next; + } else { + m->m_len -= len; + m->m_data += len; + len = 0; + } + } + m = mp; + if (mp->m_flags & M_PKTHDR) + m->m_pkthdr.len -= (req_len - len); + } else { + /* + * Trim from tail. Scan the mbuf chain, + * calculating its length and finding the last mbuf. + * If the adjustment only affects this mbuf, then just + * adjust and return. Otherwise, rescan and truncate + * after the remaining size. + */ + len = -len; + count = 0; + for (;;) { + count += m->m_len; + if (m->m_next == (struct mbuf *)0) + break; + m = m->m_next; + } + if (m->m_len > len || (m->m_len == len && m == mp)) { + m->m_len -= len; + if (mp->m_flags & M_PKTHDR) + mp->m_pkthdr.len -= len; + return; + } + count -= len; + if (count < 0) + count = 0; + /* + * Correct length for chain is "count". + * Find the mbuf with last data, adjust its length, + * and toss data from remaining mbufs on chain. + */ + m = mp; + if (m->m_flags & M_PKTHDR) + m->m_pkthdr.len = count; + for (; m; m = m->m_next) { + if (m->m_len >= count) { + m->m_len = count; + if (m->m_next != NULL) { +#ifndef VBOX + m_freem(m->m_next); +#else + m_freem(pData, m->m_next); +#endif + m->m_next = NULL; + } + break; + } + count -= m->m_len; + } + } +} + +/* + * Rearange an mbuf chain so that len bytes are contiguous + * and in the data area of an mbuf (so that mtod and dtom + * will work for a structure of size len). Returns the resulting + * mbuf chain on success, frees it and returns null on failure. + * If there is room, it will add up to max_protohdr-len extra bytes to the + * contiguous region in an attempt to avoid being called next time. + */ +struct mbuf * +#ifndef VBOX +m_pullup(struct mbuf *n, int len) +#else +m_pullup(PNATState pData, struct mbuf *n, int len) +#endif +{ + struct mbuf *m; + int count; + int space; + + /* + * If first mbuf has no cluster, and has room for len bytes + * without shifting current data, pullup into it, + * otherwise allocate a new mbuf to prepend to the chain. + */ + if ((n->m_flags & M_EXT) == 0 && + n->m_data + len < &n->m_dat[MLEN] && n->m_next) { + if (n->m_len >= len) + return (n); + m = n; + n = n->m_next; + len -= m->m_len; + } else { + if (len > MHLEN) + goto bad; + MGET(m, M_DONTWAIT, n->m_type); + if (m == NULL) + goto bad; + m->m_len = 0; + if (n->m_flags & M_PKTHDR) + M_MOVE_PKTHDR(m, n); + } + space = &m->m_dat[MLEN] - (m->m_data + m->m_len); + do { + count = min(min(max(len, max_protohdr), space), n->m_len); + bcopy(mtod(n, caddr_t), mtod(m, caddr_t) + m->m_len, + (u_int)count); + len -= count; + m->m_len += count; + n->m_len -= count; + space -= count; + if (n->m_len) + n->m_data += count; + else +#ifndef VBOX + n = m_free(n); +#else + n = m_free(pData, n); +#endif + } while (len > 0 && n); + if (len > 0) { +#ifndef VBOX + (void) m_free(m); +#else + (void) m_free(pData, m); +#endif + goto bad; + } + m->m_next = n; + return (m); +bad: +#ifndef VBOX + m_freem(n); +#else + m_freem(pData, n); +#endif + mbstat.m_mpfail++; /* XXX: No consistency. */ + return (NULL); +} + +/* + * Like m_pullup(), except a new mbuf is always allocated, and we allow + * the amount of empty space before the data in the new mbuf to be specified + * (in the event that the caller expects to prepend later). + */ +int MSFail; + +struct mbuf * +#ifndef VBOX +m_copyup(struct mbuf *n, int len, int dstoff) +#else +m_copyup(PNATState pData, struct mbuf *n, int len, int dstoff) +#endif +{ + struct mbuf *m; + int count, space; + + if (len > (int)(MHLEN - dstoff)) + goto bad; + MGET(m, M_DONTWAIT, n->m_type); + if (m == NULL) + goto bad; + m->m_len = 0; + if (n->m_flags & M_PKTHDR) + M_MOVE_PKTHDR(m, n); + m->m_data += dstoff; + space = &m->m_dat[MLEN] - (m->m_data + m->m_len); + do { + count = min(min(max(len, max_protohdr), space), n->m_len); + memcpy(mtod(m, caddr_t) + m->m_len, mtod(n, caddr_t), + (unsigned)count); + len -= count; + m->m_len += count; + n->m_len -= count; + space -= count; + if (n->m_len) + n->m_data += count; + else +#ifndef VBOX + n = m_free(n); +#else + n = m_free(pData, n); +#endif + } while (len > 0 && n); + if (len > 0) { +#ifndef VBOX + (void) m_free(m); +#else + (void) m_free(pData, m); +#endif + goto bad; + } + m->m_next = n; + return (m); + bad: +#ifndef VBOX + m_freem(n); +#else + m_freem(pData, n); +#endif + MSFail++; + return (NULL); +} + +/* + * Partition an mbuf chain in two pieces, returning the tail -- + * all but the first len0 bytes. In case of failure, it returns NULL and + * attempts to restore the chain to its original state. + * + * Note that the resulting mbufs might be read-only, because the new + * mbuf can end up sharing an mbuf cluster with the original mbuf if + * the "breaking point" happens to lie within a cluster mbuf. Use the + * M_WRITABLE() macro to check for this case. + */ +struct mbuf * +#ifndef VBOX +m_split(struct mbuf *m0, int len0, int fWait) +#else +m_split(PNATState pData, struct mbuf *m0, int len0, int fWait) +#endif +{ + struct mbuf *m, *n; + u_int len = len0, remain; + + MBUF_CHECKSLEEP(fWait); + for (m = m0; m && len > m->m_len; m = m->m_next) + len -= m->m_len; + if (m == NULL) + return (NULL); + remain = m->m_len - len; + if (m0->m_flags & M_PKTHDR) { + MGETHDR(n, fWait, m0->m_type); + if (n == NULL) + return (NULL); + n->m_pkthdr.rcvif = m0->m_pkthdr.rcvif; + n->m_pkthdr.len = m0->m_pkthdr.len - len0; + m0->m_pkthdr.len = len0; + if (m->m_flags & M_EXT) + goto extpacket; + if (remain > MHLEN) { + /* m can't be the lead packet */ + MH_ALIGN(n, 0); +#ifndef VBOX + n->m_next = m_split(m, len, fWait); +#else + n->m_next = m_split(pData, m, len, fWait); +#endif + if (n->m_next == NULL) { +#ifndef VBOX + (void) m_free(n); +#else + (void) m_free(pData, n); +#endif + return (NULL); + } else { + n->m_len = 0; + return (n); + } + } else + MH_ALIGN(n, remain); + } else if (remain == 0) { + n = m->m_next; + m->m_next = NULL; + return (n); + } else { + MGET(n, fWait, m->m_type); + if (n == NULL) + return (NULL); + M_ALIGN(n, remain); + } +extpacket: + if (m->m_flags & M_EXT) { + n->m_data = m->m_data + len; + mb_dupcl(n, m); + } else { + bcopy(mtod(m, caddr_t) + len, mtod(n, caddr_t), remain); + } + n->m_len = remain; + m->m_len = len; + n->m_next = m->m_next; + m->m_next = NULL; + return (n); +} +/* + * Routine to copy from device local memory into mbufs. + * Note that `off' argument is offset into first mbuf of target chain from + * which to begin copying the data to. + */ +#ifndef VBOX +struct mbuf * +m_devget(char *buf, int totlen, int off, struct ifnet *ifp, + void (*copy)(char *from, caddr_t to, u_int len)) +{ + struct mbuf *m; + struct mbuf *top = NULL, **mp = ⊤ + int len; + + if (off < 0 || off > MHLEN) + return (NULL); + + while (totlen > 0) { + if (top == NULL) { /* First one, must be PKTHDR */ + if (totlen + off >= MINCLSIZE) { + m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR); + len = MCLBYTES; + } else { + m = m_gethdr(M_DONTWAIT, MT_DATA); + len = MHLEN; + + /* Place initial small packet/header at end of mbuf */ + if (m && totlen + off + max_linkhdr <= MLEN) { + m->m_data += max_linkhdr; + len -= max_linkhdr; + } + } + if (m == NULL) + return NULL; + m->m_pkthdr.rcvif = ifp; + m->m_pkthdr.len = totlen; + } else { + if (totlen + off >= MINCLSIZE) { + m = m_getcl(M_DONTWAIT, MT_DATA, 0); + len = MCLBYTES; + } else { + m = m_get(M_DONTWAIT, MT_DATA); + len = MLEN; + } + if (m == NULL) { + m_freem(top); + return NULL; + } + } + if (off) { + m->m_data += off; + len -= off; + off = 0; + } + m->m_len = len = min(totlen, len); + if (copy) + copy(buf, mtod(m, caddr_t), (u_int)len); + else + bcopy(buf, mtod(m, caddr_t), (u_int)len); + buf += len; + *mp = m; + mp = &m->m_next; + totlen -= len; + } + return (top); +} +#endif + +/* + * Copy data from a buffer back into the indicated mbuf chain, + * starting "off" bytes from the beginning, extending the mbuf + * chain if necessary. + */ +void +#ifndef VBOX +m_copyback(struct mbuf *m0, int off, int len, c_caddr_t cp) +#else +m_copyback(PNATState pData, struct mbuf *m0, int off, int len, c_caddr_t cp) +#endif +{ + int mlen; + struct mbuf *m = m0, *n; + int totlen = 0; + + if (m0 == NULL) + return; + while (off > (mlen = m->m_len)) { + off -= mlen; + totlen += mlen; + if (m->m_next == NULL) { +#ifndef VBOX + n = m_get(M_DONTWAIT, m->m_type); +#else + n = m_get(pData, M_DONTWAIT, m->m_type); +#endif + if (n == NULL) + goto out; + bzero(mtod(n, caddr_t), MLEN); + n->m_len = min(MLEN, len + off); + m->m_next = n; + } + m = m->m_next; + } + while (len > 0) { + if (m->m_next == NULL && (len > m->m_len - off)) { + m->m_len += min(len - (m->m_len - off), + M_TRAILINGSPACE(m)); + } + mlen = min (m->m_len - off, len); + bcopy(cp, off + mtod(m, caddr_t), (u_int)mlen); + cp += mlen; + len -= mlen; + mlen += off; + off = 0; + totlen += mlen; + if (len == 0) + break; + if (m->m_next == NULL) { +#ifndef VBOX + n = m_get(M_DONTWAIT, m->m_type); +#else + n = m_get(pData, M_DONTWAIT, m->m_type); +#endif + if (n == NULL) + break; + n->m_len = min(MLEN, len); + m->m_next = n; + } + m = m->m_next; + } +out: if (((m = m0)->m_flags & M_PKTHDR) && (m->m_pkthdr.len < totlen)) + m->m_pkthdr.len = totlen; +} + +/* + * Append the specified data to the indicated mbuf chain, + * Extend the mbuf chain if the new data does not fit in + * existing space. + * + * Return 1 if able to complete the job; otherwise 0. + */ +int +#ifndef VBOX +m_append(struct mbuf *m0, int len, c_caddr_t cp) +#else +m_append(PNATState pData, struct mbuf *m0, int len, c_caddr_t cp) +#endif +{ + struct mbuf *m, *n; + int remainder, space; + + for (m = m0; m->m_next != NULL; m = m->m_next) + ; + remainder = len; + space = M_TRAILINGSPACE(m); + if (space > 0) { + /* + * Copy into available space. + */ + if (space > remainder) + space = remainder; + bcopy(cp, mtod(m, caddr_t) + m->m_len, space); + m->m_len += space; + cp += space, remainder -= space; + } + while (remainder > 0) { + /* + * Allocate a new mbuf; could check space + * and allocate a cluster instead. + */ +#ifndef VBOX + n = m_get(M_DONTWAIT, m->m_type); +#else + n = m_get(pData, M_DONTWAIT, m->m_type); +#endif + if (n == NULL) + break; + n->m_len = min(MLEN, remainder); + bcopy(cp, mtod(n, caddr_t), n->m_len); + cp += n->m_len, remainder -= n->m_len; + m->m_next = n; + m = n; + } + if (m0->m_flags & M_PKTHDR) + m0->m_pkthdr.len += len - remainder; + return (remainder == 0); +} + +/* + * Apply function f to the data in an mbuf chain starting "off" bytes from + * the beginning, continuing for "len" bytes. + */ +int +m_apply(struct mbuf *m, int off, int len, + int (*f)(void *, void *, u_int), void *arg) +{ + u_int count; + int rval; + + KASSERT(off >= 0, ("m_apply, negative off %d", off)); + KASSERT(len >= 0, ("m_apply, negative len %d", len)); + while (off > 0) { + KASSERT(m != NULL, ("m_apply, offset > size of mbuf chain")); + if (off < m->m_len) + break; + off -= m->m_len; + m = m->m_next; + } + while (len > 0) { + KASSERT(m != NULL, ("m_apply, offset > size of mbuf chain")); + count = min(m->m_len - off, len); + rval = (*f)(arg, mtod(m, caddr_t) + off, count); + if (rval) + return (rval); + len -= count; + off = 0; + m = m->m_next; + } + return (0); +} + +/* + * Return a pointer to mbuf/offset of location in mbuf chain. + */ +struct mbuf * +m_getptr(struct mbuf *m, int loc, int *off) +{ + + while (loc >= 0) { + /* Normal end of search. */ + if (m->m_len > loc) { + *off = loc; + return (m); + } else { + loc -= m->m_len; + if (m->m_next == NULL) { + if (loc == 0) { + /* Point at the end of valid data. */ + *off = m->m_len; + return (m); + } + return (NULL); + } + m = m->m_next; + } + } + return (NULL); +} + +void +m_print(const struct mbuf *m, int maxlen) +{ + int len; + int pdata; + const struct mbuf *m2; + + if (m->m_flags & M_PKTHDR) + len = m->m_pkthdr.len; + else + len = -1; + m2 = m; + while (m2 != NULL && (len == -1 || len)) { + pdata = m2->m_len; + if (maxlen != -1 && pdata > maxlen) + pdata = maxlen; + printf("mbuf: %p len: %d, next: %p, %b%s", m2, m2->m_len, + m2->m_next, m2->m_flags, "\20\20freelist\17skipfw" + "\11proto5\10proto4\7proto3\6proto2\5proto1\4rdonly" + "\3eor\2pkthdr\1ext", pdata ? "" : "\n"); + if (pdata) + printf(", %*D\n", pdata, (u_char *)m2->m_data, "-"); + if (len != -1) + len -= m2->m_len; + m2 = m2->m_next; + } + if (len > 0) + printf("%d bytes unaccounted for.\n", len); + return; +} + +u_int +m_fixhdr(struct mbuf *m0) +{ + u_int len; + + len = m_length(m0, NULL); + m0->m_pkthdr.len = len; + return (len); +} + +u_int +m_length(struct mbuf *m0, struct mbuf **last) +{ + struct mbuf *m; + u_int len; + + len = 0; + for (m = m0; m != NULL; m = m->m_next) { + len += m->m_len; + if (m->m_next == NULL) + break; + } + if (last != NULL) + *last = m; + return (len); +} + +/* + * Defragment a mbuf chain, returning the shortest possible + * chain of mbufs and clusters. If allocation fails and + * this cannot be completed, NULL will be returned, but + * the passed in chain will be unchanged. Upon success, + * the original chain will be freed, and the new chain + * will be returned. + * + * If a non-packet header is passed in, the original + * mbuf (chain?) will be returned unharmed. + */ +struct mbuf * +#ifndef VBOX +m_defrag(struct mbuf *m0, int how) +#else +m_defrag(PNATState pData, struct mbuf *m0, int how) +#endif +{ + struct mbuf *m_new = NULL, *m_final = NULL; + int progress = 0, length; + + MBUF_CHECKSLEEP(how); + if (!(m0->m_flags & M_PKTHDR)) + return (m0); + + m_fixhdr(m0); /* Needed sanity check */ + +#ifdef MBUF_STRESS_TEST + if (m_defragrandomfailures) { + int temp = arc4random() & 0xff; + if (temp == 0xba) + goto nospace; + } +#endif + + if (m0->m_pkthdr.len > MHLEN) +#ifndef VBOX + m_final = m_getcl(how, MT_DATA, M_PKTHDR); +#else + m_final = m_getcl(pData, how, MT_DATA, M_PKTHDR); +#endif + else +#ifndef VBOX + m_final = m_gethdr(how, MT_DATA); +#else + m_final = m_gethdr(pData, how, MT_DATA); +#endif + + if (m_final == NULL) + goto nospace; + + if (m_dup_pkthdr(m_final, m0, how) == 0) + goto nospace; + + m_new = m_final; + + while (progress < m0->m_pkthdr.len) { + length = m0->m_pkthdr.len - progress; + if (length > MCLBYTES) + length = MCLBYTES; + + if (m_new == NULL) { + if (length > MLEN) +#ifndef VBOX + m_new = m_getcl(how, MT_DATA, 0); +#else + m_new = m_getcl(pData, how, MT_DATA, 0); +#endif + else +#ifndef VBOX + m_new = m_get(how, MT_DATA); +#else + m_new = m_get(pData, how, MT_DATA); +#endif + if (m_new == NULL) + goto nospace; + } + + m_copydata(m0, progress, length, mtod(m_new, caddr_t)); + progress += length; + m_new->m_len = length; + if (m_new != m_final) +#ifndef VBOX + m_cat(m_final, m_new); +#else + m_cat(pData, m_final, m_new); +#endif + m_new = NULL; + } +#ifdef MBUF_STRESS_TEST + if (m0->m_next == NULL) + m_defraguseless++; +#endif +#ifndef VBOX + m_freem(m0); +#else + m_freem(pData, m0); +#endif + m0 = m_final; +#ifdef MBUF_STRESS_TEST + m_defragpackets++; + m_defragbytes += m0->m_pkthdr.len; +#endif + return (m0); +nospace: +#ifdef MBUF_STRESS_TEST + m_defragfailure++; +#endif + if (m_final) +#ifndef VBOX + m_freem(m_final); +#else + m_freem(pData, m_final); +#endif + return (NULL); +} + +/* + * Defragment an mbuf chain, returning at most maxfrags separate + * mbufs+clusters. If this is not possible NULL is returned and + * the original mbuf chain is left in it's present (potentially + * modified) state. We use two techniques: collapsing consecutive + * mbufs and replacing consecutive mbufs by a cluster. + * + * NB: this should really be named m_defrag but that name is taken + */ +struct mbuf * +#ifndef VBOX +m_collapse(struct mbuf *m0, int how, int maxfrags) +#else +m_collapse(PNATState pData, struct mbuf *m0, int how, int maxfrags) +#endif +{ + struct mbuf *m, *n, *n2, **prev; + u_int curfrags; + + /* + * Calculate the current number of frags. + */ + curfrags = 0; + for (m = m0; m != NULL; m = m->m_next) + curfrags++; + /* + * First, try to collapse mbufs. Note that we always collapse + * towards the front so we don't need to deal with moving the + * pkthdr. This may be suboptimal if the first mbuf has much + * less data than the following. + */ + m = m0; +again: + for (;;) { + n = m->m_next; + if (n == NULL) + break; + if ((m->m_flags & M_RDONLY) == 0 && + n->m_len < M_TRAILINGSPACE(m)) { + bcopy(mtod(n, void *), mtod(m, char *) + m->m_len, + n->m_len); + m->m_len += n->m_len; + m->m_next = n->m_next; +#ifndef VBOX + m_free(n); +#else + m_free(pData, n); +#endif + if (--curfrags <= maxfrags) + return m0; + } else + m = n; + } + KASSERT(maxfrags > 1, + ("maxfrags %u, but normal collapse failed", maxfrags)); + /* + * Collapse consecutive mbufs to a cluster. + */ + prev = &m0->m_next; /* NB: not the first mbuf */ + while ((n = *prev) != NULL) { + if ((n2 = n->m_next) != NULL && + n->m_len + n2->m_len < MCLBYTES) { +#ifndef VBOX + m = m_getcl(how, MT_DATA, 0); +#else + m = m_getcl(pData, how, MT_DATA, 0); +#endif + if (m == NULL) + goto bad; + bcopy(mtod(n, void *), mtod(m, void *), n->m_len); + bcopy(mtod(n2, void *), mtod(m, char *) + n->m_len, + n2->m_len); + m->m_len = n->m_len + n2->m_len; + m->m_next = n2->m_next; + *prev = m; +#ifndef VBOX + m_free(n); + m_free(n2); +#else + m_free(pData, n); + m_free(pData, n2); +#endif + if (--curfrags <= maxfrags) /* +1 cl -2 mbufs */ + return m0; + /* + * Still not there, try the normal collapse + * again before we allocate another cluster. + */ + goto again; + } + prev = &n->m_next; + } + /* + * No place where we can collapse to a cluster; punt. + * This can occur if, for example, you request 2 frags + * but the packet requires that both be clusters (we + * never reallocate the first mbuf to avoid moving the + * packet header). + */ +bad: + return NULL; +} + +#ifdef MBUF_STRESS_TEST + +/* + * Fragment an mbuf chain. There's no reason you'd ever want to do + * this in normal usage, but it's great for stress testing various + * mbuf consumers. + * + * If fragmentation is not possible, the original chain will be + * returned. + * + * Possible length values: + * 0 no fragmentation will occur + * > 0 each fragment will be of the specified length + * -1 each fragment will be the same random value in length + * -2 each fragment's length will be entirely random + * (Random values range from 1 to 256) + */ +struct mbuf * +m_fragment(struct mbuf *m0, int how, int length) +{ + struct mbuf *m_new = NULL, *m_final = NULL; + int progress = 0; + + if (!(m0->m_flags & M_PKTHDR)) + return (m0); + + if ((length == 0) || (length < -2)) + return (m0); + + m_fixhdr(m0); /* Needed sanity check */ + + m_final = m_getcl(how, MT_DATA, M_PKTHDR); + + if (m_final == NULL) + goto nospace; + + if (m_dup_pkthdr(m_final, m0, how) == 0) + goto nospace; + + m_new = m_final; + + if (length == -1) + length = 1 + (arc4random() & 255); + + while (progress < m0->m_pkthdr.len) { + int fraglen; + + if (length > 0) + fraglen = length; + else + fraglen = 1 + (arc4random() & 255); + if (fraglen > m0->m_pkthdr.len - progress) + fraglen = m0->m_pkthdr.len - progress; + + if (fraglen > MCLBYTES) + fraglen = MCLBYTES; + + if (m_new == NULL) { + m_new = m_getcl(how, MT_DATA, 0); + if (m_new == NULL) + goto nospace; + } + + m_copydata(m0, progress, fraglen, mtod(m_new, caddr_t)); + progress += fraglen; + m_new->m_len = fraglen; + if (m_new != m_final) + m_cat(m_final, m_new); + m_new = NULL; + } + m_freem(m0); + m0 = m_final; + return (m0); +nospace: + if (m_final) + m_freem(m_final); + /* Return the original chain on failure */ + return (m0); +} + +#endif + +/* + * Copy the contents of uio into a properly sized mbuf chain. + */ +#ifndef VBOX +struct mbuf * +m_uiotombuf(struct uio *uio, int how, int len, int align, int flags) +{ + struct mbuf *m, *mb; + int error, length, total; + int progress = 0; + + /* + * len can be zero or an arbitrary large value bound by + * the total data supplied by the uio. + */ + if (len > 0) + total = min(uio->uio_resid, len); + else + total = uio->uio_resid; + + /* + * The smallest unit returned by m_getm2() is a single mbuf + * with pkthdr. We can't align past it. + */ + if (align >= MHLEN) + return (NULL); + + /* + * Give us the full allocation or nothing. + * If len is zero return the smallest empty mbuf. + */ + m = m_getm2(NULL, max(total + align, 1), how, MT_DATA, flags); + if (m == NULL) + return (NULL); + m->m_data += align; + + /* Fill all mbufs with uio data and update header information. */ + for (mb = m; mb != NULL; mb = mb->m_next) { + length = min(M_TRAILINGSPACE(mb), total - progress); + + error = uiomove(mtod(mb, void *), length, uio); + if (error) { + m_freem(m); + return (NULL); + } + + mb->m_len = length; + progress += length; + if (flags & M_PKTHDR) + m->m_pkthdr.len += length; + } + KASSERT(progress == total, ("%s: progress != total", __func__)); + + return (m); +} +#endif + +/* + * Set the m_data pointer of a newly-allocated mbuf + * to place an object of the specified size at the + * end of the mbuf, longword aligned. + */ +void +m_align(struct mbuf *m, int len) +{ + int adjust; + + if (m->m_flags & M_EXT) + adjust = m->m_ext.ext_size - len; + else if (m->m_flags & M_PKTHDR) + adjust = MHLEN - len; + else + adjust = MLEN - len; + m->m_data += adjust &~ (sizeof(long)-1); +} + +/* + * Create a writable copy of the mbuf chain. While doing this + * we compact the chain with a goal of producing a chain with + * at most two mbufs. The second mbuf in this chain is likely + * to be a cluster. The primary purpose of this work is to create + * a writable packet for encryption, compression, etc. The + * secondary goal is to linearize the data so the data can be + * passed to crypto hardware in the most efficient manner possible. + */ +struct mbuf * +#ifndef VBOX +m_unshare(struct mbuf *m0, int how) +#else +m_unshare(PNATState pData, struct mbuf *m0, int how) +#endif +{ + struct mbuf *m, *mprev; + struct mbuf *n, *mfirst, *mlast; + int len, off; + + mprev = NULL; + for (m = m0; m != NULL; m = mprev->m_next) { + /* + * Regular mbufs are ignored unless there's a cluster + * in front of it that we can use to coalesce. We do + * the latter mainly so later clusters can be coalesced + * also w/o having to handle them specially (i.e. convert + * mbuf+cluster -> cluster). This optimization is heavily + * influenced by the assumption that we're running over + * Ethernet where MCLBYTES is large enough that the max + * packet size will permit lots of coalescing into a + * single cluster. This in turn permits efficient + * crypto operations, especially when using hardware. + */ + if ((m->m_flags & M_EXT) == 0) { + if (mprev && (mprev->m_flags & M_EXT) && + m->m_len <= M_TRAILINGSPACE(mprev)) { + /* XXX: this ignores mbuf types */ + memcpy(mtod(mprev, caddr_t) + mprev->m_len, + mtod(m, caddr_t), m->m_len); + mprev->m_len += m->m_len; + mprev->m_next = m->m_next; /* unlink from chain */ +#ifndef VBOX + m_free(m); /* reclaim mbuf */ +#else + m_free(pData, m); /* reclaim mbuf */ +#endif +#if 0 + newipsecstat.ips_mbcoalesced++; +#endif + } else { + mprev = m; + } + continue; + } + /* + * Writable mbufs are left alone (for now). + */ + if (M_WRITABLE(m)) { + mprev = m; + continue; + } + + /* + * Not writable, replace with a copy or coalesce with + * the previous mbuf if possible (since we have to copy + * it anyway, we try to reduce the number of mbufs and + * clusters so that future work is easier). + */ + KASSERT(m->m_flags & M_EXT, ("m_flags 0x%x", m->m_flags)); + /* NB: we only coalesce into a cluster or larger */ + if (mprev != NULL && (mprev->m_flags & M_EXT) && + m->m_len <= M_TRAILINGSPACE(mprev)) { + /* XXX: this ignores mbuf types */ + memcpy(mtod(mprev, caddr_t) + mprev->m_len, + mtod(m, caddr_t), m->m_len); + mprev->m_len += m->m_len; + mprev->m_next = m->m_next; /* unlink from chain */ +#ifndef VBOX + m_free(m); /* reclaim mbuf */ +#else + m_free(pData, m); /* reclaim mbuf */ +#endif +#if 0 + newipsecstat.ips_clcoalesced++; +#endif + continue; + } + + /* + * Allocate new space to hold the copy... + */ + /* XXX why can M_PKTHDR be set past the first mbuf? */ + if (mprev == NULL && (m->m_flags & M_PKTHDR)) { + /* + * NB: if a packet header is present we must + * allocate the mbuf separately from any cluster + * because M_MOVE_PKTHDR will smash the data + * pointer and drop the M_EXT marker. + */ + MGETHDR(n, how, m->m_type); + if (n == NULL) { +#ifndef VBOX + m_freem(m0); +#else + m_freem(pData, m0); +#endif + return (NULL); + } + M_MOVE_PKTHDR(n, m); + MCLGET(n, how); + if ((n->m_flags & M_EXT) == 0) { +#ifndef VBOX + m_free(n); + m_freem(m0); +#else + m_free(pData, n); + m_freem(pData, m0); +#endif + return (NULL); + } + } else { +#ifndef VBOX + n = m_getcl(how, m->m_type, m->m_flags); +#else + n = m_getcl(pData, how, m->m_type, m->m_flags); +#endif + if (n == NULL) { +#ifndef VBOX + m_freem(m0); +#else + m_freem(pData, m0); +#endif + return (NULL); + } + } + /* + * ... and copy the data. We deal with jumbo mbufs + * (i.e. m_len > MCLBYTES) by splitting them into + * clusters. We could just malloc a buffer and make + * it external but too many device drivers don't know + * how to break up the non-contiguous memory when + * doing DMA. + */ + len = m->m_len; + off = 0; + mfirst = n; + mlast = NULL; + for (;;) { + int cc = min(len, MCLBYTES); + memcpy(mtod(n, caddr_t), mtod(m, caddr_t) + off, cc); + n->m_len = cc; + if (mlast != NULL) + mlast->m_next = n; + mlast = n; +#if 0 + newipsecstat.ips_clcopied++; +#endif + + len -= cc; + if (len <= 0) + break; + off += cc; + +#ifndef VBOX + n = m_getcl(how, m->m_type, m->m_flags); +#else + n = m_getcl(pData, how, m->m_type, m->m_flags); +#endif + if (n == NULL) { +#ifndef VBOX + m_freem(mfirst); + m_freem(m0); +#else + m_freem(pData, mfirst); + m_freem(pData, m0); +#endif + return (NULL); + } + } + n->m_next = m->m_next; + if (mprev == NULL) + m0 = mfirst; /* new head of chain */ + else + mprev->m_next = mfirst; /* replace old mbuf */ +#ifndef VBOX + m_free(m); /* release old mbuf */ +#else + m_free(pData, m); /* release old mbuf */ +#endif + mprev = mfirst; + } + return (m0); +} diff --git a/src/VBox/Devices/Network/slirp/bsd/kern/uipc_mbuf2.c b/src/VBox/Devices/Network/slirp/bsd/kern/uipc_mbuf2.c new file mode 100644 index 00000000..5f3abde8 --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/kern/uipc_mbuf2.c @@ -0,0 +1,539 @@ +/* $KAME: uipc_mbuf2.c,v 1.31 2001/11/28 11:08:53 itojun Exp $ */ +/* $NetBSD: uipc_mbuf.c,v 1.40 1999/04/01 00:23:25 thorpej Exp $ */ + +/*- + * Copyright (C) 1999 WIDE Project. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the project nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ +/*- + * Copyright (c) 1982, 1986, 1988, 1991, 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)uipc_mbuf.c 8.4 (Berkeley) 2/14/95 + */ +#ifndef VBOX +#include <sys/cdefs.h> +__FBSDID("$FreeBSD: src/sys/kern/uipc_mbuf2.c,v 1.33.8.1 2009/04/15 03:14:26 kensmith Exp $"); + +/*#define PULLDOWN_DEBUG*/ + +#include "opt_mac.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/kernel.h> +#include <sys/lock.h> +#include <sys/malloc.h> +#include <sys/mbuf.h> +#include <sys/mutex.h> + +#include <security/mac/mac_framework.h> + +static MALLOC_DEFINE(M_PACKET_TAGS, MBUF_TAG_MEM_NAME, + "packet-attached information"); +#else +# include "slirp.h" +#endif + +/* can't call it m_dup(), as freebsd[34] uses m_dup() with different arg */ +#ifndef VBOX +static struct mbuf *m_dup1(struct mbuf *, int, int, int); +#else +static struct mbuf *m_dup1(PNATState, struct mbuf *, int, int, int); +#endif + +/* + * ensure that [off, off + len) is contiguous on the mbuf chain "m". + * packet chain before "off" is kept untouched. + * if offp == NULL, the target will start at <retval, 0> on resulting chain. + * if offp != NULL, the target will start at <retval, *offp> on resulting chain. + * + * on error return (NULL return value), original "m" will be freed. + * + * XXX: M_TRAILINGSPACE/M_LEADINGSPACE only permitted on writable ext_buf. + */ +struct mbuf * +#ifndef VBOX +m_pulldown(struct mbuf *m, int off, int len, int *offp) +#else +m_pulldown(PNATState pData, struct mbuf *m, int off, int len, int *offp) +#endif +{ + struct mbuf *n, *o; + int hlen, tlen, olen; + int writable; + + /* check invalid arguments. */ + if (m == NULL) + panic("m == NULL in m_pulldown()"); + if (len > MCLBYTES) { +#ifndef VBOX + m_freem(m); +#else + m_freem(pData, m); +#endif + return NULL; /* impossible */ + } + +#ifdef PULLDOWN_DEBUG + { + struct mbuf *t; + printf("before:"); + for (t = m; t; t = t->m_next) + printf(" %d", t->m_len); + printf("\n"); + } +#endif + n = m; + while (n != NULL && off > 0) { + if (n->m_len > off) + break; + off -= n->m_len; + n = n->m_next; + } + /* be sure to point non-empty mbuf */ + while (n != NULL && n->m_len == 0) + n = n->m_next; + if (!n) { +#ifndef VBOX + m_freem(m); +#else + m_freem(pData, m); +#endif + return NULL; /* mbuf chain too short */ + } + + /* + * XXX: This code is flawed because it considers a "writable" mbuf + * data region to require all of the following: + * (i) mbuf _has_ to have M_EXT set; if it is just a regular + * mbuf, it is still not considered "writable." + * (ii) since mbuf has M_EXT, the ext_type _has_ to be + * EXT_CLUSTER. Anything else makes it non-writable. + * (iii) M_WRITABLE() must evaluate true. + * Ideally, the requirement should only be (iii). + * + * If we're writable, we're sure we're writable, because the ref. count + * cannot increase from 1, as that would require posession of mbuf + * n by someone else (which is impossible). However, if we're _not_ + * writable, we may eventually become writable )if the ref. count drops + * to 1), but we'll fail to notice it unless we re-evaluate + * M_WRITABLE(). For now, we only evaluate once at the beginning and + * live with this. + */ + /* + * XXX: This is dumb. If we're just a regular mbuf with no M_EXT, + * then we're not "writable," according to this code. + */ + writable = 0; + if ((n->m_flags & M_EXT) == 0 || + (n->m_ext.ext_type == EXT_CLUSTER && M_WRITABLE(n))) + writable = 1; + + /* + * the target data is on <n, off>. + * if we got enough data on the mbuf "n", we're done. + */ + if ((off == 0 || offp) && len <= n->m_len - off && writable) + goto ok; + + /* + * when len <= n->m_len - off and off != 0, it is a special case. + * len bytes from <n, off> sits in single mbuf, but the caller does + * not like the starting position (off). + * chop the current mbuf into two pieces, set off to 0. + */ + if (len <= n->m_len - off) { +#ifndef VBOX + o = m_dup1(n, off, n->m_len - off, M_DONTWAIT); +#else + o = m_dup1(pData, n, off, n->m_len - off, M_DONTWAIT); +#endif + if (o == NULL) { +#ifndef VBOX + m_freem(m); +#else + m_freem(pData, m); +#endif + return NULL; /* ENOBUFS */ + } + n->m_len = off; + o->m_next = n->m_next; + n->m_next = o; + n = n->m_next; + off = 0; + goto ok; + } + + /* + * we need to take hlen from <n, off> and tlen from <n->m_next, 0>, + * and construct contiguous mbuf with m_len == len. + * note that hlen + tlen == len, and tlen > 0. + */ + hlen = n->m_len - off; + tlen = len - hlen; + + /* + * ensure that we have enough trailing data on mbuf chain. + * if not, we can do nothing about the chain. + */ + olen = 0; + for (o = n->m_next; o != NULL; o = o->m_next) + olen += o->m_len; + if (hlen + olen < len) { +#ifndef VBOX + m_freem(m); +#else + m_freem(pData, m); +#endif + return NULL; /* mbuf chain too short */ + } + + /* + * easy cases first. + * we need to use m_copydata() to get data from <n->m_next, 0>. + */ + if ((off == 0 || offp) && M_TRAILINGSPACE(n) >= tlen + && writable) { + m_copydata(n->m_next, 0, tlen, mtod(n, caddr_t) + n->m_len); + n->m_len += tlen; +#ifndef VBOX + m_adj(n->m_next, tlen); +#else + m_adj(pData, n->m_next, tlen); +#endif + goto ok; + } + if ((off == 0 || offp) && M_LEADINGSPACE(n->m_next) >= hlen + && writable) { + n->m_next->m_data -= hlen; + n->m_next->m_len += hlen; + bcopy(mtod(n, caddr_t) + off, mtod(n->m_next, caddr_t), hlen); + n->m_len -= hlen; + n = n->m_next; + off = 0; + goto ok; + } + + /* + * now, we need to do the hard way. don't m_copy as there's no room + * on both end. + */ + if (len > MLEN) +#ifndef VBOX + o = m_getcl(M_DONTWAIT, m->m_type, 0); +#else + o = m_getcl(pData, M_DONTWAIT, m->m_type, 0); +#endif + else +#ifndef VBOX + o = m_get(M_DONTWAIT, m->m_type); +#else + o = m_get(pData, M_DONTWAIT, m->m_type); +#endif + if (!o) { +#ifndef VBOX + m_freem(m); +#else + m_freem(pData, m); +#endif + return NULL; /* ENOBUFS */ + } + /* get hlen from <n, off> into <o, 0> */ + o->m_len = hlen; + bcopy(mtod(n, caddr_t) + off, mtod(o, caddr_t), hlen); + n->m_len -= hlen; + /* get tlen from <n->m_next, 0> into <o, hlen> */ + m_copydata(n->m_next, 0, tlen, mtod(o, caddr_t) + o->m_len); + o->m_len += tlen; +#ifndef VBOX + m_adj(n->m_next, tlen); +#else + m_adj(pData, n->m_next, tlen); +#endif + o->m_next = n->m_next; + n->m_next = o; + n = o; + off = 0; + +ok: +#ifdef PULLDOWN_DEBUG + { + struct mbuf *t; + printf("after:"); + for (t = m; t; t = t->m_next) + printf("%c%d", t == n ? '*' : ' ', t->m_len); + printf(" (off=%d)\n", off); + } +#endif + if (offp) + *offp = off; + return n; +} + +static struct mbuf * +#ifndef VBOX +m_dup1(struct mbuf *m, int off, int len, int fWait) +#else +m_dup1(PNATState pData, struct mbuf *m, int off, int len, int fWait) +#endif +{ + struct mbuf *n; + int copyhdr; + + if (len > MCLBYTES) + return NULL; + if (off == 0 && (m->m_flags & M_PKTHDR) != 0) + copyhdr = 1; + else + copyhdr = 0; + if (len >= MINCLSIZE) { + if (copyhdr == 1) +#ifndef VBOX + n = m_getcl(fWait, m->m_type, M_PKTHDR); +#else + n = m_getcl(pData, fWait, m->m_type, M_PKTHDR); +#endif + else +#ifndef VBOX + n = m_getcl(fWait, m->m_type, 0); +#else + n = m_getcl(pData, fWait, m->m_type, 0); +#endif + } else { + if (copyhdr == 1) +#ifndef VBOX + n = m_gethdr(fWait, m->m_type); +#else + n = m_gethdr(pData, fWait, m->m_type); +#endif + else +#ifndef VBOX + n = m_get(fWait, m->m_type); +#else + n = m_get(pData, fWait, m->m_type); +#endif + } + if (!n) + return NULL; /* ENOBUFS */ + + if (copyhdr && !m_dup_pkthdr(n, m, fWait)) { +#ifndef VBOX + m_free(n); +#else + m_free(pData, n); +#endif + return NULL; + } + m_copydata(m, off, len, mtod(n, caddr_t)); + n->m_len = len; + return n; +} + +/* Free a packet tag. */ +void +m_tag_free_default(struct m_tag *t) +{ +#ifdef MAC + if (t->m_tag_id == PACKET_TAG_MACLABEL) + mac_destroy_mbuf_tag(t); +#endif +#ifndef VBOX + free(t, M_PACKET_TAGS); +#else + RTMemFree(t); +#endif +} + +/* Get a packet tag structure along with specified data following. */ +struct m_tag * +m_tag_alloc(u_int32_t cookie, int type, int len, int fWait) +{ + struct m_tag *t; + + MBUF_CHECKSLEEP(fWait); + if (len < 0) + return NULL; +#ifndef VBOX + t = malloc(len + sizeof(struct m_tag), M_PACKET_TAGS, fWait); +#else + NOREF(fWait); + t = RTMemAllocZ(len + sizeof(struct m_tag)); +#endif + if (t == NULL) + return NULL; + m_tag_setup(t, cookie, type, len); + t->m_tag_free = m_tag_free_default; + return t; +} + +/* Unlink and free a packet tag. */ +void +m_tag_delete(struct mbuf *m, struct m_tag *t) +{ + + KASSERT(m && t, ("m_tag_delete: null argument, m %p t %p", m, t)); + m_tag_unlink(m, t); + m_tag_free(t); +} + +/* Unlink and free a packet tag chain, starting from given tag. */ +void +m_tag_delete_chain(struct mbuf *m, struct m_tag *t) +{ + struct m_tag *p, *q; + + KASSERT(m, ("m_tag_delete_chain: null mbuf")); + if (t != NULL) + p = t; + else + p = SLIST_FIRST(&m->m_pkthdr.tags); + if (p == NULL) + return; + while ((q = SLIST_NEXT(p, m_tag_link)) != NULL) + m_tag_delete(m, q); + m_tag_delete(m, p); +} + +/* + * Strip off all tags that would normally vanish when + * passing through a network interface. Only persistent + * tags will exist after this; these are expected to remain + * so long as the mbuf chain exists, regardless of the + * path the mbufs take. + */ +void +m_tag_delete_nonpersistent(struct mbuf *m) +{ + struct m_tag *p, *q; + + SLIST_FOREACH_SAFE(p, &m->m_pkthdr.tags, m_tag_link, q) + if ((p->m_tag_id & MTAG_PERSISTENT) == 0) + m_tag_delete(m, p); +} + +/* Find a tag, starting from a given position. */ +struct m_tag * +m_tag_locate(struct mbuf *m, u_int32_t cookie, int type, struct m_tag *t) +{ + struct m_tag *p; + + KASSERT(m, ("m_tag_locate: null mbuf")); + if (t == NULL) + p = SLIST_FIRST(&m->m_pkthdr.tags); + else + p = SLIST_NEXT(t, m_tag_link); + while (p != NULL) { + if (p->m_tag_cookie == cookie && p->m_tag_id == type) + return p; + p = SLIST_NEXT(p, m_tag_link); + } + return NULL; +} + +/* Copy a single tag. */ +struct m_tag * +m_tag_copy(struct m_tag *t, int how) +{ + struct m_tag *p; + + MBUF_CHECKSLEEP(how); + KASSERT(t, ("m_tag_copy: null tag")); + p = m_tag_alloc(t->m_tag_cookie, t->m_tag_id, t->m_tag_len, how); + if (p == NULL) + return (NULL); +#ifdef MAC + /* + * XXXMAC: we should probably pass off the initialization, and + * copying here? can we hide that PACKET_TAG_MACLABEL is + * special from the mbuf code? + */ + if (t->m_tag_id == PACKET_TAG_MACLABEL) { + if (mac_init_mbuf_tag(p, how) != 0) { + m_tag_free(p); + return (NULL); + } + mac_copy_mbuf_tag(t, p); + } else +#endif + bcopy(t + 1, p + 1, t->m_tag_len); /* Copy the data */ + return p; +} + +/* + * Copy two tag chains. The destination mbuf (to) loses any attached + * tags even if the operation fails. This should not be a problem, as + * m_tag_copy_chain() is typically called with a newly-allocated + * destination mbuf. + */ +int +m_tag_copy_chain(struct mbuf *to, struct mbuf *from, int how) +{ + struct m_tag *p, *t, *tprev = NULL; + + MBUF_CHECKSLEEP(how); + KASSERT(to && from, + ("m_tag_copy_chain: null argument, to %p from %p", to, from)); + m_tag_delete_chain(to, NULL); + SLIST_FOREACH(p, &from->m_pkthdr.tags, m_tag_link) { + t = m_tag_copy(p, how); + if (t == NULL) { + m_tag_delete_chain(to, NULL); + return 0; + } + if (tprev == NULL) + SLIST_INSERT_HEAD(&to->m_pkthdr.tags, t, m_tag_link); + else + SLIST_INSERT_AFTER(tprev, t, m_tag_link); + tprev = t; + } + return 1; +} diff --git a/src/VBox/Devices/Network/slirp/bsd/sys/mbuf.h b/src/VBox/Devices/Network/slirp/bsd/sys/mbuf.h new file mode 100644 index 00000000..909f300b --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/sys/mbuf.h @@ -0,0 +1,1177 @@ +/*- + * Copyright (c) 1982, 1986, 1988, 1993 + * The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * @(#)mbuf.h 8.5 (Berkeley) 2/19/95 + * $FreeBSD: src/sys/sys/mbuf.h,v 1.217.2.3.4.1 2009/04/15 03:14:26 kensmith Exp $ + */ + +#ifndef _SYS_MBUF_H_ +#define _SYS_MBUF_H_ + +#ifndef VBOX +/* XXX: These includes suck. Sorry! */ +#include <sys/queue.h> +#ifdef _KERNEL +#include <sys/systm.h> +#include <vm/uma.h> +#ifdef WITNESS +#include <sys/lock.h> +#endif +#endif +#else /* VBOX */ +# include <VBox/param.h> +# include "misc.h" +# include "ext.h" + +typedef const char *c_caddr_t; + +DECL_NO_RETURN(static void) panic (char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + vbox_slirp_printV(fmt, args); + va_end(args); + AssertFatalFailed(); +} +/* for non-gnu compilers */ +# define __func__ RT_GCC_EXTENSION __FUNCTION__ +# ifndef __inline +# ifdef __GNUC__ +# define __inline __inline__ +# else +# define __inline +# endif +# endif + +# undef bzero +# define bzero(a1, len) memset((a1), 0, (len)) + +/* (vvl) some definitions from sys/param.h */ +/* + * Constants related to network buffer management. + * MCLBYTES must be no larger than HOST_PAGE_SIZE. + */ +# ifndef MSIZE +# define MSIZE 256 /* size of an mbuf */ +# endif /* MSIZE */ + +# ifndef MCLSHIFT +# define MCLSHIFT 11 /* convert bytes to mbuf clusters */ +# endif /* MCLSHIFT */ + +# ifndef MCLBYTES +# define MCLBYTES (1 << MCLSHIFT) /* size of an mbuf cluster */ +# endif /*MCLBYTES*/ + +# if defined(RT_ARCH_AMD64) || defined(RT_ARCH_X86) +# define MJUMPAGESIZE HOST_PAGE_SIZE /* jumbo cluster 4k */ +# else +# define MJUMPAGESIZE (4 * 1024) /* jumbo cluster 4k */ +# endif +# define MJUM9BYTES (9 * 1024) /* jumbo cluster 9k */ +# define MJUM16BYTES (16 * 1024) /* jumbo cluster 16k */ +#endif /* VBOX */ + +/* + * Mbufs are of a single size, MSIZE (sys/param.h), which includes overhead. + * An mbuf may add a single "mbuf cluster" of size MCLBYTES (also in + * sys/param.h), which has no additional overhead and is used instead of the + * internal data area; this is done when at least MINCLSIZE of data must be + * stored. Additionally, it is possible to allocate a separate buffer + * externally and attach it to the mbuf in a way similar to that of mbuf + * clusters. + */ +#define MLEN (MSIZE - sizeof(struct m_hdr)) /* normal data len */ +#define MHLEN (MLEN - sizeof(struct pkthdr)) /* data len w/pkthdr */ +#define MINCLSIZE (MHLEN + 1) /* smallest amount to put in cluster */ +#define M_MAXCOMPRESS (MHLEN / 2) /* max amount to copy for compression */ + +#if defined(_KERNEL) || defined(VBOX) +/*- + * Macros for type conversion: + * mtod(m, t) -- Convert mbuf pointer to data pointer of correct type. + * dtom(x) -- Convert data pointer within mbuf to mbuf pointer (XXX). + */ +#define mtod(m, t) ((t)((m)->m_data)) +#define dtom(x) ((struct mbuf *)((intptr_t)(x) & ~(MSIZE-1))) + +/* + * Argument structure passed to UMA routines during mbuf and packet + * allocations. + */ +struct mb_args { + int flags; /* Flags for mbuf being allocated */ + short type; /* Type of mbuf being allocated */ +}; +#endif /* _KERNEL */ + +#if defined(__LP64__) +#define M_HDR_PAD 6 +#else +#define M_HDR_PAD 2 +#endif + +/* + * Header present at the beginning of every mbuf. + */ +struct m_hdr { + struct mbuf *mh_next; /* next buffer in chain */ + struct mbuf *mh_nextpkt; /* next chain in queue/record */ + caddr_t mh_data; /* location of data */ + int mh_len; /* amount of data in this mbuf */ + int mh_flags; /* flags; see below */ + short mh_type; /* type of data in this mbuf */ +#ifdef VBOX + struct socket *mh_so; /*socket assotiated with mbuf*/ + TAILQ_ENTRY(mbuf) mh_ifq; +#endif + uint8_t pad[M_HDR_PAD];/* word align */ +}; + +/* + * Packet tag structure (see below for details). + */ +struct m_tag { + SLIST_ENTRY(m_tag) m_tag_link; /* List of packet tags */ + u_int16_t m_tag_id; /* Tag ID */ + u_int16_t m_tag_len; /* Length of data */ + u_int32_t m_tag_cookie; /* ABI/Module ID */ + void (*m_tag_free)(struct m_tag *); +}; + +/* + * Record/packet header in first mbuf of chain; valid only if M_PKTHDR is set. + */ +struct pkthdr { + struct ifnet *rcvif; /* rcv interface */ + /* variables for ip and tcp reassembly */ + void *header; /* pointer to packet header */ + int len; /* total packet length */ + /* variables for hardware checksum */ + int csum_flags; /* flags regarding checksum */ + int csum_data; /* data field used by csum routines */ + u_int16_t tso_segsz; /* TSO segment size */ + u_int16_t ether_vtag; /* Ethernet 802.1p+q vlan tag */ + SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */ +}; + +/* + * Description of external storage mapped into mbuf; valid only if M_EXT is + * set. + */ +struct m_ext { + caddr_t ext_buf; /* start of buffer */ + void (*ext_free) /* free routine if not the usual */ + (void *, void *); + void *ext_args; /* optional argument pointer */ + u_int ext_size; /* size of buffer, for ext_free */ +#ifdef VBOX + volatile uint32_t *ref_cnt; /* pointer to ref count info */ +#else + volatile u_int *ref_cnt; /* pointer to ref count info */ +#endif + int ext_type; /* type of external storage */ +}; + +/* + * The core of the mbuf object along with some shortcut defines for practical + * purposes. + */ +struct mbuf { + struct m_hdr m_hdr; + union { + struct { + struct pkthdr MH_pkthdr; /* M_PKTHDR set */ + union { + struct m_ext MH_ext; /* M_EXT set */ + char MH_databuf[MHLEN]; + } MH_dat; + } MH; + char M_databuf[MLEN]; /* !M_PKTHDR, !M_EXT */ + } M_dat; +}; +#define m_next m_hdr.mh_next +#define m_len m_hdr.mh_len +#define m_data m_hdr.mh_data +#define m_type m_hdr.mh_type +#define m_flags m_hdr.mh_flags +#define m_nextpkt m_hdr.mh_nextpkt +#define m_act m_nextpkt +#define m_pkthdr M_dat.MH.MH_pkthdr +#define m_ext M_dat.MH.MH_dat.MH_ext +#define m_pktdat M_dat.MH.MH_dat.MH_databuf +#define m_dat M_dat.M_databuf +#ifdef VBOX +# define m_so m_hdr.mh_so +# define ifq_so m_hdr.mh_so +# define m_ifq m_hdr.mh_ifq +#endif + +/* + * mbuf flags. + */ +#define M_EXT 0x00000001 /* has associated external storage */ +#define M_PKTHDR 0x00000002 /* start of record */ +#define M_EOR 0x00000004 /* end of record */ +#define M_RDONLY 0x00000008 /* associated data is marked read-only */ +#define M_PROTO1 0x00000010 /* protocol-specific */ +#define M_PROTO2 0x00000020 /* protocol-specific */ +#define M_PROTO3 0x00000040 /* protocol-specific */ +#define M_PROTO4 0x00000080 /* protocol-specific */ +#define M_PROTO5 0x00000100 /* protocol-specific */ +#define M_BCAST 0x00000200 /* send/received as link-level broadcast */ +#define M_MCAST 0x00000400 /* send/received as link-level multicast */ +#define M_FRAG 0x00000800 /* packet is a fragment of a larger packet */ +#define M_FIRSTFRAG 0x00001000 /* packet is first fragment */ +#define M_LASTFRAG 0x00002000 /* packet is last fragment */ +#define M_SKIP_FIREWALL 0x00004000 /* skip firewall processing */ +#define M_FREELIST 0x00008000 /* mbuf is on the free list */ +#define M_VLANTAG 0x00010000 /* ether_vtag is valid */ +#define M_PROMISC 0x00020000 /* packet was not for us */ +#define M_NOFREE 0x00040000 /* do not free mbuf, embedded in cluster */ +#define M_PROTO6 0x00080000 /* protocol-specific */ +#define M_PROTO7 0x00100000 /* protocol-specific */ +#define M_PROTO8 0x00200000 /* protocol-specific */ +/* + * For RELENG_{6,7} steal these flags for limited multiple routing table + * support. In RELENG_8 and beyond, use just one flag and a tag. + */ +#define M_FIB 0xF0000000 /* steal some bits to store fib number. */ + +#define M_NOTIFICATION M_PROTO5 /* SCTP notification */ + +/* + * Flags to purge when crossing layers. + */ +#define M_PROTOFLAGS \ + (M_PROTO1|M_PROTO2|M_PROTO3|M_PROTO4|M_PROTO5|M_PROTO6|M_PROTO7|M_PROTO8) + +/* + * Flags preserved when copying m_pkthdr. + */ +#define M_COPYFLAGS \ + (M_PKTHDR|M_EOR|M_RDONLY|M_PROTOFLAGS|M_SKIP_FIREWALL|M_BCAST|M_MCAST|\ + M_FRAG|M_FIRSTFRAG|M_LASTFRAG|M_VLANTAG|M_PROMISC|M_FIB) + +/* + * External buffer types: identify ext_buf type. + */ +#define EXT_CLUSTER 1 /* mbuf cluster */ +#define EXT_SFBUF 2 /* sendfile(2)'s sf_bufs */ +#define EXT_JUMBOP 3 /* jumbo cluster 4096 bytes */ +#define EXT_JUMBO9 4 /* jumbo cluster 9216 bytes */ +#define EXT_JUMBO16 5 /* jumbo cluster 16184 bytes */ +#define EXT_PACKET 6 /* mbuf+cluster from packet zone */ +#define EXT_MBUF 7 /* external mbuf reference (M_IOVEC) */ +#define EXT_NET_DRV 100 /* custom ext_buf provided by net driver(s) */ +#define EXT_MOD_TYPE 200 /* custom module's ext_buf type */ +#define EXT_DISPOSABLE 300 /* can throw this buffer away w/page flipping */ +#define EXT_EXTREF 400 /* has externally maintained ref_cnt ptr */ + +/* + * Flags indicating hw checksum support and sw checksum requirements. This + * field can be directly tested against if_data.ifi_hwassist. + */ +#define CSUM_IP 0x0001 /* will csum IP */ +#define CSUM_TCP 0x0002 /* will csum TCP */ +#define CSUM_UDP 0x0004 /* will csum UDP */ +#define CSUM_IP_FRAGS 0x0008 /* will csum IP fragments */ +#define CSUM_FRAGMENT 0x0010 /* will do IP fragmentation */ +#define CSUM_TSO 0x0020 /* will do TSO */ + +#define CSUM_IP_CHECKED 0x0100 /* did csum IP */ +#define CSUM_IP_VALID 0x0200 /* ... the csum is valid */ +#define CSUM_DATA_VALID 0x0400 /* csum_data field is valid */ +#define CSUM_PSEUDO_HDR 0x0800 /* csum_data has pseudo hdr */ + +#define CSUM_DELAY_DATA (CSUM_TCP | CSUM_UDP) +#define CSUM_DELAY_IP (CSUM_IP) /* XXX add ipv6 here too? */ + +/* + * mbuf types. + */ +#define MT_NOTMBUF 0 /* USED INTERNALLY ONLY! Object is not mbuf */ +#define MT_DATA 1 /* dynamic (data) allocation */ +#define MT_HEADER MT_DATA /* packet header, use M_PKTHDR instead */ +#define MT_SONAME 8 /* socket name */ +#define MT_CONTROL 14 /* extra-data protocol message */ +#define MT_OOBDATA 15 /* expedited data */ +#define MT_NTYPES 16 /* number of mbuf types for mbtypes[] */ + +#define MT_NOINIT 255 /* Not a type but a flag to allocate + a non-initialized mbuf */ + +#define MB_NOTAGS 0x1UL /* no tags attached to mbuf */ + +/* + * General mbuf allocator statistics structure. + * + * Many of these statistics are no longer used; we instead track many + * allocator statistics through UMA's built in statistics mechanism. + */ +struct mbstat { + u_long m_mbufs; /* XXX */ + u_long m_mclusts; /* XXX */ + + u_long m_drain; /* times drained protocols for space */ + u_long m_mcfail; /* XXX: times m_copym failed */ + u_long m_mpfail; /* XXX: times m_pullup failed */ + u_long m_msize; /* length of an mbuf */ + u_long m_mclbytes; /* length of an mbuf cluster */ + u_long m_minclsize; /* min length of data to allocate a cluster */ + u_long m_mlen; /* length of data in an mbuf */ + u_long m_mhlen; /* length of data in a header mbuf */ + + /* Number of mbtypes (gives # elems in mbtypes[] array) */ + short m_numtypes; + + /* XXX: Sendfile stats should eventually move to their own struct */ + u_long sf_iocnt; /* times sendfile had to do disk I/O */ + u_long sf_allocfail; /* times sfbuf allocation failed */ + u_long sf_allocwait; /* times sfbuf allocation had to wait */ +}; + +/* + * Flags specifying how an allocation should be made. + * + * The flag to use is as follows: + * - M_DONTWAIT or M_NOWAIT from an interrupt handler to not block allocation. + * - M_WAIT or M_WAITOK or M_TRYWAIT from wherever it is safe to block. + * + * M_DONTWAIT/M_NOWAIT means that we will not block the thread explicitly and + * if we cannot allocate immediately we may return NULL, whereas + * M_WAIT/M_WAITOK/M_TRYWAIT means that if we cannot allocate resources we + * will block until they are available, and thus never return NULL. + * + * XXX Eventually just phase this out to use M_WAITOK/M_NOWAIT. + */ +#define MBTOM(how) (how) +#ifndef VBOX +#define M_DONTWAIT M_NOWAIT +#define M_TRYWAIT M_WAITOK +#define M_WAIT M_WAITOK +#else +/* @todo (r=vvl) not sure we can do it in NAT */ +# define M_WAITOK 0 +# define M_NOWAIT 0 +# define M_DONTWAIT 0 +# define M_TRYWAI 0 +# define M_WAIT 0 +#endif + +/* + * String names of mbuf-related UMA(9) and malloc(9) types. Exposed to + * !_KERNEL so that monitoring tools can look up the zones with + * libmemstat(3). + */ +#define MBUF_MEM_NAME "mbuf" +#define MBUF_CLUSTER_MEM_NAME "mbuf_cluster" +#define MBUF_PACKET_MEM_NAME "mbuf_packet" +#define MBUF_JUMBOP_MEM_NAME "mbuf_jumbo_pagesize" +#define MBUF_JUMBO9_MEM_NAME "mbuf_jumbo_9k" +#define MBUF_JUMBO16_MEM_NAME "mbuf_jumbo_16k" +#define MBUF_TAG_MEM_NAME "mbuf_tag" +#define MBUF_EXTREFCNT_MEM_NAME "mbuf_ext_refcnt" + +#if defined(_KERNEL) || defined(VBOX) + +#ifdef WITNESS +#define MBUF_CHECKSLEEP(how) do { \ + if (how == M_WAITOK) \ + WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, \ + "Sleeping in \"%s\"", __func__); \ +} while (0) +#else +#define MBUF_CHECKSLEEP(how) +#endif + +/* + * Network buffer allocation API + * + * The rest of it is defined in kern/kern_mbuf.c + */ + +#ifndef VBOX +extern uma_zone_t zone_mbuf; +extern uma_zone_t zone_clust; +extern uma_zone_t zone_pack; +extern uma_zone_t zone_jumbop; +extern uma_zone_t zone_jumbo9; +extern uma_zone_t zone_jumbo16; +extern uma_zone_t zone_ext_refcnt; +#endif + +#ifndef VBOX +static __inline struct mbuf *m_getcl(int how, short type, int flags); +static __inline struct mbuf *m_get(int how, short type); +static __inline struct mbuf *m_gethdr(int how, short type); +static __inline struct mbuf *m_getjcl(int how, short type, int flags, + int size); +static __inline struct mbuf *m_getclr(int how, short type); /* XXX */ +static __inline struct mbuf *m_free(struct mbuf *m); +static __inline void m_clget(struct mbuf *m, int how); +static __inline void *m_cljget(struct mbuf *m, int how, int size); +void mb_free_ext(struct mbuf *); +#else +static __inline struct mbuf *m_getcl(PNATState pData, int how, short type, int flags); +static __inline struct mbuf *m_get(PNATState pData, int how, short type); +static __inline struct mbuf *m_gethdr(PNATState pData, int how, short type); +static __inline struct mbuf *m_getjcl(PNATState pData, int how, + short type, int flags, int size); +static __inline struct mbuf *m_getclr(PNATState pData, int how, short type); /* XXX */ +static __inline struct mbuf *m_free(PNATState pData, struct mbuf *m); +static __inline void m_clget(PNATState pData, struct mbuf *m, int how); +static __inline void *m_cljget(PNATState pData, struct mbuf *m, int how, int size); +void mb_free_ext(PNATState, struct mbuf *); +#endif +static __inline void m_chtype(struct mbuf *m, short new_type); +static __inline struct mbuf *m_last(struct mbuf *m); + +static __inline int +m_gettype(int size) +{ + int type; + + switch (size) { + case MSIZE: + type = EXT_MBUF; + break; + case MCLBYTES: + type = EXT_CLUSTER; + break; +#if MJUMPAGESIZE != MCLBYTES + case MJUMPAGESIZE: + type = EXT_JUMBOP; + break; +#endif + case MJUM9BYTES: + type = EXT_JUMBO9; + break; + case MJUM16BYTES: + type = EXT_JUMBO16; + break; + default: + panic("%s: m_getjcl: invalid cluster size", __func__); + } + + return (type); +} + +static __inline uma_zone_t +#ifndef VBOX +m_getzone(int size) +#else +m_getzone(PNATState pData, int size) +#endif +{ + uma_zone_t zone; + + switch (size) { + case MSIZE: + zone = zone_mbuf; + break; + case MCLBYTES: + zone = zone_clust; + break; +#if MJUMPAGESIZE != MCLBYTES + case MJUMPAGESIZE: + zone = zone_jumbop; + break; +#endif + case MJUM9BYTES: + zone = zone_jumbo9; + break; + case MJUM16BYTES: + zone = zone_jumbo16; + break; + default: + panic("%s: m_getjcl: invalid cluster type", __func__); + } + + return (zone); +} + +static __inline struct mbuf * +#ifndef VBOX +m_get(int how, short type) +#else +m_get(PNATState pData, int how, short type) +#endif +{ + struct mb_args args; + + args.flags = 0; + args.type = type; + return ((struct mbuf *)(uma_zalloc_arg(zone_mbuf, &args, how))); +} + +/* + * XXX This should be deprecated, very little use. + */ +static __inline struct mbuf * +#ifndef VBOX +m_getclr(int how, short type) +#else +m_getclr(PNATState pData, int how, short type) +#endif +{ + struct mbuf *m; + struct mb_args args; + + args.flags = 0; + args.type = type; + m = uma_zalloc_arg(zone_mbuf, &args, how); + if (m != NULL) + bzero(m->m_data, MLEN); + return (m); +} + +static __inline struct mbuf * +#ifndef VBOX +m_gethdr(int how, short type) +#else +m_gethdr(PNATState pData, int how, short type) +#endif +{ + struct mb_args args; + + args.flags = M_PKTHDR; + args.type = type; + return ((struct mbuf *)(uma_zalloc_arg(zone_mbuf, &args, how))); +} + +static __inline struct mbuf * +#ifndef VBOX +m_getcl(int how, short type, int flags) +#else +m_getcl(PNATState pData, int how, short type, int flags) +#endif +{ + struct mb_args args; + + args.flags = flags; + args.type = type; + return ((struct mbuf *)(uma_zalloc_arg(zone_pack, &args, how))); +} + +/* + * m_getjcl() returns an mbuf with a cluster of the specified size attached. + * For size it takes MCLBYTES, MJUMPAGESIZE, MJUM9BYTES, MJUM16BYTES. + * + * XXX: This is rather large, should be real function maybe. + */ +static __inline struct mbuf * +#ifndef VBOX +m_getjcl(int how, short type, int flags, int size) +#else +m_getjcl(PNATState pData, int how, short type, int flags, int size) +#endif +{ + struct mb_args args; + struct mbuf *m, *n; + uma_zone_t zone; + + args.flags = flags; + args.type = type; + + m = uma_zalloc_arg(zone_mbuf, &args, how); + if (m == NULL) + return (NULL); + +#ifndef VBOX + zone = m_getzone(size); +#else + zone = m_getzone(pData, size); +#endif + n = uma_zalloc_arg(zone, m, how); + if (n == NULL) { + uma_zfree(zone_mbuf, m); + return (NULL); + } + return (m); +} + +#ifndef VBOX +static __inline void +m_free_fast(struct mbuf *m) +{ + KASSERT(SLIST_EMPTY(&m->m_pkthdr.tags), ("doing fast free of mbuf with tags")); + + uma_zfree_arg(zone_mbuf, m, (void *)MB_NOTAGS); +} +#else +static __inline void +m_free_fast(PNATState pData, struct mbuf *m) +{ + AssertMsg(SLIST_EMPTY(&m->m_pkthdr.tags), ("doing fast free of mbuf with tags")); + + uma_zfree_arg(zone_mbuf, m, (void *)(uintptr_t)MB_NOTAGS); +} +#endif + +static __inline struct mbuf * +#ifndef VBOX +m_free(struct mbuf *m) +#else +m_free(PNATState pData, struct mbuf *m) +#endif +{ + struct mbuf *n = m->m_next; + + if (m->m_flags & M_EXT) +#ifndef VBOX + mb_free_ext(m); +#else + mb_free_ext(pData, m); +#endif + else if ((m->m_flags & M_NOFREE) == 0) + uma_zfree(zone_mbuf, m); + return (n); +} + +static __inline void +#ifndef VBOX +m_clget(struct mbuf *m, int how) +#else +m_clget(PNATState pData, struct mbuf *m, int how) +#endif +{ + + if (m->m_flags & M_EXT) + printf("%s: %p mbuf already has cluster\n", __func__, m); + m->m_ext.ext_buf = (char *)NULL; + uma_zalloc_arg(zone_clust, m, how); + /* + * On a cluster allocation failure, drain the packet zone and retry, + * we might be able to loosen a few clusters up on the drain. + */ + if ((how & M_NOWAIT) && (m->m_ext.ext_buf == NULL)) { + zone_drain(zone_pack); + uma_zalloc_arg(zone_clust, m, how); + } +} + +/* + * m_cljget() is different from m_clget() as it can allocate clusters without + * attaching them to an mbuf. In that case the return value is the pointer + * to the cluster of the requested size. If an mbuf was specified, it gets + * the cluster attached to it and the return value can be safely ignored. + * For size it takes MCLBYTES, MJUMPAGESIZE, MJUM9BYTES, MJUM16BYTES. + */ +static __inline void * +#ifndef VBOX +m_cljget(struct mbuf *m, int how, int size) +#else +m_cljget(PNATState pData, struct mbuf *m, int how, int size) +#endif +{ + uma_zone_t zone; + + if (m && m->m_flags & M_EXT) + printf("%s: %p mbuf already has cluster\n", __func__, m); + if (m != NULL) + m->m_ext.ext_buf = NULL; + +#ifndef VBOX + zone = m_getzone(size); +#else + zone = m_getzone(pData, size); +#endif + return (uma_zalloc_arg(zone, m, how)); +} + +static __inline void +#ifndef VBOX +m_cljset(struct mbuf *m, void *cl, int type) +#else +m_cljset(PNATState pData, struct mbuf *m, void *cl, int type) +#endif +{ + uma_zone_t zone; + int size; + + switch (type) { + case EXT_CLUSTER: + size = MCLBYTES; + zone = zone_clust; + break; +#if MJUMPAGESIZE != MCLBYTES + case EXT_JUMBOP: + size = MJUMPAGESIZE; + zone = zone_jumbop; + break; +#endif + case EXT_JUMBO9: + size = MJUM9BYTES; + zone = zone_jumbo9; + break; + case EXT_JUMBO16: + size = MJUM16BYTES; + zone = zone_jumbo16; + break; + default: + panic("unknown cluster type"); + break; + } + + m->m_data = m->m_ext.ext_buf = cl; +#ifdef VBOX + m->m_ext.ext_free = (void (*)(void *, void *))0; + m->m_ext.ext_args = NULL; +#else + m->m_ext.ext_free = m->m_ext.ext_args = NULL; +#endif + m->m_ext.ext_size = size; + m->m_ext.ext_type = type; + m->m_ext.ref_cnt = uma_find_refcnt(zone, cl); + m->m_flags |= M_EXT; + +} + +static __inline void +m_chtype(struct mbuf *m, short new_type) +{ + + m->m_type = new_type; +} + +static __inline struct mbuf * +m_last(struct mbuf *m) +{ + + while (m->m_next) + m = m->m_next; + return (m); +} + +/* + * mbuf, cluster, and external object allocation macros (for compatibility + * purposes). + */ +#define M_MOVE_PKTHDR(to, from) m_move_pkthdr((to), (from)) +#ifndef VBOX +#define MGET(m, how, type) ((m) = m_get((how), (type))) +#define MGETHDR(m, how, type) ((m) = m_gethdr((how), (type))) +#define MCLGET(m, how) m_clget((m), (how)) +#define MEXTADD(m, buf, size, free, args, flags, type) \ + m_extadd((m), (caddr_t)(buf), (size), (free), (args), (flags), (type)) +#define m_getm(m, len, how, type) \ + m_getm2((m), (len), (how), (type), M_PKTHDR) +#else /*!VBOX*/ +#define MGET(m, how, type) ((m) = m_get(pData, (how), (type))) +#define MGETHDR(m, how, type) ((m) = m_gethdr(pData, (how), (type))) +#define MCLGET(m, how) m_clget(pData, (m), (how)) +#define MEXTADD(m, buf, size, free, args, flags, type) \ + m_extadd(pData, (m), (caddr_t)(buf), (size), (free), (args), (flags), (type)) +#define m_getm(m, len, how, type) \ + m_getm2(pData, (m), (len), (how), (type), M_PKTHDR) +#endif + +/* + * Evaluate TRUE if it's safe to write to the mbuf m's data region (this can + * be both the local data payload, or an external buffer area, depending on + * whether M_EXT is set). + */ +#define M_WRITABLE(m) (!((m)->m_flags & M_RDONLY) && \ + (!(((m)->m_flags & M_EXT)) || \ + (*((m)->m_ext.ref_cnt) == 1)) ) \ + +/* Check if the supplied mbuf has a packet header, or else panic. */ +#define M_ASSERTPKTHDR(m) \ + KASSERT(m != NULL && m->m_flags & M_PKTHDR, \ + ("%s: no mbuf packet header!", __func__)) + +/* + * Ensure that the supplied mbuf is a valid, non-free mbuf. + * + * XXX: Broken at the moment. Need some UMA magic to make it work again. + */ +#define M_ASSERTVALID(m) \ + KASSERT((((struct mbuf *)m)->m_flags & 0) == 0, \ + ("%s: attempted use of a free mbuf!", __func__)) + +/* + * Set the m_data pointer of a newly-allocated mbuf (m_get/MGET) to place an + * object of the specified size at the end of the mbuf, longword aligned. + */ +#define M_ALIGN(m, len) do { \ + KASSERT(!((m)->m_flags & (M_PKTHDR|M_EXT)), \ + ("%s: M_ALIGN not normal mbuf", __func__)); \ + KASSERT((m)->m_data == (m)->m_dat, \ + ("%s: M_ALIGN not a virgin mbuf", __func__)); \ + (m)->m_data += (MLEN - (len)) & ~(sizeof(long) - 1); \ +} while (0) + +/* + * As above, for mbufs allocated with m_gethdr/MGETHDR or initialized by + * M_DUP/MOVE_PKTHDR. + */ +#define MH_ALIGN(m, len) do { \ + KASSERT((m)->m_flags & M_PKTHDR && !((m)->m_flags & M_EXT), \ + ("%s: MH_ALIGN not PKTHDR mbuf", __func__)); \ + KASSERT((m)->m_data == (m)->m_pktdat, \ + ("%s: MH_ALIGN not a virgin mbuf", __func__)); \ + (m)->m_data += (MHLEN - (len)) & ~(sizeof(long) - 1); \ +} while (0) + +/* + * Compute the amount of space available before the current start of data in + * an mbuf. + * + * The M_WRITABLE() is a temporary, conservative safety measure: the burden + * of checking writability of the mbuf data area rests solely with the caller. + */ +#define M_LEADINGSPACE(m) \ + ((m)->m_flags & M_EXT ? \ + (M_WRITABLE(m) ? (m)->m_data - (m)->m_ext.ext_buf : 0): \ + (m)->m_flags & M_PKTHDR ? (m)->m_data - (m)->m_pktdat : \ + (m)->m_data - (m)->m_dat) + +/* + * Compute the amount of space available after the end of data in an mbuf. + * + * The M_WRITABLE() is a temporary, conservative safety measure: the burden + * of checking writability of the mbuf data area rests solely with the caller. + */ +#define M_TRAILINGSPACE(m) \ + ((m)->m_flags & M_EXT ? \ + (M_WRITABLE(m) ? (m)->m_ext.ext_buf + (m)->m_ext.ext_size \ + - ((m)->m_data + (m)->m_len) : 0) : \ + &(m)->m_dat[MLEN] - ((m)->m_data + (m)->m_len)) + +/* + * Arrange to prepend space of size plen to mbuf m. If a new mbuf must be + * allocated, how specifies whether to wait. If the allocation fails, the + * original mbuf chain is freed and m is set to NULL. + */ +#define M_PREPEND(m, plen, how) do { \ + struct mbuf **_mmp = &(m); \ + struct mbuf *_mm = *_mmp; \ + int _mplen = (plen); \ + int __mhow = (how); \ + \ + MBUF_CHECKSLEEP(how); \ + if (M_LEADINGSPACE(_mm) >= _mplen) { \ + _mm->m_data -= _mplen; \ + _mm->m_len += _mplen; \ + } else \ + _mm = m_prepend(_mm, _mplen, __mhow); \ + if (_mm != NULL && _mm->m_flags & M_PKTHDR) \ + _mm->m_pkthdr.len += _mplen; \ + *_mmp = _mm; \ +} while (0) + +/* + * Change mbuf to new type. This is a relatively expensive operation and + * should be avoided. + */ +#define MCHTYPE(m, t) m_chtype((m), (t)) + +/* Length to m_copy to copy all. */ +#define M_COPYALL 1000000000 + +/* Compatibility with 4.3. */ +#define m_copy(m, o, l) m_copym((m), (o), (l), M_DONTWAIT) + +extern int max_datalen; /* MHLEN - max_hdr */ +extern int max_hdr; /* Largest link + protocol header */ +extern int max_linkhdr; /* Largest link-level header */ +extern int max_protohdr; /* Largest protocol header */ +extern struct mbstat mbstat; /* General mbuf stats/infos */ +extern int nmbclusters; /* Maximum number of clusters */ + +struct uio; + +void m_align(struct mbuf *, int); +int m_apply(struct mbuf *, int, int, + int (*)(void *, void *, u_int), void *); +#ifndef VBOX +void m_adj(struct mbuf *, int); +int m_append(struct mbuf *, int, c_caddr_t); +struct mbuf *m_defrag(struct mbuf *, int); +struct mbuf *m_dup(struct mbuf *, int); +void m_cat(struct mbuf *, struct mbuf *); +struct mbuf *m_collapse(struct mbuf *, int, int); +void m_copyback(struct mbuf *, int, int, c_caddr_t); +struct mbuf *m_copym(struct mbuf *, int, int, int); +struct mbuf *m_copymdata(struct mbuf *, struct mbuf *, + int, int, int, int); +struct mbuf *m_copypacket(struct mbuf *, int); +struct mbuf *m_copyup(struct mbuf *n, int len, int dstoff); +void m_extadd(struct mbuf *, caddr_t, u_int, + void (*)(void *, void *), void *, int, int); +#else +void m_adj(PNATState, struct mbuf *, int); +int m_append(PNATState pData, struct mbuf *, int, c_caddr_t); +struct mbuf *m_defrag(PNATState, struct mbuf *, int); +struct mbuf *m_dup(PNATState, struct mbuf *, int); +void m_cat(PNATState, struct mbuf *, struct mbuf *); +struct mbuf *m_collapse(PNATState, struct mbuf *, int, int); +void m_copyback(PNATState, struct mbuf *, int, int, c_caddr_t); +struct mbuf *m_copym(PNATState, struct mbuf *, int, int, int); +struct mbuf *m_copymdata(PNATState, struct mbuf *, struct mbuf *, + int, int, int, int); +struct mbuf *m_copypacket(PNATState, struct mbuf *, int); +struct mbuf *m_copyup(PNATState, struct mbuf *n, int len, int dstoff); +void m_extadd(PNATState pData, struct mbuf *, caddr_t, u_int, + void (*)(void *, void *), void *, int, int); +#endif +void m_copydata(const struct mbuf *, int, int, caddr_t); +void m_copy_pkthdr(struct mbuf *, struct mbuf *); +void m_demote(struct mbuf *, int); +struct mbuf *m_devget(char *, int, int, struct ifnet *, + void (*)(char *, caddr_t, u_int)); +int m_dup_pkthdr(struct mbuf *, struct mbuf *, int); +u_int m_fixhdr(struct mbuf *); +struct mbuf *m_fragment(struct mbuf *, int, int); +#ifndef VBOX +void m_freem(struct mbuf *); +struct mbuf *m_getm2(struct mbuf *, int, int, short, int); +struct mbuf *m_prepend(struct mbuf *, int, int); +struct mbuf *m_pulldown(struct mbuf *, int, int, int *); +struct mbuf *m_pullup(struct mbuf *, int); +int m_sanity(struct mbuf *, int); +struct mbuf *m_split(struct mbuf *, int, int); +struct mbuf *m_unshare(struct mbuf *, int how); +#else +void m_freem(PNATState pData, struct mbuf *); +struct mbuf *m_getm2(PNATState pData, struct mbuf *, int, int, short, int); +struct mbuf *m_prepend(PNATState, struct mbuf *, int, int); +struct mbuf *m_pulldown(PNATState, struct mbuf *, int, int, int *); +struct mbuf *m_pullup(PNATState, struct mbuf *, int); +int m_sanity(PNATState, struct mbuf *, int); +struct mbuf *m_split(PNATState, struct mbuf *, int, int); +struct mbuf *m_unshare(PNATState, struct mbuf *, int how); +#endif +struct mbuf *m_getptr(struct mbuf *, int, int *); +u_int m_length(struct mbuf *, struct mbuf **); +void m_move_pkthdr(struct mbuf *, struct mbuf *); +void m_print(const struct mbuf *, int); +struct mbuf *m_uiotombuf(struct uio *, int, int, int, int); + +/*- + * Network packets may have annotations attached by affixing a list of + * "packet tags" to the pkthdr structure. Packet tags are dynamically + * allocated semi-opaque data structures that have a fixed header + * (struct m_tag) that specifies the size of the memory block and a + * <cookie,type> pair that identifies it. The cookie is a 32-bit unique + * unsigned value used to identify a module or ABI. By convention this value + * is chosen as the date+time that the module is created, expressed as the + * number of seconds since the epoch (e.g., using date -u +'%s'). The type + * value is an ABI/module-specific value that identifies a particular + * annotation and is private to the module. For compatibility with systems + * like OpenBSD that define packet tags w/o an ABI/module cookie, the value + * PACKET_ABI_COMPAT is used to implement m_tag_get and m_tag_find + * compatibility shim functions and several tag types are defined below. + * Users that do not require compatibility should use a private cookie value + * so that packet tag-related definitions can be maintained privately. + * + * Note that the packet tag returned by m_tag_alloc has the default memory + * alignment implemented by malloc. To reference private data one can use a + * construct like: + * + * struct m_tag *mtag = m_tag_alloc(...); + * struct foo *p = (struct foo *)(mtag+1); + * + * if the alignment of struct m_tag is sufficient for referencing members of + * struct foo. Otherwise it is necessary to embed struct m_tag within the + * private data structure to insure proper alignment; e.g., + * + * struct foo { + * struct m_tag tag; + * ... + * }; + * struct foo *p = (struct foo *) m_tag_alloc(...); + * struct m_tag *mtag = &p->tag; + */ + +/* + * Persistent tags stay with an mbuf until the mbuf is reclaimed. Otherwise + * tags are expected to ``vanish'' when they pass through a network + * interface. For most interfaces this happens normally as the tags are + * reclaimed when the mbuf is free'd. However in some special cases + * reclaiming must be done manually. An example is packets that pass through + * the loopback interface. Also, one must be careful to do this when + * ``turning around'' packets (e.g., icmp_reflect). + * + * To mark a tag persistent bit-or this flag in when defining the tag id. + * The tag will then be treated as described above. + */ +#define MTAG_PERSISTENT 0x800 + +#define PACKET_TAG_NONE 0 /* Nadda */ + +/* Packet tags for use with PACKET_ABI_COMPAT. */ +#define PACKET_TAG_IPSEC_IN_DONE 1 /* IPsec applied, in */ +#define PACKET_TAG_IPSEC_OUT_DONE 2 /* IPsec applied, out */ +#define PACKET_TAG_IPSEC_IN_CRYPTO_DONE 3 /* NIC IPsec crypto done */ +#define PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED 4 /* NIC IPsec crypto req'ed */ +#define PACKET_TAG_IPSEC_IN_COULD_DO_CRYPTO 5 /* NIC notifies IPsec */ +#define PACKET_TAG_IPSEC_PENDING_TDB 6 /* Reminder to do IPsec */ +#define PACKET_TAG_BRIDGE 7 /* Bridge processing done */ +#define PACKET_TAG_GIF 8 /* GIF processing done */ +#define PACKET_TAG_GRE 9 /* GRE processing done */ +#define PACKET_TAG_IN_PACKET_CHECKSUM 10 /* NIC checksumming done */ +#define PACKET_TAG_ENCAP 11 /* Encap. processing */ +#define PACKET_TAG_IPSEC_SOCKET 12 /* IPSEC socket ref */ +#define PACKET_TAG_IPSEC_HISTORY 13 /* IPSEC history */ +#define PACKET_TAG_IPV6_INPUT 14 /* IPV6 input processing */ +#define PACKET_TAG_DUMMYNET 15 /* dummynet info */ +#define PACKET_TAG_DIVERT 17 /* divert info */ +#define PACKET_TAG_IPFORWARD 18 /* ipforward info */ +#define PACKET_TAG_MACLABEL (19 | MTAG_PERSISTENT) /* MAC label */ +#define PACKET_TAG_PF 21 /* PF + ALTQ information */ +#define PACKET_TAG_RTSOCKFAM 25 /* rtsock sa family */ +#define PACKET_TAG_IPOPTIONS 27 /* Saved IP options */ +#define PACKET_TAG_CARP 28 /* CARP info */ +#ifdef VBOX +# define PACKET_TAG_ALIAS 0xab01 +# define PACKET_TAG_ETHER 0xab02 +# define PACKET_SERVICE 0xab03 +#endif + +/* Specific cookies and tags. */ + +/* Packet tag routines. */ +struct m_tag *m_tag_alloc(u_int32_t, int, int, int); +void m_tag_delete(struct mbuf *, struct m_tag *); +void m_tag_delete_chain(struct mbuf *, struct m_tag *); +void m_tag_free_default(struct m_tag *); +struct m_tag *m_tag_locate(struct mbuf *, u_int32_t, int, struct m_tag *); +struct m_tag *m_tag_copy(struct m_tag *, int); +int m_tag_copy_chain(struct mbuf *, struct mbuf *, int); +void m_tag_delete_nonpersistent(struct mbuf *); + +/* + * Initialize the list of tags associated with an mbuf. + */ +static __inline void +m_tag_init(struct mbuf *m) +{ + + SLIST_INIT(&m->m_pkthdr.tags); +} + +/* + * Set up the contents of a tag. Note that this does not fill in the free + * method; the caller is expected to do that. + * + * XXX probably should be called m_tag_init, but that was already taken. + */ +static __inline void +m_tag_setup(struct m_tag *t, u_int32_t cookie, int type, int len) +{ + + t->m_tag_id = type; + t->m_tag_len = len; + t->m_tag_cookie = cookie; +} + +/* + * Reclaim resources associated with a tag. + */ +static __inline void +m_tag_free(struct m_tag *t) +{ + + (*t->m_tag_free)(t); +} + +/* + * Return the first tag associated with an mbuf. + */ +static __inline struct m_tag * +m_tag_first(struct mbuf *m) +{ + + return (SLIST_FIRST(&m->m_pkthdr.tags)); +} + +/* + * Return the next tag in the list of tags associated with an mbuf. + */ +static __inline struct m_tag * +m_tag_next(struct mbuf *m, struct m_tag *t) +{ + NOREF(m); + return (SLIST_NEXT(t, m_tag_link)); +} + +/* + * Prepend a tag to the list of tags associated with an mbuf. + */ +static __inline void +m_tag_prepend(struct mbuf *m, struct m_tag *t) +{ + + SLIST_INSERT_HEAD(&m->m_pkthdr.tags, t, m_tag_link); +} + +/* + * Unlink a tag from the list of tags associated with an mbuf. + */ +static __inline void +m_tag_unlink(struct mbuf *m, struct m_tag *t) +{ + + SLIST_REMOVE(&m->m_pkthdr.tags, t, m_tag, m_tag_link); +} + +/* These are for OpenBSD compatibility. */ +#define MTAG_ABI_COMPAT 0 /* compatibility ABI */ + +static __inline struct m_tag * +m_tag_get(int type, int length, int fWait) +{ + return (m_tag_alloc(MTAG_ABI_COMPAT, type, length, fWait)); +} + +static __inline struct m_tag * +m_tag_find(struct mbuf *m, int type, struct m_tag *start) +{ + return (SLIST_EMPTY(&m->m_pkthdr.tags) ? (struct m_tag *)NULL : + m_tag_locate(m, MTAG_ABI_COMPAT, type, start)); +} + +/* XXX temporary FIB methods probably eventually use tags.*/ +#define M_FIBSHIFT 28 +#define M_FIBMASK 0x0F + +/* get the fib from an mbuf and if it is not set, return the default */ +#define M_GETFIB(_m) \ + ((((_m)->m_flags & M_FIB) >> M_FIBSHIFT) & M_FIBMASK) + +#define M_SETFIB(_m, _fib) do { \ + _m->m_flags &= ~M_FIB; \ + _m->m_flags |= (((_fib) << M_FIBSHIFT) & M_FIB); \ +} while (0) + +#endif /* _KERNEL */ + +#endif /* !_SYS_MBUF_H_ */ diff --git a/src/VBox/Devices/Network/slirp/bsd/sys/sbuf.h b/src/VBox/Devices/Network/slirp/bsd/sys/sbuf.h new file mode 100644 index 00000000..3f59c46f --- /dev/null +++ b/src/VBox/Devices/Network/slirp/bsd/sys/sbuf.h @@ -0,0 +1,95 @@ +/*- + * Copyright (c) 2000 Poul-Henning Kamp and Dag-Erling Coïdan Smørgrav + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: src/sys/sys/sbuf.h,v 1.14.18.1.2.1 2009/04/15 03:14:26 kensmith Exp $ + */ + +#ifndef _SYS_SBUF_H_ +#define _SYS_SBUF_H_ + +#ifndef VBOX +#include <sys/_types.h> +#else +# include <iprt/types.h> +#endif + +/* + * Structure definition + */ +struct sbuf { + char *s_buf; /* storage buffer */ + void *s_unused; /* binary compatibility. */ + int s_size; /* size of storage buffer */ + int s_len; /* current length of string */ +#define SBUF_FIXEDLEN 0x00000000 /* fixed length buffer (default) */ +#define SBUF_AUTOEXTEND 0x00000001 /* automatically extend buffer */ +#define SBUF_USRFLAGMSK 0x0000ffff /* mask of flags the user may specify */ +#define SBUF_DYNAMIC 0x00010000 /* s_buf must be freed */ +#define SBUF_FINISHED 0x00020000 /* set by sbuf_finish() */ +#define SBUF_OVERFLOWED 0x00040000 /* sbuf overflowed */ +#define SBUF_DYNSTRUCT 0x00080000 /* sbuf must be freed */ + int s_flags; /* flags */ +}; + +__BEGIN_DECLS +/* + * API functions + */ +struct sbuf *sbuf_new(struct sbuf *, char *, int, int); +#define sbuf_new_auto() \ + sbuf_new(NULL, NULL, 0, SBUF_AUTOEXTEND) +void sbuf_clear(struct sbuf *); +int sbuf_setpos(struct sbuf *, int); +int sbuf_bcat(struct sbuf *, const void *, size_t); +int sbuf_bcpy(struct sbuf *, const void *, size_t); +int sbuf_cat(struct sbuf *, const char *); +int sbuf_cpy(struct sbuf *, const char *); +#ifndef VBOX +int sbuf_printf(struct sbuf *, const char *, ...) __printflike(2, 3); +int sbuf_vprintf(struct sbuf *, const char *, __va_list) __printflike(2, 0); +#else +int sbuf_printf(struct sbuf *, const char *, ...); +int sbuf_vprintf(struct sbuf *, const char *, va_list); +#endif +int sbuf_putc(struct sbuf *, int); +int sbuf_trim(struct sbuf *); +int sbuf_overflowed(struct sbuf *); +void sbuf_finish(struct sbuf *); +char *sbuf_data(struct sbuf *); +int sbuf_len(struct sbuf *); +int sbuf_done(struct sbuf *); +void sbuf_delete(struct sbuf *); + +#ifdef _KERNEL +struct uio; +struct sbuf *sbuf_uionew(struct sbuf *, struct uio *, int *); +int sbuf_bcopyin(struct sbuf *, const void *, size_t); +int sbuf_copyin(struct sbuf *, const void *, size_t); +#endif +__END_DECLS + +#endif |