Aren
/
NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
NetBSD/sys/net/if_spppsubr.c

5258 lines
128 KiB
C
Raw Normal View History

Initialize the activity timestamp when opening a connection. Only idle- timeout connection that made it to phase NETWORK yet. (For drivers using the internal timeout mechanism; isdnd, that does the timeout handling for ISDN drivers, still needs to be fixed.) Thanks to Wolfgang Solfrank for finding this.
2002-01-14 10:39:14 +03:00
/* $NetBSD: if_spppsubr.c,v 1.40 2002/01/14 07:39:14 martin Exp $ */
Add NetBSD rcsid tags, and preserve old ones from i4b source
1999-04-04 10:57:03 +04:00
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/*
* Synchronous PPP/Cisco link level subroutines.
* Keepalive protocol implemented in both Cisco and PPP modes.
*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* Copyright (C) 1994-1996 Cronyx Engineering Ltd.
* Author: Serge Vakulenko, <vak@cronyx.ru>
*
* Heavily revamped to conform to RFC 1661.
* Copyright (C) 1997, Joerg Wunsch.
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
* RFC2472 IPv6CP support.
* Copyright (C) 2000, Jun-ichiro itojun Hagino <itojun@iijlab.net>.
*
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
* This software is distributed with NO WARRANTIES, not even the implied
* warranties for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
*
* Authors grant any other persons or organisations permission to use
* or modify this software as long as this message is kept with the software,
* all derivative works or modified versions.
*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* From: Version 2.4, Thu Apr 30 17:17:21 MSD 1997
*
* From: if_spppsubr.c,v 1.39 1998/04/04 13:26:03 phk Exp
*
Add NetBSD rcsid tags, and preserve old ones from i4b source
1999-04-04 10:57:03 +04:00
* From: Id: if_spppsubr.c,v 1.23 1999/02/23 14:47:50 hm Exp
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
add RCSIDs
2001-11-13 02:49:33 +03:00
#include <sys/cdefs.h>
Initialize the activity timestamp when opening a connection. Only idle- timeout connection that made it to phase NETWORK yet. (For drivers using the internal timeout mechanism; isdnd, that does the timeout handling for ISDN drivers, still needs to be fixed.) Thanks to Wolfgang Solfrank for finding this.
2002-01-14 10:39:14 +03:00
__KERNEL_RCSID(0, "$NetBSD: if_spppsubr.c,v 1.40 2002/01/14 07:39:14 martin Exp $");
add RCSIDs
2001-11-13 02:49:33 +03:00
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include "opt_inet.h"
#include "opt_ipx.h"
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include "opt_iso.h"
#include "opt_ns.h"
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <sys/param.h>
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
#include <sys/proc.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <sys/systm.h>
#include <sys/kernel.h>
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include <sys/sockio.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <sys/socket.h>
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include <sys/syslog.h>
#include <sys/malloc.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <sys/mbuf.h>
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
#include <sys/callout.h>
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include <sys/md5.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <net/if.h>
#include <net/netisr.h>
#include <net/if_types.h>
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include <net/route.h>
Merge if_spppsubr.c PPP protocol declarations list with the one found in ppp_defs.h, and have if_spppsubr.c include ppp_defs.h rather than duplicate its definitions. [This is a stopgap measure to clean up build lossage.]
2000-07-02 21:38:09 +04:00
#include <net/ppp_defs.h>
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include <machine/stdarg.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
fix compilation without INET
2000-10-02 07:53:47 +04:00
#ifdef INET
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#include <netinet/ip.h>
#include <netinet/tcp.h>
fix compilation without INET
2000-10-02 07:53:47 +04:00
#endif
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#include <net/ethertypes.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef IPX
#include <netipx/ipx.h>
#include <netipx/ipx_if.h>
#endif
#ifdef NS
#include <netns/ns.h>
#include <netns/ns_if.h>
#endif
#ifdef ISO
#include <netiso/argo_debug.h>
#include <netiso/iso.h>
#include <netiso/iso_var.h>
#include <netiso/iso_snpac.h>
#endif
#include <net/if_sppp.h>
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
#include <net/if_spppvar.h>
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
#define MAXALIVECNT 3 /* max. alive packets */
#define DEFAULT_MAX_AUTH_FAILURES 5 /* max. auth. failures */
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Interface flags that can be set in an ifconfig command.
*
* Setting link0 will make the link passive, i.e. it will be marked
* as being administrative openable, but won't be opened to begin
* with. Incoming calls will be answered, or subsequent calls with
* -link1 will cause the administrative open of the LCP layer.
*
* Setting link1 will cause the link to auto-dial only as packets
* arrive to be sent.
*
* Setting IFF_DEBUG will syslog the option negotiation and state
* transitions at level kern.debug. Note: all logs consistently look
* like
*
* <if-name><unit>: <proto-name> <additional info...>
*
* with <if-name><unit> being something like "bppp0", and <proto-name>
* being one of "lcp", "ipcp", "cisco", "chap", "pap", etc.
*/
#define IFF_PASSIVE IFF_LINK0 /* wait passively for connection */
#define IFF_AUTO IFF_LINK1 /* auto-dial on output */
#define CONF_REQ 1 /* PPP configure request */
#define CONF_ACK 2 /* PPP configure acknowledge */
#define CONF_NAK 3 /* PPP configure negative ack */
#define CONF_REJ 4 /* PPP configure reject */
#define TERM_REQ 5 /* PPP terminate request */
#define TERM_ACK 6 /* PPP terminate acknowledge */
#define CODE_REJ 7 /* PPP code reject */
#define PROTO_REJ 8 /* PPP protocol reject */
#define ECHO_REQ 9 /* PPP echo request */
#define ECHO_REPLY 10 /* PPP echo reply */
#define DISC_REQ 11 /* PPP discard request */
#define LCP_OPT_MRU 1 /* maximum receive unit */
#define LCP_OPT_ASYNC_MAP 2 /* async control character map */
#define LCP_OPT_AUTH_PROTO 3 /* authentication protocol */
#define LCP_OPT_QUAL_PROTO 4 /* quality protocol */
#define LCP_OPT_MAGIC 5 /* magic number */
#define LCP_OPT_RESERVED 6 /* reserved */
#define LCP_OPT_PROTO_COMP 7 /* protocol field compression */
#define LCP_OPT_ADDR_COMP 8 /* address/control field compression */
#define IPCP_OPT_ADDRESSES 1 /* both IP addresses; deprecated */
fix comment (s/IPv6/IP/)
2000-10-10 15:43:51 +04:00
#define IPCP_OPT_COMPRESSION 2 /* IP compression protocol */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#define IPCP_OPT_ADDRESS 3 /* local IP address */
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
#define IPCP_OPT_PRIMDNS 129 /* primary remote dns address */
#define IPCP_OPT_SECDNS 131 /* secondary remote dns address */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
#define IPV6CP_OPT_IFID 1 /* interface identifier */
#define IPV6CP_OPT_COMPRESSION 2 /* IPv6 compression protocol */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#define PAP_REQ 1 /* PAP name/password request */
#define PAP_ACK 2 /* PAP acknowledge */
#define PAP_NAK 3 /* PAP fail */
#define CHAP_CHALLENGE 1 /* CHAP challenge request */
#define CHAP_RESPONSE 2 /* CHAP challenge response */
#define CHAP_SUCCESS 3 /* CHAP response ok */
#define CHAP_FAILURE 4 /* CHAP response failed */
#define CHAP_MD5 5 /* hash algorithm - MD5 */
#define CISCO_MULTICAST 0x8f /* Cisco multicast address */
#define CISCO_UNICAST 0x0f /* Cisco unicast address */
#define CISCO_KEEPALIVE 0x8035 /* Cisco keepalive protocol */
#define CISCO_ADDR_REQ 0 /* Cisco address request */
#define CISCO_ADDR_REPLY 1 /* Cisco address reply */
#define CISCO_KEEPALIVE_REQ 2 /* Cisco keepalive request */
/* states are named and numbered according to RFC 1661 */
#define STATE_INITIAL 0
#define STATE_STARTING 1
#define STATE_CLOSED 2
#define STATE_STOPPED 3
#define STATE_CLOSING 4
#define STATE_STOPPING 5
#define STATE_REQ_SENT 6
#define STATE_ACK_RCVD 7
#define STATE_ACK_SENT 8
#define STATE_OPENED 9
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct ppp_header {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
u_char address;
u_char control;
u_short protocol;
Add the `packed' attribute to structures which describe wire protocol data formats.
1999-11-19 23:41:19 +03:00
} __attribute__((__packed__));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#define PPP_HEADER_LEN sizeof (struct ppp_header)
struct lcp_header {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
u_char type;
u_char ident;
u_short len;
Add the `packed' attribute to structures which describe wire protocol data formats.
1999-11-19 23:41:19 +03:00
} __attribute__((__packed__));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#define LCP_HEADER_LEN sizeof (struct lcp_header)
struct cisco_packet {
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t type;
u_int32_t par1;
u_int32_t par2;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
u_short rel;
u_short time0;
u_short time1;
Add the `packed' attribute to structures which describe wire protocol data formats.
1999-11-19 23:41:19 +03:00
} __attribute__((__packed__));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#define CISCO_PACKET_LEN 18
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* We follow the spelling and capitalization of RFC 1661 here, to make
* it easier comparing with the standard. Please refer to this RFC in
* case you can't make sense out of these abbreviation; it will also
* explain the semantics related to the various events and actions.
*/
struct cp {
u_short proto; /* PPP control protocol number */
u_char protoidx; /* index into state table in struct sppp */
u_char flags;
#define CP_LCP 0x01 /* this is the LCP */
#define CP_AUTH 0x02 /* this is an authentication protocol */
#define CP_NCP 0x04 /* this is a NCP */
#define CP_QUAL 0x08 /* this is a quality reporting protocol */
const char *name; /* name of this control protocol */
/* event handlers */
void (*Up)(struct sppp *sp);
void (*Down)(struct sppp *sp);
void (*Open)(struct sppp *sp);
void (*Close)(struct sppp *sp);
void (*TO)(void *sp);
int (*RCR)(struct sppp *sp, struct lcp_header *h, int len);
void (*RCN_rej)(struct sppp *sp, struct lcp_header *h, int len);
void (*RCN_nak)(struct sppp *sp, struct lcp_header *h, int len);
/* actions */
void (*tlu)(struct sppp *sp);
void (*tld)(struct sppp *sp);
void (*tls)(struct sppp *sp);
void (*tlf)(struct sppp *sp);
void (*scr)(struct sppp *sp);
};
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
static struct sppp *spppq;
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
static struct callout keepalive_ch;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
#ifdef __FreeBSD__
#define SPP_FMT "%s%d: "
#define SPP_ARGS(ifp) (ifp)->if_name, (ifp)->if_unit
#else
#define SPP_FMT "%s: "
#define SPP_ARGS(ifp) (ifp)->if_xname
#endif
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
fix compilation without INET
2000-10-02 07:53:47 +04:00
#ifdef INET
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/*
* The following disgusting hack gets around the problem that IP TOS
* can't be set yet. We want to put "interactive" traffic on a high
* priority queue. To decide if traffic is interactive, we check that
* a) it is TCP and b) one of its ports is telnet, rlogin or ftp control.
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
*
* XXX is this really still necessary? - joerg -
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static u_short interactive_ports[8] = {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
0, 513, 0, 0,
0, 21, 0, 23,
};
#define INTERACTIVE(p) (interactive_ports[(p) & 7] == (p))
fix compilation without INET
2000-10-02 07:53:47 +04:00
#endif
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* almost every function needs these */
#define STDDCL \
struct ifnet *ifp = &sp->pp_if; \
int debug = ifp->if_flags & IFF_DEBUG
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static int sppp_output(struct ifnet *ifp, struct mbuf *m,
struct sockaddr *dst, struct rtentry *rt);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
code cleanup for portability
2001-12-05 00:32:15 +03:00
static void sppp_cisco_send(struct sppp *sp, int type, int32_t par1, int32_t par2);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static void sppp_cisco_input(struct sppp *sp, struct mbuf *m);
static void sppp_cp_input(const struct cp *cp, struct sppp *sp,
struct mbuf *m);
static void sppp_cp_send(struct sppp *sp, u_short proto, u_char type,
u_char ident, u_short len, void *data);
/* static void sppp_cp_timeout(void *arg); */
static void sppp_cp_change_state(const struct cp *cp, struct sppp *sp,
int newstate);
static void sppp_auth_send(const struct cp *cp,
struct sppp *sp, unsigned int type, unsigned int id,
...);
static void sppp_up_event(const struct cp *cp, struct sppp *sp);
static void sppp_down_event(const struct cp *cp, struct sppp *sp);
static void sppp_open_event(const struct cp *cp, struct sppp *sp);
static void sppp_close_event(const struct cp *cp, struct sppp *sp);
static void sppp_to_event(const struct cp *cp, struct sppp *sp);
static void sppp_null(struct sppp *sp);
static void sppp_lcp_init(struct sppp *sp);
static void sppp_lcp_up(struct sppp *sp);
static void sppp_lcp_down(struct sppp *sp);
static void sppp_lcp_open(struct sppp *sp);
static void sppp_lcp_close(struct sppp *sp);
static void sppp_lcp_TO(void *sp);
static int sppp_lcp_RCR(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_lcp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_lcp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_lcp_tlu(struct sppp *sp);
static void sppp_lcp_tld(struct sppp *sp);
static void sppp_lcp_tls(struct sppp *sp);
static void sppp_lcp_tlf(struct sppp *sp);
static void sppp_lcp_scr(struct sppp *sp);
static void sppp_lcp_check_and_close(struct sppp *sp);
static int sppp_ncp_check(struct sppp *sp);
static void sppp_ipcp_init(struct sppp *sp);
static void sppp_ipcp_up(struct sppp *sp);
static void sppp_ipcp_down(struct sppp *sp);
static void sppp_ipcp_open(struct sppp *sp);
static void sppp_ipcp_close(struct sppp *sp);
static void sppp_ipcp_TO(void *sp);
static int sppp_ipcp_RCR(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_ipcp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_ipcp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_ipcp_tlu(struct sppp *sp);
static void sppp_ipcp_tld(struct sppp *sp);
static void sppp_ipcp_tls(struct sppp *sp);
static void sppp_ipcp_tlf(struct sppp *sp);
static void sppp_ipcp_scr(struct sppp *sp);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
static void sppp_ipv6cp_init(struct sppp *sp);
static void sppp_ipv6cp_up(struct sppp *sp);
static void sppp_ipv6cp_down(struct sppp *sp);
static void sppp_ipv6cp_open(struct sppp *sp);
static void sppp_ipv6cp_close(struct sppp *sp);
static void sppp_ipv6cp_TO(void *sp);
static int sppp_ipv6cp_RCR(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_ipv6cp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_ipv6cp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len);
static void sppp_ipv6cp_tlu(struct sppp *sp);
static void sppp_ipv6cp_tld(struct sppp *sp);
static void sppp_ipv6cp_tls(struct sppp *sp);
static void sppp_ipv6cp_tlf(struct sppp *sp);
static void sppp_ipv6cp_scr(struct sppp *sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static void sppp_pap_input(struct sppp *sp, struct mbuf *m);
static void sppp_pap_init(struct sppp *sp);
static void sppp_pap_open(struct sppp *sp);
static void sppp_pap_close(struct sppp *sp);
static void sppp_pap_TO(void *sp);
static void sppp_pap_my_TO(void *sp);
static void sppp_pap_tlu(struct sppp *sp);
static void sppp_pap_tld(struct sppp *sp);
static void sppp_pap_scr(struct sppp *sp);
static void sppp_chap_input(struct sppp *sp, struct mbuf *m);
static void sppp_chap_init(struct sppp *sp);
static void sppp_chap_open(struct sppp *sp);
static void sppp_chap_close(struct sppp *sp);
static void sppp_chap_TO(void *sp);
static void sppp_chap_tlu(struct sppp *sp);
static void sppp_chap_tld(struct sppp *sp);
static void sppp_chap_scr(struct sppp *sp);
static const char *sppp_auth_type_name(u_short proto, u_char type);
static const char *sppp_cp_type_name(u_char type);
code cleanup for portability
2001-12-05 00:32:15 +03:00
static const char *sppp_dotted_quad(u_int32_t addr);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static const char *sppp_ipcp_opt_name(u_char opt);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
#ifdef INET6
static const char *sppp_ipv6cp_opt_name(u_char opt);
#endif
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static const char *sppp_lcp_opt_name(u_char opt);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
static const char *sppp_phase_name(int phase);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static const char *sppp_proto_name(u_short proto);
static const char *sppp_state_name(int state);
static int sppp_params(struct sppp *sp, int cmd, void *data);
code cleanup for portability
2001-12-05 00:32:15 +03:00
static void sppp_get_ip_addrs(struct sppp *sp, u_int32_t *src, u_int32_t *dst,
u_int32_t *srcmask);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static void sppp_keepalive(void *dummy);
static void sppp_phase_network(struct sppp *sp);
static void sppp_print_bytes(const u_char *p, u_short len);
static void sppp_print_string(const char *p, u_short len);
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
static void sppp_set_ip_addrs(struct sppp *sp, u_int32_t myaddr, u_int32_t hisaddr);
static void sppp_clear_ip_addrs(struct sppp *sp);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
#ifdef INET6
static void sppp_get_ip6_addrs(struct sppp *sp, struct in6_addr *src,
struct in6_addr *dst, struct in6_addr *srcmask);
#ifdef IPV6CP_MYIFID_DYN
static void sppp_set_ip6_addr(struct sppp *sp, const struct in6_addr *src);
static void sppp_gen_ip6_addr(struct sppp *sp, const struct in6_addr *src);
#endif
static void sppp_suggest_ip6_addr(struct sppp *sp, struct in6_addr *src);
#endif
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* our control protocol descriptors */
static const struct cp lcp = {
PPP_LCP, IDX_LCP, CP_LCP, "lcp",
sppp_lcp_up, sppp_lcp_down, sppp_lcp_open, sppp_lcp_close,
sppp_lcp_TO, sppp_lcp_RCR, sppp_lcp_RCN_rej, sppp_lcp_RCN_nak,
sppp_lcp_tlu, sppp_lcp_tld, sppp_lcp_tls, sppp_lcp_tlf,
sppp_lcp_scr
};
static const struct cp ipcp = {
run IPCP only if we have IPv4 in kernel
2001-01-16 18:28:27 +03:00
PPP_IPCP, IDX_IPCP,
#ifdef INET
CP_NCP, /*don't run IPCP if there's no IPv4 support*/
#else
0,
#endif
"ipcp",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_ipcp_up, sppp_ipcp_down, sppp_ipcp_open, sppp_ipcp_close,
sppp_ipcp_TO, sppp_ipcp_RCR, sppp_ipcp_RCN_rej, sppp_ipcp_RCN_nak,
sppp_ipcp_tlu, sppp_ipcp_tld, sppp_ipcp_tls, sppp_ipcp_tlf,
sppp_ipcp_scr
};
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
static const struct cp ipv6cp = {
PPP_IPV6CP, IDX_IPV6CP,
#ifdef INET6 /*don't run IPv6CP if there's no IPv6 support*/
CP_NCP,
#else
0,
#endif
"ipv6cp",
sppp_ipv6cp_up, sppp_ipv6cp_down, sppp_ipv6cp_open, sppp_ipv6cp_close,
sppp_ipv6cp_TO, sppp_ipv6cp_RCR, sppp_ipv6cp_RCN_rej, sppp_ipv6cp_RCN_nak,
sppp_ipv6cp_tlu, sppp_ipv6cp_tld, sppp_ipv6cp_tls, sppp_ipv6cp_tlf,
sppp_ipv6cp_scr
};
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static const struct cp pap = {
PPP_PAP, IDX_PAP, CP_AUTH, "pap",
sppp_null, sppp_null, sppp_pap_open, sppp_pap_close,
sppp_pap_TO, 0, 0, 0,
sppp_pap_tlu, sppp_pap_tld, sppp_null, sppp_null,
sppp_pap_scr
};
static const struct cp chap = {
PPP_CHAP, IDX_CHAP, CP_AUTH, "chap",
sppp_null, sppp_null, sppp_chap_open, sppp_chap_close,
sppp_chap_TO, 0, 0, 0,
sppp_chap_tlu, sppp_chap_tld, sppp_null, sppp_null,
sppp_chap_scr
};
static const struct cp *cps[IDX_COUNT] = {
&lcp, /* IDX_LCP */
&ipcp, /* IDX_IPCP */
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
&ipv6cp, /* IDX_IPV6CP */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
&pap, /* IDX_PAP */
&chap, /* IDX_CHAP */
};
/*
* Exported functions, comprising our interface to the lower layer.
*/
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/*
* Process the received packet.
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
void
sppp_input(struct ifnet *ifp, struct mbuf *m)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
struct ppp_header *h = NULL;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct ifqueue *inq = 0;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
u_int16_t protocol;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
int s;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
struct sppp *sp = (struct sppp *)ifp;
int debug = ifp->if_flags & IFF_DEBUG;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (ifp->if_flags & IFF_UP)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
/* Count received bytes, add hardware framing */
ifp->if_ibytes += m->m_pkthdr.len + sp->pp_framebytes;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (m->m_pkthdr.len <= PPP_HEADER_LEN) {
/* Too small packet, drop it. */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG,
SPP_FMT "input packet is too small, %d bytes\n",
SPP_ARGS(ifp), m->m_pkthdr.len);
drop:
++ifp->if_ierrors;
++ifp->if_iqdrops;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
m_freem (m);
return;
}
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if (sp->pp_flags & PP_NOFRAMING) {
code cleanup for portability
2001-12-05 00:32:15 +03:00
memcpy(&protocol, mtod(m, void *), 2);
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = ntohs(protocol);
m_adj(m, 2);
} else {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
/* Get PPP header. */
h = mtod (m, struct ppp_header*);
m_adj (m, PPP_HEADER_LEN);
switch (h->address) {
case PPP_ALLSTATIONS:
if (h->control != PPP_UI)
goto invalid;
if (sp->pp_flags & PP_CISCO) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "PPP packet in Cisco mode "
"<addr=0x%x ctrl=0x%x proto=0x%x>\n",
SPP_ARGS(ifp),
h->address, h->control, ntohs(h->protocol));
goto drop;
}
break;
case CISCO_MULTICAST:
case CISCO_UNICAST:
/* Don't check the control field here (RFC 1547). */
if (! (sp->pp_flags & PP_CISCO)) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "Cisco packet in PPP mode "
"<addr=0x%x ctrl=0x%x proto=0x%x>\n",
SPP_ARGS(ifp),
h->address, h->control, ntohs(h->protocol));
goto drop;
}
switch (ntohs (h->protocol)) {
default:
++ifp->if_noproto;
goto invalid;
case CISCO_KEEPALIVE:
sppp_cisco_input ((struct sppp*) ifp, m);
m_freem (m);
return;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef INET
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case ETHERTYPE_IP:
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
schednetisr (NETISR_IP);
inq = &ipintrq;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#endif
add more IPv6 cases. not tested. TODO: IPv6CP support. currently IPv6 packet will be generated right after link up (spec violation)
2000-04-12 14:51:15 +04:00
#ifdef INET6
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case ETHERTYPE_IPV6:
add more IPv6 cases. not tested. TODO: IPv6CP support. currently IPv6 packet will be generated right after link up (spec violation)
2000-04-12 14:51:15 +04:00
schednetisr (NETISR_IPV6);
inq = &ip6intrq;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
break;
add more IPv6 cases. not tested. TODO: IPv6CP support. currently IPv6 packet will be generated right after link up (spec violation)
2000-04-12 14:51:15 +04:00
#endif
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef IPX
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case ETHERTYPE_IPX:
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
schednetisr (NETISR_IPX);
inq = &ipxintrq;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#endif
#ifdef NS
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case ETHERTYPE_NS:
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
schednetisr (NETISR_NS);
inq = &nsintrq;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#endif
}
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
goto queue_pkt;
default: /* Invalid PPP packet. */
invalid:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG,
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
SPP_FMT "invalid input packet "
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
"<addr=0x%x ctrl=0x%x proto=0x%x>\n",
SPP_ARGS(ifp),
h->address, h->control, ntohs(h->protocol));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
goto drop;
}
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = ntohs (h->protocol);
}
switch (protocol) {
default:
if (sp->state[IDX_LCP] == STATE_OPENED) {
u_int16_t prot = htons(protocol);
sppp_cp_send (sp, PPP_LCP, PROTO_REJ,
++sp->pp_seq[IDX_LCP], m->m_pkthdr.len + 2,
&prot);
}
if (debug)
log(LOG_DEBUG,
SPP_FMT "invalid input protocol "
"<proto=0x%x>\n", SPP_ARGS(ifp), ntohs(protocol));
++ifp->if_noproto;
goto drop;
case PPP_LCP:
sppp_cp_input(&lcp, sp, m);
m_freem (m);
return;
case PPP_PAP:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase >= SPPP_PHASE_AUTHENTICATE)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
sppp_pap_input(sp, m);
m_freem (m);
return;
case PPP_CHAP:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase >= SPPP_PHASE_AUTHENTICATE)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
sppp_chap_input(sp, m);
m_freem (m);
return;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef INET
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case PPP_IPCP:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
sppp_cp_input(&ipcp, sp, m);
m_freem (m);
return;
case PPP_IP:
if (sp->state[IDX_IPCP] == STATE_OPENED) {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
schednetisr (NETISR_IP);
inq = &ipintrq;
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
sp->pp_last_activity = time.tv_sec;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
}
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#endif
add more IPv6 cases. not tested. TODO: IPv6CP support. currently IPv6 packet will be generated right after link up (spec violation)
2000-04-12 14:51:15 +04:00
#ifdef INET6
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case PPP_IPV6CP:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
sppp_cp_input(&ipv6cp, sp, m);
m_freem (m);
return;
case PPP_IPV6:
if (sp->state[IDX_IPV6CP] == STATE_OPENED) {
add more IPv6 cases. not tested. TODO: IPv6CP support. currently IPv6 packet will be generated right after link up (spec violation)
2000-04-12 14:51:15 +04:00
schednetisr (NETISR_IPV6);
inq = &ip6intrq;
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
sp->pp_last_activity = time.tv_sec;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
}
break;
add more IPv6 cases. not tested. TODO: IPv6CP support. currently IPv6 packet will be generated right after link up (spec violation)
2000-04-12 14:51:15 +04:00
#endif
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef IPX
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case PPP_IPX:
/* IPX IPXCP not implemented yet */
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK) {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
schednetisr (NETISR_IPX);
inq = &ipxintrq;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
}
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#endif
#ifdef NS
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
case PPP_XNS:
/* XNS IDPCP not implemented yet */
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK) {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
schednetisr (NETISR_NS);
inq = &nsintrq;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
}
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#endif
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
#ifdef ISO
case PPP_ISO:
/* OSI NLCP not implemented yet */
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK) {
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
schednetisr (NETISR_ISO);
inq = &clnlintrq;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
break;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
#endif
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
queue_pkt:
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (! (ifp->if_flags & IFF_UP) || ! inq)
goto drop;
/* Check queue. */
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (IF_QFULL (inq)) {
/* Queue overflow. */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
IF_DROP(inq);
splx(s);
if (debug)
log(LOG_DEBUG, SPP_FMT "protocol queue overflow\n",
SPP_ARGS(ifp));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
goto drop;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
IF_ENQUEUE(inq, m);
splx(s);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
/*
* Enqueue transmit packet.
*/
static int
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_output(struct ifnet *ifp, struct mbuf *m,
struct sockaddr *dst, struct rtentry *rt)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
struct sppp *sp = (struct sppp*) ifp;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
struct ppp_header *h = NULL;
Add ALTQ glue.
2000-12-14 01:07:50 +03:00
struct ifqueue *ifq = NULL; /* XXX */
int s, len, rv = 0;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
u_int16_t protocol;
Add ALTQ glue.
2000-12-14 01:07:50 +03:00
ALTQ_DECL(struct altq_pktattr pktattr;)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
sp->pp_last_activity = time.tv_sec;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if ((ifp->if_flags & IFF_UP) == 0 ||
(ifp->if_flags & (IFF_RUNNING | IFF_AUTO)) == 0) {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
m_freem (m);
splx (s);
return (ENETDOWN);
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if ((ifp->if_flags & (IFF_RUNNING | IFF_AUTO)) == IFF_AUTO) {
/*
* Interface is not yet running, but auto-dial. Need
* to start LCP for it.
*/
ifp->if_flags |= IFF_RUNNING;
splx(s);
lcp.Open(sp);
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
Add ALTQ glue.
2000-12-14 01:07:50 +03:00
/*
* If the queueing discipline needs packet classification,
* do it before prepending link headers.
*/
IFQ_CLASSIFY(&ifp->if_snd, m, dst->sa_family, &pktattr);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef INET
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (dst->sa_family == AF_INET)
{
/* Check mbuf length here??? */
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct ip *ip = mtod (m, struct ip*);
code cleanup for portability
2001-12-05 00:32:15 +03:00
struct tcphdr *tcp = (struct tcphdr*) ((int32_t*)ip + ip->ip_hl);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* When using dynamic local IP address assignment by using
* 0.0.0.0 as a local address, the first TCP session will
* not connect because the local TCP checksum is computed
* using 0.0.0.0 which will later become our real IP address
* so the TCP checksum computed at the remote end will
* become invalid. So we
* - don't let packets with src ip addr 0 thru
* - we flag TCP packets with src ip 0 as an error
*/
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(ip->ip_src.s_addr == INADDR_ANY) /* -hm */
{
m_freem(m);
splx(s);
if(ip->ip_p == IPPROTO_TCP)
return(EADDRNOTAVAIL);
else
return(0);
}
/*
* Put low delay, telnet, rlogin and ftp control packets
* in front of the queue.
*/
if (! IF_QFULL (&sp->pp_fastq) &&
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
((ip->ip_tos & IPTOS_LOWDELAY) ||
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
((ip->ip_p == IPPROTO_TCP &&
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
m->m_len >= sizeof (struct ip) + sizeof (struct tcphdr) &&
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
(INTERACTIVE (ntohs (tcp->th_sport)))) ||
INTERACTIVE (ntohs (tcp->th_dport)))))
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
ifq = &sp->pp_fastq;
}
#endif
remove reference to in6_systm.h (file itself will be removed afterwords)
1999-07-30 14:35:34 +04:00
#ifdef INET6
if (dst->sa_family == AF_INET6) {
/* XXX do something tricky here? */
}
#endif
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if ((sp->pp_flags & PP_NOFRAMING) == 0) {
/*
* Prepend general data packet PPP header. For now, IP only.
*/
M_PREPEND (m, PPP_HEADER_LEN, M_DONTWAIT);
if (! m) {
if (ifp->if_flags & IFF_DEBUG)
log(LOG_DEBUG, SPP_FMT "no memory for transmit header\n",
SPP_ARGS(ifp));
++ifp->if_oerrors;
splx (s);
return (ENOBUFS);
}
/*
* May want to check size of packet
* (albeit due to the implementation it's always enough)
*/
h = mtod (m, struct ppp_header*);
if (sp->pp_flags & PP_CISCO) {
h->address = CISCO_UNICAST; /* unicast address */
h->control = 0;
} else {
h->address = PPP_ALLSTATIONS; /* broadcast address */
h->control = PPP_UI; /* Unnumbered Info */
}
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch (dst->sa_family) {
#ifdef INET
case AF_INET: /* Internet Protocol */
if (sp->pp_flags & PP_CISCO)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons (ETHERTYPE_IP);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
else {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Don't choke with an ENETDOWN early. It's
* possible that we just started dialing out,
* so don't drop the packet immediately. If
* we notice that we run out of buffer space
* below, we will however remember that we are
* not ready to carry IP packets, and return
* ENETDOWN, as opposed to ENOBUFS.
*/
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons(PPP_IP);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (sp->state[IDX_IPCP] != STATE_OPENED)
rv = ENETDOWN;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
break;
#endif
remove reference to in6_systm.h (file itself will be removed afterwords)
1999-07-30 14:35:34 +04:00
#ifdef INET6
case AF_INET6: /* Internet Protocol version 6 */
if (sp->pp_flags & PP_CISCO)
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons (ETHERTYPE_IPV6);
remove reference to in6_systm.h (file itself will be removed afterwords)
1999-07-30 14:35:34 +04:00
else {
/*
* Don't choke with an ENETDOWN early. It's
* possible that we just started dialing out,
* so don't drop the packet immediately. If
* we notice that we run out of buffer space
* below, we will however remember that we are
* not ready to carry IP packets, and return
* ENETDOWN, as opposed to ENOBUFS.
*/
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons(PPP_IPV6);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
if (sp->state[IDX_IPV6CP] != STATE_OPENED)
remove reference to in6_systm.h (file itself will be removed afterwords)
1999-07-30 14:35:34 +04:00
rv = ENETDOWN;
}
break;
#endif
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
#ifdef NS
case AF_NS: /* Xerox NS Protocol */
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons ((sp->pp_flags & PP_CISCO) ?
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
ETHERTYPE_NS : PPP_XNS);
break;
#endif
#ifdef IPX
case AF_IPX: /* Novell IPX Protocol */
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons ((sp->pp_flags & PP_CISCO) ?
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
ETHERTYPE_IPX : PPP_IPX);
break;
#endif
#ifdef ISO
case AF_ISO: /* ISO OSI Protocol */
if (sp->pp_flags & PP_CISCO)
goto nosupport;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
protocol = htons (PPP_ISO);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
nosupport:
#endif
default:
m_freem (m);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
++ifp->if_oerrors;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
splx (s);
return (EAFNOSUPPORT);
}
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if (sp->pp_flags & PP_NOFRAMING) {
M_PREPEND (m, 2, M_DONTWAIT);
if (m == NULL) {
if (ifp->if_flags & IFF_DEBUG)
log(LOG_DEBUG, SPP_FMT "no memory for transmit header\n",
SPP_ARGS(ifp));
++ifp->if_oerrors;
splx (s);
return (ENOBUFS);
}
*mtod(m, u_int16_t*) = protocol;
} else {
h->protocol = protocol;
}
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/*
* Queue message on interface, and start output if interface
* not yet active.
*/
Add ALTQ glue.
2000-12-14 01:07:50 +03:00
len = m->m_pkthdr.len;
if (ifq != NULL
#ifdef ALTQ
&& ALTQ_IS_ENABLED(&ifp->if_snd) == 0
#endif
) {
if (IF_QFULL (ifq)) {
IF_DROP (&ifp->if_snd);
m_freem (m);
if (rv == 0)
rv = ENOBUFS;
}
IF_ENQUEUE(ifq, m);
} else
IFQ_ENQUEUE(&ifp->if_snd, m, &pktattr, rv);
if (rv != 0) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
++ifp->if_oerrors;
Add ALTQ glue.
2000-12-14 01:07:50 +03:00
splx(s);
return (rv);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
Add ALTQ glue.
2000-12-14 01:07:50 +03:00
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (! (ifp->if_flags & IFF_OACTIVE))
(*ifp->if_start) (ifp);
/*
* Count output packets and bytes.
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
* The packet length includes header + additional hardware framing
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
* according to RFC 1333.
*/
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
ifp->if_obytes += len + sp->pp_framebytes;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
splx (s);
return (0);
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
void
sppp_attach(struct ifnet *ifp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
struct sppp *sp = (struct sppp*) ifp;
/* Initialize keepalive handler. */
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
if (! spppq) {
callout_init(&keepalive_ch);
callout_reset(&keepalive_ch, hz * 10, sppp_keepalive, NULL);
}
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/* Insert new entry into the keepalive list. */
sp->pp_next = spppq;
spppq = sp;
sp->pp_if.if_type = IFT_PPP;
sp->pp_if.if_output = sppp_output;
sp->pp_fastq.ifq_maxlen = 32;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->pp_cpq.ifq_maxlen = 20;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
sp->pp_loopcnt = 0;
sp->pp_alivecnt = 0;
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
sp->pp_last_activity = 0;
sp->pp_idle_timeout = 0;
bzero -> memset
2001-07-18 20:43:09 +04:00
memset(&sp->pp_seq[0], 0, sizeof(sp->pp_seq));
memset(&sp->pp_rseq[0], 0, sizeof(sp->pp_rseq));
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
sp->pp_max_auth_fail = DEFAULT_MAX_AUTH_FAILURES;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_DEAD;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->pp_up = lcp.Up;
sp->pp_down = lcp.Down;
Fix a rather annoying problem where the sockaddr_dl which holds the link level name for the interface (ifp->if_sadl) is allocated before ifp->if_addrlen is initialized, which could lead to allocating too little space for the link level address. Do this by splitting allocation of the link level name out of if_attach() and into if_alloc_sadl(), which is normally called by functions like ether_ifattach(). Network interfaces which don't have a link-specific attach routine must call if_alloc_sadl() themselves (example: gif). Link level names are freed by if_free_sadl(), which can be called from e.g. ether_ifdetach(). Drivers never need call if_free_sadl() themselves as if_detach() will do it if it is not already done. While here, add the ability to pass an AF_LINK address to SIOCSIFADDR in ether_ioctl() (this is what caused me to notice the problem that the above fixes).
2001-01-17 03:30:49 +03:00
if_alloc_sadl(ifp);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
memset(&sp->myauth, 0, sizeof sp->myauth);
memset(&sp->hisauth, 0, sizeof sp->hisauth);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_lcp_init(sp);
sppp_ipcp_init(sp);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_ipv6cp_init(sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_pap_init(sp);
sppp_chap_init(sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
void
sppp_detach(struct ifnet *ifp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
struct sppp **q, *p, *sp = (struct sppp*) ifp;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int i;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/* Remove the entry from the keepalive list. */
for (q = &spppq; (p = *q); q = &p->pp_next)
if (p == sp) {
*q = p->pp_next;
break;
}
/* Stop keepalive handler. */
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
if (! spppq) {
callout_stop(&keepalive_ch);
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
for (i = 0; i < IDX_COUNT; i++) {
callout_stop(&sp->ch[i]);
}
callout_stop(&sp->pap_my_to_ch);
Fix a rather annoying problem where the sockaddr_dl which holds the link level name for the interface (ifp->if_sadl) is allocated before ifp->if_addrlen is initialized, which could lead to allocating too little space for the link level address. Do this by splitting allocation of the link level name out of if_attach() and into if_alloc_sadl(), which is normally called by functions like ether_ifattach(). Network interfaces which don't have a link-specific attach routine must call if_alloc_sadl() themselves (example: gif). Link level names are freed by if_free_sadl(), which can be called from e.g. ether_ifdetach(). Drivers never need call if_free_sadl() themselves as if_detach() will do it if it is not already done. While here, add the ability to pass an AF_LINK address to SIOCSIFADDR in ether_ioctl() (this is what caused me to notice the problem that the above fixes).
2001-01-17 03:30:49 +03:00
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
/* free authentication info */
if (sp->myauth.name) free(sp->myauth.name, M_DEVBUF);
if (sp->myauth.secret) free(sp->myauth.secret, M_DEVBUF);
if (sp->hisauth.name) free(sp->hisauth.name, M_DEVBUF);
if (sp->hisauth.secret) free(sp->hisauth.secret, M_DEVBUF);
Fix a rather annoying problem where the sockaddr_dl which holds the link level name for the interface (ifp->if_sadl) is allocated before ifp->if_addrlen is initialized, which could lead to allocating too little space for the link level address. Do this by splitting allocation of the link level name out of if_attach() and into if_alloc_sadl(), which is normally called by functions like ether_ifattach(). Network interfaces which don't have a link-specific attach routine must call if_alloc_sadl() themselves (example: gif). Link level names are freed by if_free_sadl(), which can be called from e.g. ether_ifdetach(). Drivers never need call if_free_sadl() themselves as if_detach() will do it if it is not already done. While here, add the ability to pass an AF_LINK address to SIOCSIFADDR in ether_ioctl() (this is what caused me to notice the problem that the above fixes).
2001-01-17 03:30:49 +03:00
if_free_sadl(ifp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
/*
* Flush the interface output queue.
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
void
sppp_flush(struct ifnet *ifp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
struct sppp *sp = (struct sppp*) ifp;
Use IFQ_PURGE().
2000-12-18 23:50:36 +03:00
IFQ_PURGE (&sp->pp_if.if_snd);
IFQ_PURGE cannot be used against ifqueue. use IF_PURGE.
2001-08-23 04:56:03 +04:00
IF_PURGE (&sp->pp_fastq);
IF_PURGE (&sp->pp_cpq);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
/*
* Check if the output queue is empty.
*/
int
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_isempty(struct ifnet *ifp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
struct sppp *sp = (struct sppp*) ifp;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int empty, s;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
empty = !sp->pp_fastq.ifq_head && !sp->pp_cpq.ifq_head &&
!sp->pp_if.if_snd.ifq_head;
splx(s);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
return (empty);
}
/*
* Get next packet to send.
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
struct mbuf *
sppp_dequeue(struct ifnet *ifp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
struct sppp *sp = (struct sppp*) ifp;
struct mbuf *m;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int s;
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Process only the control protocol queue until we have at
* least one NCP open.
*
* Do always serve all three queues in Cisco mode.
*/
IF_DEQUEUE(&sp->pp_cpq, m);
if (m == NULL &&
(sppp_ncp_check(sp) || (sp->pp_flags & PP_CISCO) != 0)) {
IF_DEQUEUE(&sp->pp_fastq, m);
if (m == NULL)
IF_DEQUEUE (&sp->pp_if.if_snd, m);
}
splx(s);
return m;
}
/*
* Pick the next packet, do not remove it from the queue.
*/
struct mbuf *
sppp_pick(struct ifnet *ifp)
{
struct sppp *sp = (struct sppp*)ifp;
struct mbuf *m;
int s;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s= splnet ();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
m = sp->pp_cpq.ifq_head;
if (m == NULL &&
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
(sp->pp_phase == SPPP_PHASE_NETWORK ||
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
(sp->pp_flags & PP_CISCO) != 0))
if ((m = sp->pp_fastq.ifq_head) == NULL)
m = sp->pp_if.if_snd.ifq_head;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
splx (s);
return (m);
}
/*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* Process an ioctl request. Called on low priority level.
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int
sppp_ioctl(struct ifnet *ifp, u_long cmd, void *data)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
struct ifreq *ifr = (struct ifreq*) data;
struct sppp *sp = (struct sppp*) ifp;
int s, rv, going_up, going_down, newmode;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
rv = 0;
switch (cmd) {
case SIOCAIFADDR:
case SIOCSIFDSTADDR:
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case SIOCSIFADDR:
if_up(ifp);
/* fall through... */
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case SIOCSIFFLAGS:
going_up = ifp->if_flags & IFF_UP &&
(ifp->if_flags & IFF_RUNNING) == 0;
going_down = (ifp->if_flags & IFF_UP) == 0 &&
ifp->if_flags & IFF_RUNNING;
newmode = ifp->if_flags & (IFF_AUTO | IFF_PASSIVE);
if (newmode == (IFF_AUTO | IFF_PASSIVE)) {
/* sanity */
newmode = IFF_PASSIVE;
ifp->if_flags &= ~IFF_AUTO;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (going_up || going_down)
lcp.Close(sp);
if (going_up && newmode == 0) {
/* neither auto-dial nor passive */
ifp->if_flags |= IFF_RUNNING;
if (!(sp->pp_flags & PP_CISCO))
lcp.Open(sp);
} else if (going_down) {
sppp_flush(ifp);
ifp->if_flags &= ~IFF_RUNNING;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
#ifdef SIOCSIFMTU
#ifndef ifr_mtu
#define ifr_mtu ifr_metric
#endif
case SIOCSIFMTU:
if (ifr->ifr_mtu < 128 || ifr->ifr_mtu > sp->lcp.their_mru)
return (EINVAL);
ifp->if_mtu = ifr->ifr_mtu;
break;
#endif
#ifdef SLIOCSETMTU
case SLIOCSETMTU:
if (*(short*)data < 128 || *(short*)data > sp->lcp.their_mru)
return (EINVAL);
ifp->if_mtu = *(short*)data;
break;
#endif
#ifdef SIOCGIFMTU
case SIOCGIFMTU:
ifr->ifr_mtu = ifp->if_mtu;
break;
#endif
#ifdef SLIOCGETMTU
case SLIOCGETMTU:
*(short*)data = ifp->if_mtu;
break;
#endif
case SIOCADDMULTI:
case SIOCDELMULTI:
break;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
case SPPPSETAUTHCFG:
case SPPPSETLCPCFG:
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
case SPPPSETIDLETO:
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
case SPPPSETAUTHFAILURE:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
{
struct proc *p = curproc; /* XXX */
if ((rv = suser(p->p_ucred, &p->p_acflag)) != 0)
break;
}
/* FALLTHROUGH */
case SPPPGETAUTHCFG:
case SPPPGETLCPCFG:
case SPPPGETSTATUS:
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
case SPPPGETIDLETO:
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
case SPPPGETAUTHFAILURES:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
rv = sppp_params(sp, cmd, data);
break;
default:
rv = ENOTTY;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
splx(s);
return rv;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Cisco framing implementation.
*/
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* Handle incoming Cisco keepalive protocol packets.
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static void
sppp_cisco_input(struct sppp *sp, struct mbuf *m)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
struct cisco_packet *h;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t me, mymask;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (m->m_pkthdr.len < CISCO_PACKET_LEN) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "cisco invalid packet length: %d bytes\n",
SPP_ARGS(ifp), m->m_pkthdr.len);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
return;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
h = mtod (m, struct cisco_packet*);
if (debug)
log(LOG_DEBUG,
SPP_FMT "cisco input: %d bytes "
code cleanup for portability
2001-12-05 00:32:15 +03:00
"<0x%x 0x%x 0x%x 0x%x 0x%x-0x%x>\n",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp), m->m_pkthdr.len,
code cleanup for portability
2001-12-05 00:32:15 +03:00
ntohl (h->type), h->par1, h->par2, (u_int)h->rel,
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
(u_int)h->time0, (u_int)h->time1);
switch (ntohl (h->type)) {
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
default:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
code cleanup for portability
2001-12-05 00:32:15 +03:00
addlog(SPP_FMT "cisco unknown packet type: 0x%x\n",
SPP_ARGS(ifp), ntohl (h->type));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case CISCO_ADDR_REPLY:
/* Reply on address request, ignore */
break;
case CISCO_KEEPALIVE_REQ:
sp->pp_alivecnt = 0;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->pp_rseq[IDX_LCP] = ntohl (h->par1);
if (sp->pp_seq[IDX_LCP] == sp->pp_rseq[IDX_LCP]) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* Local and remote sequence numbers are equal.
* Probably, the line is in loopback mode. */
if (sp->pp_loopcnt >= MAXALIVECNT) {
printf (SPP_FMT "loopback\n",
SPP_ARGS(ifp));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
sp->pp_loopcnt = 0;
if (ifp->if_flags & IFF_UP) {
if_down (ifp);
IFQ_PURGE cannot be used against ifqueue. use IF_PURGE.
2001-08-23 04:56:03 +04:00
IF_PURGE (&sp->pp_cpq);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
++sp->pp_loopcnt;
/* Generate new local sequence number */
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->pp_seq[IDX_LCP] = random();
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->pp_loopcnt = 0;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (! (ifp->if_flags & IFF_UP) &&
(ifp->if_flags & IFF_RUNNING)) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if_up(ifp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
break;
case CISCO_ADDR_REQ:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_get_ip_addrs(sp, &me, 0, &mymask);
if (me != 0L)
sppp_cisco_send(sp, CISCO_ADDR_REPLY, me, mymask);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
}
}
/*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* Send Cisco keepalive packet.
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
static void
code cleanup for portability
2001-12-05 00:32:15 +03:00
sppp_cisco_send(struct sppp *sp, int type, int32_t par1, int32_t par2)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct ppp_header *h;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
struct cisco_packet *ch;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct mbuf *m;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t t = (time.tv_sec - boottime.tv_sec) * 1000;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
MGETHDR (m, M_DONTWAIT, MT_DATA);
if (! m)
return;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
m->m_pkthdr.len = m->m_len = PPP_HEADER_LEN + CISCO_PACKET_LEN;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
m->m_pkthdr.rcvif = 0;
h = mtod (m, struct ppp_header*);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
h->address = CISCO_MULTICAST;
h->control = 0;
h->protocol = htons (CISCO_KEEPALIVE);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
ch = (struct cisco_packet*) (h + 1);
ch->type = htonl (type);
ch->par1 = htonl (par1);
ch->par2 = htonl (par2);
ch->rel = -1;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
ch->time0 = htons ((u_short) (t >> 16));
ch->time1 = htons ((u_short) t);
if (debug)
log(LOG_DEBUG,
code cleanup for portability
2001-12-05 00:32:15 +03:00
SPP_FMT "cisco output: <0x%x 0x%x 0x%x 0x%x 0x%x-0x%x>\n",
SPP_ARGS(ifp), ntohl (ch->type), ch->par1,
ch->par2, (u_int)ch->rel, (u_int)ch->time0,
(u_int)ch->time1);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (IF_QFULL (&sp->pp_cpq)) {
IF_DROP (&sp->pp_fastq);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
IF_DROP (&ifp->if_snd);
m_freem (m);
} else
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
IF_ENQUEUE (&sp->pp_cpq, m);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (! (ifp->if_flags & IFF_OACTIVE))
(*ifp->if_start) (ifp);
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
ifp->if_obytes += m->m_pkthdr.len + sp->pp_framebytes;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* PPP protocol implementation.
*/
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
/*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* Send PPP control protocol packet.
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
static void
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_cp_send(struct sppp *sp, u_short proto, u_char type,
u_char ident, u_short len, void *data)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
struct lcp_header *lh;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct mbuf *m;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
size_t pkthdrlen;
pkthdrlen = (sp->pp_flags & PP_NOFRAMING) ? 2 : PPP_HEADER_LEN;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if (len > MHLEN - pkthdrlen - LCP_HEADER_LEN)
len = MHLEN - pkthdrlen - LCP_HEADER_LEN;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
MGETHDR (m, M_DONTWAIT, MT_DATA);
if (! m)
return;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
m->m_pkthdr.len = m->m_len = pkthdrlen + LCP_HEADER_LEN + len;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
m->m_pkthdr.rcvif = 0;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if (sp->pp_flags & PP_NOFRAMING) {
*mtod(m, u_int16_t*) = htons(proto);
lh = (struct lcp_header*)(mtod(m, u_int8_t*) + 2);
} else {
struct ppp_header *h;
h = mtod (m, struct ppp_header*);
h->address = PPP_ALLSTATIONS; /* broadcast address */
h->control = PPP_UI; /* Unnumbered Info */
h->protocol = htons (proto); /* Link Control Protocol */
lh = (struct lcp_header*) (h + 1);
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
lh->type = type;
lh->ident = ident;
lh->len = htons (LCP_HEADER_LEN + len);
if (len)
bcopy (data, lh+1, len);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug) {
log(LOG_DEBUG, SPP_FMT "%s output <%s id=0x%x len=%d",
SPP_ARGS(ifp),
sppp_proto_name(proto),
sppp_cp_type_name (lh->type), lh->ident,
code cleanup for portability
2001-12-05 00:32:15 +03:00
ntohs (lh->len));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (len)
sppp_print_bytes ((u_char*) (lh+1), len);
addlog(">\n");
}
if (IF_QFULL (&sp->pp_cpq)) {
IF_DROP (&sp->pp_fastq);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
IF_DROP (&ifp->if_snd);
m_freem (m);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
++ifp->if_oerrors;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
} else
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
IF_ENQUEUE (&sp->pp_cpq, m);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (! (ifp->if_flags & IFF_OACTIVE))
(*ifp->if_start) (ifp);
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
ifp->if_obytes += m->m_pkthdr.len + sp->pp_framebytes;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
/*
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* Handle incoming PPP control protocol packets.
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
*/
static void
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_cp_input(const struct cp *cp, struct sppp *sp, struct mbuf *m)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
struct lcp_header *h;
int len = m->m_pkthdr.len;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int rv;
u_char *p;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t u32;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (len < 4) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG,
SPP_FMT "%s invalid packet length: %d bytes\n",
SPP_ARGS(ifp), cp->name, len);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
return;
}
h = mtod (m, struct lcp_header*);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug) {
log(LOG_DEBUG,
SPP_FMT "%s input(%s): <%s id=0x%x len=%d",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]),
sppp_cp_type_name (h->type), h->ident, ntohs (h->len));
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (len > 4)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_print_bytes ((u_char*) (h+1), len-4);
addlog(">\n");
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
if (len > ntohs (h->len))
len = ntohs (h->len);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
p = (u_char *)(h + 1);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch (h->type) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case CONF_REQ:
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
if (len < 4) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
addlog(SPP_FMT "%s invalid conf-req length %d\n",
SPP_ARGS(ifp), cp->name,
len);
++ifp->if_ierrors;
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* handle states where RCR doesn't get a SCA/SCN */
switch (sp->state[cp->protoidx]) {
case STATE_CLOSING:
case STATE_STOPPING:
return;
case STATE_CLOSED:
sppp_cp_send(sp, cp->proto, TERM_ACK, h->ident,
0, 0);
return;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
rv = (cp->RCR)(sp, h, len);
switch (sp->state[cp->protoidx]) {
case STATE_OPENED:
(cp->tld)(sp);
(cp->scr)(sp);
/* fall through... */
case STATE_ACK_SENT:
case STATE_REQ_SENT:
sppp_cp_change_state(cp, sp, rv?
STATE_ACK_SENT: STATE_REQ_SENT);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_STOPPED:
sp->rst_counter[cp->protoidx] = sp->lcp.max_configure;
(cp->scr)(sp);
sppp_cp_change_state(cp, sp, rv?
STATE_ACK_SENT: STATE_REQ_SENT);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_ACK_RCVD:
if (rv) {
sppp_cp_change_state(cp, sp, STATE_OPENED);
if (debug)
log(LOG_DEBUG, SPP_FMT "%s tlu\n",
SPP_ARGS(ifp),
cp->name);
(cp->tlu)(sp);
} else
sppp_cp_change_state(cp, sp, STATE_ACK_RCVD);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
}
break;
case CONF_ACK:
if (h->ident != sp->confid[cp->protoidx]) {
if (debug)
addlog(SPP_FMT "%s id mismatch 0x%x != 0x%x\n",
SPP_ARGS(ifp), cp->name,
h->ident, sp->confid[cp->protoidx]);
++ifp->if_ierrors;
break;
}
switch (sp->state[cp->protoidx]) {
case STATE_CLOSED:
case STATE_STOPPED:
sppp_cp_send(sp, cp->proto, TERM_ACK, h->ident, 0, 0);
break;
case STATE_CLOSING:
case STATE_STOPPING:
break;
case STATE_REQ_SENT:
sp->rst_counter[cp->protoidx] = sp->lcp.max_configure;
sppp_cp_change_state(cp, sp, STATE_ACK_RCVD);
break;
case STATE_OPENED:
(cp->tld)(sp);
/* fall through */
case STATE_ACK_RCVD:
(cp->scr)(sp);
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
break;
case STATE_ACK_SENT:
sp->rst_counter[cp->protoidx] = sp->lcp.max_configure;
sppp_cp_change_state(cp, sp, STATE_OPENED);
if (debug)
log(LOG_DEBUG, SPP_FMT "%s tlu\n",
SPP_ARGS(ifp), cp->name);
(cp->tlu)(sp);
break;
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
}
break;
case CONF_NAK:
case CONF_REJ:
if (h->ident != sp->confid[cp->protoidx]) {
if (debug)
addlog(SPP_FMT "%s id mismatch 0x%x != 0x%x\n",
SPP_ARGS(ifp), cp->name,
h->ident, sp->confid[cp->protoidx]);
++ifp->if_ierrors;
break;
}
if (h->type == CONF_NAK)
(cp->RCN_nak)(sp, h, len);
else /* CONF_REJ */
(cp->RCN_rej)(sp, h, len);
switch (sp->state[cp->protoidx]) {
case STATE_CLOSED:
case STATE_STOPPED:
sppp_cp_send(sp, cp->proto, TERM_ACK, h->ident, 0, 0);
break;
case STATE_REQ_SENT:
case STATE_ACK_SENT:
sp->rst_counter[cp->protoidx] = sp->lcp.max_configure;
(cp->scr)(sp);
break;
case STATE_OPENED:
(cp->tld)(sp);
/* fall through */
case STATE_ACK_RCVD:
sppp_cp_change_state(cp, sp, STATE_ACK_SENT);
(cp->scr)(sp);
break;
case STATE_CLOSING:
case STATE_STOPPING:
break;
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
}
break;
case TERM_REQ:
switch (sp->state[cp->protoidx]) {
case STATE_ACK_RCVD:
case STATE_ACK_SENT:
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
/* fall through */
case STATE_CLOSED:
case STATE_STOPPED:
case STATE_CLOSING:
case STATE_STOPPING:
case STATE_REQ_SENT:
sta:
/* Send Terminate-Ack packet. */
if (debug)
log(LOG_DEBUG, SPP_FMT "%s send terminate-ack\n",
SPP_ARGS(ifp), cp->name);
sppp_cp_send(sp, cp->proto, TERM_ACK, h->ident, 0, 0);
break;
case STATE_OPENED:
(cp->tld)(sp);
sp->rst_counter[cp->protoidx] = 0;
sppp_cp_change_state(cp, sp, STATE_STOPPING);
goto sta;
break;
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
}
break;
case TERM_ACK:
switch (sp->state[cp->protoidx]) {
case STATE_CLOSED:
case STATE_STOPPED:
case STATE_REQ_SENT:
case STATE_ACK_SENT:
break;
case STATE_CLOSING:
(cp->tlf)(sp);
sppp_cp_change_state(cp, sp, STATE_CLOSED);
Fix a slight bug introduced with revision 1.9 (IPv6 integration) where the bit mask of open NCPs got out of sync. Defer the (potential) closing of LCP after a NCP went down until after the state machines got updated. This fixes PR kern/11161.
2001-07-17 23:12:02 +04:00
sppp_lcp_check_and_close(sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
case STATE_STOPPING:
(cp->tlf)(sp);
sppp_cp_change_state(cp, sp, STATE_STOPPED);
Fix a slight bug introduced with revision 1.9 (IPv6 integration) where the bit mask of open NCPs got out of sync. Defer the (potential) closing of LCP after a NCP went down until after the state machines got updated. This fixes PR kern/11161.
2001-07-17 23:12:02 +04:00
sppp_lcp_check_and_close(sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
case STATE_ACK_RCVD:
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
break;
case STATE_OPENED:
(cp->tld)(sp);
(cp->scr)(sp);
sppp_cp_change_state(cp, sp, STATE_ACK_RCVD);
break;
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
}
break;
case CODE_REJ:
/* XXX catastrophic rejects (RXJ-) aren't handled yet. */
log(LOG_INFO,
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
SPP_FMT "%s: ignoring RXJ (%s) for code ?, "
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
"danger will robinson\n",
SPP_ARGS(ifp), cp->name,
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_cp_type_name(h->type));
switch (sp->state[cp->protoidx]) {
case STATE_CLOSED:
case STATE_STOPPED:
case STATE_REQ_SENT:
case STATE_ACK_SENT:
case STATE_CLOSING:
case STATE_STOPPING:
case STATE_OPENED:
break;
case STATE_ACK_RCVD:
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
break;
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
}
break;
case PROTO_REJ:
{
int catastrophic;
const struct cp *upper;
int i;
u_int16_t proto;
catastrophic = 0;
upper = NULL;
code cleanup for portability
2001-12-05 00:32:15 +03:00
proto = p[0] << 8 | p[1];
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
for (i = 0; i < IDX_COUNT; i++) {
if (cps[i]->proto == proto) {
upper = cps[i];
break;
}
}
if (upper == NULL)
catastrophic++;
log(LOG_INFO,
SPP_FMT "%s: RXJ%c (%s) for proto 0x%x (%s/%s)\n",
SPP_ARGS(ifp), cp->name, catastrophic ? '-' : '+',
sppp_cp_type_name(h->type), proto,
upper ? upper->name : "unknown",
upper ? sppp_state_name(sp->state[upper->protoidx]) : "?");
/*
* if we got RXJ+ against conf-req, the peer does not implement
* this particular protocol type. terminate the protocol.
*/
if (upper && !catastrophic) {
if (sp->state[upper->protoidx] == STATE_REQ_SENT) {
upper->Close(sp);
break;
}
}
/* XXX catastrophic rejects (RXJ-) aren't handled yet. */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (sp->state[cp->protoidx]) {
case STATE_CLOSED:
case STATE_STOPPED:
case STATE_REQ_SENT:
case STATE_ACK_SENT:
case STATE_CLOSING:
case STATE_STOPPING:
case STATE_OPENED:
break;
case STATE_ACK_RCVD:
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
break;
default:
printf(SPP_FMT "%s illegal %s in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_cp_type_name(h->type),
sppp_state_name(sp->state[cp->protoidx]));
++ifp->if_ierrors;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
break;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case DISC_REQ:
if (cp->proto != PPP_LCP)
goto illegal;
/* Discard the packet. */
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case ECHO_REQ:
if (cp->proto != PPP_LCP)
goto illegal;
if (sp->state[cp->protoidx] != STATE_OPENED) {
if (debug)
addlog(SPP_FMT "lcp echo req but lcp closed\n",
SPP_ARGS(ifp));
++ifp->if_ierrors;
break;
}
if (len < 8) {
if (debug)
addlog(SPP_FMT "invalid lcp echo request "
"packet length: %d bytes\n",
SPP_ARGS(ifp), len);
break;
}
code cleanup for portability
2001-12-05 00:32:15 +03:00
memcpy(&u32, h + 1, sizeof u32);
if (ntohl(u32) == sp->lcp.magic) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* Line loopback mode detected. */
printf(SPP_FMT "loopback\n", SPP_ARGS(ifp));
if_down (ifp);
IFQ_PURGE cannot be used against ifqueue. use IF_PURGE.
2001-08-23 04:56:03 +04:00
IF_PURGE (&sp->pp_cpq);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* Shut down the PPP link. */
/* XXX */
lcp.Down(sp);
lcp.Up(sp);
break;
}
code cleanup for portability
2001-12-05 00:32:15 +03:00
u32 = htonl(sp->lcp.magic);
memcpy(h + 1, &u32, sizeof u32);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
addlog(SPP_FMT "got lcp echo req, sending echo rep\n",
SPP_ARGS(ifp));
sppp_cp_send (sp, PPP_LCP, ECHO_REPLY, h->ident, len-4, h+1);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case ECHO_REPLY:
if (cp->proto != PPP_LCP)
goto illegal;
if (h->ident != sp->lcp.echoid) {
++ifp->if_ierrors;
break;
}
if (len < 8) {
if (debug)
addlog(SPP_FMT "lcp invalid echo reply "
"packet length: %d bytes\n",
SPP_ARGS(ifp), len);
break;
}
if (debug)
addlog(SPP_FMT "lcp got echo rep\n",
SPP_ARGS(ifp));
code cleanup for portability
2001-12-05 00:32:15 +03:00
memcpy(&u32, h + 1, sizeof u32);
if (ntohl(u32) != sp->lcp.magic)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->pp_alivecnt = 0;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
default:
/* Unknown packet type -- send Code-Reject packet. */
illegal:
if (debug)
addlog(SPP_FMT "%s send code-rej for 0x%x\n",
SPP_ARGS(ifp), cp->name, h->type);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_cp_send(sp, cp->proto, CODE_REJ,
++sp->pp_seq[cp->protoidx], m->m_pkthdr.len, h);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
++ifp->if_ierrors;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* The generic part of all Up/Down/Open/Close/TO event handlers.
* Basically, the state transition handling in the automaton.
*/
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
static void
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_up_event(const struct cp *cp, struct sppp *sp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG, SPP_FMT "%s up(%s)\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]));
switch (sp->state[cp->protoidx]) {
case STATE_INITIAL:
sppp_cp_change_state(cp, sp, STATE_CLOSED);
break;
case STATE_STARTING:
sp->rst_counter[cp->protoidx] = sp->lcp.max_configure;
(cp->scr)(sp);
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
break;
default:
printf(SPP_FMT "%s illegal up in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]));
}
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
static void
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_down_event(const struct cp *cp, struct sppp *sp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
if (debug)
log(LOG_DEBUG, SPP_FMT "%s down(%s)\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]));
switch (sp->state[cp->protoidx]) {
case STATE_CLOSED:
case STATE_CLOSING:
sppp_cp_change_state(cp, sp, STATE_INITIAL);
break;
case STATE_STOPPED:
(cp->tls)(sp);
/* fall through */
case STATE_STOPPING:
case STATE_REQ_SENT:
case STATE_ACK_RCVD:
case STATE_ACK_SENT:
sppp_cp_change_state(cp, sp, STATE_STARTING);
break;
case STATE_OPENED:
(cp->tld)(sp);
sppp_cp_change_state(cp, sp, STATE_STARTING);
break;
default:
printf(SPP_FMT "%s illegal down in state %s\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]));
}
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
static void
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_open_event(const struct cp *cp, struct sppp *sp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG, SPP_FMT "%s open(%s)\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]));
switch (sp->state[cp->protoidx]) {
case STATE_INITIAL:
(cp->tls)(sp);
sppp_cp_change_state(cp, sp, STATE_STARTING);
break;
case STATE_STARTING:
break;
case STATE_CLOSED:
sp->rst_counter[cp->protoidx] = sp->lcp.max_configure;
(cp->scr)(sp);
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
break;
case STATE_STOPPED:
case STATE_STOPPING:
case STATE_REQ_SENT:
case STATE_ACK_RCVD:
case STATE_ACK_SENT:
case STATE_OPENED:
break;
case STATE_CLOSING:
sppp_cp_change_state(cp, sp, STATE_STOPPING);
break;
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
static void
sppp_close_event(const struct cp *cp, struct sppp *sp)
{
STDDCL;
if (debug)
log(LOG_DEBUG, SPP_FMT "%s close(%s)\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]));
switch (sp->state[cp->protoidx]) {
case STATE_INITIAL:
case STATE_CLOSED:
case STATE_CLOSING:
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_STARTING:
(cp->tlf)(sp);
sppp_cp_change_state(cp, sp, STATE_INITIAL);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_STOPPED:
sppp_cp_change_state(cp, sp, STATE_CLOSED);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_STOPPING:
sppp_cp_change_state(cp, sp, STATE_CLOSING);
break;
case STATE_OPENED:
(cp->tld)(sp);
/* fall through */
case STATE_REQ_SENT:
case STATE_ACK_RCVD:
case STATE_ACK_SENT:
sp->rst_counter[cp->protoidx] = sp->lcp.max_terminate;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_cp_send(sp, cp->proto, TERM_REQ,
++sp->pp_seq[cp->protoidx], 0, 0);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_cp_change_state(cp, sp, STATE_CLOSING);
break;
}
}
static void
sppp_to_event(const struct cp *cp, struct sppp *sp)
{
STDDCL;
int s;
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG, SPP_FMT "%s TO(%s) rst_counter = %d\n",
SPP_ARGS(ifp), cp->name,
sppp_state_name(sp->state[cp->protoidx]),
sp->rst_counter[cp->protoidx]);
if (--sp->rst_counter[cp->protoidx] < 0)
/* TO- event */
switch (sp->state[cp->protoidx]) {
case STATE_CLOSING:
(cp->tlf)(sp);
sppp_cp_change_state(cp, sp, STATE_CLOSED);
Fix a slight bug introduced with revision 1.9 (IPv6 integration) where the bit mask of open NCPs got out of sync. Defer the (potential) closing of LCP after a NCP went down until after the state machines got updated. This fixes PR kern/11161.
2001-07-17 23:12:02 +04:00
sppp_lcp_check_and_close(sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
case STATE_STOPPING:
(cp->tlf)(sp);
sppp_cp_change_state(cp, sp, STATE_STOPPED);
Fix a slight bug introduced with revision 1.9 (IPv6 integration) where the bit mask of open NCPs got out of sync. Defer the (potential) closing of LCP after a NCP went down until after the state machines got updated. This fixes PR kern/11161.
2001-07-17 23:12:02 +04:00
sppp_lcp_check_and_close(sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
case STATE_REQ_SENT:
case STATE_ACK_RCVD:
case STATE_ACK_SENT:
(cp->tlf)(sp);
sppp_cp_change_state(cp, sp, STATE_STOPPED);
Fix a slight bug introduced with revision 1.9 (IPv6 integration) where the bit mask of open NCPs got out of sync. Defer the (potential) closing of LCP after a NCP went down until after the state machines got updated. This fixes PR kern/11161.
2001-07-17 23:12:02 +04:00
sppp_lcp_check_and_close(sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
else
/* TO+ event */
switch (sp->state[cp->protoidx]) {
case STATE_CLOSING:
case STATE_STOPPING:
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_cp_send(sp, cp->proto, TERM_REQ,
++sp->pp_seq[cp->protoidx], 0, 0);
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_reset(&sp->ch[cp->protoidx], sp->lcp.timeout,
cp->TO, sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_REQ_SENT:
case STATE_ACK_RCVD:
(cp->scr)(sp);
/* sppp_cp_change_state() will restart the timer */
sppp_cp_change_state(cp, sp, STATE_REQ_SENT);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
case STATE_ACK_SENT:
(cp->scr)(sp);
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_reset(&sp->ch[cp->protoidx], sp->lcp.timeout,
cp->TO, sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
splx(s);
}
/*
* Change the state of a control protocol in the state automaton.
* Takes care of starting/stopping the restart timer.
*/
void
sppp_cp_change_state(const struct cp *cp, struct sppp *sp, int newstate)
{
sp->state[cp->protoidx] = newstate;
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_stop(&sp->ch[cp->protoidx]);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (newstate) {
case STATE_INITIAL:
case STATE_STARTING:
case STATE_CLOSED:
case STATE_STOPPED:
case STATE_OPENED:
break;
case STATE_CLOSING:
case STATE_STOPPING:
case STATE_REQ_SENT:
case STATE_ACK_RCVD:
case STATE_ACK_SENT:
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_reset(&sp->ch[cp->protoidx], sp->lcp.timeout,
cp->TO, sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
break;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
/*
*--------------------------------------------------------------------------*
* *
* The LCP implementation. *
* *
*--------------------------------------------------------------------------*
*/
static void
sppp_lcp_init(struct sppp *sp)
{
sp->lcp.opts = (1 << LCP_OPT_MAGIC);
sp->lcp.magic = 0;
sp->state[IDX_LCP] = STATE_INITIAL;
sp->fail_counter[IDX_LCP] = 0;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->pp_seq[IDX_LCP] = 0;
sp->pp_rseq[IDX_LCP] = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->lcp.protos = 0;
sp->lcp.mru = sp->lcp.their_mru = PP_MTU;
/*
* Initialize counters and timeout values. Note that we don't
* use the 3 seconds suggested in RFC 1661 since we are likely
* running on a fast link. XXX We should probably implement
* the exponential backoff option. Note that these values are
* relevant for all control protocols, not just LCP only.
*/
sp->lcp.timeout = 1 * hz;
sp->lcp.max_terminate = 2;
sp->lcp.max_configure = 10;
sp->lcp.max_failure = 10;
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_init(&sp->ch[IDX_LCP]);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static void
sppp_lcp_up(struct sppp *sp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
Initialize the activity timestamp when opening a connection. Only idle- timeout connection that made it to phase NETWORK yet. (For drivers using the internal timeout mechanism; isdnd, that does the timeout handling for ISDN drivers, still needs to be fixed.) Thanks to Wolfgang Solfrank for finding this.
2002-01-14 10:39:14 +03:00
/* Initialize activity timestamp: opening a connection is an activity */
sp->pp_last_activity = time.tv_sec;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* If this interface is passive or dial-on-demand, and we are
* still in Initial state, it means we've got an incoming
* call. Activate the interface.
*/
if ((ifp->if_flags & (IFF_AUTO | IFF_PASSIVE)) != 0) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "Up event", SPP_ARGS(ifp));
ifp->if_flags |= IFF_RUNNING;
if (sp->state[IDX_LCP] == STATE_INITIAL) {
if (debug)
addlog("(incoming call)\n");
sp->pp_flags |= PP_CALLIN;
lcp.Open(sp);
} else if (debug)
addlog("\n");
We explicitly close LCP when going to state CLOSED, so we better open it again when going from INITIAL to STARTING. This has been done for passive or auto-conecting interfaces always, but not for permanent ones. This fixes session reestablishement for PPPoE interfaces without LINK1 set, and probably also closes PR kern/11161. Thanks to Jared D. McNeill and Ross Harvey for sugesting debug methology.
2001-12-10 03:22:21 +03:00
} else if ((ifp->if_flags & (IFF_AUTO | IFF_PASSIVE)) == 0 &&
(sp->state[IDX_LCP] == STATE_INITIAL)) {
ifp->if_flags |= IFF_RUNNING;
lcp.Open(sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_up_event(&lcp, sp);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
static void
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_lcp_down(struct sppp *sp)
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
{
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
STDDCL;
sppp_down_event(&lcp, sp);
/*
* If this is neither a dial-on-demand nor a passive
* interface, simulate an ``ifconfig down'' action, so the
* administrator can force a redial by another ``ifconfig
* up''. XXX For leased line operation, should we immediately
* try to reopen the connection here?
*/
if ((ifp->if_flags & (IFF_AUTO | IFF_PASSIVE)) == 0) {
log(LOG_INFO,
SPP_FMT "Down event (carrier loss), taking interface down.\n",
SPP_ARGS(ifp));
if_down(ifp);
} else {
if (debug)
log(LOG_DEBUG,
SPP_FMT "Down event (carrier loss)\n",
SPP_ARGS(ifp));
}
sp->pp_flags &= ~PP_CALLIN;
if (sp->state[IDX_LCP] != STATE_INITIAL)
lcp.Close(sp);
ifp->if_flags &= ~IFF_RUNNING;
}
static void
sppp_lcp_open(struct sppp *sp)
{
/*
* If we are authenticator, negotiate LCP_AUTH
*/
if (sp->hisauth.proto != 0)
sp->lcp.opts |= (1 << LCP_OPT_AUTH_PROTO);
else
sp->lcp.opts &= ~(1 << LCP_OPT_AUTH_PROTO);
sp->pp_flags &= ~PP_NEEDAUTH;
sppp_open_event(&lcp, sp);
}
static void
sppp_lcp_close(struct sppp *sp)
{
sppp_close_event(&lcp, sp);
}
static void
sppp_lcp_TO(void *cookie)
{
sppp_to_event(&lcp, (struct sppp *)cookie);
}
/*
* Analyze a configure request. Return true if it was agreeable, and
* caused action sca, false if it has been rejected or nak'ed, and
* caused action scn. (The return value is used to make the state
* transition decision in the state automaton.)
*/
static int
sppp_lcp_RCR(struct sppp *sp, struct lcp_header *h, int len)
{
STDDCL;
u_char *buf, *r, *p;
int origlen, rlen;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t nmagic;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
u_short authproto;
len -= 4;
origlen = len;
buf = r = malloc (len, M_TEMP, M_NOWAIT);
if (! buf)
return (0);
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
log(LOG_DEBUG, SPP_FMT "lcp parse opts:",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp));
/* pass 1: check for things that need to be rejected */
p = (void*) (h+1);
for (rlen=0; len>1 && p[1]; len-=p[1], p+=p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_lcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
case LCP_OPT_MAGIC:
/* Magic number. */
/* fall through, both are same length */
case LCP_OPT_ASYNC_MAP:
/* Async control character map. */
if (len >= 6 || p[1] == 6)
continue;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [invalid]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
case LCP_OPT_MRU:
/* Maximum receive unit. */
if (len >= 4 && p[1] == 4)
continue;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [invalid]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
case LCP_OPT_AUTH_PROTO:
if (len < 4) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [invalid]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
}
authproto = (p[2] << 8) + p[3];
if (authproto == PPP_CHAP && p[1] != 5) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [invalid chap len]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
}
if (sp->myauth.proto == 0) {
/* we are not configured to do auth */
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [not configured]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
}
/*
* Remote want us to authenticate, remember this,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
* so we stay in SPPP_PHASE_AUTHENTICATE after LCP got
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* up.
*/
sp->pp_flags |= PP_NEEDAUTH;
continue;
default:
/* Others not supported. */
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [rej]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
}
/* Add the option to rejected list. */
bcopy (p, r, p[1]);
r += p[1];
rlen += p[1];
}
if (rlen) {
if (debug)
addlog(" send conf-rej\n");
sppp_cp_send (sp, PPP_LCP, CONF_REJ, h->ident, rlen, buf);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
goto end;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
} else if (debug)
addlog("\n");
/*
* pass 2: check for option values that are unacceptable and
* thus require to be nak'ed.
*/
if (debug)
log(LOG_DEBUG, SPP_FMT "lcp parse opt values: ",
SPP_ARGS(ifp));
p = (void*) (h+1);
len = origlen;
for (rlen=0; len>1 && p[1]; len-=p[1], p+=p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_lcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
case LCP_OPT_MAGIC:
/* Magic number -- extract. */
code cleanup for portability
2001-12-05 00:32:15 +03:00
nmagic = (u_int32_t)p[2] << 24 |
(u_int32_t)p[3] << 16 | p[4] << 8 | p[5];
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (nmagic != sp->lcp.magic) {
if (debug)
code cleanup for portability
2001-12-05 00:32:15 +03:00
addlog(" 0x%x", nmagic);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
continue;
}
/*
* Local and remote magics equal -- loopback?
*/
if (sp->pp_loopcnt >= MAXALIVECNT*5) {
printf (SPP_FMT "loopback\n",
SPP_ARGS(ifp));
sp->pp_loopcnt = 0;
if (ifp->if_flags & IFF_UP) {
if_down(ifp);
IFQ_PURGE cannot be used against ifqueue. use IF_PURGE.
2001-08-23 04:56:03 +04:00
IF_PURGE(&sp->pp_cpq);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* XXX ? */
lcp.Down(sp);
lcp.Up(sp);
}
} else if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [glitch]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
++sp->pp_loopcnt;
/*
* We negate our magic here, and NAK it. If
* we see it later in an NAK packet, we
* suggest a new one.
*/
nmagic = ~sp->lcp.magic;
/* Gonna NAK it. */
p[2] = nmagic >> 24;
p[3] = nmagic >> 16;
p[4] = nmagic >> 8;
p[5] = nmagic;
break;
case LCP_OPT_ASYNC_MAP:
/* Async control character map -- check to be zero. */
if (! p[2] && ! p[3] && ! p[4] && ! p[5]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [empty]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
continue;
}
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [non-empty]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* suggest a zero one */
p[2] = p[3] = p[4] = p[5] = 0;
break;
case LCP_OPT_MRU:
/*
* Maximum receive unit. Always agreeable,
* but ignored by now.
*/
sp->lcp.their_mru = p[2] * 256 + p[3];
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %ld", sp->lcp.their_mru);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
continue;
case LCP_OPT_AUTH_PROTO:
authproto = (p[2] << 8) + p[3];
if (sp->myauth.proto != authproto) {
/* not agreed, nak */
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [mine %s != his %s]",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_proto_name(sp->hisauth.proto),
sppp_proto_name(authproto));
p[2] = sp->myauth.proto >> 8;
p[3] = sp->myauth.proto;
break;
}
if (authproto == PPP_CHAP && p[4] != CHAP_MD5) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [chap not MD5]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
p[4] = CHAP_MD5;
break;
}
continue;
}
/* Add the option to nak'ed list. */
bcopy (p, r, p[1]);
r += p[1];
rlen += p[1];
}
if (rlen) {
if (++sp->fail_counter[IDX_LCP] >= sp->lcp.max_failure) {
if (debug)
addlog(" max_failure (%d) exceeded, "
"send conf-rej\n",
sp->lcp.max_failure);
sppp_cp_send(sp, PPP_LCP, CONF_REJ, h->ident, rlen, buf);
} else {
if (debug)
addlog(" send conf-nak\n");
sppp_cp_send (sp, PPP_LCP, CONF_NAK, h->ident, rlen, buf);
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
goto end;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
} else {
if (debug)
addlog(" send conf-ack\n");
sp->fail_counter[IDX_LCP] = 0;
sp->pp_loopcnt = 0;
sppp_cp_send (sp, PPP_LCP, CONF_ACK,
h->ident, origlen, h+1);
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
end:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
free (buf, M_TEMP);
return (rlen == 0);
port FreeBSD's serial ppp layer to NetBSD. The PPP part seems broken still, but the lmc driver uses the HDLC bits from here anyway.
1999-03-25 06:38:00 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Analyze the LCP Configure-Reject option list, and adjust our
* negotiation.
*/
static void
sppp_lcp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len)
{
STDDCL;
u_char *buf, *p;
len -= 4;
buf = malloc (len, M_TEMP, M_NOWAIT);
if (!buf)
return;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
log(LOG_DEBUG, SPP_FMT "lcp rej opts:",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp));
p = (void*) (h+1);
for (; len > 1 && p[1]; len -= p[1], p += p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_lcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
case LCP_OPT_MAGIC:
/* Magic number -- can't use it, use 0 */
sp->lcp.opts &= ~(1 << LCP_OPT_MAGIC);
sp->lcp.magic = 0;
break;
case LCP_OPT_MRU:
/*
* Should not be rejected anyway, since we only
* negotiate a MRU if explicitly requested by
* peer.
*/
sp->lcp.opts &= ~(1 << LCP_OPT_MRU);
break;
case LCP_OPT_AUTH_PROTO:
/*
* Peer doesn't want to authenticate himself,
* deny unless this is a dialout call, and
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
* SPPP_AUTHFLAG_NOCALLOUT is set.
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
*/
if ((sp->pp_flags & PP_CALLIN) == 0 &&
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
(sp->hisauth.flags & SPPP_AUTHFLAG_NOCALLOUT) != 0) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [don't insist on auth "
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
"for callout]");
sp->lcp.opts &= ~(1 << LCP_OPT_AUTH_PROTO);
break;
}
if (debug)
addlog("[access denied]\n");
lcp.Close(sp);
break;
}
}
if (debug)
addlog("\n");
free (buf, M_TEMP);
return;
}
/*
* Analyze the LCP Configure-NAK option list, and adjust our
* negotiation.
*/
static void
sppp_lcp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len)
{
STDDCL;
u_char *buf, *p;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t magic;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
len -= 4;
buf = malloc (len, M_TEMP, M_NOWAIT);
if (!buf)
return;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
log(LOG_DEBUG, SPP_FMT "lcp nak opts:",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp));
p = (void*) (h+1);
for (; len > 1 && p[1]; len -= p[1], p += p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_lcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
case LCP_OPT_MAGIC:
/* Magic number -- renegotiate */
if ((sp->lcp.opts & (1 << LCP_OPT_MAGIC)) &&
len >= 6 && p[1] == 6) {
code cleanup for portability
2001-12-05 00:32:15 +03:00
magic = (u_int32_t)p[2] << 24 |
(u_int32_t)p[3] << 16 | p[4] << 8 | p[5];
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* If the remote magic is our negated one,
* this looks like a loopback problem.
* Suggest a new magic to make sure.
*/
if (magic == ~sp->lcp.magic) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" magic glitch");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->lcp.magic = random();
} else {
sp->lcp.magic = magic;
if (debug)
code cleanup for portability
2001-12-05 00:32:15 +03:00
addlog(" %d", magic);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
}
break;
case LCP_OPT_MRU:
/*
* Peer wants to advise us to negotiate an MRU.
* Agree on it if it's reasonable, or use
* default otherwise.
*/
if (len >= 4 && p[1] == 4) {
u_int mru = p[2] * 256 + p[3];
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %d", mru);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (mru < PP_MTU || mru > PP_MAX_MRU)
mru = PP_MTU;
sp->lcp.mru = mru;
sp->lcp.opts |= (1 << LCP_OPT_MRU);
}
break;
case LCP_OPT_AUTH_PROTO:
/*
* Peer doesn't like our authentication method,
* deny.
*/
if (debug)
addlog("[access denied]\n");
lcp.Close(sp);
break;
}
}
if (debug)
addlog("\n");
free (buf, M_TEMP);
return;
}
static void
sppp_lcp_tlu(struct sppp *sp)
{
STDDCL;
int i;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t mask;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* XXX ? */
if (! (ifp->if_flags & IFF_UP) &&
(ifp->if_flags & IFF_RUNNING)) {
/* Coming out of loopback mode. */
if_up(ifp);
}
for (i = 0; i < IDX_COUNT; i++)
if ((cps[i])->flags & CP_QUAL)
(cps[i])->Open(sp);
if ((sp->lcp.opts & (1 << LCP_OPT_AUTH_PROTO)) != 0 ||
(sp->pp_flags & PP_NEEDAUTH) != 0)
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_AUTHENTICATE;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
else
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_NETWORK;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(debug)
{
log(LOG_INFO, SPP_FMT "phase %s\n", SPP_ARGS(ifp),
sppp_phase_name(sp->pp_phase));
}
/*
* Open all authentication protocols. This is even required
* if we already proceeded to network phase, since it might be
* that remote wants us to authenticate, so we might have to
* send a PAP request. Undesired authentication protocols
* don't do anything when they get an Open event.
*/
for (i = 0; i < IDX_COUNT; i++)
if ((cps[i])->flags & CP_AUTH)
(cps[i])->Open(sp);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* Notify all NCPs. */
for (i = 0; i < IDX_COUNT; i++)
if ((cps[i])->flags & CP_NCP)
(cps[i])->Open(sp);
}
/* Send Up events to all started protos. */
for (i = 0, mask = 1; i < IDX_COUNT; i++, mask <<= 1)
fix operator precedence (& and &&). do not transmit too much message from LCP layer to NCP layer. PR 11161.
2000-10-08 06:37:42 +04:00
if ((sp->lcp.protos & mask) && ((cps[i])->flags & CP_LCP) == 0)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
(cps[i])->Up(sp);
/* notify low-level driver of state change */
if (sp->pp_chg)
sp->pp_chg(sp, (int)sp->pp_phase);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase == SPPP_PHASE_NETWORK)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* if no NCP is starting, close down */
sppp_lcp_check_and_close(sp);
}
static void
sppp_lcp_tld(struct sppp *sp)
{
STDDCL;
int i;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t mask;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_TERMINATE;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(debug)
{
log(LOG_INFO, SPP_FMT "phase %s\n", SPP_ARGS(ifp),
sppp_phase_name(sp->pp_phase));
}
/*
* Take upper layers down. We send the Down event first and
* the Close second to prevent the upper layers from sending
* ``a flurry of terminate-request packets'', as the RFC
* describes it.
*/
for (i = 0, mask = 1; i < IDX_COUNT; i++, mask <<= 1)
fix operator precedence (& and &&). do not transmit too much message from LCP layer to NCP layer. PR 11161.
2000-10-08 06:37:42 +04:00
if ((sp->lcp.protos & mask) && ((cps[i])->flags & CP_LCP) == 0) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
(cps[i])->Down(sp);
(cps[i])->Close(sp);
}
}
static void
sppp_lcp_tls(struct sppp *sp)
{
STDDCL;
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
if (sp->pp_max_auth_fail != 0 && sp->pp_auth_failures >= sp->pp_max_auth_fail) {
printf("%s: authentication failed %d times, not retrying again\n",
sp->pp_if.if_xname, sp->pp_auth_failures);
if_down(&sp->pp_if);
return;
}
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_ESTABLISH;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(debug)
{
log(LOG_INFO, SPP_FMT "phase %s\n", SPP_ARGS(ifp),
sppp_phase_name(sp->pp_phase));
}
/* Notify lower layer if desired. */
if (sp->pp_tls)
(sp->pp_tls)(sp);
}
static void
sppp_lcp_tlf(struct sppp *sp)
{
STDDCL;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_DEAD;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(debug)
{
log(LOG_INFO, SPP_FMT "phase %s\n", SPP_ARGS(ifp),
sppp_phase_name(sp->pp_phase));
}
/* Notify lower layer if desired. */
if (sp->pp_tlf)
(sp->pp_tlf)(sp);
}
static void
sppp_lcp_scr(struct sppp *sp)
{
char opt[6 /* magicnum */ + 4 /* mru */ + 5 /* chap */];
int i = 0;
u_short authproto;
if (sp->lcp.opts & (1 << LCP_OPT_MAGIC)) {
if (! sp->lcp.magic)
sp->lcp.magic = random();
opt[i++] = LCP_OPT_MAGIC;
opt[i++] = 6;
opt[i++] = sp->lcp.magic >> 24;
opt[i++] = sp->lcp.magic >> 16;
opt[i++] = sp->lcp.magic >> 8;
opt[i++] = sp->lcp.magic;
}
if (sp->lcp.opts & (1 << LCP_OPT_MRU)) {
opt[i++] = LCP_OPT_MRU;
opt[i++] = 4;
opt[i++] = sp->lcp.mru >> 8;
opt[i++] = sp->lcp.mru;
}
if (sp->lcp.opts & (1 << LCP_OPT_AUTH_PROTO)) {
authproto = sp->hisauth.proto;
opt[i++] = LCP_OPT_AUTH_PROTO;
opt[i++] = authproto == PPP_CHAP? 5: 4;
opt[i++] = authproto >> 8;
opt[i++] = authproto;
if (authproto == PPP_CHAP)
opt[i++] = CHAP_MD5;
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->confid[IDX_LCP] = ++sp->pp_seq[IDX_LCP];
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_cp_send (sp, PPP_LCP, CONF_REQ, sp->confid[IDX_LCP], i, &opt);
}
/*
* Check the open NCPs, return true if at least one NCP is open.
*/
static int
sppp_ncp_check(struct sppp *sp)
{
int i, mask;
for (i = 0, mask = 1; i < IDX_COUNT; i++, mask <<= 1)
fix operator precedence (& and &&). do not transmit too much message from LCP layer to NCP layer. PR 11161.
2000-10-08 06:37:42 +04:00
if ((sp->lcp.protos & mask) && (cps[i])->flags & CP_NCP)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
return 1;
return 0;
}
/*
* Re-check the open NCPs and see if we should terminate the link.
* Called by the NCPs during their tlf action handling.
*/
static void
sppp_lcp_check_and_close(struct sppp *sp)
{
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase < SPPP_PHASE_NETWORK)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* don't bother, we are already going down */
return;
if (sppp_ncp_check(sp))
return;
lcp.Close(sp);
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
*--------------------------------------------------------------------------*
* *
* The IPCP implementation. *
* *
*--------------------------------------------------------------------------*
*/
static void
sppp_ipcp_init(struct sppp *sp)
{
sp->ipcp.opts = 0;
sp->ipcp.flags = 0;
sp->state[IDX_IPCP] = STATE_INITIAL;
sp->fail_counter[IDX_IPCP] = 0;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->pp_seq[IDX_IPCP] = 0;
sp->pp_rseq[IDX_IPCP] = 0;
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_init(&sp->ch[IDX_IPCP]);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
static void
sppp_ipcp_up(struct sppp *sp)
{
sppp_up_event(&ipcp, sp);
}
static void
sppp_ipcp_down(struct sppp *sp)
{
sppp_down_event(&ipcp, sp);
}
static void
sppp_ipcp_open(struct sppp *sp)
{
STDDCL;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t myaddr, hisaddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
sp->ipcp.flags &= ~(IPCP_HISADDR_SEEN|IPCP_MYADDR_SEEN|IPCP_MYADDR_DYN|IPCP_HISADDR_DYN);
sp->ipcp.req_myaddr = 0;
sp->ipcp.req_hisaddr = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_get_ip_addrs(sp, &myaddr, &hisaddr, 0);
/*
* If we don't have his address, this probably means our
* interface doesn't want to talk IP at all. (This could
* be the case if somebody wants to speak only IPX, for
* example.) Don't open IPCP in this case.
*/
if (hisaddr == 0L) {
/* XXX this message should go away */
if (debug)
log(LOG_DEBUG, SPP_FMT "ipcp_open(): no IP interface\n",
SPP_ARGS(ifp));
return;
}
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
if (myaddr == 0) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* I don't have an assigned address, so i need to
* negotiate my address.
*/
sp->ipcp.flags |= IPCP_MYADDR_DYN;
sp->ipcp.opts |= (1 << IPCP_OPT_ADDRESS);
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
}
if (hisaddr == 1) {
/*
* XXX - remove this hack!
* remote has no valid adress, we need to get one assigned.
*/
sp->ipcp.flags |= IPCP_HISADDR_DYN;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_open_event(&ipcp, sp);
}
static void
sppp_ipcp_close(struct sppp *sp)
{
sppp_close_event(&ipcp, sp);
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
if (sp->ipcp.flags & (IPCP_MYADDR_DYN|IPCP_HISADDR_DYN))
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
* Some address was dynamic, clear it again.
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
*/
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
sppp_clear_ip_addrs(sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
static void
sppp_ipcp_TO(void *cookie)
{
sppp_to_event(&ipcp, (struct sppp *)cookie);
}
/*
* Analyze a configure request. Return true if it was agreeable, and
* caused action sca, false if it has been rejected or nak'ed, and
* caused action scn. (The return value is used to make the state
* transition decision in the state automaton.)
*/
static int
sppp_ipcp_RCR(struct sppp *sp, struct lcp_header *h, int len)
{
u_char *buf, *r, *p;
struct ifnet *ifp = &sp->pp_if;
int rlen, origlen, debug = ifp->if_flags & IFF_DEBUG;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t hisaddr, desiredaddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
len -= 4;
origlen = len;
/*
* Make sure to allocate a buf that can at least hold a
* conf-nak with an `address' option. We might need it below.
*/
buf = r = malloc ((len < 6? 6: len), M_TEMP, M_NOWAIT);
if (! buf)
return (0);
/* pass 1: see if we can recognize them */
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
log(LOG_DEBUG, SPP_FMT "ipcp parse opts:",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp));
p = (void*) (h+1);
for (rlen=0; len>1 && p[1]; len-=p[1], p+=p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_ipcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
#ifdef notyet
case IPCP_OPT_COMPRESSION:
if (len >= 6 && p[1] >= 6) {
/* correctly formed compress option */
continue;
}
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [invalid]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
#endif
case IPCP_OPT_ADDRESS:
if (len >= 6 && p[1] == 6) {
/* correctly formed address option */
continue;
}
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [invalid]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
default:
/* Others not supported. */
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [rej]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
break;
}
/* Add the option to rejected list. */
bcopy (p, r, p[1]);
r += p[1];
rlen += p[1];
}
if (rlen) {
if (debug)
addlog(" send conf-rej\n");
sppp_cp_send (sp, PPP_IPCP, CONF_REJ, h->ident, rlen, buf);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
goto end;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
} else if (debug)
addlog("\n");
/* pass 2: parse option values */
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
if (sp->ipcp.flags & IPCP_HISADDR_SEEN)
hisaddr = sp->ipcp.req_hisaddr; /* we already aggreed on that */
else
sppp_get_ip_addrs(sp, 0, &hisaddr, 0); /* user configuration */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG, SPP_FMT "ipcp parse opt values: ",
SPP_ARGS(ifp));
p = (void*) (h+1);
len = origlen;
for (rlen=0; len>1 && p[1]; len-=p[1], p+=p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_ipcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
#ifdef notyet
case IPCP_OPT_COMPRESSION:
continue;
#endif
case IPCP_OPT_ADDRESS:
desiredaddr = p[2] << 24 | p[3] << 16 |
p[4] << 8 | p[5];
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
if (desiredaddr == hisaddr ||
((sp->ipcp.flags & IPCP_HISADDR_DYN) && desiredaddr != 0)) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
* Peer's address is same as our value,
* this is agreeable. Gonna conf-ack
* it.
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
addlog(" %s [ack]",
sppp_dotted_quad(hisaddr));
/* record that we've seen it already */
sp->ipcp.flags |= IPCP_HISADDR_SEEN;
sp->ipcp.req_hisaddr = desiredaddr;
hisaddr = desiredaddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
continue;
}
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
/*
* The address wasn't agreeable. This is either
* he sent us 0.0.0.0, asking to assign him an
* address, or he send us another address not
* matching our value. Either case, we gonna
* conf-nak it with our value.
*/
if (debug) {
if (desiredaddr == 0)
addlog(" [addr requested]");
else
addlog(" %s [not agreed]",
sppp_dotted_quad(desiredaddr));
}
p[2] = hisaddr >> 24;
p[3] = hisaddr >> 16;
p[4] = hisaddr >> 8;
p[5] = hisaddr;
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
/* Add the option to nak'ed list. */
bcopy (p, r, p[1]);
r += p[1];
rlen += p[1];
}
/*
* If we are about to conf-ack the request, but haven't seen
* his address so far, gonna conf-nak it instead, with the
* `address' option present and our idea of his address being
* filled in there, to request negotiation of both addresses.
*
* XXX This can result in an endless req - nak loop if peer
* doesn't want to send us his address. Q: What should we do
* about it? XXX A: implement the max-failure counter.
*/
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
if (rlen == 0 && !(sp->ipcp.flags & IPCP_HISADDR_SEEN)) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
buf[0] = IPCP_OPT_ADDRESS;
buf[1] = 6;
buf[2] = hisaddr >> 24;
buf[3] = hisaddr >> 16;
buf[4] = hisaddr >> 8;
buf[5] = hisaddr;
rlen = 6;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" still need hisaddr");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
if (rlen) {
if (debug)
addlog(" send conf-nak\n");
sppp_cp_send (sp, PPP_IPCP, CONF_NAK, h->ident, rlen, buf);
} else {
if (debug)
addlog(" send conf-ack\n");
sppp_cp_send (sp, PPP_IPCP, CONF_ACK,
h->ident, origlen, h+1);
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
end:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
free (buf, M_TEMP);
return (rlen == 0);
}
/*
* Analyze the IPCP Configure-Reject option list, and adjust our
* negotiation.
*/
static void
sppp_ipcp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len)
{
u_char *buf, *p;
struct ifnet *ifp = &sp->pp_if;
int debug = ifp->if_flags & IFF_DEBUG;
len -= 4;
buf = malloc (len, M_TEMP, M_NOWAIT);
if (!buf)
return;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
log(LOG_DEBUG, SPP_FMT "ipcp rej opts:",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp));
p = (void*) (h+1);
for (; len > 1 && p[1]; len -= p[1], p += p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_ipcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
case IPCP_OPT_ADDRESS:
/*
* Peer doesn't grok address option. This is
* bad. XXX Should we better give up here?
*/
sp->ipcp.opts &= ~(1 << IPCP_OPT_ADDRESS);
break;
#ifdef notyet
case IPCP_OPT_COMPRESS:
sp->ipcp.opts &= ~(1 << IPCP_OPT_COMPRESS);
break;
#endif
}
}
if (debug)
addlog("\n");
free (buf, M_TEMP);
return;
}
/*
* Analyze the IPCP Configure-NAK option list, and adjust our
* negotiation.
*/
static void
sppp_ipcp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len)
{
u_char *buf, *p;
struct ifnet *ifp = &sp->pp_if;
int debug = ifp->if_flags & IFF_DEBUG;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t wantaddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
len -= 4;
buf = malloc (len, M_TEMP, M_NOWAIT);
if (!buf)
return;
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
log(LOG_DEBUG, SPP_FMT "ipcp nak opts:",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp));
p = (void*) (h+1);
for (; len > 1 && p[1]; len -= p[1], p += p[1]) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" %s", sppp_ipcp_opt_name(*p));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
switch (*p) {
case IPCP_OPT_ADDRESS:
/*
* Peer doesn't like our local IP address. See
* if we can do something for him. We'll drop
* him our address then.
*/
if (len >= 6 && p[1] == 6) {
wantaddr = p[2] << 24 | p[3] << 16 |
p[4] << 8 | p[5];
sp->ipcp.opts |= (1 << IPCP_OPT_ADDRESS);
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [wantaddr %s]",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_dotted_quad(wantaddr));
/*
* When doing dynamic address assignment,
* we accept his offer. Otherwise, we
* ignore it and thus continue to negotiate
* our already existing value.
*/
if (sp->ipcp.flags & IPCP_MYADDR_DYN) {
if (debug)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
addlog(" [agree]");
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->ipcp.flags |= IPCP_MYADDR_SEEN;
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
sp->ipcp.req_myaddr = wantaddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
}
break;
#ifdef notyet
case IPCP_OPT_COMPRESS:
/*
* Peer wants different compression parameters.
*/
break;
#endif
}
}
if (debug)
addlog("\n");
free (buf, M_TEMP);
return;
}
static void
sppp_ipcp_tlu(struct sppp *sp)
{
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
/* we are up. Set addresses and notify anyone interested */
u_int32_t myaddr, hisaddr;
sppp_get_ip_addrs(sp, &myaddr, &hisaddr, 0);
if ((sp->ipcp.flags & IPCP_MYADDR_DYN) && (sp->ipcp.flags & IPCP_MYADDR_SEEN))
myaddr = sp->ipcp.req_myaddr;
if ((sp->ipcp.flags & IPCP_HISADDR_DYN) && (sp->ipcp.flags & IPCP_HISADDR_SEEN))
hisaddr = sp->ipcp.req_hisaddr;
sppp_set_ip_addrs(sp, myaddr, hisaddr);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (sp->pp_con)
sp->pp_con(sp);
}
static void
sppp_ipcp_tld(struct sppp *sp)
{
}
static void
sppp_ipcp_tls(struct sppp *sp)
{
/* indicate to LCP that it must stay alive */
sp->lcp.protos |= (1 << IDX_IPCP);
}
static void
sppp_ipcp_tlf(struct sppp *sp)
{
/* we no longer need LCP */
sp->lcp.protos &= ~(1 << IDX_IPCP);
}
static void
sppp_ipcp_scr(struct sppp *sp)
{
char opt[6 /* compression */ + 6 /* address */];
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t ouraddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int i = 0;
#ifdef notyet
if (sp->ipcp.opts & (1 << IPCP_OPT_COMPRESSION)) {
opt[i++] = IPCP_OPT_COMPRESSION;
opt[i++] = 6;
opt[i++] = 0; /* VJ header compression */
opt[i++] = 0x2d; /* VJ header compression */
opt[i++] = max_slot_id;
opt[i++] = comp_slot_id;
}
#endif
if (sp->ipcp.opts & (1 << IPCP_OPT_ADDRESS)) {
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
if (sp->ipcp.flags & IPCP_MYADDR_SEEN)
ouraddr = sp->ipcp.req_myaddr; /* not sure if this can ever happen */
else
sppp_get_ip_addrs(sp, &ouraddr, 0, 0);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
opt[i++] = IPCP_OPT_ADDRESS;
opt[i++] = 6;
opt[i++] = ouraddr >> 24;
opt[i++] = ouraddr >> 16;
opt[i++] = ouraddr >> 8;
opt[i++] = ouraddr;
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->confid[IDX_IPCP] = ++sp->pp_seq[IDX_IPCP];
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_cp_send(sp, PPP_IPCP, CONF_REQ, sp->confid[IDX_IPCP], i, &opt);
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
/*
*--------------------------------------------------------------------------*
* *
* The IPv6CP implementation. *
* *
*--------------------------------------------------------------------------*
*/
#ifdef INET6
static void
sppp_ipv6cp_init(struct sppp *sp)
{
sp->ipv6cp.opts = 0;
sp->ipv6cp.flags = 0;
sp->state[IDX_IPV6CP] = STATE_INITIAL;
sp->fail_counter[IDX_IPV6CP] = 0;
sp->pp_seq[IDX_IPV6CP] = 0;
sp->pp_rseq[IDX_IPV6CP] = 0;
callout_init(&sp->ch[IDX_IPV6CP]);
}
static void
sppp_ipv6cp_up(struct sppp *sp)
{
sppp_up_event(&ipv6cp, sp);
}
static void
sppp_ipv6cp_down(struct sppp *sp)
{
sppp_down_event(&ipv6cp, sp);
}
static void
sppp_ipv6cp_open(struct sppp *sp)
{
STDDCL;
struct in6_addr myaddr, hisaddr;
#ifdef IPV6CP_MYIFID_DYN
sp->ipv6cp.flags &= ~(IPV6CP_MYIFID_SEEN|IPV6CP_MYIFID_DYN);
#else
sp->ipv6cp.flags &= ~IPV6CP_MYIFID_SEEN;
#endif
sppp_get_ip6_addrs(sp, &myaddr, &hisaddr, 0);
/*
* If we don't have our address, this probably means our
* interface doesn't want to talk IPv6 at all. (This could
* be the case if somebody wants to speak only IPX, for
* example.) Don't open IPv6CP in this case.
*/
if (IN6_IS_ADDR_UNSPECIFIED(&myaddr)) {
/* XXX this message should go away */
if (debug)
log(LOG_DEBUG, SPP_FMT "ipv6cp_open(): no IPv6 interface\n",
SPP_ARGS(ifp));
return;
}
sp->ipv6cp.flags |= IPV6CP_MYIFID_SEEN;
sp->ipv6cp.opts |= (1 << IPV6CP_OPT_IFID);
sppp_open_event(&ipv6cp, sp);
}
static void
sppp_ipv6cp_close(struct sppp *sp)
{
sppp_close_event(&ipv6cp, sp);
}
static void
sppp_ipv6cp_TO(void *cookie)
{
sppp_to_event(&ipv6cp, (struct sppp *)cookie);
}
/*
* Analyze a configure request. Return true if it was agreeable, and
* caused action sca, false if it has been rejected or nak'ed, and
* caused action scn. (The return value is used to make the state
* transition decision in the state automaton.)
*/
static int
sppp_ipv6cp_RCR(struct sppp *sp, struct lcp_header *h, int len)
{
u_char *buf, *r, *p;
struct ifnet *ifp = &sp->pp_if;
int rlen, origlen, debug = ifp->if_flags & IFF_DEBUG;
struct in6_addr myaddr, desiredaddr, suggestaddr;
int ifidcount;
int type;
int collision, nohisaddr;
len -= 4;
origlen = len;
/*
* Make sure to allocate a buf that can at least hold a
* conf-nak with an `address' option. We might need it below.
*/
buf = r = malloc ((len < 6? 6: len), M_TEMP, M_NOWAIT);
if (! buf)
return (0);
/* pass 1: see if we can recognize them */
if (debug)
log(LOG_DEBUG, SPP_FMT "ipv6cp parse opts:",
SPP_ARGS(ifp));
p = (void*) (h+1);
ifidcount = 0;
for (rlen=0; len>1 && p[1]; len-=p[1], p+=p[1]) {
if (debug)
addlog(" %s", sppp_ipv6cp_opt_name(*p));
switch (*p) {
case IPV6CP_OPT_IFID:
if (len >= 10 && p[1] == 10 && ifidcount == 0) {
/* correctly formed address option */
ifidcount++;
continue;
}
if (debug)
addlog(" [invalid]");
break;
#ifdef notyet
case IPV6CP_OPT_COMPRESSION:
if (len >= 4 && p[1] >= 4) {
/* correctly formed compress option */
continue;
}
if (debug)
addlog(" [invalid]");
break;
#endif
default:
/* Others not supported. */
if (debug)
addlog(" [rej]");
break;
}
/* Add the option to rejected list. */
bcopy (p, r, p[1]);
r += p[1];
rlen += p[1];
}
if (rlen) {
if (debug)
addlog(" send conf-rej\n");
sppp_cp_send (sp, PPP_IPV6CP, CONF_REJ, h->ident, rlen, buf);
goto end;
} else if (debug)
addlog("\n");
/* pass 2: parse option values */
sppp_get_ip6_addrs(sp, &myaddr, 0, 0);
if (debug)
log(LOG_DEBUG, SPP_FMT "ipv6cp parse opt values: ",
SPP_ARGS(ifp));
p = (void*) (h+1);
len = origlen;
type = CONF_ACK;
for (rlen=0; len>1 && p[1]; len-=p[1], p+=p[1]) {
if (debug)
addlog(" %s", sppp_ipv6cp_opt_name(*p));
switch (*p) {
#ifdef notyet
case IPV6CP_OPT_COMPRESSION:
continue;
#endif
case IPV6CP_OPT_IFID:
bzero -> memset
2001-07-18 20:43:09 +04:00
memset(&desiredaddr, 0, sizeof(desiredaddr));
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
bcopy(&p[2], &desiredaddr.s6_addr[8], 8);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
collision = (memcmp(&desiredaddr.s6_addr[8],
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
&myaddr.s6_addr[8], 8) == 0);
nohisaddr = IN6_IS_ADDR_UNSPECIFIED(&desiredaddr);
desiredaddr.s6_addr16[0] = htons(0xfe80);
desiredaddr.s6_addr16[1] = htons(sp->pp_if.if_index);
if (!collision && !nohisaddr) {
/* no collision, hisaddr known - Conf-Ack */
type = CONF_ACK;
if (debug) {
addlog(" %s [%s]",
ip6_sprintf(&desiredaddr),
sppp_cp_type_name(type));
}
continue;
}
bzero -> memset
2001-07-18 20:43:09 +04:00
memset(&suggestaddr, 0, sizeof(&suggestaddr));
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
if (collision && nohisaddr) {
/* collision, hisaddr unknown - Conf-Rej */
type = CONF_REJ;
bzero -> memset
2001-07-18 20:43:09 +04:00
memset(&p[2], 0, 8);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
} else {
/*
* - no collision, hisaddr unknown, or
* - collision, hisaddr known
* Conf-Nak, suggest hisaddr
*/
type = CONF_NAK;
sppp_suggest_ip6_addr(sp, &suggestaddr);
bcopy(&suggestaddr.s6_addr[8], &p[2], 8);
}
if (debug)
addlog(" %s [%s]", ip6_sprintf(&desiredaddr),
sppp_cp_type_name(type));
break;
}
/* Add the option to nak'ed list. */
bcopy (p, r, p[1]);
r += p[1];
rlen += p[1];
}
if (rlen == 0 && type == CONF_ACK) {
if (debug)
addlog(" send %s\n", sppp_cp_type_name(type));
sppp_cp_send (sp, PPP_IPV6CP, type, h->ident, origlen, h+1);
} else {
#ifdef DIAGNOSTIC
if (type == CONF_ACK)
panic("IPv6CP RCR: CONF_ACK with non-zero rlen");
#endif
if (debug) {
addlog(" send %s suggest %s\n",
sppp_cp_type_name(type), ip6_sprintf(&suggestaddr));
}
sppp_cp_send (sp, PPP_IPV6CP, type, h->ident, rlen, buf);
}
end:
free (buf, M_TEMP);
return (rlen == 0);
}
/*
* Analyze the IPv6CP Configure-Reject option list, and adjust our
* negotiation.
*/
static void
sppp_ipv6cp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len)
{
u_char *buf, *p;
struct ifnet *ifp = &sp->pp_if;
int debug = ifp->if_flags & IFF_DEBUG;
len -= 4;
buf = malloc (len, M_TEMP, M_NOWAIT);
if (!buf)
return;
if (debug)
log(LOG_DEBUG, SPP_FMT "ipv6cp rej opts:",
SPP_ARGS(ifp));
p = (void*) (h+1);
for (; len > 1 && p[1]; len -= p[1], p += p[1]) {
if (debug)
addlog(" %s", sppp_ipv6cp_opt_name(*p));
switch (*p) {
case IPV6CP_OPT_IFID:
/*
* Peer doesn't grok address option. This is
* bad. XXX Should we better give up here?
*/
sp->ipv6cp.opts &= ~(1 << IPV6CP_OPT_IFID);
break;
#ifdef notyet
case IPV6CP_OPT_COMPRESS:
sp->ipv6cp.opts &= ~(1 << IPV6CP_OPT_COMPRESS);
break;
#endif
}
}
if (debug)
addlog("\n");
free (buf, M_TEMP);
return;
}
/*
* Analyze the IPv6CP Configure-NAK option list, and adjust our
* negotiation.
*/
static void
sppp_ipv6cp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len)
{
u_char *buf, *p;
struct ifnet *ifp = &sp->pp_if;
int debug = ifp->if_flags & IFF_DEBUG;
struct in6_addr suggestaddr;
len -= 4;
buf = malloc (len, M_TEMP, M_NOWAIT);
if (!buf)
return;
if (debug)
log(LOG_DEBUG, SPP_FMT "ipv6cp nak opts:",
SPP_ARGS(ifp));
p = (void*) (h+1);
for (; len > 1 && p[1]; len -= p[1], p += p[1]) {
if (debug)
addlog(" %s", sppp_ipv6cp_opt_name(*p));
switch (*p) {
case IPV6CP_OPT_IFID:
/*
* Peer doesn't like our local ifid. See
* if we can do something for him. We'll drop
* him our address then.
*/
if (len < 10 || p[1] != 10)
break;
bzero -> memset
2001-07-18 20:43:09 +04:00
memset(&suggestaddr, 0, sizeof(suggestaddr));
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
suggestaddr.s6_addr16[0] = htons(0xfe80);
suggestaddr.s6_addr16[1] = htons(sp->pp_if.if_index);
bcopy(&p[2], &suggestaddr.s6_addr[8], 8);
sp->ipv6cp.opts |= (1 << IPV6CP_OPT_IFID);
if (debug)
addlog(" [suggestaddr %s]",
ip6_sprintf(&suggestaddr));
#ifdef IPV6CP_MYIFID_DYN
/*
* When doing dynamic address assignment,
* we accept his offer.
*/
if (sp->ipv6cp.flags & IPV6CP_MYIFID_DYN) {
struct in6_addr lastsuggest;
/*
* If <suggested myaddr from peer> equals to
* <hisaddr we have suggested last time>,
* we have a collision. generate new random
* ifid.
*/
sppp_suggest_ip6_addr(&lastsuggest);
if (IN6_ARE_ADDR_EQUAL(&suggestaddr,
lastsuggest)) {
if (debug)
addlog(" [random]");
sppp_gen_ip6_addr(sp, &suggestaddr);
}
sppp_set_ip6_addr(sp, &suggestaddr, 0);
if (debug)
addlog(" [agree]");
sp->ipv6cp.flags |= IPV6CP_MYIFID_SEEN;
}
#else
/*
* Since we do not do dynamic address assignment,
* we ignore it and thus continue to negotiate
* our already existing value. This can possibly
* go into infinite request-reject loop.
*
* This is not likely because we normally use
* ifid based on MAC-address.
* If you have no ethernet card on the node, too bad.
* XXX should we use fail_counter?
*/
#endif
break;
#ifdef notyet
case IPV6CP_OPT_COMPRESS:
/*
* Peer wants different compression parameters.
*/
break;
#endif
}
}
if (debug)
addlog("\n");
free (buf, M_TEMP);
return;
}
static void
sppp_ipv6cp_tlu(struct sppp *sp)
{
/* we are up - notify isdn daemon */
if (sp->pp_con)
sp->pp_con(sp);
}
static void
sppp_ipv6cp_tld(struct sppp *sp)
{
}
static void
sppp_ipv6cp_tls(struct sppp *sp)
{
/* indicate to LCP that it must stay alive */
sp->lcp.protos |= (1 << IDX_IPV6CP);
}
static void
sppp_ipv6cp_tlf(struct sppp *sp)
{
/* we no longer need LCP */
sp->lcp.protos &= ~(1 << IDX_IPV6CP);
}
static void
sppp_ipv6cp_scr(struct sppp *sp)
{
char opt[10 /* ifid */ + 4 /* compression, minimum */];
struct in6_addr ouraddr;
int i = 0;
if (sp->ipv6cp.opts & (1 << IPV6CP_OPT_IFID)) {
sppp_get_ip6_addrs(sp, &ouraddr, 0, 0);
opt[i++] = IPV6CP_OPT_IFID;
opt[i++] = 10;
bcopy(&ouraddr.s6_addr[8], &opt[i], 8);
i += 8;
}
#ifdef notyet
if (sp->ipv6cp.opts & (1 << IPV6CP_OPT_COMPRESSION)) {
opt[i++] = IPV6CP_OPT_COMPRESSION;
opt[i++] = 4;
opt[i++] = 0; /* TBD */
opt[i++] = 0; /* TBD */
/* variable length data may follow */
}
#endif
sp->confid[IDX_IPV6CP] = ++sp->pp_seq[IDX_IPV6CP];
sppp_cp_send(sp, PPP_IPV6CP, CONF_REQ, sp->confid[IDX_IPV6CP], i, &opt);
}
#else /*INET6*/
static void sppp_ipv6cp_init(struct sppp *sp)
{
}
static void sppp_ipv6cp_up(struct sppp *sp)
{
}
static void sppp_ipv6cp_down(struct sppp *sp)
{
}
static void sppp_ipv6cp_open(struct sppp *sp)
{
}
static void sppp_ipv6cp_close(struct sppp *sp)
{
}
static void sppp_ipv6cp_TO(void *sp)
{
}
static int sppp_ipv6cp_RCR(struct sppp *sp, struct lcp_header *h, int len)
{
return 0;
}
static void sppp_ipv6cp_RCN_rej(struct sppp *sp, struct lcp_header *h, int len)
{
}
static void sppp_ipv6cp_RCN_nak(struct sppp *sp, struct lcp_header *h, int len)
{
}
static void sppp_ipv6cp_tlu(struct sppp *sp)
{
}
static void sppp_ipv6cp_tld(struct sppp *sp)
{
}
static void sppp_ipv6cp_tls(struct sppp *sp)
{
}
static void sppp_ipv6cp_tlf(struct sppp *sp)
{
}
static void sppp_ipv6cp_scr(struct sppp *sp)
{
}
#endif /*INET6*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
*--------------------------------------------------------------------------*
* *
* The CHAP implementation. *
* *
*--------------------------------------------------------------------------*
*/
/*
* The authentication protocols don't employ a full-fledged state machine as
* the control protocols do, since they do have Open and Close events, but
* not Up and Down, nor are they explicitly terminated. Also, use of the
* authentication protocols may be different in both directions (this makes
* sense, think of a machine that never accepts incoming calls but only
* calls out, it doesn't require the called party to authenticate itself).
*
* Our state machine for the local authentication protocol (we are requesting
* the peer to authenticate) looks like:
*
* RCA-
* +--------------------------------------------+
* V scn,tld|
* +--------+ Close +---------+ RCA+
* | |<----------------------------------| |------+
* +--->| Closed | TO* | Opened | sca |
* | | |-----+ +-------| |<-----+
* | +--------+ irc | | +---------+
* | ^ | | ^
* | | | | |
* | | | | |
* | TO-| | | |
* | |tld TO+ V | |
* | | +------->+ | |
* | | | | | |
* | +--------+ V | |
* | | |<----+<--------------------+ |
* | | Req- | scr |
* | | Sent | |
* | | | |
* | +--------+ |
* | RCA- | | RCA+ |
* +------+ +------------------------------------------+
* scn,tld sca,irc,ict,tlu
*
*
* with:
*
* Open: LCP reached authentication phase
* Close: LCP reached terminate phase
*
* RCA+: received reply (pap-req, chap-response), acceptable
* RCN: received reply (pap-req, chap-response), not acceptable
* TO+: timeout with restart counter >= 0
* TO-: timeout with restart counter < 0
* TO*: reschedule timeout for CHAP
*
* scr: send request packet (none for PAP, chap-challenge)
* sca: send ack packet (pap-ack, chap-success)
* scn: send nak packet (pap-nak, chap-failure)
* ict: initialize re-challenge timer (CHAP only)
*
* tlu: this-layer-up, LCP reaches network phase
* tld: this-layer-down, LCP enters terminate phase
*
* Note that in CHAP mode, after sending a new challenge, while the state
* automaton falls back into Req-Sent state, it doesn't signal a tld
* event to LCP, so LCP remains in network phase. Only after not getting
* any response (or after getting an unacceptable response), CHAP closes,
* causing LCP to enter terminate phase.
*
* With PAP, there is no initial request that can be sent. The peer is
* expected to send one based on the successful negotiation of PAP as
* the authentication protocol during the LCP option negotiation.
*
* Incoming authentication protocol requests (remote requests
* authentication, we are peer) don't employ a state machine at all,
* they are simply answered. Some peers [Ascend P50 firmware rev
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
* 4.50] react allergically when sending IPCP/IPv6CP requests while they are
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
* still in authentication phase (thereby violating the standard that
* demands that these NCP packets are to be discarded), so we keep
* track of the peer demanding us to authenticate, and only proceed to
* phase network once we've seen a positive acknowledge for the
* authentication.
*/
/*
* Handle incoming CHAP packets.
*/
void
sppp_chap_input(struct sppp *sp, struct mbuf *m)
{
STDDCL;
struct lcp_header *h;
int len, x;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
u_char *value, *name, digest[sizeof(sp->myauth.challenge)], dsize;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int value_len, name_len;
MD5_CTX ctx;
len = m->m_pkthdr.len;
if (len < 4) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "chap invalid packet length: %d bytes\n",
SPP_ARGS(ifp), len);
return;
}
h = mtod (m, struct lcp_header*);
if (len > ntohs (h->len))
len = ntohs (h->len);
switch (h->type) {
/* challenge, failure and success are his authproto */
case CHAP_CHALLENGE:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->myauth.secret == NULL || sp->myauth.name == NULL) {
/* can't do anything usefull */
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures++;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
printf(SPP_FMT "chap input without my name and my secret being set\n",
SPP_ARGS(ifp));
break;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
value = 1 + (u_char*)(h+1);
value_len = value[-1];
name = value + value_len;
name_len = len - value_len - 5;
if (name_len < 0) {
if (debug) {
log(LOG_DEBUG,
SPP_FMT "chap corrupted challenge "
"<%s id=0x%x len=%d",
SPP_ARGS(ifp),
sppp_auth_type_name(PPP_CHAP, h->type),
h->ident, ntohs(h->len));
if (len > 4)
sppp_print_bytes((u_char*) (h+1), len-4);
addlog(">\n");
}
break;
}
if (debug) {
log(LOG_DEBUG,
SPP_FMT "chap input <%s id=0x%x len=%d name=",
SPP_ARGS(ifp),
sppp_auth_type_name(PPP_CHAP, h->type), h->ident,
ntohs(h->len));
sppp_print_string((char*) name, name_len);
addlog(" value-size=%d value=", value_len);
sppp_print_bytes(value, value_len);
addlog(">\n");
}
/* Compute reply value. */
MD5Init(&ctx);
MD5Update(&ctx, &h->ident, 1);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
MD5Update(&ctx, sp->myauth.secret, strlen(sp->myauth.secret));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
MD5Update(&ctx, value, value_len);
MD5Final(digest, &ctx);
dsize = sizeof digest;
sppp_auth_send(&chap, sp, CHAP_RESPONSE, h->ident,
sizeof dsize, (const char *)&dsize,
sizeof digest, digest,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
strlen(sp->myauth.name),
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->myauth.name,
0);
break;
case CHAP_SUCCESS:
if (debug) {
log(LOG_DEBUG, SPP_FMT "chap success",
SPP_ARGS(ifp));
if (len > 4) {
addlog(": ");
sppp_print_string((char*)(h + 1), len - 4);
}
addlog("\n");
}
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
x = splnet();
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->pp_flags &= ~PP_NEEDAUTH;
if (sp->myauth.proto == PPP_CHAP &&
(sp->lcp.opts & (1 << LCP_OPT_AUTH_PROTO)) &&
(sp->lcp.protos & (1 << IDX_CHAP)) == 0) {
/*
* We are authenticator for CHAP but didn't
* complete yet. Leave it to tlu to proceed
* to network phase.
*/
splx(x);
break;
}
splx(x);
sppp_phase_network(sp);
break;
case CHAP_FAILURE:
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
x = splnet();
sp->pp_auth_failures++;
splx(x);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug) {
log(LOG_INFO, SPP_FMT "chap failure",
SPP_ARGS(ifp));
if (len > 4) {
addlog(": ");
sppp_print_string((char*)(h + 1), len - 4);
}
addlog("\n");
} else
log(LOG_INFO, SPP_FMT "chap failure\n",
SPP_ARGS(ifp));
/* await LCP shutdown by authenticator */
break;
/* response is my authproto */
case CHAP_RESPONSE:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->hisauth.secret == NULL) {
/* can't do anything usefull */
printf(SPP_FMT "chap input without his secret being set\n",
SPP_ARGS(ifp));
break;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
value = 1 + (u_char*)(h+1);
value_len = value[-1];
name = value + value_len;
name_len = len - value_len - 5;
if (name_len < 0) {
if (debug) {
log(LOG_DEBUG,
SPP_FMT "chap corrupted response "
"<%s id=0x%x len=%d",
SPP_ARGS(ifp),
sppp_auth_type_name(PPP_CHAP, h->type),
h->ident, ntohs(h->len));
if (len > 4)
sppp_print_bytes((u_char*)(h+1), len-4);
addlog(">\n");
}
break;
}
if (h->ident != sp->confid[IDX_CHAP]) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "chap dropping response for old ID "
"(got %d, expected %d)\n",
SPP_ARGS(ifp),
h->ident, sp->confid[IDX_CHAP]);
break;
}
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->hisauth.name != NULL &&
(name_len != strlen(sp->hisauth.name)
|| memcmp(name, sp->hisauth.name, name_len) != 0)) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
log(LOG_INFO, SPP_FMT "chap response, his name ",
SPP_ARGS(ifp));
sppp_print_string(name, name_len);
addlog(" != expected ");
sppp_print_string(sp->hisauth.name,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
strlen(sp->hisauth.name));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
addlog("\n");
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
goto chap_failure;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug) {
log(LOG_DEBUG, SPP_FMT "chap input(%s) "
"<%s id=0x%x len=%d name=",
SPP_ARGS(ifp),
sppp_state_name(sp->state[IDX_CHAP]),
sppp_auth_type_name(PPP_CHAP, h->type),
h->ident, ntohs (h->len));
sppp_print_string((char*)name, name_len);
addlog(" value-size=%d value=", value_len);
sppp_print_bytes(value, value_len);
addlog(">\n");
}
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (value_len != sizeof(sp->myauth.challenge)) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG,
SPP_FMT "chap bad hash value length: "
Fix LP64 printf format problem.
2002-01-05 22:26:44 +03:00
"%d bytes, should be %ld\n",
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
SPP_ARGS(ifp), value_len,
Fix LP64 printf format problem.
2002-01-05 22:26:44 +03:00
(long) sizeof(sp->myauth.challenge));
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
goto chap_failure;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
MD5Init(&ctx);
MD5Update(&ctx, &h->ident, 1);
MD5Update(&ctx, sp->hisauth.secret,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
strlen(sp->hisauth.secret));
MD5Update(&ctx, sp->myauth.challenge, sizeof(sp->myauth.challenge));
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
MD5Final(digest, &ctx);
#define FAILMSG "Failed..."
#define SUCCMSG "Welcome!"
if (value_len != sizeof digest ||
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
memcmp(digest, value, value_len) != 0) {
chap_failure:
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* action scn, tld */
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
x = splnet();
sp->pp_auth_failures++;
splx(x);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_auth_send(&chap, sp, CHAP_FAILURE, h->ident,
sizeof(FAILMSG) - 1, (u_char *)FAILMSG,
0);
chap.tld(sp);
break;
}
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* action sca, perhaps tlu */
if (sp->state[IDX_CHAP] == STATE_REQ_SENT ||
sp->state[IDX_CHAP] == STATE_OPENED)
sppp_auth_send(&chap, sp, CHAP_SUCCESS, h->ident,
sizeof(SUCCMSG) - 1, (u_char *)SUCCMSG,
0);
if (sp->state[IDX_CHAP] == STATE_REQ_SENT) {
sppp_cp_change_state(&chap, sp, STATE_OPENED);
chap.tlu(sp);
}
break;
default:
/* Unknown CHAP packet type -- ignore. */
if (debug) {
log(LOG_DEBUG, SPP_FMT "chap unknown input(%s) "
"<0x%x id=0x%xh len=%d",
SPP_ARGS(ifp),
sppp_state_name(sp->state[IDX_CHAP]),
h->type, h->ident, ntohs(h->len));
if (len > 4)
sppp_print_bytes((u_char*)(h+1), len-4);
addlog(">\n");
}
break;
}
}
static void
sppp_chap_init(struct sppp *sp)
{
/* Chap doesn't have STATE_INITIAL at all. */
sp->state[IDX_CHAP] = STATE_CLOSED;
sp->fail_counter[IDX_CHAP] = 0;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->pp_seq[IDX_CHAP] = 0;
sp->pp_rseq[IDX_CHAP] = 0;
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_init(&sp->ch[IDX_CHAP]);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
static void
sppp_chap_open(struct sppp *sp)
{
if (sp->myauth.proto == PPP_CHAP &&
(sp->lcp.opts & (1 << LCP_OPT_AUTH_PROTO)) != 0) {
/* we are authenticator for CHAP, start it */
chap.scr(sp);
sp->rst_counter[IDX_CHAP] = sp->lcp.max_configure;
sppp_cp_change_state(&chap, sp, STATE_REQ_SENT);
}
/* nothing to be done if we are peer, await a challenge */
}
static void
sppp_chap_close(struct sppp *sp)
{
if (sp->state[IDX_CHAP] != STATE_CLOSED)
sppp_cp_change_state(&chap, sp, STATE_CLOSED);
}
static void
sppp_chap_TO(void *cookie)
{
struct sppp *sp = (struct sppp *)cookie;
STDDCL;
int s;
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG, SPP_FMT "chap TO(%s) rst_counter = %d\n",
SPP_ARGS(ifp),
sppp_state_name(sp->state[IDX_CHAP]),
sp->rst_counter[IDX_CHAP]);
if (--sp->rst_counter[IDX_CHAP] < 0)
/* TO- event */
switch (sp->state[IDX_CHAP]) {
case STATE_REQ_SENT:
chap.tld(sp);
sppp_cp_change_state(&chap, sp, STATE_CLOSED);
break;
}
else
/* TO+ (or TO*) event */
switch (sp->state[IDX_CHAP]) {
case STATE_OPENED:
/* TO* event */
sp->rst_counter[IDX_CHAP] = sp->lcp.max_configure;
/* fall through */
case STATE_REQ_SENT:
chap.scr(sp);
/* sppp_cp_change_state() will restart the timer */
sppp_cp_change_state(&chap, sp, STATE_REQ_SENT);
break;
}
splx(s);
}
static void
sppp_chap_tlu(struct sppp *sp)
{
STDDCL;
int i, x;
i = 0;
sp->rst_counter[IDX_CHAP] = sp->lcp.max_configure;
/*
* Some broken CHAP implementations (Conware CoNet, firmware
* 4.0.?) don't want to re-authenticate their CHAP once the
* initial challenge-response exchange has taken place.
* Provide for an option to avoid rechallenges.
*/
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if ((sp->hisauth.flags & SPPP_AUTHFLAG_NORECHALLENGE) == 0) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Compute the re-challenge timeout. This will yield
* a number between 300 and 810 seconds.
*/
i = 300 + ((unsigned)(random() & 0xff00) >> 7);
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_reset(&sp->ch[IDX_CHAP], i * hz, chap.TO, sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
if (debug) {
log(LOG_DEBUG,
SPP_FMT "chap %s, ",
SPP_ARGS(ifp),
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase == SPPP_PHASE_NETWORK? "reconfirmed": "tlu");
if ((sp->hisauth.flags & SPPP_AUTHFLAG_NORECHALLENGE) == 0)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
addlog("next re-challenge in %d seconds\n", i);
else
addlog("re-challenging supressed\n");
}
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
x = splnet();
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* indicate to LCP that we need to be closed down */
sp->lcp.protos |= (1 << IDX_CHAP);
if (sp->pp_flags & PP_NEEDAUTH) {
/*
* Remote is authenticator, but his auth proto didn't
* complete yet. Defer the transition to network
* phase.
*/
splx(x);
return;
}
splx(x);
/*
* If we are already in phase network, we are done here. This
* is the case if this is a dummy tlu event after a re-challenge.
*/
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->pp_phase != SPPP_PHASE_NETWORK)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_phase_network(sp);
}
static void
sppp_chap_tld(struct sppp *sp)
{
STDDCL;
if (debug)
log(LOG_DEBUG, SPP_FMT "chap tld\n", SPP_ARGS(ifp));
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_stop(&sp->ch[IDX_CHAP]);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->lcp.protos &= ~(1 << IDX_CHAP);
lcp.Close(sp);
}
static void
sppp_chap_scr(struct sppp *sp)
{
struct timeval tv;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t *ch, seed;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
u_char clen;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->myauth.name == NULL) {
/* can't do anything usefull */
printf(SPP_FMT "chap starting without my name being set\n",
SPP_ARGS(&sp->pp_if));
return;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* Compute random challenge. */
code cleanup for portability
2001-12-05 00:32:15 +03:00
ch = (u_int32_t *)sp->myauth.challenge;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
microtime(&tv);
seed = tv.tv_sec ^ tv.tv_usec;
ch[0] = seed ^ random();
ch[1] = seed ^ random();
ch[2] = seed ^ random();
ch[3] = seed ^ random();
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
clen = 16; /* 4 * sizeof(u_int32_t) */
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->confid[IDX_CHAP] = ++sp->pp_seq[IDX_CHAP];
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_auth_send(&chap, sp, CHAP_CHALLENGE, sp->confid[IDX_CHAP],
sizeof clen, (const char *)&clen,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sizeof(sp->myauth.challenge), sp->myauth.challenge,
strlen(sp->myauth.name),
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->myauth.name,
0);
}
/*
*--------------------------------------------------------------------------*
* *
* The PAP implementation. *
* *
*--------------------------------------------------------------------------*
*/
/*
* For PAP, we need to keep a little state also if we are the peer, not the
* authenticator. This is since we don't get a request to authenticate, but
* have to repeatedly authenticate ourself until we got a response (or the
* retry counter is expired).
*/
/*
* Handle incoming PAP packets. */
static void
sppp_pap_input(struct sppp *sp, struct mbuf *m)
{
STDDCL;
struct lcp_header *h;
int len, x;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
u_char mlen;
char *name, *passwd;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
int name_len, passwd_len;
len = m->m_pkthdr.len;
if (len < 5) {
if (debug)
log(LOG_DEBUG,
SPP_FMT "pap invalid packet length: %d bytes\n",
SPP_ARGS(ifp), len);
return;
}
h = mtod (m, struct lcp_header*);
if (len > ntohs (h->len))
len = ntohs (h->len);
switch (h->type) {
/* PAP request is my authproto */
case PAP_REQ:
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->hisauth.name == NULL || sp->hisauth.secret == NULL) {
/* can't do anything usefull */
printf(SPP_FMT "pap request without his name and his secret being set\n",
SPP_ARGS(ifp));
break;
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
name = 1 + (u_char*)(h+1);
name_len = name[-1];
passwd = name + name_len + 1;
if (name_len > len - 6 ||
(passwd_len = passwd[-1]) > len - 6 - name_len) {
if (debug) {
log(LOG_DEBUG, SPP_FMT "pap corrupted input "
"<%s id=0x%x len=%d",
SPP_ARGS(ifp),
sppp_auth_type_name(PPP_PAP, h->type),
h->ident, ntohs(h->len));
if (len > 4)
sppp_print_bytes((u_char*)(h+1), len-4);
addlog(">\n");
}
break;
}
if (debug) {
log(LOG_DEBUG, SPP_FMT "pap input(%s) "
"<%s id=0x%x len=%d name=",
SPP_ARGS(ifp),
sppp_state_name(sp->state[IDX_PAP]),
sppp_auth_type_name(PPP_PAP, h->type),
h->ident, ntohs(h->len));
sppp_print_string((char*)name, name_len);
addlog(" passwd=");
sppp_print_string((char*)passwd, passwd_len);
addlog(">\n");
}
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (memcmp(name, sp->hisauth.name, name_len) != 0 ||
memcmp(passwd, sp->hisauth.secret, passwd_len) != 0) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* action scn, tld */
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures++;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
mlen = sizeof(FAILMSG) - 1;
sppp_auth_send(&pap, sp, PAP_NAK, h->ident,
sizeof mlen, (const char *)&mlen,
sizeof(FAILMSG) - 1, (u_char *)FAILMSG,
0);
pap.tld(sp);
break;
}
/* action sca, perhaps tlu */
if (sp->state[IDX_PAP] == STATE_REQ_SENT ||
sp->state[IDX_PAP] == STATE_OPENED) {
mlen = sizeof(SUCCMSG) - 1;
sppp_auth_send(&pap, sp, PAP_ACK, h->ident,
sizeof mlen, (const char *)&mlen,
sizeof(SUCCMSG) - 1, (u_char *)SUCCMSG,
0);
}
if (sp->state[IDX_PAP] == STATE_REQ_SENT) {
sppp_cp_change_state(&pap, sp, STATE_OPENED);
pap.tlu(sp);
}
break;
/* ack and nak are his authproto */
case PAP_ACK:
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_stop(&sp->pap_my_to_ch);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug) {
log(LOG_DEBUG, SPP_FMT "pap success",
SPP_ARGS(ifp));
code cleanup for portability
2001-12-05 00:32:15 +03:00
name_len = *(char *)h;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (len > 5 && name_len) {
addlog(": ");
sppp_print_string((char*)(h+1), name_len);
}
addlog("\n");
}
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
x = splnet();
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->pp_flags &= ~PP_NEEDAUTH;
if (sp->myauth.proto == PPP_PAP &&
(sp->lcp.opts & (1 << LCP_OPT_AUTH_PROTO)) &&
(sp->lcp.protos & (1 << IDX_PAP)) == 0) {
/*
* We are authenticator for PAP but didn't
* complete yet. Leave it to tlu to proceed
* to network phase.
*/
splx(x);
break;
}
splx(x);
sppp_phase_network(sp);
break;
case PAP_NAK:
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_stop(&sp->pap_my_to_ch);
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures++;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug) {
log(LOG_INFO, SPP_FMT "pap failure",
SPP_ARGS(ifp));
code cleanup for portability
2001-12-05 00:32:15 +03:00
name_len = *(char *)h;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (len > 5 && name_len) {
addlog(": ");
sppp_print_string((char*)(h+1), name_len);
}
addlog("\n");
} else
log(LOG_INFO, SPP_FMT "pap failure\n",
SPP_ARGS(ifp));
/* await LCP shutdown by authenticator */
break;
default:
/* Unknown PAP packet type -- ignore. */
if (debug) {
log(LOG_DEBUG, SPP_FMT "pap corrupted input "
"<0x%x id=0x%x len=%d",
SPP_ARGS(ifp),
h->type, h->ident, ntohs(h->len));
if (len > 4)
sppp_print_bytes((u_char*)(h+1), len-4);
addlog(">\n");
}
break;
}
}
static void
sppp_pap_init(struct sppp *sp)
{
/* PAP doesn't have STATE_INITIAL at all. */
sp->state[IDX_PAP] = STATE_CLOSED;
sp->fail_counter[IDX_PAP] = 0;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->pp_seq[IDX_PAP] = 0;
sp->pp_rseq[IDX_PAP] = 0;
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_init(&sp->ch[IDX_PAP]);
callout_init(&sp->pap_my_to_ch);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
static void
sppp_pap_open(struct sppp *sp)
{
if (sp->hisauth.proto == PPP_PAP &&
(sp->lcp.opts & (1 << LCP_OPT_AUTH_PROTO)) != 0) {
/* we are authenticator for PAP, start our timer */
sp->rst_counter[IDX_PAP] = sp->lcp.max_configure;
sppp_cp_change_state(&pap, sp, STATE_REQ_SENT);
}
if (sp->myauth.proto == PPP_PAP) {
/* we are peer, send a request, and start a timer */
pap.scr(sp);
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_reset(&sp->pap_my_to_ch, sp->lcp.timeout,
sppp_pap_my_TO, sp);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
}
static void
sppp_pap_close(struct sppp *sp)
{
if (sp->state[IDX_PAP] != STATE_CLOSED)
sppp_cp_change_state(&pap, sp, STATE_CLOSED);
}
/*
* That's the timeout routine if we are authenticator. Since the
* authenticator is basically passive in PAP, we can't do much here.
*/
static void
sppp_pap_TO(void *cookie)
{
struct sppp *sp = (struct sppp *)cookie;
STDDCL;
int s;
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (debug)
log(LOG_DEBUG, SPP_FMT "pap TO(%s) rst_counter = %d\n",
SPP_ARGS(ifp),
sppp_state_name(sp->state[IDX_PAP]),
sp->rst_counter[IDX_PAP]);
if (--sp->rst_counter[IDX_PAP] < 0)
/* TO- event */
switch (sp->state[IDX_PAP]) {
case STATE_REQ_SENT:
pap.tld(sp);
sppp_cp_change_state(&pap, sp, STATE_CLOSED);
break;
}
else
/* TO+ event, not very much we could do */
switch (sp->state[IDX_PAP]) {
case STATE_REQ_SENT:
/* sppp_cp_change_state() will restart the timer */
sppp_cp_change_state(&pap, sp, STATE_REQ_SENT);
break;
}
splx(s);
}
/*
* That's the timeout handler if we are peer. Since the peer is active,
* we need to retransmit our PAP request since it is apparently lost.
* XXX We should impose a max counter.
*/
static void
sppp_pap_my_TO(void *cookie)
{
struct sppp *sp = (struct sppp *)cookie;
STDDCL;
if (debug)
log(LOG_DEBUG, SPP_FMT "pap peer TO\n",
SPP_ARGS(ifp));
pap.scr(sp);
}
static void
sppp_pap_tlu(struct sppp *sp)
{
STDDCL;
int x;
sp->rst_counter[IDX_PAP] = sp->lcp.max_configure;
if (debug)
log(LOG_DEBUG, SPP_FMT "%s tlu\n",
SPP_ARGS(ifp), pap.name);
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
x = splnet();
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* indicate to LCP that we need to be closed down */
sp->lcp.protos |= (1 << IDX_PAP);
if (sp->pp_flags & PP_NEEDAUTH) {
/*
* Remote is authenticator, but his auth proto didn't
* complete yet. Defer the transition to network
* phase.
*/
splx(x);
return;
}
splx(x);
sppp_phase_network(sp);
}
static void
sppp_pap_tld(struct sppp *sp)
{
STDDCL;
if (debug)
log(LOG_DEBUG, SPP_FMT "pap tld\n", SPP_ARGS(ifp));
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_stop(&sp->ch[IDX_PAP]);
callout_stop(&sp->pap_my_to_ch);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sp->lcp.protos &= ~(1 << IDX_PAP);
lcp.Close(sp);
}
static void
sppp_pap_scr(struct sppp *sp)
{
u_char idlen, pwdlen;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (sp->myauth.secret == NULL || sp->myauth.name == NULL) {
/* can't do anything usefull */
printf(SPP_FMT "pap starting without my name and secret being set\n",
SPP_ARGS(&sp->pp_if));
return;
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->confid[IDX_PAP] = ++sp->pp_seq[IDX_PAP];
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
pwdlen = strlen(sp->myauth.secret);
idlen = strlen(sp->myauth.name);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_auth_send(&pap, sp, PAP_REQ, sp->confid[IDX_PAP],
sizeof idlen, (const char *)&idlen,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
idlen, sp->myauth.name,
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sizeof pwdlen, (const char *)&pwdlen,
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
pwdlen, sp->myauth.secret,
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
0);
}
/*
* Random miscellaneous functions.
*/
/*
* Send a PAP or CHAP proto packet.
*
* Varadic function, each of the elements for the ellipsis is of type
* ``size_t mlen, const u_char *msg''. Processing will stop iff
* mlen == 0.
* NOTE: never declare variadic functions with types subject to type
* promotion (i.e. u_char). This is asking for big trouble depending
* on the architecture you are on...
*/
static void
sppp_auth_send(const struct cp *cp, struct sppp *sp,
unsigned int type, unsigned int id,
...)
{
STDDCL;
struct lcp_header *lh;
struct mbuf *m;
u_char *p;
int len;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
size_t pkthdrlen;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
unsigned int mlen;
const char *msg;
va_list ap;
MGETHDR (m, M_DONTWAIT, MT_DATA);
if (! m)
return;
m->m_pkthdr.rcvif = 0;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if (sp->pp_flags & PP_NOFRAMING) {
*mtod(m, u_int16_t*) = htons(cp->proto);
pkthdrlen = 2;
lh = (struct lcp_header*)(mtod(m, u_int8_t*)+2);
} else {
struct ppp_header *h;
h = mtod (m, struct ppp_header*);
h->address = PPP_ALLSTATIONS; /* broadcast address */
h->control = PPP_UI; /* Unnumbered Info */
h->protocol = htons(cp->proto);
pkthdrlen = PPP_HEADER_LEN;
lh = (struct lcp_header*)(h + 1);
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
lh->type = type;
lh->ident = id;
p = (u_char*) (lh+1);
va_start(ap, id);
len = 0;
while ((mlen = (unsigned int)va_arg(ap, size_t)) != 0) {
msg = va_arg(ap, const char *);
len += mlen;
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
if (len > MHLEN - pkthdrlen - LCP_HEADER_LEN) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
va_end(ap);
m_freem(m);
return;
}
bcopy(msg, p, mlen);
p += mlen;
}
va_end(ap);
Add another option for encapsulation: PP_NOFRAMING. In this mode, the PPP packets start with the protocol identifier and don't have any explicit framing (which may be added by the lower level driver). Make input/output statistics a little bit more correct by adding a hardware driver adjustable framing length for each packet (instead of the constant value "3" used before). While there, bump authentication name length from 32 to 48 (I have a connection where I need more than 32). XXX - this should not be artificialy limited at all.
2001-04-09 23:33:01 +04:00
m->m_pkthdr.len = m->m_len = pkthdrlen + LCP_HEADER_LEN + len;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
lh->len = htons (LCP_HEADER_LEN + len);
if (debug) {
log(LOG_DEBUG, SPP_FMT "%s output <%s id=0x%x len=%d",
SPP_ARGS(ifp), cp->name,
sppp_auth_type_name(cp->proto, lh->type),
lh->ident, ntohs(lh->len));
if (len)
sppp_print_bytes((u_char*) (lh+1), len);
addlog(">\n");
}
if (IF_QFULL (&sp->pp_cpq)) {
IF_DROP (&sp->pp_fastq);
IF_DROP (&ifp->if_snd);
m_freem (m);
++ifp->if_oerrors;
} else
IF_ENQUEUE (&sp->pp_cpq, m);
if (! (ifp->if_flags & IFF_OACTIVE))
(*ifp->if_start) (ifp);
ifp->if_obytes += m->m_pkthdr.len + 3;
}
/*
* Send keepalive packets, every 10 seconds.
*/
static void
sppp_keepalive(void *dummy)
{
struct sppp *sp;
int s;
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
time_t now;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
s = splnet();
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
now = time.tv_sec;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
for (sp=spppq; sp; sp=sp->pp_next) {
struct ifnet *ifp = &sp->pp_if;
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
/* check idle timeout */
Initialize the activity timestamp when opening a connection. Only idle- timeout connection that made it to phase NETWORK yet. (For drivers using the internal timeout mechanism; isdnd, that does the timeout handling for ISDN drivers, still needs to be fixed.) Thanks to Wolfgang Solfrank for finding this.
2002-01-14 10:39:14 +03:00
if ((sp->pp_idle_timeout != 0) && (ifp->if_flags & IFF_RUNNING)
&& (sp->pp_phase == SPPP_PHASE_NETWORK)) {
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
/* idle timeout is enabled for this interface */
if ((now-sp->pp_last_activity) >= sp->pp_idle_timeout) {
if (ifp->if_flags & IFF_DEBUG)
printf("%s: no activitiy for %lu seconds\n",
sp->pp_if.if_xname,
(unsigned long)(now-sp->pp_last_activity));
lcp.Close(sp);
continue;
}
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/* Keepalive mode disabled or channel down? */
if (! (sp->pp_flags & PP_KEEPALIVE) ||
! (ifp->if_flags & IFF_RUNNING))
continue;
/* No keepalive in PPP mode if LCP not opened yet. */
if (! (sp->pp_flags & PP_CISCO) &&
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase < SPPP_PHASE_AUTHENTICATE)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
continue;
if (sp->pp_alivecnt == MAXALIVECNT) {
/* No keepalive packets got. Stop the interface. */
if_down (ifp);
IFQ_PURGE cannot be used against ifqueue. use IF_PURGE.
2001-08-23 04:56:03 +04:00
IF_PURGE (&sp->pp_cpq);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (! (sp->pp_flags & PP_CISCO)) {
Make reconnects after LCP keepalive detected an error actually work.
2001-12-15 23:40:37 +03:00
printf("%s: LCP keepalive timed out, going to restart the connection\n",
ifp->if_xname);
sp->pp_alivecnt = 0;
/* we are down, close all open protocols */
lcp.Close(sp);
/* And now prepare LCP to reestablish the link, if configured to do so. */
sppp_cp_change_state(&lcp, sp, STATE_STOPPED);
/* Close connection imediatly, completition of this
* will summon the magic needed to reestablish it. */
sp->pp_tlf(sp);
continue;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
}
if (sp->pp_alivecnt <= MAXALIVECNT)
++sp->pp_alivecnt;
if (sp->pp_flags & PP_CISCO)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_cisco_send (sp, CISCO_KEEPALIVE_REQ,
++sp->pp_seq[IDX_LCP], sp->pp_rseq[IDX_LCP]);
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
else if (sp->pp_phase >= SPPP_PHASE_AUTHENTICATE) {
code cleanup for portability
2001-12-05 00:32:15 +03:00
int32_t nmagic = htonl (sp->lcp.magic);
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sp->lcp.echoid = ++sp->pp_seq[IDX_LCP];
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sppp_cp_send (sp, PPP_LCP, ECHO_REQ,
sp->lcp.echoid, 4, &nmagic);
}
}
splx(s);
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
callout_reset(&keepalive_ch, hz * 10, sppp_keepalive, NULL);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
/*
* Get both IP addresses.
*/
static void
code cleanup for portability
2001-12-05 00:32:15 +03:00
sppp_get_ip_addrs(struct sppp *sp, u_int32_t *src, u_int32_t *dst, u_int32_t *srcmask)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
{
struct ifnet *ifp = &sp->pp_if;
struct ifaddr *ifa;
struct sockaddr_in *si, *sm;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t ssrc, ddst;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
sm = NULL;
code cleanup for portability
2001-12-05 00:32:15 +03:00
ssrc = ddst = 0;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Pick the first AF_INET address from the list,
* aliases don't make any sense on a p2p link anyway.
*/
Switch to using queue access macros instead of refering to the member fields explicitly.
2001-11-05 21:02:15 +03:00
si = 0;
TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list) {
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (ifa->ifa_addr->sa_family == AF_INET) {
si = (struct sockaddr_in *)ifa->ifa_addr;
sm = (struct sockaddr_in *)ifa->ifa_netmask;
if (si)
break;
}
Switch to using queue access macros instead of refering to the member fields explicitly.
2001-11-05 21:02:15 +03:00
}
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (ifa) {
if (si && si->sin_addr.s_addr) {
ssrc = si->sin_addr.s_addr;
if (srcmask)
*srcmask = ntohl(sm->sin_addr.s_addr);
}
si = (struct sockaddr_in *)ifa->ifa_dstaddr;
if (si && si->sin_addr.s_addr)
ddst = si->sin_addr.s_addr;
}
if (dst) *dst = ntohl(ddst);
if (src) *src = ntohl(ssrc);
}
/*
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
* Set IP addresses. Must be called at splnet.
* If an address is 0, leave it the way it is.
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
*/
static void
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
sppp_set_ip_addrs(struct sppp *sp, u_int32_t myaddr, u_int32_t hisaddr)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
{
STDDCL;
struct ifaddr *ifa;
struct sockaddr_in *si;
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
struct sockaddr_in *dest;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
/*
* Pick the first AF_INET address from the list,
* aliases don't make any sense on a p2p link anyway.
*/
Switch to using queue access macros instead of refering to the member fields explicitly.
2001-11-05 21:02:15 +03:00
si = 0;
TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
{
if (ifa->ifa_addr->sa_family == AF_INET)
{
si = (struct sockaddr_in *)ifa->ifa_addr;
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
dest = (struct sockaddr_in *)ifa->ifa_dstaddr;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if (si)
break;
}
}
if (ifa && si)
{
int error;
struct sockaddr_in new_sin = *si;
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
struct sockaddr_in new_dst = *dest;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
/*
* Scrub old routes now instead of calling in_ifinit with
* scrub=1, because we may change the dstaddr
* before the call to in_ifinit.
*/
in_ifscrub(ifp, ifatoia(ifa));
if (myaddr != 0)
new_sin.sin_addr.s_addr = htonl(myaddr);
if (hisaddr != 0) {
new_dst.sin_addr.s_addr = htonl(hisaddr);
if (new_dst.sin_addr.s_addr != dest->sin_addr.s_addr) {
sp->ipcp.saved_hisaddr = dest->sin_addr.s_addr;
*dest = new_dst; /* fix dstaddr in place */
}
}
error = in_ifinit(ifp, ifatoia(ifa), &new_sin, 0);
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(debug && error)
{
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
log(LOG_DEBUG, SPP_FMT "sppp_set_ip_addrs: in_ifinit "
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
" failed, error=%d\n", SPP_ARGS(ifp), error);
}
}
}
Change the way IPCP negotiation is handled. Collect both local and remote address and set them to the interface in one step (the peer adress was not set at all before). This causes the peer address now to show up on the interface and all messages to the routing socket to be send with correct data. The latter has been the last missing piece to complete PPPoE support.
2001-12-08 22:46:39 +03:00
/*
* Clear IP addresses. Must be called at splnet.
*/
static void
sppp_clear_ip_addrs(struct sppp *sp)
{
struct ifnet *ifp = &sp->pp_if;
struct ifaddr *ifa;
struct sockaddr_in *si;
struct sockaddr_in *dest;
u_int32_t remote;
if (sp->ipcp.flags & IPCP_HISADDR_DYN)
remote = sp->ipcp.saved_hisaddr;
else
sppp_get_ip_addrs(sp, 0, &remote, 0);
/*
* Pick the first AF_INET address from the list,
* aliases don't make any sense on a p2p link anyway.
*/
si = 0;
TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list)
{
if (ifa->ifa_addr->sa_family == AF_INET)
{
si = (struct sockaddr_in *)ifa->ifa_addr;
dest = (struct sockaddr_in *)ifa->ifa_dstaddr;
if (si)
break;
}
}
if (ifa && si)
{
struct sockaddr_in new_sin = *si;
in_ifscrub(ifp, ifatoia(ifa));
if (sp->ipcp.flags & IPCP_MYADDR_DYN)
new_sin.sin_addr.s_addr = 0;
if (sp->ipcp.flags & IPCP_HISADDR_DYN)
/* replace peer addr in place */
dest->sin_addr.s_addr = sp->ipcp.saved_hisaddr;
in_ifinit(ifp, ifatoia(ifa), &new_sin, 0);
}
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
#ifdef INET6
/*
* Get both IPv6 addresses.
*/
static void
sppp_get_ip6_addrs(struct sppp *sp, struct in6_addr *src, struct in6_addr *dst,
struct in6_addr *srcmask)
{
struct ifnet *ifp = &sp->pp_if;
struct ifaddr *ifa;
struct sockaddr_in6 *si, *sm;
struct in6_addr ssrc, ddst;
sm = NULL;
bzero -> memset
2001-07-18 20:43:09 +04:00
memset(&ssrc, 0, sizeof(ssrc));
memset(&ddst, 0, sizeof(ddst));
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
/*
* Pick the first link-local AF_INET6 address from the list,
* aliases don't make any sense on a p2p link anyway.
*/
Switch to using queue access macros instead of refering to the member fields explicitly.
2001-11-05 21:02:15 +03:00
si = 0;
TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
if (ifa->ifa_addr->sa_family == AF_INET6) {
si = (struct sockaddr_in6 *)ifa->ifa_addr;
sm = (struct sockaddr_in6 *)ifa->ifa_netmask;
if (si && IN6_IS_ADDR_LINKLOCAL(&si->sin6_addr))
break;
}
if (ifa) {
if (si && !IN6_IS_ADDR_UNSPECIFIED(&si->sin6_addr)) {
bcopy(&si->sin6_addr, &ssrc, sizeof(ssrc));
if (srcmask) {
bcopy(&sm->sin6_addr, srcmask,
sizeof(*srcmask));
}
}
si = (struct sockaddr_in6 *)ifa->ifa_dstaddr;
if (si && !IN6_IS_ADDR_UNSPECIFIED(&si->sin6_addr))
bcopy(&si->sin6_addr, &ddst, sizeof(ddst));
}
if (dst)
bcopy(&ddst, dst, sizeof(*dst));
if (src)
bcopy(&ssrc, src, sizeof(*src));
}
#ifdef IPV6CP_MYIFID_DYN
/*
* Generate random ifid.
*/
static void
sppp_gen_ip6_addr(struct sppp *sp, struct in6_addr *addr)
{
/* TBD */
}
/*
Remove the use of splimp() from the NetBSD kernel. splnet() and only splnet() is allowed for the protection of data structures used by network devices.
2001-04-14 03:29:55 +04:00
* Set my IPv6 address. Must be called at splnet.
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
*/
static void
sppp_set_ip6_addr(struct sppp *sp, const struct in6_addr *src)
{
STDDCL;
struct ifaddr *ifa;
struct sockaddr_in6 *sin6;
/*
* Pick the first link-local AF_INET6 address from the list,
* aliases don't make any sense on a p2p link anyway.
*/
sin6 = NULL;
Switch to using queue access macros instead of refering to the member fields explicitly.
2001-11-05 21:02:15 +03:00
TAILQ_FOREACH(ifa, &ifp->if_addrlist, ifa_list)
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
{
if (ifa->ifa_addr->sa_family == AF_INET6)
{
sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
if (sin6 && IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr))
break;
}
}
if (ifa && sin6)
{
int error;
struct sockaddr_in6 new_sin6 = *sin6;
bcopy(src, &new_sin6.sin6_addr, sizeof(new_sin6.sin6_addr));
error = in6_ifinit(ifp, ifatoia6(ifa), &new_sin6, 1);
if (debug && error)
{
log(LOG_DEBUG, SPP_FMT "sppp_set_ip6_addr: in6_ifinit "
" failed, error=%d\n", SPP_ARGS(ifp), error);
}
}
}
#endif
/*
* Suggest a candidate address to be used by peer.
*/
static void
sppp_suggest_ip6_addr(struct sppp *sp, struct in6_addr *suggest)
{
struct in6_addr myaddr;
propose better IPv6 ifid alternative to the peer, when ifid collides during IPv6CP negotiation. it is very rare to see collision.
2000-05-16 09:45:27 +04:00
struct timeval tv;
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
sppp_get_ip6_addrs(sp, &myaddr, 0, 0);
myaddr.s6_addr[8] &= ~0x02; /* u bit to "local" */
propose better IPv6 ifid alternative to the peer, when ifid collides during IPv6CP negotiation. it is very rare to see collision.
2000-05-16 09:45:27 +04:00
microtime(&tv);
if ((tv.tv_usec & 0xff) == 0 && (tv.tv_sec & 0xff) == 0) {
myaddr.s6_addr[14] ^= 0xff;
myaddr.s6_addr[15] ^= 0xff;
} else {
myaddr.s6_addr[14] ^= (tv.tv_usec & 0xff);
myaddr.s6_addr[15] ^= (tv.tv_sec & 0xff);
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
if (suggest)
bcopy(&myaddr, suggest, sizeof(myaddr));
}
#endif /*INET6*/
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
/*
* Process ioctl requests specific to the PPP interface.
* Permissions have already been checked.
*/
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static int
sppp_params(struct sppp *sp, int cmd, void *data)
{
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
switch (cmd) {
case SPPPGETAUTHCFG:
{
struct spppauthcfg * cfg = (struct spppauthcfg*)data;
cfg->myauthflags = sp->myauth.flags;
cfg->hisauthflags = sp->hisauth.flags;
strncpy(cfg->ifname, sp->pp_if.if_xname, IFNAMSIZ);
cfg->hisauth = 0;
if (sp->hisauth.proto)
cfg->hisauth = sp->hisauth.proto == PPP_PAP ? SPPP_AUTHPROTO_PAP : SPPP_AUTHPROTO_CHAP;
cfg->myauth = 0;
if (sp->myauth.proto)
cfg->myauth = sp->myauth.proto == PPP_PAP ? SPPP_AUTHPROTO_PAP : SPPP_AUTHPROTO_CHAP;
if (cfg->myname_length == 0) {
if (sp->myauth.name != NULL)
cfg->myname_length = strlen(sp->myauth.name)+1;
} else {
int rv;
size_t len = strlen(sp->myauth.name);
if (cfg->myname_length < len+1)
return ENAMETOOLONG;
rv = copyout(sp->myauth.name, cfg->myname, len);
if (rv) return rv;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
if (cfg->hisname_length == 0) {
if(sp->hisauth.name != NULL)
cfg->hisname_length = strlen(sp->hisauth.name)+1;
} else {
int rv;
size_t len = strlen(sp->hisauth.name);
if (cfg->hisname_length < len+1)
return ENAMETOOLONG;
rv = copyout(sp->hisauth.name, cfg->hisname, len);
if (rv) return rv;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
}
break;
case SPPPSETAUTHCFG:
{
struct spppauthcfg * cfg = (struct spppauthcfg*)data;
int rv;
if (sp->myauth.name) free(sp->myauth.name, M_DEVBUF);
sp->myauth.name = NULL;
if (sp->myauth.secret) free(sp->myauth.secret, M_DEVBUF);
sp->myauth.secret = NULL;
if (sp->hisauth.name) free(sp->hisauth.name, M_DEVBUF);
sp->hisauth.name = NULL;
if (sp->hisauth.secret) free(sp->hisauth.secret, M_DEVBUF);
sp->hisauth.secret = NULL;
if (cfg->hisname != NULL && cfg->hisname_length) {
sp->hisauth.name = malloc(cfg->hisname_length, M_DEVBUF, M_WAITOK);
rv = copyin(cfg->hisname, sp->hisauth.name, cfg->hisname_length);
if (rv) return rv;
sp->hisauth.name[cfg->hisname_length-1] = 0;
}
if (cfg->hissecret != NULL && cfg->hissecret_length) {
sp->hisauth.secret = malloc(cfg->hissecret_length, M_DEVBUF, M_WAITOK);
rv = copyin(cfg->hissecret, sp->hisauth.secret, cfg->hissecret_length);
if (rv) return rv;
sp->hisauth.secret[cfg->hisname_length-1] = 0;
}
if (cfg->myname != NULL && cfg->myname_length) {
sp->myauth.name = malloc(cfg->myname_length, M_DEVBUF, M_WAITOK);
rv = copyin(cfg->myname, sp->myauth.name, cfg->myname_length);
if (rv) return rv;
sp->myauth.name[cfg->myname_length-1] = 0;
}
if (cfg->mysecret != NULL && cfg->mysecret_length) {
sp->myauth.secret = malloc(cfg->mysecret_length, M_DEVBUF, M_WAITOK);
rv = copyin(cfg->mysecret, sp->myauth.secret, cfg->mysecret_length);
if (rv) return rv;
sp->myauth.secret[cfg->myname_length-1] = 0;
}
sp->myauth.flags = cfg->myauthflags;
if (cfg->myauth)
sp->myauth.proto = cfg->myauth == SPPP_AUTHPROTO_PAP ? PPP_PAP : PPP_CHAP;
sp->hisauth.flags = cfg->hisauthflags;
if (cfg->hisauth)
sp->hisauth.proto = cfg->hisauth == SPPP_AUTHPROTO_PAP ? PPP_PAP : PPP_CHAP;
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
sp->pp_auth_failures = 0;
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
}
break;
case SPPPGETLCPCFG:
{
struct sppplcpcfg * lcp = (struct sppplcpcfg*)data;
lcp->lcp_timeout = sp->lcp.timeout;
}
break;
case SPPPSETLCPCFG:
{
struct sppplcpcfg * lcp = (struct sppplcpcfg*)data;
sp->lcp.timeout = lcp->lcp_timeout;
}
break;
case SPPPGETSTATUS:
{
struct spppstatus * status = (struct spppstatus*)data;
status->phase = sp->pp_phase;
}
break;
Implement an activity timestamp, recording the last time payload data passed through. Implement optional idle timeout.
2002-01-06 23:14:29 +03:00
case SPPPGETIDLETO:
{
struct spppidletimeout * to = (struct spppidletimeout*)data;
to->idle_seconds = sp->pp_idle_timeout;
}
break;
case SPPPSETIDLETO:
{
struct spppidletimeout * to = (struct spppidletimeout*)data;
sp->pp_idle_timeout = to->idle_seconds;
}
break;
Implement a retry counter for failed authorizations and limit it to a configurable maximum (default: 5). Some ISPs shut down accounts (at least temporarily) after to many bad retries. This hit me recently due to a stupid pilot error and the fast retry rate.
2002-01-07 13:49:02 +03:00
case SPPPSETAUTHFAILURE:
{
struct spppauthfailuresettings * afsettings = (struct spppauthfailuresettings*)data;
sp->pp_max_auth_fail = afsettings->max_failures;
sp->pp_auth_failures = 0;
}
break;
case SPPPGETAUTHFAILURES:
{
struct spppauthfailurestats * stats = (struct spppauthfailurestats*)data;
stats->auth_failures = sp->pp_auth_failures;
stats->max_failures = sp->pp_max_auth_fail;
}
break;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
default:
return EINVAL;
}
return 0;
}
static void
sppp_phase_network(struct sppp *sp)
{
STDDCL;
int i;
code cleanup for portability
2001-12-05 00:32:15 +03:00
u_int32_t mask;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sp->pp_phase = SPPP_PHASE_NETWORK;
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
if(debug)
{
log(LOG_INFO, SPP_FMT "phase %s\n", SPP_ARGS(ifp),
sppp_phase_name(sp->pp_phase));
}
/* Notify NCPs now. */
for (i = 0; i < IDX_COUNT; i++)
if ((cps[i])->flags & CP_NCP)
(cps[i])->Open(sp);
/* Send Up events to all NCPs. */
for (i = 0, mask = 1; i < IDX_COUNT; i++, mask <<= 1)
fix operator precedence (& and &&). do not transmit too much message from LCP layer to NCP layer. PR 11161.
2000-10-08 06:37:42 +04:00
if ((sp->lcp.protos & mask) && ((cps[i])->flags & CP_NCP))
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
(cps[i])->Up(sp);
/* if no NCP is starting, all this was in vain, close down */
sppp_lcp_check_and_close(sp);
}
static const char *
sppp_cp_type_name(u_char type)
{
static char buf[12];
switch (type) {
case CONF_REQ: return "conf-req";
case CONF_ACK: return "conf-ack";
case CONF_NAK: return "conf-nak";
case CONF_REJ: return "conf-rej";
case TERM_REQ: return "term-req";
case TERM_ACK: return "term-ack";
case CODE_REJ: return "code-rej";
case PROTO_REJ: return "proto-rej";
case ECHO_REQ: return "echo-req";
case ECHO_REPLY: return "echo-reply";
case DISC_REQ: return "discard-req";
}
sprintf (buf, "0x%x", type);
return buf;
}
static const char *
sppp_auth_type_name(u_short proto, u_char type)
{
static char buf[12];
switch (proto) {
case PPP_CHAP:
switch (type) {
case CHAP_CHALLENGE: return "challenge";
case CHAP_RESPONSE: return "response";
case CHAP_SUCCESS: return "success";
case CHAP_FAILURE: return "failure";
}
case PPP_PAP:
switch (type) {
case PAP_REQ: return "req";
case PAP_ACK: return "ack";
case PAP_NAK: return "nak";
}
}
sprintf (buf, "0x%x", type);
return buf;
}
static const char *
sppp_lcp_opt_name(u_char opt)
{
static char buf[12];
switch (opt) {
case LCP_OPT_MRU: return "mru";
case LCP_OPT_ASYNC_MAP: return "async-map";
case LCP_OPT_AUTH_PROTO: return "auth-proto";
case LCP_OPT_QUAL_PROTO: return "qual-proto";
case LCP_OPT_MAGIC: return "magic";
case LCP_OPT_PROTO_COMP: return "proto-comp";
case LCP_OPT_ADDR_COMP: return "addr-comp";
}
sprintf (buf, "0x%x", opt);
return buf;
}
static const char *
sppp_ipcp_opt_name(u_char opt)
{
static char buf[12];
switch (opt) {
case IPCP_OPT_ADDRESSES: return "addresses";
case IPCP_OPT_COMPRESSION: return "compression";
case IPCP_OPT_ADDRESS: return "address";
}
sprintf (buf, "0x%x", opt);
return buf;
}
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
#ifdef INET6
static const char *
sppp_ipv6cp_opt_name(u_char opt)
{
static char buf[12];
switch (opt) {
case IPV6CP_OPT_IFID: return "ifid";
case IPV6CP_OPT_COMPRESSION: return "compression";
}
sprintf (buf, "0x%x", opt);
return buf;
}
#endif
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
static const char *
sppp_state_name(int state)
{
switch (state) {
case STATE_INITIAL: return "initial";
case STATE_STARTING: return "starting";
case STATE_CLOSED: return "closed";
case STATE_STOPPED: return "stopped";
case STATE_CLOSING: return "closing";
case STATE_STOPPING: return "stopping";
case STATE_REQ_SENT: return "req-sent";
case STATE_ACK_RCVD: return "ack-rcvd";
case STATE_ACK_SENT: return "ack-sent";
case STATE_OPENED: return "opened";
}
return "illegal";
}
static const char *
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
sppp_phase_name(int phase)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
{
switch (phase) {
Move net/if_sppp.h to net/if_spppvar.h, create a new net/if_sppp.h containing the userland visible thinks (i.e. ioctl definitions). Remove all (both) old ioctls, as they had a brain dead API and made keeping binary compatibility more or less impossible. Replace by several new ioctls. While there, remove any arbitrary limits (resulting from the old, broken ioctls) and allow any length of names and passwords.
2002-01-04 15:21:24 +03:00
case SPPP_PHASE_DEAD: return "dead";
case SPPP_PHASE_ESTABLISH: return "establish";
case SPPP_PHASE_TERMINATE: return "terminate";
case SPPP_PHASE_AUTHENTICATE: return "authenticate";
case SPPP_PHASE_NETWORK: return "network";
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
return "illegal";
}
static const char *
sppp_proto_name(u_short proto)
{
static char buf[12];
switch (proto) {
case PPP_LCP: return "lcp";
case PPP_IPCP: return "ipcp";
case PPP_PAP: return "pap";
case PPP_CHAP: return "chap";
IPv6CP support. if IPv6 link-local address is configured to the interface, the interface tries to negotiate ifid with the other end by using IPv6CP. other changes: - do not share ppp sequence number across protocols. - if LCP proto-rej is received, drop the protocol mentioned by the message. this is to be friendly with non-IPv6 peer (if the peer complains due to lack of IPv6CP, drop IPv6CP). this basically implements "RXJ+" state transition in the RFC. - cleanup debugging message. always print blank just before message. CAVEAT: - if the peer uses the same MAC address as our side (pretty unlikely) the code may go into req-rej loop. - even though we negotiate ifid, we don't configure destination address onto the interface. it is not really necessary to do so (IMHO). - I've tested this code on a NetBSD 1.4.2 node, which was with fair amount of modifications. not sure if the committed code does it right... (please test and send reports)
2000-05-02 16:43:16 +04:00
case PPP_IPV6CP: return "ipv6cp";
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
}
sprintf(buf, "0x%x", (unsigned)proto);
return buf;
}
static void
sppp_print_bytes(const u_char *p, u_short len)
{
addlog(" %02x", *p++);
while (--len > 0)
addlog("-%02x", *p++);
}
static void
sppp_print_string(const char *p, u_short len)
{
u_char c;
while (len-- > 0) {
c = *p++;
/*
* Print only ASCII chars directly. RFC 1994 recommends
* using only them, but we don't rely on it. */
if (c < ' ' || c > '~')
addlog("\\x%x", c);
else
addlog("%c", c);
}
}
static const char *
code cleanup for portability
2001-12-05 00:32:15 +03:00
sppp_dotted_quad(u_int32_t addr)
switch to the i4b version of if_sppp*.[ch] (with mods)
1999-04-04 10:07:56 +04:00
{
static char s[16];
sprintf(s, "%d.%d.%d.%d",
(int)((addr >> 24) & 0xff),
(int)((addr >> 16) & 0xff),
(int)((addr >> 8) & 0xff),
(int)(addr & 0xff));
return s;
}
/* a dummy, used to drop uninteresting events */
static void
sppp_null(struct sppp *unused)
{
/* do just nothing */
}
/*
* This file is large. Tell emacs to highlight it nevertheless.
*
* Local Variables:
* hilit-auto-highlight-maxout: 120000
* End:
*/