freebsd-dev/usr.sbin/ppp/physical.c

1132 lines
29 KiB
C
Raw Normal View History

/*
* Written by Eivind Eklund <eivind@yes.no>
* for Yes Interactive
*
* Copyright (C) 1998, Yes Interactive. All rights reserved.
*
* Redistribution and use in any form is permitted. Redistribution in
* source form should include the above copyright and this set of
* conditions, because large sections american law seems to have been
* created by a bunch of jerks on drugs that are now illegal, forcing
* me to include this copyright-stuff instead of placing this in the
* public domain. The name of of 'Yes Interactive' or 'Eivind Eklund'
* may not be used to endorse or promote products derived from this
* software without specific prior written permission.
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED
* WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE.
*
1999-08-28 01:35:59 +00:00
* $FreeBSD$
*
*/
#include <sys/param.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <sys/socket.h>
#include <sys/un.h>
#include <errno.h>
#include <fcntl.h>
#include <paths.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/tty.h> /* TIOCOUTQ */
#include <sys/uio.h>
#include <time.h>
#include <unistd.h>
#include <utmp.h>
#if defined(__OpenBSD__) || defined(__NetBSD__)
#include <sys/ioctl.h>
#include <util.h>
#else
#include <libutil.h>
#endif
#include "layer.h"
#ifndef NONAT
#include "nat_cmd.h"
#endif
#include "proto.h"
#include "acf.h"
#include "vjcomp.h"
#include "defs.h"
#include "command.h"
#include "mbuf.h"
#include "log.h"
#include "id.h"
#include "timer.h"
#include "fsm.h"
#include "lqr.h"
#include "hdlc.h"
#include "lcp.h"
#include "throughput.h"
#include "sync.h"
#include "async.h"
#include "iplist.h"
#include "slcompress.h"
#include "ncpaddr.h"
#include "ipcp.h"
#include "filter.h"
#include "descriptor.h"
o Move struct lcp and struct ccp into struct link. o Remove bundle2lcp(), bundle2ccp() and bundle2link(). They're too resource-hungry and we have `owner pointers' to do their job. o Make our FSM understand LCPs that are always ST_OPENED (with a minimum code that != 1). o Send FSM code rejects for invalid codes. o Make our bundle fsm_parent deal with multiple links. o Make timer diagnostics pretty and allow access via ~t in `term' mode (not just when logging debug) and `show timers'. Only show timers every second in debug mode, otherwise we get too many diagnostics to be useful (we probably still do). Also, don't restrict ~m in term mode to depend on debug logging. o Rationalise our bundles' phases. o Create struct mp (multilink protocol). This is both an NCP and a type of struct link. It feeds off other NCPs for output, passing fragmented packets into the queues of available datalinks. It also gets PROTO_MP input, reassembles the fragments into ppp frames, and passes them back to the HDLC layer that the fragments were passed from. ** It's not yet possible to enter multilink mode :-( ** o Add `set weight' (requires context) for deciding on a links weighting in multilink mode. Weighting is simplistic (and probably badly implemented) for now. o Remove the function pointers in struct link. They ended up only applying to physical links. o Configure our tun device with an MTU equal to the MRU from struct mp's LCP and a speed equal to the sum of our link speeds. o `show {lcp,ccp,proto}' and `set deflate' now have optional context and use ChooseLink() to decide on which `struct link' to use. This allows behaviour as before when in non-multilink mode, and allows access to the MP logical link in multilink mode. o Ignore reconnect and redial values when in -direct mode and when cleaning up. Always redial when in -ddial or -dedicated mode (unless cleaning up). o Tell our links to `staydown' when we close them due to a signal. o Remove remaining `#ifdef SIGALRM's (ppp doesn't function without alarms). o Don't bother strdup()ing our physical link name. o Various other cosmetic changes.
1998-04-03 19:21:56 +00:00
#include "ccp.h"
#include "link.h"
#include "physical.h"
#include "mp.h"
#ifndef NORADIUS
#include "radius.h"
#endif
#include "ipv6cp.h"
#include "ncp.h"
#include "bundle.h"
#include "prompt.h"
#include "chat.h"
#include "auth.h"
#include "chap.h"
#include "cbcp.h"
#include "datalink.h"
#include "tcp.h"
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
#include "udp.h"
#include "exec.h"
#include "tty.h"
#ifndef NOI4B
#include "i4b.h"
#endif
#ifndef NONETGRAPH
#include "ether.h"
#include "netgraph.h"
#endif
#ifndef NOATM
#include "atm.h"
#endif
#include "tcpmss.h"
#define PPPOTCPLINE "ppp"
static int physical_DescriptorWrite(struct fdescriptor *, struct bundle *,
const fd_set *);
static int
physical_DeviceSize(void)
{
return sizeof(struct device);
}
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
struct {
struct device *(*create)(struct physical *);
struct device *(*iov2device)(int, struct physical *, struct iovec *,
int *, int, int *, int *);
int (*DeviceSize)(void);
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
} devices[] = {
#ifndef NOI4B
/*
* This must come before ``tty'' so that the probe routine is
* able to identify it as a more specific type of terminal device.
*/
{ i4b_Create, i4b_iov2device, i4b_DeviceSize },
#endif
{ tty_Create, tty_iov2device, tty_DeviceSize },
#ifndef NONETGRAPH
/*
* This must come before ``udp'' so that the probe routine is
* able to identify it as a more specific type of SOCK_DGRAM.
*/
{ ether_Create, ether_iov2device, ether_DeviceSize },
#ifdef EXPERIMENTAL_NETGRAPH
{ ng_Create, ng_iov2device, ng_DeviceSize },
#endif
#endif
#ifndef NOATM
/* Ditto for ATM devices */
{ atm_Create, atm_iov2device, atm_DeviceSize },
#endif
{ tcp_Create, tcp_iov2device, tcp_DeviceSize },
{ udp_Create, udp_iov2device, udp_DeviceSize },
{ exec_Create, exec_iov2device, exec_DeviceSize }
};
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
#define NDEVICES (sizeof devices / sizeof devices[0])
static int
physical_UpdateSet(struct fdescriptor *d, fd_set *r, fd_set *w, fd_set *e,
int *n)
{
return physical_doUpdateSet(d, r, w, e, n, 0);
}
void
physical_SetDescriptor(struct physical *p)
{
p->desc.type = PHYSICAL_DESCRIPTOR;
p->desc.UpdateSet = physical_UpdateSet;
p->desc.IsSet = physical_IsSet;
p->desc.Read = physical_DescriptorRead;
p->desc.Write = physical_DescriptorWrite;
}
struct physical *
physical_Create(struct datalink *dl, int type)
{
struct physical *p;
p = (struct physical *)malloc(sizeof(struct physical));
if (!p)
return NULL;
p->link.type = PHYSICAL_LINK;
p->link.name = dl->name;
p->link.len = sizeof *p;
/* The sample period is fixed - see physical2iov() & iov2physical() */
throughput_init(&p->link.stats.total, SAMPLE_PERIOD);
p->link.stats.parent = dl->bundle->ncp.mp.active ?
&dl->bundle->ncp.mp.link.stats.total : NULL;
p->link.stats.gather = 1;
memset(p->link.Queue, '\0', sizeof p->link.Queue);
memset(p->link.proto_in, '\0', sizeof p->link.proto_in);
memset(p->link.proto_out, '\0', sizeof p->link.proto_out);
link_EmptyStack(&p->link);
p->handler = NULL;
physical_SetDescriptor(p);
p->type = type;
hdlc_Init(&p->hdlc, &p->link.lcp);
async_Init(&p->async);
p->fd = -1;
p->out = NULL;
p->connect_count = 0;
p->dl = dl;
p->input.sz = 0;
*p->name.full = '\0';
p->name.base = p->name.full;
p->Utmp = 0;
p->session_owner = (pid_t)-1;
p->cfg.rts_cts = MODEM_CTSRTS;
p->cfg.speed = MODEM_SPEED;
p->cfg.parity = CS8;
memcpy(p->cfg.devlist, MODEM_LIST, sizeof MODEM_LIST);
p->cfg.ndev = NMODEMS;
p->cfg.cd.necessity = CD_DEFAULT;
p->cfg.cd.delay = 0; /* reconfigured or device specific default */
lcp_Init(&p->link.lcp, dl->bundle, &p->link, &dl->fsmp);
ccp_Init(&p->link.ccp, dl->bundle, &p->link, &dl->fsmp);
return p;
}
static const struct parity {
const char *name;
const char *name1;
int set;
} validparity[] = {
{ "even", "P_EVEN", CS7 | PARENB },
{ "odd", "P_ODD", CS7 | PARENB | PARODD },
{ "none", "P_ZERO", CS8 },
{ NULL, 0 },
};
static int
GetParityValue(const char *str)
{
const struct parity *pp;
for (pp = validparity; pp->name; pp++) {
if (strcasecmp(pp->name, str) == 0 ||
strcasecmp(pp->name1, str) == 0) {
return pp->set;
}
}
return (-1);
}
int
physical_SetParity(struct physical *p, const char *str)
{
struct termios rstio;
int val;
val = GetParityValue(str);
if (val > 0) {
p->cfg.parity = val;
if (p->fd >= 0) {
tcgetattr(p->fd, &rstio);
rstio.c_cflag &= ~(CSIZE | PARODD | PARENB);
rstio.c_cflag |= val;
tcsetattr(p->fd, TCSADRAIN, &rstio);
}
return 0;
}
log_Printf(LogWARN, "%s: %s: Invalid parity\n", p->link.name, str);
return -1;
}
int
physical_GetSpeed(struct physical *p)
{
if (p->handler && p->handler->speed)
return (*p->handler->speed)(p);
return 0;
}
int
physical_SetSpeed(struct physical *p, int speed)
{
if (IntToSpeed(speed) != B0) {
p->cfg.speed = speed;
return 1;
}
return 0;
}
int
physical_Raw(struct physical *p)
{
if (p->handler && p->handler->raw)
return (*p->handler->raw)(p);
return 1;
}
void
physical_Offline(struct physical *p)
{
if (p->handler && p->handler->offline)
(*p->handler->offline)(p);
log_Printf(LogPHASE, "%s: Disconnected!\n", p->link.name);
}
static int
physical_Lock(struct physical *p)
{
int res;
if (*p->name.full == '/' && p->type != PHYS_DIRECT &&
(res = ID0uu_lock(p->name.base)) != UU_LOCK_OK) {
if (res == UU_LOCK_INUSE)
log_Printf(LogPHASE, "%s: %s is in use\n", p->link.name, p->name.full);
else
log_Printf(LogPHASE, "%s: %s is in use: uu_lock: %s\n",
p->link.name, p->name.full, uu_lockerr(res));
return 0;
}
return 1;
}
static void
physical_Unlock(struct physical *p)
{
if (*p->name.full == '/' && p->type != PHYS_DIRECT &&
ID0uu_unlock(p->name.base) == -1)
log_Printf(LogALERT, "%s: Can't uu_unlock %s\n", p->link.name,
p->name.base);
}
void
physical_Close(struct physical *p)
{
int newsid;
char fn[PATH_MAX];
if (p->fd < 0)
return;
log_Printf(LogDEBUG, "%s: Close\n", p->link.name);
if (p->handler && p->handler->cooked)
(*p->handler->cooked)(p);
physical_StopDeviceTimer(p);
if (p->Utmp) {
if (p->handler && (p->handler->type == TCP_DEVICE ||
p->handler->type == UDP_DEVICE))
/* Careful - we logged in on line ``ppp'' with IP as our host */
ID0logout(PPPOTCPLINE, 1);
else
ID0logout(p->name.base, 0);
p->Utmp = 0;
}
newsid = tcgetpgrp(p->fd) == getpgrp();
close(p->fd);
p->fd = -1;
log_SetTtyCommandMode(p->dl);
throughput_stop(&p->link.stats.total);
throughput_log(&p->link.stats.total, LogPHASE, p->link.name);
if (p->session_owner != (pid_t)-1) {
log_Printf(LogPHASE, "%s: HUPing %ld\n", p->link.name,
(long)p->session_owner);
ID0kill(p->session_owner, SIGHUP);
p->session_owner = (pid_t)-1;
}
if (newsid)
bundle_setsid(p->dl->bundle, 0);
if (*p->name.full == '/') {
snprintf(fn, sizeof fn, "%s%s.if", _PATH_VARRUN, p->name.base);
#ifndef RELEASE_CRUNCH
if (ID0unlink(fn) == -1)
log_Printf(LogALERT, "%s: Can't remove %s: %s\n",
p->link.name, fn, strerror(errno));
#else
ID0unlink(fn);
#endif
}
physical_Unlock(p);
if (p->handler && p->handler->destroy)
(*p->handler->destroy)(p);
p->handler = NULL;
p->name.base = p->name.full;
*p->name.full = '\0';
}
void
physical_Destroy(struct physical *p)
{
physical_Close(p);
throughput_destroy(&p->link.stats.total);
free(p);
}
static int
physical_DescriptorWrite(struct fdescriptor *d, struct bundle *bundle,
const fd_set *fdset)
{
struct physical *p = descriptor2physical(d);
int nw, result = 0;
if (p->out == NULL)
p->out = link_Dequeue(&p->link);
if (p->out) {
nw = physical_Write(p, MBUF_CTOP(p->out), p->out->m_len);
2000-03-14 01:47:19 +00:00
log_Printf(LogDEBUG, "%s: DescriptorWrite: wrote %d(%lu) to %d\n",
p->link.name, nw, (unsigned long)p->out->m_len, p->fd);
if (nw > 0) {
p->out->m_len -= nw;
p->out->m_offset += nw;
if (p->out->m_len == 0)
p->out = m_free(p->out);
result = 1;
} else if (nw < 0) {
if (errno == EAGAIN)
result = 1;
else if (errno != ENOBUFS) {
log_Printf(LogPHASE, "%s: write (%d): %s\n", p->link.name,
p->fd, strerror(errno));
datalink_Down(p->dl, CLOSE_NORMAL);
}
}
/* else we shouldn't really have been called ! select() is broken ! */
}
return result;
}
int
physical_ShowStatus(struct cmdargs const *arg)
{
struct physical *p = arg->cx->physical;
struct cd *cd;
const char *dev;
int n, slot;
prompt_Printf(arg->prompt, "Name: %s\n", p->link.name);
prompt_Printf(arg->prompt, " State: ");
if (p->fd < 0)
prompt_Printf(arg->prompt, "closed\n");
else {
slot = physical_Slot(p);
if (p->handler && p->handler->openinfo) {
if (slot == -1)
prompt_Printf(arg->prompt, "open (%s)\n", (*p->handler->openinfo)(p));
else
prompt_Printf(arg->prompt, "open (%s, port %d)\n",
(*p->handler->openinfo)(p), slot);
} else if (slot == -1)
prompt_Printf(arg->prompt, "open\n");
else
prompt_Printf(arg->prompt, "open (port %d)\n", slot);
}
prompt_Printf(arg->prompt, " Device: %s",
*p->name.full ? p->name.full :
p->type == PHYS_DIRECT ? "unknown" : "N/A");
if (p->session_owner != (pid_t)-1)
prompt_Printf(arg->prompt, " (session owner: %ld)", (long)p->session_owner);
prompt_Printf(arg->prompt, "\n Link Type: %s\n", mode2Nam(p->type));
prompt_Printf(arg->prompt, " Connect Count: %d\n", p->connect_count);
#ifdef TIOCOUTQ
if (p->fd >= 0 && ioctl(p->fd, TIOCOUTQ, &n) >= 0)
prompt_Printf(arg->prompt, " Physical outq: %d\n", n);
#endif
prompt_Printf(arg->prompt, " Queued Packets: %lu\n",
(u_long)link_QueueLen(&p->link));
prompt_Printf(arg->prompt, " Phone Number: %s\n", arg->cx->phone.chosen);
prompt_Printf(arg->prompt, "\nDefaults:\n");
prompt_Printf(arg->prompt, " Device List: ");
dev = p->cfg.devlist;
for (n = 0; n < p->cfg.ndev; n++) {
if (n)
prompt_Printf(arg->prompt, ", ");
prompt_Printf(arg->prompt, "\"%s\"", dev);
dev += strlen(dev) + 1;
}
2002-06-15 08:03:30 +00:00
prompt_Printf(arg->prompt, "\n Characteristics: ");
if (physical_IsSync(arg->cx->physical))
prompt_Printf(arg->prompt, "sync");
else
prompt_Printf(arg->prompt, "%dbps", p->cfg.speed);
switch (p->cfg.parity & CSIZE) {
case CS7:
prompt_Printf(arg->prompt, ", cs7");
break;
case CS8:
prompt_Printf(arg->prompt, ", cs8");
break;
}
if (p->cfg.parity & PARENB) {
if (p->cfg.parity & PARODD)
prompt_Printf(arg->prompt, ", odd parity");
else
prompt_Printf(arg->prompt, ", even parity");
} else
prompt_Printf(arg->prompt, ", no parity");
prompt_Printf(arg->prompt, ", CTS/RTS %s\n", (p->cfg.rts_cts ? "on" : "off"));
prompt_Printf(arg->prompt, " CD check delay: ");
cd = p->handler ? &p->handler->cd : &p->cfg.cd;
if (cd->necessity == CD_NOTREQUIRED)
prompt_Printf(arg->prompt, "no cd");
else if (p->cfg.cd.necessity == CD_DEFAULT) {
prompt_Printf(arg->prompt, "device specific");
} else {
prompt_Printf(arg->prompt, "%d second%s", p->cfg.cd.delay,
p->cfg.cd.delay == 1 ? "" : "s");
if (p->cfg.cd.necessity == CD_REQUIRED)
prompt_Printf(arg->prompt, " (required!)");
}
prompt_Printf(arg->prompt, "\n\n");
throughput_disp(&p->link.stats.total, arg->prompt);
return 0;
}
void
physical_DescriptorRead(struct fdescriptor *d, struct bundle *bundle,
const fd_set *fdset)
{
struct physical *p = descriptor2physical(d);
u_char *rbuff;
int n, found;
rbuff = p->input.buf + p->input.sz;
/* something to read */
n = physical_Read(p, rbuff, sizeof p->input.buf - p->input.sz);
log_Printf(LogDEBUG, "%s: DescriptorRead: read %d/%d from %d\n",
p->link.name, n, (int)(sizeof p->input.buf - p->input.sz), p->fd);
if (n <= 0) {
if (n < 0)
log_Printf(LogPHASE, "%s: read (%d): %s\n", p->link.name, p->fd,
strerror(errno));
else
log_Printf(LogPHASE, "%s: read (%d): Got zero bytes\n",
p->link.name, p->fd);
datalink_Down(p->dl, CLOSE_NORMAL);
return;
}
rbuff -= p->input.sz;
n += p->input.sz;
if (p->link.lcp.fsm.state <= ST_CLOSED) {
if (p->type != PHYS_DEDICATED) {
found = hdlc_Detect((u_char const **)&rbuff, n, physical_IsSync(p));
if (rbuff != p->input.buf)
log_WritePrompts(p->dl, "%.*s", (int)(rbuff - p->input.buf),
p->input.buf);
p->input.sz = n - (rbuff - p->input.buf);
if (found) {
/* LCP packet is detected. Turn ourselves into packet mode */
log_Printf(LogPHASE, "%s: PPP packet detected, coming up\n",
p->link.name);
log_SetTtyCommandMode(p->dl);
datalink_Up(p->dl, 0, 1);
link_PullPacket(&p->link, rbuff, p->input.sz, bundle);
p->input.sz = 0;
} else
bcopy(rbuff, p->input.buf, p->input.sz);
} else
/* In -dedicated mode, we just discard input until LCP is started */
p->input.sz = 0;
} else if (n > 0)
link_PullPacket(&p->link, rbuff, n, bundle);
}
struct physical *
iov2physical(struct datalink *dl, struct iovec *iov, int *niov, int maxiov,
int fd, int *auxfd, int *nauxfd)
{
struct physical *p;
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
int len, h, type;
p = (struct physical *)iov[(*niov)++].iov_base;
p->link.name = dl->name;
memset(p->link.Queue, '\0', sizeof p->link.Queue);
p->desc.UpdateSet = physical_UpdateSet;
p->desc.IsSet = physical_IsSet;
p->desc.Read = physical_DescriptorRead;
p->desc.Write = physical_DescriptorWrite;
p->type = PHYS_DIRECT;
p->dl = dl;
len = strlen(_PATH_DEV);
p->out = NULL;
p->connect_count = 1;
physical_SetDevice(p, p->name.full);
p->link.lcp.fsm.bundle = dl->bundle;
p->link.lcp.fsm.link = &p->link;
memset(&p->link.lcp.fsm.FsmTimer, '\0', sizeof p->link.lcp.fsm.FsmTimer);
memset(&p->link.lcp.fsm.OpenTimer, '\0', sizeof p->link.lcp.fsm.OpenTimer);
memset(&p->link.lcp.fsm.StoppedTimer, '\0',
sizeof p->link.lcp.fsm.StoppedTimer);
p->link.lcp.fsm.parent = &dl->fsmp;
lcp_SetupCallbacks(&p->link.lcp);
p->link.ccp.fsm.bundle = dl->bundle;
p->link.ccp.fsm.link = &p->link;
/* Our in.state & out.state are NULL (no link-level ccp yet) */
memset(&p->link.ccp.fsm.FsmTimer, '\0', sizeof p->link.ccp.fsm.FsmTimer);
memset(&p->link.ccp.fsm.OpenTimer, '\0', sizeof p->link.ccp.fsm.OpenTimer);
memset(&p->link.ccp.fsm.StoppedTimer, '\0',
sizeof p->link.ccp.fsm.StoppedTimer);
p->link.ccp.fsm.parent = &dl->fsmp;
ccp_SetupCallbacks(&p->link.ccp);
p->hdlc.lqm.owner = &p->link.lcp;
p->hdlc.ReportTimer.state = TIMER_STOPPED;
p->hdlc.lqm.timer.state = TIMER_STOPPED;
p->fd = fd;
p->link.stats.total.in.SampleOctets = (long long *)iov[(*niov)++].iov_base;
p->link.stats.total.out.SampleOctets = (long long *)iov[(*niov)++].iov_base;
p->link.stats.parent = dl->bundle->ncp.mp.active ?
&dl->bundle->ncp.mp.link.stats.total : NULL;
p->link.stats.gather = 1;
type = (long)p->handler;
p->handler = NULL;
for (h = 0; h < NDEVICES && p->handler == NULL; h++)
p->handler = (*devices[h].iov2device)(type, p, iov, niov, maxiov,
auxfd, nauxfd);
if (p->handler == NULL) {
log_Printf(LogPHASE, "%s: Unknown link type\n", p->link.name);
free(iov[(*niov)++].iov_base);
physical_SetupStack(p, "unknown", PHYSICAL_NOFORCE);
} else
log_Printf(LogPHASE, "%s: Device %s, link type is %s\n",
p->link.name, p->name.full, p->handler->name);
if (p->hdlc.lqm.method && p->hdlc.lqm.timer.load)
lqr_reStart(&p->link.lcp);
hdlc_StartTimer(&p->hdlc);
throughput_restart(&p->link.stats.total, "physical throughput",
Enabled(dl->bundle, OPT_THROUGHPUT));
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
return p;
}
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
int
physical_MaxDeviceSize()
{
int biggest, sz, n;
biggest = sizeof(struct device);
for (sz = n = 0; n < NDEVICES; n++)
if (devices[n].DeviceSize) {
sz = (*devices[n].DeviceSize)();
if (biggest < sz)
biggest = sz;
}
return biggest;
}
int
physical2iov(struct physical *p, struct iovec *iov, int *niov, int maxiov,
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
int *auxfd, int *nauxfd)
{
struct device *h;
int sz;
h = NULL;
if (p) {
hdlc_StopTimer(&p->hdlc);
lqr_StopTimer(p);
timer_Stop(&p->link.lcp.fsm.FsmTimer);
timer_Stop(&p->link.ccp.fsm.FsmTimer);
timer_Stop(&p->link.lcp.fsm.OpenTimer);
timer_Stop(&p->link.ccp.fsm.OpenTimer);
timer_Stop(&p->link.lcp.fsm.StoppedTimer);
timer_Stop(&p->link.ccp.fsm.StoppedTimer);
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
if (p->handler) {
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
h = p->handler;
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
p->handler = (struct device *)(long)p->handler->type;
}
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
if (Enabled(p->dl->bundle, OPT_KEEPSESSION) ||
tcgetpgrp(p->fd) == getpgrp())
p->session_owner = getpid(); /* So I'll eventually get HUP'd */
else
p->session_owner = (pid_t)-1;
timer_Stop(&p->link.stats.total.Timer);
}
if (*niov + 2 >= maxiov) {
log_Printf(LogERROR, "physical2iov: No room for physical + throughput"
" + device !\n");
if (p)
free(p);
return -1;
}
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
iov[*niov].iov_base = (void *)p;
iov[*niov].iov_len = sizeof *p;
(*niov)++;
iov[*niov].iov_base = p ? (void *)p->link.stats.total.in.SampleOctets : NULL;
iov[*niov].iov_len = SAMPLE_PERIOD * sizeof(long long);
(*niov)++;
iov[*niov].iov_base = p ? (void *)p->link.stats.total.out.SampleOctets : NULL;
iov[*niov].iov_len = SAMPLE_PERIOD * sizeof(long long);
(*niov)++;
sz = physical_MaxDeviceSize();
if (p) {
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
if (h && h->device2iov)
(*h->device2iov)(h, iov, niov, maxiov, auxfd, nauxfd);
else {
iov[*niov].iov_base = malloc(sz);
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
if (h)
memcpy(iov[*niov].iov_base, h, sizeof *h);
iov[*niov].iov_len = sz;
(*niov)++;
}
} else {
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
iov[*niov].iov_base = NULL;
iov[*niov].iov_len = sz;
(*niov)++;
}
return p ? p->fd : 0;
}
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
const char *
physical_LockedDevice(struct physical *p)
{
if (p->fd >= 0 && *p->name.full == '/' && p->type != PHYS_DIRECT)
return p->name.base;
return NULL;
}
void
physical_ChangedPid(struct physical *p, pid_t newpid)
{
Rewrite the link descriptor transfer code in MP mode. Previously, ppp attempted to bind() to a local domain tcp socket based on the peer authname & enddisc. If it succeeded, it listen()ed and became MP server. If it failed, it connect()ed and became MP client. The server then select()ed on the descriptor, accept()ed it and wrote its pid to it then read the link data & link file descriptor, and finally sent an ack (``!''). The client would read() the server pid, transfer the link lock to that pid, send the link data & descriptor and read the ack. It would then close the descriptor and clean up. There was a race between the bind() and listen() where someone could attempt to connect() and fail. This change removes the race. Now ppp makes the RCVBUF big enough on a socket descriptor and attempts to bind() to a local domain *udp* socket (same name as before). If it succeeds, it becomes MP server. If it fails, it sets the SNDBUF and connect()s, becoming MP client. The server select()s on the descriptor and recvmsg()s the message, insisting on at least two descriptors (plus the link data). It uses the second descriptor to write() its pid then read()s an ack (``!''). The client creates a socketpair() and sendmsg()s the link data, link descriptor and one of the socketpair descriptors. It then read()s the server pid from the other socketpair descriptor, transfers any locks and write()s an ack. Now, there can be no race, and a connect() failure indicates a stale socket file. This also fixes MP ppp over ethernet, where the struct msghdr was being misconstructed when transferring the control socket descriptor. Also, if we fail to send the link, don't hang around in a ``session owner'' state, just do the setsid() and fork() if it's required to disown a tty. UDP idea suggested by: Chris Bennet from Mindspring at FreeBSDCon
1999-11-25 02:47:04 +00:00
if (physical_LockedDevice(p)) {
int res;
if ((res = ID0uu_lock_txfr(p->name.base, newpid)) != UU_LOCK_OK)
log_Printf(LogPHASE, "uu_lock_txfr: %s\n", uu_lockerr(res));
}
}
int
physical_IsSync(struct physical *p)
{
return p->cfg.speed == 0;
}
u_short
physical_DeviceMTU(struct physical *p)
{
return p->handler ? p->handler->mtu : 0;
}
const char *physical_GetDevice(struct physical *p)
{
return p->name.full;
}
void
physical_SetDeviceList(struct physical *p, int argc, const char *const *argv)
{
int f, pos;
p->cfg.devlist[sizeof p->cfg.devlist - 1] = '\0';
for (f = 0, pos = 0; f < argc && pos < sizeof p->cfg.devlist - 1; f++) {
if (pos)
p->cfg.devlist[pos++] = '\0';
strncpy(p->cfg.devlist + pos, argv[f], sizeof p->cfg.devlist - pos - 1);
pos += strlen(p->cfg.devlist + pos);
}
p->cfg.ndev = f;
}
void
physical_SetSync(struct physical *p)
{
p->cfg.speed = 0;
}
int
physical_SetRtsCts(struct physical *p, int enable)
{
p->cfg.rts_cts = enable ? 1 : 0;
return 1;
}
ssize_t
physical_Read(struct physical *p, void *buf, size_t nbytes)
{
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
ssize_t ret;
if (p->handler && p->handler->read)
ret = (*p->handler->read)(p, buf, nbytes);
else
ret = read(p->fd, buf, nbytes);
log_DumpBuff(LogPHYSICAL, "read", buf, ret);
return ret;
}
ssize_t
physical_Write(struct physical *p, const void *buf, size_t nbytes)
{
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
log_DumpBuff(LogPHYSICAL, "write", buf, nbytes);
if (p->handler && p->handler->write)
return (*p->handler->write)(p, buf, nbytes);
return write(p->fd, buf, nbytes);
}
1998-02-06 02:22:28 +00:00
int
physical_doUpdateSet(struct fdescriptor *d, fd_set *r, fd_set *w, fd_set *e,
int *n, int force)
{
struct physical *p = descriptor2physical(d);
int sets;
sets = 0;
if (p->fd >= 0) {
if (r) {
FD_SET(p->fd, r);
log_Printf(LogTIMER, "%s: fdset(r) %d\n", p->link.name, p->fd);
sets++;
}
if (e) {
FD_SET(p->fd, e);
log_Printf(LogTIMER, "%s: fdset(e) %d\n", p->link.name, p->fd);
sets++;
}
if (w && (force || link_QueueLen(&p->link) || p->out)) {
FD_SET(p->fd, w);
log_Printf(LogTIMER, "%s: fdset(w) %d\n", p->link.name, p->fd);
sets++;
}
if (sets && *n < p->fd + 1)
*n = p->fd + 1;
}
return sets;
}
int
physical_RemoveFromSet(struct physical *p, fd_set *r, fd_set *w, fd_set *e)
{
if (p->handler && p->handler->removefromset)
return (*p->handler->removefromset)(p, r, w, e);
else {
int sets;
sets = 0;
if (p->fd >= 0) {
if (r && FD_ISSET(p->fd, r)) {
FD_CLR(p->fd, r);
log_Printf(LogTIMER, "%s: fdunset(r) %d\n", p->link.name, p->fd);
sets++;
}
if (e && FD_ISSET(p->fd, e)) {
FD_CLR(p->fd, e);
log_Printf(LogTIMER, "%s: fdunset(e) %d\n", p->link.name, p->fd);
sets++;
}
if (w && FD_ISSET(p->fd, w)) {
FD_CLR(p->fd, w);
log_Printf(LogTIMER, "%s: fdunset(w) %d\n", p->link.name, p->fd);
sets++;
}
}
return sets;
}
}
int
physical_IsSet(struct fdescriptor *d, const fd_set *fdset)
{
struct physical *p = descriptor2physical(d);
return p->fd >= 0 && FD_ISSET(p->fd, fdset);
}
void
physical_Login(struct physical *p, const char *name)
{
if (p->type == PHYS_DIRECT && *p->name.base && !p->Utmp) {
struct utmp ut;
const char *connstr;
char *colon;
memset(&ut, 0, sizeof ut);
ut.ut_time = time(NULL);
strncpy(ut.ut_name, name, sizeof ut.ut_name);
if (p->handler && (p->handler->type == TCP_DEVICE ||
p->handler->type == UDP_DEVICE)) {
strncpy(ut.ut_line, PPPOTCPLINE, sizeof ut.ut_line);
strncpy(ut.ut_host, p->name.base, sizeof ut.ut_host);
colon = memchr(ut.ut_host, ':', sizeof ut.ut_host);
if (colon)
*colon = '\0';
} else
strncpy(ut.ut_line, p->name.base, sizeof ut.ut_line);
if ((connstr = getenv("CONNECT")))
/* mgetty sets this to the connection speed */
strncpy(ut.ut_host, connstr, sizeof ut.ut_host);
ID0login(&ut);
p->Utmp = ut.ut_time;
}
}
int
physical_SetMode(struct physical *p, int mode)
{
if ((p->type & (PHYS_DIRECT|PHYS_DEDICATED) ||
mode & (PHYS_DIRECT|PHYS_DEDICATED)) &&
(!(p->type & PHYS_DIRECT) || !(mode & PHYS_BACKGROUND))) {
/* Note: The -direct -> -background is for callback ! */
log_Printf(LogWARN, "%s: Cannot change mode %s to %s\n", p->link.name,
mode2Nam(p->type), mode2Nam(mode));
return 0;
}
p->type = mode;
return 1;
}
void
physical_DeleteQueue(struct physical *p)
{
if (p->out) {
m_freem(p->out);
p->out = NULL;
}
link_DeleteQueue(&p->link);
}
void
physical_SetDevice(struct physical *p, const char *name)
{
int len = strlen(_PATH_DEV);
if (name != p->name.full) {
strncpy(p->name.full, name, sizeof p->name.full - 1);
p->name.full[sizeof p->name.full - 1] = '\0';
}
p->name.base = *p->name.full == '!' ? p->name.full + 1 :
strncmp(p->name.full, _PATH_DEV, len) ?
p->name.full : p->name.full + len;
}
static void
physical_Found(struct physical *p)
{
FILE *lockfile;
char fn[PATH_MAX];
if (*p->name.full == '/') {
snprintf(fn, sizeof fn, "%s%s.if", _PATH_VARRUN, p->name.base);
lockfile = ID0fopen(fn, "w");
if (lockfile != NULL) {
fprintf(lockfile, "%s%d\n", TUN_NAME, p->dl->bundle->unit);
fclose(lockfile);
}
#ifndef RELEASE_CRUNCH
else
log_Printf(LogALERT, "%s: Can't create %s: %s\n",
p->link.name, fn, strerror(errno));
#endif
}
throughput_start(&p->link.stats.total, "physical throughput",
Enabled(p->dl->bundle, OPT_THROUGHPUT));
p->connect_count++;
p->input.sz = 0;
log_Printf(LogPHASE, "%s: Connected!\n", p->link.name);
}
int
physical_Open(struct physical *p, struct bundle *bundle)
{
int devno, h, wasfd, err;
char *dev;
if (p->fd >= 0)
log_Printf(LogDEBUG, "%s: Open: Modem is already open!\n", p->link.name);
/* We're going back into "term" mode */
else if (p->type == PHYS_DIRECT) {
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
physical_SetDevice(p, "");
p->fd = STDIN_FILENO;
for (h = 0; h < NDEVICES && p->handler == NULL && p->fd >= 0; h++)
p->handler = (*devices[h].create)(p);
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
if (p->fd >= 0) {
if (p->handler == NULL) {
physical_SetupStack(p, "unknown", PHYSICAL_NOFORCE);
log_Printf(LogDEBUG, "%s: stdin is unidentified\n", p->link.name);
}
physical_Found(p);
}
} else {
dev = p->cfg.devlist;
devno = 0;
while (devno < p->cfg.ndev && p->fd < 0) {
physical_SetDevice(p, dev);
if (physical_Lock(p)) {
err = 0;
if (*p->name.full == '/') {
p->fd = ID0open(p->name.full, O_RDWR | O_NONBLOCK);
if (p->fd < 0)
err = errno;
}
wasfd = p->fd;
for (h = 0; h < NDEVICES && p->handler == NULL; h++)
if ((p->handler = (*devices[h].create)(p)) == NULL && wasfd != p->fd)
break;
if (p->fd < 0) {
if (h == NDEVICES) {
if (err)
log_Printf(LogWARN, "%s: %s: %s\n", p->link.name, p->name.full,
strerror(errno));
else
log_Printf(LogWARN, "%s: Device (%s) must begin with a '/',"
" a '!' or contain at least one ':'\n", p->link.name,
p->name.full);
}
physical_Unlock(p);
} else
physical_Found(p);
}
dev += strlen(dev) + 1;
devno++;
}
}
return p->fd;
}
void
physical_SetupStack(struct physical *p, const char *who, int how)
{
link_EmptyStack(&p->link);
if (how == PHYSICAL_FORCE_SYNC || how == PHYSICAL_FORCE_SYNCNOACF ||
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
(how == PHYSICAL_NOFORCE && physical_IsSync(p)))
link_Stack(&p->link, &synclayer);
else {
link_Stack(&p->link, &asynclayer);
link_Stack(&p->link, &hdlclayer);
}
if (how != PHYSICAL_FORCE_SYNCNOACF)
link_Stack(&p->link, &acflayer);
link_Stack(&p->link, &protolayer);
link_Stack(&p->link, &lqrlayer);
link_Stack(&p->link, &ccplayer);
link_Stack(&p->link, &vjlayer);
link_Stack(&p->link, &tcpmsslayer);
#ifndef NONAT
link_Stack(&p->link, &natlayer);
#endif
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
if (how == PHYSICAL_FORCE_ASYNC && physical_IsSync(p)) {
log_Printf(LogWARN, "Sync device setting ignored for ``%s'' device\n", who);
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
p->cfg.speed = MODEM_SPEED;
} else if (how == PHYSICAL_FORCE_SYNC && !physical_IsSync(p)) {
log_Printf(LogWARN, "Async device setting ignored for ``%s'' device\n",
who);
Allow ``host:port/udp'' devices and support ``host:port/tcp'' as being the same as the previous (still supported) ``host:port'' syntax for tcp socket devices. A udp device uses synchronous ppp rather than async, and avoids the double-retransmit overhead that comes with ppp over tcp (it's usually a bad idea to transport IP over a reliable transport that itself is using an unreliable transport). PPP over UDP provides througput of ** 1.5Mb per second ** with all compression disabled, maxing out a PPro/200 when running ppp twice, back-to-back. This proves that PPPoE is plausable in userland.... This change adds a few more handler functions to struct device and allows derivations of struct device (which may contain their own data etc) to pass themselves through the unix domain socket for MP. ** At last **, struct physical has lost all the tty crud ! iov2physical() is now smart enough to restore the correct stack of layers so that MP servers will work again. The version number has bumped as our MP link transfer contents have changed (they now may contain a `struct device'). Don't extract the protocol twice in MP mode (resulting in protocol rejects for every MP packet). This was broken with my original layering changes. Add ``Physical'' and ``Sync'' log levels for logging the relevent raw packets and add protocol-tracking LogDEBUG stuff in various LayerPush & LayerPull functions. Assign our physical device name for incoming tcp connections by calling getpeername(). Assign our physical device name for incoming udp connections from the address retrieved by the first recvfrom().
1999-05-12 09:49:12 +00:00
physical_SetSync(p);
}
}
void
physical_StopDeviceTimer(struct physical *p)
{
if (p->handler && p->handler->stoptimer)
(*p->handler->stoptimer)(p);
}
int
physical_AwaitCarrier(struct physical *p)
{
if (p->handler && p->handler->awaitcarrier)
return (*p->handler->awaitcarrier)(p);
return CARRIER_OK;
}
void
physical_SetAsyncParams(struct physical *p, u_int32_t mymap, u_int32_t hismap)
{
if (p->handler && p->handler->setasyncparams)
return (*p->handler->setasyncparams)(p, mymap, hismap);
async_SetLinkParams(&p->async, mymap, hismap);
}
int
physical_Slot(struct physical *p)
{
if (p->handler && p->handler->slot)
return (*p->handler->slot)(p);
return -1;
}