d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
8f95b97072
freebsd-dev/usr.sbin/ppp/bundle.c

1932 lines
54 KiB
C
Raw Normal View History

Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
/*-
* Copyright (c) 1998 Brian Somers <brian@Awfulhak.org>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
$Id$ -> $FreeBSD$
1999-08-28 01:35:59 +00:00
* $FreeBSD$
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
*/
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
#include <sys/param.h>
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include <sys/socket.h>
#include <netinet/in.h>
#include <net/if.h>
Add ISDN support via isdnd & i4b. This requires version 0.81.1 of the i4b code - namely support of the I4B_VR_REQ ioctl via the i4brbchX device. Ppp controls the phone number, but idle timers and SYNC/RAW decisions are still made by isdnd (in isdnd.rc). This involves a new datalink state machine phase. The ``wait for carrier'' phase happens after dialing but before logging in. The whole dial state should really be abstracted so that each device type can deal with it in its own way (thinking about PPPoE) - but that'll have to wait. The ``set cd'' symantics remain the same for tty devices, but we now delay until we either get CD or timeout waiting (at which time we drop the link if we require CD). For i4b devices we always insist on carrier. Thanks to hm@ for his help, and especially for pointing out that I *don't* need to re-implement isdnd (that was a huge waste of time !) :-]
1999-08-06 20:04:08 +00:00
#include <net/if_tun.h> /* For TUNSIFMODE & TUNSLMODE */
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include <arpa/inet.h>
#include <net/route.h>
Move VJ compression state and stats into struct ipcp.
1998-03-16 22:54:35 +00:00
#include <netinet/in_systm.h>
#include <netinet/ip.h>
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
#include <sys/un.h>
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include <errno.h>
#include <fcntl.h>
Fix ``set proctitle'' by using setproctitle().
1999-11-17 21:12:35 +00:00
#ifdef __OpenBSD__
#include <util.h>
#else
#include <libutil.h>
#endif
o Use two `cat' processes to connect the modem to an already-running ppp. Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> o Use _PATH_DEVNULL rather than "/dev/null" o Be more paranoid about nuking running timers when transferring links.
1998-05-01 19:20:09 +00:00
#include <paths.h>
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include <stdio.h>
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
#include <stdlib.h>
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include <string.h>
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
#include <sys/uio.h>
o Don't try to transfer tty device descriptors as there's no way of transferring session rights with them. Instead, create two `/bin/cat' processes. A new child is spawned and disassociated from the terminal and the parent, which continues with the rest of the ppp process. Meanwhile, the parent spawns another child, and both the parent and child exec the `/bin/cat' commands with the appropriate descriptors. This way, the session is owned by the parent, and the tty is held open. o Close LCPs that have done a TLF and are now in ST_STOPPED before calling Down. This prevents them from trying to come back up again after the peer has shut them down (it seems a bit strange that the rfc says that a Down in ST_STOPPED will cause a TLS etc). o Don't try to set the physical link name pointer when we're receiving and renaming a datalink. The physical hasn't been created yet, and as it happens, the garbage physical pointer happens to be the value of another physical - so we're pointing that other physical name at ourselves. yeuck. o Re-arrange the order of things in main (DoLoop()). We now handle signals only after the select and not before the UpdateSet. It's possible that either a signal (FSM timeout) or a descriptor_Read() brings a link down, after which we'd better tidy up any dead direct and 1off descriptors before calling UpdateSet() again. o Mention when we detect a PPP packet when we see one before the link is up (then start LCP as before).
1998-05-25 02:22:38 +00:00
#include <sys/wait.h>
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
#if defined(__FreeBSD__) && !defined(NOKLDLOAD)
#include <sys/linker.h>
Use modfind() to check if a kld is already loaded. Submitted mostly by: green
1999-11-16 21:57:34 +00:00
#include <sys/module.h>
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
#endif
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include <termios.h>
#include <unistd.h>
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
#include "layer.h"
Fix a rather nasty use of `static'. This caused a SEGV when running ``link * load label'' as we ended up recursing back into command_Interpret after nuking our command arg list.
1998-06-15 19:06:25 +00:00
#include "defs.h"
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include "command.h"
#include "mbuf.h"
#include "log.h"
#include "id.h"
#include "timer.h"
#include "fsm.h"
#include "iplist.h"
De-staticise LQR information Increment OutPackets for any packet - not just LQRs MFC: o Fix a few comment typos. o Fix ``set timeout'' usage message and documentation. o Change ifOutPackets, ifOutOctets and ifOutLQRs to `u_int32_t's so that they wrap correctly. o Put the LQR in network byte order using the correct struct size (sizeof u_int32_t, not sizeof u_long). o Wrap LQR ECHO counters correctly. o Don't increment OutLQR count if the last LQR hasn't been replied to. o Initialise last received LQR in StartLqm. o Don't start the LQR timer if we're `disabled' and `accepted'. o Generate LQR responses when both sides are using a timer and we're not going to send our next LQR before the peers max timeout.
1998-03-13 00:44:26 +00:00
#include "lqr.h"
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
#include "hdlc.h"
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include "throughput.h"
Move VJ compression state and stats into struct ipcp.
1998-03-16 22:54:35 +00:00
#include "slcompress.h"
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include "ipcp.h"
Move filter sets into struct bundle.
1998-03-16 22:52:54 +00:00
#include "filter.h"
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
#include "descriptor.h"
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
#include "route.h"
#include "lcp.h"
#include "ccp.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 "link.h"
#include "mp.h"
Initial RADIUS support (using libradius). See the man page for details. Compiling with -DNORADIUS (the default for `release') removes support. TODO: The functionality in libradius::rad_send_request() needs to be supplied as a set of routines so that ppp doesn't have to wait indefinitely for the radius server(s). Instead, we need to get a descriptor back, select() on the descriptor, and ask libradius to service it when necessary. For now, ppp blocks SIGALRM while in rad_send_request(), so it misses PAP/CHAP retries & timeouts if they occur. Only PAP is functional. When CHAP is attempted, libradius complains that no User-Password has been specified... rfc2138 says that it *mustn't* be used for CHAP :-( Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-01-28 01:56:34 +00:00
#ifndef NORADIUS
#include "radius.h"
#endif
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 "bundle.h"
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
#include "async.h"
#include "physical.h"
#include "auth.h"
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
#include "proto.h"
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
#include "chap.h"
#include "tun.h"
Move the terminal/diagnostic socket stuff out of main.c and into the new `prompt.c'. struct prompt is (of course) a `sort' of descriptor.
1998-02-10 03:23:50 +00:00
#include "prompt.h"
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
#include "chat.h"
o Support callback types NONE, E.164, AUTH and CBCP. (see the new ``set callback'' and ``set cbcp'' commands) o Add a ``cbcp'' log level and mbuf type. o Don't dump core when \T is given in ``set login'' or ``set hangup''. o Allow ``*'' and blanks as placeholders in ppp.secret and allow a fifth field for specifying auth/cbcp dialback parameters. o Remove a few extraneous #includes o Define the default number of REQs (restart counter) in defs.h rather than hardcoding ``5'' all over the place. o Fix a few man page inconsistencies.
1998-08-07 18:42:51 +00:00
#include "cbcp.h"
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
#include "datalink.h"
#include "ip.h"
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
#include "iface.h"
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
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
#define SCATTER_SEGMENTS 6 /* version, datalink, name, physical,
throughput, device */
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
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
#define SEND_MAXFD 3 /* Max file descriptors passed through
the local domain socket */
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
static int bundle_RemainingIdleTime(struct bundle *);
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
static const char *PhaseNames[] = {
"Dead", "Establish", "Authenticate", "Network", "Terminate"
};
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
const char *
bundle_PhaseName(struct bundle *bundle)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
{
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
return bundle->phase <= PHASE_TERMINATE ?
PhaseNames[bundle->phase] : "unknown";
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
void
Remove the second bundle_NewPhase argument and do the tun config from bundle_LayerUp.
1998-03-13 00:44:33 +00:00
bundle_NewPhase(struct bundle *bundle, u_int new)
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
{
Introduce the ``link'' command for controlling individual links.
1998-02-17 19:28:01 +00:00
if (new == bundle->phase)
return;
Do authentication at the datalink level, not the bundle level. The bundle doesn't get a LayerUp 'till we're authenticated. Introduce DATALINK_LCP and DATALINK_AUTH phases.
1998-03-01 01:07:49 +00:00
if (new <= PHASE_TERMINATE)
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogPHASE, "bundle: %s\n", PhaseNames[new]);
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
switch (new) {
case PHASE_DEAD:
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
log_DisplayPrompts();
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
bundle->phase = new;
break;
case PHASE_ESTABLISH:
bundle->phase = new;
break;
case PHASE_AUTHENTICATE:
Do authentication at the datalink level, not the bundle level. The bundle doesn't get a LayerUp 'till we're authenticated. Introduce DATALINK_LCP and DATALINK_AUTH phases.
1998-03-01 01:07:49 +00:00
bundle->phase = new;
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
log_DisplayPrompts();
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
break;
case PHASE_NETWORK:
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
fsm_Up(&bundle->ncp.ipcp.fsm);
fsm_Open(&bundle->ncp.ipcp.fsm);
o Initialise MP correctly depending on if we're the first link in mp_Up(). o Bring MP and its CCP down when we enter phase TERMINATE, and ditch everything in the incoming packet queue. o Enable MRRU negotiation. Now, we can multilink mode, but only with one physical link. o Close the link if the peer PROTO REJs PROTO_MP. o Prepend our protocol before passing a packet to struct mp for fragmentation. o Log info messages to DEBUG, not ERROR (oops). o Align `show mp' output (again).
1998-04-23 21:50:13 +00:00
bundle->phase = new;
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
log_DisplayPrompts();
o Initialise MP correctly depending on if we're the first link in mp_Up(). o Bring MP and its CCP down when we enter phase TERMINATE, and ditch everything in the incoming packet queue. o Enable MRRU negotiation. Now, we can multilink mode, but only with one physical link. o Close the link if the peer PROTO REJs PROTO_MP. o Prepend our protocol before passing a packet to struct mp for fragmentation. o Log info messages to DEBUG, not ERROR (oops). o Align `show mp' output (again).
1998-04-23 21:50:13 +00:00
break;
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
case PHASE_TERMINATE:
bundle->phase = new;
o Initialise MP correctly depending on if we're the first link in mp_Up(). o Bring MP and its CCP down when we enter phase TERMINATE, and ditch everything in the incoming packet queue. o Enable MRRU negotiation. Now, we can multilink mode, but only with one physical link. o Close the link if the peer PROTO REJs PROTO_MP. o Prepend our protocol before passing a packet to struct mp for fragmentation. o Log info messages to DEBUG, not ERROR (oops). o Align `show mp' output (again).
1998-04-23 21:50:13 +00:00
mp_Down(&bundle->ncp.mp);
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
log_DisplayPrompts();
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
break;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
}
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
static void
bundle_LayerStart(void *v, struct fsm *fp)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
{
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
/* The given FSM is about to start up ! */
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
Tidy up the parent notification stuff for -background mode. Remove extraneous extern decls of tun_{in,out} and netfd.
1998-03-25 18:38:59 +00:00
static void
bundle_Notify(struct bundle *bundle, char c)
{
if (bundle->notify.fd != -1) {
if (write(bundle->notify.fd, &c, 1) == 1)
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogPHASE, "Parent notified of success.\n");
Tidy up the parent notification stuff for -background mode. Remove extraneous extern decls of tun_{in,out} and netfd.
1998-03-25 18:38:59 +00:00
else
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogPHASE, "Failed to notify parent of success.\n");
Tidy up the parent notification stuff for -background mode. Remove extraneous extern decls of tun_{in,out} and netfd.
1998-03-25 18:38:59 +00:00
close(bundle->notify.fd);
bundle->notify.fd = -1;
}
}
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
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
static void
bundle_ClearQueues(void *v)
{
struct bundle *bundle = (struct bundle *)v;
struct datalink *dl;
log_Printf(LogPHASE, "Clearing choked output queue\n");
timer_Stop(&bundle->choked.timer);
/*
* Emergency time:
*
* We've had a full queue for PACKET_DEL_SECS seconds without being
* able to get rid of any of the packets. We've probably given up
* on the redials at this point, and the queued data has almost
* definitely been timed out by the layer above. As this is preventing
* us from reading the TUN_NAME device (we don't want to buffer stuff
* indefinitely), we may as well nuke this data and start with a clean
* slate !
*
* Unfortunately, this has the side effect of shafting any compression
* dictionaries in use (causing the relevant RESET_REQ/RESET_ACK).
*/
Put the IP buffer queues into struct ipcp. Forgotten by: me
1998-08-26 17:39:37 +00:00
ip_DeleteQueue(&bundle->ncp.ipcp);
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
mp_DeleteQueue(&bundle->ncp.mp);
for (dl = bundle->links; dl; dl = dl->next)
physical_DeleteQueue(dl->physical);
}
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
static void
bundle_LinkAdded(struct bundle *bundle, struct datalink *dl)
{
bundle->phys_type.all |= dl->physical->type;
if (dl->state == DATALINK_OPEN)
bundle->phys_type.open |= dl->physical->type;
if ((bundle->phys_type.open & (PHYS_DEDICATED|PHYS_DDIAL))
!= bundle->phys_type.open && bundle->idle.timer.state == TIMER_STOPPED)
/* We may need to start our idle timer */
bundle_StartIdleTimer(bundle);
}
o Support callback types NONE, E.164, AUTH and CBCP. (see the new ``set callback'' and ``set cbcp'' commands) o Add a ``cbcp'' log level and mbuf type. o Don't dump core when \T is given in ``set login'' or ``set hangup''. o Allow ``*'' and blanks as placeholders in ppp.secret and allow a fifth field for specifying auth/cbcp dialback parameters. o Remove a few extraneous #includes o Define the default number of REQs (restart counter) in defs.h rather than hardcoding ``5'' all over the place. o Fix a few man page inconsistencies.
1998-08-07 18:42:51 +00:00
void
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
bundle_LinksRemoved(struct bundle *bundle)
{
struct datalink *dl;
bundle->phys_type.all = bundle->phys_type.open = 0;
for (dl = bundle->links; dl; dl = dl->next)
bundle_LinkAdded(bundle, dl);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
bundle_CalculateBandwidth(bundle);
mp_CheckAutoloadTimer(&bundle->ncp.mp);
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
if ((bundle->phys_type.open & (PHYS_DEDICATED|PHYS_DDIAL))
== bundle->phys_type.open)
bundle_StopIdleTimer(bundle);
}
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
static void
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
bundle_LayerUp(void *v, struct fsm *fp)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
{
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
/*
* The given fsm is now up
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
* If it's an LCP, adjust our phys_mode.open value and check the
* autoload timer.
* If it's the first NCP, calculate our bandwidth
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
* If it's the first NCP, set our ``upat'' time
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
* If it's the first NCP, start the idle timer.
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
* If it's an NCP, tell our -background parent to go away.
* If it's the first NCP, start the autoload timer
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
*/
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
struct bundle *bundle = (struct bundle *)v;
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
Remove the second bundle_NewPhase argument and do the tun config from bundle_LayerUp.
1998-03-13 00:44:33 +00:00
if (fp->proto == PROTO_LCP) {
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
struct physical *p = link2physical(fp->link);
bundle_LinkAdded(bundle, p->dl);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
mp_CheckAutoloadTimer(&bundle->ncp.mp);
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
} else if (fp->proto == PROTO_IPCP) {
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
bundle_CalculateBandwidth(fp->bundle);
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
time(&bundle->upat);
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
bundle_StartIdleTimer(bundle);
Tidy up the parent notification stuff for -background mode. Remove extraneous extern decls of tun_{in,out} and netfd.
1998-03-25 18:38:59 +00:00
bundle_Notify(bundle, EX_NORMAL);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
mp_CheckAutoloadTimer(&fp->bundle->ncp.mp);
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
static void
bundle_LayerDown(void *v, struct fsm *fp)
{
/*
* The given FSM has been told to come down.
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
* If it's our last NCP, stop the idle timer.
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
* If it's our last NCP, clear our ``upat'' value.
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
* If it's our last NCP, stop the autoload timer
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
* If it's an LCP, adjust our phys_type.open value and any timers.
o Make sure we adjust our min seq and process any outstanding queued incoming fragments when a link goes down. o Don't use the minimum sequence numbers of links that aren't open. o Understand sequence number wrapping when determining the minimum sequence number. o Add & adjust a few comments.
1998-05-23 17:05:28 +00:00
* If it's an LCP and we're in multilink mode, adjust our tun
* speed and make sure our minimum sequence number is adjusted.
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
*/
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
struct bundle *bundle = (struct bundle *)v;
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
if (fp->proto == PROTO_IPCP) {
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
bundle_StopIdleTimer(bundle);
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle->upat = 0;
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
mp_StopAutoloadTimer(&bundle->ncp.mp);
} else if (fp->proto == PROTO_LCP) {
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
bundle_LinksRemoved(bundle); /* adjust timers & phys_type values */
if (bundle->ncp.mp.active) {
struct datalink *dl;
struct datalink *lost;
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
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
lost = NULL;
for (dl = bundle->links; dl; dl = dl->next)
if (fp == &dl->physical->link.lcp.fsm)
lost = dl;
o Make sure we adjust our min seq and process any outstanding queued incoming fragments when a link goes down. o Don't use the minimum sequence numbers of links that aren't open. o Understand sequence number wrapping when determining the minimum sequence number. o Add & adjust a few comments.
1998-05-23 17:05:28 +00:00
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
bundle_CalculateBandwidth(bundle);
o Make sure we adjust our min seq and process any outstanding queued incoming fragments when a link goes down. o Don't use the minimum sequence numbers of links that aren't open. o Understand sequence number wrapping when determining the minimum sequence number. o Add & adjust a few comments.
1998-05-23 17:05:28 +00:00
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
if (lost)
mp_LinkLost(&bundle->ncp.mp, lost);
else
Change some log levels. ALERTs are only logged when something that can't happen happens or when everyone needs to know. ERRORs are only logged when something unexpected happens.
1998-06-16 19:40:42 +00:00
log_Printf(LogALERT, "Oops, lost an unrecognised datalink (%s) !\n",
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
fp->link->name);
}
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
}
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
}
static void
bundle_LayerFinish(void *v, struct fsm *fp)
{
/* The given fsm is now down (fp cannot be NULL)
*
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
* If it's the last LCP, fsm_Down all NCPs
* If it's the last NCP, fsm_Close all LCPs
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
*/
struct bundle *bundle = (struct bundle *)v;
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
struct datalink *dl;
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
o Don't immediately reply to the first LQR thinking it's a duplicate ('cos it compares with our initial values). o Move the LCP into struct datalink.
1998-03-13 21:07:14 +00:00
if (fp->proto == PROTO_IPCP) {
o FsmDown() and FsmClose() any NCP that's still around before going into phase DEAD when the last physical link disappears. o Don't go to phase TERMINATE if we're already phase DEAD.
1998-04-16 21:19:05 +00:00
if (bundle_Phase(bundle) != PHASE_DEAD)
bundle_NewPhase(bundle, PHASE_TERMINATE);
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
for (dl = bundle->links; dl; dl = dl->next)
When we get the last NCP TLD, close all datalinks with CLOSE_STAYDOWN rather than CLOSE_NORMAL.
1999-10-15 20:36:04 +00:00
datalink_Close(dl, CLOSE_STAYDOWN);
Create & use fsm2initial(), a function to bring a state machine back to ST_INITIAL without going through any unnecessary TLS/TLF pairs.
1998-06-20 00:19:42 +00:00
fsm2initial(fp);
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
} else if (fp->proto == PROTO_LCP) {
int others_active;
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
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
others_active = 0;
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
for (dl = bundle->links; dl; dl = dl->next)
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
if (fp != &dl->physical->link.lcp.fsm &&
dl->state != DATALINK_CLOSED && dl->state != DATALINK_HANGUP)
others_active++;
o Don't immediately reply to the first LQR thinking it's a duplicate ('cos it compares with our initial values). o Move the LCP into struct datalink.
1998-03-13 21:07:14 +00:00
Create & use fsm2initial(), a function to bring a state machine back to ST_INITIAL without going through any unnecessary TLS/TLF pairs.
1998-06-20 00:19:42 +00:00
if (!others_active)
fsm2initial(&bundle->ncp.ipcp.fsm);
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
}
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
int
bundle_LinkIsUp(const struct bundle *bundle)
{
Move the IPCP into struct bundle.
1998-03-13 21:07:46 +00:00
return bundle->ncp.ipcp.fsm.state == ST_OPENED;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
void
Make `close lcp' just close the LCP layer and not hangup. This is useful for slirp users that wish to get their shell back after the ppp session. `close' with no args still hangs up as expected. Required by: jmz
1998-06-15 19:05:27 +00:00
bundle_Close(struct bundle *bundle, const char *name, int how)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
{
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
/*
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
* Please close the given datalink.
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
* If name == NULL or name is the last datalink, fsm_Close all NCPs
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
* (except our MP)
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
* If it isn't the last datalink, just Close that datalink.
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
*/
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
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
struct datalink *dl, *this_dl;
int others_active;
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
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
others_active = 0;
this_dl = NULL;
for (dl = bundle->links; dl; dl = dl->next) {
if (name && !strcasecmp(name, dl->name))
this_dl = dl;
if (name == NULL || this_dl == dl) {
Make `close lcp' just close the LCP layer and not hangup. This is useful for slirp users that wish to get their shell back after the ppp session. `close' with no args still hangs up as expected. Required by: jmz
1998-06-15 19:05:27 +00:00
switch (how) {
case CLOSE_LCP:
datalink_DontHangup(dl);
/* fall through */
case CLOSE_STAYDOWN:
datalink_StayDown(dl);
break;
}
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
} else if (dl->state != DATALINK_CLOSED && dl->state != DATALINK_HANGUP)
others_active++;
}
if (name && this_dl == NULL) {
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogWARN, "%s: Invalid datalink name\n", name);
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
return;
}
if (!others_active) {
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
bundle_StopIdleTimer(bundle);
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
if (bundle->ncp.ipcp.fsm.state > ST_CLOSED ||
bundle->ncp.ipcp.fsm.state == ST_STARTING)
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
fsm_Close(&bundle->ncp.ipcp.fsm);
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
else {
Create & use fsm2initial(), a function to bring a state machine back to ST_INITIAL without going through any unnecessary TLS/TLF pairs.
1998-06-20 00:19:42 +00:00
fsm2initial(&bundle->ncp.ipcp.fsm);
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
for (dl = bundle->links; dl; dl = dl->next)
Make `close lcp' just close the LCP layer and not hangup. This is useful for slirp users that wish to get their shell back after the ppp session. `close' with no args still hangs up as expected. Required by: jmz
1998-06-15 19:05:27 +00:00
datalink_Close(dl, how);
Force IPCP back to initial in bundle_Close(). Add a few un-necessary includes to make us build under 2.2.5. These'll go when the code is cleaned up at the end.
1998-03-02 17:25:30 +00:00
}
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
} else if (this_dl && this_dl->state != DATALINK_CLOSED &&
this_dl->state != DATALINK_HANGUP)
Make `close lcp' just close the LCP layer and not hangup. This is useful for slirp users that wish to get their shell back after the ppp session. `close' with no args still hangs up as expected. Required by: jmz
1998-06-15 19:05:27 +00:00
datalink_Close(this_dl, how);
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
o Make sure we don't dereference NULL when we've lost all our links. o Use srandomdev() for __FreeBSD__ >= 2, not just >2. o Use srandom((time(NULL)^getpid())+random()), random() when we haven't got srandomdev().
1998-05-06 18:50:12 +00:00
void
Don't bring the modem offline or hangup when ``down lcp'' is done. Instead, behave like ``close lcp''.
1998-06-16 19:40:28 +00:00
bundle_Down(struct bundle *bundle, int how)
o Make sure we don't dereference NULL when we've lost all our links. o Use srandomdev() for __FreeBSD__ >= 2, not just >2. o Use srandom((time(NULL)^getpid())+random()), random() when we haven't got srandomdev().
1998-05-06 18:50:12 +00:00
{
struct datalink *dl;
for (dl = bundle->links; dl; dl = dl->next)
Don't bring the modem offline or hangup when ``down lcp'' is done. Instead, behave like ``close lcp''.
1998-06-16 19:40:28 +00:00
datalink_Down(dl, how);
o Make sure we don't dereference NULL when we've lost all our links. o Use srandomdev() for __FreeBSD__ >= 2, not just >2. o Use srandom((time(NULL)^getpid())+random()), random() when we haven't got srandomdev().
1998-05-06 18:50:12 +00:00
}
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
static int
bundle_UpdateSet(struct descriptor *d, fd_set *r, fd_set *w, fd_set *e, int *n)
{
struct bundle *bundle = descriptor2bundle(d);
struct datalink *dl;
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
int result, queued, nlinks;
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
result = 0;
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
/* If there are aren't many packets queued, look for some more. */
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
for (nlinks = 0, dl = bundle->links; dl; dl = dl->next)
nlinks++;
if (nlinks) {
Put the IP buffer queues into struct ipcp. Forgotten by: me
1998-08-26 17:39:37 +00:00
queued = r ? bundle_FillQueues(bundle) : ip_QueueLen(&bundle->ncp.ipcp);
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
if (r && (bundle->phase == PHASE_NETWORK ||
bundle->phys_type.all & PHYS_AUTO)) {
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
/* enough surplus so that we can tell if we're getting swamped */
o Split the two IPCP queues into three - one for FSM data (LCP/CCP/IPCP), one for urgent IP traffic and one for everything else. o Add the ``set urgent'' command for adjusting the list of urgent port numbers. The default urgent ports are 21, 22, 23, 513, 514, 543 and 544 (Ports 80 and 81 have been removed from the default priority list). o Increase the buffered packet threshold from 20 to 30. o Report the number of packets in the IP output queue and the list of urgent ports under ``show ipcp''.
1999-09-04 00:00:21 +00:00
if (queued < 30) {
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
/* Not enough - select() for more */
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
if (bundle->choked.timer.state == TIMER_RUNNING)
timer_Stop(&bundle->choked.timer); /* Not needed any more */
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
FD_SET(bundle->dev.fd, r);
if (*n < bundle->dev.fd + 1)
*n = bundle->dev.fd + 1;
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
log_Printf(LogTIMER, "%s: fdset(r) %d\n", TUN_NAME, bundle->dev.fd);
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
result++;
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
} else if (bundle->choked.timer.state == TIMER_STOPPED) {
bundle->choked.timer.func = bundle_ClearQueues;
bundle->choked.timer.name = "output choke";
bundle->choked.timer.load = bundle->cfg.choked.timeout * SECTICKS;
bundle->choked.timer.arg = bundle;
timer_Start(&bundle->choked.timer);
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
}
}
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
}
Decouple pap & chap output routines from the corresponding input routines and take advantage of the new init/continue interface in libradius. This allows a timely response on other links in an MP setup while RADIUS requests are in progress as well as the ability to handle other data from the peer in parallel. It should also make the future addition of PAM support trivial. While I'm in there, validate pap & chap header IDs if ``idcheck'' is enabled (the default) for other FSM packet types. NOTE: This involved integrating the generation of chap challenges and the validation of chap responses (and commenting what's going on in those routines). I currently have no way of testing ppps ability to respond to M$Chap CHALLENGEs correctly, so if someone could do the honours, it'd be much appreciated (it *looks* ok!). Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-02-06 02:54:47 +00:00
#ifndef NORADIUS
result += descriptor_UpdateSet(&bundle->radius.desc, r, w, e, n);
#endif
Update the write descriptor selection set for our data links *after* shuffling fragments from the IP queue into the individual link queues. This fixes a latency problem pointed out by: Bruce Evans <bde@zeta.org.au>
1998-06-07 00:16:37 +00:00
/* Which links need a select() ? */
for (dl = bundle->links; dl; dl = dl->next)
result += descriptor_UpdateSet(&dl->desc, r, w, e, n);
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
/*
* This *MUST* be called after the datalink UpdateSet()s as it
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
* might be ``holding'' one of the datalinks (death-row) and
* wants to be able to de-select() it from the descriptor set.
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
*/
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
result += descriptor_UpdateSet(&bundle->ncp.mp.server.desc, r, w, e, n);
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
return result;
}
static int
bundle_IsSet(struct descriptor *d, const fd_set *fdset)
{
struct bundle *bundle = descriptor2bundle(d);
struct datalink *dl;
for (dl = bundle->links; dl; dl = dl->next)
if (descriptor_IsSet(&dl->desc, fdset))
return 1;
Decouple pap & chap output routines from the corresponding input routines and take advantage of the new init/continue interface in libradius. This allows a timely response on other links in an MP setup while RADIUS requests are in progress as well as the ability to handle other data from the peer in parallel. It should also make the future addition of PAM support trivial. While I'm in there, validate pap & chap header IDs if ``idcheck'' is enabled (the default) for other FSM packet types. NOTE: This involved integrating the generation of chap challenges and the validation of chap responses (and commenting what's going on in those routines). I currently have no way of testing ppps ability to respond to M$Chap CHALLENGEs correctly, so if someone could do the honours, it'd be much appreciated (it *looks* ok!). Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-02-06 02:54:47 +00:00
#ifndef NORADIUS
if (descriptor_IsSet(&bundle->radius.desc, fdset))
return 1;
#endif
Check the MP server in bundle_IsSet().
1998-05-17 10:16:14 +00:00
if (descriptor_IsSet(&bundle->ncp.mp.server.desc, fdset))
return 1;
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
return FD_ISSET(bundle->dev.fd, fdset);
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
}
static void
bundle_DescriptorRead(struct descriptor *d, struct bundle *bundle,
const fd_set *fdset)
{
struct datalink *dl;
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
if (descriptor_IsSet(&bundle->ncp.mp.server.desc, fdset))
descriptor_Read(&bundle->ncp.mp.server.desc, bundle, fdset);
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
for (dl = bundle->links; dl; dl = dl->next)
if (descriptor_IsSet(&dl->desc, fdset))
descriptor_Read(&dl->desc, bundle, fdset);
Deglobalise `struct prompt': o Our diagnostic socket has its password set in the `set socket' line only (not in ppp.secret). o Passwords are per server socket (*VarAuthKey are gone) o Authority is per prompt (VarLocalAuth is gone). o Local logging is per prompt. o Add a `show who' command to see who's connected. No identd routine - just a `where the connection came from' display. o SIGUSR1 is disabled for now - we have no way of choosing a password for the socket created :-( Prompts are attached as a list of `struct descriptor's in struct bundle, and serviced under the bundles descriptor service routines. Ultimately, everything should be done like this. Cosmetic: o alphabeticalise SRCS in Makefile. o Add a few comments in command.h TODO: Start checking that we don't overflow the descriptor sets in select() now that we can have any number of descriptors.
1998-04-03 19:26:02 +00:00
Decouple pap & chap output routines from the corresponding input routines and take advantage of the new init/continue interface in libradius. This allows a timely response on other links in an MP setup while RADIUS requests are in progress as well as the ability to handle other data from the peer in parallel. It should also make the future addition of PAM support trivial. While I'm in there, validate pap & chap header IDs if ``idcheck'' is enabled (the default) for other FSM packet types. NOTE: This involved integrating the generation of chap challenges and the validation of chap responses (and commenting what's going on in those routines). I currently have no way of testing ppps ability to respond to M$Chap CHALLENGEs correctly, so if someone could do the honours, it'd be much appreciated (it *looks* ok!). Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-02-06 02:54:47 +00:00
#ifndef NORADIUS
if (descriptor_IsSet(&bundle->radius.desc, fdset))
descriptor_Read(&bundle->radius.desc, bundle, fdset);
#endif
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
if (FD_ISSET(bundle->dev.fd, fdset)) {
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
struct tun_data tun;
int n, pri;
/* something to read from tun */
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
n = read(bundle->dev.fd, &tun, sizeof tun);
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
if (n < 0) {
Change some log levels. ALERTs are only logged when something that can't happen happens or when everyone needs to know. ERRORs are only logged when something unexpected happens.
1998-06-16 19:40:42 +00:00
log_Printf(LogWARN, "read from %s: %s\n", TUN_NAME, strerror(errno));
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
return;
}
n -= sizeof tun - sizeof tun.data;
if (n <= 0) {
Change some log levels. ALERTs are only logged when something that can't happen happens or when everyone needs to know. ERRORs are only logged when something unexpected happens.
1998-06-16 19:40:42 +00:00
log_Printf(LogERROR, "read from %s: Only %d bytes read ?\n", TUN_NAME, n);
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
return;
}
if (!tun_check_header(tun, AF_INET))
return;
if (((struct ip *)tun.data)->ip_dst.s_addr ==
bundle->ncp.ipcp.my_ip.s_addr) {
/* we've been asked to send something addressed *to* us :( */
if (Enabled(bundle, OPT_LOOPBACK)) {
pri = PacketCheck(bundle, tun.data, n, &bundle->filter.in);
if (pri >= 0) {
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
n += sizeof tun - sizeof tun.data;
write(bundle->dev.fd, &tun, n);
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
log_Printf(LogDEBUG, "Looped back packet addressed to myself\n");
}
return;
} else
log_Printf(LogDEBUG, "Oops - forwarding packet addressed to myself\n");
}
/*
* Process on-demand dialup. Output packets are queued within tunnel
* device until IPCP is opened.
*/
if (bundle_Phase(bundle) == PHASE_DEAD) {
/*
* Note, we must be in AUTO mode :-/ otherwise our interface should
* *not* be UP and we can't receive data
*/
if ((pri = PacketCheck(bundle, tun.data, n, &bundle->filter.dial)) >= 0)
Allow an optional ``!'' in the open, dial & call commands. When used, the redial timer is ignored and the modem is opened immediately.
1998-07-29 18:21:17 +00:00
bundle_Open(bundle, NULL, PHYS_AUTO, 0);
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
else
/*
* Drop the packet. If we were to queue it, we'd just end up with
* a pile of timed-out data in our output queue by the time we get
* around to actually dialing. We'd also prematurely reach the
* threshold at which we stop select()ing to read() the tun
* device - breaking auto-dial.
*/
return;
}
pri = PacketCheck(bundle, tun.data, n, &bundle->filter.out);
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
if (pri >= 0)
Put the IP buffer queues into struct ipcp. Forgotten by: me
1998-08-26 17:39:37 +00:00
ip_Enqueue(&bundle->ncp.ipcp, pri, tun.data, n);
o Don't enter phase TERMINATE in the LCP TLD, leave it exclusively for the last NCP TLF. o Move tun reading from the main loop into the bundle descriptor handling routines. o Cosmetic: Add a few `const's and make some diagnostics clearer.
1998-05-06 18:49:45 +00:00
}
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
}
o If we come out of select() with only write descriptors that end up writing zero bytes, sleep for 1/10 of a second so that we don't end up using up too much cpu. This should only ever happen on systems that wrongly report a descriptor as writable despite the tty buffer being full. Discussed with: Jeff Evarts o Do an initial run-time check to see if select() alters the passed timeval. This knowledge isn't yet used, but will be soon.
1998-06-24 19:33:36 +00:00
static int
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
bundle_DescriptorWrite(struct descriptor *d, struct bundle *bundle,
const fd_set *fdset)
{
struct datalink *dl;
o If we come out of select() with only write descriptors that end up writing zero bytes, sleep for 1/10 of a second so that we don't end up using up too much cpu. This should only ever happen on systems that wrongly report a descriptor as writable despite the tty buffer being full. Discussed with: Jeff Evarts o Do an initial run-time check to see if select() alters the passed timeval. This knowledge isn't yet used, but will be soon.
1998-06-24 19:33:36 +00:00
int result = 0;
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
/* This is not actually necessary as struct mpserver doesn't Write() */
if (descriptor_IsSet(&bundle->ncp.mp.server.desc, fdset))
descriptor_Write(&bundle->ncp.mp.server.desc, bundle, fdset);
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
for (dl = bundle->links; dl; dl = dl->next)
if (descriptor_IsSet(&dl->desc, fdset))
o If we come out of select() with only write descriptors that end up writing zero bytes, sleep for 1/10 of a second so that we don't end up using up too much cpu. This should only ever happen on systems that wrongly report a descriptor as writable despite the tty buffer being full. Discussed with: Jeff Evarts o Do an initial run-time check to see if select() alters the passed timeval. This knowledge isn't yet used, but will be soon.
1998-06-24 19:33:36 +00:00
result += descriptor_Write(&dl->desc, bundle, fdset);
return result;
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
}
Put the correct pid in /var/run/tunX.pid Suggested by: many
1998-06-06 20:50:57 +00:00
void
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
bundle_LockTun(struct bundle *bundle)
{
FILE *lockfile;
char pidfile[MAXPATHLEN];
snprintf(pidfile, sizeof pidfile, "%stun%d.pid", _PATH_VARRUN, bundle->unit);
lockfile = ID0fopen(pidfile, "w");
if (lockfile != NULL) {
fprintf(lockfile, "%d\n", (int)getpid());
fclose(lockfile);
}
#ifndef RELEASE_CRUNCH
else
log_Printf(LogERROR, "Warning: Can't create %s: %s\n",
pidfile, strerror(errno));
#endif
}
static void
bundle_UnlockTun(struct bundle *bundle)
{
char pidfile[MAXPATHLEN];
snprintf(pidfile, sizeof pidfile, "%stun%d.pid", _PATH_VARRUN, bundle->unit);
ID0unlink(pidfile);
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
struct bundle *
Fix ``set proctitle'' by using setproctitle().
1999-11-17 21:12:35 +00:00
bundle_Create(const char *prefix, int type, int unit)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
{
static struct bundle bundle; /* there can be only one */
Add the -unit command line switch for specifying the tun device. Warn about -alias being depricated (but still allow it). Don't moan twice about failing to open any tun device. Fix a diagnostic and add the -quiet switch to the usage message.
1999-10-19 15:21:09 +00:00
int enoentcount, err, minunit, maxunit;
Move the code for tweaking interface flags into one function.
1999-05-27 08:42:17 +00:00
const char *ifname;
Use modfind() to check if a kld is already loaded. Submitted mostly by: green
1999-11-16 21:57:34 +00:00
#if defined(__FreeBSD__) && !defined(NOKLDLOAD)
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
int kldtried;
#endif
Disable TUNSLMODE when we first open the tun device. Submitted by: Ian West <ian@niw.com.au>
1999-06-22 11:31:42 +00:00
#if defined(TUNSIFMODE) || defined(TUNSLMODE)
Undo possible damage done by the new TUNSIFMODE ioctl in FreeBSD-current.
1999-03-25 11:37:51 +00:00
int iff;
#endif
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
if (bundle.iface != NULL) { /* Already allocated ! */
Change some log levels. ALERTs are only logged when something that can't happen happens or when everyone needs to know. ERRORs are only logged when something unexpected happens.
1998-06-16 19:40:42 +00:00
log_Printf(LogALERT, "bundle_Create: There's only one BUNDLE !\n");
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
return NULL;
}
Add the -unit command line switch for specifying the tun device. Warn about -alias being depricated (but still allow it). Don't moan twice about failing to open any tun device. Fix a diagnostic and add the -quiet switch to the usage message.
1999-10-19 15:21:09 +00:00
if (unit == -1) {
minunit = 0;
maxunit = -1;
} else {
minunit = unit;
maxunit = unit + 1;
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
err = ENOENT;
enoentcount = 0;
Use modfind() to check if a kld is already loaded. Submitted mostly by: green
1999-11-16 21:57:34 +00:00
#if defined(__FreeBSD__) && !defined(NOKLDLOAD)
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
kldtried = 0;
#endif
Add the -unit command line switch for specifying the tun device. Warn about -alias being depricated (but still allow it). Don't moan twice about failing to open any tun device. Fix a diagnostic and add the -quiet switch to the usage message.
1999-10-19 15:21:09 +00:00
for (bundle.unit = minunit; bundle.unit != maxunit; bundle.unit++) {
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
snprintf(bundle.dev.Name, sizeof bundle.dev.Name, "%s%d",
prefix, bundle.unit);
bundle.dev.fd = ID0open(bundle.dev.Name, O_RDWR);
if (bundle.dev.fd >= 0)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
break;
o Don't limit our tun device number to 256. As long as there's another /dev/ entry, keep trying to open them. o Don't allow ``open ccp'' if lcp isn't open.
1998-04-27 01:40:38 +00:00
else if (errno == ENXIO) {
Use modfind() to check if a kld is already loaded. Submitted mostly by: green
1999-11-16 21:57:34 +00:00
#if defined(__FreeBSD__) && !defined(NOKLDLOAD)
Add the -unit command line switch for specifying the tun device. Warn about -alias being depricated (but still allow it). Don't moan twice about failing to open any tun device. Fix a diagnostic and add the -quiet switch to the usage message.
1999-10-19 15:21:09 +00:00
if (bundle.unit == minunit && !kldtried++) {
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
/*
Use modfind() to check if a kld is already loaded. Submitted mostly by: green
1999-11-16 21:57:34 +00:00
* Attempt to load the tunnel interface KLD if it isn't loaded
* already.
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
*/
Use modfind() to check if a kld is already loaded. Submitted mostly by: green
1999-11-16 21:57:34 +00:00
if (modfind("if_tun") == -1) {
Do a kldload() if we get ENXIO trying to open /dev/tun0 Originally submitted by: green
1999-09-22 00:40:47 +00:00
if (ID0kldload("if_tun") != -1) {
bundle.unit--;
continue;
}
log_Printf(LogWARN, "kldload: if_tun: %s\n", strerror(errno));
}
}
#endif
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
err = errno;
o Don't limit our tun device number to 256. As long as there's another /dev/ entry, keep trying to open them. o Don't allow ``open ccp'' if lcp isn't open.
1998-04-27 01:40:38 +00:00
break;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
} else if (errno == ENOENT) {
if (++enoentcount > 2)
o Don't limit our tun device number to 256. As long as there's another /dev/ entry, keep trying to open them. o Don't allow ``open ccp'' if lcp isn't open.
1998-04-27 01:40:38 +00:00
break;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
} else
err = errno;
}
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
if (bundle.dev.fd < 0) {
Add the -unit command line switch for specifying the tun device. Warn about -alias being depricated (but still allow it). Don't moan twice about failing to open any tun device. Fix a diagnostic and add the -quiet switch to the usage message.
1999-10-19 15:21:09 +00:00
if (unit == -1)
log_Printf(LogWARN, "No available tunnel devices found (%s)\n",
strerror(err));
else
log_Printf(LogWARN, "%s%d: %s\n", prefix, unit, strerror(err));
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
return NULL;
}
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_SetTun(bundle.unit);
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
ifname = strrchr(bundle.dev.Name, '/');
if (ifname == NULL)
ifname = bundle.dev.Name;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
else
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
ifname++;
bundle.iface = iface_Create(ifname);
if (bundle.iface == NULL) {
close(bundle.dev.fd);
return NULL;
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
Undo possible damage done by the new TUNSIFMODE ioctl in FreeBSD-current.
1999-03-25 11:37:51 +00:00
#ifdef TUNSIFMODE
/* Make sure we're POINTOPOINT */
iff = IFF_POINTOPOINT;
if (ID0ioctl(bundle.dev.fd, TUNSIFMODE, &iff) < 0)
log_Printf(LogERROR, "bundle_Create: ioctl(TUNSIFMODE): %s\n",
strerror(errno));
#endif
Disable TUNSLMODE when we first open the tun device. Submitted by: Ian West <ian@niw.com.au>
1999-06-22 11:31:42 +00:00
#ifdef TUNSLMODE
/* Make sure we're POINTOPOINT */
iff = 0;
if (ID0ioctl(bundle.dev.fd, TUNSLMODE, &iff) < 0)
log_Printf(LogERROR, "bundle_Create: ioctl(TUNSLMODE): %s\n",
strerror(errno));
#endif
Move the code for tweaking interface flags into one function.
1999-05-27 08:42:17 +00:00
if (!iface_SetFlags(bundle.iface, IFF_UP)) {
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
iface_Destroy(bundle.iface);
bundle.iface = NULL;
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
close(bundle.dev.fd);
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
return NULL;
}
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
log_Printf(LogPHASE, "Using interface: %s\n", ifname);
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
bundle.bandwidth = 0;
Bring the routing seqno into struct bundle.
1998-02-02 19:33:02 +00:00
bundle.routing_seq = 0;
Move `CleaningUp' into struct bundle.
1998-03-16 22:53:15 +00:00
bundle.phase = PHASE_DEAD;
bundle.CleaningUp = 0;
o Add the -foreground switch. This switch behaves like -background except that ppp stays in the foreground. o Add the -quiet switch to quieten ppps startup o Add the -nat flag and discourage the use of the -alias flag. Both do the same thing. o Correct some nat usage strings. o Change the internal ``alias'' command to ``nat''.
1999-08-19 18:15:52 +00:00
bundle.NatEnabled = 0;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
bundle.fsm.LayerStart = bundle_LayerStart;
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
bundle.fsm.LayerUp = bundle_LayerUp;
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
bundle.fsm.LayerDown = bundle_LayerDown;
bundle.fsm.LayerFinish = bundle_LayerFinish;
bundle.fsm.object = &bundle;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle.cfg.idle.timeout = NCP_IDLE_TIMEOUT;
bundle.cfg.idle.min_timeout = 0;
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
*bundle.cfg.auth.name = '\0';
*bundle.cfg.auth.key = '\0';
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
bundle.cfg.opt = OPT_SROUTES | OPT_IDCHECK | OPT_LOOPBACK |
OPT_THROUGHPUT | OPT_UTMP;
Negotiate MRRU, SHORTSEQ and ENDDISC. ENDDISC doesn't imply multilink ('cos I've seen my ISP REQ it without multilink). Setting MRRU is ifdef'd out until it's debugged and we can merge -direct links with other running programs. Fix MTU setting.
1998-04-23 03:23:03 +00:00
*bundle.cfg.label = '\0';
bundle.cfg.mtu = DEF_MTU;
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
bundle.cfg.choked.timeout = CHOKED_TIMEOUT;
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
bundle.phys_type.all = type;
bundle.phys_type.open = 0;
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle.upat = 0;
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
bundle.links = datalink_Create("deflink", &bundle, type);
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
if (bundle.links == NULL) {
Change some log levels. ALERTs are only logged when something that can't happen happens or when everyone needs to know. ERRORs are only logged when something unexpected happens.
1998-06-16 19:40:42 +00:00
log_Printf(LogALERT, "Cannot create data link: %s\n", strerror(errno));
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
iface_Destroy(bundle.iface);
bundle.iface = NULL;
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
close(bundle.dev.fd);
Move pppVars.physical into the bundle (for the moment).
1998-02-06 02:23:48 +00:00
return NULL;
}
Make struct bundle into a sort of `struct descriptor'. It does the fdsets/reads/writes for each of it's datalinks.
1998-03-20 19:48:28 +00:00
bundle.desc.type = BUNDLE_DESCRIPTOR;
bundle.desc.UpdateSet = bundle_UpdateSet;
bundle.desc.IsSet = bundle_IsSet;
bundle.desc.Read = bundle_DescriptorRead;
bundle.desc.Write = bundle_DescriptorWrite;
Negotiate MRRU, SHORTSEQ and ENDDISC. ENDDISC doesn't imply multilink ('cos I've seen my ISP REQ it without multilink). Setting MRRU is ifdef'd out until it's debugged and we can merge -direct links with other running programs. Fix MTU setting.
1998-04-23 03:23:03 +00:00
mp_Init(&bundle.ncp.mp, &bundle);
/* Send over the first physical link by default */
Move the IPCP into struct bundle.
1998-03-13 21:07:46 +00:00
ipcp_Init(&bundle.ncp.ipcp, &bundle, &bundle.links->physical->link,
&bundle.fsm);
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
Move filter sets into struct bundle.
1998-03-16 22:52:54 +00:00
memset(&bundle.filter, '\0', sizeof bundle.filter);
bundle.filter.in.fragok = bundle.filter.in.logok = 1;
bundle.filter.in.name = "IN";
bundle.filter.out.fragok = bundle.filter.out.logok = 1;
bundle.filter.out.name = "OUT";
bundle.filter.dial.name = "DIAL";
o Drop packets that fail the dial filter when we're in phase DEAD. They'll almost definitely have timed out by the time we dial anyway. o Log dial filters again (LogTCPIP). o Make DEBUG diagnostics for filter checking actually mean something to the common observer. o Do our best to keep any already-configured IP numbers at IPCP negotiation time. We always first request our configured IP, and if the peer asks for an invalid IP, we NAK with HISADDR Cosmetic: o Add a linefeed to the `set timeout' arg count error message. o Log unacceptable address errors to LogPHASE if LogIPCP is switched off. o Fix ``destination system not found'' error message. o Get out immediately if we get a fatal error before entering the main loop.
1998-04-03 19:25:07 +00:00
bundle.filter.dial.logok = 1;
Move filter sets into struct bundle.
1998-03-16 22:52:54 +00:00
bundle.filter.alive.name = "ALIVE";
bundle.filter.alive.logok = 1;
o Overhaul filtering, adding facilities to jump over rules and to negate the sense of rules. o Remove the redundant (and undocumented) ``host'' and ``port'' words (README.changes updated). o Don't permit (and ignore) garbage instead of the protocol. Mostly submitted by: Peter Jeremy <jeremyp@gsmx07.alcatel.com.au>
1999-07-27 23:44:00 +00:00
{
int i;
for (i = 0; i < MAXFILTERS; i++) {
bundle.filter.in.rule[i].f_action = A_NONE;
bundle.filter.out.rule[i].f_action = A_NONE;
bundle.filter.dial.rule[i].f_action = A_NONE;
bundle.filter.alive.rule[i].f_action = A_NONE;
}
}
Display the correct amount of remaining time in `show time'.
1998-03-18 23:15:31 +00:00
memset(&bundle.idle.timer, '\0', sizeof bundle.idle.timer);
bundle.idle.done = 0;
Tidy up the parent notification stuff for -background mode. Remove extraneous extern decls of tun_{in,out} and netfd.
1998-03-25 18:38:59 +00:00
bundle.notify.fd = -1;
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
memset(&bundle.choked.timer, '\0', sizeof bundle.choked.timer);
Initial RADIUS support (using libradius). See the man page for details. Compiling with -DNORADIUS (the default for `release') removes support. TODO: The functionality in libradius::rad_send_request() needs to be supplied as a set of routines so that ppp doesn't have to wait indefinitely for the radius server(s). Instead, we need to get a descriptor back, select() on the descriptor, and ask libradius to service it when necessary. For now, ppp blocks SIGALRM while in rad_send_request(), so it misses PAP/CHAP retries & timeouts if they occur. Only PAP is functional. When CHAP is attempted, libradius complains that no User-Password has been specified... rfc2138 says that it *mustn't* be used for CHAP :-( Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-01-28 01:56:34 +00:00
#ifndef NORADIUS
radius_Init(&bundle.radius);
#endif
Display the correct amount of remaining time in `show time'.
1998-03-18 23:15:31 +00:00
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
/* Clean out any leftover crud */
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
iface_Clear(bundle.iface, IFACE_CLEAR_ALL);
Create struct fsm_parent. FSMs are created with one of these, and the FSM passes subsequent events to them. The datalink now hides its CCP from the bundle layer.
1998-02-27 01:22:39 +00:00
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
bundle_LockTun(&bundle);
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
return &bundle;
}
o Add bundle_Destroy(). o Don't remove routes and DOWN the interface when we're closed in auto mode. o Initialize the FSMs in bundle_Create. o Initialize ipcp::if_mine & ipcp::if_peer only once (so that we don't forget that we've SIOCAIFADDR'd the interface). o Do a SIOCDIFADDR on the specific address to avoid hurting any other (still non-existent) NCPs. o Fix some error/diagnostic messages.
1998-02-08 19:29:45 +00:00
static void
bundle_DownInterface(struct bundle *bundle)
{
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
route_IfDelete(bundle, 1);
Move the code for tweaking interface flags into one function.
1999-05-27 08:42:17 +00:00
iface_ClearFlags(bundle->iface, IFF_UP);
o Add bundle_Destroy(). o Don't remove routes and DOWN the interface when we're closed in auto mode. o Initialize the FSMs in bundle_Create. o Initialize ipcp::if_mine & ipcp::if_peer only once (so that we don't forget that we've SIOCAIFADDR'd the interface). o Do a SIOCDIFADDR on the specific address to avoid hurting any other (still non-existent) NCPs. o Fix some error/diagnostic messages.
1998-02-08 19:29:45 +00:00
}
void
bundle_Destroy(struct bundle *bundle)
{
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
struct datalink *dl;
Deglobalise `struct prompt': o Our diagnostic socket has its password set in the `set socket' line only (not in ppp.secret). o Passwords are per server socket (*VarAuthKey are gone) o Authority is per prompt (VarLocalAuth is gone). o Local logging is per prompt. o Add a `show who' command to see who's connected. No identd routine - just a `where the connection came from' display. o SIGUSR1 is disabled for now - we have no way of choosing a password for the socket created :-( Prompts are attached as a list of `struct descriptor's in struct bundle, and serviced under the bundles descriptor service routines. Ultimately, everything should be done like this. Cosmetic: o alphabeticalise SRCS in Makefile. o Add a few comments in command.h TODO: Start checking that we don't overflow the descriptor sets in select() now that we can have any number of descriptors.
1998-04-03 19:26:02 +00:00
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
/*
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
* Clean up the interface. We don't need to timer_Stop()s, mp_Down(),
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
* ipcp_CleanInterface() and bundle_DownInterface() unless we're getting
* out under exceptional conditions such as a descriptor exception.
*/
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
timer_Stop(&bundle->idle.timer);
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
timer_Stop(&bundle->choked.timer);
o Do an mp_Down() when aborting ppp. o Re-assign the link name when strdup()ing the datalink name. o Add the link name to some more hdlc log messages.
1998-05-11 23:39:30 +00:00
mp_Down(&bundle->ncp.mp);
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
ipcp_CleanInterface(&bundle->ncp.ipcp);
bundle_DownInterface(bundle);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
Initial RADIUS support (using libradius). See the man page for details. Compiling with -DNORADIUS (the default for `release') removes support. TODO: The functionality in libradius::rad_send_request() needs to be supplied as a set of routines so that ppp doesn't have to wait indefinitely for the radius server(s). Instead, we need to get a descriptor back, select() on the descriptor, and ask libradius to service it when necessary. For now, ppp blocks SIGALRM while in rad_send_request(), so it misses PAP/CHAP retries & timeouts if they occur. Only PAP is functional. When CHAP is attempted, libradius complains that no User-Password has been specified... rfc2138 says that it *mustn't* be used for CHAP :-( Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-01-28 01:56:34 +00:00
#ifndef NORADIUS
/* Tell the radius server the bad news */
radius_Destroy(&bundle->radius);
#endif
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
/* Again, these are all DATALINK_CLOSED unless we're abending */
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
dl = bundle->links;
while (dl)
dl = datalink_Destroy(dl);
o Split the two IPCP queues into three - one for FSM data (LCP/CCP/IPCP), one for urgent IP traffic and one for everything else. o Add the ``set urgent'' command for adjusting the list of urgent port numbers. The default urgent ports are 21, 22, 23, 513, 514, 543 and 544 (Ports 80 and 81 have been removed from the default priority list). o Increase the buffered packet threshold from 20 to 30. o Report the number of packets in the IP output queue and the list of urgent ports under ``show ipcp''.
1999-09-04 00:00:21 +00:00
ipcp_Destroy(&bundle->ncp.ipcp);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
close(bundle->dev.fd);
bundle_UnlockTun(bundle);
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
/* In case we never made PHASE_NETWORK */
Tidy up the parent notification stuff for -background mode. Remove extraneous extern decls of tun_{in,out} and netfd.
1998-03-25 18:38:59 +00:00
bundle_Notify(bundle, EX_ERRDEAD);
Deglobalise `struct prompt': o Our diagnostic socket has its password set in the `set socket' line only (not in ppp.secret). o Passwords are per server socket (*VarAuthKey are gone) o Authority is per prompt (VarLocalAuth is gone). o Local logging is per prompt. o Add a `show who' command to see who's connected. No identd routine - just a `where the connection came from' display. o SIGUSR1 is disabled for now - we have no way of choosing a password for the socket created :-( Prompts are attached as a list of `struct descriptor's in struct bundle, and serviced under the bundles descriptor service routines. Ultimately, everything should be done like this. Cosmetic: o alphabeticalise SRCS in Makefile. o Add a few comments in command.h TODO: Start checking that we don't overflow the descriptor sets in select() now that we can have any number of descriptors.
1998-04-03 19:26:02 +00:00
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
iface_Destroy(bundle->iface);
bundle->iface = NULL;
o Add bundle_Destroy(). o Don't remove routes and DOWN the interface when we're closed in auto mode. o Initialize the FSMs in bundle_Create. o Initialize ipcp::if_mine & ipcp::if_peer only once (so that we don't forget that we've SIOCAIFADDR'd the interface). o Do a SIOCDIFADDR on the specific address to avoid hurting any other (still non-existent) NCPs. o Fix some error/diagnostic messages.
1998-02-08 19:29:45 +00:00
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
struct rtmsg {
struct rt_msghdr m_rtm;
char m_space[64];
};
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
int
Bring the routing seqno into struct bundle.
1998-02-02 19:33:02 +00:00
bundle_SetRoute(struct bundle *bundle, int cmd, struct in_addr dst,
Silence ``Network unreachable'' warnings when using ``add .... HISADDR''. The network will never be reachable at this point unless we're in -auto or reading the command from ppp.linkup. We can now run the following lines and get the expected results: set ifaddr 1.2.3.4/0 5.6.7.8/0 add default HISADDR where a route is added immediately in auto mode and the whole thing is delayed 'till the IP numbers have been agreed in other modes. Essentially, ppp.linkup is no longer required.
1998-07-28 21:54:54 +00:00
struct in_addr gateway, struct in_addr mask, int bang, int ssh)
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
{
struct rtmsg rtmes;
int s, nb, wb;
char *cp;
const char *cmdstr;
struct sockaddr_in rtdata;
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
int result = 1;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
if (bang)
cmdstr = (cmd == RTM_ADD ? "Add!" : "Delete!");
else
cmdstr = (cmd == RTM_ADD ? "Add" : "Delete");
s = ID0socket(PF_ROUTE, SOCK_RAW, 0);
if (s < 0) {
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogERROR, "bundle_SetRoute: socket(): %s\n", strerror(errno));
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
return result;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
memset(&rtmes, '\0', sizeof rtmes);
rtmes.m_rtm.rtm_version = RTM_VERSION;
rtmes.m_rtm.rtm_type = cmd;
rtmes.m_rtm.rtm_addrs = RTA_DST;
Bring the routing seqno into struct bundle.
1998-02-02 19:33:02 +00:00
rtmes.m_rtm.rtm_seq = ++bundle->routing_seq;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
rtmes.m_rtm.rtm_pid = getpid();
rtmes.m_rtm.rtm_flags = RTF_UP | RTF_GATEWAY | RTF_STATIC;
Add ``enable proxyall'' support. This adds proxy ARP entries for every machine on every class C or smaller subnet that we route to. Add ``set {send,recv}pipe'' for controlling our socket buffer sizes. Mention the IP number with the problem in a few error messages. All submitted by: Craig Leres <leres@ee.lbl.gov> Modified slightly by: me
1998-10-26 19:07:39 +00:00
if (cmd == RTM_ADD || cmd == RTM_CHANGE) {
if (bundle->ncp.ipcp.cfg.sendpipe > 0) {
rtmes.m_rtm.rtm_rmx.rmx_sendpipe = bundle->ncp.ipcp.cfg.sendpipe;
rtmes.m_rtm.rtm_inits |= RTV_SPIPE;
}
if (bundle->ncp.ipcp.cfg.recvpipe > 0) {
rtmes.m_rtm.rtm_rmx.rmx_recvpipe = bundle->ncp.ipcp.cfg.recvpipe;
rtmes.m_rtm.rtm_inits |= RTV_RPIPE;
}
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
memset(&rtdata, '\0', sizeof rtdata);
Fix the worst sort of bug - a SEGV in a diagnostic ! Also, use sizeof sockaddr_in instead of 16 when writing to the routing socket.
1998-04-16 23:06:53 +00:00
rtdata.sin_len = sizeof rtdata;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
rtdata.sin_family = AF_INET;
rtdata.sin_port = 0;
rtdata.sin_addr = dst;
cp = rtmes.m_space;
Fix the worst sort of bug - a SEGV in a diagnostic ! Also, use sizeof sockaddr_in instead of 16 when writing to the routing socket.
1998-04-16 23:06:53 +00:00
memcpy(cp, &rtdata, rtdata.sin_len);
cp += rtdata.sin_len;
Make gcc-2.8.1 build ppp cleanly. Support OpenBSD again.
1998-04-25 10:49:52 +00:00
if (cmd == RTM_ADD) {
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
if (gateway.s_addr == INADDR_ANY) {
Don't warn when ``add xxxx HISADDR'' fails because HISADDR is 0.0.0.0. This isn't a problem as the sticky route is still added. Suggested by: Chris Timmons <skynyrd@opus.cts.cwu.edu>
1999-01-06 00:08:03 +00:00
if (!ssh)
log_Printf(LogERROR, "bundle_SetRoute: Cannot add a route with"
" destination 0.0.0.0\n");
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
close(s);
return result;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
} else {
rtdata.sin_addr = gateway;
Fix the worst sort of bug - a SEGV in a diagnostic ! Also, use sizeof sockaddr_in instead of 16 when writing to the routing socket.
1998-04-16 23:06:53 +00:00
memcpy(cp, &rtdata, rtdata.sin_len);
cp += rtdata.sin_len;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
rtmes.m_rtm.rtm_addrs |= RTA_GATEWAY;
}
Make gcc-2.8.1 build ppp cleanly. Support OpenBSD again.
1998-04-25 10:49:52 +00:00
}
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
if (dst.s_addr == INADDR_ANY)
mask.s_addr = INADDR_ANY;
if (cmd == RTM_ADD || dst.s_addr == INADDR_ANY) {
rtdata.sin_addr = mask;
Fix the worst sort of bug - a SEGV in a diagnostic ! Also, use sizeof sockaddr_in instead of 16 when writing to the routing socket.
1998-04-16 23:06:53 +00:00
memcpy(cp, &rtdata, rtdata.sin_len);
cp += rtdata.sin_len;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
rtmes.m_rtm.rtm_addrs |= RTA_NETMASK;
}
nb = cp - (char *) &rtmes;
rtmes.m_rtm.rtm_msglen = nb;
wb = ID0write(s, &rtmes, nb);
if (wb < 0) {
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogTCPIP, "bundle_SetRoute failure:\n");
log_Printf(LogTCPIP, "bundle_SetRoute: Cmd = %s\n", cmdstr);
log_Printf(LogTCPIP, "bundle_SetRoute: Dst = %s\n", inet_ntoa(dst));
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
log_Printf(LogTCPIP, "bundle_SetRoute: Gateway = %s\n",
inet_ntoa(gateway));
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogTCPIP, "bundle_SetRoute: Mask = %s\n", inet_ntoa(mask));
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
failed:
if (cmd == RTM_ADD && (rtmes.m_rtm.rtm_errno == EEXIST ||
Make gcc-2.8.1 build ppp cleanly. Support OpenBSD again.
1998-04-25 10:49:52 +00:00
(rtmes.m_rtm.rtm_errno == 0 && errno == EEXIST))) {
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
if (!bang) {
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogWARN, "Add route failed: %s already exists\n",
Add ``enable proxyall'' support. This adds proxy ARP entries for every machine on every class C or smaller subnet that we route to. Add ``set {send,recv}pipe'' for controlling our socket buffer sizes. Mention the IP number with the problem in a few error messages. All submitted by: Craig Leres <leres@ee.lbl.gov> Modified slightly by: me
1998-10-26 19:07:39 +00:00
dst.s_addr == 0 ? "default" : inet_ntoa(dst));
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
result = 0; /* Don't add to our dynamic list */
} else {
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
rtmes.m_rtm.rtm_type = cmd = RTM_CHANGE;
if ((wb = ID0write(s, &rtmes, nb)) < 0)
goto failed;
}
Make gcc-2.8.1 build ppp cleanly. Support OpenBSD again.
1998-04-25 10:49:52 +00:00
} else if (cmd == RTM_DELETE &&
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
(rtmes.m_rtm.rtm_errno == ESRCH ||
(rtmes.m_rtm.rtm_errno == 0 && errno == ESRCH))) {
if (!bang)
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogWARN, "Del route failed: %s: Non-existent\n",
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
inet_ntoa(dst));
Silence ``Network unreachable'' warnings when using ``add .... HISADDR''. The network will never be reachable at this point unless we're in -auto or reading the command from ppp.linkup. We can now run the following lines and get the expected results: set ifaddr 1.2.3.4/0 5.6.7.8/0 add default HISADDR where a route is added immediately in auto mode and the whole thing is delayed 'till the IP numbers have been agreed in other modes. Essentially, ppp.linkup is no longer required.
1998-07-28 21:54:54 +00:00
} else if (rtmes.m_rtm.rtm_errno == 0) {
if (!ssh || errno != ENETUNREACH)
log_Printf(LogWARN, "%s route failed: %s: errno: %s\n", cmdstr,
inet_ntoa(dst), strerror(errno));
} else
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogWARN, "%s route failed: %s: %s\n",
Silence ``Network unreachable'' warnings when using ``add .... HISADDR''. The network will never be reachable at this point unless we're in -auto or reading the command from ppp.linkup. We can now run the following lines and get the expected results: set ifaddr 1.2.3.4/0 5.6.7.8/0 add default HISADDR where a route is added immediately in auto mode and the whole thing is delayed 'till the IP numbers have been agreed in other modes. Essentially, ppp.linkup is no longer required.
1998-07-28 21:54:54 +00:00
cmdstr, inet_ntoa(dst), strerror(rtmes.m_rtm.rtm_errno));
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogDEBUG, "wrote %d: cmd = %s, dst = %x, gateway = %x\n",
o Use __attribute__ (#ifdef __GNUC__) to type-check LogPrintf() and prompt_Printf(), and fix the bits identified as being wrong as a result.
1998-04-19 23:09:03 +00:00
wb, cmdstr, (unsigned)dst.s_addr, (unsigned)gateway.s_addr);
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
close(s);
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
return result;
Create `struct bundle' - the top level control structure. This structure will eventually contain a list of NCPs (currently only IPCP is supported) and a list of physical `struct link's. It will also derive from a struct link itself. Make ModemTimeout() static - it's way to dangerous to be called from outside ! Bump version to 1.9. Our first MP release should be 2.0.
1998-02-02 19:32:16 +00:00
}
Isolate fsm callback functions Derive {lcp,ipcp,ccp}state structures from struct fsm where convenient. Create bundle_Down() and remove LcpDown()
1998-02-06 02:24:29 +00:00
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
void
bundle_LinkClosed(struct bundle *bundle, struct datalink *dl)
{
/*
* Our datalink has closed.
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
* CleanDatalinks() (called from DoLoop()) will remove closed
Make -foreground a proper option (allowing ``allow mode foreground'', ``set mode foreground'' etc.
1999-11-28 15:50:08 +00:00
* BACKGROUND, FOREGROUND and DIRECT links.
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
* If it's the last data link, enter phase DEAD.
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
*
* NOTE: dl may not be in our list (bundle_SendDatalink()) !
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
*/
Move the dial, login and hangup scripts into struct datalink. Don't set these scripts in -direct mode. Always set reconnect_tries to zero for -direct mode.
1998-02-16 19:10:03 +00:00
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
struct datalink *odl;
int other_links;
Move the dial, login and hangup scripts into struct datalink. Don't set these scripts in -direct mode. Always set reconnect_tries to zero for -direct mode.
1998-02-16 19:10:03 +00:00
When entering ``term'' mode, don't output any messages until the device is successfully opened. If we fail to open it, mention the fact. Also go back into command mode as soon as the device is closed rather than waiting for the user to type something before noticing.
1998-08-09 15:34:11 +00:00
log_SetTtyCommandMode(dl);
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
other_links = 0;
for (odl = bundle->links; odl; odl = odl->next)
if (odl != dl && odl->state != DATALINK_CLOSED)
other_links++;
Sort out (fix) the `term' command. datalink_Up() can now be told to skip the dial/login/hangup scripts and can be told whether to enter packet mode when entering the DATALINK_OPENED state.
1998-02-17 01:05:47 +00:00
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
if (!other_links) {
o Make modes consistent throughout ppp. The same strings are used in `set mode', `allow modes', on the command line and when outputting mode names. The strings are matched so that only enough characters to uniquely identify the string are required, so you can now ppp -a mylabel (for auto mode) ppp -b mylabel (for background mode) ppp -dd mylabel (for direct dial mode) etc. o Make -ddial dial when specified on the command line (oops). Pointed out by: Alex <garbanzo@hooked.net>
1998-05-29 18:32:11 +00:00
if (dl->physical->type != PHYS_AUTO) /* Not in -auto mode */
Don't bring the interface down with the last link when we're in -auto mode (broken a few commits ago).
1998-05-19 23:05:10 +00:00
bundle_DownInterface(bundle);
Create & use fsm2initial(), a function to bring a state machine back to ST_INITIAL without going through any unnecessary TLS/TLF pairs.
1998-06-20 00:19:42 +00:00
fsm2initial(&bundle->ncp.ipcp.fsm);
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
bundle_NewPhase(bundle, PHASE_DEAD);
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
bundle_StopIdleTimer(bundle);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
}
o Hook the FSMs into our bundle. o The FSM layering is now more sane. o Move a lot of the NCP stuff into our ipcpstate rather than having it in the bundle, including control of the configured IP addresses. We don't need hacks like the global `linkup' variable any more as the FSM decides when our ppp.link* files get run. This is going to eventually be configurable based on FSM events anyway. o Fix a few inconsistencies when both sides require authentication. o We now have "Ppp..." and "PPp" prompts, reflecting authentication and network phase. We don't print loads of spurious prompts as we change phases any more. o Our phase is part of the bundle now. o Fix a bug where the FSM wasn't calling LayerFinish. o Close the FSM down correctly with a signal rather than slamming it down as if the line was dropped (the undocumented ``down'' command is still available though). o Remove the forgotten `tunno' variable and fix references to it.
1998-02-07 20:50:08 +00:00
}
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
void
Allow an optional ``!'' in the open, dial & call commands. When used, the redial timer is ignored and the modem is opened immediately.
1998-07-29 18:21:17 +00:00
bundle_Open(struct bundle *bundle, const char *name, int mask, int force)
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
{
/*
* Please open the given datalink, or all if name == NULL
*/
struct datalink *dl;
for (dl = bundle->links; dl; dl = dl->next)
if (name == NULL || !strcasecmp(dl->name, name)) {
Allow an optional ``!'' in the open, dial & call commands. When used, the redial timer is ignored and the modem is opened immediately.
1998-07-29 18:21:17 +00:00
if ((mask & dl->physical->type) &&
(dl->state == DATALINK_CLOSED ||
(force && dl->state == DATALINK_OPENING &&
Extend the ``set redial'' command to allow incremental redial timeouts.
1999-03-04 17:42:15 +00:00
dl->dial.timer.state == TIMER_RUNNING))) {
if (force) /* Ignore redial timeout ? */
timer_Stop(&dl->dial.timer);
o Remove the `mode' global - it's now per physical device. o Shuffle things that live at the datalink level into ``show link'' rather than ``show modem''. o Make both ``show'' commands prettier and more consistent, and display carrier status, link type and our name in ``show modem''. o Show redial and reconnect information in ``show link'' and remove ``show redial'' and ``show reconnect''. o Down the correct link in bundle_LinkLost(). o Remove stale -direct and -background links at the end of our main loop, not when we know they're going. This prevents unexpected pointer-invalidations... o If we ``set server'' with the same values twice, notice and don't moan about failure. o Record dial script despite our link mode. The mode may be changed later (next mod) :-) We never run scripts in -direct and -dedicated modes. o Make ``set server none'' functional again. o Correct datalink state array so that we don't report an ``unknown'' state. o Pass struct ipcp to IpcpCleanInterface, not struct fsm. o Create TUN_PREFIX define rather than hard-coding in main.c o prompt_TtyInit now handles a NULL prompt for -direct mode rather than having to create one then destroy it uncleanly. o Mention our mode in the "PPP Started" LogPHASE message. o Bring all auto links up when we have something to send. o Remove some redundant Physical_*() functions. o Show which connection is running a command when logging commands. o Initialise throughput uptime correctly.
1998-04-10 13:19:23 +00:00
datalink_Up(dl, 1, 1);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
if (mask & PHYS_AUTO)
/* Only one AUTO link at a time */
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
break;
}
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
if (name != NULL)
break;
}
}
struct datalink *
bundle2datalink(struct bundle *bundle, const char *name)
{
struct datalink *dl;
if (name != NULL) {
for (dl = bundle->links; dl; dl = dl->next)
if (!strcasecmp(dl->name, name))
return dl;
} else if (bundle->links && !bundle->links->next)
return bundle->links;
return NULL;
}
int
bundle_FillQueues(struct bundle *bundle)
{
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
int total;
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
o Make sure we don't dereference NULL when we've lost all our links. o Use srandomdev() for __FreeBSD__ >= 2, not just >2. o Use srandom((time(NULL)^getpid())+random()), random() when we haven't got srandomdev().
1998-05-06 18:50:12 +00:00
if (bundle->ncp.mp.active)
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
total = mp_FillQueues(bundle);
o Make sure we don't dereference NULL when we've lost all our links. o Use srandomdev() for __FreeBSD__ >= 2, not just >2. o Use srandom((time(NULL)^getpid())+random()), random() when we haven't got srandomdev().
1998-05-06 18:50:12 +00:00
else {
struct datalink *dl;
int add;
for (total = 0, dl = bundle->links; dl; dl = dl->next)
if (dl->state == DATALINK_OPEN) {
add = link_QueueLen(&dl->physical->link);
if (add == 0 && dl->physical->out == NULL)
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
add = ip_PushPacket(&dl->physical->link, bundle);
o Make sure we don't dereference NULL when we've lost all our links. o Use srandomdev() for __FreeBSD__ >= 2, not just >2. o Use srandom((time(NULL)^getpid())+random()), random() when we haven't got srandomdev().
1998-05-06 18:50:12 +00:00
total += add;
}
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
}
Put the IP buffer queues into struct ipcp. Forgotten by: me
1998-08-26 17:39:37 +00:00
return total + ip_QueueLen(&bundle->ncp.ipcp);
Create struct datalink. This is a type of physical link that can chat and talk LCP & CCP. A bundle contains a list of these (only one in the list for the moment). The datalink is a type of descriptor, and dials, enters LCP (& does CCP), kicks the bundle when its FSMs do something interesting and does the hangup chat script on the way down. It also handles redials and reconnects. There are lots of loose ends, and probably lots of bugs, but the data structures are getting there !
1998-02-16 00:01:12 +00:00
}
Introduce the ``link'' command for controlling individual links.
1998-02-17 19:28:01 +00:00
int
bundle_ShowLinks(struct cmdargs const *arg)
{
o Defer setting up pap/chap based IP numbers & labels until after we've determined if we're going to join another ppp invocation. o Make ``show link'' show all link details, and ``show links'' just give a list of links and their current status. o Show our current label in ``show bundle''. o Allow link cloning and removal as soon as our MRRU is set. o Make ``show lcp'' require context as nothing will ever change in our MP LCP (it's auto-configured as per rfc1990). o Initialise our LQM owner in hdlc_Init(). o Store our endpoint discriminator and authentication name at both the datalink and multilink level and compare them when we've finished AUTHENTICATE phase and before entering NETWORK phase. If they don't match, close the link again. Display the information in the appropriate ``show'' command. o Initialise datalink::phone and datalink::fsmp.object properly when we're cloning the link. o Show which link we're passing LQRs on in our diagnostics. o Reject endpoint discriminator REQs at the logical multilink level. o Remove the rest of our CARRIER and LINK logging setup.
1998-04-24 19:16:15 +00:00
struct datalink *dl;
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
struct pppThroughput *t;
int secs;
Introduce the ``link'' command for controlling individual links.
1998-02-17 19:28:01 +00:00
o Display current link throughput in `show links' (assuming throughput measurement is enabled). o Load balance the links based on weight *and* on a round-robin basis. This makes things fairly even on an output basis. We don't try to allow for the peer sending all his data down one link (and try to send ours back up the other). o Show the number of pending input buffers that can't be processed in ``show mp''. o Fix a typo in the man page.
1998-05-03 22:13:14 +00:00
for (dl = arg->bundle->links; dl; dl = dl->next) {
Show the link mode in `show links'.
1998-05-23 22:28:19 +00:00
prompt_Printf(arg->prompt, "Name: %s [%s, %s]",
dl->name, mode2Nam(dl->physical->type), datalink_State(dl));
o Use a tun speed of 115200 for each network link. o Display the link weight in `show links', and only show throughput for open links.
1998-05-05 03:01:32 +00:00
if (dl->physical->link.throughput.rolling && dl->state == DATALINK_OPEN)
Change printf formats %q[du] -> %ll[du]
1999-08-09 22:56:18 +00:00
prompt_Printf(arg->prompt, " bandwidth %d, %llu bps (%llu bytes/sec)",
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
dl->mp.bandwidth ? dl->mp.bandwidth :
physical_GetSpeed(dl->physical),
dl->physical->link.throughput.OctetsPerSecond * 8,
o Display current link throughput in `show links' (assuming throughput measurement is enabled). o Load balance the links based on weight *and* on a round-robin basis. This makes things fairly even on an output basis. We don't try to allow for the peer sending all his data down one link (and try to send ours back up the other). o Show the number of pending input buffers that can't be processed in ``show mp''. o Fix a typo in the man page.
1998-05-03 22:13:14 +00:00
dl->physical->link.throughput.OctetsPerSecond);
prompt_Printf(arg->prompt, "\n");
}
Introduce the ``link'' command for controlling individual links.
1998-02-17 19:28:01 +00:00
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
t = &arg->bundle->ncp.mp.link.throughput;
secs = t->downtime ? 0 : throughput_uptime(t);
if (secs > t->SamplePeriod)
secs = t->SamplePeriod;
if (secs)
Change printf formats %q[du] -> %ll[du]
1999-08-09 22:56:18 +00:00
prompt_Printf(arg->prompt, "Currently averaging %llu bps (%llu bytes/sec)"
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
" over the last %d secs\n", t->OctetsPerSecond * 8,
t->OctetsPerSecond, secs);
Introduce the ``link'' command for controlling individual links.
1998-02-17 19:28:01 +00:00
return 0;
}
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
static const char *
optval(struct bundle *bundle, int bit)
{
return (bundle->cfg.opt & bit) ? "enabled" : "disabled";
}
Remove ``show timeout'' and ``show auth'' and add ``show bundle''. Mention ``show timer'' in the man page.
1998-04-11 21:50:47 +00:00
int
bundle_ShowStatus(struct cmdargs const *arg)
{
int remaining;
prompt_Printf(arg->prompt, "Phase %s\n", bundle_PhaseName(arg->bundle));
Categorize the fields in struct bundle, and make the tunnel device speed available in ``show bundle''.
1998-05-06 23:50:25 +00:00
prompt_Printf(arg->prompt, " Device: %s\n", arg->bundle->dev.Name);
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
prompt_Printf(arg->prompt, " Interface: %s @ %lubps",
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
arg->bundle->iface->name, arg->bundle->bandwidth);
Remove ``show timeout'' and ``show auth'' and add ``show bundle''. Mention ``show timer'' in the man page.
1998-04-11 21:50:47 +00:00
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
if (arg->bundle->upat) {
int secs = time(NULL) - arg->bundle->upat;
prompt_Printf(arg->prompt, ", up time %d:%02d:%02d", secs / 3600,
(secs / 60) % 60, secs % 60);
}
prompt_Printf(arg->prompt, "\n\nDefaults:\n");
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
prompt_Printf(arg->prompt, " Label: %s\n", arg->bundle->cfg.label);
prompt_Printf(arg->prompt, " Auth name: %s\n",
arg->bundle->cfg.auth.name);
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
If we've got a full output buffer queue and cannot send anything for two mintues (see ``set choked'' and ``show bundle''), nuke the ip, mp and link level buffer queues. This should fix problems where ``ppp -auto'' seems to stop responding after failing to connect to the peer a few times.
1998-08-25 17:48:43 +00:00
prompt_Printf(arg->prompt, " Choked Timer: %ds\n",
arg->bundle->cfg.choked.timeout);
Initial RADIUS support (using libradius). See the man page for details. Compiling with -DNORADIUS (the default for `release') removes support. TODO: The functionality in libradius::rad_send_request() needs to be supplied as a set of routines so that ppp doesn't have to wait indefinitely for the radius server(s). Instead, we need to get a descriptor back, select() on the descriptor, and ask libradius to service it when necessary. For now, ppp blocks SIGALRM while in rad_send_request(), so it misses PAP/CHAP retries & timeouts if they occur. Only PAP is functional. When CHAP is attempted, libradius complains that no User-Password has been specified... rfc2138 says that it *mustn't* be used for CHAP :-( Sponsored by: Internet Business Solutions Ltd., Switzerland
1999-01-28 01:56:34 +00:00
#ifndef NORADIUS
radius_Show(&arg->bundle->radius, arg->prompt);
#endif
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
prompt_Printf(arg->prompt, " Idle Timer: ");
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
if (arg->bundle->cfg.idle.timeout) {
prompt_Printf(arg->prompt, "%ds", arg->bundle->cfg.idle.timeout);
if (arg->bundle->cfg.idle.min_timeout)
prompt_Printf(arg->prompt, ", min %ds",
arg->bundle->cfg.idle.min_timeout);
Remove ``show timeout'' and ``show auth'' and add ``show bundle''. Mention ``show timer'' in the man page.
1998-04-11 21:50:47 +00:00
remaining = bundle_RemainingIdleTime(arg->bundle);
if (remaining != -1)
prompt_Printf(arg->prompt, " (%ds remaining)", remaining);
prompt_Printf(arg->prompt, "\n");
} else
prompt_Printf(arg->prompt, "disabled\n");
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
prompt_Printf(arg->prompt, " MTU: ");
o Add ipcp_SetLink() for attaching IPCP to the correct link, and call it after link authentication. o Pretty print our bundle MTU. o Correct MP header encoding and decoding (should be network byte order). o Add some debug diagnostics so that we can see MP fragment sending and re-assembly.
1998-04-23 18:56:21 +00:00
if (arg->bundle->cfg.mtu)
prompt_Printf(arg->prompt, "%d\n", arg->bundle->cfg.mtu);
else
prompt_Printf(arg->prompt, "unspecified\n");
Remove ``show timeout'' and ``show auth'' and add ``show bundle''. Mention ``show timer'' in the man page.
1998-04-11 21:50:47 +00:00
Add ``enable proxyall'' support. This adds proxy ARP entries for every machine on every class C or smaller subnet that we route to. Add ``set {send,recv}pipe'' for controlling our socket buffer sizes. Mention the IP number with the problem in a few error messages. All submitted by: Craig Leres <leres@ee.lbl.gov> Modified slightly by: me
1998-10-26 19:07:39 +00:00
prompt_Printf(arg->prompt, " sendpipe: ");
if (arg->bundle->ncp.ipcp.cfg.sendpipe > 0)
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, "%-20ld", arg->bundle->ncp.ipcp.cfg.sendpipe);
Add ``enable proxyall'' support. This adds proxy ARP entries for every machine on every class C or smaller subnet that we route to. Add ``set {send,recv}pipe'' for controlling our socket buffer sizes. Mention the IP number with the problem in a few error messages. All submitted by: Craig Leres <leres@ee.lbl.gov> Modified slightly by: me
1998-10-26 19:07:39 +00:00
else
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, "unspecified ");
Add ``enable proxyall'' support. This adds proxy ARP entries for every machine on every class C or smaller subnet that we route to. Add ``set {send,recv}pipe'' for controlling our socket buffer sizes. Mention the IP number with the problem in a few error messages. All submitted by: Craig Leres <leres@ee.lbl.gov> Modified slightly by: me
1998-10-26 19:07:39 +00:00
prompt_Printf(arg->prompt, " recvpipe: ");
if (arg->bundle->ncp.ipcp.cfg.recvpipe > 0)
prompt_Printf(arg->prompt, "%ld\n", arg->bundle->ncp.ipcp.cfg.recvpipe);
else
prompt_Printf(arg->prompt, "unspecified\n");
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, " Sticky Routes: %-20.20s",
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
optval(arg->bundle, OPT_SROUTES));
prompt_Printf(arg->prompt, " ID check: %s\n",
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
optval(arg->bundle, OPT_IDCHECK));
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, " Keep-Session: %-20.20s",
Introduce the ``keep-session'' option. Refer to the man page for details. This allows MP over non-tty devices where the original ppp process must not exit (such as sshd-spawned ppp sessions).
1999-06-02 00:46:55 +00:00
optval(arg->bundle, OPT_KEEPSESSION));
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
prompt_Printf(arg->prompt, " Loopback: %s\n",
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
optval(arg->bundle, OPT_LOOPBACK));
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, " PasswdAuth: %-20.20s",
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
optval(arg->bundle, OPT_PASSWDAUTH));
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
prompt_Printf(arg->prompt, " Proxy: %s\n",
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
optval(arg->bundle, OPT_PROXY));
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, " Proxyall: %-20.20s",
Add ``enable proxyall'' support. This adds proxy ARP entries for every machine on every class C or smaller subnet that we route to. Add ``set {send,recv}pipe'' for controlling our socket buffer sizes. Mention the IP number with the problem in a few error messages. All submitted by: Craig Leres <leres@ee.lbl.gov> Modified slightly by: me
1998-10-26 19:07:39 +00:00
optval(arg->bundle, OPT_PROXYALL));
Create `struct sticky_route'. Any `add' or `delete' command that uses MYADDR or HISADDR will be added to the sticky route list (show ipcp). When MYADDR or HISADDR change due to IPCP negotiations, and if `sroutes' is enabled (the default), all sticky route entries are updated in the routing table. The end result is that `add default hisaddr' will ``stick'', as will ``add myaddr 255.255.255.255 127.0.0.1'' and ``add 1.2.3.4 255.255.255.0 hisaddr''.
1998-05-05 23:30:13 +00:00
prompt_Printf(arg->prompt, " Throughput: %s\n",
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
optval(arg->bundle, OPT_THROUGHPUT));
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
prompt_Printf(arg->prompt, " Utmp Logging: %-20.20s",
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
optval(arg->bundle, OPT_UTMP));
Solve the ``first connection'' problem that occurs on demand-dial links with dynamic IP numbers where the program that causes the dial bind()s to an interface address that is subsequently changed after ppp negotiation. The problem is defeated by adding negotiated addresses to the tun interface as additional alias addresses and providing a set of ``iface'' commands for managing the interface. Libalias is also required (and what a name clash!) - it happily IP-aliases the address so that the source is that of the primary (negotiated) interface and un-IP-aliases it on the way back. An ``enable iface-alias'' is done implicitly by the -alias command line switch. If -alias isn't given, iface-aliasing is disabled by default and can't be enabled 'till an ``alias enable yes'' is done. ``alias enable no'' silently disables iface-alias. So, for dynamic-IP-type-connections, running ``ppp -alias -auto blah'' will work for the first connection, although existing bindings will not survive a disconnect/connect as the TCP peer will be trying to send to the old IP address - the packets won't route. It's now a lot easier to add IPXCP to ppp with minor updates to the new iface.[ch] (if anyone ever gets 'round to it). It's also now possible to manually add interface aliases with something like ``iface add 1.2.3.4/24 5.6.7.8''. This allows multi-homed ppp links :-)
1998-10-22 02:32:50 +00:00
prompt_Printf(arg->prompt, " Iface-Alias: %s\n",
optval(arg->bundle, OPT_IFACEALIAS));
o Move the accept/deny/disable/enable globals either to the bundle (non-negotiated vars) or to their respective IPCP, LCP or CCP. o Enable rolling throughput statistics by default. o Remove the `display' command. These values now appear in `show bundle', `show ipcp', `show ccp' and `show lcp'. o Initialise auth name & key at bundle create time (oops). o Rename pppd-deflate (the id-24 hack) to deflate24. o Don't send both a REJ and a NAK to an IPCP or LCP REQ. Favour the REJ (already done at the CCP level). o Recurse in datalink_UpdateSet() when we change state, otherwise we end up setting no descriptors and getting jammed in the imminent select() instead of doing the dial/login/hangup. o Display our CHAP encryption method despite being built with DES. o Display VJ as not negotiated in ``show ipcp'' when necessary.
1998-04-16 00:26:21 +00:00
Remove ``show timeout'' and ``show auth'' and add ``show bundle''. Mention ``show timer'' in the man page.
1998-04-11 21:50:47 +00:00
return 0;
}
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
static void
bundle_IdleTimeout(void *v)
{
struct bundle *bundle = (struct bundle *)v;
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogPHASE, "Idle timer expired.\n");
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
bundle_StopIdleTimer(bundle);
Make `close lcp' just close the LCP layer and not hangup. This is useful for slirp users that wish to get their shell back after the ppp session. `close' with no args still hangs up as expected. Required by: jmz
1998-06-15 19:05:27 +00:00
bundle_Close(bundle, NULL, CLOSE_STAYDOWN);
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
}
/*
* Start Idle timer. If timeout is reached, we call bundle_Close() to
* close LCP and link.
*/
void
bundle_StartIdleTimer(struct bundle *bundle)
{
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
timer_Stop(&bundle->idle.timer);
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
if ((bundle->phys_type.open & (PHYS_DEDICATED|PHYS_DDIAL)) !=
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle->phys_type.open && bundle->cfg.idle.timeout) {
int secs;
secs = bundle->cfg.idle.timeout;
if (bundle->cfg.idle.min_timeout > secs && bundle->upat) {
int up = time(NULL) - bundle->upat;
if ((long long)bundle->cfg.idle.min_timeout - up > (long long)secs)
secs = bundle->cfg.idle.min_timeout - up;
}
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
bundle->idle.timer.func = bundle_IdleTimeout;
bundle->idle.timer.name = "idle";
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle->idle.timer.load = secs * SECTICKS;
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
bundle->idle.timer.arg = bundle;
timer_Start(&bundle->idle.timer);
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle->idle.done = time(NULL) + secs;
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
}
}
void
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle_SetIdleTimer(struct bundle *bundle, int timeout, int min_timeout)
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
{
Implement a minimum idle time value as an optional second argument to ``set timeout''. This is useful for situations where your minimum call charge is (say) 5 minutes (like mine is)
1999-08-17 17:22:46 +00:00
bundle->cfg.idle.timeout = timeout;
if (min_timeout >= 0)
bundle->cfg.idle.min_timeout = min_timeout;
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
if (bundle_LinkIsUp(bundle))
bundle_StartIdleTimer(bundle);
}
void
bundle_StopIdleTimer(struct bundle *bundle)
{
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
timer_Stop(&bundle->idle.timer);
Initialise bundle::idle::done when the timer's stopped.
1998-04-07 01:49:24 +00:00
bundle->idle.done = 0;
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
}
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
static int
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
bundle_RemainingIdleTime(struct bundle *bundle)
{
Display the correct amount of remaining time in `show time'.
1998-03-18 23:15:31 +00:00
if (bundle->idle.done)
return bundle->idle.done - time(NULL);
return -1;
Move the idle timer into struct bundle. There's no link-level idle timer for the moment.
1998-03-09 19:26:41 +00:00
}
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
int
bundle_IsDead(struct bundle *bundle)
{
return !bundle->links || (bundle->phase == PHASE_DEAD && bundle->CleaningUp);
}
Deglobalise `struct prompt': o Our diagnostic socket has its password set in the `set socket' line only (not in ppp.secret). o Passwords are per server socket (*VarAuthKey are gone) o Authority is per prompt (VarLocalAuth is gone). o Local logging is per prompt. o Add a `show who' command to see who's connected. No identd routine - just a `where the connection came from' display. o SIGUSR1 is disabled for now - we have no way of choosing a password for the socket created :-( Prompts are attached as a list of `struct descriptor's in struct bundle, and serviced under the bundles descriptor service routines. Ultimately, everything should be done like this. Cosmetic: o alphabeticalise SRCS in Makefile. o Add a few comments in command.h TODO: Start checking that we don't overflow the descriptor sets in select() now that we can have any number of descriptors.
1998-04-03 19:26:02 +00:00
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
static struct datalink *
bundle_DatalinkLinkout(struct bundle *bundle, struct datalink *dl)
Add the `clone' and `remove' commands for creating and destroying links.
1998-04-05 22:48:25 +00:00
{
struct datalink **dlp;
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
for (dlp = &bundle->links; *dlp; dlp = &(*dlp)->next)
if (*dlp == dl) {
*dlp = dl->next;
dl->next = NULL;
bundle_LinksRemoved(bundle);
return dl;
}
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
return NULL;
}
static void
bundle_DatalinkLinkin(struct bundle *bundle, struct datalink *dl)
{
struct datalink **dlp = &bundle->links;
while (*dlp)
dlp = &(*dlp)->next;
*dlp = dl;
dl->next = NULL;
bundle_LinkAdded(bundle, dl);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
mp_CheckAutoloadTimer(&bundle->ncp.mp);
o Remove the `mode' global - it's now per physical device. o Shuffle things that live at the datalink level into ``show link'' rather than ``show modem''. o Make both ``show'' commands prettier and more consistent, and display carrier status, link type and our name in ``show modem''. o Show redial and reconnect information in ``show link'' and remove ``show redial'' and ``show reconnect''. o Down the correct link in bundle_LinkLost(). o Remove stale -direct and -background links at the end of our main loop, not when we know they're going. This prevents unexpected pointer-invalidations... o If we ``set server'' with the same values twice, notice and don't moan about failure. o Record dial script despite our link mode. The mode may be changed later (next mod) :-) We never run scripts in -direct and -dedicated modes. o Make ``set server none'' functional again. o Correct datalink state array so that we don't report an ``unknown'' state. o Pass struct ipcp to IpcpCleanInterface, not struct fsm. o Create TUN_PREFIX define rather than hard-coding in main.c o prompt_TtyInit now handles a NULL prompt for -direct mode rather than having to create one then destroy it uncleanly. o Mention our mode in the "PPP Started" LogPHASE message. o Bring all auto links up when we have something to send. o Remove some redundant Physical_*() functions. o Show which connection is running a command when logging commands. o Initialise throughput uptime correctly.
1998-04-10 13:19:23 +00:00
}
void
bundle_CleanDatalinks(struct bundle *bundle)
{
struct datalink **dlp = &bundle->links;
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
int found = 0;
o Remove the `mode' global - it's now per physical device. o Shuffle things that live at the datalink level into ``show link'' rather than ``show modem''. o Make both ``show'' commands prettier and more consistent, and display carrier status, link type and our name in ``show modem''. o Show redial and reconnect information in ``show link'' and remove ``show redial'' and ``show reconnect''. o Down the correct link in bundle_LinkLost(). o Remove stale -direct and -background links at the end of our main loop, not when we know they're going. This prevents unexpected pointer-invalidations... o If we ``set server'' with the same values twice, notice and don't moan about failure. o Record dial script despite our link mode. The mode may be changed later (next mod) :-) We never run scripts in -direct and -dedicated modes. o Make ``set server none'' functional again. o Correct datalink state array so that we don't report an ``unknown'' state. o Pass struct ipcp to IpcpCleanInterface, not struct fsm. o Create TUN_PREFIX define rather than hard-coding in main.c o prompt_TtyInit now handles a NULL prompt for -direct mode rather than having to create one then destroy it uncleanly. o Mention our mode in the "PPP Started" LogPHASE message. o Bring all auto links up when we have something to send. o Remove some redundant Physical_*() functions. o Show which connection is running a command when logging commands. o Initialise throughput uptime correctly.
1998-04-10 13:19:23 +00:00
while (*dlp)
if ((*dlp)->state == DATALINK_CLOSED &&
Make -foreground a proper option (allowing ``allow mode foreground'', ``set mode foreground'' etc.
1999-11-28 15:50:08 +00:00
(*dlp)->physical->type &
(PHYS_DIRECT|PHYS_BACKGROUND|PHYS_FOREGROUND)) {
o Remove the `mode' global - it's now per physical device. o Shuffle things that live at the datalink level into ``show link'' rather than ``show modem''. o Make both ``show'' commands prettier and more consistent, and display carrier status, link type and our name in ``show modem''. o Show redial and reconnect information in ``show link'' and remove ``show redial'' and ``show reconnect''. o Down the correct link in bundle_LinkLost(). o Remove stale -direct and -background links at the end of our main loop, not when we know they're going. This prevents unexpected pointer-invalidations... o If we ``set server'' with the same values twice, notice and don't moan about failure. o Record dial script despite our link mode. The mode may be changed later (next mod) :-) We never run scripts in -direct and -dedicated modes. o Make ``set server none'' functional again. o Correct datalink state array so that we don't report an ``unknown'' state. o Pass struct ipcp to IpcpCleanInterface, not struct fsm. o Create TUN_PREFIX define rather than hard-coding in main.c o prompt_TtyInit now handles a NULL prompt for -direct mode rather than having to create one then destroy it uncleanly. o Mention our mode in the "PPP Started" LogPHASE message. o Bring all auto links up when we have something to send. o Remove some redundant Physical_*() functions. o Show which connection is running a command when logging commands. o Initialise throughput uptime correctly.
1998-04-10 13:19:23 +00:00
*dlp = datalink_Destroy(*dlp);
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
found++;
} else
o Remove the `mode' global - it's now per physical device. o Shuffle things that live at the datalink level into ``show link'' rather than ``show modem''. o Make both ``show'' commands prettier and more consistent, and display carrier status, link type and our name in ``show modem''. o Show redial and reconnect information in ``show link'' and remove ``show redial'' and ``show reconnect''. o Down the correct link in bundle_LinkLost(). o Remove stale -direct and -background links at the end of our main loop, not when we know they're going. This prevents unexpected pointer-invalidations... o If we ``set server'' with the same values twice, notice and don't moan about failure. o Record dial script despite our link mode. The mode may be changed later (next mod) :-) We never run scripts in -direct and -dedicated modes. o Make ``set server none'' functional again. o Correct datalink state array so that we don't report an ``unknown'' state. o Pass struct ipcp to IpcpCleanInterface, not struct fsm. o Create TUN_PREFIX define rather than hard-coding in main.c o prompt_TtyInit now handles a NULL prompt for -direct mode rather than having to create one then destroy it uncleanly. o Mention our mode in the "PPP Started" LogPHASE message. o Bring all auto links up when we have something to send. o Remove some redundant Physical_*() functions. o Show which connection is running a command when logging commands. o Initialise throughput uptime correctly.
1998-04-10 13:19:23 +00:00
dlp = &(*dlp)->next;
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
if (found)
bundle_LinksRemoved(bundle);
}
int
bundle_DatalinkClone(struct bundle *bundle, struct datalink *dl,
const char *name)
{
if (bundle2datalink(bundle, name)) {
log_Printf(LogWARN, "Clone: %s: name already exists\n", name);
return 0;
}
bundle_DatalinkLinkin(bundle, datalink_Clone(dl, name));
return 1;
}
void
bundle_DatalinkRemove(struct bundle *bundle, struct datalink *dl)
{
dl = bundle_DatalinkLinkout(bundle, dl);
if (dl)
datalink_Destroy(dl);
Add the `clone' and `remove' commands for creating and destroying links.
1998-04-05 22:48:25 +00:00
}
Negotiate MRRU, SHORTSEQ and ENDDISC. ENDDISC doesn't imply multilink ('cos I've seen my ISP REQ it without multilink). Setting MRRU is ifdef'd out until it's debugged and we can merge -direct links with other running programs. Fix MTU setting.
1998-04-23 03:23:03 +00:00
void
bundle_SetLabel(struct bundle *bundle, const char *label)
{
if (label)
strncpy(bundle->cfg.label, label, sizeof bundle->cfg.label - 1);
else
*bundle->cfg.label = '\0';
}
const char *
bundle_GetLabel(struct bundle *bundle)
{
return *bundle->cfg.label ? bundle->cfg.label : NULL;
}
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
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
bundle_LinkSize()
{
struct iovec iov[SCATTER_SEGMENTS];
int niov, expect, f;
iov[0].iov_len = strlen(Version) + 1;
iov[0].iov_base = NULL;
niov = 1;
if (datalink2iov(NULL, iov, &niov, SCATTER_SEGMENTS, NULL, NULL) == -1) {
log_Printf(LogERROR, "Cannot determine space required for link\n");
return 0;
}
for (f = expect = 0; f < niov; f++)
expect += iov[f].iov_len;
return expect;
}
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
void
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
bundle_ReceiveDatalink(struct bundle *bundle, int s)
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
{
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
char cmsgbuf[sizeof(struct cmsghdr) + sizeof(int) * SEND_MAXFD];
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
int niov, expect, f, *fd, nfd, onfd, got;
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
struct iovec iov[SCATTER_SEGMENTS];
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
struct cmsghdr *cmsg;
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
struct msghdr msg;
o Rename datalinks as soon as the name has been received so that LQM and HDLC timer diagnostics come out with the correct name. o Don't send an LQR immediately upon reviving a datalink. Leave it 'till the next timeout. o Add the link name to some more LQR diagnostics. o Break out of the main loop when a descriptor exception is seen in select(). o Remove the evil nointr_[u]sleep() functions. Timers should be (and are) used instead. o Treat a read() of 0 bytes as an error that's fatal to the link on which the read() is done. We should never read() 0 after select() says there's something there - not unless the link has been closed by the other side. o Write the data seen before a HDLC header to the terminal in `term' mode, *not* back to the modem :-/ o Initialise our transmitted file descriptor before starting any timers. o Only send data links that have *no* pending output data. This means that our final ACK will be written rather than being nuked with the datalink transmission.
1998-05-08 01:15:19 +00:00
struct datalink *dl;
o Don't forget to close our transfer socket if we cannot generate the data to transfer. o Transfer uucp lock ownership for the transferred device. o Don't assume we know the correct values of dev_is_modem and mbits after the transfer.
1998-05-28 23:15:40 +00:00
pid_t pid;
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogPHASE, "Receiving datalink\n");
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
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
/*
* Create our scatter/gather array - passing NULL gets the space
* allocation requirement rather than actually flattening the
* structures.
*/
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
iov[0].iov_len = strlen(Version) + 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[0].iov_base = NULL;
niov = 1;
if (datalink2iov(NULL, iov, &niov, SCATTER_SEGMENTS, NULL, NULL) == -1) {
log_Printf(LogERROR, "Cannot determine space required for link\n");
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
return;
o Don't try to transfer tty device descriptors as there's no way of transferring session rights with them. Instead, create two `/bin/cat' processes. A new child is spawned and disassociated from the terminal and the parent, which continues with the rest of the ppp process. Meanwhile, the parent spawns another child, and both the parent and child exec the `/bin/cat' commands with the appropriate descriptors. This way, the session is owned by the parent, and the tty is held open. o Close LCPs that have done a TLF and are now in ST_STOPPED before calling Down. This prevents them from trying to come back up again after the peer has shut them down (it seems a bit strange that the rfc says that a Down in ST_STOPPED will cause a TLS etc). o Don't try to set the physical link name pointer when we're receiving and renaming a datalink. The physical hasn't been created yet, and as it happens, the garbage physical pointer happens to be the value of another physical - so we're pointing that other physical name at ourselves. yeuck. o Re-arrange the order of things in main (DoLoop()). We now handle signals only after the select and not before the UpdateSet. It's possible that either a signal (FSM timeout) or a descriptor_Read() brings a link down, after which we'd better tidy up any dead direct and 1off descriptors before calling UpdateSet() again. o Mention when we detect a PPP packet when we see one before the link is up (then start LCP as before).
1998-05-25 02:22:38 +00:00
}
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
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
/* Allocate the scatter/gather array for recvmsg() */
for (f = expect = 0; f < niov; f++) {
if ((iov[f].iov_base = malloc(iov[f].iov_len)) == NULL) {
log_Printf(LogERROR, "Cannot allocate space to receive link\n");
return;
}
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
if (f)
expect += iov[f].iov_len;
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
}
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
/* Set up our message */
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
cmsg = (struct cmsghdr *)cmsgbuf;
cmsg->cmsg_len = sizeof cmsgbuf;
cmsg->cmsg_level = SOL_SOCKET;
cmsg->cmsg_type = 0;
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
memset(&msg, '\0', sizeof msg);
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
msg.msg_name = NULL;
msg.msg_namelen = 0;
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
msg.msg_iov = iov;
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
msg.msg_iovlen = 1; /* Only send the version at the first pass */
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
msg.msg_control = cmsgbuf;
msg.msg_controllen = sizeof cmsgbuf;
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
log_Printf(LogDEBUG, "Expecting %d scatter/gather bytes\n", iov[0].iov_len);
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
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
if ((got = recvmsg(s, &msg, MSG_WAITALL)) != iov[0].iov_len) {
if (got == -1)
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
log_Printf(LogERROR, "Failed recvmsg: %s\n", strerror(errno));
else
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
log_Printf(LogERROR, "Failed recvmsg: Got %d, not %d\n",
got, iov[0].iov_len);
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
while (niov--)
free(iov[niov].iov_base);
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
return;
}
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
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 (cmsg->cmsg_level != SOL_SOCKET || cmsg->cmsg_type != SCM_RIGHTS) {
log_Printf(LogERROR, "Recvmsg: no descriptors received !\n");
while (niov--)
free(iov[niov].iov_base);
return;
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
}
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
fd = (int *)(cmsg + 1);
nfd = (cmsg->cmsg_len - sizeof *cmsg) / sizeof(int);
if (nfd < 2) {
log_Printf(LogERROR, "Recvmsg: %d descriptor%s received (too few) !\n",
nfd, nfd == 1 ? "" : "s");
while (nfd--)
close(fd[nfd]);
Make sure our recvmsg() is a SCM_RIGHTS message when we receive a link from another ppp.
1998-06-18 23:38:04 +00:00
while (niov--)
free(iov[niov].iov_base);
return;
o Don't try to transfer tty device descriptors as there's no way of transferring session rights with them. Instead, create two `/bin/cat' processes. A new child is spawned and disassociated from the terminal and the parent, which continues with the rest of the ppp process. Meanwhile, the parent spawns another child, and both the parent and child exec the `/bin/cat' commands with the appropriate descriptors. This way, the session is owned by the parent, and the tty is held open. o Close LCPs that have done a TLF and are now in ST_STOPPED before calling Down. This prevents them from trying to come back up again after the peer has shut them down (it seems a bit strange that the rfc says that a Down in ST_STOPPED will cause a TLS etc). o Don't try to set the physical link name pointer when we're receiving and renaming a datalink. The physical hasn't been created yet, and as it happens, the garbage physical pointer happens to be the value of another physical - so we're pointing that other physical name at ourselves. yeuck. o Re-arrange the order of things in main (DoLoop()). We now handle signals only after the select and not before the UpdateSet. It's possible that either a signal (FSM timeout) or a descriptor_Read() brings a link down, after which we'd better tidy up any dead direct and 1off descriptors before calling UpdateSet() again. o Mention when we detect a PPP packet when we see one before the link is up (then start LCP as before).
1998-05-25 02:22:38 +00:00
}
When transmitting link file descriptors in multilink mode, ACK our received file descriptor, and don't close a sent descriptor 'till the ack has been received.
1998-05-09 14:44:11 +00:00
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
/*
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
* We've successfully received two or more open file descriptors
* through our socket, plus a version string. Make sure it's the
* correct version, and drop the connection if it's not.
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
*/
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
if (strncmp(Version, iov[0].iov_base, iov[0].iov_len)) {
log_Printf(LogWARN, "Cannot receive datalink, incorrect version"
" (\"%.*s\", not \"%s\")\n", (int)iov[0].iov_len,
Cast void * to char * for log_Printf().
1998-06-27 12:03:46 +00:00
(char *)iov[0].iov_base, Version);
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
while (nfd--)
close(fd[nfd]);
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
while (niov--)
free(iov[niov].iov_base);
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
return;
}
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
/*
* Everything looks good. Send the other side our process id so that
* they can transfer lock ownership, and wait for them to send the
* actual link data.
*/
pid = getpid();
if ((got = write(fd[1], &pid, sizeof pid)) != sizeof pid) {
if (got == -1)
log_Printf(LogERROR, "Failed write: %s\n", strerror(errno));
else
log_Printf(LogERROR, "Failed write: Got %d, not %d\n", got,
(int)(sizeof pid));
while (nfd--)
close(fd[nfd]);
while (niov--)
free(iov[niov].iov_base);
return;
}
if ((got = readv(fd[1], iov + 1, niov - 1)) != expect) {
if (got == -1)
log_Printf(LogERROR, "Failed write: %s\n", strerror(errno));
else
log_Printf(LogERROR, "Failed write: Got %d, not %d\n", got, expect);
while (nfd--)
close(fd[nfd]);
while (niov--)
free(iov[niov].iov_base);
return;
}
close(fd[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
onfd = nfd; /* We've got this many in our array */
nfd -= 2; /* Don't include p->fd and our reply descriptor */
niov = 1; /* Skip the version id */
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
dl = iov2datalink(bundle, iov, &niov, sizeof iov / sizeof *iov, 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
fd + 2, &nfd);
o Rename datalinks as soon as the name has been received so that LQM and HDLC timer diagnostics come out with the correct name. o Don't send an LQR immediately upon reviving a datalink. Leave it 'till the next timeout. o Add the link name to some more LQR diagnostics. o Break out of the main loop when a descriptor exception is seen in select(). o Remove the evil nointr_[u]sleep() functions. Timers should be (and are) used instead. o Treat a read() of 0 bytes as an error that's fatal to the link on which the read() is done. We should never read() 0 after select() says there's something there - not unless the link has been closed by the other side. o Write the data seen before a HDLC header to the terminal in `term' mode, *not* back to the modem :-/ o Initialise our transmitted file descriptor before starting any timers. o Only send data links that have *no* pending output data. This means that our final ACK will be written rather than being nuked with the datalink transmission.
1998-05-08 01:15:19 +00:00
if (dl) {
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
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
if (nfd) {
log_Printf(LogERROR, "bundle_ReceiveDatalink: Failed to handle %d "
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
"auxiliary file descriptors (%d remain)\n", onfd, nfd);
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
datalink_Destroy(dl);
while (nfd--)
close(fd[onfd--]);
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
close(fd[0]);
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
} else {
bundle_DatalinkLinkin(bundle, dl);
datalink_AuthOk(dl);
bundle_CalculateBandwidth(dl->bundle);
}
} else {
while (nfd--)
close(fd[onfd--]);
close(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
close(fd[1]);
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
}
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
free(iov[0].iov_base);
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
}
void
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
bundle_SendDatalink(struct datalink *dl, int s, struct sockaddr_un *sun)
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
{
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
char cmsgbuf[sizeof(struct cmsghdr) + sizeof(int) * SEND_MAXFD];
const char *constlock;
char *lock;
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
struct cmsghdr *cmsg;
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
struct msghdr msg;
struct iovec iov[SCATTER_SEGMENTS];
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 niov, f, expect, newsid, fd[SEND_MAXFD], nfd, reply[2], got;
o Don't forget to close our transfer socket if we cannot generate the data to transfer. o Transfer uucp lock ownership for the transferred device. o Don't assume we know the correct values of dev_is_modem and mbits after the transfer.
1998-05-28 23:15:40 +00:00
pid_t newpid;
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
Cosmetic: Make our external function names consistent.
1998-05-01 19:26:12 +00:00
log_Printf(LogPHASE, "Transmitting datalink %s\n", dl->name);
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
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
/* Record the base device name for a lock transfer later */
constlock = physical_LockedDevice(dl->physical);
if (constlock) {
lock = alloca(strlen(constlock) + 1);
strcpy(lock, constlock);
} else
lock = NULL;
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
bundle_LinkClosed(dl->bundle, dl);
o Move our prompt descriptor list outside of the bundle. It's now dealt with by the `server' object. This simplifies things as we only have one list of prompt descriptors and the log_ routines check prompt::logactive to determine whether it should be used for output. o Include the MP socket UpdateSet() result in bundle::UpdateSet(). o Don't select on the tun device unless we're in NETWORK phase or AUTO mode. o Stop the idle timer when we go to DEAD phase. We may have transferred a link and not had a chance to kill it. o Don't fail when trying to unlink our transferred datalink from our descriptor lists just before the transfer. o Add our link descriptor to the write set if we got a short write the last time (physical::out is set). o Log the connection source address when a connection is closed. o Remove descriptor::next field. Descriptor lists are not required any more.
1998-05-23 22:24:50 +00:00
bundle_DatalinkLinkout(dl->bundle, dl);
o Call bundle_LinkClosed() when transferring a datalink so that the bundle has the opportunity to go PHASE_DEAD and cleanup the interface (if it's the last link). o Regnerate our phys_type value when we transfer the link. o Always clean up the interface when destroying our bundle in case we're abending. o Always clean up our interface when the last link is gone rather than delaying things 'till exit time in the -direct case (the interface is useless anyway). Do this *after* slamming down our NCPs (if they're still around). o Our MP server descriptor now clears the relevent device descriptor from our descriptor [fd]sets when a datalink is on death-row (to be transferred to another running ppp), thus avoiding the possibility of passing a bum descriptor to select() and having ppp abend. o Handle the MP socket descriptor functions from within the bundle descriptor functions. Now we ensure that the MP socket descriptor functions see the descriptor sets *after* they've been seen by our datalinks. o Add/fix a few more comments.
1998-05-15 18:21:45 +00:00
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
/* Build our scatter/gather array */
iov[0].iov_len = strlen(Version) + 1;
iov[0].iov_base = strdup(Version);
niov = 1;
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
nfd = 0;
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
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
fd[0] = datalink2iov(dl, iov, &niov, SCATTER_SEGMENTS, fd + 2, &nfd);
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
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 (fd[0] != -1 && socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, reply) != -1) {
/*
* fd[1] is used to get the peer process id back, then to confirm that
* we've transferred any device locks to that process id.
*/
fd[1] = reply[1];
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
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
nfd += 2; /* Include fd[0] and fd[1] */
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
memset(&msg, '\0', sizeof msg);
o Don't try to transfer tty device descriptors as there's no way of transferring session rights with them. Instead, create two `/bin/cat' processes. A new child is spawned and disassociated from the terminal and the parent, which continues with the rest of the ppp process. Meanwhile, the parent spawns another child, and both the parent and child exec the `/bin/cat' commands with the appropriate descriptors. This way, the session is owned by the parent, and the tty is held open. o Close LCPs that have done a TLF and are now in ST_STOPPED before calling Down. This prevents them from trying to come back up again after the peer has shut them down (it seems a bit strange that the rfc says that a Down in ST_STOPPED will cause a TLS etc). o Don't try to set the physical link name pointer when we're receiving and renaming a datalink. The physical hasn't been created yet, and as it happens, the garbage physical pointer happens to be the value of another physical - so we're pointing that other physical name at ourselves. yeuck. o Re-arrange the order of things in main (DoLoop()). We now handle signals only after the select and not before the UpdateSet. It's possible that either a signal (FSM timeout) or a descriptor_Read() brings a link down, after which we'd better tidy up any dead direct and 1off descriptors before calling UpdateSet() again. o Mention when we detect a PPP packet when we see one before the link is up (then start LCP as before).
1998-05-25 02:22:38 +00:00
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
msg.msg_name = NULL;
msg.msg_namelen = 0;
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
/*
* Only send the version to start... We used to send the whole lot, but
* this caused problems with our RECVBUF size as a single link is about
* 22k ! This way, we should bump into no limits.
*/
msg.msg_iovlen = 1;
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
msg.msg_iov = iov;
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
msg.msg_control = cmsgbuf;
msg.msg_controllen = sizeof *cmsg + sizeof(int) * nfd;
msg.msg_flags = 0;
o Don't try to transfer tty device descriptors as there's no way of transferring session rights with them. Instead, create two `/bin/cat' processes. A new child is spawned and disassociated from the terminal and the parent, which continues with the rest of the ppp process. Meanwhile, the parent spawns another child, and both the parent and child exec the `/bin/cat' commands with the appropriate descriptors. This way, the session is owned by the parent, and the tty is held open. o Close LCPs that have done a TLF and are now in ST_STOPPED before calling Down. This prevents them from trying to come back up again after the peer has shut them down (it seems a bit strange that the rfc says that a Down in ST_STOPPED will cause a TLS etc). o Don't try to set the physical link name pointer when we're receiving and renaming a datalink. The physical hasn't been created yet, and as it happens, the garbage physical pointer happens to be the value of another physical - so we're pointing that other physical name at ourselves. yeuck. o Re-arrange the order of things in main (DoLoop()). We now handle signals only after the select and not before the UpdateSet. It's possible that either a signal (FSM timeout) or a descriptor_Read() brings a link down, after which we'd better tidy up any dead direct and 1off descriptors before calling UpdateSet() again. o Mention when we detect a PPP packet when we see one before the link is up (then start LCP as before).
1998-05-25 02:22:38 +00:00
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
cmsg = (struct cmsghdr *)cmsgbuf;
cmsg->cmsg_len = msg.msg_controllen;
cmsg->cmsg_level = SOL_SOCKET;
cmsg->cmsg_type = SCM_RIGHTS;
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
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
for (f = 0; f < nfd; f++)
*((int *)(cmsg + 1) + f) = fd[f];
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
for (f = 1, expect = 0; f < niov; f++)
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
expect += iov[f].iov_len;
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
if (setsockopt(reply[0], SOL_SOCKET, SO_SNDBUF, &expect, sizeof(int)) == -1)
log_Printf(LogERROR, "setsockopt(SO_RCVBUF, %d): %s\n", expect,
strerror(errno));
if (setsockopt(reply[1], SOL_SOCKET, SO_RCVBUF, &expect, sizeof(int)) == -1)
log_Printf(LogERROR, "setsockopt(SO_RCVBUF, %d): %s\n", expect,
strerror(errno));
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
log_Printf(LogDEBUG, "Sending %d descriptor%s and %d bytes in scatter"
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
"/gather array\n", nfd, nfd == 1 ? "" : "s", iov[0].iov_len);
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
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 ((got = sendmsg(s, &msg, 0)) == -1)
log_Printf(LogERROR, "Failed sendmsg: %s: %s\n",
sun->sun_path, strerror(errno));
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
else if (got != iov[0].iov_len)
log_Printf(LogERROR, "%s: Failed initial sendmsg: Only sent %d of %d\n",
sun->sun_path, got, iov[0].iov_len);
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
else {
/* We must get the ACK before closing the descriptor ! */
int res;
Change the way we transfer links (again). The previous method avoided all race conditions, but suffered from sometimes running out of buffer space if enough clients were piled up at the same time. Now, the client pushes the link descriptor, one end of a socketpair() and the ppp version via sendmsg() at the server. The server replies with a pid. The client then transfers any link lock with uu_lock_txfr() and writev()s the actual link contents. The socketpair is now the only place we need to have large socket buffers and the bind()ed socket can keep the default 4k buffer while still handling around 90 racing clients.
1999-11-30 23:52:37 +00:00
if ((got = read(reply[0], &newpid, sizeof newpid)) == sizeof newpid) {
log_Printf(LogDEBUG, "Received confirmation from pid %d\n",
(int)newpid);
if (lock && (res = ID0uu_lock_txfr(lock, newpid)) != UU_LOCK_OK)
log_Printf(LogPHASE, "uu_lock_txfr: %s\n", uu_lockerr(res));
log_Printf(LogDEBUG, "Transmitting link (%d bytes)\n", expect);
if ((got = writev(reply[0], iov + 1, niov - 1)) != expect) {
if (got == -1)
log_Printf(LogERROR, "%s: Failed writev: %s\n",
sun->sun_path, strerror(errno));
else
log_Printf(LogERROR, "%s: Failed writev: Wrote %d of %d\n",
sun->sun_path, got, expect);
}
} else if (got == -1)
log_Printf(LogERROR, "%s: Failed socketpair read: %s\n",
sun->sun_path, strerror(errno));
else
log_Printf(LogERROR, "%s: Failed socketpair read: Got %d of %d\n",
sun->sun_path, got, (int)(sizeof 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
}
close(reply[0]);
close(reply[1]);
o Don't try to transfer tty device descriptors as there's no way of transferring session rights with them. Instead, create two `/bin/cat' processes. A new child is spawned and disassociated from the terminal and the parent, which continues with the rest of the ppp process. Meanwhile, the parent spawns another child, and both the parent and child exec the `/bin/cat' commands with the appropriate descriptors. This way, the session is owned by the parent, and the tty is held open. o Close LCPs that have done a TLF and are now in ST_STOPPED before calling Down. This prevents them from trying to come back up again after the peer has shut them down (it seems a bit strange that the rfc says that a Down in ST_STOPPED will cause a TLS etc). o Don't try to set the physical link name pointer when we're receiving and renaming a datalink. The physical hasn't been created yet, and as it happens, the garbage physical pointer happens to be the value of another physical - so we're pointing that other physical name at ourselves. yeuck. o Re-arrange the order of things in main (DoLoop()). We now handle signals only after the select and not before the UpdateSet. It's possible that either a signal (FSM timeout) or a descriptor_Read() brings a link down, after which we'd better tidy up any dead direct and 1off descriptors before calling UpdateSet() again. o Mention when we detect a PPP packet when we see one before the link is up (then start LCP as before).
1998-05-25 02:22:38 +00:00
Introduce the ``keep-session'' option. Refer to the man page for details. This allows MP over non-tty devices where the original ppp process must not exit (such as sshd-spawned ppp sessions).
1999-06-02 00:46:55 +00:00
newsid = Enabled(dl->bundle, OPT_KEEPSESSION) ||
Support PPPoE Help (lots) from: julian, archie Facilities from: ahebert@pubnix.net
1999-11-06 22:50:59 +00:00
tcgetpgrp(fd[0]) == getpgrp();
while (nfd)
close(fd[--nfd]);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
if (newsid)
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
bundle_setsid(dl->bundle, got != -1);
o Create a new ``timer'' log level. This lets us ``set log debug'' without filling our filesystem/screen with junk that we don't really want to see. o change PHYS_STDIN to PHYS_DIRECT - we can handle incoming connections that aren't on STDIN_FILENO now. o Allow return values from our FSM LayerUp functions. If LayerUp() fails, the FSM does an immediate FsmDown() without calling the fsm_parent's Layer{Up,Down} functions. o Clear the close-on-exec flag of file descriptor 3 when executing chat programs so that our documented ability to communicate with /dev/tty via that descriptor works. Also document it as descriptor 3, not 4 :-O o Allow a ``rm'' command as an alias for ``remove''. o Fix the bind()/connect()/accept() calls made by the MP server. o Create bundle_SendDatalink() and bundle_ReceiveDatalink(). This allows `struct datalink's to flatten themselves, pass through a pipe (read: the eye of a needle !) and come alive at the other end. The donator then fork()s & exec()s pppmpipe, ``passing'' the connection to another ppp instance. *** PPP NOW TALKS MULTILINK :-))) *** Our link utilization is hideous, and lots of code needs tidying still. It's also probably riddled with bugs ! It's been tested against itself only, and has hung once, so confidence isn't high....
1998-04-30 23:53:56 +00:00
}
o Don't forget to close our transfer socket if we cannot generate the data to transfer. o Transfer uucp lock ownership for the transferred device. o Don't assume we know the correct values of dev_is_modem and mbits after the transfer.
1998-05-28 23:15:40 +00:00
close(s);
o Transfer file descriptors using {send,recv}msg & SCM_RIGHTS, using the scatter/gather array to transfer the link information. The whole link is now passed in one message. This is far better than the two `/bin/cat' processes per additional link :-) I remember years ago thinking that file descriptor transferral would be a really amazing facility ! Suggested by: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> and: Eivind Eklund <eivind@yes.no>
1998-05-02 21:57:50 +00:00
while (niov--)
free(iov[niov].iov_base);
o Add the link name to modem diagnostics. o Create struct mpserver as part of struct mp. mpserver creates a unix-domain socket based on the peers auth name and endpoint discriminator. If it already exists, ppp will ``pass the link'' over to the owner of the socket, joining it into the bundle of another ppp invocation, otherwise ppp waits for other invocations to pass it links through this socket. The final piece of code will be the code that flattens our datalink info and passes it down this channel (not yet implemented).
1998-04-28 01:25:46 +00:00
}
o Add the `set mode' command for change a links current mode. It is not possible to switch to or from dedicated or direct mode, but all other combinations are ok (eg. -auto -> -ddial). o Cope with the fact that commands with optional context may not be able to obtain a link with command_ChooseLink() (if all links have been deleted for example). o Allow `clone'ing in non-multilink mode. We may for example want to configure two links in unilink mode and dial them both, using the one that comes up first. It's also possible to rename ``deflink'' by cloning it, deleting the original, then setting the mode of the new link.
1998-05-15 23:58:30 +00:00
Add the `rename' command for renaming links.
1998-05-16 23:47:28 +00:00
int
bundle_RenameDatalink(struct bundle *bundle, struct datalink *ndl,
const char *name)
{
struct datalink *dl;
if (!strcasecmp(ndl->name, name))
return 1;
for (dl = bundle->links; dl; dl = dl->next)
if (!strcasecmp(dl->name, name))
return 0;
datalink_Rename(ndl, name);
return 1;
}
o Add the `set mode' command for change a links current mode. It is not possible to switch to or from dedicated or direct mode, but all other combinations are ok (eg. -auto -> -ddial). o Cope with the fact that commands with optional context may not be able to obtain a link with command_ChooseLink() (if all links have been deleted for example). o Allow `clone'ing in non-multilink mode. We may for example want to configure two links in unilink mode and dial them both, using the one that comes up first. It's also possible to rename ``deflink'' by cloning it, deleting the original, then setting the mode of the new link.
1998-05-15 23:58:30 +00:00
int
bundle_SetMode(struct bundle *bundle, struct datalink *dl, int mode)
{
int omode;
omode = dl->physical->type;
if (omode == mode)
return 1;
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
if (mode == PHYS_AUTO && !(bundle->phys_type.all & PHYS_AUTO))
/* First auto link */
o Add the `set mode' command for change a links current mode. It is not possible to switch to or from dedicated or direct mode, but all other combinations are ok (eg. -auto -> -ddial). o Cope with the fact that commands with optional context may not be able to obtain a link with command_ChooseLink() (if all links have been deleted for example). o Allow `clone'ing in non-multilink mode. We may for example want to configure two links in unilink mode and dial them both, using the one that comes up first. It's also possible to rename ``deflink'' by cloning it, deleting the original, then setting the mode of the new link.
1998-05-15 23:58:30 +00:00
if (bundle->ncp.ipcp.peer_ip.s_addr == INADDR_ANY) {
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
log_Printf(LogWARN, "You must `set ifaddr' or `open' before"
" changing mode to %s\n", mode2Nam(mode));
o Add the `set mode' command for change a links current mode. It is not possible to switch to or from dedicated or direct mode, but all other combinations are ok (eg. -auto -> -ddial). o Cope with the fact that commands with optional context may not be able to obtain a link with command_ChooseLink() (if all links have been deleted for example). o Allow `clone'ing in non-multilink mode. We may for example want to configure two links in unilink mode and dial them both, using the one that comes up first. It's also possible to rename ``deflink'' by cloning it, deleting the original, then setting the mode of the new link.
1998-05-15 23:58:30 +00:00
return 0;
}
if (!datalink_SetMode(dl, mode))
return 0;
o Maintain a link-type mask for open datalinks as well as for all datalinks in a bundle. Ppp now deals correctly with link types that are changed while open o When changing the type of the last AUTO link, only clear the interface if we're not in PHASE_NETWORK. This allows us to switch to -ddial mode while we have a connection without suddenly unexpectedly throttling ourselves by clearing the interface configuration. Problem area noted by: Aaron Jeremias Luz <aaron@csh.rit.edu>
1998-06-12 17:45:10 +00:00
if (mode == PHYS_AUTO && !(bundle->phys_type.all & PHYS_AUTO) &&
bundle->phase != PHASE_NETWORK)
/* First auto link, we need an interface */
o Add the `set mode' command for change a links current mode. It is not possible to switch to or from dedicated or direct mode, but all other combinations are ok (eg. -auto -> -ddial). o Cope with the fact that commands with optional context may not be able to obtain a link with command_ChooseLink() (if all links have been deleted for example). o Allow `clone'ing in non-multilink mode. We may for example want to configure two links in unilink mode and dial them both, using the one that comes up first. It's also possible to rename ``deflink'' by cloning it, deleting the original, then setting the mode of the new link.
1998-05-15 23:58:30 +00:00
ipcp_InterfaceUp(&bundle->ncp.ipcp);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
/* Regenerate phys_type and adjust idle timer */
o Add `set autoload'. You can now set the minimum and maximum thresholds (in terms of queued packets for a period of time) where -auto links will be brought up and down. By default, all auto links come up when we reach NETWORK phase and never go down. o Display current autoload state in `show bundle'. o Disable the idle timer as soon as it's called. o Disable the idle and autoload timers when exiting (in case we're abending).
1998-05-21 01:13:32 +00:00
bundle_LinksRemoved(bundle);
o Add the `set mode' command for change a links current mode. It is not possible to switch to or from dedicated or direct mode, but all other combinations are ok (eg. -auto -> -ddial). o Cope with the fact that commands with optional context may not be able to obtain a link with command_ChooseLink() (if all links have been deleted for example). o Allow `clone'ing in non-multilink mode. We may for example want to configure two links in unilink mode and dial them both, using the one that comes up first. It's also possible to rename ``deflink'' by cloning it, deleting the original, then setting the mode of the new link.
1998-05-15 23:58:30 +00:00
return 1;
}
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
void
bundle_setsid(struct bundle *bundle, int holdsession)
{
/*
* Lose the current session. This means getting rid of our pid
* too so that the tty device will really go away, and any getty
* etc will be allowed to restart.
*/
pid_t pid, orig;
int fds[2];
char done;
struct datalink *dl;
orig = getpid();
if (pipe(fds) == -1) {
log_Printf(LogERROR, "pipe: %s\n", strerror(errno));
return;
}
switch ((pid = fork())) {
case -1:
log_Printf(LogERROR, "fork: %s\n", strerror(errno));
close(fds[0]);
close(fds[1]);
return;
case 0:
close(fds[1]);
Read from fd[0] and write to fd[1] where `fd' is the result of a pipe(). This matters under OpenBSD.
1999-03-07 01:41:40 +00:00
read(fds[0], &done, 1); /* uu_locks are mine ! */
close(fds[0]);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
if (pipe(fds) == -1) {
log_Printf(LogERROR, "pipe(2): %s\n", strerror(errno));
return;
}
switch ((pid = fork())) {
case -1:
Change some log levels. ALERTs are only logged when something that can't happen happens or when everyone needs to know. ERRORs are only logged when something unexpected happens.
1998-06-16 19:40:42 +00:00
log_Printf(LogERROR, "fork(2): %s\n", strerror(errno));
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
close(fds[0]);
close(fds[1]);
return;
case 0:
close(fds[1]);
Read from fd[0] and write to fd[1] where `fd' is the result of a pipe(). This matters under OpenBSD.
1999-03-07 01:41:40 +00:00
bundle_LockTun(bundle); /* update pid */
read(fds[0], &done, 1); /* uu_locks are mine ! */
close(fds[0]);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
setsid();
log_Printf(LogPHASE, "%d -> %d: %s session control\n",
(int)orig, (int)getpid(),
holdsession ? "Passed" : "Dropped");
Rather than interrupting 10 times per second then checking to see if there's anything to do, schedule the next alarm based on the next required timeout. This decreases the load when there are lots of relatively idle ppp processes. While I'm in there, handle the possibility that a timeout makes the timer element go out of scope by grabbing the enext pointer before executing the timer function.
1998-12-14 19:24:30 +00:00
timer_InitService(0); /* Start the Timer Service */
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
break;
default:
Read from fd[0] and write to fd[1] where `fd' is the result of a pipe(). This matters under OpenBSD.
1999-03-07 01:41:40 +00:00
close(fds[0]);
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
/* Give away all our physical locks (to the final process) */
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
for (dl = bundle->links; dl; dl = dl->next)
if (dl->state != DATALINK_CLOSED)
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
physical_ChangedPid(dl->physical, pid);
Read from fd[0] and write to fd[1] where `fd' is the result of a pipe(). This matters under OpenBSD.
1999-03-07 01:41:40 +00:00
write(fds[1], "!", 1); /* done */
close(fds[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
_exit(0);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
break;
}
break;
default:
Read from fd[0] and write to fd[1] where `fd' is the result of a pipe(). This matters under OpenBSD.
1999-03-07 01:41:40 +00:00
close(fds[0]);
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
/* Give away all our physical locks (to the intermediate process) */
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
for (dl = bundle->links; dl; dl = dl->next)
if (dl->state != DATALINK_CLOSED)
o Redesign the layering mechanism and make the aliasing code part of the layering. We now ``stack'' layers as soon as we open the device (when we figure out what we're dealing with). A static set of `dispatch' routines are also declared for dealing with incoming packets after they've been `pulled' up through the stacked layers. Physical devices are now assigned handlers based on the device type when they're opened. For the moment there are three device types; ttys, execs and tcps. o Increment version number to 2.2 o Make an entry in [uw]tmp for non-tty -direct invocations (after pap/chap authentication). o Make throughput counters quad_t's o Account for the absolute number of mbuf malloc()s and free()s in ``show mem''. o ``show modem'' becomes ``show physical''.
1999-05-08 11:07:56 +00:00
physical_ChangedPid(dl->physical, pid);
Read from fd[0] and write to fd[1] where `fd' is the result of a pipe(). This matters under OpenBSD.
1999-03-07 01:41:40 +00:00
write(fds[1], "!", 1); /* done */
close(fds[1]);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
if (holdsession) {
int fd, status;
timer_TermService();
signal(SIGPIPE, SIG_DFL);
signal(SIGALRM, SIG_DFL);
signal(SIGHUP, SIG_DFL);
signal(SIGTERM, SIG_DFL);
signal(SIGINT, SIG_DFL);
signal(SIGQUIT, SIG_DFL);
for (fd = getdtablesize(); fd >= 0; fd--)
close(fd);
/*
* Reap the intermediate process. As we're not exiting but the
* intermediate is, we don't want it to become defunct.
*/
waitpid(pid, &status, 0);
o If there's a session leader left running for a descriptor that we're now closing, manually HUP that session leader so that the tty is fully released. o Always restart our carrier detect timer in the receiving process if it was running in the sending process (as we now *always* pass the descriptor). o Tweak argv when we go into pause() mode to keep our session so that ps can see what's going on (without checking for a `pause' state in `ps -l').
1998-05-29 18:33:10 +00:00
/* Tweak our process arguments.... */
Fix ``set proctitle'' by using setproctitle().
1999-11-17 21:12:35 +00:00
ID0setproctitle("session owner");
Don't setuid() 'till we've called ID0setproctitle()
1999-11-23 00:14:50 +00:00
setuid(geteuid());
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
/*
* Hang around for a HUP. This should happen as soon as the
* ppp that we passed our ctty descriptor to closes it.
* NOTE: If this process dies, the passed descriptor becomes
* invalid and will give a select() error by setting one
* of the error fds, aborting the other ppp. We don't
* want that to happen !
*/
pause();
}
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
_exit(0);
o #define the name "tun" in defs.h against the future possibility of supporting architectures with different device names. o Close /dev/tunX when destroying the bundle. o Don't forget to close the parent end of the pipe in the child process when exec'ing a program from a chat script. o If we close our controlling terminal, ditch the current session with it, allowing getty(8) (or whatever) to regain control. o After transferring our controlling terminal descriptor to another ppp instance, we now fork a new ppp to continue where we left off, transferring ownership of all uucp locks and the /var/run/tunX.pid file. Meanwhile the parent closes all file descriptors, defaults all signals and does a pause() to wait for a HUP after the transferred descriptor is finally closed. We don't run /bin/cat any more (again!). Suggested by: bde TODO: It seems clocal devices need their pause()d session leader to be given a manual HUP, as closing the last open descriptor doesn't do the job.
1998-05-28 23:17:51 +00:00
break;
}
}
Loosen our restrictions on setting enddisc, mrru, shortseq, authname and authkey. o Auth{name,key} may additionally be set in PHASE_ESTABLISH. o The others may be set in PHASE_ESTABLISH as long as no links have yet reached DATALINK_LCP.
1998-10-24 01:08:45 +00:00
int
bundle_HighestState(struct bundle *bundle)
{
struct datalink *dl;
int result = DATALINK_CLOSED;
for (dl = bundle->links; dl; dl = dl->next)
if (result < dl->state)
result = dl->state;
return result;
}
When using ``set device !someprogram'', when ``someprogram'' exits, it causes a select() exception. Handle these select() exceptions on link descriptors in pretty much the same way as loss of carrier rather than dropping out in confusion.
1998-12-10 18:36:30 +00:00
int
bundle_Exception(struct bundle *bundle, int fd)
{
struct datalink *dl;
for (dl = bundle->links; dl; dl = dl->next)
if (dl->physical->fd == fd) {
datalink_Down(dl, CLOSE_NORMAL);
return 1;
}
return 0;
}
Remember if MYADDR or HISADDR is used in a filter add tweak all filters any time either value changes.
1999-05-31 23:57:40 +00:00
void
bundle_AdjustFilters(struct bundle *bundle, struct in_addr *my_ip,
struct in_addr *peer_ip)
{
filter_AdjustAddr(&bundle->filter.in, my_ip, peer_ip);
filter_AdjustAddr(&bundle->filter.out, my_ip, peer_ip);
filter_AdjustAddr(&bundle->filter.dial, my_ip, peer_ip);
filter_AdjustAddr(&bundle->filter.alive, my_ip, peer_ip);
}
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
void
bundle_CalculateBandwidth(struct bundle *bundle)
{
struct datalink *dl;
int mtu, sp;
bundle->bandwidth = 0;
mtu = 0;
for (dl = bundle->links; dl; dl = dl->next)
if (dl->state == DATALINK_OPEN) {
if ((sp = dl->mp.bandwidth) == 0 &&
(sp = physical_GetSpeed(dl->physical)) == 0)
log_Printf(LogDEBUG, "%s: %s: Cannot determine bandwidth\n",
dl->name, dl->physical->name.full);
else
bundle->bandwidth += sp;
if (!bundle->ncp.mp.active) {
mtu = dl->physical->link.lcp.his_mru;
break;
}
}
if(bundle->bandwidth == 0)
bundle->bandwidth = 115200; /* Shrug */
if (bundle->ncp.mp.active)
mtu = bundle->ncp.mp.peer_mrru;
else if (!mtu)
mtu = 1500;
#ifndef NORADIUS
if (bundle->radius.valid && bundle->radius.mtu && bundle->radius.mtu < mtu) {
log_Printf(LogLCP, "Reducing MTU to radius value %lu\n",
bundle->radius.mtu);
mtu = bundle->radius.mtu;
}
#endif
tun_configure(bundle, mtu);
}
void
bundle_AutoAdjust(struct bundle *bundle, int percent, int what)
{
struct datalink *dl, *choice, *otherlinkup;
choice = otherlinkup = NULL;
for (dl = bundle->links; dl; dl = dl->next)
if (dl->physical->type == PHYS_AUTO) {
if (dl->state == DATALINK_OPEN) {
if (what == AUTO_DOWN) {
if (choice)
otherlinkup = choice;
choice = dl;
}
} else if (dl->state == DATALINK_CLOSED) {
if (what == AUTO_UP) {
choice = dl;
break;
}
} else {
/* An auto link in an intermediate state - forget it for the moment */
choice = NULL;
break;
}
} else if (dl->state == DATALINK_OPEN && what == AUTO_DOWN)
otherlinkup = dl;
if (choice) {
if (what == AUTO_UP) {
log_Printf(LogPHASE, "%d%% saturation -> Opening link ``%s''\n",
percent, choice->name);
datalink_Up(choice, 1, 1);
mp_StopAutoloadTimer(&bundle->ncp.mp);
} else if (otherlinkup) { /* Only bring the second-last link down */
log_Printf(LogPHASE, "%d%% saturation -> Closing link ``%s''\n",
percent, choice->name);
Bring links down gently when their throughput isn't enough for the ``set autoload'' value. Don't just ``down'' the link. Don't get stuck in ``ready'' the first time we ``close'' a link.
1999-10-05 05:01:44 +00:00
datalink_Close(choice, CLOSE_STAYDOWN);
o Obsolete the undocumented ``set weight'' command. o If we're using RADIUS and the RADIUS mtu is less than our peers mru/mrru, reduce our mtu to this value for NetBSD too. o Make struct throughput's sample period dynamic and tweak the ppp version number to reflect the extra stuff being passed through the local domain socket as a result (MP mode). o Measure the current throughput based on the number of samples actually taken rather than on the full sample period. o Keep the throughput statisics persistent while being passed to another ppp invocation through the local domain socket. o When showing throughput statistics after the timer has stopped, use the stopped time for overall calculations, not the current time. Also show the stopped time and how long the current throughput has been sampled for. o Use time() consistently in throughput.c o Tighten up the ``show bundle'' output. o Introduce the ``set bandwidth'' command. o Rewrite the ``set autoload'' command. It now takes three arguments and works based on a rolling bundle throughput average compared against the theoretical bundle bandwidth over a given period (read: it's now functional).
1999-08-05 10:32:16 +00:00
mp_StopAutoloadTimer(&bundle->ncp.mp);
}
}
}
int
bundle_WantAutoloadTimer(struct bundle *bundle)
{
struct datalink *dl;
int autolink, opened;
if (bundle->phase == PHASE_NETWORK) {
for (autolink = opened = 0, dl = bundle->links; dl; dl = dl->next)
if (dl->physical->type == PHYS_AUTO) {
if (++autolink == 2 || (autolink == 1 && opened))
/* Two auto links or one auto and one open in NETWORK phase */
return 1;
} else if (dl->state == DATALINK_OPEN) {
opened++;
if (autolink)
/* One auto and one open link in NETWORK phase */
return 1;
}
}
return 0;
}
Reference in New Issue Copy Permalink