d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
7d29df5931
freebsd-dev/sys/netatalk/at_control.c

887 lines
23 KiB
C
Raw Normal View History

/* -> /*- for license, minor formatting changes, insert COPYRIGHT into files
2005-01-07 02:35:34 +00:00
/*-
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
* Copyright (c) 1990,1991 Regents of The University of Michigan.
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
* Copyright (c) 2009 Robert N. M. Watson
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
* All Rights Reserved.
Move suser() and suser_xxx() prototypes and a related #define from <sys/proc.h> to <sys/systm.h>. Correctly document the #includes needed in the manpage. Add one now needed #include of <sys/systm.h>. Remove the consequent 48 unused #includes of <sys/proc.h>.
2000-10-29 16:06:56 +00:00
*
/* -> /*- for license, minor formatting changes, insert COPYRIGHT into files
2005-01-07 02:35:34 +00:00
* Permission to use, copy, modify, and distribute this software and
* its documentation for any purpose and without fee is hereby granted,
* provided that the above copyright notice appears in all copies and
* that both that copyright notice and this permission notice appear
* in supporting documentation, and that the name of The University
* of Michigan not be used in advertising or publicity pertaining to
* distribution of the software without specific, written prior
* permission. This software is supplied as is without expressed or
* implied warranties of any kind.
*
* This product includes software developed by the University of
* California, Berkeley and its contributors.
*
* Research Systems Unix Group
* The University of Michigan
* c/o Wesley Craig
* 535 W. William Street
* Ann Arbor, Michigan
* +1-313-764-2278
* netatalk@umich.edu
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
*/
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Don't include <sys/ioctl.h> in the kernel. Stage 2: include <sys/sockio.h> instead of <sys/ioctl.h> in network files.
1997-03-24 11:33:46 +00:00
#include <sys/sockio.h>
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
#include <sys/lock.h>
Added used #include - don't depend on <sys/mbuf.h> including <sys/malloc.h> (unless we only use the bogusly shared M*WAIT flags).
1997-09-02 01:19:47 +00:00
#include <sys/malloc.h>
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
#include <sys/kernel.h>
Use privilege PRIV_NET_ADDIFADDR rather than suser() to authorize adding a netatalk address to an interface.
2007-02-19 22:40:02 +00:00
#include <sys/priv.h>
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
#include <sys/rwlock.h>
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
#include <sys/socket.h>
#include <net/if.h>
#include <net/route.h>
#include <netinet/in.h>
#undef s_net
#include <netinet/if_ether.h>
Updated #includes to 4.4Lite style.
1996-09-10 08:32:01 +00:00
#include <netatalk/at.h>
#include <netatalk/at_var.h>
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
#include <netatalk/at_extern.h>
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
struct rwlock at_ifaddr_rw;
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
struct at_ifaddrhead at_ifaddrhead;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
RW_SYSINIT(at_ifaddr_rw, &at_ifaddr_rw, "at_ifaddr_rw");
Fixed sloppy common-style declarations.
1997-09-07 07:33:30 +00:00
Further function forward declaration white space tweaks.
2004-07-19 17:18:58 +00:00
static int aa_dorangeroute(struct ifaddr *ifa, u_int first, u_int last,
int cmd);
static int aa_addsingleroute(struct ifaddr *ifa, struct at_addr *addr,
struct at_addr *mask);
static int aa_delsingleroute(struct ifaddr *ifa, struct at_addr *addr,
struct at_addr *mask);
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
static int aa_dosingleroute(struct ifaddr *ifa, struct at_addr *addr,
Further function forward declaration white space tweaks.
2004-07-19 17:18:58 +00:00
struct at_addr *mask, int cmd, int flags);
Since I have my hands all over netatalk adding locking and restructuring it, cinch the file's style closer to style(9) with regard to parenthesis: s/( /(/g s/ )/)/g s/return(/return (/g s/return 0/return (0)/ s/return 1/return (1)/
2004-03-22 03:24:10 +00:00
static int at_scrub(struct ifnet *ifp, struct at_ifaddr *aa);
static int at_ifinit(struct ifnet *ifp, struct at_ifaddr *aa,
Further function forward declaration white space tweaks.
2004-07-19 17:18:58 +00:00
struct sockaddr_at *sat);
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
static int aa_claim_addr(struct ifaddr *ifa, struct sockaddr *gw);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Further function forward declaration white space tweaks.
2004-07-19 17:18:58 +00:00
#define sateqaddr(a,b) \
((a)->sat_len == (b)->sat_len && \
(a)->sat_family == (b)->sat_family && \
(a)->sat_addr.s_net == (b)->sat_addr.s_net && \
(a)->sat_addr.s_node == (b)->sat_addr.s_node)
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
at_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
struct thread *td)
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
{
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
struct ifreq *ifr = (struct ifreq *)data;
struct sockaddr_at *sat;
struct netrange *nr;
struct at_aliasreq *ifra = (struct at_aliasreq *)data;
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
struct at_ifaddr *aa;
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
struct ifaddr *ifa;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
int error;
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* If we have an ifp, then find the matching at_ifaddr if it exists
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
aa = NULL;
AT_IFADDR_RLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (ifp != NULL) {
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
TAILQ_FOREACH(aa, &at_ifaddrhead, aa_link) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_ifp == ifp)
break;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (aa != NULL)
ifa_ref(&aa->aa_ifa);
AT_IFADDR_RUNLOCK();
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* In this first switch table we are basically getting ready for
* the second one, by getting the atalk-specific things set up
* so that they start to look more similar to other protocols etc.
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
error = 0;
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
switch (cmd) {
case SIOCAIFADDR:
case SIOCDIFADDR:
/*
* If we have an appletalk sockaddr, scan forward of where we
* are now on the at_ifaddr list to find one with a matching
* address on this interface. This may leave aa pointing to
* the first address on the NEXT interface!
*/
if (ifra->ifra_addr.sat_family == AF_APPLETALK) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
struct at_ifaddr *oaa;
AT_IFADDR_RLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
for (oaa = aa; aa; aa = TAILQ_NEXT(aa, aa_link)) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_ifp == ifp &&
sateqaddr(&aa->aa_addr, &ifra->ifra_addr))
break;
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (oaa != NULL && oaa != aa)
ifa_free(&oaa->aa_ifa);
if (aa != NULL && oaa != aa)
ifa_ref(&aa->aa_ifa);
AT_IFADDR_RUNLOCK();
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* If we a retrying to delete an addres but didn't find such,
* then rewurn with an error
*/
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
if (cmd == SIOCDIFADDR && aa == NULL) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
error = EADDRNOTAVAIL;
goto out;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*FALLTHROUGH*/
case SIOCSIFADDR:
/*
* If we are not superuser, then we don't get to do these ops.
Sweep kernel replacing suser(9) calls with priv(9) calls, assigning specific privilege names to a broad range of privileges. These may require some future tweaking. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:42:10 +00:00
*
* XXXRW: Layering?
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
*/
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
if (priv_check(td, PRIV_NET_ADDIFADDR)) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
error = EPERM;
goto out;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
sat = satosat(&ifr->ifr_addr);
nr = (struct netrange *)sat->sat_zero;
if (nr->nr_phase == 1) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
struct at_ifaddr *oaa;
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* Look for a phase 1 address on this interface.
* This may leave aa pointing to the first address on
* the NEXT interface!
*/
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
AT_IFADDR_RLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
for (oaa = aa; aa; aa = TAILQ_NEXT(aa, aa_link)) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_ifp == ifp &&
(aa->aa_flags & AFA_PHASE2) == 0)
break;
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (oaa != NULL && oaa != aa)
ifa_free(&oaa->aa_ifa);
if (aa != NULL && oaa != aa)
ifa_ref(&aa->aa_ifa);
AT_IFADDR_RUNLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
} else { /* default to phase 2 */
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
struct at_ifaddr *oaa;
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* Look for a phase 2 address on this interface.
* This may leave aa pointing to the first address on
* the NEXT interface!
*/
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
AT_IFADDR_RLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
for (oaa = aa; aa; aa = TAILQ_NEXT(aa, aa_link)) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_ifp == ifp && (aa->aa_flags &
AFA_PHASE2))
break;
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (oaa != NULL && oaa != aa)
ifa_free(&oaa->aa_ifa);
if (aa != NULL && oaa != aa)
ifa_ref(&aa->aa_ifa);
AT_IFADDR_RUNLOCK();
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (ifp == NULL)
panic("at_control");
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* If we failed to find an existing at_ifaddr entry, then we
* allocate a fresh one.
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa == NULL) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
aa = malloc(sizeof(struct at_ifaddr), M_IFADDR,
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
M_NOWAIT | M_ZERO);
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (aa == NULL) {
error = ENOBUFS;
goto out;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
callout_init(&aa->aa_callout, CALLOUT_MPSAFE);
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
ifa = (struct ifaddr *)aa;
Clean up common ifaddr management: - Unify reference count and lock initialization in a single function, ifa_init(). - Move tear-down from a macro (IFAFREE) to a function ifa_free(). - Move reference count bump from a macro (IFAREF) to a function ifa_ref(). - Instead of using a u_int protected by a mutex to refcount(9) for reference count management. The ifa_mtx is now used for exactly one ioctl, and possibly should be removed. MFC after: 3 weeks
2009-06-21 19:30:33 +00:00
ifa_init(ifa);
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* As the at_ifaddr contains the actual sockaddrs,
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
* and the ifaddr itself, link them all together
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* correctly.
*/
ifa->ifa_addr = (struct sockaddr *)&aa->aa_addr;
ifa->ifa_dstaddr = (struct sockaddr *)&aa->aa_addr;
ifa->ifa_netmask = (struct sockaddr *)&aa->aa_netmask;
/*
* Set/clear the phase 2 bit.
*/
if (nr->nr_phase == 1)
aa->aa_flags &= ~AFA_PHASE2;
else
aa->aa_flags |= AFA_PHASE2;
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
ifa_ref(&aa->aa_ifa); /* at_ifaddrhead */
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
AT_IFADDR_WLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
if (!TAILQ_EMPTY(&at_ifaddrhead)) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
/*
* Don't let the loopback be first, since the
* first address is the machine's default
* address for binding. If it is, stick
* ourself in front, otherwise go to the back
* of the list.
*/
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
if (TAILQ_FIRST(&at_ifaddrhead)->aa_ifp->
if_flags & IFF_LOOPBACK)
TAILQ_INSERT_HEAD(&at_ifaddrhead, aa,
aa_link);
else
TAILQ_INSERT_TAIL(&at_ifaddrhead, aa,
aa_link);
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
} else
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
TAILQ_INSERT_HEAD(&at_ifaddrhead, aa,
aa_link);
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
AT_IFADDR_WUNLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* and link it all together
*/
aa->aa_ifp = ifp;
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
ifa_ref(&aa->aa_ifa); /* if_addrhead */
Lock interface address list lock around ifaddr inserts and deletes in at_control(). This locking is not yet complete but should prevent some classes of race conditions. MFC after: 2 weeks
2009-04-19 22:01:38 +00:00
IF_ADDR_LOCK(ifp);
TAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
IF_ADDR_UNLOCK(ifp);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
} else {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* If we DID find one then we clobber any routes
* dependent on it..
*/
at_scrub(ifp, aa);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
break;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
case SIOCGIFADDR :
sat = satosat(&ifr->ifr_addr);
nr = (struct netrange *)sat->sat_zero;
if (nr->nr_phase == 1) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
struct at_ifaddr *oaa;
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* If the request is specifying phase 1, then
* only look at a phase one address
*/
Reverse misordered unlock and lock in at_control for netatalk phase I addresses. Submitted by: Russell Cattelan <cattelan at thebarn.com> Approved by: re (kib)
2009-08-12 10:44:13 +00:00
AT_IFADDR_RLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
for (oaa = aa; aa; aa = TAILQ_NEXT(aa, aa_link)) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_ifp == ifp &&
(aa->aa_flags & AFA_PHASE2) == 0)
break;
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (oaa != NULL && oaa != aa)
ifa_free(&oaa->aa_ifa);
if (aa != NULL && oaa != aa)
ifa_ref(&aa->aa_ifa);
Reverse misordered unlock and lock in at_control for netatalk phase I addresses. Submitted by: Russell Cattelan <cattelan at thebarn.com> Approved by: re (kib)
2009-08-12 10:44:13 +00:00
AT_IFADDR_RUNLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
} else {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
struct at_ifaddr *oaa;
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* default to phase 2
*/
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
AT_IFADDR_RLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
for (oaa = aa; aa; aa = TAILQ_NEXT(aa, aa_link)) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_ifp == ifp && (aa->aa_flags &
AFA_PHASE2))
break;
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (oaa != NULL && oaa != aa)
ifa_free(&oaa->aa_ifa);
if (aa != NULL && oaa != aa)
ifa_ref(&aa->aa_ifa);
AT_IFADDR_RUNLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
}
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
if (aa == NULL) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
error = EADDRNOTAVAIL;
goto out;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
break;
}
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* By the time this switch is run we should be able to assume that
* the "aa" pointer is valid when needed.
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
switch (cmd) {
case SIOCGIFADDR:
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* copy the contents of the sockaddr blindly.
*/
sat = (struct sockaddr_at *)&ifr->ifr_addr;
*sat = aa->aa_addr;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* and do some cleanups
*/
((struct netrange *)&sat->sat_zero)->nr_phase
= (aa->aa_flags & AFA_PHASE2) ? 2 : 1;
((struct netrange *)&sat->sat_zero)->nr_firstnet =
aa->aa_firstnet;
((struct netrange *)&sat->sat_zero)->nr_lastnet =
aa->aa_lastnet;
break;
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
case SIOCSIFADDR:
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
error = at_ifinit(ifp, aa,
(struct sockaddr_at *)&ifr->ifr_addr);
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
goto out;
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
case SIOCAIFADDR:
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
if (sateqaddr(&ifra->ifra_addr, &aa->aa_addr)) {
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
error = 0;
goto out;
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
}
error = at_ifinit(ifp, aa,
(struct sockaddr_at *)&ifr->ifr_addr);
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
goto out;
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
case SIOCDIFADDR:
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* remove the ifaddr from the interface
*/
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
ifa = (struct ifaddr *)aa;
Lock interface address list lock around ifaddr inserts and deletes in at_control(). This locking is not yet complete but should prevent some classes of race conditions. MFC after: 2 weeks
2009-04-19 22:01:38 +00:00
IF_ADDR_LOCK(ifp);
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
TAILQ_REMOVE(&ifp->if_addrhead, ifa, ifa_link);
Lock interface address list lock around ifaddr inserts and deletes in at_control(). This locking is not yet complete but should prevent some classes of race conditions. MFC after: 2 weeks
2009-04-19 22:01:38 +00:00
IF_ADDR_UNLOCK(ifp);
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
ifa_free(ifa); /* if_addrhead */
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* Now remove the at_ifaddr from the parallel structure
* as well, or we'd be in deep trouble
*/
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
AT_IFADDR_WLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
TAILQ_REMOVE(&at_ifaddrhead, aa, aa_link);
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
AT_IFADDR_WUNLOCK();
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
ifa_free(ifa); /* at_ifaddrhead */
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
break;
default:
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
if (ifp == NULL || ifp->if_ioctl == NULL) {
error = EOPNOTSUPP;
goto out;
}
error = ((*ifp->if_ioctl)(ifp, cmd, data));
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
}
Break at_ifawithnet() into two variants: - at_ifawithnet(), which acquires an locks it needs and returns an at_ifaddr reference. - at_ifawithnet_locked(), which relies on the caller locking at_ifaddr_list, and returns a pointer rather than a reference. Update various consumers to prefer one or the other, including ether and fddi output, to properly release at_ifaddr references. Rework at_control() to manage locking and references in a manner identical to in_control(). MFC after: 6 weeks
2009-06-24 10:32:44 +00:00
out:
if (aa != NULL)
ifa_free(&aa->aa_ifa);
return (error);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
* Given an interface and an at_ifaddr (supposedly on that interface)
* remove any routes that depend on this.
* Why ifp is needed I'm not sure,
* as aa->at_ifaddr.ifa_ifp should be the same.
*/
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
static int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
at_scrub(struct ifnet *ifp, struct at_ifaddr *aa)
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
{
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
int error;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_flags & AFA_ROUTE) {
if (ifp->if_flags & IFF_LOOPBACK) {
if ((error = aa_delsingleroute(&aa->aa_ifa,
&aa->aa_addr.sat_addr, &aa->aa_netmask.sat_addr))
!= 0)
return (error);
} else if (ifp->if_flags & IFF_POINTOPOINT) {
if ((error = rtinit(&aa->aa_ifa, RTM_DELETE,
RTF_HOST)) != 0)
return (error);
} else if (ifp->if_flags & IFF_BROADCAST) {
error = aa_dorangeroute(&aa->aa_ifa,
ntohs(aa->aa_firstnet), ntohs(aa->aa_lastnet),
RTM_DELETE);
allow a new loopback route to overwrite an old one.. this allows the atalkd to be restarted.. a better fix will come later.
1996-09-03 06:23:13 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa->aa_ifa.ifa_flags &= ~IFA_ROUTE;
aa->aa_flags &= ~AFA_ROUTE;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
return (0);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
* given an at_ifaddr,a sockaddr_at and an ifp,
* bang them all together at high speed and see what happens
*/
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
static int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
at_ifinit(struct ifnet *ifp, struct at_ifaddr *aa, struct sockaddr_at *sat)
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
{
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
struct netrange nr, onr;
struct sockaddr_at oldaddr;
int error = 0, i, j;
int netinc, nodeinc, nnets;
u_short net;
/*
* save the old addresses in the at_ifaddr just in case we need them.
*/
oldaddr = aa->aa_addr;
onr.nr_firstnet = aa->aa_firstnet;
onr.nr_lastnet = aa->aa_lastnet;
/*
* take the address supplied as an argument, and add it to the
* at_ifnet (also given). Remember ing to update
* those parts of the at_ifaddr that need special processing
*/
bzero(AA_SAT(aa), sizeof(struct sockaddr_at));
bcopy(sat->sat_zero, &nr, sizeof(struct netrange));
bcopy(sat->sat_zero, AA_SAT(aa)->sat_zero, sizeof(struct netrange));
nnets = ntohs(nr.nr_lastnet) - ntohs(nr.nr_firstnet) + 1;
aa->aa_firstnet = nr.nr_firstnet;
aa->aa_lastnet = nr.nr_lastnet;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
/* XXX ALC */
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
#if 0
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
printf("at_ifinit: %s: %u.%u range %u-%u phase %d\n",
ifp->if_name,
ntohs(sat->sat_addr.s_net), sat->sat_addr.s_node,
ntohs(aa->aa_firstnet), ntohs(aa->aa_lastnet),
(aa->aa_flags & AFA_PHASE2) ? 2 : 1);
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
#endif
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* We could eliminate the need for a second phase 1 probe (post
* autoconf) if we check whether we're resetting the node. Note
* that phase 1 probes use only nodes, not net.node pairs. Under
* phase 2, both the net and node must be the same.
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (ifp->if_flags & IFF_LOOPBACK) {
AA_SAT(aa)->sat_len = sat->sat_len;
AA_SAT(aa)->sat_family = AF_APPLETALK;
AA_SAT(aa)->sat_addr.s_net = sat->sat_addr.s_net;
AA_SAT(aa)->sat_addr.s_node = sat->sat_addr.s_node;
#if 0
} else if (fp->if_flags & IFF_POINTOPOINT) {
/* unimplemented */
/*
* we'd have to copy the dstaddr field over from the sat
* but it's not clear that it would contain the right info..
*/
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
#endif
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
} else {
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* We are a normal (probably ethernet) interface.
* apply the new address to the interface structures etc.
* We will probe this address on the net first, before
* applying it to ensure that it is free.. If it is not, then
* we will try a number of other randomly generated addresses
* in this net and then increment the net. etc.etc. until
* we find an unused address.
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa->aa_flags |= AFA_PROBING; /* not loopback we Must probe? */
AA_SAT(aa)->sat_len = sizeof(struct sockaddr_at);
AA_SAT(aa)->sat_family = AF_APPLETALK;
if (aa->aa_flags & AFA_PHASE2) {
if (sat->sat_addr.s_net == ATADDR_ANYNET) {
/*
* If we are phase 2, and the net was not
* specified then we select a random net
* within the supplied netrange.
* XXX use /dev/random?
*/
if (nnets != 1)
net = ntohs(nr.nr_firstnet) +
time_second % (nnets - 1);
else
net = ntohs(nr.nr_firstnet);
} else {
/*
* if a net was supplied, then check that it
* is within the netrange. If it is not then
* replace the old values and return an error
*/
if (ntohs(sat->sat_addr.s_net) <
ntohs(nr.nr_firstnet) ||
ntohs(sat->sat_addr.s_net) >
ntohs(nr.nr_lastnet)) {
aa->aa_addr = oldaddr;
aa->aa_firstnet = onr.nr_firstnet;
aa->aa_lastnet = onr.nr_lastnet;
return (EINVAL);
}
/*
* otherwise just use the new net number..
*/
net = ntohs(sat->sat_addr.s_net);
}
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
} else {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* we must be phase one, so just use whatever we were
* given. I guess it really isn't going to be
* used... RIGHT?
*/
net = ntohs(sat->sat_addr.s_net);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* set the node part of the address into the ifaddr.
* If it's not specified, be random about it...
* XXX use /dev/random?
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (sat->sat_addr.s_node == ATADDR_ANYNODE)
AA_SAT(aa)->sat_addr.s_node = time_second;
else
AA_SAT(aa)->sat_addr.s_node = sat->sat_addr.s_node;
/*
* Copy the phase.
*/
AA_SAT(aa)->sat_range.r_netrange.nr_phase =
((aa->aa_flags & AFA_PHASE2) ? 2:1);
/*
* step through the nets in the range
* starting at the (possibly random) start point.
*/
for (i = nnets, netinc = 1; i > 0; net =
ntohs(nr.nr_firstnet) + ((net - ntohs(nr.nr_firstnet) +
netinc) % nnets), i--) {
AA_SAT(aa)->sat_addr.s_net = htons(net);
/*
* using a rather strange stepping method,
* stagger through the possible node addresses
* Once again, starting at the (possibly random)
* initial node address.
*/
for (j = 0, nodeinc = time_second | 1; j < 256;
j++, AA_SAT(aa)->sat_addr.s_node += nodeinc) {
if (AA_SAT(aa)->sat_addr.s_node > 253 ||
AA_SAT(aa)->sat_addr.s_node < 1)
continue;
aa->aa_probcnt = 10;
/*
* start off the probes as an asynchronous
* activity. though why wait 200mSec?
*/
Convert the aa_ifaddr timeout to a callout, and run the aarprobe callout MPSAFE. Acquire the aarptab_mtx to make sure that the callout and msleep in the ioctl thread don't race. MFC after: 1 week
2005-02-22 14:20:29 +00:00
AARPTAB_LOCK();
callout_reset(&aa->aa_callout, hz / 5,
aarpprobe, ifp);
if (msleep(aa, &aarptab_mtx, PPAUSE|PCATCH,
"at_ifinit", 0)) {
AARPTAB_UNLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* theoretically we shouldn't time
* out here so if we returned with an
* error..
*/
printf("at_ifinit: why did this "
"happen?!\n");
aa->aa_addr = oldaddr;
aa->aa_firstnet = onr.nr_firstnet;
aa->aa_lastnet = onr.nr_lastnet;
return (EINTR);
}
Convert the aa_ifaddr timeout to a callout, and run the aarprobe callout MPSAFE. Acquire the aarptab_mtx to make sure that the callout and msleep in the ioctl thread don't race. MFC after: 1 week
2005-02-22 14:20:29 +00:00
AARPTAB_UNLOCK();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* The async activity should have woken us
* up. We need to see if it was successful
* in finding a free spot, or if we need to
* iterate to the next address to try.
*/
if ((aa->aa_flags & AFA_PROBING) == 0)
break;
}
/*
* of course we need to break out through two loops...
*/
if ((aa->aa_flags & AFA_PROBING) == 0)
break;
/* reset node for next network */
AA_SAT(aa)->sat_addr.s_node = time_second;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* if we are still trying to probe, then we have finished all
* the possible addresses, so we need to give up
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (aa->aa_flags & AFA_PROBING) {
aa->aa_addr = oldaddr;
aa->aa_firstnet = onr.nr_firstnet;
aa->aa_lastnet = onr.nr_lastnet;
return (EADDRINUSE);
}
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* Now that we have selected an address, we need to tell the interface
* about it, just in case it needs to adjust something.
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (ifp->if_ioctl != NULL &&
(error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)aa))) {
/*
* of course this could mean that it objects violently
* so if it does, we back out again..
*/
aa->aa_addr = oldaddr;
aa->aa_firstnet = onr.nr_firstnet;
aa->aa_lastnet = onr.nr_lastnet;
return (error);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* set up the netmask part of the at_ifaddr
* and point the appropriate pointer in the ifaddr to it.
* probably pointless, but what the heck.. XXX
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
bzero(&aa->aa_netmask, sizeof(aa->aa_netmask));
aa->aa_netmask.sat_len = sizeof(struct sockaddr_at);
aa->aa_netmask.sat_family = AF_APPLETALK;
aa->aa_netmask.sat_addr.s_net = 0xffff;
aa->aa_netmask.sat_addr.s_node = 0;
aa->aa_ifa.ifa_netmask =(struct sockaddr *) &(aa->aa_netmask); /* XXX */
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* Initialize broadcast (or remote p2p) address
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
bzero(&aa->aa_broadaddr, sizeof(aa->aa_broadaddr));
aa->aa_broadaddr.sat_len = sizeof(struct sockaddr_at);
aa->aa_broadaddr.sat_family = AF_APPLETALK;
aa->aa_ifa.ifa_metric = ifp->if_metric;
if (ifp->if_flags & IFF_BROADCAST) {
aa->aa_broadaddr.sat_addr.s_net = htons(0);
aa->aa_broadaddr.sat_addr.s_node = 0xff;
aa->aa_ifa.ifa_broadaddr = (struct sockaddr *)
&aa->aa_broadaddr;
/* add the range of routes needed */
error = aa_dorangeroute(&aa->aa_ifa, ntohs(aa->aa_firstnet),
ntohs(aa->aa_lastnet), RTM_ADD);
} else if (ifp->if_flags & IFF_POINTOPOINT) {
struct at_addr rtaddr, rtmask;
bzero(&rtaddr, sizeof(rtaddr));
bzero(&rtmask, sizeof(rtmask));
/* fill in the far end if we know it here XXX */
aa->aa_ifa.ifa_dstaddr = (struct sockaddr *) &aa->aa_dstaddr;
error = aa_addsingleroute(&aa->aa_ifa, &rtaddr, &rtmask);
} else if (ifp->if_flags & IFF_LOOPBACK) {
struct at_addr rtaddr, rtmask;
bzero(&rtaddr, sizeof(rtaddr));
bzero(&rtmask, sizeof(rtmask));
rtaddr.s_net = AA_SAT(aa)->sat_addr.s_net;
rtaddr.s_node = AA_SAT(aa)->sat_addr.s_node;
rtmask.s_net = 0xffff;
/* XXX should not be so.. should be HOST route */
rtmask.s_node = 0x0;
error = aa_addsingleroute(&aa->aa_ifa, &rtaddr, &rtmask);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* set the address of our "check if this addr is ours" routine.
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa->aa_ifa.ifa_claim_addr = aa_claim_addr;
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* of course if we can't add these routes we back out, but it's
* getting risky by now XXX
*/
if (error) {
at_scrub(ifp, aa);
aa->aa_addr = oldaddr;
aa->aa_firstnet = onr.nr_firstnet;
aa->aa_lastnet = onr.nr_lastnet;
return (error);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
* note that the address has a route associated with it....
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
*/
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa->aa_ifa.ifa_flags |= IFA_ROUTE;
aa->aa_flags |= AFA_ROUTE;
return (0);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Massively COMMENT at_control.c and fix some bugs.. also fix a bug in aarp.c that didn't take netranges into account. default routes now work with appletalk, which is a poor-man's way of being able to access netranges if you only have one network :) Hopefully the full netranges fix will happen soon.
1996-08-29 23:16:34 +00:00
/*
* check whether a given address is a broadcast address for us..
*/
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
at_broadcast(struct sockaddr_at *sat)
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
{
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
struct at_ifaddr *aa;
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
Add a global rwlock, at_ifaddr_rw, to protect the global netatalk address lists, at_ifaddr_list. Acquire the lock, and use ifaddr refcounts where necessary, to close most known address-related races in netatalk. Annotate one potential race in at_control() where we acquire an ifaddr reference, drop the global lock, and scrub the address from the ifnet before re-acquiring the global lock, which could allow for a writer-writer race. MFC after: 3 weeks
2009-06-22 10:23:54 +00:00
AT_IFADDR_LOCK_ASSERT();
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
/*
* If the node is not right, it can't be a broadcast
*/
if (sat->sat_addr.s_node != ATADDR_BCAST)
return (0);
/*
* If the node was right then if the net is right, it's a broadcast
*/
if (sat->sat_addr.s_net == ATADDR_ANYNET)
return (1);
/*
* failing that, if the net is one we have, it's a broadcast as well.
*/
Use queue(9) instead of hand-crafted link lists for the global netatalk address list. Generally follow the style and convention of similar parts in netinet. MFC after: 6 weeks
2009-06-24 21:36:09 +00:00
TAILQ_FOREACH(aa, &at_ifaddrhead, aa_link) {
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if ((aa->aa_ifp->if_flags & IFF_BROADCAST)
&& (ntohs(sat->sat_addr.s_net) >= ntohs(aa->aa_firstnet)
&& ntohs(sat->sat_addr.s_net) <= ntohs(aa->aa_lastnet)))
Since I have my hands all over netatalk adding locking and restructuring it, cinch the file's style closer to style(9) with regard to parenthesis: s/( /(/g s/ )/)/g s/return(/return (/g s/return 0/return (0)/ s/return 1/return (1)/
2004-03-22 03:24:10 +00:00
return (1);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
return (0);
Obtained from: netatalk distribution netatalk@itd.umich.edu Kernel Appletalk protocol support both CAP and netatalk can make use of this.. still needs some owrk but it seemd the right tiime to commit it so other can experiment.
1996-05-24 01:35:45 +00:00
}
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
/*
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
* aa_dorangeroute()
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
*
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
* Add a route for a range of networks from bot to top - 1.
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
* Algorithm:
*
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
* Split the range into two subranges such that the middle
* of the two ranges is the point where the highest bit of difference
Seventy-odd "its" / "it's" typos in comments fixed as per kern/6108.
1998-04-17 22:37:19 +00:00
* between the two addresses makes its transition.
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
* Each of the upper and lower ranges might not exist, or might be
* representable by 1 or more netmasks. In addition, if both
Typo fix: teh --> (the|they)
1998-07-30 02:27:41 +00:00
* ranges can be represented by the same netmask, then they can be merged
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
* by using the next higher netmask..
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
*/
static int
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
aa_dorangeroute(struct ifaddr *ifa, u_int bot, u_int top, int cmd)
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
{
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
u_int mask1;
struct at_addr addr;
struct at_addr mask;
int error;
/*
* slight sanity check
*/
if (bot > top) return (EINVAL);
addr.s_node = 0;
mask.s_node = 0;
/*
* just start out with the lowest boundary
* and keep extending the mask till it's too big.
*/
while (bot <= top) {
mask1 = 1;
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
while (((bot & ~mask1) >= bot) && ((bot | mask1) <= top)) {
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
mask1 <<= 1;
mask1 |= 1;
}
mask1 >>= 1;
mask.s_net = htons(~mask1);
addr.s_net = htons(bot);
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
if (cmd == RTM_ADD) {
error = aa_addsingleroute(ifa,&addr,&mask);
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
if (error) {
/* XXX clean up? */
return (error);
}
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
} else
wow we can even make net ranges go away now (how sophisticated!)
1996-08-31 08:56:36 +00:00
error = aa_delsingleroute(ifa,&addr,&mask);
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
bot = (bot | mask1) + 1;
}
Since I have my hands all over netatalk adding locking and restructuring it, cinch the file's style closer to style(9) with regard to parenthesis: s/( /(/g s/ )/)/g s/return(/return (/g s/return 0/return (0)/ s/return 1/return (1)/
2004-03-22 03:24:10 +00:00
return (0);
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
}
static int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa_addsingleroute(struct ifaddr *ifa, struct at_addr *addr,
struct at_addr *mask)
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
{
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
#if 0
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
printf("aa_addsingleroute: %x.%x mask %x.%x ...\n",
ntohs(addr->s_net), addr->s_node, ntohs(mask->s_net),
mask->s_node);
Add code to automaticall support subnets on ethertalk networks Subnets are represented in the routing table as a set of binary routing nets using the standard netmask algorythm. The code produces the minimum possible set of standard netmasks and net addresses to be able to represent a given netrange.
1996-08-31 06:18:27 +00:00
#endif
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
Printf fewer warnings when adding a route to an atalk address fails; userspace will print the error. MFC after: 3 days
2009-06-24 10:46:03 +00:00
return (aa_dosingleroute(ifa, addr, mask, RTM_ADD, RTF_UP));
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
}
static int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa_delsingleroute(struct ifaddr *ifa, struct at_addr *addr,
struct at_addr *mask)
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
{
Printf fewer warnings when adding a route to an atalk address fails; userspace will print the error. MFC after: 3 days
2009-06-24 10:46:03 +00:00
return (aa_dosingleroute(ifa, addr, mask, RTM_DELETE, 0));
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
}
static int
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
aa_dosingleroute(struct ifaddr *ifa, struct at_addr *at_addr,
struct at_addr *at_mask, int cmd, int flags)
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
{
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
struct sockaddr_at addr, mask;
bzero(&addr, sizeof(addr));
bzero(&mask, sizeof(mask));
addr.sat_family = AF_APPLETALK;
addr.sat_len = sizeof(struct sockaddr_at);
addr.sat_addr.s_net = at_addr->s_net;
addr.sat_addr.s_node = at_addr->s_node;
mask.sat_family = AF_APPLETALK;
mask.sat_len = sizeof(struct sockaddr_at);
mask.sat_addr.s_net = at_mask->s_net;
mask.sat_addr.s_node = at_mask->s_node;
if (at_mask->s_node)
flags |= RTF_HOST;
return (rtrequest(cmd, (struct sockaddr *) &addr,
(flags & RTF_HOST)?(ifa->ifa_dstaddr):(ifa->ifa_addr),
(struct sockaddr *) &mask, flags, NULL));
Submitted by: archie@whistle.com appletalk cleanups
1996-07-23 01:18:47 +00:00
}
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
static int
aa_claim_addr(struct ifaddr *ifa, struct sockaddr *gw0)
{
struct sockaddr_at *addr = (struct sockaddr_at *)ifa->ifa_addr;
struct sockaddr_at *gw = (struct sockaddr_at *)gw0;
switch (gw->sat_range.r_netrange.nr_phase) {
case 1:
if(addr->sat_range.r_netrange.nr_phase == 1)
Since I have my hands all over netatalk adding locking and restructuring it, cinch the file's style closer to style(9) with regard to parenthesis: s/( /(/g s/ )/)/g s/return(/return (/g s/return 0/return (0)/ s/return 1/return (1)/
2004-03-22 03:24:10 +00:00
return (1);
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
case 0:
case 2:
/*
* if it's our net (including 0),
* or netranges are valid, and we are in the range,
* then it's ours.
*/
if ((addr->sat_addr.s_net == gw->sat_addr.s_net)
Re-style at_control.c to bring it closer to style(9), primarily with regard to function prototypes and indentation. The lack of indentation in if clauses and case statements made this code extremely difficult to read.
2004-07-19 17:15:51 +00:00
|| ((addr->sat_range.r_netrange.nr_lastnet)
&& (ntohs(gw->sat_addr.s_net) >=
ntohs(addr->sat_range.r_netrange.nr_firstnet))
&& (ntohs(gw->sat_addr.s_net) <=
ntohs(addr->sat_range.r_netrange.nr_lastnet))))
Since I have my hands all over netatalk adding locking and restructuring it, cinch the file's style closer to style(9) with regard to parenthesis: s/( /(/g s/ )/)/g s/return(/return (/g s/return 0/return (0)/ s/return 1/return (1)/
2004-03-22 03:24:10 +00:00
return (1);
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
break;
default:
printf("atalk: bad phase\n");
}
Since I have my hands all over netatalk adding locking and restructuring it, cinch the file's style closer to style(9) with regard to parenthesis: s/( /(/g s/ )/)/g s/return(/return (/g s/return 0/return (0)/ s/return 1/return (1)/
2004-03-22 03:24:10 +00:00
return (0);
Add a per-interface-address pointer to a function that can be supplied by a protocol, to detirmine if an address matches the net this address is part of. This is needed by protocols for which netmasks "just don't work", for example appletalk. Also add the code in appletalk to make use of this new feature. Thsi fixes one of the longest standing bugs in appletalk. The inability to talk to machines to which the path is via a router which is on a different net, but the same netrange, as your interface. Protocols that do not supply this function (e.g. IP) should not be affected.
1997-08-28 01:17:12 +00:00
}
Reference in New Issue Copy Permalink