d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

protosw: cleanup protocols that existed merely to provide pr_input

Browse Source 
Since 4.4BSD the protosw was used to implement socket types created
by socket(2) syscall and at the same to demultiplex incoming IPv4
datagrams (later copied to IPv6).  This story ended with 78b1fc05b2.

These entries (e.g. IPPROTO_ICMP) in inetsw that were added to catch
packets in ip_input(), they would also be returned by pffindproto()
if user says socket(AF_INET, SOCK_RAW, IPPROTO_ICMP).  Thus, for raw
sockets to work correctly, all the entries were pointing at raw_usrreq
differentiating only in the value of pr_protocol.

With 78b1fc05b2 all these entries are no longer needed, as ip_protox
is independent of protosw.  Any socket syscall requesting SOCK_RAW type
would end up with rip_protosw.  And this protosw has its pr_protocol
set to 0, allowing to mark socket with any protocol.

For IPv6 raw socket the change required two small fixes:
o Validate user provided protocol value
o Always use protocol number stored in inp in rip6_attach, instead
  of protosw value, which is now always 0.

Differential revision:	https://reviews.freebsd.org/D36380
This commit is contained in:
Gleb Smirnoff 2022-08-30 15:09:21 -07:00
parent 8624f4347e
commit 61f7427f02
7 changed files with 49 additions and 219 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
sys/kern/uipc_domain.c
View File

@ -306,32 +306,6 @@ pffindtype(int family, int type)
return (NULL);
}
struct protosw *
pffindproto(int family, int protocol, int type)
{
struct domain *dp;
struct protosw *pr;
struct protosw *maybe;
dp = pffinddomain(family);
if (dp == NULL)
return (NULL);
maybe = NULL;
for (int i = 0; i < dp->dom_nprotosw; i++) {
if ((pr = dp->dom_protosw[i]) == NULL)
continue;
if ((pr->pr_protocol == protocol) && (pr->pr_type == type))
return (pr);
/* XXX: raw catches all. Why? */
if (type == SOCK_RAW && pr->pr_type == SOCK_RAW &&
pr->pr_protocol == 0 && maybe == NULL)
maybe = pr;
}
return (maybe);
}
/*
* The caller must make sure that the new protocol is fully set up and ready to
* accept requests before it is registered.

10
sys/kern/uipc_socket.c
View File

@ -520,11 +520,7 @@ socreate(int dom, struct socket **aso, int type, int proto,
td->td_proc->p_comm);
}
if (proto)
prp = pffindproto(dom, proto, type);
else
prp = pffindtype(dom, type);
prp = pffindtype(dom, type);
if (prp == NULL) {
/* No support for domain. */
if (pffinddomain(dom) == NULL)
@ -534,6 +530,8 @@ socreate(int dom, struct socket **aso, int type, int proto,
return (EPROTOTYPE);
return (EPROTONOSUPPORT);
}
if (prp->pr_protocol != 0 && proto != 0 && prp->pr_protocol != proto)
return (EPROTONOSUPPORT);
MPASS(prp->pr_attach);
@ -543,8 +541,6 @@ socreate(int dom, struct socket **aso, int type, int proto,
if (prison_check_af(cred, prp->pr_domain->dom_family) != 0)
return (EPROTONOSUPPORT);
if (prp->pr_type != type)
return (EPROTOTYPE);
so = soalloc(CRED_TO_VNET(cred));
if (so == NULL)
return (ENOBUFS);

26
sys/netinet/in_proto.c
View File

@ -94,9 +94,7 @@ __FBSDID("$FreeBSD$");
#endif
/* netinet/raw_ip.c */
extern struct protosw rip_protosw, rsvp_protosw, rawipv4_protosw,
rawipv6_protosw, mobile_protosw, etherip_protosw, icmp_protosw,
igmp_protosw, gre_protosw, pim_protosw, ripwild_protosw;
extern struct protosw rip_protosw;
/* netinet/udp_usrreq.c */
extern struct protosw udp_protosw, udplite_protosw;
@ -111,7 +109,7 @@ struct domain inetdomain = {
#endif
.dom_ifattach = in_domifattach,
.dom_ifdetach = in_domifdetach,
.dom_nprotosw = 24,
.dom_nprotosw = 14,
.dom_protosw = {
&tcp_protosw,
&udp_protosw,
@ -123,28 +121,8 @@ struct domain inetdomain = {
#endif
&udplite_protosw,
&rip_protosw,
/*
* XXXGL: it is entirely possible that all below raw-based
* protosw definitions are not needed. They could have existed
* just to define pr_input, pr_drain, pr_*timo or PR_LASTHDR
* flag, and were never supposed to create a special socket.
*/
&icmp_protosw,
&igmp_protosw,
&rsvp_protosw,
&rawipv4_protosw,
&mobile_protosw,
&etherip_protosw,
&gre_protosw,
#ifdef INET6
&rawipv6_protosw,
#else
NULL,
#endif
&pim_protosw,
/* Spacer 8 times for loadable protocols. XXXGL: why 8? */
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
&ripwild_protosw,
},
};

80
sys/netinet/raw_ip.c
View File

@ -1068,70 +1068,22 @@ SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist,
"List of active raw IP sockets");
#ifdef INET
/*
* See comment in in_proto.c containing "protosw definitions are not needed".
*/
#define RAW_PROTOSW \
.pr_type = SOCK_RAW, \
.pr_flags = PR_ATOMIC|PR_ADDR, \
.pr_ctloutput = rip_ctloutput, \
.pr_abort = rip_abort, \
.pr_attach = rip_attach, \
.pr_bind = rip_bind, \
.pr_connect = rip_connect, \
.pr_control = in_control, \
.pr_detach = rip_detach, \
.pr_disconnect = rip_disconnect, \
.pr_peeraddr = in_getpeeraddr, \
.pr_send = rip_send, \
.pr_shutdown = rip_shutdown, \
.pr_sockaddr = in_getsockaddr, \
.pr_sosetlabel = in_pcbsosetlabel, \
.pr_close = rip_close
struct protosw rip_protosw = {
.pr_protocol = IPPROTO_RAW,
RAW_PROTOSW
};
struct protosw icmp_protosw = {
.pr_protocol = IPPROTO_ICMP,
RAW_PROTOSW
};
struct protosw igmp_protosw = {
.pr_protocol = IPPROTO_IGMP,
RAW_PROTOSW
};
struct protosw rsvp_protosw = {
.pr_protocol = IPPROTO_RSVP,
RAW_PROTOSW
};
struct protosw rawipv4_protosw = {
.pr_protocol = IPPROTO_IPV4,
RAW_PROTOSW
};
struct protosw mobile_protosw = {
.pr_protocol = IPPROTO_MOBILE,
RAW_PROTOSW
};
struct protosw etherip_protosw = {
.pr_protocol = IPPROTO_ETHERIP,
RAW_PROTOSW
};
struct protosw gre_protosw = {
.pr_protocol = IPPROTO_GRE,
RAW_PROTOSW
};
#ifdef INET6
struct protosw rawipv6_protosw = {
.pr_protocol = IPPROTO_IPV6,
RAW_PROTOSW
};
#endif
struct protosw pim_protosw = {
.pr_protocol = IPPROTO_PIM,
RAW_PROTOSW
};
struct protosw ripwild_protosw = {
RAW_PROTOSW
.pr_type = SOCK_RAW,
.pr_flags = PR_ATOMIC|PR_ADDR,
.pr_ctloutput = rip_ctloutput,
.pr_abort = rip_abort,
.pr_attach = rip_attach,
.pr_bind = rip_bind,
.pr_connect = rip_connect,
.pr_control = in_control,
.pr_detach = rip_detach,
.pr_disconnect = rip_disconnect,
.pr_peeraddr = in_getpeeraddr,
.pr_send = rip_send,
.pr_shutdown = rip_shutdown,
.pr_sockaddr = in_getsockaddr,
.pr_sosetlabel = in_pcbsosetlabel,
.pr_close = rip_close
};
#endif /* INET */

26
sys/netinet6/in6_proto.c
View File

@ -122,9 +122,7 @@ __FBSDID("$FreeBSD$");
#include <netinet6/ip6protosw.h>
/* netinet6/raw_ip6.c */
extern struct protosw rip6_protosw, icmp6_protosw, dstopts6_protosw,
routing6_protosw, frag6_protosw, rawipv4in6_protosw, rawipv6in6_protosw,
etherip6_protosw, gre6_protosw, pim6_protosw, rip6wild_protosw;
extern struct protosw rip6_protosw;
/* netinet6/udp6_usrreq.c */
extern struct protosw udp6_protosw, udplite6_protosw;
@ -143,7 +141,7 @@ struct domain inet6domain = {
.dom_ifattach = in6_domifattach,
.dom_ifdetach = in6_domifdetach,
.dom_ifmtu = in6_domifmtu,
.dom_nprotosw = 24,
.dom_nprotosw = 14,
.dom_protosw = {
&tcp6_protosw,
&udp6_protosw,
@ -155,28 +153,8 @@ struct domain inet6domain = {
#endif
&udplite6_protosw,
&rip6_protosw,
/*
* XXXGL: it is entirely possible that all below raw-based
* protosw definitions are not needed. They could have existed
* just to define pr_input, pr_drain, pr_*timo or PR_LASTHDR
* flag, and were never supposed to create a special socket.
*/
&icmp6_protosw,
&dstopts6_protosw,
&routing6_protosw,
&frag6_protosw,
#ifdef INET
&rawipv4in6_protosw,
#else
NULL,
#endif
&rawipv6in6_protosw,
&etherip6_protosw,
&gre6_protosw,
&pim6_protosw,
/* Spacer 8 times for loadable protocols. XXXGL: why 8? */
NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
&rip6wild_protosw,
},
};

99
sys/netinet6/raw_ip6.c
View File

@ -424,9 +424,8 @@ rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
if (control != NULL) {
NET_EPOCH_ENTER(et);
error = ip6_setpktopts(control, &opt,
inp->in6p_outputopts, so->so_cred,
so->so_proto->pr_protocol);
error = ip6_setpktopts(control, &opt, inp->in6p_outputopts,
so->so_cred, inp->inp_ip_p);
NET_EPOCH_EXIT(et);
if (error != 0) {
@ -455,7 +454,7 @@ rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
* For an ICMPv6 packet, we should know its type and code to update
* statistics.
*/
if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
if (inp->inp_ip_p == IPPROTO_ICMPV6) {
struct icmp6_hdr *icmp6;
if (m->m_len < sizeof(struct icmp6_hdr) &&
(m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
@ -479,8 +478,7 @@ rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
uint32_t hash_type, hash_val;
hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
&dstsock->sin6_addr, 0, 0, so->so_proto->pr_protocol,
&hash_type);
&dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
inp->inp_flowid = hash_val;
inp->inp_flowtype = hash_type;
}
@ -516,14 +514,13 @@ rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
ip6->ip6_nxt = inp->inp_ip_p;
ip6->ip6_hlim = hlim;
if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
inp->in6p_cksum != -1) {
if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
struct mbuf *n;
int off;
u_int16_t *p;
/* Compute checksum. */
if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
if (inp->inp_ip_p == IPPROTO_ICMPV6)
off = offsetof(struct icmp6_hdr, icmp6_cksum);
else
off = inp->in6p_cksum;
@ -550,7 +547,7 @@ rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
* them to rtadvd/rtsol.
*/
if ((send_sendso_input_hook != NULL) &&
so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
inp->inp_ip_p == IPPROTO_ICMPV6) {
switch (type) {
case ND_ROUTER_ADVERT:
case ND_ROUTER_SOLICIT:
@ -565,7 +562,7 @@ rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
NET_EPOCH_ENTER(et);
error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
NET_EPOCH_EXIT(et);
if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
if (inp->inp_ip_p == IPPROTO_ICMPV6) {
if (oifp)
icmp6_ifoutstat_inc(oifp, type, code);
ICMP6STAT_INC(icp6s_outhist[type]);
@ -682,6 +679,8 @@ rip6_attach(struct socket *so, int proto, struct thread *td)
error = priv_check(td, PRIV_NETINET_RAW);
if (error)
return (error);
if (proto >= IPPROTO_MAX || proto < 0)
return (EPROTONOSUPPORT);
error = soreserve(so, rip_sendspace, rip_recvspace);
if (error)
return (error);
@ -694,7 +693,7 @@ rip6_attach(struct socket *so, int proto, struct thread *td)
return (error);
}
inp = (struct inpcb *)so->so_pcb;
inp->inp_ip_p = (long)proto;
inp->inp_ip_p = proto;
inp->in6p_cksum = -1;
inp->in6p_icmp6filt = filter;
ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
@ -869,66 +868,20 @@ rip6_shutdown(struct socket *so)
return (0);
}
/*
* See comment in in6_proto.c containing "protosw definitions are not needed".
*/
#define RAW6_PROTOSW \
.pr_type = SOCK_RAW, \
.pr_flags = PR_ATOMIC|PR_ADDR, \
.pr_ctloutput = rip6_ctloutput, \
.pr_abort = rip6_abort, \
.pr_attach = rip6_attach, \
.pr_bind = rip6_bind, \
.pr_connect = rip6_connect, \
.pr_control = in6_control, \
.pr_detach = rip6_detach, \
.pr_disconnect = rip6_disconnect, \
.pr_peeraddr = in6_getpeeraddr, \
.pr_send = rip6_send, \
.pr_shutdown = rip6_shutdown, \
.pr_sockaddr = in6_getsockaddr, \
.pr_close = rip6_close
struct protosw rip6_protosw = {
.pr_protocol = IPPROTO_RAW,
RAW6_PROTOSW
};
struct protosw icmp6_protosw = {
.pr_protocol = IPPROTO_ICMPV6,
RAW6_PROTOSW
};
struct protosw dstopts6_protosw = {
.pr_protocol = IPPROTO_DSTOPTS,
RAW6_PROTOSW
};
struct protosw routing6_protosw = {
.pr_protocol = IPPROTO_ROUTING,
RAW6_PROTOSW
};
struct protosw frag6_protosw = {
.pr_protocol = IPPROTO_FRAGMENT,
RAW6_PROTOSW
};
struct protosw rawipv4in6_protosw = {
.pr_protocol = IPPROTO_IPV4,
RAW6_PROTOSW
};
struct protosw rawipv6in6_protosw = {
.pr_protocol = IPPROTO_IPV6,
RAW6_PROTOSW
};
struct protosw etherip6_protosw = {
.pr_protocol = IPPROTO_ETHERIP,
RAW6_PROTOSW
};
struct protosw gre6_protosw = {
.pr_protocol = IPPROTO_GRE,
RAW6_PROTOSW
};
struct protosw pim6_protosw = {
.pr_protocol = IPPROTO_PIM,
RAW6_PROTOSW
};
struct protosw rip6wild_protosw = {
RAW6_PROTOSW
.pr_type = SOCK_RAW,
.pr_flags = PR_ATOMIC|PR_ADDR,
.pr_ctloutput = rip6_ctloutput,
.pr_abort = rip6_abort,
.pr_attach = rip6_attach,
.pr_bind = rip6_bind,
.pr_connect = rip6_connect,
.pr_control = in6_control,
.pr_detach = rip6_detach,
.pr_disconnect = rip6_disconnect,
.pr_peeraddr = in6_getpeeraddr,
.pr_send = rip6_send,
.pr_shutdown = rip6_shutdown,
.pr_sockaddr = in6_getsockaddr,
.pr_close = rip6_close
};

1
sys/sys/protosw.h
View File

@ -236,7 +236,6 @@ char *prcorequests[] = {
#ifdef _KERNEL
struct domain *pffinddomain(int family);
struct protosw *pffindproto(int family, int protocol, int type);
struct protosw *pffindtype(int family, int type);
int protosw_register(struct domain *, struct protosw *);
int protosw_unregister(struct protosw *);