Diff below changes the type of `so_state' and `so_error' to int. It
also introduces two new wrappers to set & clear state bits atomically.
My goal is to prevent a second CPU from reading garbage when looking
at these two fields in filt_soread() and filt_sowrite(). The values
might be outdated though.
This should be the last requirement before unlocking the receiving path.
ok?
Index: kern/uipc_socket.c
===================================================================
RCS file: /cvs/src/sys/kern/uipc_socket.c,v
retrieving revision 1.213
diff -u -p -r1.213 uipc_socket.c
--- kern/uipc_socket.c 2 Jan 2018 12:54:07 -0000 1.213
+++ kern/uipc_socket.c 8 Jan 2018 14:32:33 -0000
@@ -139,7 +139,7 @@ socreate(int dom, struct socket **aso, i
s = solock(so);
error = (*prp->pr_attach)(so, proto);
if (error) {
- so->so_state |= SS_NOFDREF;
+ sosetstate(so, SS_NOFDREF);
sofree(so);
sounlock(s);
return (error);
@@ -275,7 +275,7 @@ drop:
discard:
if (so->so_state & SS_NOFDREF)
panic("soclose NOFDREF: so %p, so_type %d", so, so->so_type);
- so->so_state |= SS_NOFDREF;
+ sosetstate(so, SS_NOFDREF);
sofree(so);
sounlock(s);
return (error);
@@ -299,7 +299,7 @@ soaccept(struct socket *so, struct mbuf
if ((so->so_state & SS_NOFDREF) == 0)
panic("soaccept !NOFDREF: so %p, so_type %d", so, so->so_type);
- so->so_state &= ~SS_NOFDREF;
+ soclrstate(so, SS_NOFDREF);
if ((so->so_state & SS_ISDISCONNECTED) == 0 ||
(so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0)
error = (*so->so_proto->pr_usrreq)(so, PRU_ACCEPT, NULL,
@@ -425,7 +425,7 @@ sosend(struct socket *so, struct mbuf *a
restart:
if ((error = sblock(so, &so->so_snd, SBLOCKWAIT(flags))) != 0)
goto out;
- so->so_state |= SS_ISSENDING;
+ sosetstate(so, SS_ISSENDING);
do {
if (so->so_state & SS_CANTSENDMORE)
snderr(EPIPE);
@@ -455,7 +455,7 @@ restart:
snderr(EWOULDBLOCK);
sbunlock(so, &so->so_snd);
error = sbwait(so, &so->so_snd);
- so->so_state &= ~SS_ISSENDING;
+ soclrstate(so, SS_ISSENDING);
if (error)
goto out;
goto restart;
@@ -481,7 +481,7 @@ restart:
top->m_flags |= M_EOR;
}
if (resid == 0)
- so->so_state &= ~SS_ISSENDING;
+ soclrstate(so, SS_ISSENDING);
if (top && so->so_options & SO_ZEROIZE)
top->m_flags |= M_ZEROIZE;
error = (*so->so_proto->pr_usrreq)(so,
@@ -496,7 +496,7 @@ restart:
} while (resid);
release:
- so->so_state &= ~SS_ISSENDING;
+ soclrstate(so, SS_ISSENDING);
sbunlock(so, &so->so_snd);
out:
sounlock(s);
@@ -857,7 +857,7 @@ dontblock:
panic("receive 3: so %p, so_type %d, m %p, m_type %d",
so, so->so_type, m, m->m_type);
#endif
- so->so_state &= ~SS_RCVATMARK;
+ soclrstate(so, SS_RCVATMARK);
len = uio->uio_resid;
if (so->so_oobmark && len > so->so_oobmark - offset)
len = so->so_oobmark - offset;
@@ -932,7 +932,7 @@ dontblock:
if ((flags & MSG_PEEK) == 0) {
so->so_oobmark -= len;
if (so->so_oobmark == 0) {
- so->so_state |= SS_RCVATMARK;
+ sosetstate(so, SS_RCVATMARK);
break;
}
} else {
@@ -1292,7 +1292,7 @@ somove(struct socket *so, int wait)
goto release;
len = space;
}
- sosp->so_state |= SS_ISSENDING;
+ sosetstate(sosp, SS_ISSENDING);
SBLASTRECORDCHK(&so->so_rcv, "somove 1");
SBLASTMBUFCHK(&so->so_rcv, "somove 1");
@@ -1422,12 +1422,12 @@ somove(struct socket *so, int wait)
/* Receive buffer did shrink by len bytes, adjust oob. */
state = so->so_state;
- so->so_state &= ~SS_RCVATMARK;
+ soclrstate(so, SS_RCVATMARK);
oobmark = so->so_oobmark;
so->so_oobmark = oobmark > len ? oobmark - len : 0;
if (oobmark) {
if (oobmark == len)
- so->so_state |= SS_RCVATMARK;
+ sosetstate(so, SS_RCVATMARK);
if (oobmark >= len)
oobmark = 0;
}
@@ -1485,7 +1485,7 @@ somove(struct socket *so, int wait)
/* Append all remaining data to drain socket. */
if (so->so_rcv.sb_cc == 0 || maxreached)
- sosp->so_state &= ~SS_ISSENDING;
+ soclrstate(sosp, SS_ISSENDING);
error = (*sosp->so_proto->pr_usrreq)(sosp, PRU_SEND, m, NULL, NULL,
NULL);
if (error) {
@@ -1500,7 +1500,7 @@ somove(struct socket *so, int wait)
goto nextpkt;
release:
- sosp->so_state &= ~SS_ISSENDING;
+ soclrstate(sosp, SS_ISSENDING);
if (!error && maxreached && so->so_splicemax == so->so_splicelen)
error = EFBIG;
if (error)
Index: kern/sys_socket.c
===================================================================
RCS file: /cvs/src/sys/kern/sys_socket.c,v
retrieving revision 1.35
diff -u -p -r1.35 sys_socket.c
--- kern/sys_socket.c 10 Dec 2017 11:31:54 -0000 1.35
+++ kern/sys_socket.c 8 Jan 2018 14:25:21 -0000
@@ -80,20 +80,20 @@ soo_ioctl(struct file *fp, u_long cmd, c
case FIONBIO:
s = solock(so);
if (*(int *)data)
- so->so_state |= SS_NBIO;
+ sosetstate(so, SS_NBIO);
else
- so->so_state &= ~SS_NBIO;
+ soclrstate(so, SS_NBIO);
sounlock(s);
break;
case FIOASYNC:
s = solock(so);
if (*(int *)data) {
- so->so_state |= SS_ASYNC;
+ sosetstate(so, SS_ASYNC);
so->so_rcv.sb_flags |= SB_ASYNC;
so->so_snd.sb_flags |= SB_ASYNC;
} else {
- so->so_state &= ~SS_ASYNC;
+ soclrstate(so, SS_ASYNC);
so->so_rcv.sb_flags &= ~SB_ASYNC;
so->so_snd.sb_flags &= ~SB_ASYNC;
}
Index: kern/uipc_usrreq.c
===================================================================
RCS file: /cvs/src/sys/kern/uipc_usrreq.c,v
retrieving revision 1.123
diff -u -p -r1.123 uipc_usrreq.c
--- kern/uipc_usrreq.c 4 Jan 2018 10:45:30 -0000 1.123
+++ kern/uipc_usrreq.c 8 Jan 2018 14:30:45 -0000
@@ -575,7 +575,7 @@ unp_disconnect(struct unpcb *unp)
case SOCK_DGRAM:
SLIST_REMOVE(&unp2->unp_refs, unp, unpcb, unp_nextref);
- unp->unp_socket->so_state &= ~SS_ISCONNECTED;
+ soclrstate(unp->unp_socket, SS_ISCONNECTED);
break;
case SOCK_STREAM:
Index: kern/uipc_socket2.c
===================================================================
RCS file: /cvs/src/sys/kern/uipc_socket2.c,v
retrieving revision 1.89
diff -u -p -r1.89 uipc_socket2.c
--- kern/uipc_socket2.c 30 Dec 2017 20:47:00 -0000 1.89
+++ kern/uipc_socket2.c 8 Jan 2018 14:30:01 -0000
@@ -87,8 +87,8 @@ void
soisconnecting(struct socket *so)
{
soassertlocked(so);
- so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING);
- so->so_state |= SS_ISCONNECTING;
+ soclrstate(so, SS_ISCONNECTED|SS_ISDISCONNECTING);
+ sosetstate(so, SS_ISCONNECTING);
}
void
@@ -97,8 +97,8 @@ soisconnected(struct socket *so)
struct socket *head = so->so_head;
soassertlocked(so);
- so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING);
- so->so_state |= SS_ISCONNECTED;
+ soclrstate(so, SS_ISCONNECTING|SS_ISDISCONNECTING);
+ sosetstate(so, SS_ISCONNECTED);
if (head && soqremque(so, 0)) {
soqinsque(head, so, 1);
sorwakeup(head);
@@ -114,8 +114,8 @@ void
soisdisconnecting(struct socket *so)
{
soassertlocked(so);
- so->so_state &= ~SS_ISCONNECTING;
- so->so_state |= (SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE);
+ soclrstate(so, SS_ISCONNECTING);
+ sosetstate(so, SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE);
wakeup(&so->so_timeo);
sowwakeup(so);
sorwakeup(so);
@@ -125,8 +125,8 @@ void
soisdisconnected(struct socket *so)
{
soassertlocked(so);
- so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
- so->so_state |= (SS_CANTRCVMORE|SS_CANTSENDMORE|SS_ISDISCONNECTED);
+ soclrstate(so, SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
+ sosetstate(so, SS_CANTRCVMORE|SS_CANTSENDMORE|SS_ISDISCONNECTED);
wakeup(&so->so_timeo);
sowwakeup(so);
sorwakeup(so);
@@ -198,7 +198,7 @@ sonewconn(struct socket *head, int conns
if (connstatus) {
sorwakeup(head);
wakeup(&head->so_timeo);
- so->so_state |= connstatus;
+ sosetstate(so, connstatus);
}
return (so);
}
@@ -260,7 +260,7 @@ void
socantsendmore(struct socket *so)
{
soassertlocked(so);
- so->so_state |= SS_CANTSENDMORE;
+ sosetstate(so, SS_CANTSENDMORE);
sowwakeup(so);
}
@@ -268,7 +268,7 @@ void
socantrcvmore(struct socket *so)
{
soassertlocked(so);
- so->so_state |= SS_CANTRCVMORE;
+ sosetstate(so, SS_CANTRCVMORE);
sorwakeup(so);
}
Index: kern/uipc_syscalls.c
===================================================================
RCS file: /cvs/src/sys/kern/uipc_syscalls.c,v
retrieving revision 1.161
diff -u -p -r1.161 uipc_syscalls.c
--- kern/uipc_syscalls.c 2 Jan 2018 06:38:45 -0000 1.161
+++ kern/uipc_syscalls.c 8 Jan 2018 14:29:48 -0000
@@ -112,8 +112,8 @@ sys_socket(struct proc *p, void *v, regi
fdpunlock(fdp);
} else {
if (type & SOCK_NONBLOCK)
- so->so_state |= SS_NBIO;
- so->so_state |= ss;
+ sosetstate(so, SS_NBIO);
+ sosetstate(so, ss);
fp->f_data = so;
FILE_SET_MATURE(fp, p);
*retval = fd;
@@ -340,9 +340,9 @@ doaccept(struct proc *p, int sock, struc
error = copyaddrout(p, nam, name, namelen, anamelen);
if (!error) {
if (nflag & FNONBLOCK)
- so->so_state |= SS_NBIO;
+ sosetstate(so, SS_NBIO);
else
- so->so_state &= ~SS_NBIO;
+ soclrstate(so, SS_NBIO);
fp->f_data = so;
FILE_SET_MATURE(fp, p);
*retval = tmpfd;
@@ -424,7 +424,7 @@ sys_connect(struct proc *p, void *v, reg
}
bad:
if (!interrupted)
- so->so_state &= ~SS_ISCONNECTING;
+ soclrstate(so, SS_ISCONNECTING);
out:
sounlock(s);
FRELE(fp, p);
Index: miscfs/fifofs/fifo_vnops.c
===================================================================
RCS file: /cvs/src/sys/miscfs/fifofs/fifo_vnops.c,v
retrieving revision 1.62
diff -u -p -r1.62 fifo_vnops.c
--- miscfs/fifofs/fifo_vnops.c 2 Jan 2018 06:38:45 -0000 1.62
+++ miscfs/fifofs/fifo_vnops.c 8 Jan 2018 14:31:49 -0000
@@ -158,7 +158,7 @@ fifo_open(void *v)
return (error);
}
fip->fi_readers = fip->fi_writers = 0;
- wso->so_state |= SS_CANTSENDMORE;
+ sosetstate(wso, SS_CANTSENDMORE);
wso->so_snd.sb_lowat = PIPE_BUF;
} else {
rso = fip->fi_readsock;
@@ -168,7 +168,7 @@ fifo_open(void *v)
if (ap->a_mode & FREAD) {
fip->fi_readers++;
if (fip->fi_readers == 1) {
- wso->so_state &= ~SS_CANTSENDMORE;
+ soclrstate(wso, SS_CANTSENDMORE);
if (fip->fi_writers > 0)
wakeup(&fip->fi_writers);
}
@@ -181,7 +181,7 @@ fifo_open(void *v)
goto bad;
}
if (fip->fi_writers == 1) {
- rso->so_state &= ~(SS_CANTRCVMORE|SS_ISDISCONNECTED);
+ soclrstate(rso, (SS_CANTRCVMORE|SS_ISDISCONNECTED));
if (fip->fi_readers > 0)
wakeup(&fip->fi_readers);
}
@@ -231,12 +231,12 @@ fifo_read(void *v)
if (uio->uio_resid == 0)
return (0);
if (ap->a_ioflag & IO_NDELAY)
- rso->so_state |= SS_NBIO;
+ sosetstate(rso, SS_NBIO);
VOP_UNLOCK(ap->a_vp, p);
error = soreceive(rso, NULL, uio, NULL, NULL, NULL, 0);
vn_lock(ap->a_vp, LK_EXCLUSIVE | LK_RETRY, p);
if (ap->a_ioflag & IO_NDELAY) {
- rso->so_state &= ~SS_NBIO;
+ soclrstate(rso, SS_NBIO);
if (error == EWOULDBLOCK &&
ap->a_vp->v_fifoinfo->fi_writers == 0)
error = 0;
@@ -260,14 +260,13 @@ fifo_write(void *v)
if (ap->a_uio->uio_rw != UIO_WRITE)
panic("fifo_write mode");
#endif
- /* XXXSMP changing state w/o lock isn't safe. */
if (ap->a_ioflag & IO_NDELAY)
- wso->so_state |= SS_NBIO;
+ sosetstate(wso, SS_NBIO);
VOP_UNLOCK(ap->a_vp, p);
error = sosend(wso, NULL, ap->a_uio, NULL, NULL, 0);
vn_lock(ap->a_vp, LK_EXCLUSIVE | LK_RETRY, p);
if (ap->a_ioflag & IO_NDELAY)
- wso->so_state &= ~SS_NBIO;
+ soclrstate(wso, SS_NBIO);
return (error);
}
@@ -387,7 +386,7 @@ fifo_close(void *v)
s = solock(rso);
/* SS_ISDISCONNECTED will result in POLLHUP */
- rso->so_state |= SS_ISDISCONNECTED;
+ sosetstate(rso, SS_ISDISCONNECTED);
socantrcvmore(rso);
sounlock(s);
}
Index: netinet/tcp_usrreq.c
===================================================================
RCS file: /cvs/src/sys/netinet/tcp_usrreq.c,v
retrieving revision 1.162
diff -u -p -r1.162 tcp_usrreq.c
--- netinet/tcp_usrreq.c 1 Dec 2017 10:33:33 -0000 1.162
+++ netinet/tcp_usrreq.c 8 Jan 2018 14:32:00 -0000
@@ -246,7 +246,7 @@ tcp_usrreq(struct socket *so, int req, s
break;
}
- so->so_state |= SS_CONNECTOUT;
+ sosetstate(so, SS_CONNECTOUT);
/* Compute window scaling to request. */
tcp_rscale(tp, sb_max);
@@ -585,9 +585,9 @@ tcp_attach(struct socket *so, int proto)
if (tp == NULL) {
int nofd = so->so_state & SS_NOFDREF; /* XXX */
- so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */
+ soclrstate(so, SS_NOFDREF); /* don't free the socket yet */
in_pcbdetach(inp);
- so->so_state |= nofd;
+ sosetstate(so, nofd);
return (ENOBUFS);
}
tp->t_state = TCPS_CLOSED;
Index: netinet/udp_usrreq.c
===================================================================
RCS file: /cvs/src/sys/netinet/udp_usrreq.c,v
retrieving revision 1.245
diff -u -p -r1.245 udp_usrreq.c
--- netinet/udp_usrreq.c 1 Dec 2017 10:33:33 -0000 1.245
+++ netinet/udp_usrreq.c 8 Jan 2018 14:28:40 -0000
@@ -1144,7 +1144,7 @@ udp_usrreq(struct socket *so, int req, s
inp->inp_laddr.s_addr = INADDR_ANY;
in_pcbdisconnect(inp);
- so->so_state &= ~SS_ISCONNECTED; /* XXX */
+ soclrstate(so, SS_ISCONNECTED); /* XXX */
break;
case PRU_SHUTDOWN:
Index: netinet/tcp_input.c
===================================================================
RCS file: /cvs/src/sys/netinet/tcp_input.c,v
retrieving revision 1.354
diff -u -p -r1.354 tcp_input.c
--- netinet/tcp_input.c 4 Dec 2017 13:40:34 -0000 1.354
+++ netinet/tcp_input.c 8 Jan 2018 14:28:45 -0000
@@ -1872,7 +1872,7 @@ step6:
so->so_oobmark = so->so_rcv.sb_cc +
(tp->rcv_up - tp->rcv_nxt) - 1;
if (so->so_oobmark == 0)
- so->so_state |= SS_RCVATMARK;
+ sosetstate(so, SS_RCVATMARK);
sohasoutofband(so);
tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA);
}
Index: netinet6/raw_ip6.c
===================================================================
RCS file: /cvs/src/sys/netinet6/raw_ip6.c,v
retrieving revision 1.125
diff -u -p -r1.125 raw_ip6.c
--- netinet6/raw_ip6.c 4 Dec 2017 13:40:35 -0000 1.125
+++ netinet6/raw_ip6.c 8 Jan 2018 14:28:50 -0000
@@ -560,7 +560,7 @@ rip6_usrreq(struct socket *so, int req,
break;
}
in6p->inp_faddr6 = in6addr_any;
- so->so_state &= ~SS_ISCONNECTED; /* XXX */
+ soclrstate(so, SS_ISCONNECTED); /* XXX */
break;
case PRU_ABORT:
Index: nfs/nfs_socket.c
===================================================================
RCS file: /cvs/src/sys/nfs/nfs_socket.c,v
retrieving revision 1.128
diff -u -p -r1.128 nfs_socket.c
--- nfs/nfs_socket.c 7 Sep 2017 11:35:34 -0000 1.128
+++ nfs/nfs_socket.c 8 Jan 2018 14:28:55 -0000
@@ -319,7 +319,7 @@ nfs_connect(struct nfsmount *nmp, struct
if ((so->so_state & SS_ISCONNECTING) &&
so->so_error == 0 && rep &&
(error = nfs_sigintr(nmp, rep, rep->r_procp)) != 0){
- so->so_state &= ~SS_ISCONNECTING;
+ soclrstate(so, SS_ISCONNECTING);
goto bad;
}
}
Index: sys/socketvar.h
===================================================================
RCS file: /cvs/src/sys/sys/socketvar.h,v
retrieving revision 1.81
diff -u -p -r1.81 socketvar.h
--- sys/socketvar.h 2 Jan 2018 12:54:07 -0000 1.81
+++ sys/socketvar.h 8 Jan 2018 14:20:40 -0000
@@ -51,12 +51,12 @@ TAILQ_HEAD(soqhead, socket);
* private data and error information.
*/
struct socket {
+ const struct protosw *so_proto; /* protocol handle */
+ void *so_pcb; /* protocol control block */
+ u_int so_state; /* internal state flags SS_*, below */
short so_type; /* generic type, see socket.h */
short so_options; /* from socket call, see socket.h */
short so_linger; /* time to linger while closing */
- short so_state; /* internal state flags SS_*, below */
- void *so_pcb; /* protocol control block */
- const struct protosw *so_proto; /* protocol handle */
/*
* Variables for connection queueing.
* Socket where accepts occur is so_head in all subsidiary sockets.
@@ -77,7 +77,7 @@ struct socket {
short so_qlen; /* number of connections on so_q */
short so_qlimit; /* max number queued connections */
short so_timeo; /* connection timeout */
- u_short so_error; /* error affecting connection */
+ u_int so_error; /* error affecting connection */
pid_t so_pgid; /* pgid for signals */
uid_t so_siguid; /* uid of process who set so_pgid */
uid_t so_sigeuid; /* euid of process who set so_pgid */
@@ -156,6 +156,7 @@ struct socket {
#ifdef _KERNEL
+#include <sys/atomic.h>
#include <lib/libkern/libkern.h>
void soassertlocked(struct socket *);
@@ -167,6 +168,18 @@ void soassertlocked(struct socket *);
#define isspliced(so) ((so)->so_sp && (so)->so_sp->ssp_socket)
#define issplicedback(so) ((so)->so_sp && (so)->so_sp->ssp_soback)
+static inline void
+sosetstate(struct socket *so, unsigned int state)
+{
+ atomic_setbits_int(&so->so_state, state);
+}
+
+static inline void
+soclrstate(struct socket *so, unsigned int state)
+{
+ atomic_clearbits_int(&so->so_state, state);
+}
+
/*
* Do we need to notify the other side when I/O is possible?
*/
@@ -190,10 +203,6 @@ static inline long
sbspace(struct socket *so, struct sockbuf *sb)
{
KASSERT(sb == &so->so_rcv || sb == &so->so_snd);
-#if 0
- /* XXXSMP kqueue_scan() calling filt_sowrite() cannot sleep. */
- soassertlocked(so);
-#endif
return lmin(sb->sb_hiwat - sb->sb_cc, sb->sb_mbmax - sb->sb_mbcnt);
}
