/*
* Define the code needed before returning to user mode, for trap and
* syscall.
*/
void
userret(struct thread *td, struct trapframe *frame)
{
struct proc *p = td->td_proc;
CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid,
td->td_name);
KASSERT((p->p_flag & P_WEXIT) == 0,
("Exiting process returns to usermode"));
#if 0
#ifdef DIAGNOSTIC
/* Check that we called signotify() enough. */
PROC_LOCK(p);
thread_lock(td);
if (SIGPENDING(td) && ((td->td_flags & TDF_NEEDSIGCHK) == 0 ||
(td->td_flags & TDF_ASTPENDING) == 0))
printf("failed to set signal flags properly for ast()\n");
thread_unlock(td);
PROC_UNLOCK(p);
#endif
#endif
#ifdef KTRACE
KTRUSERRET(td);
#endif
if (softdep_ast_cleanup != NULL)
softdep_ast_cleanup();
/*
* If this thread tickled GEOM, we need to wait for the giggling to
* stop before we return to userland
*/
if (td->td_pflags & TDP_GEOM)
g_waitidle();
/*
* Charge system time if profiling.
*/
if (p->p_flag & P_PROFIL)
addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio);
/*
* Let the scheduler adjust our priority etc.
*/
sched_userret(td);
#ifdef XEN
PT_UPDATES_FLUSH();
#endif
/*
* Check for misbehavior.
*
* In case there is a callchain tracing ongoing because of
* hwpmc(4), skip the scheduler pinning check.
* hwpmc(4) subsystem, infact, will collect callchain informations
* at ast() checkpoint, which is past userret().
*/
WITNESS_WARN(WARN_PANIC, NULL, "userret: returning");
KASSERT(td->td_critnest == 0,
("userret: Returning in a critical section"));
KASSERT(td->td_locks == 0,
("userret: Returning with %d locks held", td->td_locks));
KASSERT((td->td_pflags & TDP_NOFAULTING) == 0,
("userret: Returning with pagefaults disabled"));
KASSERT(td->td_no_sleeping == 0,
("userret: Returning with sleep disabled"));
KASSERT(td->td_pinned == 0 || (td->td_pflags & TDP_CALLCHAIN) != 0,
("userret: Returning with with pinned thread"));
KASSERT(td->td_vp_reserv == 0,
("userret: Returning while holding vnode reservation"));
KASSERT((td->td_flags & TDF_SBDRY) == 0,
("userret: Returning with stop signals deferred"));
KASSERT(td->td_su == NULL,
("userret: Returning with SU cleanup request not handled"));
#ifdef VIMAGE
/* Unfortunately td_vnet_lpush needs VNET_DEBUG. */
VNET_ASSERT(curvnet == NULL,
("%s: Returning on td %p (pid %d, %s) with vnet %p set in %s",
__func__, td, p->p_pid, td->td_name, curvnet,
(td->td_vnet_lpush != NULL) ? td->td_vnet_lpush : "N/A"));
#endif
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(p);
while (p->p_throttled == 1)
msleep(p->p_racct, &p->p_mtx, 0, "racct", 0);
PROC_UNLOCK(p);
}
#endif
}
static void
nfssvc_program(struct svc_req *rqst, SVCXPRT *xprt)
{
struct nfsrv_descript nd;
struct nfsrvcache *rp = NULL;
int cacherep, credflavor;
memset(&nd, 0, sizeof(nd));
if (rqst->rq_vers == NFS_VER2) {
if (rqst->rq_proc > NFSV2PROC_STATFS ||
newnfs_nfsv3_procid[rqst->rq_proc] == NFSPROC_NOOP) {
svcerr_noproc(rqst);
svc_freereq(rqst);
goto out;
}
nd.nd_procnum = newnfs_nfsv3_procid[rqst->rq_proc];
nd.nd_flag = ND_NFSV2;
} else if (rqst->rq_vers == NFS_VER3) {
if (rqst->rq_proc >= NFS_V3NPROCS) {
svcerr_noproc(rqst);
svc_freereq(rqst);
goto out;
}
nd.nd_procnum = rqst->rq_proc;
nd.nd_flag = ND_NFSV3;
} else {
if (rqst->rq_proc != NFSPROC_NULL &&
rqst->rq_proc != NFSV4PROC_COMPOUND) {
svcerr_noproc(rqst);
svc_freereq(rqst);
goto out;
}
nd.nd_procnum = rqst->rq_proc;
nd.nd_flag = ND_NFSV4;
}
/*
* Note: we want rq_addr, not svc_getrpccaller for nd_nam2 -
* NFS_SRVMAXDATA uses a NULL value for nd_nam2 to detect TCP
* mounts.
*/
nd.nd_mrep = rqst->rq_args;
rqst->rq_args = NULL;
newnfs_realign(&nd.nd_mrep, M_WAITOK);
nd.nd_md = nd.nd_mrep;
nd.nd_dpos = mtod(nd.nd_md, caddr_t);
nd.nd_nam = svc_getrpccaller(rqst);
nd.nd_nam2 = rqst->rq_addr;
nd.nd_mreq = NULL;
nd.nd_cred = NULL;
if (nfs_privport && (nd.nd_flag & ND_NFSV4) == 0) {
/* Check if source port is privileged */
u_short port;
struct sockaddr *nam = nd.nd_nam;
struct sockaddr_in *sin;
sin = (struct sockaddr_in *)nam;
/*
* INET/INET6 - same code:
* sin_port and sin6_port are at same offset
*/
port = ntohs(sin->sin_port);
if (port >= IPPORT_RESERVED &&
nd.nd_procnum != NFSPROC_NULL) {
#ifdef INET6
char b6[INET6_ADDRSTRLEN];
#if defined(KLD_MODULE)
/* Do not use ip6_sprintf: the nfs module should work without INET6. */
#define ip6_sprintf(buf, a) \
(sprintf((buf), "%x:%x:%x:%x:%x:%x:%x:%x", \
(a)->s6_addr16[0], (a)->s6_addr16[1], \
(a)->s6_addr16[2], (a)->s6_addr16[3], \
(a)->s6_addr16[4], (a)->s6_addr16[5], \
(a)->s6_addr16[6], (a)->s6_addr16[7]), \
(buf))
#endif
#endif
printf("NFS request from unprivileged port (%s:%d)\n",
#ifdef INET6
sin->sin_family == AF_INET6 ?
ip6_sprintf(b6, &satosin6(sin)->sin6_addr) :
#if defined(KLD_MODULE)
#undef ip6_sprintf
#endif
#endif
inet_ntoa(sin->sin_addr), port);
svcerr_weakauth(rqst);
svc_freereq(rqst);
m_freem(nd.nd_mrep);
goto out;
}
}
if (nd.nd_procnum != NFSPROC_NULL) {
if (!svc_getcred(rqst, &nd.nd_cred, &credflavor)) {
svcerr_weakauth(rqst);
svc_freereq(rqst);
m_freem(nd.nd_mrep);
goto out;
}
/* Set the flag based on credflavor */
if (credflavor == RPCSEC_GSS_KRB5) {
nd.nd_flag |= ND_GSS;
} else if (credflavor == RPCSEC_GSS_KRB5I) {
nd.nd_flag |= (ND_GSS | ND_GSSINTEGRITY);
} else if (credflavor == RPCSEC_GSS_KRB5P) {
nd.nd_flag |= (ND_GSS | ND_GSSPRIVACY);
} else if (credflavor != AUTH_SYS) {
svcerr_weakauth(rqst);
svc_freereq(rqst);
m_freem(nd.nd_mrep);
goto out;
}
#ifdef MAC
mac_cred_associate_nfsd(nd.nd_cred);
#endif
/*
* Get a refcnt (shared lock) on nfsd_suspend_lock.
* NFSSVC_SUSPENDNFSD will take an exclusive lock on
* nfsd_suspend_lock to suspend these threads.
* This must be done here, before the check of
* nfsv4root exports by nfsvno_v4rootexport().
*/
NFSLOCKV4ROOTMUTEX();
nfsv4_getref(&nfsd_suspend_lock, NULL, NFSV4ROOTLOCKMUTEXPTR,
NULL);
NFSUNLOCKV4ROOTMUTEX();
if ((nd.nd_flag & ND_NFSV4) != 0) {
nd.nd_repstat = nfsvno_v4rootexport(&nd);
if (nd.nd_repstat != 0) {
NFSLOCKV4ROOTMUTEX();
nfsv4_relref(&nfsd_suspend_lock);
NFSUNLOCKV4ROOTMUTEX();
svcerr_weakauth(rqst);
svc_freereq(rqst);
m_freem(nd.nd_mrep);
goto out;
}
}
cacherep = nfs_proc(&nd, rqst->rq_xid, xprt, &rp);
NFSLOCKV4ROOTMUTEX();
nfsv4_relref(&nfsd_suspend_lock);
NFSUNLOCKV4ROOTMUTEX();
} else {
NFSMGET(nd.nd_mreq);
nd.nd_mreq->m_len = 0;
cacherep = RC_REPLY;
}
if (nd.nd_mrep != NULL)
m_freem(nd.nd_mrep);
if (nd.nd_cred != NULL)
crfree(nd.nd_cred);
if (cacherep == RC_DROPIT) {
if (nd.nd_mreq != NULL)
m_freem(nd.nd_mreq);
svc_freereq(rqst);
goto out;
}
if (nd.nd_mreq == NULL) {
svcerr_decode(rqst);
svc_freereq(rqst);
goto out;
}
if (nd.nd_repstat & NFSERR_AUTHERR) {
svcerr_auth(rqst, nd.nd_repstat & ~NFSERR_AUTHERR);
if (nd.nd_mreq != NULL)
m_freem(nd.nd_mreq);
} else if (!svc_sendreply_mbuf(rqst, nd.nd_mreq)) {
svcerr_systemerr(rqst);
}
if (rp != NULL) {
nfsrvd_sentcache(rp, (rqst->rq_reply_seq != 0 ||
SVC_ACK(xprt, NULL)), rqst->rq_reply_seq);
}
svc_freereq(rqst);
out:
if (softdep_ast_cleanup != NULL)
softdep_ast_cleanup();
NFSEXITCODE(0);
}
