gmane.linux.nfs

[patch 1/1] lockd: convert reclaimer thread to kthread interface

2008-12-01T22:27:48

From: Jeff Layton public.gmane.org> My understanding is that there is a push to turn the kernel_thread interface into a non-exported symbol and move all kernel threads to use the kthread API. This patch changes lockd to use kthread_run to spawn the reclaimer thread. I've made the assumption here that the extra module references taken when we spawn this thread are unnecessary and removed them. I've also added a KERN_ERR printk that pops if the thread can't be spawned to warn the admin that the locks won't be reclaimed. In the future, it would be nice to be able to notify userspace that locks have been lost (probably by implementing SIGLOST), and adding some good policies about how long we should reattempt to reclaim the locks. Finally, I removed a comment about memory leaks that I believe is obsolete and added a new one to clarify the result of sending a SIGKILL to the reclaimer thread. As best I can tell, doing so doesn't actually cause a memory leak. Signed-off-by: Jeff Layton public.gmane.org> Cc: Trond Myklebust public.gmane.org> Cc: "J. Bruce Fields" public.gmane.org> Signed-off-by: Andrew Morton public.gmane.org> --- fs/lockd/clntlock.c | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) diff -puN fs/lockd/clntlock.c~lockd-convert-reclaimer-thread-to-kthread-interface fs/lockd/clntlock.c --- a/fs/lockd/clntlock.c~lockd-convert-reclaimer-thread-to-kthread-interface +++ a/fs/lockd/clntlock.c < at >< at > -14,6 +14,7 < at >< at > #include #include #include +#include #define NLMDBG_FACILITYNLMDBG_CLIENT < at >< at > -191,11 +192,15 < at >< at > __be32 nlmclnt_grant(const struct sockad void nlmclnt_recovery(struct nlm_host *host) { +struct task_struct *task; + if (!host->h_reclaiming++) { nlm_get_host(host); -__module_get(THIS_MODULE); -if (kernel_thread(reclaimer, host, CLONE_FS | CLONE_FILES) < 0) -module_put(THIS_MODULE); +task = kthread_run(reclaimer, host, "%s-reclaim", host->h_name); +if (IS_ERR(task)) +printk(KERN_ERR "lockd: unable to spawn reclaimer " +"thread. Locks for %s won't be reclaimed! " +"(%ld)\n", host->h_name, PTR_ERR(task)); } } < at >< at > -207,7 +212,6 < at >< at > reclaimer(void *ptr) struct file_lock *fl, *next; u32 nsmstate; -daemonize("%s-reclaim", host->h_name); allow_signal(SIGKILL); down_write(&host->h_rwsem); < at >< at > -233,7 +237,12 < at >< at > restart: list_for_each_entry_safe(fl, next, &host->h_reclaim, fl_u.nfs_fl.list) { list_del_init(&fl->fl_u.nfs_fl.list); -/* Why are we leaking memory here? --okir */ +/* + * sending this thread a SIGKILL will result in any unreclaimed + * locks being removed from the h_granted list. This means that + * the kernel will not attempt to reclaim them again if a new + * reclaimer thread is spawned for this host. + */ if (signalled()) continue; if (nlmclnt_reclaim(host, fl) != 0) < at >< at > -261,5 +270,5 < at >< at > restart: nlm_release_host(host); lockd_down(); unlock_kernel(); -module_put_and_exit(0); +return 0; } _ -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/3] NFS regression in 2.6.26?, "task blocked for morethan 120 seconds"

Ian Campbell — 2008-12-01T22:09:43

On Sun, 2008-11-30 at 19:17 -0500, Trond Myklebust wrote: Thanks. The server was actually running 2.6.25.7 but the matching sources have since been removed the backports.org so I've reproduce with 2.6.26 and now I'll add the patches. Ian.

[NFS] D entists Contact List for the USA

Strong O Karin — 2008-12-01T22:04:01

Comes with unlimited use license and at a very resonable price: --> 164,574 |>entists --> 158,242 Physical Addresses --> 163,185 Phone Numbers --> 77,878 Office Fax Numbers --> 45,954 e mails only during this week you won't pay the regular $593 price, you get it for $296 Write an email to Monroe-YiFMxoQ3Mhodw2Fo+rOw49BPR1lH4CV8< at >public.gmane.org for additional info to get off please email exclude-YiFMxoQ3Mhodw2Fo+rOw49BPR1lH4CV8< at >public.gmane.org ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ NFS maillist - NFS-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org https://lists.sourceforge.net/lists/listinfo/nfs _______________________________________________ Please note that nfs-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org is being discontinued. Please subscribe to linux-nfs-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org instead. http://vger.kernel.org/vger-lists.html#linux-nfs -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] Support for IPv6 in the showmount command

Steve Dickson — 2008-12-01T19:30:22

Chuck Lever wrote: I believe everything that's needed should be there... let me know if something is missing... steved. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH] gssd: unblock DNOTIFY_SIGNAL in case it was blocked.

Steve Dickson — 2008-12-01T19:29:01

Neil Brown wrote: Committed... steved. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: [PATCH nfs-utils] Ensure statd gets started if required when non-root user mounts an NFS filesystem.

Steve Dickson — 2008-12-01T19:28:19

[PATCH 5/6] NLM: Clean up nsm_monitor() call

Chuck Lever — 2008-12-01T18:58:05

Clean up: A few minor clean-ups for nsm_unmonitor(): o Arrange dprintk and check of sm_monitored consistently with nsm_monitor(). o nsm_unmonitor()'s only caller doesn't care about the return code, so eliminate it (thanks, Bruce). o The nsm_handle's reference count is bumped in nlm_lookup_host() so it should be decremented in nlm_destroy_host() to make it easier to see the balance of these two operations. For the moment we leave nsm_release() public. o Use nsm->sm_name instead of host->h_name to be consistent with other functions in fs/lockd/mon.c. o Collect the public declaration of nsm_unmonitor() in lockd.h with other NSM public function declarations. o Add documenting comments. Signed-off-by: Chuck Lever public.gmane.org> --- fs/lockd/host.c | 7 +++---- fs/lockd/mon.c | 33 +++++++++++++++------------------ include/linux/lockd/lockd.h | 1 + include/linux/lockd/sm_inter.h | 1 - 4 files changed, 19 insertions(+), 23 deletions(-) diff --git a/fs/lockd/host.c b/fs/lockd/host.c index cd7edc3..8d26d0c 100644 --- a/fs/lockd/host.c +++ b/fs/lockd/host.c < at >< at > -263,10 +263,9 < at >< at > nlm_destroy_host(struct nlm_host *host) BUG_ON(!list_empty(&host->h_lockowners)); BUG_ON(atomic_read(&host->h_count)); -/* - * Release NSM handle and unmonitor host. - */ -nsm_unmonitor(host); +nsm_unmonitor(host->h_nsmhandle); +nsm_release(host->h_nsmhandle); +host->h_nsmhandle = NULL; clnt = host->h_rpcclnt; if (clnt != NULL) diff --git a/fs/lockd/mon.c b/fs/lockd/mon.c index 78d5f59..b76d9b2 100644 --- a/fs/lockd/mon.c +++ b/fs/lockd/mon.c < at >< at > -107,33 +107,30 < at >< at > int nsm_monitor(const struct nlm_host *host) return status; } -/* - * Cease to monitor remote host +/** + * nsm_unmonitor - Unregister peer notification + * < at >nsm: pointer to nsm_handle of peer to stop monitoring + * + * If this peer is monitored, this function sends an upcall to + * tell the local rpc.statd not to send this peer a notification + * when we reboot. */ -int -nsm_unmonitor(struct nlm_host *host) +void nsm_unmonitor(struct nsm_handle *nsm) { -struct nsm_handle *nsm = host->h_nsmhandle; -struct nsm_resres; -intstatus = 0; +struct nsm_res res; -if (nsm == NULL) -return 0; -host->h_nsmhandle = NULL; +dprintk("lockd: nsm_unmonitor(%s)\n", nsm->sm_name); -if (atomic_read(&nsm->sm_count) == 1 - && nsm->sm_monitored && !nsm->sm_sticky) { -dprintk("lockd: nsm_unmonitor(%s)\n", host->h_name); +if (!nsm->sm_monitored) +return; -status = nsm_mon_unmon(nsm, SM_UNMON, &res); -if (status < 0) +if (atomic_read(&nsm->sm_count) == 1 && !nsm->sm_sticky) { +if (nsm_mon_unmon(nsm, SM_UNMON, &res) < 0) printk(KERN_NOTICE "lockd: cannot unmonitor %s\n", -host->h_name); +nsm->sm_name); else nsm->sm_monitored = 0; } -nsm_release(nsm); -return status; } /* diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h index 4ca6f39..e2fa968 100644 --- a/include/linux/lockd/lockd.h +++ b/include/linux/lockd/lockd.h < at >< at > -237,6 +237,7 < at >< at > void nsm_release(struct nsm_handle *); * Host monitoring */ int nsm_monitor(const struct nlm_host *host); +void nsm_unmonitor(struct nsm_handle *nsm); /* * This is used in garbage collection and resource reclaim diff --git a/include/linux/lockd/sm_inter.h b/include/linux/lockd/sm_inter.h index 546b610..896a5e3 100644 --- a/include/linux/lockd/sm_inter.h +++ b/include/linux/lockd/sm_inter.h < at >< at > -41,7 +41,6 < at >< at > struct nsm_res { u32state; }; -intnsm_unmonitor(struct nlm_host *); extern intnsm_local_state; #endif /* LINUX_LOCKD_SM_INTER_H */ -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 2/6] NLM: Support IPv6 scope IDs in nlm_display_address()

Chuck Lever — 2008-12-01T18:57:42

Scope ID support is needed since the kernel's NSM implementation is about to use these displayed addresses as a mon_name in some cases. If nsm_use_hostnames is zero, NSM will fail to handle peers that contact us via a link-local address without scope ID support. Link- local addresses do not work without an interface ID, which is stored in the sockaddr's sin6_scope_id field. Signed-off-by: Chuck Lever public.gmane.org> --- fs/lockd/host.c | 31 +++++++++++++++++++++++-------- include/linux/lockd/lockd.h | 2 +- 2 files changed, 24 insertions(+), 9 deletions(-) diff --git a/fs/lockd/host.c b/fs/lockd/host.c index 68e00c0..cd7edc3 100644 --- a/fs/lockd/host.c +++ b/fs/lockd/host.c < at >< at > -105,22 +105,37 < at >< at > static void nlm_clear_port(struct sockaddr *sap) } } -static void nlm_display_address(const struct sockaddr *sap, -char *buf, const size_t len) +static void nlm_display_ipv4_address(const struct sockaddr *sap, char *buf, + const size_t len) { const struct sockaddr_in *sin = (struct sockaddr_in *)sap; +snprintf(buf, len, NIPQUAD_FMT, NIPQUAD(sin->sin_addr.s_addr)); +} + +static void nlm_display_ipv6_address(const struct sockaddr *sap, char *buf, + const size_t len) +{ const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap; +if (ipv6_addr_v4mapped(&sin6->sin6_addr)) +snprintf(buf, len, NIPQUAD_FMT, +NIPQUAD(sin6->sin6_addr.s6_addr32[3])); +else if (sin6->sin6_scope_id != 0) +snprintf(buf, len, NIP6_FMT "%%%u", NIP6(sin6->sin6_addr), +sin6->sin6_scope_id); +else +snprintf(buf, len, NIP6_FMT, NIP6(sin6->sin6_addr)); +} + +static void nlm_display_address(const struct sockaddr *sap, +char *buf, const size_t len) +{ switch (sap->sa_family) { case AF_INET: -snprintf(buf, len, NIPQUAD_FMT, NIPQUAD(sin->sin_addr.s_addr)); +nlm_display_ipv4_address(sap, buf, len); break; case AF_INET6: -if (ipv6_addr_v4mapped(&sin6->sin6_addr)) -snprintf(buf, len, NIPQUAD_FMT, - NIPQUAD(sin6->sin6_addr.s6_addr32[3])); -else -snprintf(buf, len, NIP6_FMT, NIP6(sin6->sin6_addr)); +nlm_display_ipv6_address(sap, buf, len); break; default: snprintf(buf, len, "unsupported address family"); diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h index 4467b83..307a8f0 100644 --- a/include/linux/lockd/lockd.h +++ b/include/linux/lockd/lockd.h < at >< at > -75,7 +75,7 < at >< at > struct nsm_handle { size_tsm_addrlen; unsigned intsm_monitored : 1, sm_sticky : 1;/* don't unmonitor */ -charsm_addrbuf[48];/* address eyecatcher */ +charsm_addrbuf[63];/* presentation address */ }; /* -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 4/6] NLM: Clean up nsm_monitor() call

Chuck Lever — 2008-12-01T18:57:58

Clean up: A few minor clean-ups for nsm_monitor(): o Make sure to return an error if the SM_MON call result is not zero. o Remove the BUG_ON() -- the code will die anyway if nsm is NULL. o Use nsm->sm_name instead of host->h_name to be consistent with other functions in fs/lockd/mon.c. o Collect the public declaration of nsm_monitor() in lockd.h with other NSM public function declarations (eg. nsm_release). o Add documenting comments. Signed-off-by: Chuck Lever public.gmane.org> --- fs/lockd/mon.c | 29 ++++++++++++++++++----------- include/linux/lockd/lockd.h | 4 ++++ include/linux/lockd/sm_inter.h | 1 - 3 files changed, 22 insertions(+), 12 deletions(-) diff --git a/fs/lockd/mon.c b/fs/lockd/mon.c index a606fbb..78d5f59 100644 --- a/fs/lockd/mon.c +++ b/fs/lockd/mon.c < at >< at > -69,18 +69,24 < at >< at > nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res) return status; } -/* - * Set up monitoring of a remote host +/** + * nsm_monitor - Notify a peer in case we reboot + * < at >host: pointer to nlm_host of peer to notify + * + * If this peer is not already monitored, this function sends an + * upcall to the local rpc.statd to record the name/address of + * the peer to notify in case we reboot. + * + * Returns zero if the peer is monitored by the local rpc.statd; + * otherwise a negative errno value is returned. */ -int -nsm_monitor(struct nlm_host *host) +int nsm_monitor(const struct nlm_host *host) { struct nsm_handle *nsm = host->h_nsmhandle; -struct nsm_resres; -intstatus; +struct nsm_res res; +int status; -dprintk("lockd: nsm_monitor(%s)\n", host->h_name); -BUG_ON(nsm == NULL); +dprintk("lockd: nsm_monitor(%s)\n", nsm->sm_name); if (nsm->sm_monitored) return 0; < at >< at > -92,9 +98,10 < at >< at > nsm_monitor(struct nlm_host *host) nsm->sm_mon_name = nsm_use_hostnames ? nsm->sm_name : nsm->sm_addrbuf; status = nsm_mon_unmon(nsm, SM_MON, &res); - -if (status < 0 || res.status != 0) -printk(KERN_NOTICE "lockd: cannot monitor %s\n", host->h_name); +if (res.status != 0) +status = -EIO; +if (status < 0) +printk(KERN_NOTICE "lockd: cannot monitor %s\n", nsm->sm_name); else nsm->sm_monitored = 1; return status; diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h index de9ea7b..4ca6f39 100644 --- a/include/linux/lockd/lockd.h +++ b/include/linux/lockd/lockd.h < at >< at > -233,6 +233,10 < at >< at > extern void nlm_host_rebooted(const struct sockaddr_in *, const char *, unsigned int, u32); void nsm_release(struct nsm_handle *); +/* + * Host monitoring + */ +int nsm_monitor(const struct nlm_host *host); /* * This is used in garbage collection and resource reclaim diff --git a/include/linux/lockd/sm_inter.h b/include/linux/lockd/sm_inter.h index 5a5448b..546b610 100644 --- a/include/linux/lockd/sm_inter.h +++ b/include/linux/lockd/sm_inter.h < at >< at > -41,7 +41,6 < at >< at > struct nsm_res { u32state; }; -intnsm_monitor(struct nlm_host *); intnsm_unmonitor(struct nlm_host *); extern intnsm_local_state; -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 6/6] NSM: Serialize SM_MON and SM_UNMON upcalls

Chuck Lever — 2008-12-01T18:58:13

The Linux NSM implementation currently does nothing to prevent concurrent SM_MON and SM_UNMON upcalls regarding the same peer. Races could cause duplicate SM_MON upcalls for the same peer, or locking activity during server shutdown might cause SM_MON and SM_UNMON upcalls to occur in an arbitrary order. To avoid confusing rpc.statd, add a per-nsm_handle mutex to serialize the upcalls. To prevent duplicate upcalls, the mutex is held while the upcall is outstanding, and is not released until we can tell if the upcall succeeded. The nsm_monitor() function is called frequently. The extra inline function mitigates the overhead of taking the mutex. It is only taken if the nsm_handle is not already monitored. Signed-off-by: Chuck Lever public.gmane.org> --- fs/lockd/host.c | 1 + fs/lockd/mon.c | 23 ++++++++++++++++------- include/linux/lockd/lockd.h | 12 +++++++++++- 3 files changed, 28 insertions(+), 8 deletions(-) diff --git a/fs/lockd/host.c b/fs/lockd/host.c index 8d26d0c..bf705d0 100644 --- a/fs/lockd/host.c +++ b/fs/lockd/host.c < at >< at > -697,6 +697,7 < at >< at > retry: nsm->sm_name = (char *) (nsm + 1); memcpy(nsm->sm_name, hostname, hostname_len); nsm->sm_name[hostname_len] = '\0'; +mutex_init(&nsm->sm_mutex); nlm_display_address((struct sockaddr *)&nsm->sm_addr, nsm->sm_addrbuf, sizeof(nsm->sm_addrbuf)); atomic_set(&nsm->sm_count, 1); diff --git a/fs/lockd/mon.c b/fs/lockd/mon.c index b76d9b2..301aeb7 100644 --- a/fs/lockd/mon.c +++ b/fs/lockd/mon.c < at >< at > -70,8 +70,8 < at >< at > nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res) } /** - * nsm_monitor - Notify a peer in case we reboot - * < at >host: pointer to nlm_host of peer to notify + * __nsm_monitor - Notify a peer in case we reboot + * < at >nsm: pointer to nsm_handle of peer to notify * * If this peer is not already monitored, this function sends an * upcall to the local rpc.statd to record the name/address of < at >< at > -80,16 +80,17 < at >< at > nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res) * Returns zero if the peer is monitored by the local rpc.statd; * otherwise a negative errno value is returned. */ -int nsm_monitor(const struct nlm_host *host) +int __nsm_monitor(struct nsm_handle *nsm) { -struct nsm_handle *nsm = host->h_nsmhandle; struct nsm_res res; -int status; +int status = 0; dprintk("lockd: nsm_monitor(%s)\n", nsm->sm_name); +mutex_lock(&nsm->sm_mutex); + if (nsm->sm_monitored) -return 0; +goto out; /* * Choose whether to record the caller_name or IP address of < at >< at > -104,6 +105,9 < at >< at > int nsm_monitor(const struct nlm_host *host) printk(KERN_NOTICE "lockd: cannot monitor %s\n", nsm->sm_name); else nsm->sm_monitored = 1; + +out: +mutex_unlock(&nsm->sm_mutex); return status; } < at >< at > -121,8 +125,10 < at >< at > void nsm_unmonitor(struct nsm_handle *nsm) dprintk("lockd: nsm_unmonitor(%s)\n", nsm->sm_name); +mutex_lock(&nsm->sm_mutex); + if (!nsm->sm_monitored) -return; +goto out; if (atomic_read(&nsm->sm_count) == 1 && !nsm->sm_sticky) { if (nsm_mon_unmon(nsm, SM_UNMON, &res) < 0) < at >< at > -131,6 +137,9 < at >< at > void nsm_unmonitor(struct nsm_handle *nsm) else nsm->sm_monitored = 0; } + +out: +mutex_unlock(&nsm->sm_mutex); } /* diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h index e2fa968..bf6c1c1 100644 --- a/include/linux/lockd/lockd.h +++ b/include/linux/lockd/lockd.h < at >< at > -76,6 +76,7 < at >< at > struct nsm_handle { char *sm_mon_name; unsigned intsm_monitored : 1, sm_sticky : 1;/* don't unmonitor */ +struct mutexsm_mutex; charsm_addrbuf[63];/* presentation address */ }; < at >< at > -236,9 +237,18 < at >< at > void nsm_release(struct nsm_handle *); /* * Host monitoring */ -int nsm_monitor(const struct nlm_host *host); +int __nsm_monitor(struct nsm_handle *nsm); void nsm_unmonitor(struct nsm_handle *nsm); +static inline int nsm_monitor(const struct nlm_host *host) +{ +struct nsm_handle *nsm = host->h_nsmhandle; + +if (likely(nsm->sm_monitored)) +return 0; +return __nsm_monitor(nsm); +} + /* * This is used in garbage collection and resource reclaim * A return value != 0 means destroy the lock/block/share -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 3/6] NSM: Support IPv6 version of mon_name

Chuck Lever — 2008-12-01T18:57:50

The "mon_name" argument of the NSMPROC_MON and NSMPROC_UNMON upcalls is a string that contains the hostname or IP address of the remote peer to be notified when this host has rebooted. The sm-notify command uses this identifier to contact the peer when we reboot, so it must be either a well-qualified DNS hostname or a presentation format IP address string. When the "nsm_use_hostnames" sysctl is set to zero, the kernel's NSM provides a presentation format IP address in the "mon_name" argument. Otherwise, the "caller_name" argument from NLM requests is used, which is usually just the DNS hostname of the peer. To support IPv6 addresses for the mon_name argument, we use the nsm_handle's address eye-catcher, which already contains an appropriate presentation format address string. Using the eye-catcher string obviates the need to use a large buffer on the stack to form the presentation address string for the upcall. This patch also addresses a subtle bug. An NSMPROC_MON request and the subsequent NSMPROC_UNMON request for the same peer are required to use the same value for the "mon_name" argument. Otherwise, rpc.statd's NSMPROC_UNMON processing cannot locate the database entry for that peer and remove it. If the setting of nsm_use_hostnames is changed between the time the kernel sends an NSMPROC_MON request and the time it sends the NSMPROC_UNMON request for the same peer, the "mon_name" argument for these two requests may not be the same. This is because the value of "mon_name" is currently chosen at the moment the call is made based on the setting of nsm_use_hostnames To ensure both requests pass identical contents in the "mon_name" argument, we now select which string to use for the argument in the nsm_monitor() function. A pointer to this string is saved in the nsm_handle so it can be used for the subsequent NSMPROC_UNMON upcall. Signed-off-by: Chuck Lever public.gmane.org> --- fs/lockd/mon.c | 45 +++++++++++++++++-------------------------- include/linux/lockd/lockd.h | 1 + 2 files changed, 19 insertions(+), 27 deletions(-) diff --git a/fs/lockd/mon.c b/fs/lockd/mon.c index 4e7e958..a606fbb 100644 --- a/fs/lockd/mon.c +++ b/fs/lockd/mon.c < at >< at > -18,8 +18,6 < at >< at > #define NLMDBG_FACILITYNLMDBG_MONITOR -#define XDR_ADDRBUF_LEN(20) - static struct rpc_clnt *nsm_create(void); static struct rpc_programnsm_program; < at >< at > -37,7 +35,13 < at >< at > nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res) { struct rpc_clnt*clnt; intstatus; -struct nsm_argsargs; +struct nsm_args args = { +.addr= nsm_addr_in(nsm)->sin_addr.s_addr, +.prog= NLM_PROGRAM, +.vers= 3, +.proc= NLMPROC_NSM_NOTIFY, +.mon_name= nsm->sm_mon_name, +}; struct rpc_message msg = { .rpc_argp= &args, .rpc_resp= res, < at >< at > -46,22 +50,18 < at >< at > nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res) clnt = nsm_create(); if (IS_ERR(clnt)) { status = PTR_ERR(clnt); +dprintk("lockd: failed to create NSM upcall transport, " +"status=%d\n", status); goto out; } -memset(&args, 0, sizeof(args)); -args.mon_name = nsm->sm_name; -args.addr = nsm_addr_in(nsm)->sin_addr.s_addr; -args.prog = NLM_PROGRAM; -args.vers = 3; -args.proc = NLMPROC_NSM_NOTIFY; memset(res, 0, sizeof(*res)); msg.rpc_proc = &clnt->cl_procinfo[proc]; status = rpc_call_sync(clnt, &msg, 0); if (status < 0) -printk(KERN_DEBUG "nsm_mon_unmon: rpc failed, status=%d\n", -status); +dprintk("lockd: NSM upcall RPC failed, status=%d\n", +status); else status = 0; rpc_shutdown_client(clnt); < at >< at > -85,6 +85,12 < at >< at > nsm_monitor(struct nlm_host *host) if (nsm->sm_monitored) return 0; +/* + * Choose whether to record the caller_name or IP address of + * this peer in the local rpc.statd's database. + */ +nsm->sm_mon_name = nsm_use_hostnames ? nsm->sm_name : nsm->sm_addrbuf; + status = nsm_mon_unmon(nsm, SM_MON, &res); if (status < 0 || res.status != 0) < at >< at > -165,25 +171,10 < at >< at > static __be32 *xdr_encode_nsm_string(__be32 *p, char *string) /* * "mon_name" specifies the host to be monitored. - * - * Linux uses a text version of the IP address of the remote - * host as the host identifier (the "mon_name" argument). - * - * Linux statd always looks up the canonical hostname first for - * whatever remote hostname it receives, so this works alright. */ static __be32 *xdr_encode_mon_name(__be32 *p, struct nsm_args *argp) { -charbuffer[XDR_ADDRBUF_LEN + 1]; -char*name = argp->mon_name; - -if (!nsm_use_hostnames) { -snprintf(buffer, XDR_ADDRBUF_LEN, - NIPQUAD_FMT, NIPQUAD(argp->addr)); -name = buffer; -} - -return xdr_encode_nsm_string(p, name); +return xdr_encode_nsm_string(p, argp->mon_name); } /* diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h index 307a8f0..de9ea7b 100644 --- a/include/linux/lockd/lockd.h +++ b/include/linux/lockd/lockd.h < at >< at > -73,6 +73,7 < at >< at > struct nsm_handle { char *sm_name; struct sockaddr_storagesm_addr; size_tsm_addrlen; +char *sm_mon_name; unsigned intsm_monitored : 1, sm_sticky : 1;/* don't unmonitor */ charsm_addrbuf[63];/* presentation address */ -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 1/6] NLM: Remove address eye-catcher buffers from nlm_host

Chuck Lever — 2008-12-01T18:57:34

The h_name field in struct nlm_host is a just copy of h_nsmhandle->sm_name. Likewise, the contents of the h_addrbuf field should be identical to the sm_addrbuf field. The h_srcaddrbuf field is used only in one place for debugging. We can live without this until we get %pI formatting for printk(). Currently these buffers are 48 bytes, but we need to support scope IDs in IPv6 presentation addresses, which means making the buffers even larger. Instead, let's find ways to eliminate them to save space. Finally, AF_UNSPEC support is no longer needed in nlm_display_address() now that it is not used for the h_srcaddr field. Signed-off-by: Chuck Lever public.gmane.org> --- fs/lockd/host.c | 14 +++----------- include/linux/lockd/lockd.h | 3 --- 2 files changed, 3 insertions(+), 14 deletions(-) diff --git a/fs/lockd/host.c b/fs/lockd/host.c index acc2aa5..68e00c0 100644 --- a/fs/lockd/host.c +++ b/fs/lockd/host.c < at >< at > -112,9 +112,6 < at >< at > static void nlm_display_address(const struct sockaddr *sap, const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap; switch (sap->sa_family) { -case AF_UNSPEC: -snprintf(buf, len, "unspecified"); -break; case AF_INET: snprintf(buf, len, NIPQUAD_FMT, NIPQUAD(sin->sin_addr.s_addr)); break; < at >< at > -178,7 +175,7 < at >< at > static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni) nlm_get_host(host); dprintk("lockd: nlm_lookup_host found host %s (%s)\n", -host->h_name, host->h_addrbuf); +host->h_name, nsm->sm_addrbuf); goto out; } < at >< at > -232,11 +229,6 < at >< at > static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni) nrhosts++; -nlm_display_address((struct sockaddr *)&host->h_addr, -host->h_addrbuf, sizeof(host->h_addrbuf)); -nlm_display_address((struct sockaddr *)&host->h_srcaddr, -host->h_srcaddrbuf, sizeof(host->h_srcaddrbuf)); - dprintk("lockd: nlm_lookup_host created host %s\n", host->h_name); < at >< at > -378,8 +370,8 < at >< at > nlm_bind_host(struct nlm_host *host) { struct rpc_clnt*clnt; -dprintk("lockd: nlm_bind_host %s (%s), my addr=%s\n", -host->h_name, host->h_addrbuf, host->h_srcaddrbuf); +dprintk("lockd: nlm_bind_host %s (%s)\n", +host->h_name, host->h_nsmhandle->sm_addrbuf); /* Lock host handle */ mutex_lock(&host->h_mutex); diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h index 23da3fa..4467b83 100644 --- a/include/linux/lockd/lockd.h +++ b/include/linux/lockd/lockd.h < at >< at > -65,9 +65,6 < at >< at > struct nlm_host { struct list_headh_granted;/* Locks in GRANTED state */ struct list_headh_reclaim;/* Locks in RECLAIM state */ struct nsm_handle *h_nsmhandle;/* NSM status handle */ - -charh_addrbuf[48],/* address eyecatchers */ -h_srcaddrbuf[48]; }; struct nsm_handle { -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 0/6] First set of patches for 2.6.29

Chuck Lever — 2008-12-01T18:57:25

These patches add some IPv6 support to the Linux kernel's NSM upcall implementation, and address a couple of shortcomings in the existing NLM/NSM implementation. I'd like you to consider these for 2.6.29. There are 40 more patches I have queued for 2.6.29 that finish IPv6 support in the NSM upcall implementation, enable IPv6 for lockd, and finish and enable IPv6 support in NFSD. --- Chuck Lever (6): NSM: Serialize SM_MON and SM_UNMON upcalls NLM: Clean up nsm_monitor() call NLM: Clean up nsm_monitor() call NSM: Support IPv6 version of mon_name NLM: Support IPv6 scope IDs in nlm_display_address() NLM: Remove address eye-catcher buffers from nlm_host fs/lockd/host.c | 53 ++++++++++-------- fs/lockd/mon.c | 120 +++++++++++++++++++++------------------- include/linux/lockd/lockd.h | 21 ++++++- include/linux/lockd/sm_inter.h | 2 - 4 files changed, 109 insertions(+), 87 deletions(-)

[PATCH 4/4] mount command: remove local getport() implementation

Chuck Lever — 2008-12-01T18:48:56

Eliminate local getport() implementation from utils/mount/network.c, as it is no longer used. Signed-off-by: Chuck Lever public.gmane.org> --- utils/mount/network.c | 76 +------------------------------------------------ 1 files changed, 2 insertions(+), 74 deletions(-) diff --git a/utils/mount/network.c b/utils/mount/network.c index 8d96e12..0c68993 100644 --- a/utils/mount/network.c +++ b/utils/mount/network.c < at >< at > -449,78 +449,6 < at >< at > err_connect: return RPC_ANYSOCK; } -/* - * getport() is very similar to pmap_getport() with the exception that - * this version tries to use an ephemeral port, since reserved ports are - * not needed for GETPORT queries. This conserves the very limited - * reserved port space, which helps reduce failed socket binds - * during mount storms. - * - * A side effect of calling this function is that rpccreateerr is set. - */ -static unsigned short getport(struct sockaddr_in *saddr, -unsigned long program, -unsigned long version, -unsigned int proto) -{ -struct sockaddr_in bind_saddr; -unsigned short port = 0; -int socket; -CLIENT *clnt = NULL; -enum clnt_stat stat; - -bind_saddr = *saddr; -bind_saddr.sin_port = htons(PMAPPORT); - -socket = get_socket(&bind_saddr, proto, PMAP_TIMEOUT, FALSE, TRUE); -if (socket == RPC_ANYSOCK) { -if (proto == IPPROTO_TCP && - rpc_createerr.cf_error.re_errno == ETIMEDOUT) -rpc_createerr.cf_stat = RPC_TIMEDOUT; -return 0; -} - -switch (proto) { -case IPPROTO_UDP: -clnt = clntudp_bufcreate(&bind_saddr, - PMAPPROG, PMAPVERS, - RETRY_TIMEOUT, &socket, - RPCSMALLMSGSIZE, - RPCSMALLMSGSIZE); -break; -case IPPROTO_TCP: -clnt = clnttcp_create(&bind_saddr, - PMAPPROG, PMAPVERS, - &socket, - RPCSMALLMSGSIZE, RPCSMALLMSGSIZE); -break; -} -if (clnt != NULL) { -struct pmap parms = { -.pm_prog= program, -.pm_vers= version, -.pm_prot= proto, -}; - -stat = clnt_call(clnt, PMAPPROC_GETPORT, - (xdrproc_t)xdr_pmap, (caddr_t)&parms, - (xdrproc_t)xdr_u_short, (caddr_t)&port, - TIMEOUT); -if (stat) { -clnt_geterr(clnt, &rpc_createerr.cf_error); -rpc_createerr.cf_stat = stat; -} -clnt_destroy(clnt); -if (stat != RPC_SUCCESS) -port = 0; -else if (port == 0) -rpc_createerr.cf_stat = RPC_PROGNOTREGISTERED; -} -close(socket); - -return port; -} - static void nfs_pp_debug(const struct sockaddr *sap, const socklen_t salen, const rpcprog_t program, const rpcvers_t version, const unsigned short protocol, < at >< at > -843,9 +771,9 < at >< at > void mnt_closeclnt(CLIENT *clnt, int msock) * < at >prot: target RPC protocol * < at >caddr: filled in with our network address * - * Sigh... getport() doesn't actually check the version number. + * Sigh... GETPORT queries don't actually check the version number. * In order to make sure that the server actually supports the service - * we're requesting, we open and RPC client, and fire off a NULL + * we're requesting, we open an RPC client, and fire off a NULL * RPC call. * * caddr is the network address that the server will use to call us back. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 3/4] mount command: Replace clnt_ping() and getport() calls inprobe_port()

Chuck Lever — 2008-12-01T18:48:51

Update the mount command's probe_port() function to call the new shared rpcbind query and RPC ping functions. This provides immediate support for rpcbind v3/v4 queries, and paves the way for supporting AF_INET6 in the probe_bothports() path. Signed-off-by: Chuck Lever public.gmane.org> --- utils/mount/network.c | 43 +++++++++++++++++++++++++++++-------------- 1 files changed, 29 insertions(+), 14 deletions(-) diff --git a/utils/mount/network.c b/utils/mount/network.c index 2231210..8d96e12 100644 --- a/utils/mount/network.c +++ b/utils/mount/network.c < at >< at > -521,6 +521,27 < at >< at > static unsigned short getport(struct sockaddr_in *saddr, return port; } +static void nfs_pp_debug(const struct sockaddr *sap, const socklen_t salen, + const rpcprog_t program, const rpcvers_t version, + const unsigned short protocol, + const unsigned short port) +{ +char buf[NI_MAXHOST]; + +if (!verbose) +return; + +if (nfs_present_sockaddr(sap, salen, buf, sizeof(buf)) == 0) { +buf[0] = '\0'; +strcat(buf, "unknown host"); +} + +fprintf(stderr, _("%s: trying %s prog %ld vers %ld prot %s port %d\n"), +progname, buf, program, version, +(protocol == IPPROTO_UDP ? _("UDP") : _("TCP")), +port); +} + /* * Use the portmapper to discover whether or not the service we want is * available. The lists 'versions' and 'protos' define ordered sequences < at >< at > -529,7 +550,8 < at >< at > static unsigned short getport(struct sockaddr_in *saddr, static int probe_port(clnt_addr_t *server, const unsigned long *versions, const unsigned int *protos) { -struct sockaddr_in *saddr = &server->saddr; +const struct sockaddr *saddr = (struct sockaddr *)&server->saddr; +const socklen_t salen = sizeof(server->saddr); struct pmap *pmap = &server->pmap; const unsigned long prog = pmap->pm_prog, *p_vers; const unsigned int prot = (u_int)pmap->pm_prot, *p_prot; < at >< at > -541,21 +563,14 < at >< at > static int probe_port(clnt_addr_t *server, const unsigned long *versions, p_vers = vers ? &vers : versions; rpc_createerr.cf_stat = 0; for (;;) { -p_port = getport(saddr, prog, *p_vers, *p_prot); +p_port = nfs_getport(saddr, salen, prog, *p_vers, *p_prot); if (p_port) { if (!port || port == p_port) { -saddr->sin_port = htons(p_port); -if (verbose) { -printf(_("%s: trying %s prog %ld vers " -"%ld prot %s port %d\n"), -progname, -inet_ntoa(saddr->sin_addr), -prog, *p_vers, -*p_prot == IPPROTO_UDP ? -_("UDP") : _("TCP"), -p_port); - } -if (clnt_ping(saddr, prog, *p_vers, *p_prot, NULL)) +server->saddr.sin_port = htons(p_port); +nfs_pp_debug(saddr, salen, prog, *p_vers, +*p_prot, p_port); +if (nfs_rpc_ping(saddr, salen, prog, +*p_vers, *p_prot, NULL)) goto out_ok; } } -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 2/4] mount command: Use nfs_error() instead of perror()

Chuck Lever — 2008-12-01T18:48:45

So we can ensure that error output is directed appropriately, use nfs_error() instead of perror() in start_statd(). Signed-off-by: Chuck Lever public.gmane.org> --- utils/mount/network.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/utils/mount/network.c b/utils/mount/network.c index 0ba7bdb..2231210 100644 --- a/utils/mount/network.c +++ b/utils/mount/network.c < at >< at > -710,7 +710,8 < at >< at > int start_statd(void) execl(START_STATD, START_STATD, NULL); exit(1); case -1: /* error */ -perror("Fork failed"); +nfs_error(_("fork failed: %s"), +strerror(errno)); break; default: /* parent */ waitpid(pid, NULL,0); -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 1/4] mount command: Use nfs_pmap_getport() in probe_statd()

Chuck Lever — 2008-12-01T18:48:40

Repace the getport() and clnt_ping() calls in probe_statd() with their new shared equivalents. Signed-off-by: Chuck Lever public.gmane.org> --- utils/mount/network.c | 34 ++++++++++++++++------------------ 1 files changed, 16 insertions(+), 18 deletions(-) diff --git a/utils/mount/network.c b/utils/mount/network.c index 806344c..0ba7bdb 100644 --- a/utils/mount/network.c +++ b/utils/mount/network.c < at >< at > -47,6 +47,7 < at >< at > #include "nls.h" #include "nfs_mount.h" #include "mount_constants.h" +#include "nfsrpc.h" #include "network.h" /* < at >< at > -79,6 +80,11 < at >< at > extern int nfs_mount_data_version; extern char *progname; extern int verbose; +static const char *nfs_ns_pgmtbl[] = { +"status", +NULL, +}; + static const unsigned long nfs_to_mnt[] = { 0, 0, < at >< at > -669,24 +675,16 < at >< at > version_fixed: return probe_mntport(mnt_server); } -static int probe_statd(void) +static int nfs_probe_statd(void) { -struct sockaddr_in addr; -unsigned short port; - -memset(&addr, 0, sizeof(addr)); -addr.sin_family = AF_INET; -addr.sin_addr.s_addr = htonl(INADDR_LOOPBACK); -port = getport(&addr, 100024, 1, IPPROTO_UDP); - -if (port == 0) -return 0; -addr.sin_port = htons(port); - -if (clnt_ping(&addr, 100024, 1, IPPROTO_UDP, NULL) <= 0) -return 0; +struct sockaddr_in addr = { +.sin_family= AF_INET, +.sin_addr.s_addr= htonl(INADDR_LOOPBACK), +}; +rpcprog_t program = nfs_getrpcbyname(NSMPROG, nfs_ns_pgmtbl); -return 1; +return nfs_getport_ping((struct sockaddr *)&addr, sizeof(addr), +program, (rpcvers_t)1, IPPROTO_UDP); } /** < at >< at > -700,7 +698,7 < at >< at > int start_statd(void) struct stat stb; #endif -if (probe_statd()) +if (nfs_probe_statd()) return 1; #ifdef START_STATD < at >< at > -718,7 +716,7 < at >< at > int start_statd(void) waitpid(pid, NULL,0); break; } -if (probe_statd()) +if (nfs_probe_statd()) return 1; } } -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html

[PATCH 0/4] Make mount.nfs use new getport() function

Chuck Lever — 2008-12-01T18:48:33

Hi Steve- Four new IPv6-related patches for mount.nfs. These replace the existing PMAP_GETPORT implementation in utils/mount/network.c with a call to the new library functions we just added. The probe_statd() function is used by both the legacy and the text- based paths, but replacing the mount.nfs command's getport() function affects only the *legacy* NFS mount path, not text-based mounts. For anyone testing this code, please try it on pre-2.6.23 kernels, or patch mount.nfs to use legacy binary style mounts. This will exercise the new PMAP_GETPORT implementation using all the special cases needed by mount.nfs. The new RPCB_GETADDR implementation will not be used until mount.nfs can pass in IPv6 addresses. --- Chuck Lever (4): mount command: remove local getport() implementation mount command: Replace clnt_ping() and getport() calls in probe_port() mount command: Use nfs_error() instead of perror() mount command: Use nfs_pmap_getport() in probe_statd() utils/mount/network.c | 144 +++++++++++++++---------------------------------- 1 files changed, 43 insertions(+), 101 deletions(-)

Re: ipv6 status

Brian J. Murrell — 2008-12-01T18:40:02

Indeed. :-) Thanx for the update. Looking forward to it finally finishing. b.

Re: ipv6 status

Chuck Lever — 2008-12-01T18:33:25

There isn't a published roadmap. Linux developers tend to eschew deadlines. The NFS client in 2.6.28 (and maybe 2.6.27) should support NFSv4 over IPv6. Use the latest nfs-utils. I'm hoping 2.6.29 will see all the kernel pieces we need for NFSv2/v3 client and server, and NFSv4 server support for IPv6. Support for NFSv2/v3 over IPv6, and server-side support for NFSv4 over IPv6, requires user space changes. We are pushing those into nfs- utils over time, but I don't expect that work to be complete until sometime in 1H2009.

Fwd: nfs related backtrace. (.27.5)

Chuck Lever — 2008-12-01T17:47:47

Forwarding to linux-nfs-u79uwXL29TY76Z2rM5mHXA< at >public.gmane.org Begin forwarded message: