user/sven/linux.git/include/linux/lockd, branch v2.6.31.2

lockd: Update NSM state from SM_MON replies

2009-06-18T01:02:10Z

When rpc.statd starts up in user space at boot time, it attempts to write the latest NSM local state number into /proc/sys/fs/nfs/nsm_local_state. If lockd.ko isn't loaded yet (as is the case in most configurations), that file doesn't exist, thus the kernel's NSM state remains set to its initial value of zero during lockd operation. This is a problem because rpc.statd and lockd use the NSM state number to prevent repeated lock recovery on rebooted hosts. If lockd sends a zero NSM state, but then a delayed SM_NOTIFY with a real NSM state number is received, there is no way for lockd or rpc.statd to distinguish that stale SM_NOTIFY from an actual reboot. Thus lock recovery could be performed after the rebooted host has already started reclaiming locks, and those locks will be lost. We could change /etc/init.d/nfslock so it always modprobes lockd.ko before starting rpc.statd. However, if lockd.ko is ever unloaded and reloaded, we are back at square one, since the NSM state is not preserved across an unload/reload cycle. This may happen frequently on clients that use automounter. A period of NFS inactivity causes lockd.ko to be unloaded, and the kernel loses its NSM state setting. Instead, let's use the fact that rpc.statd plants the local system's NSM state in every SM_MON (and SM_UNMON) reply. lockd performs a synchronous SM_MON upcall to the local rpc.statd _before_ sending its first NLM request to a new remote. This would permit rpc.statd to provide the current NSM state to lockd, even after lockd.ko had been unloaded and reloaded. Note that NLMPROC_LOCK arguments are constructed before the nsm_monitor() call, so we have to rearrange argument construction very slightly to make this all work out. And, the kernel appears to treat NSM state as a u32 (see struct nlm_args and nsm_res). Make nsm_local_state a u32 as well, to ensure we don't get bogus comparison results. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust

nfs: replace uses of __constant_{endian}

2009-03-18T21:30:51Z

The base versions handle constant folding now, none of these headers are exported to userspace, so the __ prefixed versions are not necessary. Signed-off-by: Harvey Harrison Reviewed-by: NeilBrown Signed-off-by: J. Bruce Fields

NLM: Shrink the IPv4-only version of nlm_cmp_addr()

2009-03-11T00:33:19Z

Clean up/micro-optimatization: Make the AF_INET-only version of nlm_cmp_addr() smaller. This matches the style of nlm_privileged_requester(), and makes the AF_INET-only version of nlm_cmp_addr() nearly the same size as it was before IPv6 support. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust

NLM: Rewrite IPv4 privileged requester's check

2009-01-06T16:53:56Z

Clean up. For consistency, rewrite the IPv4 check to match the same style as the new IPv6 check. Note that ipv4_is_loopback() is somewhat broader in its interpretation of what is a loopback address than simply "127.0.0.1". Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields

NLM: nlm_privileged_requester() doesn't recognize mapped loopback address

2009-01-06T16:53:56Z

Commit b85e4676 added the nlm_privileged_requester() helper to check whether an RPC request was sent from a local privileged caller. It recognizes IPv4 privileged callers (from "127.0.0.1"), and IPv6 privileged callers (from "::1"). However, IPV6_ADDR_LOOPBACK is not set for the mapped IPv4 loopback address (::ffff:7f00:0001), so the test breaks when the kernel's RPC service is IPv6-enabled but user space is calling via the IPv4 loopback address. This is actually the most common case for IPv6- enabled RPC services on Linux. Rewrite the IPv6 check to handle the mapped IPv4 loopback address as well as a normal IPv6 loopback address. Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields

NSM: Move nsm_addr() to fs/lockd/mon.c

2009-01-06T16:53:55Z

Clean up: nsm_addr_in() is no longer used, and nsm_addr() is used only in fs/lockd/mon.c, so move it there. Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields

NSM: Remove include/linux/lockd/sm_inter.h

2009-01-06T16:53:55Z

Clean up: The include/linux/lockd/sm_inter.h header is nearly empty now. Remove it. Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields

NLM: Remove "create" argument from nsm_find()

2009-01-06T16:53:54Z

Clean up: nsm_find() now has only one caller, and that caller unconditionally sets the @create argument. Thus the @create argument is no longer needed. Since nsm_find() now has a more specific purpose, pick a more appropriate name for it. Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields

NSM: Add nsm_lookup() function

2009-01-06T16:53:54Z

Introduce a new API to fs/lockd/mon.c that allows nlm_host_rebooted() to lookup up nsm_handles via the contents of an nlm_reboot struct. The new function is equivalent to calling nsm_find() with @create set to zero, but it takes a struct nlm_reboot instead of separate arguments. Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields

NLM: Decode "priv" argument of NLMPROC_SM_NOTIFY as an opaque

2009-01-06T16:53:54Z

The NLM XDR decoders for the NLMPROC_SM_NOTIFY procedure should treat their "priv" argument truly as an opaque, as defined by the protocol, and let the upper layers figure out what is in it. This will make it easier to modify the contents and interpretation of the "priv" argument, and keep knowledge about what's in "priv" local to fs/lockd/mon.c. For now, the NLM and NSM implementations should behave exactly as they did before. The formation of the address of the rebooted host in nlm_host_rebooted() may look a little strange, but it is the inverse of how nsm_init_private() forms the private cookie. Plus, it's going away soon anyway. Signed-off-by: Chuck Lever Signed-off-by: J. Bruce Fields