This patch moves next_gc to per-net data.

Note: passed network can be NULL (when Lockd kthread is exiting of Lockd
module is removing).
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

This is required for per-network NLM shutdown and cleanup.
This patch passes init_net for a while.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

This is just a code move, which from my POV makes the code look better.
I.e. now on start we have 3 different stages:
1) Service creation.
2) Service per-net data allocation.
3) Service start.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

This function creates service if it doesn't exist, or increases usage
counter if it does, and returns a pointer to it.  The usage counter will
be droppepd by svc_destroy() later in lockd_up().
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

This patch also replaces svc_rpcb_setup() with svc_bind().
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

The idea is to separate service destruction and per-net operations,
because these are two different things and the mix looks ugly.

Notes:

1) For NFS server this patch looks ugly (sorry for that). But these
place will be rewritten soon during NFSd containerization.

2) LockD per-net counter increase int lockd_up() was moved prior to
make_socks() to make lockd_down_net() call safe in case of error.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

This new routine is responsible for service registration in a specified
network context.

The idea is to separate service creation from per-net operations.

Note also: since registering service with svc_bind() can fail, the
service will be destroyed and during destruction it will try to
unregister itself from rpcbind. In this case unregistration has to be
skipped.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

comparing be32 values for < is not doing the right thing...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

v2: dereference of most probably already released nlm_host removed in
nlmclnt_done() and reclaimer().

These routines are called from locks reclaimer() kernel thread. This thread
works in "init_net" network context and currently relays on persence on lockd
thread and it's per-net resources. Thus lockd_up() and lockd_down() can't relay
on current network context. So let's pass corrent one into them.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Stephen Rothwell reports:
net/sunrpc/rpcb_clnt.c: In function 'rpcb_enc_mapping':
net/sunrpc/rpcb_clnt.c:820:19: warning: unused variable 'task' [-Wunused-variable]
net/sunrpc/rpcb_clnt.c: In function 'rpcb_dec_getport':
net/sunrpc/rpcb_clnt.c:837:19: warning: unused variable 'task' [-Wunused-variable]
net/sunrpc/rpcb_clnt.c: In function 'rpcb_dec_set':
net/sunrpc/rpcb_clnt.c:860:19: warning: unused variable 'task' [-Wunused-variable]
net/sunrpc/rpcb_clnt.c: In function 'rpcb_enc_getaddr':
net/sunrpc/rpcb_clnt.c:892:19: warning: unused variable 'task' [-Wunused-variable]
net/sunrpc/rpcb_clnt.c: In function 'rpcb_dec_getaddr':
net/sunrpc/rpcb_clnt.c:914:19: warning: unused variable 'task' [-Wunused-variable]
fs/lockd/svclock.c:49:20: warning: 'nlmdbg_cookie2a' declared 'static' but never defined [-Wunused-function]
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

If you try to set grace_period or timeout via a module parameter
to lockd, and do this on a big-endian machine where

   sizeof(int) != sizeof(unsigned long)

it won't work.  This number given will be effectively shifted right
by the difference in those two sizes.

So cast kp->arg properly to get correct result.

Cc: stable@kernel.org
Signed-off-by: NNeilBrown <neilb@suse.de>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Lockd now managed in network namespace context. And this patch introduces
network namespace related NLM hosts shutdown in case of releasing per-net Lockd
resources.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

NLM host is network namespace aware now.
So NSM have to take it into account.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

This object depends on RPC client, and thus on network namespace.
So let's make it's allocation and lookup in network namespace context.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

This patch introduces per-net Lockd initialization and destruction routines.
The logic is the same as in global Lockd up and down routines. Probably the
solution is not the best one. But at least it looks clear.
So per-net "up" routine are called only in case of lockd is running already. If
per-net resources are not allocated yet, then service is being registered with
local portmapper and lockd sockets created.
Per-net "down" routine is called on every lockd_down() call in case of global
users counter is not zero.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Lockd is going to be shared between network namespaces - i.e. going to be able
to handle lock requests from different network namespaces. This means, that
network namespace related resources have to be allocated not once (like now),
but for every network namespace context, from which service is requested to
operate.
This patch implements Lockd per-net users accounting. New per-net counter is
used to determine, when per-net resources have to be freed.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

This patch parametrizes Lockd permanent sockets creation routine by network
namespace context.
It also replaces hard-coded init_net with current network namespace context in
Lockd sockets creation routines.
This approach looks safe, because Lockd is created during NFS mount (or NFS
server start) and thus socket is required exactly in current network namespace
context. But in the same time it means, that Lockd sockets inherits first Lockd
requester network namespace. This issue will be fixed in further patches of the
series.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Service transports are parametrized by network namespace. And thus lookup of
transport instance have to take network namespace into account.
Signed-off-by: NStanislav Kinsbursky <skinsbursky@parallels.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: NJ. Bruce Fields <bfields@redhat.com>

module_param(bool) used to counter-intuitively take an int.  In
fddd5201 (mid-2009) we allowed bool or int/unsigned int using a messy
trick.

It's time to remove the int/unsigned int option.  For this version
it'll simply give a warning, but it'll break next kernel version.
Acked-by: NMauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

For IPv6 local address, lockd can not callback to client for
missing scope id when binding address at inet6_bind:

 324       if (addr_type & IPV6_ADDR_LINKLOCAL) {
 325               if (addr_len >= sizeof(struct sockaddr_in6) &&
 326                   addr->sin6_scope_id) {
 327                       /* Override any existing binding, if another one
 328                        * is supplied by user.
 329                        */
 330                       sk->sk_bound_dev_if = addr->sin6_scope_id;
 331               }
 332
 333               /* Binding to link-local address requires an interface */
 334               if (!sk->sk_bound_dev_if) {
 335                       err = -EINVAL;
 336                       goto out_unlock;
 337               }

Replacing svc_addr_u by sockaddr_storage, let rqstp->rq_daddr contains more info
besides address.
Reviewed-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NMi Jinlong <mijinlong@cn.fujitsu.com>
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Use NUMA aware allocations to reduce latencies and increase throughput.

sunrpc kthreads can use kthread_create_on_node() if pool_mode is
"percpu" or "pernode", and svc_prepare_thread()/svc_init_buffer() can
also take into account NUMA node affinity for memory allocations.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
CC: "J. Bruce Fields" <bfields@fieldses.org>
CC: Neil Brown <neilb@suse.de>
CC: David Miller <davem@davemloft.net>
Reviewed-by: NGreg Banks <gnb@fastmail.fm>
[bfields@redhat.com: fix up caller nfs41_callback_up]
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

Both the filesystem and the lock manager can associate operations with a
lock.  Confusingly, one of them (fl_release_private) actually has the
same name in both operation structures.

It would save some confusion to give the lock-manager ops different
names.
Signed-off-by: NJ. Bruce Fields <bfields@redhat.com>

lockd: server returns status 50331648
it's quite hard to understand that number in this message is 3 in big endian
Signed-off-by: NVasily Averin <vvs@sw.ru>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

If the NLM daemon is killed on the NFS server, we can currently end up
hanging forever on an 'unlock' request, instead of aborting. Basically,
if the rpcbind request fails, or the server keeps returning garbage, we
really want to quit instead of retrying.
Tested-by: NVasily Averin <vvs@sw.ru>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org

Nick Bowler <nbowler@elliptictech.com> reports:

> We were just having some NFS server troubles, and my client machine
> running 2.6.38-rc1+ (specifically, commit 2b1caf6e) crashed
> hard (syslog output appended to this mail).
>
> I'm not sure what the exact timeline was or how to reproduce this,
> but the server was rebooted during all this.  Since I've never seen
> this happen before, it is possibly a regression from previous kernel
> releases.  However, I recently updated my nfs-utils (on the client) to
> version 1.2.3, so that might be related as well.

  [ BUG output redacted ]

When done searching, the for_each_host loop in next_host_state() falls
through and returns the final host on the host chain without bumping
it's reference count.

Since the host's ref count is only one at that point, releasing the
host in nlm_host_rebooted() attempts to destroy the host prematurely,
and therefore hits a BUG().

Likely, the original intent of the for_each_host behavior in
next_host_state() was to handle the case when the host chain is empty.
Searching the chain and finding no suitable host to return needs to be
handled as well.

Defensively restructure next_host_state() always to return NULL when
the loop falls through.

Introduced by commit b10e30f6 "lockd: reorganize nlm_host_rebooted".

Cc: J. Bruce Fields <bfields@fieldses.org>
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

We unlock again after we goto out.
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.

The contents of the src_sap field is not used in nlm_alloc_host().
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.

Remove the now unused helper nlm_lookup_host().
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.

nlm_hosts now contains only server-side entries.  Rename it to match
convention of client side cache.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.

Change nlmsvc_lookup_host() to be purpose-built for server-side
nlm_host management.  This replaces the generic nlm_lookup_host()
helper function, just like on the client side.  The lookup logic is
specialized for server host lookups.

The server side cache also gets its own specialized equivalent of the
nlm_release_host() function.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

NFS clients don't need the garbage collection processing that is
performed on nlm_host structures.  The client picks up an nlm_host at
mount time and holds a reference to it until the file system is
unmounted.

Servers, on the other hand, don't have a precise way to tell when an
nlm_host is no longer being used, so zero refcount nlm_host entries
are left to expire in the cache after a time.

Basically there's nothing holding a reference to an nlm_host between
individual server-side NLM requests, but we can't afford the expense
of recreating them for every new NLM request from a client.  The
nlm_host cache adds some lifetime hysteresis to entries in the cache
so the next time a particular nlm_host is needed, it's likely to be
discovered by a lookup rather than created from whole cloth.

With the new implementation, client nlm_host cache items are no longer
garbage collected, and are destroyed directly by a new release
function specialized for client entries, nlmclnt_release_host().  They
are cached in their own data structure, and have their own lookup
logic, simplified and specialized for client nlm_host entries.

However, the client nlm_host cache still shares reboot recovery logic
with the server nlm_host cache.  The NSM "peer rebooted" downcall for
clients and servers still come through the same RPC call.  This is a
legacy formal API that would be difficult to alter, and besides, the
user space NSM implementation can't tell the difference between peers
that are clients or servers.

For this reason, the client cache continues to share the
nlm_host_mutex (and reboot recovery logic) with the server cache.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The nlm_release_call() function is invoked from both the server and
the client side.  We're about to introduce a distinct server- and
client-side nlm_release_host(), so nlm_release_call() must first be
split into a client-side and a server-side version.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Refactor the tail of nlm_gc_hosts() into nlm_destroy_host() so that
this logic can be used separately from garbage collection.

Rename it _locked() to document that it must be called with the hosts
cache mutex held.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>