Helge Deller [Mon, 2 Dec 2013 18:59:31 +0000 (19:59 +0100)]
nfs: fix do_div() warning by instead using sector_div()
When compiling a 32bit kernel with CONFIG_LBDAF=n the compiler complains like
shown below. Fix this warning by instead using sector_div() which is provided
by the kernel.h header file.
fs/nfs/blocklayout/extents.c: In function ‘normalize’:
include/asm-generic/div64.h:43:28: warning: comparison of distinct pointer types lacks a cast [enabled by default]
fs/nfs/blocklayout/extents.c:47:13: note: in expansion of macro ‘do_div’
nfs/blocklayout/extents.c:47:2: warning: right shift count >= width of type [enabled by default]
fs/nfs/blocklayout/extents.c:47:2: warning: passing argument 1 of ‘__div64_32’ from incompatible pointer type [enabled by default]
include/asm-generic/div64.h:35:17: note: expected ‘uint64_t *’ but argument is of type ‘sector_t *’
extern uint32_t __div64_32(uint64_t *dividend, uint32_t divisor);
Trond Myklebust [Wed, 4 Dec 2013 17:09:45 +0000 (12:09 -0500)]
NFSv4.1: Prevent a 3-way deadlock between layoutreturn, open and state recovery
Andy Adamson reports:
The state manager is recovering expired state and recovery OPENs are being
processed. If kswapd is pruning inodes at the same time, a deadlock can occur
when kswapd calls evict_inode on an NFSv4.1 inode with a layout, and the
resultant layoutreturn gets an error that the state mangager is to handle,
causing the layoutreturn to wait on the (NFS client) cl_rpcwaitq.
At the same time an open is waiting for the inode deletion to complete in
__wait_on_freeing_inode.
If the open is either the open called by the state manager, or an open from
the same open owner that is holding the NFSv4 sequence id which causes the
OPEN from the state manager to wait for the sequence id on the Seqid_waitqueue,
then the state is deadlocked with kswapd.
The fix is simply to have layoutreturn ignore all errors except NFS4ERR_DELAY.
We already know that layouts are dropped on all server reboots, and that
it has to be coded to deal with the "forgetful client model" that doesn't
send layoutreturns.
Andy Adamson [Fri, 15 Nov 2013 21:36:16 +0000 (16:36 -0500)]
NFSv4 wait on recovery for async session errors
When the state manager is processing the NFS4CLNT_DELEGRETURN flag, session
draining is off, but DELEGRETURN can still get a session error.
The async handler calls nfs4_schedule_session_recovery returns -EAGAIN, and
the DELEGRETURN done then restarts the RPC task in the prepare state.
With the state manager still processing the NFS4CLNT_DELEGRETURN flag with
session draining off, these DELEGRETURNs will cycle with errors filling up the
session slots.
This prevents OPEN reclaims (from nfs_delegation_claim_opens) required by the
NFS4CLNT_DELEGRETURN state manager processing from completing, hanging the
state manager in the __rpc_wait_for_completion_task in nfs4_run_open_task
as seen in this kernel thread dump:
Anna Schumaker [Wed, 13 Nov 2013 17:29:08 +0000 (12:29 -0500)]
NFS: Enabling v4.2 should not recompile nfsd and lockd
When CONFIG_NFS_V4_2 is toggled nfsd and lockd will be recompiled,
instead of only the nfs client. This patch moves a small amount of code
into the client directory to avoid unnecessary recompiles.
Signed-off-by: Anna Schumaker <bjschuma@netapp.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Defaulting to m seem to prevent building the pnfs layout modules into the
kernel. Default to the value of CONFIG_NFS_V4 make sure they are
built in for built-in NFSv4 support and modular for a modular NFSv4.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
NeilBrown [Thu, 14 Nov 2013 02:00:17 +0000 (13:00 +1100)]
NFS: correctly report misuse of "migration" mount option.
The current test on valid use of the "migration" mount option can never
report an error as it will only do so if
mnt->version !=4 && mnt->minor_version != 0
(and some other condition), but if that test would succeed, then the previous
test has already gone-to out_minorversion_mismatch.
So change the && to an || to get correct semantics.
Jeff Layton [Wed, 13 Nov 2013 14:08:21 +0000 (09:08 -0500)]
nfs: don't retry detect_trunking with RPC_AUTH_UNIX more than once
Currently, when we try to mount and get back NFS4ERR_CLID_IN_USE or
NFS4ERR_WRONGSEC, we create a new rpc_clnt and then try the call again.
There is no guarantee that doing so will work however, so we can end up
retrying the call in an infinite loop.
Worse yet, we create the new client using rpc_clone_client_set_auth,
which creates the new client as a child of the old one. Thus, we can end
up with a *very* long lineage of rpc_clnts. When we go to put all of the
references to them, we can end up with a long call chain that can smash
the stack as each rpc_free_client() call can recurse back into itself.
This patch fixes this by simply ensuring that the SETCLIENTID call will
only be retried in this situation if the last attempt did not use
RPC_AUTH_UNIX.
Note too that with this change, we don't need the (i > 2) check in the
-EACCES case since we now have a more reliable test as to whether we
should reattempt.
Trond Myklebust [Tue, 12 Nov 2013 22:24:36 +0000 (17:24 -0500)]
SUNRPC: Avoid deep recursion in rpc_release_client
In cases where an rpc client has a parent hierarchy, then
rpc_free_client may end up calling rpc_release_client() on the
parent, thus recursing back into rpc_free_client. If the hierarchy
is deep enough, then we can get into situations where the stack
simply overflows.
The fix is to have rpc_release_client() loop so that it can take
care of the parent rpc client hierarchy without needing to
recurse.
Trond Myklebust [Fri, 8 Nov 2013 21:03:50 +0000 (16:03 -0500)]
SUNRPC: Fix a data corruption issue when retransmitting RPC calls
The following scenario can cause silent data corruption when doing
NFS writes. It has mainly been observed when doing database writes
using O_DIRECT.
1) The RPC client uses sendpage() to do zero-copy of the page data.
2) Due to networking issues, the reply from the server is delayed,
and so the RPC client times out.
3) The client issues a second sendpage of the page data as part of
an RPC call retransmission.
4) The reply to the first transmission arrives from the server
_before_ the client hardware has emptied the TCP socket send
buffer.
5) After processing the reply, the RPC state machine rules that
the call to be done, and triggers the completion callbacks.
6) The application notices the RPC call is done, and reuses the
pages to store something else (e.g. a new write).
7) The client NIC drains the TCP socket send buffer. Since the
page data has now changed, it reads a corrupted version of the
initial RPC call, and puts it on the wire.
This patch fixes the problem in the following manner:
The ordering guarantees of TCP ensure that when the server sends a
reply, then we know that the _first_ transmission has completed. Using
zero-copy in that situation is therefore safe.
If a time out occurs, we then send the retransmission using sendmsg()
(i.e. no zero-copy), We then know that the socket contains a full copy of
the data, and so it will retransmit a faithful reproduction even if the
RPC call completes, and the application reuses the O_DIRECT buffer in
the meantime.
Trond Myklebust [Mon, 4 Nov 2013 18:46:40 +0000 (13:46 -0500)]
NFSv4.2: Remove redundant checks in nfs_setsecurity+nfs4_label_init_security
We already check for nfs_server_capable(inode, NFS_CAP_SECURITY_LABEL)
in nfs4_label_alloc()
We check the minor version in _nfs4_server_capabilities before setting
NFS_CAP_SECURITY_LABEL.
Trond Myklebust [Mon, 4 Nov 2013 20:20:20 +0000 (15:20 -0500)]
NFSv4: Sanity check the server reply in _nfs4_server_capabilities
We don't want to be setting capabilities and/or requesting attributes
that are not appropriate for the NFSv4 minor version.
- Ensure that we clear the NFS_CAP_SECURITY_LABEL capability when appropriate
- Ensure that we limit the attribute bitmasks to the mounted_on_fileid
attribute and less for NFSv4.0
- Ensure that we limit the attribute bitmasks to suppattr_exclcreat and
less for NFSv4.1
- Ensure that we limit it to change_sec_label or less for NFSv4.2
Trond Myklebust [Mon, 4 Nov 2013 18:23:59 +0000 (13:23 -0500)]
NFSv4.2: encode_readdir - only ask for labels when doing readdirplus
Currently, if the server is doing NFSv4.2 and supports labeled NFS, then
our on-the-wire READDIR request ends up asking for the label information,
which is then ignored unless we're doing readdirplus.
This patch ensures that READDIR doesn't ask the server for label information
at all unless the readdir->bitmask contains the FATTR4_WORD2_SECURITY_LABEL
attribute, and the readdir->plus flag is set.
While we're at it, optimise away the 3rd bitmap field if it is zero.
Jeff Layton [Sat, 2 Nov 2013 10:57:18 +0000 (06:57 -0400)]
nfs: set security label when revalidating inode
Currently, we fetch the security label when revalidating an inode's
attributes, but don't apply it. This is in contrast to the readdir()
codepath where we do apply label changes.
Cc: Dave Quigley <dpquigl@davequigley.com> Signed-off-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The problem is that _nfs4_do_set_security_label calls rpc_call_sync()
directly which fails to do any setup of the SEQUENCE call. Have it use
nfs4_call_sync() instead which does the right thing. While we're at it
change the name of "args" to "arg" to better match the pattern in
_nfs4_do_setattr.
Reported-by: Chao Ye <cye@redhat.com> Cc: David Quigley <dpquigl@davequigley.com> Signed-off-by: Jeff Layton <jlayton@redhat.com> Cc: stable@vger.kernel.org # 3.11+ Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Jeff Layton [Thu, 31 Oct 2013 17:03:04 +0000 (13:03 -0400)]
nfs: fix inverted test for delegation in nfs4_reclaim_open_state
commit 6686390bab6a0e0 (NFS: remove incorrect "Lock reclaim failed!"
warning.) added a test for a delegation before checking to see if any
reclaimed locks failed. The test however is backward and is only doing
that check when a delegation is held instead of when one isn't.
All calls to xs_close are protected by XPRT_LOCKED as are most
activations of the workqueue which runs xs_tcp_setup_socket.
The exception is xs_tcp_schedule_linger_timeout.
So presumably the timeout queued by the later fires exactly when some
other code runs xs_close().
To protect against this we can move the cancel_delayed_work_sync()
call from xs_destory() to xs_close().
As xs_close is never called from the worker scheduled on
->connect_worker, this can never deadlock.
Signed-off-by: NeilBrown <neilb@suse.de>
[Trond: Make it safe to call cancel_delayed_work_sync() on AF_LOCAL sockets] Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Trond Myklebust [Mon, 28 Oct 2013 23:33:36 +0000 (19:33 -0400)]
Merge branch 'fscache' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs into linux-next
Pull fs-cache fixes from David Howells:
Can you pull these commits to fix an issue with NFS whereby caching can be
enabled on a file that is open for writing by subsequently opening it for
reading. This can be made to crash by opening it for writing again if you're
quick enough.
The gist of the patchset is that the cookie should be acquired at inode
creation only and subsequently enabled and disabled as appropriate (which
dispenses with the backing objects when they're not needed).
The extra synchronisation that NFS does can then be dispensed with as it is
thenceforth managed by FS-Cache.
Could you send these on to Linus?
This likely will need fixing also in CIFS and 9P also once the FS-Cache
changes are upstream. AFS and Ceph are probably safe.
* 'fscache' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
NFS: Use i_writecount to control whether to get an fscache cookie in nfs_open()
FS-Cache: Provide the ability to enable/disable cookies
FS-Cache: Add use/unuse/wake cookie wrappers
J. Bruce Fields [Tue, 15 Oct 2013 21:03:16 +0000 (17:03 -0400)]
nfs: use IS_ROOT not DCACHE_DISCONNECTED
This check was added by Al Viro with d9e80b7de91db05c1c4d2e5ebbfd70b3b3ba0e0f "nfs d_revalidate() is too
trigger-happy with d_drop()", with the explanation that we don't want to
remove the root of a disconnected tree, which will still be included on
the s_anon list.
But DCACHE_DISCONNECTED does *not* actually identify dentries that are
disconnected from the dentry tree or hashed on s_anon. IS_ROOT() is the
way to do that.
Also add a comment from Al's commit to remind us why this check is
there.
Signed-off-by: J. Bruce Fields <bfields@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Trond Myklebust [Mon, 28 Oct 2013 22:18:00 +0000 (18:18 -0400)]
SUNRPC: Fix buffer overflow checking in gss_encode_v0_msg/gss_encode_v1_msg
In gss_encode_v1_msg, it is pointless to BUG() after the overflow has
happened. Replace the existing sprintf()-based code with scnprintf(),
and warn if an overflow is ever triggered.
In gss_encode_v0_msg, replace the runtime BUG_ON() with an appropriate
compile-time BUILD_BUG_ON.
Reported-by: Bruce Fields <bfields@fieldses.org> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This patch adds support for multiple security options which can be
specified using a colon-delimited list of security flavors (the same
syntax as nfsd's exports file).
This is useful, for instance, when NFSv4.x mounts cross SECINFO
boundaries. With this patch a user can use "sec=krb5i,krb5p"
to mount a remote filesystem using krb5i, but can still cross
into krb5p-only exports.
New mounts will try all security options before failing. NFSv4.x
SECINFO results will be compared against the sec= flavors to
find the first flavor in both lists or if no match is found will
return -EPERM.
Since the parsed sec= flavor is now stored in nfs_server->auth_info,
we no longer need an nfs_server flag to determine if a sec= option was
used.
This flag has not been completely removed because it is still needed for
the (old but still supported) non-text parsed mount options ABI
compatability.
Chuck Lever [Thu, 17 Oct 2013 18:14:10 +0000 (14:14 -0400)]
NFS: Fix possible endless state recovery wait
In nfs4_wait_clnt_recover(), hold a reference to the clp being
waited on. The state manager can reduce clp->cl_count to 1, in
which case the nfs_put_client() in nfs4_run_state_manager() can
free *clp before wait_on_bit() returns and allows
nfs4_wait_clnt_recover() to run again.
The behavior at that point is non-deterministic. If the waited-on
bit still happens to be zero, wait_on_bit() will wake the waiter as
expected. If the bit is set again (say, if the memory was poisoned
when freed) wait_on_bit() can leave the waiter asleep.
This is a narrow fix which ensures the safety of accessing *clp in
nfs4_wait_clnt_recover(), but does not address the continued use
of a possibly freed *clp after nfs4_wait_clnt_recover() returns
(see nfs_end_delegation_return(), for example).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:14:04 +0000 (14:14 -0400)]
NFS: Set EXCHGID4_FLAG_SUPP_MOVED_MIGR
Broadly speaking, v4.1 migration is untested. There are no servers
in the wild that support NFSv4.1 migration. However, as server
implementations become available, we do want to enable testing by
developers, while leaving it disabled for environments for which
broken migration support would be an unpleasant surprise.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:13:58 +0000 (14:13 -0400)]
NFS: Handle SEQ4_STATUS_LEASE_MOVED
With the advent of NFSv4 sessions in NFSv4.1 and following, a "lease
moved" condition is reported differently than it is in NFSv4.0.
NFSv4 minor version 0 servers return an error status code,
NFS4ERR_LEASE_MOVED, to signal that a lease has moved. This error
causes the whole compound operation to fail. Normal compounds
against this server continue to fail until the client performs
migration recovery on the migrated share.
Minor version 1 and later servers assert a bit flag in the reply to
a compound's SEQUENCE operation to signal LEASE_MOVED. This is not
a fatal condition: operations against this server continue normally.
The server asserts this flag until the client performs migration
recovery on the migrated share.
Note that servers MUST NOT return NFS4ERR_LEASE_MOVED to NFSv4
clients not using NFSv4.0.
After the server asserts any of the sr_status_flags in the SEQUENCE
operation in a typical compound, our client initiates standard lease
recovery. For NFSv4.1+, a stand-alone SEQUENCE operation is
performed to discover what recovery is needed.
If SEQ4_STATUS_LEASE_MOVED is asserted in this stand-alone SEQUENCE
operation, our client attempts to discover which FSIDs have been
migrated, and then performs migration recovery on each.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:13:53 +0000 (14:13 -0400)]
NFS: Handle NFS4ERR_LEASE_MOVED during async RENEW
With NFSv4 minor version 0, the asynchronous lease RENEW
heartbeat can return NFS4ERR_LEASE_MOVED. Error recovery logic for
async RENEW is a separate code path from the generic NFS proc paths,
so it must be updated to handle NFS4ERR_LEASE_MOVED as well.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:13:47 +0000 (14:13 -0400)]
NFS: Migration support for RELEASE_LOCKOWNER
Currently the Linux NFS client ignores the operation status code for
the RELEASE_LOCKOWNER operation. Like NFSv3's UMNT operation,
RELEASE_LOCKOWNER is a courtesy to help servers manage their
resources, and the outcome is not consequential for the client.
During a migration, a server may report NFS4ERR_LEASE_MOVED, in
which case the client really should retry, since typically
LEASE_MOVED has nothing to do with the current operation, but does
prevent it from going forward.
Also, it's important for a client to respond as soon as possible to
a moved lease condition, since the client's lease could expire on
the destination without further action by the client.
NFS4ERR_DELAY is not included in the list of valid status codes for
RELEASE_LOCKOWNER in RFC 3530bis. However, rfc3530-migration-update
does permit migration-capable servers to return DELAY to clients,
but only in the context of an ongoing migration. In this case the
server has frozen lock state in preparation for migration, and a
client retry would help the destination server purge unneeded state
once migration recovery is complete.
Interestly, NFS4ERR_MOVED is not valid for RELEASE_LOCKOWNER, even
though lock owners can be migrated with Transparent State Migration.
Note that RFC 3530bis section 9.5 includes RELEASE_LOCKOWNER in the
list of operations that renew a client's lease on the server if they
succeed. Now that our client pays attention to the operation's
status code, we can note that renewal appropriately.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:13:35 +0000 (14:13 -0400)]
NFS: Support NFS4ERR_LEASE_MOVED recovery in state manager
A migration on the FSID in play for the current NFS operation
is reported via the error status code NFS4ERR_MOVED.
"Lease moved" means that a migration has occurred on some other
FSID than the one for the current operation. It's a signal that
the client should take action immediately to handle a migration
that it may not have noticed otherwise. This is so that the
client's lease does not expire unnoticed on the destination server.
In NFSv4.0, a moved lease is reported with the NFS4ERR_LEASE_MOVED
error status code.
To recover from NFS4ERR_LEASE_MOVED, check each FSID for that server
to see if it is still present. Invoke nfs4_try_migration() if the
FSID is no longer present on the server.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:13:30 +0000 (14:13 -0400)]
NFS: Add method to detect whether an FSID is still on the server
Introduce a mechanism for probing a server to determine if an FSID
is present or absent.
The on-the-wire compound is different between minor version 0 and 1.
Minor version 0 appends a RENEW operation to identify which client
ID is probing. Minor version 1 has a SEQUENCE operation in the
compound which effectively carries the same information.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:12:50 +0000 (14:12 -0400)]
NFS: Add method to retrieve fs_locations during migration recovery
The nfs4_proc_fs_locations() function is invoked during referral
processing to perform a GETATTR(fs_locations) on an object's parent
directory in order to discover the target of the referral. It
performs a LOOKUP in the compound, so the client needs to know the
parent's file handle a priori.
Unfortunately this function is not adequate for handling migration
recovery. We need to probe fs_locations information on an FSID, but
there's no parent directory available for many operations that
can return NFS4ERR_MOVED.
Another subtlety: recovering from NFS4ERR_LEASE_MOVED is a process
of walking over a list of known FSIDs that reside on the server, and
probing whether they have migrated. Once the server has detected
that the client has probed all migrated file systems, it stops
returning NFS4ERR_LEASE_MOVED.
A minor version zero server needs to know what client ID is
requesting fs_locations information so it can clear the flag that
forces it to continue returning NFS4ERR_LEASE_MOVED. This flag is
set per client ID and per FSID. However, the client ID is not an
argument of either the PUTFH or GETATTR operations. Later minor
versions have client ID information embedded in the compound's
SEQUENCE operation.
Therefore, by convention, minor version zero clients send a RENEW
operation in the same compound as the GETATTR(fs_locations), since
RENEW's one argument is a clientid4. This allows a minor version
zero server to identify correctly the client that is probing for a
migration.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:12:34 +0000 (14:12 -0400)]
NFS: Add functions to swap transports during migration recovery
Introduce functions that can walk through an array of returned
fs_locations information and connect a transport to one of the
destination servers listed therein.
Note that NFS minor version 1 introduces "fs_locations_info" which
extends the locations array sorting criteria available to clients.
This is not supported yet.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:12:28 +0000 (14:12 -0400)]
NFS: Add nfs4_update_server
New function nfs4_update_server() moves an nfs_server to a different
nfs_client. This is done as part of migration recovery.
Though it may be appealing to think of them as the same thing,
migration recovery is not the same as following a referral.
For a referral, the client has not descended into the file system
yet: it has no nfs_server, no super block, no inodes or open state.
It is enough to simply instantiate the nfs_server and super block,
and perform a referral mount.
For a migration, however, we have all of those things already, and
they have to be moved to a different nfs_client. No local namespace
changes are needed here.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Chuck Lever [Thu, 17 Oct 2013 18:12:17 +0000 (14:12 -0400)]
SUNRPC: Modify synopsis of rpc_client_register()
The rpc_client_register() helper was added in commit e73f4cc0,
"SUNRPC: split client creation routine into setup and registration,"
Mon Jun 24 11:52:52 2013. In a subsequent patch, I'd like to invoke
rpc_client_register() from a context where a struct rpc_create_args
is not available.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cached opens have already been handled by _nfs4_opendata_reclaim_to_nfs4_state
and can safely skip being reprocessed, but must still call update_open_stateid
to make sure that all active fmodes are recovered.
Trond Myklebust [Mon, 28 Oct 2013 18:57:12 +0000 (14:57 -0400)]
NFSv4: Fix state reference counting in _nfs4_opendata_reclaim_to_nfs4_state
Currently, if the call to nfs_refresh_inode fails, then we end up leaking
a reference count, due to the call to nfs4_get_open_state.
While we're at it, replace nfs4_get_open_state with a simple call to
atomic_inc(); there is no need to do a full lookup of the struct nfs_state
since it is passed as an argument in the struct nfs4_opendata, and
is already assigned to the variable 'state'.
_nfs4_opendata_reclaim_to_nfs4_state doesn't expect to see a cached
open CLAIM_PREVIOUS, but this can happen. An example is when there are
RDWR openers and RDONLY openers on a delegation stateid. The recovery
path will first try an open CLAIM_PREVIOUS for the RDWR openers, this
marks the delegation as not needing RECLAIM anymore, so the open
CLAIM_PREVIOUS for the RDONLY openers will not actually send an rpc.
The NULL dereference is due to _nfs4_opendata_reclaim_to_nfs4_state
returning PTR_ERR(rpc_status) when !rpc_done. When the open is
cached, rpc_done == 0 and rpc_status == 0, thus
_nfs4_opendata_reclaim_to_nfs4_state returns NULL - this is unexpected
by callers of nfs4_opendata_to_nfs4_state().
This can be reproduced easily by opening the same file two times on an
NFSv4.0 mount with delegations enabled, once as RDWR and once as RDONLY then
sleeping for a long time. While the files are held open, kick off state
recovery and this NULL dereference will be hit every time.
Trond Myklebust [Mon, 28 Oct 2013 18:50:38 +0000 (14:50 -0400)]
NFSv4.1: Don't change the security label as part of open reclaim.
The current caching model calls for the security label to be set on
first lookup and/or on any subsequent label changes. There is no
need to do it as part of an open reclaim.
SUNRPC: call_connect_status should recheck bind and connect status on error
Currently, we go directly to call_transmit which sends us to call_status
on error. If we know that the connect attempt failed, we should rather
just jump straight back to call_bind and call_connect.
SUNRPC: Don't set the request connect_cookie until a successful transmit
We're using the request connect_cookie to track whether or not a
request was successfully transmitted on the current transport
connection or not. For that reason we should ensure that it is
only set after we've successfully transmitted the request.
SUNRPC: Enable the keepalive option for TCP sockets
For NFSv4 we want to avoid retransmitting RPC calls unless the TCP
connection breaks. However we still want to detect TCP connection
breakage as soon as possible. Do this by setting the keepalive option
with the idle timeout and count set to the 'timeo' and 'retrans' mount
options.
The problem is that we don't clear the request->fl_ops after the first
try and so when we retry, nfs4_set_lock_state() exits early without
setting the lock stateid.
Regression introduced by commit 70cc6487a4e08b8698c0e2ec935fb48d10490162
(locks: make ->lock release private data before returning in GETLK case)
Linus Torvalds [Tue, 1 Oct 2013 00:10:26 +0000 (17:10 -0700)]
Merge tag 'nfs-for-3.12-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client bugfixes from Trond Myklebust:
- Stable fix for Oopses in the pNFS files layout driver
- Fix a regression when doing a non-exclusive file create on NFSv4.x
- NFSv4.1 security negotiation fixes when looking up the root
filesystem
- Fix a memory ordering issue in the pNFS files layout driver
* tag 'nfs-for-3.12-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
NFS: Give "flavor" an initial value to fix a compile warning
NFSv4.1: try SECINFO_NO_NAME flavs until one works
NFSv4.1: Ensure memory ordering between nfs4_ds_connect and nfs4_fl_prepare_ds
NFSv4.1: nfs4_fl_prepare_ds - fix bugs when the connect attempt fails
NFSv4: Honour the 'opened' parameter in the atomic_open() filesystem method
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (22 commits)
pidns: fix free_pid() to handle the first fork failure
ipc,msg: prevent race with rmid in msgsnd,msgrcv
ipc/sem.c: update sem_otime for all operations
mm/hwpoison: fix the lack of one reference count against poisoned page
mm/hwpoison: fix false report on 2nd attempt at page recovery
mm/hwpoison: fix test for a transparent huge page
mm/hwpoison: fix traversal of hugetlbfs pages to avoid printk flood
block: change config option name for cmdline partition parsing
mm/mlock.c: prevent walking off the end of a pagetable in no-pmd configuration
mm: avoid reinserting isolated balloon pages into LRU lists
arch/parisc/mm/fault.c: fix uninitialized variable usage
include/asm-generic/vtime.h: avoid zero-length file
nilfs2: fix issue with race condition of competition between segments for dirty blocks
Documentation/kernel-parameters.txt: replace kernelcore with Movable
mm/bounce.c: fix a regression where MS_SNAP_STABLE (stable pages snapshotting) was ignored
kernel/kmod.c: check for NULL in call_usermodehelper_exec()
ipc/sem.c: synchronize the proc interface
ipc/sem.c: optimize sem_lock()
ipc/sem.c: fix race in sem_lock()
mm/compaction.c: periodically schedule when freeing pages
...
This fixes a race in both msgrcv() and msgsnd() between finding the msg
and actually dealing with the queue, as another thread can delete shmid
underneath us if we are preempted before acquiring the
kern_ipc_perm.lock.
Manfred illustrates this nicely:
Assume a preemptible kernel that is preempted just after
msq = msq_obtain_object_check(ns, msqid)
in do_msgrcv(). The only lock that is held is rcu_read_lock().
Now the other thread processes IPC_RMID. When the first task is
resumed, then it will happily wait for messages on a deleted queue.
Fix this by checking for if the queue has been deleted after taking the
lock.
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com> Reported-by: Manfred Spraul <manfred@colorfullife.com> Cc: Rik van Riel <riel@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: <stable@vger.kernel.org> [3.11] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process out of the
spinlock section"), the update of semaphore's sem_otime(last semop time)
was moved to one central position (do_smart_update).
But since do_smart_update() is only called for operations that modify
the array, this means that wait-for-zero semops do not update sem_otime
anymore.
The fix is simple:
Non-alter operations must update sem_otime.
[akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Manfred Spraul <manfred@colorfullife.com> Reported-by: Jia He <jiakernel@gmail.com> Tested-by: Jia He <jiakernel@gmail.com> Cc: Davidlohr Bueso <davidlohr.bueso@hp.com> Cc: Mike Galbraith <efault@gmx.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Wanpeng Li [Mon, 30 Sep 2013 20:45:24 +0000 (13:45 -0700)]
mm/hwpoison: fix the lack of one reference count against poisoned page
The lack of one reference count against poisoned page for hwpoison_inject
w/o hwpoison_filter enabled result in hwpoison detect -1 users still
referenced the page, however, the number should be 0 except the poison
handler held one after successfully unmap. This patch fix it by hold one
referenced count against poisoned page for hwpoison_inject w/ and w/o
hwpoison_filter enabled.
Before patch:
[ 71.902112] Injecting memory failure at pfn 224706
[ 71.902137] MCE 0x224706: dirty LRU page recovery: Failed
[ 71.902138] MCE 0x224706: dirty LRU page still referenced by -1 users
Wanpeng Li [Mon, 30 Sep 2013 20:45:23 +0000 (13:45 -0700)]
mm/hwpoison: fix false report on 2nd attempt at page recovery
If the page is poisoned by software injection w/ MF_COUNT_INCREASED
flag, there is a false report during the 2nd attempt at page recovery
which is not truthful.
This patch fixes it by reporting the first attempt to try free buddy
page recovery if MF_COUNT_INCREASED is set.
Wanpeng Li [Mon, 30 Sep 2013 20:45:21 +0000 (13:45 -0700)]
mm/hwpoison: fix traversal of hugetlbfs pages to avoid printk flood
madvise_hwpoison won't check if the page is small page or huge page and
traverses in small page granularity against the range unconditionally,
which result in a printk flood "MCE xxx: already hardware poisoned" if
the page is a huge page.
This patch fixes it by using compound_order(compound_head(page)) for
huge page iterator.
Paul Gortmaker [Mon, 30 Sep 2013 20:45:19 +0000 (13:45 -0700)]
block: change config option name for cmdline partition parsing
Recently commit bab55417b10c ("block: support embedded device command
line partition") introduced CONFIG_CMDLINE_PARSER. However, that name
is too generic and sounds like it enables/disables generic kernel boot
arg processing, when it really is block specific.
Before this option becomes a part of a full/final release, add the BLK_
prefix to it so that it is clear in absence of any other context that it
is block specific.
In addition, fix up the following less critical items:
- help text was not really at all helpful.
- index file for Documentation was not updated
- add the new arg to Documentation/kernel-parameters.txt
- clarify wording in source comments
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Cai Zhiyong <caizhiyong@huawei.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/mlock.c: prevent walking off the end of a pagetable in no-pmd configuration
The function __munlock_pagevec_fill() introduced in commit 7a8010cd3627
("mm: munlock: manual pte walk in fast path instead of
follow_page_mask()") uses pmd_addr_end() for restricting its operation
within current page table.
This is insufficient on architectures/configurations where pmd is folded
and pmd_addr_end() just returns the end of the full range to be walked.
In this case, it allows pte++ to walk off the end of a page table
resulting in unpredictable behaviour.
This patch fixes the function by using pgd_addr_end() and pud_addr_end()
before pmd_addr_end(), which will yield correct page table boundary on
all configurations. This is similar to what existing page walkers do
when walking each level of the page table.
Additionaly, the patch clarifies a comment for get_locked_pte() call in the
function.
Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Reported-by: Fengguang Wu <fengguang.wu@intel.com> Reviewed-by: Bob Liu <bob.liu@oracle.com> Cc: Jörn Engel <joern@logfs.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Michel Lespinasse <walken@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@suse.cz> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Rafael Aquini [Mon, 30 Sep 2013 20:45:16 +0000 (13:45 -0700)]
mm: avoid reinserting isolated balloon pages into LRU lists
Isolated balloon pages can wrongly end up in LRU lists when
migrate_pages() finishes its round without draining all the isolated
page list.
The same issue can happen when reclaim_clean_pages_from_list() tries to
reclaim pages from an isolated page list, before migration, in the CMA
path. Such balloon page leak opens a race window against LRU lists
shrinkers that leads us to the following kernel panic:
This patch fixes the issue, by assuring the proper tests are made at
putback_movable_pages() & reclaim_clean_pages_from_list() to avoid
isolated balloon pages being wrongly reinserted in LRU lists.
But such error messages are consequence of file system's issue that takes
place more earlier. Fortunately, Jerome Poulin <jeromepoulin@gmail.com>
and Anton Eliasson <devel@antoneliasson.se> were reported about another
issue not so recently. These reports describe the issue with segctor
thread's crash:
BUG: unable to handle kernel paging request at 0000000000004c83
IP: nilfs_end_page_io+0x12/0xd0 [nilfs2]
These two issues have one reason. This reason can raise third issue
too. Third issue results in hanging of segctor thread with eating of
100% CPU.
REPRODUCING PATH:
One of the possible way or the issue reproducing was described by
Jermoe me Poulin <jeromepoulin@gmail.com>:
1. init S to get to single user mode.
2. sysrq+E to make sure only my shell is running
3. start network-manager to get my wifi connection up
4. login as root and launch "screen"
5. cd /boot/log/nilfs which is a ext3 mount point and can log when NILFS dies.
6. lscp | xz -9e > lscp.txt.xz
7. mount my snapshot using mount -o cp=3360839,ro /dev/vgUbuntu/root /mnt/nilfs
8. start a screen to dump /proc/kmsg to text file since rsyslog is killed
9. start a screen and launch strace -f -o find-cat.log -t find
/mnt/nilfs -type f -exec cat {} > /dev/null \;
10. start a screen and launch strace -f -o apt-get.log -t apt-get update
11. launch the last command again as it did not crash the first time
12. apt-get crashes
13. ps aux > ps-aux-crashed.log
13. sysrq+W
14. sysrq+E wait for everything to terminate
15. sysrq+SUSB
Simplified way of the issue reproducing is starting kernel compilation
task and "apt-get update" in parallel.
REPRODUCIBILITY:
The issue is reproduced not stable [60% - 80%]. It is very important to
have proper environment for the issue reproducing. The critical
conditions for successful reproducing:
(1) It should have big modified file by mmap() way.
(2) This file should have the count of dirty blocks are greater that
several segments in size (for example, two or three) from time to time
during processing.
(3) It should be intensive background activity of files modification
in another thread.
INVESTIGATION:
First of all, it is possible to see that the reason of crash is not valid
page address:
Moreover, value of b_page (0x1a82) is 6786. This value looks like segment
number. And b_blocknr with b_size values look like block numbers. So,
buffer_head's pointer points on not proper address value.
Detailed investigation of the issue is discovered such picture:
Usually, for every segment we collect dirty files in list. Then, dirty
blocks are gathered for every dirty file, prepared for write and
submitted by means of nilfs_segbuf_submit_bh() call. Finally, it takes
place complete write phase after calling nilfs_end_bio_write() on the
block layer. Buffers/pages are marked as not dirty on final phase and
processed files removed from the list of dirty files.
It is possible to see that we had three prepare_write and submit_bio
phases before segbuf_wait and complete_write phase. Moreover, segments
compete between each other for dirty blocks because on every iteration
of segments processing dirty buffer_heads are added in several lists of
payload_buffers:
The next pointer is the same but prev pointer has changed. It means
that buffer_head has next pointer from one list but prev pointer from
another. Such modification can be made several times. And, finally, it
can be resulted in various issues: (1) segctor hanging, (2) segctor
crashing, (3) file system metadata corruption.
FIX:
This patch adds:
(1) setting of BH_Async_Write flag in nilfs_segctor_prepare_write()
for every proccessed dirty block;
(2) checking of BH_Async_Write flag in
nilfs_lookup_dirty_data_buffers() and
nilfs_lookup_dirty_node_buffers();
(3) clearing of BH_Async_Write flag in nilfs_segctor_complete_write(),
nilfs_abort_logs(), nilfs_forget_buffer(), nilfs_clear_dirty_page().
Reported-by: Jerome Poulin <jeromepoulin@gmail.com> Reported-by: Anton Eliasson <devel@antoneliasson.se> Cc: Paul Fertser <fercerpav@gmail.com> Cc: ARAI Shun-ichi <hermes@ceres.dti.ne.jp> Cc: Piotr Szymaniak <szarpaj@grubelek.pl> Cc: Juan Barry Manuel Canham <Linux@riotingpacifist.net> Cc: Zahid Chowdhury <zahid.chowdhury@starsolutions.com> Cc: Elmer Zhang <freeboy6716@gmail.com> Cc: Kenneth Langga <klangga@gmail.com> Signed-off-by: Vyacheslav Dubeyko <slava@dubeyko.com> Acked-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Darrick J. Wong [Mon, 30 Sep 2013 20:45:09 +0000 (13:45 -0700)]
mm/bounce.c: fix a regression where MS_SNAP_STABLE (stable pages snapshotting) was ignored
The "force" parameter in __blk_queue_bounce was being ignored, which
means that stable page snapshots are not always happening (on ext3).
This of course leads to DIF disks reporting checksum errors, so fix this
regression.
The regression was introduced in commit 6bc454d15004 ("bounce: Refactor
__blk_queue_bounce to not use bi_io_vec")
Reported-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Cc: Kent Overstreet <koverstreet@google.com> Cc: <stable@vger.kernel.org> [3.10+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/kmod.c: check for NULL in call_usermodehelper_exec()
If /proc/sys/kernel/core_pattern contains only "|", a NULL pointer
dereference happens upon core dump because argv_split("") returns
argv[0] == NULL.
This bug was once fixed by commit 264b83c07a84 ("usermodehelper: check
subprocess_info->path != NULL") but was by error reintroduced by commit 7f57cfa4e2aa ("usermodehelper: kill the sub_info->path[0] check").
This bug seems to exist since 2.6.19 (the version which core dump to
pipe was added). Depending on kernel version and config, some side
effect might happen immediately after this oops (e.g. kernel panic with
2.6.32-358.18.1.el6).
The proc interface is not aware of sem_lock(), it instead calls
ipc_lock_object() directly. This means that simple semop() operations
can run in parallel with the proc interface. Right now, this is
uncritical, because the implementation doesn't do anything that requires
a proper synchronization.
But it is dangerous and therefore should be fixed.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com> Cc: Davidlohr Bueso <davidlohr.bueso@hp.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Operations that need access to the whole array must guarantee that there
are no simple operations ongoing. Right now this is achieved by
spin_unlock_wait(sem->lock) on all semaphores.
If complex_count is nonzero, then this spin_unlock_wait() is not
necessary, because it was already performed in the past by the thread
that increased complex_count and even though sem_perm.lock was dropped
inbetween, no simple operation could have started, because simple
operations cannot start when complex_count is non-zero.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com> Cc: Mike Galbraith <bitbucket@online.de> Cc: Rik van Riel <riel@redhat.com> Reviewed-by: Davidlohr Bueso <davidlohr@hp.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>