GitHub/LineageOS/G12/android_kernel_amlogic_linux-4.9.git
13 years agoNFS: Ensure that we update the readdir filp->f_pos correctly
Trond Myklebust [Wed, 23 Mar 2011 12:43:09 +0000 (08:43 -0400)]
NFS: Ensure that we update the readdir filp->f_pos correctly

If we're doing a search by readdir cookie, we need to ensure that the
resulting f_pos is updated. To do so, we need to update the
desc->current_index, in the same way that we do in the search by
file offset case.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoRemove pointless memset in nfsacl_encode()
Jesper Juhl [Mon, 21 Mar 2011 19:35:49 +0000 (20:35 +0100)]
Remove pointless memset in nfsacl_encode()

Remove pointless memset() in nfsacl_encode().

Thanks to Trond Myklebust <Trond.Myklebust@netapp.com> for pointing out
that it is not needed since posix_acl_init() will set everything
regardless..

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agonfs4: Fix NULL dereference at d_alloc_and_lookup()
Gusev Vitaliy [Tue, 22 Mar 2011 21:40:25 +0000 (00:40 +0300)]
nfs4: Fix NULL dereference at d_alloc_and_lookup()

d_alloc_and_lookup() calls i_op->lookup method due to
rootfh changes his fsid.

During mount i_op of NFS root inode is set to
nfs_mountpoint_inode_operations, if rpc_ops->getroot()
and rpc_ops->getattr() return different fsid.

After that  nfs_follow_remote_path() raised oops:

   BUG: unable to handle kernel NULL pointer dereference at (null)
   IP: [<          (null)>]           (null)

stack trace:

     d_alloc_and_lookup+0x4c/0x74
     do_lookup+0x1e3/0x280
     link_path_walk+0x12e/0xab0
     nfs4_remote_get_sb+0x56/0x2c0 [nfs]
     path_walk+0x67/0xe0
     vfs_path_lookup+0x8e/0x100
     nfs_follow_remote_path+0x16f/0x3e0 [nfs]
     nfs4_try_mount+0x6f/0xd0 [nfs]
     nfs_get_sb+0x269/0x400 [nfs]
     vfs_kern_mount+0x8a/0x1f0
     do_kern_mount+0x52/0x130
     do_mount+0x20a/0x260
     sys_mount+0x90/0xe0
     system_call_fastpath+0x16/0x1b

So just refresh fsid, as RFC3530 doesn't specify behavior
in case of rootfh changes fsid.

Signed-off-by: Vitaliy Gusev <gusev.vitaliy@nexenta.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Never reuse the socket port after an xs_close()
Trond Myklebust [Tue, 22 Mar 2011 22:40:10 +0000 (18:40 -0400)]
SUNRPC: Never reuse the socket port after an xs_close()

If we call xs_close(), we're in one of two situations:
 - Autoclose, which means we don't expect to resend a request
 - bind+connect failed, which probably means the port is in use

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org
13 years agoNFS: Fix a hang/infinite loop in nfs_wb_page()
Trond Myklebust [Mon, 21 Mar 2011 19:37:01 +0000 (15:37 -0400)]
NFS: Fix a hang/infinite loop in nfs_wb_page()

When one of the two waits in nfs_commit_inode() is interrupted, it
returns a non-negative value, which causes nfs_wb_page() to think
that the operation was successful causing it to busy-loop rather
than exiting.
It also causes nfs_file_fsync() to incorrectly report the file as
being successfully committed to disk.

This patch fixes both problems by ensuring that we return an error
if the attempts to wait fail.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org
13 years agoFS: Use stable writes when not doing a bulk flush
Trond Myklebust [Mon, 21 Mar 2011 21:02:00 +0000 (17:02 -0400)]
FS: Use stable writes when not doing a bulk flush

If we're only doing a single write, and there are no other unstable
writes being queued up, we might want to just flip to using a stable
write RPC call.

Reviewed-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoRPC: killing RPC tasks races fixed
Stanislav Kinsbursky [Thu, 17 Mar 2011 15:54:23 +0000 (18:54 +0300)]
RPC: killing RPC tasks races fixed

RPC task RPC_TASK_QUEUED bit is set must be checked before trying to wake up
task rpc_killall_tasks() because task->tk_waitqueue can not be set (equal to
NULL).
Also, as Trond Myklebust mentioned, such approach (instead of checking
tk_waitqueue to NULL) allows us to "optimise away the call to
rpc_wake_up_queued_task() altogether for those
tasks that aren't queued".

Here is an example of dereferencing of tk_waitqueue equal to NULL:

CPU 0                CPU 1 CPU 2
-------------------- --------------------- --------------------------
nfs4_run_open_task
rpc_run_task
rpc_execute
rpc_set_active
rpc_make_runnable
(waiting)
rpc_async_schedule
nfs4_open_prepare
nfs_wait_on_sequence
nfs_umount_begin
rpc_killall_tasks
rpc_wake_up_task
rpc_wake_up_queued_task
spin_lock(tk_waitqueue == NULL)
BUG()
rpc_sleep_on
spin_lock(&q->lock)
__rpc_sleep_on
task->tk_waitqueue = q

Signed-off-by: Stanislav Kinsbursky <skinsbursky@openvz.org>
Cc: stable@kernel.org
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoxprt: remove redundant check
j223yang@asset.uwaterloo.ca [Wed, 16 Mar 2011 15:16:22 +0000 (11:16 -0400)]
xprt: remove redundant check

remove redundant check.

Signed-off-by: Jinqiu Yang <crindy646@gmail.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Convert struct rpc_xprt to use atomic_t counters
Trond Myklebust [Tue, 15 Mar 2011 23:56:30 +0000 (19:56 -0400)]
SUNRPC: Convert struct rpc_xprt to use atomic_t counters

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Ensure we always run the tk_callback before tk_action
Trond Myklebust [Tue, 15 Mar 2011 23:56:30 +0000 (19:56 -0400)]
SUNRPC: Ensure we always run the tk_callback before tk_action

This fixes a race in which the task->tk_callback() puts the rpc_task
to sleep, setting a new callback. Under certain circumstances, the current
code may end up executing the task->tk_action before it gets round to the
callback.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org
13 years agosunrpc: fix printk format warning
Randy Dunlap [Wed, 16 Mar 2011 00:11:59 +0000 (17:11 -0700)]
sunrpc: fix printk format warning

Fix printk format build warning:

net/sunrpc/xprtrdma/verbs.c:1463: warning: format '%llx' expects type 'long long unsigned int', but argument 3 has type 'dma_addr_t'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoxprt: remove redundant null check
j223yang@asset.uwaterloo.ca [Thu, 10 Mar 2011 17:40:28 +0000 (12:40 -0500)]
xprt: remove redundant null check

'req' is dereferenced before checked for NULL.
The patch simply removes the check.

Signed-off-by: Jinqiu Yang<crindy646@gmail.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agonfs: BKL is no longer needed, so remove the include
Stephen Rothwell [Tue, 15 Mar 2011 07:37:09 +0000 (18:37 +1100)]
nfs: BKL is no longer needed, so remove the include

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFS: Fix a warning in fs/nfs/idmap.c
Trond Myklebust [Mon, 14 Mar 2011 22:20:01 +0000 (18:20 -0400)]
NFS: Fix a warning in fs/nfs/idmap.c

Commit 45a52a02072b2a7e265f024cfdb00127e08dd9f2 (NFS move nfs_client
initialization into nfs_get_client) introduces a new warning in
fs/nfs/idmap.c:

‘struct rpc_timeout’ declared inside parameter list

Fix it by adding a forward declaration for the struct rpc_timeout
in include/linux/nfs_xdr.h

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoCleanup: Factor out some cut-and-paste code.
Rob Landley [Wed, 9 Mar 2011 22:02:37 +0000 (16:02 -0600)]
Cleanup: Factor out some cut-and-paste code.

Factor out some cut-and-paste code in options parsing.
Saves about 800 bytes on x86-64.

Signed-off-by: Rob Landley <rlandley@parallels.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agocleanup: save 60 lines/100 bytes by combining two mostly duplicate functions.
Rob Landley [Wed, 9 Mar 2011 21:54:13 +0000 (15:54 -0600)]
cleanup: save 60 lines/100 bytes by combining two mostly duplicate functions.

Eliminate two mostly duplicate functions (nfs_parse_simple_hostname()
and nfs_parse_protected_hostname()) and instead just make the calling
function (nfs_parse_devname()) do everything.

Signed-off-by: Rob Landley <rlandley@parallels.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFS: account direct-io into task io accounting
Konstantin Khlebnikov [Mon, 21 Feb 2011 21:28:34 +0000 (00:28 +0300)]
NFS: account direct-io into task io accounting

Account NFS direct-io reads and writes into Task I/O Accounting.
Do it before complition to handle aio.

NFS have unusual direct-io implementation,
thus accounting in generic code does not work.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agogss:krb5 only include enctype numbers in gm_upcall_enctypes
Kevin Coffman [Thu, 3 Mar 2011 00:51:41 +0000 (00:51 +0000)]
gss:krb5 only include enctype numbers in gm_upcall_enctypes

Make the value in gm_upcall_enctypes just the enctype values.
This allows the values to be used more easily elsewhere.

Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoRPCRDMA: Fix FRMR registration/invalidate handling.
Tom Tucker [Wed, 9 Feb 2011 19:45:34 +0000 (19:45 +0000)]
RPCRDMA: Fix FRMR registration/invalidate handling.

When the rpc_memreg_strategy is 5, FRMR are used to map RPC data.
This mode uses an FRMR to map the RPC data, then invalidates
(i.e. unregisers) the data in xprt_rdma_free. These FRMR are used
across connections on the same mount, i.e. if the connection goes
away on an idle timeout and reconnects later, the FRMR are not
destroyed and recreated.

This creates a problem for transport errors because the WR that
invalidate an FRMR may be flushed (i.e. fail) leaving the
FRMR valid. When the FRMR is later used to map an RPC it will fail,
tearing down the transport and starting over. Over time, more and
more of the FRMR pool end up in the wrong state resulting in
seemingly random disconnects.

This fix keeps track of the FRMR state explicitly by setting it's
state based on the successful completion of a reg/inv WR. If the FRMR
is ever used and found to be in the wrong state, an invalidate WR
is prepended, re-syncing the FRMR state and avoiding the connection loss.

Signed-off-by: Tom Tucker <tom@ogc.us>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoRPCRDMA: Fix to XDR page base interpretation in marshalling logic.
Tom Tucker [Wed, 9 Feb 2011 19:45:28 +0000 (19:45 +0000)]
RPCRDMA: Fix to XDR page base interpretation in marshalling logic.

The RPCRDMA marshalling logic assumed that xdr->page_base was an
offset into the first page of xdr->page_list. It is in fact an
offset into the xdr->page_list itself, that is, it selects the
first page in the page_list and the offset into that page.

The symptom depended in part on the rpc_memreg_strategy, if it was
FRMR, or some other one-shot mapping mode, the connection would get
torn down on a base and bounds error. When the badly marshalled RPC
was retransmitted it would reconnect, get the error, and tear down the
connection again in a loop forever. This resulted in a hung-mount. For
the other modes, it would result in silent data corruption. This bug is
most easily reproduced by writing more data than the filesystem
has space for.

This fix corrects the page_base assumption and otherwise simplifies
the iov mapping logic.

Signed-off-by: Tom Tucker <tom@ogc.us>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: Send unmapped uid/gids to the server when using auth_sys
Trond Myklebust [Tue, 22 Feb 2011 23:44:32 +0000 (15:44 -0800)]
NFSv4: Send unmapped uid/gids to the server when using auth_sys

The new behaviour is enabled using the new module parameter
'nfs4_disable_idmapping'.

Note that if the server rejects an unmapped uid or gid, then
the client will automatically switch back to using the idmapper.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: Propagate the error NFS4ERR_BADOWNER to nfs4_do_setattr
Trond Myklebust [Tue, 22 Feb 2011 23:44:31 +0000 (15:44 -0800)]
NFSv4: Propagate the error NFS4ERR_BADOWNER to nfs4_do_setattr

This will be required in order to switch uid/gid mapping back on if the
admin has tried to disable it.

Note that we also propagate NFS4ERR_BADNAME at the same time, in order to
work around a Linux server bug.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: cleanup idmapper functions to take an nfs_server argument
Trond Myklebust [Tue, 22 Feb 2011 23:44:31 +0000 (15:44 -0800)]
NFSv4: cleanup idmapper functions to take an nfs_server argument

...instead of the nfs_client.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: Send unmapped uid/gids to the server if the idmapper fails
Trond Myklebust [Tue, 22 Feb 2011 23:44:31 +0000 (15:44 -0800)]
NFSv4: Send unmapped uid/gids to the server if the idmapper fails

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: If the server sends us a numeric uid/gid then accept it
Trond Myklebust [Tue, 22 Feb 2011 23:44:31 +0000 (15:44 -0800)]
NFSv4: If the server sends us a numeric uid/gid then accept it

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: reject zero layout with zeroed stripe unit
Benny Halevy [Tue, 22 Feb 2011 23:56:01 +0000 (15:56 -0800)]
NFSv4.1: reject zero layout with zeroed stripe unit

Allowing stripe_unit==0 causes the client to crash later on
when dividing by zero.

Reported-by: Marc Eshel <eshel@almaden.ibm.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: Clear lseg pointer in ->doio function
Fred Isaman [Thu, 3 Mar 2011 15:13:49 +0000 (15:13 +0000)]
NFSv4.1: Clear lseg pointer in ->doio function

Now that we have access to the pointer, clear it immediately after
the put, instead of in caller.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: rearrange ->doio args
Fred Isaman [Thu, 3 Mar 2011 15:13:48 +0000 (15:13 +0000)]
NFSv4.1: rearrange ->doio args

This will make it possible to clear the lseg pointer in the same
function as it is put, instead of in the caller nfs_pageio_doio().

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: pnfs filelayout driver write
Fred Isaman [Thu, 3 Mar 2011 15:13:47 +0000 (15:13 +0000)]
NFSv4.1: pnfs filelayout driver write

Allows the pnfs filelayout driver to write to the data servers.

Note that COMMIT to data servers will be implemented in a future
patch.  To avoid improper behavior, for the moment any WRITE to a data
server that would also require a COMMIT to the data server is sent
NFS_FILE_SYNC.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
Signed-off-by: Mingyang Guo <guomingyang@nrchpc.ac.cn>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Ricardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: remove GETATTR from ds writes
Fred Isaman [Thu, 3 Mar 2011 15:13:46 +0000 (15:13 +0000)]
NFSv4.1: remove GETATTR from ds writes

Any WRITE compound directed to a data server needs to have the
GETATTR calls suppressed.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: implement generic pnfs layer write switch
Andy Adamson [Thu, 3 Mar 2011 15:13:45 +0000 (15:13 +0000)]
NFSv4.1: implement generic pnfs layer write switch

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
Signed-off-by: J. Bruce Fields <bfields@fieldses.org>
Signed-off-by: Mike Sager <sager@netapp.com>
Signed-off-by: Ricardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: Tao Guo <guotao@nrchpc.ac.cn>
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: trigger LAYOUTGET for writes
Fred Isaman [Thu, 3 Mar 2011 15:13:44 +0000 (15:13 +0000)]
NFSv4.1: trigger LAYOUTGET for writes

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: Send lseg down into nfs_write_rpcsetup
Fred Isaman [Thu, 3 Mar 2011 15:13:43 +0000 (15:13 +0000)]
NFSv4.1: Send lseg down into nfs_write_rpcsetup

We grab the lseg sent in from the doio function and attach it to
each struct nfs_write_data created.  This is how the lseg will be
sent to the layout driver.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: add callback to nfs4_write_done
Fred Isaman [Thu, 3 Mar 2011 15:13:42 +0000 (15:13 +0000)]
NFSv4.1: add callback to nfs4_write_done

Add callback that pnfs layout driver can use to do its own handling
of data server WRITE response.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: rearrange nfs_write_rpcsetup
Andy Adamson [Thu, 3 Mar 2011 15:13:41 +0000 (15:13 +0000)]
NFSv4.1: rearrange nfs_write_rpcsetup

Reorder nfs_write_rpcsetup, preparing for a pnfs entry point.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: lseg documentation
Fred Isaman [Tue, 1 Mar 2011 01:34:23 +0000 (01:34 +0000)]
NFSv4.1: lseg documentation

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: turn off pNFS on ds connection failure
Andy Adamson [Tue, 1 Mar 2011 01:34:22 +0000 (01:34 +0000)]
NFSv4.1: turn off pNFS on ds connection failure

If a data server is unavailable, go through MDS.

Mark the deviceid containing the data server as a negative cache entry.
Do not try to connect to any data server on a deviceid marked as a negative
cache entry. Mark any layout that tries to use the marked deviceid as failed.

Inodes with a layout marked as fails will not use the layout for I/O, and will
not perform any more layoutgets.
Inodes without a layout will still do layoutget, but the layout will get
marked immediately.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1 move deviceid cache to filelayout driver
Christoph Hellwig [Tue, 1 Mar 2011 01:34:21 +0000 (01:34 +0000)]
NFSv4.1 move deviceid cache to filelayout driver

No need for generic cache with only one user.
Keep a simple hash of deviceids in the filelayout driver.

Signed-off-by: Christoph Hellwig <hch@infradead.org>
Acked-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: filelayout async error handler
Andy Adamson [Tue, 1 Mar 2011 01:34:20 +0000 (01:34 +0000)]
NFSv4.1: filelayout async error handler

Use our own async error handler.
Mark the layout as failed and retry i/o through the MDS on specified errors.

Update the mds_offset in nfs_readpage_retry so that a failed short-read retry
to a DS gets correctly resent through the MDS.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: filelayout read
Andy Adamson [Tue, 1 Mar 2011 01:34:19 +0000 (01:34 +0000)]
NFSv4.1: filelayout read

Attempt a pNFS file layout read by setting up the nfs_read_data struct and
calling nfs_initiate_read with the data server rpc client and the
filelayout rpc call ops.

Error handling is implemented in a subsequent patch.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Mingyang Guo <guomingyang@nrchpc.ac.cn>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Ricardo Labiaga <Ricardo.Labiaga@netapp.com>
Tested-by: Guo Mingyang <guomingyang@nrchpc.ac.cn>
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: filelayout i/o helpers
Fred Isaman [Tue, 1 Mar 2011 01:34:18 +0000 (01:34 +0000)]
NFSv4.1: filelayout i/o helpers

Prepare for filelayout_read_pagelist with helper functions that find the correct
data server, filehandle, and offset.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Marc Eshel <eshel@almaden.ibm.com>
Signed-off-by: Mike Sager <sager@netapp.com>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Tao Guo <guotao@nrchpc.ac.cn>
Signed-off-by: Tigran Mkrtchyan <tigran@anahit.desy.de>
Signed-off-by: Tigran Mkrtchyan <tigran.mkrtchyan@desy.de>
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: data server connection
Andy Adamson [Tue, 1 Mar 2011 01:34:17 +0000 (01:34 +0000)]
NFSv4.1: data server connection

Introduce a data server set_client and init session following the
nfs4_set_client and  nfs4_init_session convention.

Once a new nfs_client is on the nfs_client_list, the nfs_client cl_cons_state
serializes access to creating an nfs_client struct with matching properties.

Use the new nfs_get_client() that initializes new clients.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: generic read
Andy Adamson [Tue, 1 Mar 2011 01:34:16 +0000 (01:34 +0000)]
NFSv4.1: generic read

Separate the rpc run portion of nfs_read_rpcsetup into a new function
nfs_initiate_read that is called for normal NFS I/O.

Add a pNFS read_pagelist function that is called instead of nfs_intitate_read
for pNFS reads.

Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Mike Sager <sager@netapp.com>
Signed-off-by: Mingyang Guo <guomingyang@nrchpc.ac.cn>
Signed-off-by: Ricardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: Tao Guo <guotao@nrchpc.ac.cn>
Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: shift pnfs_update_layout locations
Fred Isaman [Tue, 1 Mar 2011 01:34:15 +0000 (01:34 +0000)]
NFSv4.1: shift pnfs_update_layout locations

Move the pnfs_update_layout call location to nfs_pageio_do_add_request().
Grab the lseg sent in the doio function to nfs_read_rpcsetup and attach
it to each nfs_read_data so it can be sent to the layout driver.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Tao Guo <guotao@nrchpc.ac.cn>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: coelesce across layout stripes
Fred Isaman [Tue, 1 Mar 2011 01:34:14 +0000 (01:34 +0000)]
NFSv4.1: coelesce across layout stripes

Add a pg_test layout driver hook which is used to avoid coelescing I/O across
layout stripes.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Andy Adamson <andros@citi.umich.edu>
Signed-off-by: Dean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Tao Guo <guotao@nrchpc.ac.cn>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: lseg refcounting
Fred Isaman [Tue, 1 Mar 2011 01:34:13 +0000 (01:34 +0000)]
NFSv4.1: lseg refcounting

Prepare put_lseg and get_lseg to be called from the pNFS I/O code.
Pull common code from pnfs_lseg_locked to call from pnfs_lseg.
Inline pnfs_lseg_locked into it's only caller.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Benny Halevy <bhalevy@panasas.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: add MDS mount DS only check
Andy Adamson [Tue, 1 Mar 2011 01:34:12 +0000 (01:34 +0000)]
NFSv4.1: add MDS mount DS only check

The DS only role cannot be used to mount.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: new flag for lease time check
Andy Adamson [Tue, 1 Mar 2011 01:34:11 +0000 (01:34 +0000)]
NFSv4.1: new flag for lease time check

Data servers cannot send nfs4_proc_get_lease_time. but still need to setup
state renewal. Add the NFS_CS_CHECK_LEASE_TIME bit to indicate if the lease
time can be checked.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: new flag for state renewal check
Andy Adamson [Tue, 1 Mar 2011 01:34:10 +0000 (01:34 +0000)]
NFSv4.1: new flag for state renewal check

Data servers not sharing a session with the mount MDS always have an empty
cl_superblocks list.
Replace the cl_superblocks empty list check to see if it is time to shut down
renewd with the NFS_CS_STOP_RENEW bit which is not set by such a data server.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: send zero stateid seqid on v4.1 i/o
Andy Adamson [Tue, 1 Mar 2011 01:34:09 +0000 (01:34 +0000)]
NFSv4.1: send zero stateid seqid on v4.1 i/o

Data servers require a zero stateid seqid, and there is no advantage to not
doing the same for all NFSv4.1

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFS move nfs_client initialization into nfs_get_client
Andy Adamson [Tue, 1 Mar 2011 01:34:08 +0000 (01:34 +0000)]
NFS move nfs_client initialization into nfs_get_client

Now nfs_get_client returns an nfs_client ready to be used no matter if it was
found or created.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: put_layout_hdr can remove nfsi->layout
Andy Adamson [Tue, 1 Mar 2011 01:34:07 +0000 (01:34 +0000)]
NFSv4.1: put_layout_hdr can remove nfsi->layout

Prevents an Oops triggered by CB_LAYOUTRECALL and LAYOUTGET race on a
pnfs_layout_hdr first pnfs_layout_segment.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: remove CONFIG_NFS_V4 from nfs_read_data
Andy Adamson [Tue, 1 Mar 2011 01:34:06 +0000 (01:34 +0000)]
NFSv4: remove CONFIG_NFS_V4 from nfs_read_data

Cleanup nfs_read_data. We also won't use CONFIG_NFS_V4_1 for additional
NFSv4.1 fields in subsequent patches.

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFS: change nfs_writeback_done to return void
Fred Isaman [Fri, 11 Feb 2011 15:42:38 +0000 (15:42 +0000)]
NFS: change nfs_writeback_done to return void

The return values are not used by any callers.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFS: remove pointless if statement in nfs_direct_write_result
Fred Isaman [Fri, 11 Feb 2011 15:42:37 +0000 (15:42 +0000)]
NFS: remove pointless if statement in nfs_direct_write_result

The code was doing nothing more in either branch of the if.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoRPC: clarify rpc_run_task error handling
Fred Isaman [Fri, 11 Feb 2011 15:42:36 +0000 (15:42 +0000)]
RPC: clarify rpc_run_task error handling

rpc_run_task can only fail if it is not passed in a preallocated task.
However, that is not at all clear with the current code.  So
remove several impossible to occur failure checks.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoRPC: remove check for impossible condition in rpc_make_runnable
Fred Isaman [Fri, 11 Feb 2011 15:42:35 +0000 (15:42 +0000)]
RPC: remove check for impossible condition in rpc_make_runnable

queue_work() only returns 0 or 1, never a negative value.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agopnfs: fix pnfs lock inversion of i_lock and cl_lock
Fred Isaman [Thu, 3 Feb 2011 18:28:52 +0000 (18:28 +0000)]
pnfs: fix pnfs lock inversion of i_lock and cl_lock

The pnfs code was using throughout the lock order i_lock, cl_lock.
This conflicts with the nfs delegation code.  Rework the pnfs code
to avoid taking both locks simultaneously.

Currently the code takes the double lock to add/remove the layout to a
nfs_client list, while atomically checking that the list of lsegs is
empty.  To avoid this, we rely on existing serializations.  When a
layout is initialized with lseg count equal zero, LAYOUTGET's
openstateid serialization is in effect, making it safe to assume it
stays zero unless we change it.  And once a layout's lseg count drops
to zero, it is set as DESTROYED and so will stay at zero.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agopnfs: do not need to clear NFS_LAYOUT_BULK_RECALL flag
Fred Isaman [Thu, 3 Feb 2011 18:28:51 +0000 (18:28 +0000)]
pnfs: do not need to clear NFS_LAYOUT_BULK_RECALL flag

We do not need to clear the NFS_LAYOUT_BULK_RECALL, as setting it
guarantees that NFS_LAYOUT_DESTROYED will be set once any outstanding
io is finished.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agopnfs: avoid incorrect use of layout stateid
Fred Isaman [Thu, 3 Feb 2011 18:28:50 +0000 (18:28 +0000)]
pnfs: avoid incorrect use of layout stateid

The code could violate the following from RFC5661, section 12.5.3:
"Once a client has no more layouts on a file, the layout stateid is no
longer valid and MUST NOT be used."

This can occur when a layout already has a lseg, starts another
non-everlapping LAYOUTGET, and a CB_LAYOUTRECALL for the existing lseg
is processed before we hit pnfs_layout_process().

Solve by setting, each time the client has no more lsegs for a file, a
flag which blocks further use of the layout and triggers its removal.

This also fixes a second bug which occurs in the same instance as
above.  If we actually use pnfs_layout_process, we add the new lseg to
the layout, but the layout has been removed from the nfs_client list
by the intervening CB_LAYOUTRECALL and will not be added back.  Thus
the newly acquired lseg will not be properly returned in the event of
a subsequent CB_LAYOUTRECALL.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFS: NFSROOT should default to "proto=udp"
Chuck Lever [Fri, 11 Mar 2011 20:31:06 +0000 (15:31 -0500)]
NFS: NFSROOT should default to "proto=udp"

There have been a number of recent reports that NFSROOT is no longer
working with default mount options, but fails only with certain NICs.

Brian Downing <bdowning@lavos.net> bisected to commit 56463e50 "NFS:
Use super.c for NFSROOT mount option parsing".  Among other things,
this commit changes the default mount options for NFSROOT to use TCP
instead of UDP as the underlying transport.

TCP seems less able to deal with NICs that are slow to initialize.
The system logs that have accompanied reports of problems all show
that NFSROOT attempts to establish a TCP connection before the NIC is
fully initialized, and thus the TCP connection attempt fails.

When a TCP connection attempt fails during a mount operation, the
NFS stack needs to fail the operation.  Usually user space knows how
and when to retry it.  The network layer does not report a distinct
error code for this particular failure mode.  Thus, there isn't a
clean way for the RPC client to see that it needs to retry in this
case, but not in others.

Because NFSROOT is used in some environments where it is not possible
to update the kernel command line to specify "udp", the proper thing
to do is change NFSROOT to use UDP by default, as it did before commit
56463e50.

To make it easier to see how to change default mount options for
NFSROOT and to distinguish default settings from mandatory settings,
I've adjusted a couple of areas to document the specifics.

root_nfs_cat() is also modified to deal with commas properly when
concatenating strings containing mount option lists.  This keeps
root_nfs_cat() call sites simpler, now that we may be concatenating
multiple mount option strings.

Tested-by: Brian Downing <bdowning@lavos.net>
Tested-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Cc: <stable@kernel.org> # 2.6.37
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agonfs4: remove duplicated #include
Huang Weiyi [Tue, 8 Mar 2011 23:11:30 +0000 (23:11 +0000)]
nfs4: remove duplicated #include

Remove duplicated #include('s) in
  fs/nfs/nfs4proc.c

Signed-off-by: Huang Weiyi <weiyi.huang@gmail.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: nfs4_state_mark_reclaim_nograce() should be static
Trond Myklebust [Wed, 9 Mar 2011 21:12:46 +0000 (16:12 -0500)]
NFSv4: nfs4_state_mark_reclaim_nograce() should be static

There are no more external users of nfs4_state_mark_reclaim_nograce() or
nfs4_state_mark_reclaim_reboot(), so mark them as static.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: Fix the setlk error handler
Trond Myklebust [Wed, 9 Mar 2011 21:00:56 +0000 (16:00 -0500)]
NFSv4: Fix the setlk error handler

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: Fix the handling of the SEQUENCE status bits
Trond Myklebust [Wed, 9 Mar 2011 21:00:55 +0000 (16:00 -0500)]
NFSv4.1: Fix the handling of the SEQUENCE status bits

We want SEQUENCE status bits to be handled by the state manager in order
to avoid threading issues.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4/4.1: Fix nfs4_schedule_state_recovery abuses
Trond Myklebust [Wed, 9 Mar 2011 21:00:53 +0000 (16:00 -0500)]
NFSv4/4.1: Fix nfs4_schedule_state_recovery abuses

nfs4_schedule_state_recovery() should only be used when we need to force
the state manager to check the lease. If we just want to start the
state manager in order to handle a state recovery situation, we should be
using nfs4_schedule_state_manager().

This patch fixes the abuses of nfs4_schedule_state_recovery() by replacing
its use with a set of helper functions that do the right thing.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1 reclaim complete must wait for completion
Andy Adamson [Wed, 9 Mar 2011 18:13:46 +0000 (13:13 -0500)]
NFSv4.1 reclaim complete must wait for completion

Signed-off-by: Andy Adamson <andros@netapp.com>
[Trond: fix whitespace errors]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4: remove duplicate clientid in struct nfs_client
Andy Adamson [Wed, 9 Mar 2011 18:13:45 +0000 (13:13 -0500)]
NFSv4: remove duplicate clientid in struct nfs_client

Signed-off-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoNFSv4.1: Retry CREATE_SESSION on NFS4ERR_DELAY
Ricardo Labiaga [Wed, 9 Mar 2011 18:13:44 +0000 (13:13 -0500)]
NFSv4.1: Retry CREATE_SESSION on NFS4ERR_DELAY

Fix bug where we currently retry the EXCHANGEID call again, eventhough
we already have a valid clientid.  Instead, delay and retry the CREATE_SESSION
call.

Signed-off-by: Ricardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agosunrpc: Propagate errors from xs_bind() through xs_create_sock()
Ben Hutchings [Tue, 22 Feb 2011 21:54:34 +0000 (21:54 +0000)]
sunrpc: Propagate errors from xs_bind() through xs_create_sock()

xs_create_sock() is supposed to return a pointer or an ERR_PTR-encoded
error, but it currently returns 0 if xs_bind() fails.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Cc: stable@kernel.org [v2.6.37]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years ago(try3-resend) Fix nfs_compat_user_ino64 so it doesn't cause problems if bit 31 or...
Frank Filz [Thu, 2 Dec 2010 19:31:23 +0000 (19:31 +0000)]
(try3-resend) Fix nfs_compat_user_ino64 so it doesn't cause problems if bit 31 or 63 are set in fileid

The problem was use of an int32, which when converted to a uint64
is sign extended resulting in a fileid that doesn't fit in 32 bits
even though the intent of the function is to fit the fileid into
32 bits.

Signed-off-by: Frank Filz <ffilzlnx@us.ibm.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
[Trond: Added an include for compat.h]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agonfs: fix compilation warning
Jovi Zhang [Wed, 2 Mar 2011 23:19:37 +0000 (23:19 +0000)]
nfs: fix compilation warning

this commit fix compilation warning as following:
linux-2.6/fs/nfs/nfs4proc.c:3265: warning: comparison of distinct pointer types lacks a cast

Signed-off-by: Jovi Zhang <bookjovi@gmail.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agonfs: add kmalloc return value check in decode_and_add_ds
Stanislav Fomichev [Sat, 5 Feb 2011 23:13:01 +0000 (23:13 +0000)]
nfs: add kmalloc return value check in decode_and_add_ds

add kmalloc return value check in decode_and_add_ds

Signed-off-by: Stanislav Fomichev <kernel@fomichev.me>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Remove resource leak in svc_rdma_send_error()
Jesper Juhl [Sat, 22 Jan 2011 21:40:20 +0000 (21:40 +0000)]
SUNRPC: Remove resource leak in svc_rdma_send_error()

We leak the memory allocated to 'ctxt' when we return after
'ib_dma_mapping_error()' returns !=0.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agonfs: close NFSv4 COMMIT vs. CLOSE race
Jeff Layton [Sun, 6 Mar 2011 17:14:13 +0000 (17:14 +0000)]
nfs: close NFSv4 COMMIT vs. CLOSE race

I've been adding in more artificial delays in the NFSv4 commit and close
codepaths to uncover races. The kernel I'm testing has the patch to
close the race in __rpc_wait_for_completion_task that's in Trond's
cthon2011 branch. The reproducer I've been using does this in a loop:

mkdir("DIR");
fd = open("DIR/FILE", O_WRONLY|O_CREAT|O_EXCL, 0644);
write(fd, "abcdefg", 7);
close(fd);
unlink("DIR/FILE");
rmdir("DIR");

The above reproducer shouldn't result in any silly-renaming. However,
when I add a "msleep(100)" just after the nfs_commit_clear_lock call in
nfs_commit_release, I can almost always force one to occur. If I can
force it to occur with that, then it can happen without that delay
given the right timing.

nfs_commit_inode waits for the NFS_INO_COMMIT bit to clear when called
with FLUSH_SYNC set. nfs_commit_rpcsetup on the other hand does not wait
for the task to complete before putting its reference to it, so the last
reference get put in rpc_release task and gets queued to a workqueue.

In this situation, the last open context reference may be put by the
COMMIT release instead of the close() syscall. The close() syscall
returns too quickly and the unlink runs while the d_count is still
high since the COMMIT release hasn't put its dentry reference yet.

Fix this by having rpc_commit_rpcsetup wait for the RPC call to complete
before putting the task reference when FLUSH_SYNC is set. With this, the
last reference is put by the process that's initiating the FLUSH_SYNC
commit and the race is closed.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Close a race in __rpc_wait_for_completion_task()
Trond Myklebust [Mon, 21 Feb 2011 19:05:41 +0000 (11:05 -0800)]
SUNRPC: Close a race in __rpc_wait_for_completion_task()

Although they run as rpciod background tasks, under normal operation
(i.e. no SIGKILL), functions like nfs_sillyrename(), nfs4_proc_unlck()
and nfs4_do_close() want to be fully synchronous. This means that when we
exit, we want all references to the rpc_task to be gone, and we want
any dentry references etc. held by that task to be released.

For this reason these functions call __rpc_wait_for_completion_task(),
followed by rpc_put_task() in the expectation that the latter will be
releasing the last reference to the rpc_task, and thus ensuring that the
callback_ops->rpc_release() has been called synchronously.

This patch fixes a race which exists due to the fact that
rpciod calls rpc_complete_task() (in order to wake up the callers of
__rpc_wait_for_completion_task()) and then subsequently calls
rpc_put_task() without ensuring that these two steps are done atomically.

In order to avoid adding new spin locks, the patch uses the existing
waitqueue spin lock to order the rpc_task reference count releases between
the waiting process and rpciod.
The common case where nobody is waiting for completion is optimised for by
checking if the RPC_TASK_ASYNC flag is cleared and/or if the rpc_task
reference count is 1: in those cases we drop trying to grab the spin lock,
and immediately free up the rpc_task.

Those few processes that need to put the rpc_task from inside an
asynchronous context and that do not care about ordering are given a new
helper: rpc_put_task_async().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoMerge branch 'omap-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 7 Mar 2011 21:15:02 +0000 (13:15 -0800)]
Merge branch 'omap-fixes-for-linus' of git://git./linux/kernel/git/tmlind/linux-omap-2.6

* 'omap-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6:
  omap: mailbox: resolve hang issue
  OMAP2+: PM: SmartReflex: fix memory leaks in Smartreflex driver
  arm: mach-omap2: smartreflex: fix another memory leak

13 years agoMerge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Mon, 7 Mar 2011 21:14:19 +0000 (13:14 -0800)]
Merge branch 'for-linus' of git://git390.marist.edu/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6:
  [S390] tape: deadlock on system work queue
  [S390] keyboard: integer underflow bug
  [S390] xpram: remove __initdata attribute from module parameters

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Sun, 6 Mar 2011 18:44:49 +0000 (10:44 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda - Don't set to D3 in Cirrus errata init verbs
  ALSA: hda - add new Fermi 5xx codec IDs to snd-hda
  ASoC: WM8994: Ensure late enable events are processed for the ADCs
  ASoC: WM8994: Don't disable the AIF[1|2]CLK_ENA unconditionaly
  ASoC: Fix WM9081 platform data initialisation
  ALSA: hda - Fix unable to record issue on ASUS N82JV
  ALSA: HDA: Realtek: Fixup jack detection to input subsystem

13 years agovirtio: console: Don't access vqs if device was unplugged
Amit Shah [Fri, 4 Mar 2011 03:34:33 +0000 (14:04 +1030)]
virtio: console: Don't access vqs if device was unplugged

If a virtio-console device gets unplugged while a port is open, a
subsequent close() call on the port accesses vqs to free up buffers.
This can lead to a crash.

The buffers are already freed up as a result of the call to
unplug_ports() from virtcons_remove().  The fix is to simply not access
vq information if port->portdev is NULL.

Reported-by: juzhang <juzhang@redhat.com>
CC: stable@kernel.org
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'fix/asoc' into for-linus
Takashi Iwai [Sun, 6 Mar 2011 11:37:42 +0000 (12:37 +0100)]
Merge branch 'fix/asoc' into for-linus

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Sat, 5 Mar 2011 18:43:22 +0000 (10:43 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  ceph: no .snap inside of snapped namespace
  libceph: fix msgr standby handling
  libceph: fix msgr keepalive flag
  libceph: fix msgr backoff
  libceph: retry after authorization failure
  libceph: fix handling of short returns from get_user_pages
  ceph: do not clear I_COMPLETE from d_release
  ceph: do not set I_COMPLETE
  Revert "ceph: keep reference to parent inode on ceph_dentry"

13 years agomm: use correct numa policy node for transparent hugepages
Andi Kleen [Sat, 5 Mar 2011 01:36:32 +0000 (17:36 -0800)]
mm: use correct numa policy node for transparent hugepages

Pass down the correct node for a transparent hugepage allocation.  Most
callers continue to use the current node, however the hugepaged daemon
now uses the previous node of the first to be collapsed page instead.
This ensures that khugepaged does not mess up local memory for an
existing process which uses local policy.

The choice of node is somewhat primitive currently: it just uses the
node of the first page in the pmd range.  An alternative would be to
look at multiple pages and use the most popular node.  I used the
simplest variant for now which should work well enough for the case of
all pages being on the same node.

[akpm@linux-foundation.org: coding-style fixes]
Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomm: preserve original node for transparent huge page copies
Andi Kleen [Sat, 5 Mar 2011 01:36:31 +0000 (17:36 -0800)]
mm: preserve original node for transparent huge page copies

This makes a difference for LOCAL policy, where the node cannot be
determined from the policy itself, but has to be gotten from the original
page.

Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomm: add alloc_page_vma_node()
Andi Kleen [Sat, 5 Mar 2011 01:36:30 +0000 (17:36 -0800)]
mm: add alloc_page_vma_node()

Add a alloc_page_vma_node that allows passing the "local" node in.  Used
in a followon patch.

Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomm: change alloc_pages_vma to pass down the policy node for local policy
Andi Kleen [Sat, 5 Mar 2011 01:36:29 +0000 (17:36 -0800)]
mm: change alloc_pages_vma to pass down the policy node for local policy

Currently alloc_pages_vma() always uses the local node as policy node for
the LOCAL policy.  Pass this node down as an argument instead.

No behaviour change from this patch, but will be needed for followons.

Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoRapidIO: Update MAINTAINERS
Alexandre Bounine [Sat, 5 Mar 2011 01:36:28 +0000 (17:36 -0800)]
RapidIO: Update MAINTAINERS

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/video/backlight/ltv350qv.c: fix a memory leak
Axel Lin [Sat, 5 Mar 2011 01:36:27 +0000 (17:36 -0800)]
drivers/video/backlight/ltv350qv.c: fix a memory leak

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Cc: Haavard Skinnemoen <hskinnemoen@atmel.com>
Cc: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMAINTAINERS: add maintainer of Samsung Mobile Machine support
Kyungmin Park [Sat, 5 Mar 2011 01:36:26 +0000 (17:36 -0800)]
MAINTAINERS: add maintainer of Samsung Mobile Machine support

Add maintainer of Samsung Mobile machine support.  Currently, Aquila,
Goni, Universal (C210), and Nuri board are supported.

Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Joe Perches <joe@perches.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopps: make pps_gen_parport depend on BROKEN
Thomas Gleixner [Sat, 5 Mar 2011 01:36:23 +0000 (17:36 -0800)]
pps: make pps_gen_parport depend on BROKEN

This driver causes hard lockups, when the active clock soure is jiffies.

The reason is that it loops with interrupts disabled waiting for a
timestamp to be reached by polling getnstimeofday().  Though with a
jiffies clocksource, when that code runs on the same CPU which is
responsible for updating jiffies, then we loop in circles for ever
simply because the timer interrupt cannot update jiffies.  So both UP
and SMP can be affected.

There is no easy fix for that problem so make it depend on BROKEN for
now.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Alexander Gordeev <lasaine@lvk.cs.msu.su>
Cc: Rodolfo Giometti <giometti@linux.it>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/misc/bmp085.c: add MODULE_DEVICE_TABLE
Axel Lin [Sat, 5 Mar 2011 01:36:22 +0000 (17:36 -0800)]
drivers/misc/bmp085.c: add MODULE_DEVICE_TABLE

The device table is required to load modules based on modaliases.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Cc: Shubhrajyoti D <shubhrajyoti@ti.com>
Cc: Christoph Mair <christoph.mair@gmail.com>
Cc: Jonathan Cameron <jic23@cam.ac.uk>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpuset: add a missing unlock in cpuset_write_resmask()
Li Zefan [Sat, 5 Mar 2011 01:36:21 +0000 (17:36 -0800)]
cpuset: add a missing unlock in cpuset_write_resmask()

Don't forget to release cgroup_mutex if alloc_trial_cpuset() fails.

[akpm@linux-foundation.org: avoid multiple return points]
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/rtc/rtc-s3c.c: fix prototype for s3c_rtc_setaie()
Axel Lin [Sat, 5 Mar 2011 01:36:19 +0000 (17:36 -0800)]
drivers/rtc/rtc-s3c.c: fix prototype for s3c_rtc_setaie()

Fix s3c_rtc_setaie() prototype to eliminate the following compile
warning:

  drivers/rtc/rtc-s3c.c:383: warning: initialization from incompatible pointer type

(akpm: the rtc_class_ops.alarm_irq_enable() handler is being passed two
arguments where it expects just one, presumably with undesired effects)

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier...
Linus Torvalds [Sat, 5 Mar 2011 01:31:43 +0000 (17:31 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/vapier/blackfin

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin:
  Blackfin: iflush: update anomaly 05000491 workaround
  Blackfin: outs[lwb]: make sure count is greater than 0

13 years agoMerge branch 'rmobile-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 5 Mar 2011 01:31:19 +0000 (17:31 -0800)]
Merge branch 'rmobile-fixes-for-linus' of git://git./linux/kernel/git/lethal/sh-2.6

* 'rmobile-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  ARM: mach-shmobile: mackerel: modify LCDC clock divider value
  ARM: mach-shmobile: ap4evb: modify LCDC clock divider value
  ARM: mach-shmobile: mackerel: fixup memory initialize for zboot
  ARM: mach-shmobile: ap4evb: fixup memory initialize for zboot
  ARM: mach-shmobile: Add sh73a0 MIPI-CSI and CEU clocks
  ARM: mach-shmobile: AG5EVM MIPI-DSI LCD reset delay fix

13 years agoMerge branch 'sh-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 5 Mar 2011 01:31:01 +0000 (17:31 -0800)]
Merge branch 'sh-fixes-for-linus' of git://git./linux/kernel/git/lethal/sh-2.6

* 'sh-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  sh: Change __nosave_XXX symbols to long
  sh: Flush executable pages in copy_user_highpage
  sh: Ensure ST40-300 BogoMIPS value is consistent
  sh: sh7750: Fix incompatible pointer type
  sh: sh7750: move machtypes.h to include/generated

13 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Sat, 5 Mar 2011 01:30:32 +0000 (17:30 -0800)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/nouveau: allocate kernel's notifier object at end of block

13 years agonfs4: Ensure that ACL pages sent over NFS were not allocated from the slab (v3)
Neil Horman [Sat, 5 Mar 2011 00:26:03 +0000 (19:26 -0500)]
nfs4: Ensure that ACL pages sent over NFS were not allocated from the slab (v3)

The "bad_page()" page allocator sanity check was reported recently (call
chain as follows):

  bad_page+0x69/0x91
  free_hot_cold_page+0x81/0x144
  skb_release_data+0x5f/0x98
  __kfree_skb+0x11/0x1a
  tcp_ack+0x6a3/0x1868
  tcp_rcv_established+0x7a6/0x8b9
  tcp_v4_do_rcv+0x2a/0x2fa
  tcp_v4_rcv+0x9a2/0x9f6
  do_timer+0x2df/0x52c
  ip_local_deliver+0x19d/0x263
  ip_rcv+0x539/0x57c
  netif_receive_skb+0x470/0x49f
  :virtio_net:virtnet_poll+0x46b/0x5c5
  net_rx_action+0xac/0x1b3
  __do_softirq+0x89/0x133
  call_softirq+0x1c/0x28
  do_softirq+0x2c/0x7d
  do_IRQ+0xec/0xf5
  default_idle+0x0/0x50
  ret_from_intr+0x0/0xa
  default_idle+0x29/0x50
  cpu_idle+0x95/0xb8
  start_kernel+0x220/0x225
  _sinittext+0x22f/0x236

It occurs because an skb with a fraglist was freed from the tcp
retransmit queue when it was acked, but a page on that fraglist had
PG_Slab set (indicating it was allocated from the Slab allocator (which
means the free path above can't safely free it via put_page.

We tracked this back to an nfsv4 setacl operation, in which the nfs code
attempted to fill convert the passed in buffer to an array of pages in
__nfs4_proc_set_acl, which gets used by the skb->frags list in
xs_sendpages.  __nfs4_proc_set_acl just converts each page in the buffer
to a page struct via virt_to_page, but the vfs allocates the buffer via
kmalloc, meaning the PG_slab bit is set.  We can't create a buffer with
kmalloc and free it later in the tcp ack path with put_page, so we need
to either:

1) ensure that when we create the list of pages, no page struct has
   PG_Slab set

 or

2) not use a page list to send this data

Given that these buffers can be multiple pages and arbitrarily sized, I
think (1) is the right way to go.  I've written the below patch to
allocate a page from the buddy allocator directly and copy the data over
to it.  This ensures that we have a put_page free-able page for every
entry that winds up on an skb frag list, so it can be safely freed when
the frame is acked.  We do a put page on each entry after the
rpc_call_sync call so as to drop our own reference count to the page,
leaving only the ref count taken by tcp_sendpages.  This way the data
will be properly freed when the ack comes in

Successfully tested by myself to solve the above oops.

Note, as this is the result of a setacl operation that exceeded a page
of data, I think this amounts to a local DOS triggerable by an
uprivlidged user, so I'm CCing security on this as well.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: Trond Myklebust <Trond.Myklebust@netapp.com>
CC: security@kernel.org
CC: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoceph: no .snap inside of snapped namespace
Sage Weil [Thu, 3 Mar 2011 21:44:35 +0000 (13:44 -0800)]
ceph: no .snap inside of snapped namespace

Otherwise you can do things like

# mkdir .snap/foo
# cd .snap/foo/.snap
# ls
<badness>

Signed-off-by: Sage Weil <sage@newdream.net>
13 years agolibceph: fix msgr standby handling
Sage Weil [Fri, 4 Mar 2011 20:25:05 +0000 (12:25 -0800)]
libceph: fix msgr standby handling

The standby logic used to be pretty dependent on the work requeueing
behavior that changed when we switched to WQ_NON_REENTRANT.  It was also
very fragile.

Restructure things so that:
 - We clear WRITE_PENDING when we set STANDBY.  This ensures we will
   requeue work when we wake up later.
 - con_work backs off if STANDBY is set.  There is nothing to do if we are
   in standby.
 - clear_standby() helper is called by both con_send() and con_keepalive(),
   the two actions that can wake us up again.  Move the connect_seq++
   logic here.

Signed-off-by: Sage Weil <sage@newdream.net>