For pnfs pagelist write failure, we need to pg_recoalesce and resend IO to
mds.
Signed-off-by: NPeng Tao <peng_tao@emc.com>
Signed-off-by: NJim Rees <rees@umich.edu>
Cc: stable@kernel.org [3.0]
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

nfs_find_and_lock_request will take a reference to the nfs_page and
will then put it if the req is already locked. It's possible though
that the reference will be the last one. That put then can kick off
a whole series of reference puts:

nfs_page
   nfs_open_context
      dentry
          inode

If the inode ends up being deleted, then the VFS will call
truncate_inode_pages. That function will try to take the page lock, but
it was already locked when migrate_page was called. The code
deadlocks.

Fix this by simply refusing the migration request if PagePrivate is
already set, indicating that the page is already associated with an
active read or write request.

We've had a customer test a backported version of this patch and
the preliminary results seem good.

Cc: stable@kernel.org
Cc: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: NHarshula Jayasuriya <harshula@redhat.com>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

commit 420e3646 allowed the kernel to reduce the number of unnecessary
commit calls by skipping the commit when there are a large number of
outstanding pages.

However, the current test in nfs_commit_unstable_pages does not handle
the edge condition properly. When ncommit == 0, then that means that the
kernel doesn't need to do anything more for the inode. The current test
though in the WB_SYNC_NONE case will return true, and the inode will end
up being marked dirty. Once that happens the inode will never be clean
until there's a WB_SYNC_ALL flush.

Fix this by immediately returning from nfs_commit_unstable_pages when
ncommit == 0.

Mike noticed this problem initially in RHEL5 (2.6.18-based kernel) which
has a backported version of 420e3646. The inode cache there was growing
very large. The inode cache was unable to be shrunk since the inodes
were all marked dirty. Calling sync() would essentially "fix" the
problem -- the WB_SYNC_ALL flush would result in the inodes all being
marked clean.

What I'm not clear on is how big a problem this is in mainline kernels
as the writeback code there is very different. Either way, it seems
incorrect to re-mark the inode dirty in this case.
Reported-by: NMike McLean <mikem@redhat.com>
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Cc: stable@kernel.org [2.6.34+]
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

This reverts commit b80c3cb6.

The reverted commit was rendered obsolete by a VFS fix: commit
5547e8aa (writeback: Update dirty flags in
two steps). We now no longer need to worry about writeback_single_inode()
missing our marking the inode for COMMIT in 'do_writepages()' call.

Reverting this patch, fixes a performance regression in which the inode
would continuously get queued to the dirty list, causing the writeback
code to unnecessarily try to send a COMMIT.

Signed-off-by: Trond Myklebust <Trond.Myklebust>
Tested-by: NSimon Kirby <sim@hostway.ca>
Cc: stable@kernel.org [2.6.35+]

Fix a typo which causes an Oops in the RPC layer, when using wsize < 4k.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Tested-by: NSricharan R <r.sricharan@ti.com>

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

Use nfs_pageio_reset_read_mds and nfs_pageio_reset_write_mds instead of
completely reinitialising the struct nfs_pageio_descriptor.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

...and ensure that we recoalese to take into account differences in
differences in block sizes when falling back to write through the MDS.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Instead of looking up the rsize and wsize, the routines that generate the
RPC requests should really be using the pg_bsize, since that is what we
use when deciding whether or not to coalesce write requests...
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

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

...and do the same for nfs_flush_multi/one.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Signed-off-by: NPeng Tao <peng_tao@emc.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Split them up into two parts: one which sets up the struct nfs_read/write_data,
the other which sets up the actual RPC call or pNFS call.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

If we're writing back data, and the FLUSH_STABLE flag is set, then we
always want to use NFS_FILE_SYNC, since we're always in a situation where
we're doing page reclaim, and so we want to free up the page as quickly
as possible.

If we're in the FLUSH_COND_STABLE case, then we either want to use another
unstable write (if we have to do a commit anyway) or again, we want to
use NFS_FILE_SYNC because we know that we have no more pages to write
out.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Ask for whole file layouts. Until support for layout segments is fully
supported in the file layout code, discard non-whole file layouts.
Signed-off-by: NAndy Adamson <andros@netapp.com>
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

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

Ensure that we always get a layout before setting up the i/o request.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

We need to ensure that the layouts are set up before we can decide to
coalesce requests. To do so, we want to further split up the struct
nfs_pageio_descriptor operations into an initialisation callback, a
coalescing test callback, and a 'do i/o' callback.

This patch cleans up the existing callback methods before adding the
'initialisation' callback.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

In current pnfs tree, all the layouts set mds_offset in their
.write_pagelist member.
mds_offset is only used by generic layer and should be handled by it.

This patch is for upstream. It is needed in this -rc series to fix a
bug in objects layout_commit.

I'll send patches for objects and blocks to be
squashed into current pnfs tree.

TODO: It looks like the read path needs the same patch.
Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Remove two unused struct writeback_control fields:

	.encountered_congestion	(completely unused)
	.nonblocking		(never set, checked/showed in XFS,NFS/btrfs)

The .for_background check in nfs_write_inode() is also removed btw,
as .for_background implies WB_SYNC_NONE.
Reviewed-by: NJan Kara <jack@suse.cz>
Proposed-by: NChristoph Hellwig <hch@infradead.org>
Signed-off-by: NWu Fengguang <fengguang.wu@intel.com>

Use common code for pnfs_pageio_init_{read,write} and use
a common generic pg_test function.

Note that this function always assumes the the layout driver's
pg_test method is implemented.

[Fix BUG]
Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NBenny Halevy <bhalevy@panasas.com>

Add offset and count parameters to pnfs_update_layout and use them to get
the layout in the pageio path.

Order cache layout segments in the following order:
* offset (ascending)
* length (descending)
* iomode (RW before READ)

Test byte range against the layout segment in use in pnfs_{read,write}_pg_test
so not to coalesce pages not using the same layout segment.

[fix lseg ordering]
[clean up pnfs_find_lseg lseg arg]
[remove unnecessary FIXME]
[fix ordering in pnfs_insert_layout]
[clean up pnfs_insert_layout]
Signed-off-by: NBenny Halevy <bhalevy@panasas.com>

Currently, writebacks may end up recursing back into the filesystem due to
GFP_KERNEL direct reclaims in the pnfs subsystem.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

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

We only need to call nfs_mark_request_dirty() once in nfs_writepage_setup().
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

nfs_scan_commit() is called with the inode->i_lock held, but it then
calls __mark_inode_dirty() while still holding the lock. This causes
a deadlock.

Push the inode->i_lock into nfs_scan_commit() so it can protect only
the parts of the code it needs to and can be dropped before the call
to __mark_inode_dirty() to avoid the deadlock.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Tested-by: NWill Simoneau <simoneau@ele.uri.edu>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Now that the inode scalability patches have been merged, it is no longer
safe to call igrab() under the inode->i_lock.
Now that we no longer call nfs_clear_request() until the nfs_page is
being freed, we know that we are always holding a reference to the
nfs_open_context, which again holds a reference to the path, and so
the inode cannot be freed until the last nfs_page has been removed
from the radix tree and freed.

We can therefore skip the igrab()/iput() altogether.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

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

The filelayout driver sends LAYOUTCOMMIT only when COMMIT goes to
the data server (as opposed to the MDS) and the data server WRITE
is not NFS_FILE_SYNC.

Only whole file layout support means that there is only one IOMODE_RW layout
segment.
Signed-off-by: NAndy Adamson <andros@netapp.com>
Signed-off-by: NAlexandros Batsakis <batsakis@netapp.com>
Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NDean Hildebrand <dhildeb@us.ibm.com>
Signed-off-by: NFred Isaman <iisaman@citi.umich.edu>
Signed-off-by: NMingyang Guo <guomingyang@nrchpc.ac.cn>
Signed-off-by: NTao Guo <guotao@nrchpc.ac.cn>
Signed-off-by: NZhang Jingwang <zhangjingwang@nrchpc.ac.cn>
Tested-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NBenny Halevy <bhalevy@panasas.com>
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Implement all the hooks created in the previous patches.
This requires exporting quite a few functions and adding a few
structure fields.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Any COMMIT compound directed to a data server needs to have the
GETATTR calls suppressed.  We here, make sure the field we are testing
(data->lseg) is set and refcounted correctly.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

We create three major hooks for the pnfs code.

pnfs_mark_request_commit() is called during writeback_done from
nfs_mark_request_commit, which gives the driver an opportunity to
claim it wants control over commiting a particular req.

pnfs_choose_commit_list() is called from nfs_scan_list
to choose which list a given req should be added to, based on
where we intend to send it for COMMIT.  It is up to the driver
to have preallocated list headers for each destination it may need.

pnfs_commit_list() is how the driver actually takes control, it is
used instead of nfs_commit_list().

In order to pass information between the above functions, we create
a union in nfs_page to hold a lseg (which is possible because the req is
not on any list while in transition), and add some flags to indicate
if we need to use the pnfs code.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Create a separate support function for later use by data server
commit code.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Create a separate support function for later use by data server
commit code.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Reorder nfs_commit_rpcsetup, preparing for a pnfs entry point.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Based on consensus reached in Feb 2011 interim IETF meeting regarding
use of LAYOUTCOMMIT, it has been decided that a NFS_DATA_SYNC return
from a WRITE to data server should not initiate a COMMIT.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

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: NTrond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org

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: NNeilBrown <neilb@suse.de>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Now that we have access to the pointer, clear it immediately after
the put, instead of in caller.
Signed-off-by: NFred Isaman <iisaman@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

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