An osd request currently has two callbacks.  They inform the
initiator of the request when we've received confirmation for the
target osd that a request was received, and when the osd indicates
all changes described by the request are durable.

The only time the second callback is used is in the ceph file system
for a synchronous write.  There's a race that makes some handling of
this case unsafe.  This patch addresses this problem.  The error
handling for this callback is also kind of gross, and this patch
changes that as well.

In ceph_sync_write(), if a safe callback is requested we want to add
the request on the ceph inode's unsafe items list.  Because items on
this list must have their tid set (by ceph_osd_start_request()), the
request added *after* the call to that function returns.  The
problem with this is that there's a race between starting the
request and adding it to the unsafe items list; the request may
already be complete before ceph_sync_write() even begins to put it
on the list.

To address this, we change the way the "safe" callback is used.
Rather than just calling it when the request is "safe", we use it to
notify the initiator the bounds (start and end) of the period during
which the request is *unsafe*.  So the initiator gets notified just
before the request gets sent to the osd (when it is "unsafe"), and
again when it's known the results are durable (it's no longer
unsafe).  The first call will get made in __send_request(), just
before the request message gets sent to the messenger for the first
time.  That function is only called by __send_queued(), which is
always called with the osd client's request mutex held.

We then have this callback function insert the request on the ceph
inode's unsafe list when we're told the request is unsafe.  This
will avoid the race because this call will be made under protection
of the osd client's request mutex.  It also nicely groups the setup
and cleanup of the state associated with managing unsafe requests.

The name of the "safe" callback field is changed to "unsafe" to
better reflect its new purpose.  It has a Boolean "unsafe" parameter
to indicate whether the request is becoming unsafe or is now safe.
Because the "msg" parameter wasn't used, we drop that.

This resolves the original problem reportedin:
    http://tracker.ceph.com/issues/4706Reported-by: NYan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NYan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: NSage Weil <sage@inktank.com>

In ceph_sync_write(), if a safe callback is supplied with a request,
and an error is returned by ceph_osdc_wait_request(), a block of
code is executed to remove the request from the unsafe writes list
and drop references to capabilities acquired just prior to a call to
ceph_osdc_wait_request().

The only function used for this callback is sync_write_commit(),
and it does *exactly* what that block of error handling code does.

Now in ceph_osdc_wait_request(), if an error occurs (due to an
interupt during a wait_for_completion_interruptible() call),
complete_request() gets called, and that calls the request's
safe_callback method if it's defined.

So this means that this cleanup activity gets called twice in this
case, which is erroneous (and in fact leads to a crash).

Fix this by just letting the osd client handle the cleanup in
the event of an interrupt.

This resolves one problem mentioned in:
    http://tracker.ceph.com/issues/4706Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NYan, Zheng <zheng.z.yan@intel.com>

This ends up being a rather large patch but what it's doing is
somewhat straightforward.

Basically, this is replacing two calls with one.  The first of the
two calls is initializing a struct ceph_osd_data with data (either a
page array, a page list, or a bio list); the second is setting an
osd request op so it associates that data with one of the op's
parameters.  In place of those two will be a single function that
initializes the op directly.

That means we sort of fan out a set of the needed functions:
    - extent ops with pages data
    - extent ops with pagelist data
    - extent ops with bio list data
and
    - class ops with page data for receiving a response

We also have define another one, but it's only used internally:
    - class ops with pagelist data for request parameters

Note that we *still* haven't gotten rid of the osd request's
r_data_in and r_data_out fields.  All the osd ops refer to them for
their data.  For now, these data fields are pointers assigned to the
appropriate r_data_* field when these new functions are called.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

An extent type osd operation currently implies that there will
be corresponding data supplied in the data portion of the request
(for write) or response (for read) message.  Similarly, an osd class
method operation implies a data item will be supplied to receive
the response data from the operation.

Add a ceph_osd_data pointer to each of those structures, and assign
it to point to eithre the incoming or the outgoing data structure in
the osd message.  The data is not always available when an op is
initially set up, so add two new functions to allow setting them
after the op has been initialized.

Begin to make use of the data item pointer available in the osd
operation rather than the request data in or out structure in
places where it's convenient.  Add some assertions to verify
pointers are always set the way they're expected to be.

This is a sort of stepping stone toward really moving the data
into the osd request ops, to allow for some validation before
making that jump.

This is the first in a series of patches that resolve:
    http://tracker.ceph.com/issues/4657Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

An osd request keeps a pointer to the osd operations (ops) array
that it builds in its request message.

In order to allow each op in the array to have its own distinct
data, we will need to keep track of each op's data, and that
information does not go over the wire.

As long as we're tracking the data we might as well just track the
entire (source) op definition for each of the ops.  And if we're
doing that, we'll have no more need to keep a pointer to the
wire-encoded version.

This patch makes the array of source ops be kept with the osd
request structure, and uses that instead of the version encoded in
the message in places where that was previously used.  The array
will be embedded in the request structure, and the maximum number of
ops we ever actually use is currently 2.  So reduce CEPH_OSD_MAX_OP
to 2 to reduce the size of the structure.

The result of doing this sort of ripples back up, and as a result
various function parameters and local variables become unnecessary.

Make r_num_ops be unsigned, and move the definition of struct
ceph_osd_req_op earlier to ensure it's defined where needed.

It does not yet add per-op data, that's coming soon.

This resolves:
    http://tracker.ceph.com/issues/4656Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

Define and use functions that encapsulate the initializion of a
ceph_osd_data structure.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

Defer building the osd request until just before submitting it in
all callers except ceph_writepages_start().  (That caller will be
handed in the next patch.)
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

This patch moves the call to ceph_osdc_build_request() out of
ceph_osdc_new_request() and into its caller.

This is in order to defer formatting osd operation information into
the request message until just before request is started.

The only unusual (ab)user of ceph_osdc_build_request() is
ceph_writepages_start(), where the final length of write request may
change (downward) based on the current inode size or the oldest
snapshot context with dirty data for the inode.

The remaining callers don't change anything in the request after has
been built.

This means the ops array is now supplied by the caller.  It also
means there is no need to pass the mtime to ceph_osdc_new_request()
(it gets provided to ceph_osdc_build_request()).  And rather than
passing a do_sync flag, have the number of ops in the ops array
supplied imply adding a second STARTSYNC operation after the READ or
WRITE requested.

This and some of the patches that follow are related to having the
messenger (only) be responsible for filling the content of the
message header, as described here:
    http://tracker.ceph.com/issues/4589Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

We should advance the user data pointer by _len_ instead of _written_.
_len_ is the data length written in each iteration while _written_ is the
accumulated data length we have writtent out.
Signed-off-by: NHenry C Chang <henry.cy.chang@gmail.com>
Reviewed-by: NGreg Farnum <greg@inktank.com>
Tested-by: NSage Weil <sage@inktank.com>

Record the byte count for an osd request rather than the page count.
The number of pages can always be derived from the byte count (and
alignment/offset) but the reverse is not true.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

An osd request defines information about where data to be read
should be placed as well as where data to write comes from.
Currently these are represented by common fields.

Keep information about data for writing separate from data to be
read by splitting these into data_in and data_out fields.

This is the key patch in this whole series, in that it actually
identifies which osd requests generate outgoing data and which
generate incoming data.  It's less obvious (currently) that an osd
CALL op generates both outgoing and incoming data; that's the focus
of some upcoming work.

This resolves:
    http://tracker.ceph.com/issues/4127Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

An osd request uses either pages or a bio list for its data.  Use a
union to record information about the two, and add a data type
tag to select between them.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

Pull the fields in an osd request structure that define the data for
the request out into a separate structure.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

Currently ceph_osdc_new_request() assigns an osd request's
r_num_pages and r_alignment fields.  The only thing it does
after that is call ceph_osdc_build_request(), and that doesn't
need those fields to be assigned.

Move the assignment of those fields out of ceph_osdc_new_request()
and into its caller.  As a result, the page_align parameter is no
longer used, so get rid of it.

Note that in ceph_sync_write(), the value for req->r_num_pages had
already been calculated earlier (as num_pages, and fortunately
it was computed the same way).  So don't bother recomputing it,
but because it's not needed earlier, move that calculation after the
call to ceph_osdc_new_request().  Hold off making the assignment to
r_alignment, doing it instead r_pages and r_num_pages are
getting set.

Similarly, in start_read(), nr_pages already holds the number of
pages in the array (and is calculated the same way), so there's no
need to recompute it.  Move the assignment of the page alignment
down with the others there as well.

This and the next few patches are preparation work for:
    http://tracker.ceph.com/issues/4127Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

(This is being reposted.  The first one had a problem because it
erroneously added a similar change elsewhere; that change has been
dropped.)

The next patch in this series points out that the calculation for
the number of pages in an osd request is getting done twice.  It
is not obvious, but the result of both calculations is identical.
This patch simplifies one of them--as a separate step--to make
it clear that the transformation in the next patch is valid.

In ceph_sync_write() there is some magic that computes page_align
for an osd request.  But a little analysis shows it can be
simplified.

First, we have:
 	io_align = pos & ~PAGE_MASK;
which is used here:
	page_align = (pos - io_align + buf_align) & ~PAGE_MASK;

Note (pos - io_align) simply rounds "pos" down to the nearest multiple
of the page size.

We also have:
 	buf_align = (unsigned long)data & ~PAGE_MASK;

Adding buf_align to that rounded-down "pos" value will stay within
the same page; the result will just be offset by the page offset for
the "data" pointer.  The final mask therefore leaves just the value
of "buf_align".

One more simplification.  Note that the result of calc_pages_for()
is invariant of which page the offset starts in--the only thing that
matters is the offset within the starting page.  We will have
put the proper page offset to use into "page_align", so just use
that in calculating num_pages.

This resolves:
    http://tracker.ceph.com/issues/4166Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

make __ceph_do_pending_vmtruncate() acquire the i_mutex if the caller
does not hold the i_mutex, so ceph_aio_read() can call safely.
Signed-off-by: NYan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: NGreg Farnum <greg@inktank.com>

ceph_aio_write() has an optimization that marks CEPH_CAP_FILE_WR
cap dirty before data is copied to page cache and inode size is
updated. The optimization avoids slow cap revocation caused by
balance_dirty_pages(), but introduces inode size update race. If
ceph_check_caps() flushes the dirty cap before the inode size is
updated, MDS can miss the new inode size. So just remove the
optimization.
Signed-off-by: NYan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: NGreg Farnum <greg@inktank.com>

commit 22cddde1 breaks the atomicity of write operation, it also
introduces a deadlock between write and truncate.
Signed-off-by: NYan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: NGreg Farnum <greg@inktank.com>

Conflicts:
	fs/ceph/addr.c

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

The "num_reply" parameter to ceph_osdc_new_request() is never
used inside that function, so get rid of it.

Note that ceph_sync_write() passes 2 for that argument, while all
other callers pass 1.  It doesn't matter, but perhaps someone should
verify this doesn't indicate a problem.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NJosh Durgin <josh.durgin@inktank.com>

The mds now sends back a created inode if the create request
performed the create.  If the file already existed, no inode is
returned in the reply.  This allows ceph to set the created flag
in atomic_open so that permissions are properly checked in the case
that the file wasn't created by the create call to the mds.

To ensure compability with previous kernels, a feature for sending
back the inode in the create reply was added, so that the mds will
only send back the inode if the client indicates it supports the
feature.
Signed-off-by: NSam Lang <sam.lang@inktank.com>
Reviewed-by: NSage Weil <sage@inktank.com>

The error returned by ceph_mdsc_do_request includes errors sending the
request, errors on timeout, or any errors coming from the mds.  If
ceph_mdsc_do_request returns an error, the reply struct will most likely
be bogus.  We need to bail out and propogate the error instead of
overwriting it.
Signed-off-by: NSam Lang <sam.lang@inktank.com>
Reviewed-by: NSage Weil <sage@inktank.com>

But the kernel decided to call it "origin" instead.  Fix most of the
sites.
Acked-by: NHugh Dickins <hughd@google.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

ceph_aio_write() has an optimization that marks cap EPH_CAP_FILE_WR
dirty before data is copied to page cache and inode size is updated.
If ceph_check_caps() flushes the dirty cap before the inode size is
updated, MDS can miss the new inode size. The fix is move
ceph_{get,put}_cap_refs() into ceph_write_{begin,end}() and call
__ceph_mark_dirty_caps() after inode size is updated.
Signed-off-by: NYan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: NSage Weil <sage@inktank.com>

If we are creating an osd request and get an invalid layout, return
an EINVAL to the caller.  We switch up the return to have an error
code instead of NULL implying -ENOMEM.
Signed-off-by: NSage Weil <sage@inktank.com>
Reviewed-by: NAlex Elder <elder@inktank.com>

The initial ->atomic_open op was carried over from the old intent code,
which was incomplete and didn't really work.  Replace it with a fresh
method.  In particular:

 * always attempt to do an atomic open+lookup, both for the create case
   and for lookups of existing files.
 * fix symlink handling by returning 1 to the VFS so that we can follow
   the link to its destination. This fixes a longstanding ceph bug (#2392).
Signed-off-by: NSage Weil <sage@inktank.com>

Just pass struct file *.  Methods are happier that way...
There's no need to return struct file * from finish_open() now,
so let it return int.  Next: saner prototypes for parts in
namei.c
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Change of calling conventions:
old		new
NULL		1
file		0
ERR_PTR(-ve)	-ve

Caller *knows* that struct file *; no need to return it.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

... and let finish_open() report having opened the file via that sucker.
Next step: don't modify od->filp at all.

[AV: FILE_CREATE was already used by cifs; Miklos' fix folded]
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Add an ->atomic_open implementation which replaces the atomic lookup+open+create
operation implemented via ->lookup and ->create operations.
Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
CC: Sage Weil <sage@newdream.net>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

What was the purpose of this?
Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
CC: Sage Weil <sage@newdream.net>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

This was an ill-conceived feature that has been removed from Ceph.  Do
this gracefully:

 - reject attempts to specify a preferred_osd via the ioctl
 - stop exposing this information via virtual xattrs
 - always fill in -1 for requests, in case we talk to an older server
 - don't calculate preferred_osd placements/pgids
Reviewed-by: NAlex Elder <elder@inktank.com>
Signed-off-by: NSage Weil <sage@inktank.com>

Commit 06222e49 got the if wrong so that
it always evaluates as true.  This is semantically harmless, but makes
SEEK_CUR and SEEK_SET needlessly query the server.

Rewrite the if to explicitly enumerate the cases we DO need a valid i_size
to make this code less fragile.
Reported-by: NRoel Kluin <roel.kluin@gmail.com>
Signed-off-by: NSage Weil <sage@newdream.net>

We have been using i_lock to protect all kinds of data structures in the
ceph_inode_info struct, including lists of inodes that we need to iterate
over while avoiding races with inode destruction.  That requires grabbing
a reference to the inode with the list lock protected, but igrab() now
takes i_lock to check the inode flags.

Changing the list lock ordering would be a painful process.

However, using a ceph-specific i_ceph_lock in the ceph inode instead of
i_lock is a simple mechanical change and avoids the ordering constraints
imposed by igrab().
Reported-by: NAmon Ott <a.ott@m-privacy.de>
Signed-off-by: NSage Weil <sage@newdream.net>

d_parent is protected by d_lock: use it when looking up a dentry's parent
directory inode.  Also take a reference and drop it in the caller to avoid
a use-after-free.
Reported-by: NAl Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>

We weren't properly calling lookup_instantiate_filp when setting up the
lookup intent, which could lead to file leakage on errors.  So:

 - use separate helper for the hidden snapdir translation, immediately
   following the mds request
 - use ceph_finish_lookup for the final dentry/return value dance in the
   exit path
 - lookup_instantiate_filp on success
Reported-by: NAl Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>

We only need to put these on the directory unsafe list if they have
side effects that fsync(2) should flush out.
Reviewed-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>

We were always getting NULL here because the intent file f_dentry is always
NULL at this point, which means we were always passing NULL to
ceph_mdsc_do_request.  In reality, this was fine, since this isn't
currently ever a write operation that needs to get strung on the dir's
unsafe list.

Use the dir explicitly, and only pass it if this open has side-effects that
a dir fsync should flush.
Reviewed-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>

The generic_file_aio_write call may block on balance_dirty_pages while we
flush data to the OSDs.  If we hold a reference to the FILE_WR cap during
that interval revocation by the MDS (e.g., to do a stat(2)) may be very
slow.
Reviewed-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>

This allows us to force IO through the sync path which you normally only
get when multiple clients are reading/writing to the same file or by
mounting with -o sync.  Among other things, this lets test programs verify
correctness with a single mount.
Reviewed-by: NYehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: NSage Weil <sage@newdream.net>