Hold off setting the EXISTS rbd device flag until just before we
announce the disk as available for use.  There's no point in doing
so any earlier than that, and at that point the device truly is
fully set up and ready to use.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

This just tweaks a few things in the routines that implement
rbd sysfs files.

All of the entries for an rbd device in /sys/bus/rbd/devices/<id>/
will represent information whose valid values are known by the time
they are accessible.

Right now we get the size of the mapped image by a call to
get_capacity().  There's no need to do this, because that will
return what we last set the capacity to, which is just the size
recorded for the mapping.  So just show that value instead.

We also get this under protection of the header semaphore, in order
to provide a precisely correct value.  This isn't really necessary;
these files are really informational only and it's not necessary to
be so careful.

Finally, print a special value in case the major device number is
not recorded.  Right now that won't matter much but soon the parent
images won't have devices associated with them.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

When a snapshot context update occurs, rbd_update_mapping_size() is
called to set the capacity of the disk to record the updated
size of the image in case it has changed.

There's a bug though.  The mapping size is in units of *bytes*.  The
code that updates the mapping size field is assigning a value that
has been scaled down to *sectors*.

Fix that.  Also, check to see if the size has actually changed, and
don't bother updating things (specifically, calling set_capacity())
if it has not.

This resolves:
    http://tracker.ceph.com/issues/4833Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Fairly straightforward refactoring of rbd_dev_probe_update_spec().
The name is changed to rbd_dev_spec_update().

Rearrange it so nothing gets assigned to the spec until all of the
names have been successfully acquired.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Rename rbd_dev_probe() to be rbd_dev_image_probe().  Its purpose
will eventually be to probe for the existence of a valid rbd image
for the rbd device--focusing only on the ceph side and not the Linux
device side of initialization.

For now the two "sides" are not fully separated, and this function
is still the entry point for initializing the full rbd device.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Currently, rbd_dev_destroy() does more than just the inverse of what
rbd_dev_create() does.  Stop doing that, and move the two extra
things it does into the three call sites.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Encapsulate the creation of a snapshot context for rbd in a new
function rbd_snap_context_create().  Define rbd wrappers for getting
and dropping references to them once they're created.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Change some calls to WARN_ON() so they use rbd_warn() instead, so we
get consistent messaging.  A few remain but they can probably just
go away eventually.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

This commit added fetching if fancy striping parameters:
    09186ddb rbd: get and check striping parameters

They are almost unused, but the two fields storing the information
really belonged in the rbd_image_header structure.

This patch moves them there.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Make the names and image id in an rbd_spec be pointers to constant
data.  This required the use of a local variable to hold the
snapshot name in rbd_add_parse_args() to avoid a warning.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Set the rbd spec's snapshot id for an image getting mapped in
rbd_dev_probe_update_spec() rather than rbd_dev_set_mapping().
This is the more logical place for that to happen (even though
it means we might look up the snapshot by name twice).
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

A function called snap_by_name() ought to just look up a snapshot by
name.  It does that, but then it assigns some stuff to the rbd
device structure as well.

Change the function to do just the lookup, and have the caller do
the assignments that follow.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

If a format 2 image id is found for an image being mapped, but the
subsequent probe of the image fails, rbd_dev_probe() quits without
freeing the image id.  Fix that.

Also drop a redundant hunk of code in rbd_dev_image_id().
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Currently, rbd_dev_probe() assumes that any error returned by
rbd_dev_image_id() is most likely -ENOENT, and responds by
calling the format 1 probe routine, rbd_dev_v1_probe().  Then,
at the top of rbd_dev_v1_probe(), an empty string is allocated
for the image id.

This is sort of unbalanced.  Fix this by having rbd_dev_image_id()
look for -ENOENT from its "get_id" method call.  If that is seen,
have it allocate the empty string there rather than depending on
rbd_dev_v1_probe() to do it.

Given that this is effectively defining the format of the image,
set rbd_dev->image_format inside rbd_dev_image_id() rather than in
the format-specific probe routines.

Also drop a redundant hunk of code in rbd_dev_image_id().
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

I found during some failure injection testing that the call to
rbd_free_disk() in the error path of rbd_dev_probe_finish() was
dropping an extra reference to the disk queue.  The problem
occurred when put_disk tried to drop a reference to the disk's
queue.  A call to blk_cleanup_queue() just prior to that will have
also dropped a reference to the queue.

The problem is that the reference dropped by put_disk() is assumed
to have been taken by add_disk().  Our code has error paths that can
occur after the disk and its queue are initialized, but before the
call to add_disk(), and in those paths we won't have that extra
reference.

The fix is easy though.  In rbd_free_disk() we're already checking
the disk's GENHD_FL_UP flag.  That flag is an indication that
add_disk() has been called, so just call blk_cleanup_queue()
conditional on that flag being set.

This resolves:
    http://tracker.ceph.com/issues/4800Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Now that rbd_obj_method_sync() returns the number of bytes
returned by the method call, that value should be used by
callers to ensure we don't overrun the valid portion of the
buffer.

Fix the two spots that remained that weren't doing that,
rbd_dev_image_name() and rbd_dev_v2_snap_name().

Rearrange the error path slightly in rbd_dev_v2_snap_name().
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

When the snapshot context for an rbd device gets updated (or the
initial one is recorded) a a list of snapshot structures is created
to represent them, one entry per snapshot.  Each entry includes a
dynamically-allocated copy of the snapshot name.

Currently the name is allocated in rbd_snap_create(), as a duplicate
of the passed-in name.

For format 1 images, the snapshot name provided is just a pointer to
an existing name.  But for format 2 images, the passed-in name is
already dynamically allocated, and in the the process of duplicating
it here we are leaking the passed-in name.

Fix this by dynamically allocating the name for format 1 snapshots
also, and then stop allocating a duplicate in rbd_snap_create().

Change rbd_dev_v1_snap_info() so none of its parameters is
side-effected unless it's going to return success.

This is part of:
    http://tracker.ceph.com/issues/4803Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Rename __rbd_add_snap_dev() to be rbd_snap_create().  We no longer
have devices for non-mapped snapshots, and we're not actually
"adding" it to the list in this function, just creating it.

Rename rbd_remove_snap_dev() to be rbd_snap_destroy() for reasons
similar to the above.  Stop having this function delete the snapshot
from its list (to be symmetrical with its create counterpart) and do
that in the caller instead.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Change rbd_dev_v2_snap_info() so it only ever sets values of the
size and features parameters if looking up the snapshot name was
successful.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Only one of the two callers of _rbd_dev_v2_snap_size() needs the
order value returned.  So make that an optional argument--a null
pointer if the caller doesn't need it.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

When an rbd image is initially mapped, its snapshot context is
collected, and then a list of snapshot entries representing the
snapshots in that context is created.  The list is created using
rbd_dev_snaps_update().  (This function also supports updating an
existing snapshot list based on a new snapshot context.)

If an error occurs, updating the list is aborted, and the list is
currently left as-is, in an inconsistent state.  At that point,
there may be a partially-constructed list, but the calling functions
(rbd_dev_probe_finish() from rbd_dev_probe() from rbd_add()) never
clean them up.  So this constitutes a leak.

A snapshot list that is inconsistent with the current snapshot
context is of no use, and might even be actively bad.  So rather
than just having the caller clean it up, have rbd_dev_snaps_update()
just clear out the entire snapshot list in the event an error
occurs.

The other place rbd_dev_snaps_update() is used is when a refresh is
triggered, either because of a watch callback or via a write to the
/sys/bus/rbd/devices/<id>/refresh interface.  An error while
updating the snapshots has no substantive effect in either of those
cases, but one of them issues a warning.  Move that warning to the
common rbd_dev_refresh() function so it gets issued regardless of
how it got initiated.

This is part of:
    http://tracker.ceph.com/issues/4803Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

When an rbd image gets mapped a device entry gets created for it
under /sys/bus/rbd/devices/<id>/.  Inside that directory there are
sysfs files that contain information about the image: its size,
feature bits, major device number, and so on.

Additionally, if that image has any snapshots, a device entry gets
created for each of those as a "child" of the mapped device.  Each
of these is a subdirectory of the mapped device, and each directory
contains a few files with information about the snapshot (its
snapshot id, size, and feature mask).

There is no clear benefit to having those device entries for the
snapshots.  The information provided via sysfs of of little real
value--and all of it is available via rbd CLI commands.  If we
still wanted to see the kernel's view of this information it could
be done much more simply by including it in a single sysfs file for
the mapped image.

But there *is* a clear cost to supporting them.  Every time a snapshot
context changes, these entries need to be updated (deleted snapshots
removed, new snapshots created).  The rbd driver is notified of
changes to the snapshot context via callbacks from an osd, and care
must be taken to coordinate removal of snapshot data structures
with the possibility of one these notifications occurring.

Things would be considerably simpler if we just didn't have to
maintain device entries for the snapshots.

So get rid of them.

The ability to map a snapshot of an rbd image will remain; the only
thing lost will be the ability to query these sysfs directories for
information about snapshots of mapped images.

This resolves:
    http://tracker.ceph.com/issues/4796Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

A WATCH op includes an object version.  The version that's supplied
is incorrectly byte-swapped osd_req_op_watch_init() where it's first
assigned (it's been this way since that code was first added).

The result is that the version sent to the osd is wrong, because
that value gets byte-swapped again in osd_req_encode_op().  This
is the source of a sparse warning related to improper byte order in
the assignment.

The approach of using the version to avoid a race is deprecated
(see http://tracker.ceph.com/issues/3871), and the watch parameter
is no longer even examined by the osd.  So fix the assignment in
osd_req_op_watch_init() so it no longer does the byte swap.

This resolves:
    http://tracker.ceph.com/issues/3847Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Now that we have most everything in place to support layered rbd
images, enable support for them in the kernel client.  Issue a
warning to the log that the support is considered experimental
whenever a format 2 layered image is mapped.

Note that we also have to claim to support the STRIPINGV2 feature,
due to a mistake in the way the rbd CLI set up those flags.  This
feature can work if it has the right parameters, and safeguards
have been put in place to reject those images that do not have
compatible parameters.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

If an rbd format 2 image indicates it supports the STRIPINGV2
feature we need to find out its stripe unit and stripe count in
order to know whether we can use it.  We don't yet support fancy
striping fully, but if the default parameters are used the behavior
is indistinguishible from non-fancy striping.

This is necessary because some images require the STRIPINGV2 feature
even if they use the default parameters.  (Which is to say the feature
bit was erroneously set even if the feature was not used.)

This resolves:
    http://tracker.ceph.com/issues/4709Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Callers of rbd_obj_method_sync() don't know how many bytes of data
got returned by the class method call.  As a result, they have been
assuming enough got returned to decode whatever was expected.

This isn't safe.  We know how many bytes got transferred, so have
rbd_obj_method_sync() return that amount (rather than just 0) if
the call is successful.

Change all callers to use this return value to ensure decoding of
the results is done safely.

On the other hand, most callers of rbd_obj_method_sync() only
indicate success or failure, so all of *their* callers can simply
test for non-zero result.

This resolves:
    http://tracker.ceph.com/issues/4773Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Make the inbound and outbound data parameters have void rather than
character type for rbd_obj_method_sync().  This makes it more clear
they don't expect typed data, and eliminates the need for some silly
type casts.

One more unrelated change: define the features buffer used in
_rbd_dev_v2_snap_features() to be a packed data structure.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Make the buf parameter into which the data is to be read have type
void pointer.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

A ceph timespec contains 32-bit unsigned values for its seconds and
nanoseconds components.  For a standard timespec, both fields are
signed, and the seconds field is almost surely 64 bits.

Add some explicit casts so the fact that this conversion is taking
place is obvious.  Also trip a bug if we ever try to put out of
range (negative or too big) values into a ceph timespec.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Flesh out the limits defined in <linux/ceph/decode.h> to include the
maximum and minimum values for signed type S8, S16, S32, and S64.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

A clone image has a defined overlap point with its parent image.
That is the byte offset beyond which the parent image has no
defined data to back the clone, and anything thereafter can be
viewed as being zero-filled by the clone image.

This is needed because a clone image can be resized.  If it gets
resized larger than the snapshot it is based on, the overlap defines
the original size.  If the clone gets resized downward below the
original size the new clone size defines the overlap.  If the clone
is subsequently resized to be larger, the overlap won't be increased
because the previous resize invalidated any parent data beyond that
point.

This resolves:
    http://tracker.ceph.com/issues/4724Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

This implements the main copyup functionality for layered writes.

Here we add a copyup_pages field to the object request, which is
used only for copyup requests to keep track of the page array
containing data read from the parent image.

A copyup request is currently the only request rbd has that requires
two osd operations.  Because of this we handle copyup specially.
All image object requests get an osd request allocated when they are
created.  For a write request, if a copyup is required, the osd
request originally allocated is released, and a new one (with room
for two osd ops) is allocated to replace it.  A new function
rbd_osd_req_create_copyup() allocates an osd request suitable for
a copyup request.

The first op is then filled with a copyup object class method call,
supplying the array of pages containing data read from the parent.
The second op is filled in with the original write request.

The original request otherwise remains intact, and it describes the
original write request (found in the second osd op).  The presence
of the copyup op is sort of implicit; a non-null copyup_pages field
could be used to distinguish between a "normal" write request and a
request containing both a copyup call and a write.

This resolves:
    http://tracker.ceph.com/issues/3419Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

As a step toward implementing layered writes, implement reading the
data for a target object from the parent image for a write request
whose target object is known to not exist.  Add a copyup_pages field
to an image request to track the page array used (only) for such a
request.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

If rbd disk is open and rbd resize is done, new size is not
visible by filesystem.  Like is done in virtio-blk and dm driver,
revalidate_disk() permits to update the bd_inode size.
Signed-off-by: Laurent Barbe <laurent@ksperis.com>
Reviewed-by: Alex Elder <elder@inktank.com>

This patch adds the ability to build an image request whose data
will be written from or read into memory described by a page array.
(Previously only bio lists were supported.)

Originally this was going to define a new function for this purpose
but it was largely identical to the rbd_img_request_fill_bio().  So
instead, rbd_img_request_fill_bio() has been generalized to handle
both types of image request.

For the moment we still only fill image requests with bio data.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Define a new function zero_pages() that zeroes a range of memory
defined by a page array, along the lines of zero_bio_chain().  It
saves and the irq flags like bvec_kmap_irq() does, though I'm not
sure at this point that it's necessary.

Update rbd_img_obj_request_read_callback() to use the new function
if the object request contains page rather than bio data.

For the moment, only bio data is used for osd READ ops.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Object requests that are part of an image request are subject to
some additional handling.  Define rbd_img_obj_request_submit() to
encapsulate that, and use it when initially submitting an image
object request, and when re-submitting it during callback of
an object existence check.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Separate rbd_osd_req_format() into two functions, one for read
requests and the other for write requests.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Add the ability to provide an array of pages as outbound request
data for object class method calls.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

This patch makes four small changes in the ceph messenger.

While getting copyup functionality working I found two bugs in the
messenger.  Existing paths through the code did not trigger these
problems, but they're fixed here:
    - In ceph_msg_data_pagelist_cursor_init(), the cursor's
      last_piece field was being checked against the length
      supplied.  This was OK until this commit: ccba6d98 libceph:
      implement multiple data items in a message That commit changed
      the cursor init routines to allow lengths to be supplied that
      exceeded the size of the current data item. Because of this,
      we have to use the assigned cursor resid field rather than the
      provided length in determining whether the cursor points to
      the last piece of a data item.
    - In ceph_msg_data_add_pages(), a BUG_ON() was erroneously
      catching attempts to add page data to a message if the message
      already had data assigned to it. That was OK until that same
      commit, at which point it was fine for messages to have
      multiple data items. It slipped through because that BUG_ON()
      call was present twice in that function. (You can never be too
      careful.)

In addition two other minor things are changed:
    - In ceph_msg_data_cursor_init(), the local variable "data" was
      getting assigned twice.
    - In ceph_msg_data_advance(), it was assumed that the
      type-specific advance routine would set new_piece to true
      after it advanced past the last piece. That may have been
      fine, but since we check for that case we might as well set it
      explicitly in ceph_msg_data_advance().

This resolves:
    http://tracker.ceph.com/issues/4762Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>