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/3847
Signed-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/4709
Signed-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/4773
Signed-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/4724
Signed-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/3419
Signed-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/4762
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
This is a step toward fully implementing layered writes.
Add checks before request submission for the object(s) associated
with an image request. For write requests, if we don't know that
the target object exists, issue a STAT request to find out. When
that request completes, mark the known and exists flags for the
original object request accordingly and re-submit the object
request. (Note that this still does the existence check only; the
copyup operation is not yet done.)
A new object request is created to perform the existence check. A
pointer to the original request is added to that object request to
allow the stat request to re-issue the original request after
updating its flags. If there is a failure with the stat request
the error code is stored with the original request, which is then
completed.
This resolves:
http://tracker.ceph.com/issues/3418
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
This creates two new flags for object requests to indicate what is
known about the existence of the object to which a request is to be
sent. The KNOWN flag will be true if the the EXISTS flag is
meaningful. That is:
KNOWN EXISTS
----- ------
0 0 don't know whether the object exists
0 1 (not used/invalid)
1 0 object is known to not exist
1 0 object is known to exist
This will be used in determining how to handle write requests for
data objects for layered rbd images.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
In a few spots, whether the an object request's img_request pointer
is null is used to determine whether an object request is being done
as part of an image data request.
Stop doing that, and instead always use the object request IMG_DATA
flag for that purpose. Swap the order of the definition of the
IMG_DATA and DONE flag helpers, because obj_request_done_set() now
refers to obj_request_img_data_set() to get its rbd_dev value.
This will become important because the img_request pointer is
about to become part of a union.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
An extra reference is taken when an object request is added as one
of the requests making up an image object. A reference is dropped
again when the image's object requests get submitted.
The original reference for the object request will remain throughout
this period, so we don't need to add and then take away an extra
one.
This can be interpreted as the image request inheriting the original
object request's reference.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Allow osd request ops that aren't otherwise structured (not class,
extent, or watch ops) to specify "raw" data to be used to hold
incoming data for the op. Make use of this capability for the osd
STAT op.
Prefix the name of the private function osd_req_op_init() with "_",
and expose a new function by that (earlier) name whose purpose is to
initialize osd ops with (only) implied data.
For now we'll just support the use of a page array for an osd op
with incoming raw data.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
There are a bunch of functions defined to encapsulate getting the
address of a data field for a particular op in an osd request.
They're all defined the same way, so create a macro to take the
place of all of them.
Two of these are used outside the osd client code, so preserve them
(but convert them to use the new macro internally). Stop exporting
the ones that aren't used elsewhere.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
In the incremental move toward supporting distinct data items in an
osd request some of the functions had "write_request" parameters to
indicate, basically, whether the data belonged to in_data or the
out_data. Now that we maintain the data fields in the op structure
there is no need to indicate the direction, so get rid of the
"write_request" parameters.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Fix printk format warnings by using %zd for 'ssize_t' variables:
fs/ceph/file.c:751:2: warning: format '%ld' expects argument of type 'long int', but argument 11 has type 'ssize_t' [-Wformat]
fs/ceph/file.c:762:2: warning: format '%ld' expects argument of type 'long int', but argument 11 has type 'ssize_t' [-Wformat]
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: ceph-devel@vger.kernel.org
Signed-off-by: Sage Weil <sage@inktank.com>
ceph_writepages_start() reads inode->i_size in two places. It can get
different values between successive read, because truncate can change
inode->i_size at any time. The race can lead to mismatch between data
length of osd request and pages marked as writeback. When osd request
finishes, it clear writeback page according to its data length. So
some pages can be left in writeback state forever. The fix is only
read inode->i_size once, save its value to a local variable and use
the local variable when i_size is needed.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: Alex Elder <elder@inktank.com>
copy write checks in __generic_file_aio_write to ceph_aio_write.
To make these checks cover sync write path.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: Alex Elder <elder@inktank.com>
There is deadlock as illustrated bellow. The fix is taking i_mutex
before getting Fw cap reference.
write truncate MDS
--------------------- -------------------- --------------
get Fw cap
lock i_mutex
lock i_mutex (blocked)
request setattr.size ->
<- revoke Fw cap
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: Alex Elder <elder@inktank.com>
Reviewed-by: Sage Weil <sage@inktank.com>
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/4706
Reported-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Yan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: Sage 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/4706
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Yan, Zheng <zheng.z.yan@intel.com>
We don't need to use up entropy to choose an mds,
so use prandom_u32() to get a pseudo-random number.
Also, we don't need to choose a random mds if only
one mds is available, so add special casing for the
common case.
Fixes http://tracker.ceph.com/issues/3579
Signed-off-by: Sam Lang <sam.lang@inktank.com>
Reviewed-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Alex Elder <elder@inktank.com>
Implement layered read requests for format 2 rbd images.
If an rbd image is a clone of a snapshot, the snapshot will be the
clone's "parent" image. When an object read request on a clone
comes back with ENOENT it indicates that the clone is not yet
populated with that portion of the image's data, and the parent
image should be consulted to satisfy the read.
When this occurs, a new image request is created, directed to the
parent image. The offset and length of the image are the same as
the image-relative offset and length of the object request that
produced ENOENT. Data from the parent image therefore satisfies the
object read request for the original image request.
While this code works, it will not be active until we enable the
layering feature (by adding RBD_FEATURE_LAYERING to the value of
RBD_FEATURES_SUPPORTED).
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Call the probe function for the parent device if one is present.
Since we don't formally support the layering feature we won't
be using this functionality just yet.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Add a flag to distinguish between object requests being done on
standalone objects and requests being sent for objects representing
rbd image data (i.e., object requests that are the result of image
request).
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
We're going to need some more Boolean values for object requests,
so create a flags bit field and use it to record whether the request
is done.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Encapsulate the code that completes processing of an object request
that's part of an image request.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Define a flag indicating whether an image request is for a layered
image (one with a parent image to which requests will be redirected
if the target object of a request does not exist). The code that
checks this flag will be added shortly.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Define a flag indicating whether an image request originated from
the Linux block layer (from blk_fetch_request()) or whether it was
initiated in order to satisfy an object request for a child image
of a layered rbd device. For image requests initiated by objects of
child images we'll save a pointer to the object request rather than
the Linux block request.
For now, only block requests are used.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
There are several Boolean values we'll be maintaining for image
requests. Switch from the single write_request field to a
general-purpose flags field, and use one if its bits to represent
the direction of I/O for the image request. Define helper functions
for setting and testing that flag.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
For an image object request we will need to know what offset within
the rbd image the request covers. Record that when the object
request gets created.
Update the I/O error warnings so they use this so what's reported
is more informative.
Rename a local variable to fit the convention used everywhere else.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Compute the total number of bytes transferred for an image
request--the sum across each of the request's object requests.
To avoid contention do it only when all object requests are
complete, in rbd_img_request_complete().
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
If any image object request produces a non-zero result, preserve
that as the result of the overall image request. If multiple
objects have non-zero results, save only the first one.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
There is a new rbd feature bit defined for "fancy striping." Add
it to the ones defined in the kernel client.
Change RBD_FEATURES_ALL so it represents the set of all feature
bits (rather than just the ones we support). Define a new symbol
RBD_FEATURES_SUPPORTED to indicate the supported ones.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Right now the data for a method call is specified via a pointer and
length, and it's copied--along with the class and method name--into
a pagelist data item to be sent to the osd. Instead, encode the
data in a data item separate from the class and method names.
This will allow large amounts of data to be supplied to methods
without copying. Only rbd uses the class functionality right now,
and when it really needs this it will probably need to use a page
array rather than a page list. But this simple implementation
demonstrates the functionality on the osd client, and that's enough
for now.
This resolves:
http://tracker.ceph.com/issues/4104
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Change the names of the functions that put data on a pagelist to
reflect that we're adding to whatever's already there rather than
just setting it to the one thing. Currently only one data item is
ever added to a message, but that's about to change.
This resolves:
http://tracker.ceph.com/issues/2770
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
This patch adds support to the messenger for more than one data item
in its data list.
A message data cursor has two more fields to support this:
- a count of the number of bytes left to be consumed across
all data items in the list, "total_resid"
- a pointer to the head of the list (for validation only)
The cursor initialization routine has been split into two parts: the
outer one, which initializes the cursor for traversing the entire
list of data items; and the inner one, which initializes the cursor
to start processing a single data item.
When a message cursor is first initialized, the outer initialization
routine sets total_resid to the length provided. The data pointer
is initialized to the first data item on the list. From there, the
inner initialization routine finishes by setting up to process the
data item the cursor points to.
Advancing the cursor consumes bytes in total_resid. If the resid
field reaches zero, it means the current data item is fully
consumed. If total_resid indicates there is more data, the cursor
is advanced to point to the next data item, and then the inner
initialization routine prepares for using that. (A check is made at
this point to make sure we don't wrap around the front of the list.)
The type-specific init routines are modified so they can be given a
length that's larger than what the data item can support. The resid
field is initialized to the smaller of the provided length and the
length of the entire data item.
When total_resid reaches zero, we're done.
This resolves:
http://tracker.ceph.com/issues/3761
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
In place of the message data pointer, use a list head which links
through message data items. For now we only support a single entry
on that list.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Rather than having a ceph message data item point to the cursor it's
associated with, have the cursor point to a data item. This will
allow a message cursor to be used for more than one data item.
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
Alex Elder