linux-toradex.git/include/linux/ceph, branch v6.0-rc7

libceph: clean up ceph_osdc_start_request prototype

2022-08-03T12:05:39+00:00

This function always returns 0, and ignores the nofail boolean. Drop the
nofail argument, make the function void return and fix up the callers.

Signed-off-by: Jeff Layton 
Reviewed-by: Ilya Dryomov 
Signed-off-by: Ilya Dryomov

ceph: fix incorrect old_size length in ceph_mds_request_args

2022-08-02T22:54:12+00:00

The 'old_size' is a __le64 type since birth, not sure why the
kclient incorrectly switched it to __le32. This change is okay
won't break anything because union will always allocate more memory
than the 'open' member needed.

Rename 'file_replication' to 'pool' as ceph did. Though this 'open'
struct may never be used in kclient in future, it's confusing when
going through the ceph code.

Signed-off-by: Xiubo Li 
Reviewed-by: Ilya Dryomov 
Signed-off-by: Ilya Dryomov

ceph: fix the incorrect comment for the ceph_mds_caps struct

2022-08-02T22:54:12+00:00

The incorrect comment is misleading. Acutally the last members
in ceph_mds_caps strcut is a union for none export and export
bodies.

Signed-off-by: Xiubo Li 
Reviewed-by: Jeff Layton 
Signed-off-by: Ilya Dryomov

ceph: prevent a client from exceeding the MDS maximum xattr size

2022-08-02T22:54:12+00:00

The MDS tries to enforce a limit on the total key/values in extended
attributes.  However, this limit is enforced only if doing a synchronous
operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
doesn't have a chance to enforce these limits.

This patch adds support for decoding the xattrs maximum size setting that is
distributed in the mdsmap.  Then, when setting an xattr, the kernel client
will revert to do a synchronous operation if that maximum size is exceeded.

While there, fix a dout() that would trigger a printk warning:

[   98.718078] ------------[ cut here ]------------
[   98.719012] precision 65536 too large
[   98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
...

Link: https://tracker.ceph.com/issues/55725
Signed-off-by: Luís Henriques 
Reviewed-by: Xiubo Li 
Signed-off-by: Ilya Dryomov

libceph: fix potential use-after-free on linger ping and resends

2022-05-18T19:21:05+00:00

request_reinit() is not only ugly as the comment rightfully suggests,
but also unsafe.  Even though it is called with osdc->lock held for
write in all cases, resetting the OSD request refcount can still race
with handle_reply() and result in use-after-free.  Taking linger ping
as an example:

    handle_timeout thread                     handle_reply thread

                                              down_read(&osdc->lock)
                                              req = lookup_request(...)
                                              ...
                                              finish_request(req)  # unregisters
                                              up_read(&osdc->lock)
                                              __complete_request(req)
                                                linger_ping_cb(req)

      # req->r_kref == 2 because handle_reply still holds its ref

    down_write(&osdc->lock)
    send_linger_ping(lreq)
      req = lreq->ping_req  # same req
      # cancel_linger_request is NOT
      # called - handle_reply already
      # unregistered
      request_reinit(req)
        WARN_ON(req->r_kref != 1)  # fires
        request_init(req)
          kref_init(req->r_kref)

                   # req->r_kref == 1 after kref_init

                                              ceph_osdc_put_request(req)
                                                kref_put(req->r_kref)

            # req->r_kref == 0 after kref_put, req is freed

         !!!

This happens because send_linger_ping() always (re)uses the same OSD
request for watch ping requests, relying on cancel_linger_request() to
unregister it from the OSD client and rip its messages out from the
messenger.  send_linger() does the same for watch/notify registration
and watch reconnect requests.  Unfortunately cancel_request() doesn't
guarantee that after it returns the OSD client would be completely done
with the OSD request -- a ref could still be held and the callback (if
specified) could still be invoked too.

The original motivation for request_reinit() was inability to deal with
allocation failures in send_linger() and send_linger_ping().  Switching
to using osdc->req_mempool (currently only used by CephFS) respects that
and allows us to get rid of request_reinit().

Cc: stable@vger.kernel.org
Signed-off-by: Ilya Dryomov 
Reviewed-by: Xiubo Li 
Acked-by: Jeff Layton

ceph: do not release the global snaprealm until unmounting

2022-03-01T17:26:37+00:00

The global snaprealm would be created and then destroyed immediately
every time when updating it.

URL: https://tracker.ceph.com/issues/54362
Signed-off-by: Xiubo Li 
Reviewed-by: Jeff Layton 
Signed-off-by: Ilya Dryomov

ceph: remove incorrect and unused CEPH_INO_DOTDOT macro

2022-03-01T17:26:37+00:00

Ceph have removed this macro and the 0x3 will be use for global dummy
snaprealm.

Signed-off-by: Xiubo Li 
Reviewed-by: Jeff Layton 
Signed-off-by: Ilya Dryomov

ceph: move to a dedicated slabcache for ceph_cap_snap

2022-03-01T17:26:37+00:00

There could be huge number of capsnaps around at any given time. On
x86_64 the structure is 248 bytes, which will be rounded up to 256 bytes
by kzalloc. Move this to a dedicated slabcache to save 8 bytes for each.

[ jlayton: use kmem_cache_zalloc ]

Signed-off-by: Xiubo Li 
Signed-off-by: Jeff Layton 
Signed-off-by: Ilya Dryomov

ceph: add getvxattr op

2022-03-01T17:26:37+00:00

Problem:
Some directory vxattrs (e.g. ceph.dir.pin.random) are governed by
information that isn't necessarily shared with the client. Add support
for the new GETVXATTR operation, which allows the client to query the
MDS directly for vxattrs.
When the client is queried for a vxattr that doesn't have a special
handler, have it issue a GETVXATTR to the MDS directly.

Solution:
Adds new getvxattr op to fetch ceph.dir.pin*, ceph.dir.layout* and
ceph.file.layout* vxattrs.
If the entire layout for a dir or a file is being set, then it is
expected that the layout be set in standard JSON format. Individual
field value retrieval is not wrapped in JSON. The JSON format also
applies while setting the vxattr if the entire layout is being set in
one go.
As a temporary measure, setting a vxattr can also be done in the old
format. The old format will be deprecated in the future.

URL: https://tracker.ceph.com/issues/51062
Signed-off-by: Milind Changire 
Reviewed-by: Jeff Layton 
Signed-off-by: Ilya Dryomov

libceph: optionally use bounce buffer on recv path in crc mode

2022-02-02T17:50:36+00:00

Both msgr1 and msgr2 in crc mode are zero copy in the sense that
message data is read from the socket directly into the destination
buffer.  We assume that the destination buffer is stable (i.e. remains
unchanged while it is being read to) though.  Otherwise, CRC errors
ensue:

  libceph: read_partial_message 0000000048edf8ad data crc 1063286393 != exp. 228122706
  libceph: osd1 (1)192.168.122.1:6843 bad crc/signature

  libceph: bad data crc, calculated 57958023, expected 1805382778
  libceph: osd2 (2)192.168.122.1:6876 integrity error, bad crc

Introduce rxbounce option to enable use of a bounce buffer when
receiving message data.  In particular this is needed if a mapped
image is a Windows VM disk, passed to QEMU.  Windows has a system-wide
"dummy" page that may be mapped into the destination buffer (potentially
more than once into the same buffer) by the Windows Memory Manager in
an effort to generate a single large I/O [1][2].  QEMU makes a point of
preserving overlap relationships when cloning I/O vectors, so krbd gets
exposed to this behaviour.

[1] "What Is Really in That MDL?"
    https://docs.microsoft.com/en-us/previous-versions/windows/hardware/design/dn614012(v=vs.85)
[2] https://blogs.msmvps.com/kernelmustard/2005/05/04/dummy-pages/

URL: https://bugzilla.redhat.com/show_bug.cgi?id=1973317
Signed-off-by: Ilya Dryomov 
Reviewed-by: Jeff Layton