IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
This allows the vfs backend to detect a retry and keep state between
the retries.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14113
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Returning NT_STATUS_FILE_LOCK_CONFLICT is a SMB1 only detail
for delayed brlock requests, which is handled in
smbd_smb1_do_locks*().
The brlock layer should be consistent even for posix locks.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14113
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
They implement the logic from [MS-FSA].
The following commits will use these functions in other locations.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=14113
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Directly initialized variables give compilers less reason to complain
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Böhme <slow@samba.org>
remove_lease_if_stale() does not have the check
if (e == e2) {
/* Not ourselves. */
continue;
}
that remove_share_mode_lease() had. However, remove_share_mode_lease()
has already set e->op_type=NO_OPLOCK, so that the
if (e->op_type != LEASE_OPLOCK) {
continue;
}
statement has the same effect.
Why? The next commit will need it for proper error path cleanup.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The fsp carries all required information also for leases. There's no
need to pass that as additional parameters
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Yes, this adds another peek from locking/ back into smbd/proto.h, but
locking/locking.c does the same already.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Makes the interface more obvious to me. Also, I want to remove
fsp->share_access, which is not really used anywhere after the fsp has
been fully established.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is not a real error, it happens when the share mode record is not
around.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Uri Simchoni <uri@samba.org>
This removes a kludgy implementation that worked around a locking
hierarchy problem: Setting a byte range lock had to contend the level2
oplocks, which are stored in locking.tdb/leases.tdb. We could not
access locking.tdb in the brlock.tdb code, as brlock.tdb might have
been locked first without locking.tdb, violating the locking hierarchy
locking.tdb->brlock.tdb. Now that that problem is gone (see the commit
wrapping do_lock() in share_mode_do_locked()), we can remove this
kludge.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This caches share_mode_data->flags in the fsp, cache flush happening
on tdb_seqnum change.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
They are put at the beginning for easy parsing without reading the
full struct. First step to remove the number of read oplocks/leases
from brlock.tdb, where it does not belong.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
We need to maintain the locking hierarchy locking.tdb->brlock.tdb at
all times. vfs_fruit directly calls do_lock(), which might fail to
maintain the locking hierarchy: In brlock.c we call
contend_level2_oplocks_begin(), which will soon look at the
locking.tdb record.
For the SMB1 and SMB2 callers we already have the share mode locked,
we might want to watch that record for unlocks. For those callers
share_mode_do_locked() is practically free to call, we share the
underlying db_record.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This is made for efficient locking of share mode records in
locking.tdb. Right now we already need that when accessing leases.tdb,
and soon it will be required for brlock.tdb as well. It does not give
direct access to the parsed share mode entry, but the record is
available for dbwrap_watched_wakeup() within downgrade_lease().
It can be freely nested with get_share_mode_lock calls, the record
will be shared and proper nesting should be checked.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
The next commit will introduce share_mode_do_locked(), which allocates
a share mode record on the stack. We have to expect nested
get_share_mode_lock() calls from within share_mode_do_locked() for
which we need to share a db_record.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Do explicit refcounting instead of talloc_reference(). A later patch
will introduce a share_mode_do_locked() routine that can be nested
arbitrarily with get_share_mode_lock(). To do sanity checks for proper
nesting, share_mode_do_locked needs to be aware of the reference
counts for "static_share_mode_lock".
Why is share_mode_memcache_delete() gone? In parse_share_modes() we
already move the data out of the cache, share_mode_lock_destructor()
we don't even bother re-adding the share_mode_data to the cache if
it does not have share entries, because the next opener will invent a
new seqnum anyway.
Also: Less talloc_reference(), less lines of code.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This routine did a NO-GO: It returned something on talloc_tos(), for
later consumption by push_blocking_lock_request. This is now gone, no
caller uses the "struct byte_range_lock" returned anymore.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This should have been in f11c5887f4, sorry. We now always need
TDB_SEQNUM on brlock.tdb.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
This is not called anymore, bump the VFS version number in a separate
commit
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Jun 20 18:34:20 UTC 2019 on sn-devel-184
This is now handled in smbd_smb1_do_locks_send/recv.
From here on for a few commits we won't survive make test.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Quite a few callers set "psmblctx" to NULL, and I could not really
follow 100% that brl_lock only assigns that in the blocking lock
case. Too many layers :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Soon we will wait on a conflicting lock to become free via
dbwrap_watched_watch_send. That routine can take a server_id that
blocks us, watching it to go away. To use that, we need to know which
PID it is that blocks us.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Just reformatting, no behaviour change. This just looked too ugly to me.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue May 28 22:22:28 UTC 2019 on sn-devel-184
With e7424897a1 we don't need this special case for clustering
anymore.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Fix ubsan warning null pointer passed as argument 2 when the source
pointer is NULL. The calls to memcpy are now guarded by an
if (len > 0)
Signed-off-by: Gary Lockyer <gary@catalyst.net.nz>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Gary Lockyer <gary@samba.org>
Autobuild-Date(master): Mon May 27 01:29:48 UTC 2019 on sn-devel-184
file_id plus share_file_id remotely specify the fsp. This avoids the
explicit loop in the receiver.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Sat May 18 20:18:55 UTC 2019 on sn-devel-184
This also removes the temporary functions introduced during the patchset.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Christof Schmitt <cs@samba.org>
Autobuild-User(master): Christof Schmitt <cs@samba.org>
Autobuild-Date(master): Sun Apr 14 05:18:14 UTC 2019 on sn-devel-144
Function to walk all leases for a file exactly once. This used to be simpler
with the leases[] array, thus this function that encapsulates the complexity.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Christof Schmitt <cs@samba.org>
Temporary patch to keep the code running. The new code in set_share_mode() will
leave again once the patchset to remove share_mode_lease and thus the lease_idx
in share_mode_entry goes away.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Christof Schmitt <cs@samba.org>
Instead of indexing into the leases[] array, put the lease_db reference into
the share_mode_entry. For simplicity, put in the client guid as well. We
*might* be able to retrieve that from somewhere else, but as other smbd
processes have to look at the lease values, put in the full leases_db index
data.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Christof Schmitt <cs@samba.org>
This is the data stored in share_mode_lease inside the leases[] array in
locking.tdb. This and all the following patches move all leases array to
looking at the leases.tdb.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Christof Schmitt <cs@samba.org>
Do the rename under one lock to protect against potential races while
we don't hold it.
Factor out the NDR marshalling into leases_db_do_locked(), leaving the
rename function pretty simple.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We don't need to talloc the blob, it's always the same size
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Oct 3 04:11:59 CEST 2018 on sn-devel-144
gcc complains that the "const" is ignored on function return
types. Right now I'm compiling this file a lot, so silence this
warning :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This function is pretty closely entangled with its only caller. In
particular the NT_STATUS_OPLOCK_BREAK_IN_PROGRESS triggers acitivity
in the caller, and that's the only case where "*_l" is being set to
non-NULL. Prepare for cleanup
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
lock reference count is always increased and reduced by a value of 1.
But lock_ref_count variable holds the old value prior to change and
was being logged wrongly under debug level 10. DEBUG statement must
log lock_ref_count+1 and lock_ref_count-1 respectively when value
gets increased and decreased.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
* posix_pending_close_db is an in-memory database and not a tdb.
Therefore adjusting comments to convey the correct meaning of the
database.
* we do not have posix_locking_close_file() any more which got renamed
to locking_close_file(). Thus fixing comment to mention the new name.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Sep 19 22:23:28 CEST 2018 on sn-devel-144
"the_lock_id" is not required here. The share mode data carry the file
id, so use that.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This loop does not need to count valid share modes. A single valid one
is sufficient for keeping the delete token around
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
This reflects that the messaging context is also used outside of the
server processes.
The command used for the rename:
find . -name '*.[hc]' -print0 | xargs -0 sed -i 's/server_messaging_context/global_messaging_context/'
Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
fetch_share_mode_unlocked_parser() takes a "struct
fetch_share_mode_unlocked_state *" as
"private_data". fetch_share_mode_send() used a talloc_zero'ed "struct
share_mode_lock". This lead to the parser putting a "struct
share_mode_lock on the NULL talloc_context where nobody really picked it
up.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=13602
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Point out in the API that "backend" talloc_moves into the watched
database.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Aug 17 21:29:15 CEST 2018 on sn-devel-144
Fix a confusing API: Many places TALLOC_FREE the path where it's not
clear you have to do it.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
lock_ref_count will always hold the old value prior to change. Thus it
would mean that if lock_ref_count is 0 the new value is already -1 which
is not expected here. Therefore it is better to make sure that it is
always greater than 0 rather than >= 0.
Signed-off-by: Anoop C S <anoopcs@redhat.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Jul 31 04:23:47 CEST 2018 on sn-devel-144
smbstatus was the only user, and this could be solved by adapting
share_entry_forall.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Jul 27 01:42:31 CEST 2018 on sn-devel-144
Quite a bit of the contents have been passed explicitly anyway.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This can be used to check if a file opened by fsp also has stream opens.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=13451
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Feb 13 05:01:38 CET 2018 on sn-devel-144
The file ids in all share modes match the share_mode_data's one
We don't have a paranoia check for this, but the share mode is per inode.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
It's also in the share_entry, but that is redundant and will go
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
As per MS-SMB2 and MS-FSA and our SMB_VFS_STRICT_LOCK implementation,
we're merely testing for locks, not setting any.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12887
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Jul 11 03:37:44 CEST 2017 on sn-devel-144
It's just a noop, so let's remove it. SMB_VFS_STRICT_LOCK doesn't set
logs, it just checks for the presence of incompatible locks.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12887
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This is currently marked 'skip', which means it isn't stored in the
db, but printed out in ndr dump. However, this pointer can be invalid
if the lease_idx is set to 0xFFFFFFFF (invalid).
This is fixed up inside parse_share_modes(), but not until after
ndr_pull_share_mode_data() is called. If lease_idx == 0xFFFFFFFF
then ndr_print_share_mode_lease() prints an invalid value and
crashes.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12793
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Alexander Bokovoy <ab@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu May 18 03:01:40 CEST 2017 on sn-devel-144
Add some helper functions that will be used to update a bunch of checks
for exclusive oplocks to the lease area.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12766
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The boolean out parameter "queued" tells the caller whether the
async request is blocked in a full send queue:
false := request is dispatched
true := send queue is full, request waiting to be dispatched
This is useful in a clustered Samba environment where the async dbwrap
request is sent over a socket to the local ctdbd.
If the send queue is full and the caller was issuing multiple async
dbwrap requests in a loop, the caller knows it's probably time to stop
sending requests for now and try again later.
This will be used in subsequent commits in
smbd_smb2_query_directory_send() when implementing async write time
updates. Directories may contain umpteen files so we send many requests
to ctdb without going through tevent and reading the responses which
has the potential to deadlock.
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=12628
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Wed Mar 15 20:04:32 CET 2017 on sn-devel-144
dbwrap_parse_record() can return ctdb tombstone records from the lctdb,
ignore them.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=12005
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Not freeing up(and reparenting to NULL context) ndr buffer
used for TDB updates resulting in huge memory leak when there
in high volume of opens and closes happening on same object.
Free the buffer before reparenting its parent to NULL context.
https://bugzilla.samba.org/show_bug.cgi?id=11934
Signed-off-by: Hemanth Thummala <hemanth.thummala@nutanix.com>
Signed-off-by: Saji VR <saji.vr@nutanix.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Fri May 27 18:43:31 CEST 2016 on sn-devel-144
Only in the default VFS. Gpfs, Ceph, Gluster and other modern
backend VFS filesystems might want to do the same.
Allow tuneable "smbd:force process locks = true" to turn
off OFD locks if in use and the kernel doesn't support them.
Display debug message showing admins what to do in this case.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Jeff Layton <jlayton@samba.org>
Not yet used. We will set this if we translate a process-associated
lock operation to a open file description lock operation.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Jeff Layton <jlayton@samba.org>
This means locks are associated with the SMB handle
they were created on, not the inode. In all other ways
they behave like UNIX extensions fcntl (process-associated)
locks. Torture test to follow.
When a handle is closed all locks attached to that handle
are closed, not all locks on the underlying inode. In
this respect they now behave like Windows locks.
The key to this in the UNIX extensions locking codepath is modifying
the reference count only when a new locking context is seen
on any lock request, and decrementing the reference count
when the last instance of a locking context is seen on any
unlock request. For SMB2+ the persistent part of a file handle
is used as the locking context so this behavior becomes
natural.
This is a behavior change but after consultation with
Jeff Layton and Steve French the only client that implements
UNIX extensions POSIX locks - the cifsfs client - already
expects these locks to behave like open file description
(ofd) locks. With our previous behavior Linux ofd-locks
fail against smbd.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Jeff Layton <jlayton@samba.org>
We will need this to implement open file description record locks.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Jeff Layton <jlayton@samba.org>
We will be using this to also ref count a posix lock applied
to a file handle when changing to open file description lock
semantics.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Jeff Layton <jlayton@samba.org>
This is in preperation of a more fine grained control of POSIX behaviour
in the SMB and VFS layers.
Inititally we use an uint8_t for the flags bitmap and add a define
posix_flags as posix_open in order to avoid breaking the VFS ABI.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=11065
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
mostly "Conditional jump or move depends on uninitialised value" &
"Use of uninitialised value of size 8" errors, suspect this is
related to compiler padding for the struct
Signed-off-by: Noel Power <noel.power@suse.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Nov 6 00:16:53 CET 2015 on sn-devel-104
Initialize lease pointer to point to the share_mode_data leases array
entry at index lease_idx.
This fixes a bug in smbstatus where the lease info is not printed.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=11549
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Oct 13 01:14:09 CEST 2015 on sn-devel-104
Add stream name argument to share_entry_forall machinery so smbstatus
can print the stream name of a file.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=11550
Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Replace all callers with direct calls to server_id_str_buf without
talloc_tos()
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We now have reset_delete_on_close_lck, this was called with "true"
everywhere now.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Boolean flags passed down make things more complex than necessary...
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Based on an idea by Volker to optimize cpu usage when
parsing struct share_mode_data entries.
Add a 64-bit sequence number to the share mode entry,
and after the entry is stored back in the db, cache
the in-memory version using talloc reparenting into the
memcache. The memcache key used is the same struct file_id
used as the key into the locking db.
On read, check if the locking db version sequence number
matches the in-memory cache, and if so reparent the
memcache version back onto the required memory context.
Saves all the ndr decoding when multiple accesses to the
same lock entry happen in succession.
Design also improved by Metze and Ira.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Ira Cooper <ira@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Apr 17 22:38:34 CEST 2015 on sn-devel-104
C does not allow empty initializer lists. Although gcc accepts that, the
SunOS compiler fails in this case with an error.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=11153
Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Christof Schmitt <cs@samba.org>
Autobuild-Date(master): Thu Mar 12 02:49:36 CET 2015 on sn-devel-104
Also check for allocation failures.
Signed-off-by: David Disseldorp <ddiss@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Mon Jan 12 19:22:31 CET 2015 on sn-devel-104
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Sun Jan 11 20:34:56 CET 2015 on sn-devel-104
If dbwrap_fetch_locked failed, we did a TALLOC_FREE(value). Fix this with a
talloc hierarchy.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
In a cluster this can be called with an empty record, while
brl_parse_data() relies on an initialized structure.
This is a regression in commit 837e29035c.
Bug: https://bugzilla.samba.org/show_bug.cgi?id=10911
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Pair-Programmed-With: Jeremy Allison <jra@samba.org>
Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Will enable us to solve the dynamic share path problem
with leases on [homes].
We're also able to give the correct error codes when a
lease key is re-used with a different file name.
Pair-Programmed-With: Jeremy Allison <jra@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
struct byte_range_lock *rw = NULL; will never change...
commit 1057240733 removed the
possible assigment of 'rw'.
So we can remove all code under if (rw != NULL) { ...
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Oct 31 06:07:43 CET 2014 on sn-devel-104
This is now only called during brl_forall. It does not really hurt if we list
dead processes here. If the upper layers really care, they can filter it out
themselves. The real lock conflicts are not removed on-demand.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This avoids the need to do sweeping validate_lock_entries calls
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Aug 6 03:27:36 CEST 2014 on sn-devel-104
Pass "struct lock_struct" as a parameter. This had to be destructured
before the call and re-constructed inside brl_locktest.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
It took me really long to grasp what's going on in this routine. I hope
its logic is easier to understand now
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
lck1 and lck2 are treated differently. They should carry more descriptive
names.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Consider:
lock = start=110,size=10
pend_lock = 100, size=10
Do not overlap. However,
(lock->start <= pend_lock->start + pend_lock->size)
110 100 10
is true, so it returns true (overlap).
lock->start <= pend_lock->start + pend_lock->size
should be:
lock->start < pend_lock->start + pend_lock->size
BUG: https://bugzilla.samba.org/show_bug.cgi?id=10685
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Wed Jul 2 10:18:17 CEST 2014 on sn-devel-104
memmove calculations are never nice, and this is going to be used in
validate_lock_entries soon :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We only print valid share mode entries, stale ones don't count. In
traverse, let the callback decide about staleness.
https://bugzilla.samba.org/show_bug.cgi?id=10680
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Fix several occurences of using printf conversion to fload when
printing offset and count variables in locking debug messages and
smbstatus.
Conversion to float may lead to wrong results with very large values.
Signed-off-by: Ralph Boehme <rb@sernet.de>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
In changes to come this will be possible for an INTERNAL_OPEN_ONLY.
The protection was already in place for some code paths, this
makes the coverage compete.
Bug 10564 - Lock order violation and file lost
https://bugzilla.samba.org/show_bug.cgi?id=10564
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri Mar 21 21:22:24 CET 2014 on sn-devel-104
This might be the reason for a few flaky builds.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Mon Mar 3 16:30:53 CET 2014 on sn-devel-104
brl_get_locks_readonly() currently returns NULL when it can't
find any byte range locks on the file. This is an error - it
should return a valid struct byte_range_lock containing num_locks == 0
so it can be cached.
Returning NULL when there are no locks causes POSIX lock
tests to fail returning NT_STATUS_NO_MEMORY (as it thinks
it can't allocate the struct) instead of NT_STATUS_OK.
Fixes bug:
Bug 10431 - STATUS_NO_MEMORY response from Query File Posix Lock request
https://bugzilla.samba.org/show_bug.cgi?id=10431
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <Volker.Lendecke@SerNet.DE>
This is in preparation to support handing flags to backends,
in particular activating read only record support for ctdb
databases. For a start, this does nothing but adding the
parameter, and all databases use DBWRAP_FLAG_NONE.
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
This parameter was originally set for removal in 2007 in 28210588ed
Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Thu Jan 30 13:25:22 CET 2014 on sn-devel-104
This does two things: It gets rid of a talloc_stackframe in a hot
code path and to me it makes the code easier to understand. It makes
the talloc hierarchy more obvious to follow.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Looks scary, but the only effect of this bug is too many UNLOCK messages
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Ira Cooper <ira@samba.org>
FAKE_LEVEL_II_OPLOCK was an indicator to break level2 oplock holders
on write. This information is now being held in brlock.tdb, which makes
the FAKE_LEVEL_II_OPLOCK type unnecessary.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This implements an idea by metze: Right now Samba does not grant level2
oplocks where it should: After an initial no-oplock open that has been
written to, we don't have the FAKE_LEVEL2_OPLOCK entry in locking.tdb
around anymore, this downgraded to NO_OPLOCK. Windows in this case will
grant level2 if being asked, we don't. Part of the reason for this
is that we don't have a proper mechanism to communicate the fact that
level2 needs to be broken to other smbds. Metze's insight was that we
have to look into brlock.tdb for every write anyway, so this might be
the right place to store this information.
My first reaction was that this is really hackish, but on further thought
this is not. oplocks depend on brlocks anyway, and we have the proper
mechanisms in place for brlocks.
The format for this change is to add one byte to the end of the brlock.tdb
record with value 1 if we have level2 oplocks around. Thus this patch
effectively reverts 8f41142 which I discovered while writing this
change. We now legally have unaligned records.
We can certainly talk about the format, but I'm not yet convinced we
need an idl for this yet. This is a potentially very hot code path,
and ndr marshalling has a cost.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
With the find_share_mode simplification we don't need fill_share_mode anymore.
So this coalesces add_share_mode as well.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
All callers used fill_share_mode_entry before calling
find_share_mode_entry. Remove that requirement.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
share_mode_stale_pid internally only has to deal with uint32_t. Make
the parameter match this.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
With the rewritten brl_get_lock_readonly we only set the destructor for
r/w lock records anyway.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Sun Oct 6 22:20:05 CEST 2013 on sn-devel-104
This is step 1 to get rid of brl_get_locks_internal with its complex readonly
business. It also optimizes 2 things: First, it uses dbwrap_parse_record to
avoid a talloc and memcpy, and second it uses talloc_pooled_object.
And -- hopefully it is easier to understand the caching logic with
fsp->brlock_rec and the clustering escape.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Without clustering, fsp->brlock_rec will never be set anyway. In the
clustering case we can't use the seqnum trick, so this is slow enough
that the additional if-statement does not matter in this case anyway. In
the non-clustered case it might. Have not measured it, but every little
bit helps I guess.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Stefan Metzmacher <metze@samba.org>
Autobuild-Date(master): Sun Oct 6 15:49:43 CEST 2013 on sn-devel-104
If someone messes with brlock.tdb and inserts an invalid record length,
this will lead to memcpy overwriting a few bytes behind malloc'ed data.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Sep 12 03:26:45 CEST 2013 on sn-devel-104
There's no point checking the validity of the "entry" argument more
than once
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
The comment for this routine said:
> Modifies d->num_share_modes, watch out in routines iterating over
> that array.
Well, it turns out that *every* caller of this API got it wrong. So I
think it's better to change the routine.
This leaves the array untouched while iterating but filters out the
deleted ones while saving them back to disk.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
For a given file, clean share mode entries for a given persistent file id.
Pair-Programmed-With: Michael Adam <obnox@samba.org>
Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Gregor Beck <gbeck@sernet.de>
Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
For a given file, clean up brl entries belonging to a given persistent file id.
Signed-off-by: Gregor Beck <gbeck@sernet.de>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
We should not remove locks of disconnected opens just like that.
When getting the byte range lock record for a newly connected file
handle, we still do the clean up, because in that situation,
disconnected entries are not valid any more.
Signed-off-by: Gregor Beck <gbeck@sernet.de>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
...instead of checking each server-id separately which can
be expensive in a cluster.
Signed-off-by: Gregor Beck <gbeck@sernet.de>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
fsp->fnum and lock->fnum are uint64_t already and we should not truncate the value here.
Currently this doesn't matter as we only use 16-bit.
But as 'int' is int32_t and we later compare fnum with lock->fnum == fnum,
the cast from int32_t to uint64_t goes via int64_t instead of uint32_t.
This means even if fsp->fnum just uses 32-bit of the uint64_t
we'll get the wrong result, as the implicit cast from a negative int32_t
value to uint64_t adds 0xFFFFFFFF00000000.
metze
