| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Called, from key_exists, scan_sorted_subkeys re-creates the sorted
subkeys record of the given key and then searches through it.
The race is that between creation and parsing of the sorted subkey
record, another process that stores some other subkey of the same
parent key will delete the sorted subkey record, resulting in an
WERR_BADFILE of an operation that should actually succeed.
This patch fixes the issue by wrapping the creation and parsing
into a transaction.
Michael
|
|
Michael
|
|
This made smbd crash in g_lock_lock() when trying to start a
transaction on a db with an already started transaction,
e.g. in a tcon_and_X where the share_info.tdb was not yet
initialized but share_info.tdb was already locked by another
process or writing acces to the winreg rpc pipe where the
registry tdb was already locked by another process.
What we really _want_ to do here by design is to react to
MSG_DBWRAP_G_LOCK_RETRY messages that are either sent
by a client doing g_lock_unlock or by ourselves when
we receive a CTDB_SRVID_SAMBA_NOTIFY or
CTDB_SRVID_RECONFIGURE message from ctdbd, i.e. when
either a client holding a lock or a complete node
has died.
Doing this properly involves calling tevent_loop_once(),
but doing this here with the main ctdbd messaging context
creates a nested event loop when g_lock_lock() is called
from the main event loop.
So as a quick fix, we act a little corasely here: we do
a select on the ctdb connection fd and when it is readable
or we get EINTR, then we retry without actually parsing
any ctdb packages or dispatching messages. This means that
we retry more often than necessary and intended by design,
but this does not harm and it is unobtrusive. When we have
finished, the main loop will pick up all the messages and
ctdb packets. The only extra twist is that we cannot use
timed events here but have to handcode a timeout for select.
Michael
|
|
Michael
|
|
Michael
|
|
The key for reading and writing was inconsistent due to a
off by one data length.
Michael
|
|
This skips update of the __db_sequence_number__ record when nothing else has
been written. There are transactions that are just openend and then nothing
is written until transaction_commit is called. This is for instance the case
with registry initialization routines: They start a transaction and only
write somthing when the registry has not been initialized yet.
So this change will skip many db_seqnum bumps and TRANS3_COMMIT roundtrips.
Michael
|
|
I carefully prepared the return value only to "return 0;" at the bottom. :-(
This may well have hit us for instance in the nested cancel case
and produced random errors.
Michael
|
|
The logic bug was that if a record was found in the marshall buffer,
then always the ctdb header of tha last record in the marshall buffer
was returned, and not the ctdb header of the last occurrence of the
requested record.
This is fixed by introducing an additional temporary variable.
Michael
|
|
Michael
|
|
Michael
|
|
Don't treat this as an error but return seqnum 0 instead.
Michael
|
|
|
|
Michael
|
|
For persistent databases, 64bit integer is kept in a special record
__db_sequence_number__. This record is incremented with each completed
transaction.
The retry mechanism for failing TRANS3_COMMIT controls inside the
db_ctdb_transaction_commit() function now relies one a modified
behaviour of ctdbd's treatment of persistent databases in recoveries.
Recently, a special treatment for persistent databases had been
introduced in ctdb (1.0.108) to work around the problems with the
orinal design of persistent transactions.
Now with the rewrite we need to revert to the old behaviour that
ctdb always takes the newest copies of all records.
This change also paves the way for a next step, which will make
recovery use the db seqnum to tell which node has the newest copy
of a persistent db and use that node's copy. This will greatly
reduce the amount of data transferred with each recovery.
Michael
|
|
The return values calculated by the callers were wrong anyways since
the new marshalling code does not set the local tdbs tdb error code.
Michael
|
|
Michael
|
|
This is the new implementation of ctdb transactions using the
global lock feature. It is needed by the current dbwrap_ctdb code.
Michael
|
|
|
|
This simplifies the transaction code a lot:
* transaction_start essentially consists of acquiring a global lock.
* No write operations at all are performed on the local database
until the transaction is committed: Every store operation is just
going into the marshall buffer.
* The commit operation calls a new simplified TRANS3_COMMIT control
in ctdb which rolls out thae changes to all nodes including the
node that is performing the transaction.
Michael
|
|
|
|
|
|
This is the basis to implement global locks in ctdb without depending on a
shared file system. The initial goal is to make ctdb persistent transactions
deterministic without too many timeouts.
|
|
|
|
Jeremy.
|
|
Volker.
Create widelinks_warning(int snum) to cover the message needed in make_connection.
Jeremy.
|
|
|
|
are incompatible.
Volker pointed out that the preexec scripts get passed the conn->connectpath
as a parameter, so call canonicalize_connect_path() both *before* and after
the preexec scripts. Ignore errors on the call before the preexec scripts,
as the path may not exist until created by the preexec scripts.
Jeremy.
|
|
This way we avoid any chance that a configuration reload may turn
back on wide links when unix extensions are enabled.
|
|
extensions" are incompatible.
Make sure we match the previous allow widelinks behavior, in that
non-root preexec scripts can create share directories for a share
definition.
Jeremy
|
|
incompatible.
Bug reported by Ralf Zimmermann <r.zimmermann@siegnetz.de>. Reproduced by jra.
If the target directory of a share doesn't exist, but is designed to
be created by a "root preexec" script call, then the widelinks check
is done too early - thus preventing the user from connecting to the
share.
Fix is to re-arrange the order of checks in make_connection_snum()
to always do the following order of operations:
(1). Turn off wide links if unix extensions = yes.
(2). Call any root preexec scripts.
(3). Canonicalize the share path to remove any symlinks (ie. end
up with the realpath in the connection_struct).
Jeremy.
|
|
This allows to set "spoolss:architecture = 'Windows x64'" for debugging purpose.
Guenther
|
|
Level 3 has been added with NT 4.0 and Windows 7 (at least 64bit version) makes
use of it in order to display queued jobs. Windows 7 will *not* fall back to
level 2 if we just return WERR_UNKNOWN_LEVEL, instead there will be no printjobs
displayed at all.
Guenther
|
|
Jeremy.
|
|
check !
Jeremy.
|
|
This changes the meaning of the ->prev pointer in our doubly linked
lists to point at the end of the list from the front of the list. That
allows us to implement DLIST_ADD_END() and related functions in O(1)
time, which can be a huge saving in many places in Samba.
This also means that the 'type' argument to various DLIST_*() macros
is no longer needed, but I have left it in for now to keep the
patchset small, which will make it easier to revert if any problems
are found. In the future we should remove the 'type' arguments.
(jra. Move the one use of DLIST_TAIL over to the new macros).
|
|
(cherry picked from commit 365b408c458c848a818637d9b36a0423aeb1ba54)
|
|
(cherry picked from commit 3437713ad7e5bccafde30553a8232119fd2a9eb9)
|
|
(cherry picked from commit a13b507f2d8be7f90c8872094cd0732926a6fcbb)
|
|
(cherry picked from commit 6c6df527e14514027cbcaa6deac25adf04363926)
|
|
manipulating p->prev directly is not safe any more
(cherry picked from commit 3c650ac1e3e1cdbbabecfddcd29325f20b5dcb48)
|
|
we don't need a separate lru pointer any more
(cherry picked from commit 4ffd7aca3e38728077bd80c2a65c4efbcfd216fc)
|
|
(cherry picked from commit a7d8bfd373392eecf4fff33d39b85e1b55ad901d)
|
|
(cherry picked from commit 4d23d777bc6d4fad20d0f3084fe658635812bee9)
|
|
uses of (list)->prev are moved over to DLIST_PREV. This will be replaced
when the final (new) version of the dlinklist.h header is added.
Jeremy.
|
|
This reverts commit 84fba3c1bc962804259f201d465acfdf0cd3c6a8.
Now we have a "processed packet queue" in nmbd we can go back
to doing this by default.
Jeremy.
|
|
Add a simple "processed packet queue" cache to stop nmbd responding to
packets received on the broadcast and non-broadcast socket (which
it has opened when "nmbd bind explicit broadcast = yes").
This is a very simple packet queue - it only keeps the packets
processed during a single call to listen_for_packets() (i.e. one
select call). This means that if the delivery notification for a
packet received on both broadcast and non-broadcast addresses
is done in two different select calls, the packet will still be
processed twice. This is a very rare occurrance and we can just
live with it when it does as the protocol is stateless. If this
is ever flagged as a repeatable problem then we can add a longer
lived cache, using timeout processing to clear etc. etc. But without
storing all packets processed we can never be *sure* we've eliminated
the race condition so I'm going to go with this simple solution until
someone proves a more complex one is needed :-).
Jeremy.
|
|
until the double processing problem in bug #7118 is fixed.
Jeremy.
|
|
|
|
values in subsequent SMBtrans replies)
There are two problems:
1). The server is off-by-one in the end of buffer space test.
2). The server returns 0 in the totaldata (smb_vwv1) and totalparams (smb_vwv0)
fields in the second and subsequent SMBtrans replies.
This patch fixes both.
Jeremy.
|
