Age | Commit message (Collapse) | Author | Files | Lines |
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
When we're deciding what handlers to call in the select backend,
we didn't take into account the fact that the same fd may have
been added into the read FD_SET and the write FD_SET but with
different handlers.
We must match on both the file descriptor and the flags requested
before calling the handler.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
This is important to avoid races between threads if the poll_mt
backend is used.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
That has the same complexity but simplifies the code.
It also seems to fix a real problem, which was found
by the following test failure in Samba's autobuild,
while removing the epoll support from the "standard" backend.
[402/1555 in 19m8s] samba3.raw.composite(plugin_s4_dc)
smbtorture 4.1.0pre1-DEVELOPERBUILD
Using seed 1361531675
Testing savefile
Testing parallel fetchfile with 10 ops
waiting for completion
maximum runtime exceeded for smbtorture - terminating
UNEXPECTED(error): samba3.raw.composite.fetchfile (subunit.RemotedTestCase)(plugin_s4_dc)
REASON: _StringException: _StringException: was started but never finished!
UNEXPECTED(error): samba3.raw.composite.fetchfile(plugin_s4_dc) (subunit.RemotedTestCase)
REASON: was started but never finished!
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
We need to avoid passing pollfd.events == 0, to poll(),
as it will report POLLERR and POLLHUP events, but our caller
does not expect the event handler to be called.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
If we don't cleanup the array passed to poll after an
event was deleted, we may pass a bad file descriptor to poll().
This was found by the following test failure in Samba's
autobuild, while removing the epoll support from
the "standard" backend.
[48/1555 in 4m37s] samba3.smbtorture_s3.plain(s3dc).LOCK4
UNEXPECTED(failure): samba3.smbtorture_s3.plain(s3dc).LOCK4.smbtorture(s3dc)
REASON: _StringException: _StringException: using seed 1361530718
host=127.0.0.2 share=tmp user=metze myname=sn-devel-104
Running LOCK4
starting locktest4
Failed to create file: NT_STATUS_INVALID_HANDLE
finished locktest4
TEST LOCK4 FAILED!
LOCK4 took 190.492 secs
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
In a threaded environment it can happen that an tevent_fd is talloc_free'ed
while the main thread sleeps in the poll() syscall.
In such a case poll_event_fd_destructor() would set poll_ev->fdes[i] = NULL.
We then skip the removed event, but before we also skipped the one
that was located at the end of the array. We moved it to possition
'i', but the next loop uses 'i=i+1'.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
The define TEVENT_HAS_LOOP_ONCE_TRACE_POINTS can be used to
detect the new feature, without writing configure tests.
Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
When we disable python, it will cause build fail!
so, the patch will fix it.
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
|
|
This patch should address the problem with Link Local addresses
on FreeBSD and Linux.
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Richard Sharpe <rsharpe@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Feb 28 20:56:57 CET 2013 on sn-devel-104
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
|
|
Reviewed-by: Alexander Bokovoy <ab@samba.org>
|
|
Signed-off-by: Bjoern Jacke <bj@sernet.de>
Reviewed-by: Matthieu Patou <mat@samba.org>
Autobuild-User(master): Björn Jacke <bj@sernet.de>
Autobuild-Date(master): Fri Feb 22 15:38:21 CET 2013 on sn-devel-104
|
|
This reverts commit e6643fbf48afccd0acedb65fbe24d3ce84d44c40.
Signed-off-by: Bjoern Jacke <bj@sernet.de>
Reviewed-by: Matthieu Patou <mat@samba.org>
|
|
This brings in rep_timegm() on Solaris for example.
Andrew Bartlett
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
talloc is not a dependency of this library, but is required by includes.h.
By not including includes.h, we avoid needing to add an otherwise false talloc dep.
(this comes up if talloc.h is not installed as a system package).
Andrew Bartlett
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
Since we open with dbwrap, it auto-converts old tdbs (which it will
rename to secrets.tdb.bak once it's done).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Rusty Russell <rusty@rustcorp.com.au>
Autobuild-Date(master): Wed Feb 20 07:09:19 CET 2013 on sn-devel-104
|
|
For simplicity, we use this in the caller to choose between filenames.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
|
|
This is another belt-and-braces check in case someone decides to turn
on the fancy new .ntdb extension, and we haven't converted it to ntdb.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
|
|
This provides an extra safety check that everyone is using one or the
other: you can't create a tdb file if there's an ntdb file.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
|
|
This switches dbwrap_local_open() based on the extension of the
database name, so it handles both TDB and NTDB files.
Moreover, if asked to open a .ntdb, and there's no ntdb file but
there's a .tdb file, it converts that then moves it to .tdb.bak before
opening, and turn the .tdb file into a dangling symlink to make sure
it's never accidentally re-created or used:
$ ls -l secrets.tdb
lrwxrwxrwx 1 rusty rusty 23 Feb 11 11:31 secrets.tdb -> This is now in an NTDB
This provides transparent upgrade if people decide to use NTDB on a
database. Downgrade would be manual, eg:
ntdbdump foo.ntdb | tdbrestore foo.tdb && mv foo.ntdb foo.ntdb.bak
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
|
|
This is a dbwrap backend for ntdb. It's a fairly straight conversion from
the tdb version.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
|
|
ntdb's transaction code has an optimization which tdb's doesnt: it
only writes the parts of blocks whose contents have changed. This
means we can actually have a transaction which turns out to need no
recovery region.
This breaks the recovery setup logic, which sets the current recovery
size to 0 if there's no recovery area, and assumes that we'll always
create a new recovery area since the recovery will always need > 0
bytes.
In fact, if we really haven't changed anything, we can skip the
transaction commit altogether: since this happens at least once with
Samba, it's worth doing.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
|
|
realloc(NULL, ...) is equivalent to malloc. We are already using this
realloc property for tdb->lockrecs. It should not make any difference
in speed, it just makes for a little simpler code.
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): Tue Feb 19 17:30:13 CET 2013 on sn-devel-104
|
|
The transaction code uses tdb_alrecord_lock/upgrade, so it should also
use the tdb_allrecord_unlock function just for symmetry reasons
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>
|
|
For the mutex code we will have to lock the hashchain and the record
lock area independently. So we will have to call the loop twice. And,
it's a small refactoring for the better anyway I think.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
All arguments but the cmd are the same. To me this looks a bit better
and saves some bytes in the object code.
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>
|
|
lock-tracking.c mirrors the in-kernel fcntl structures to detect
unexpected use of fcntl calls. During my mutex work I changed our fcntl
use so that we unlock the allrecord_lock in two pieces: The range covering
the hash locks and the range covering the data area for the individual
traverse record locks. Splitting locks is not covered by lock-tracking.c.
This patch extends lock-tracking.c with this little piece. It's still
far from complete to track the full range of fcntl semantics. It is not
strictly needed right now, but it does not hurt either.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
If strftime() is not available, asctime() is used, and this usually
appends a newline character to the result. This is not desired for
timestamp().
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
Signed-off-by: Gregor Beck <gbeck@sernet.de>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
redundent -> redundant
Signed-off-by: Karolin Seeger <kseeger@samba.org>
|
|
Signed-off-by: Volker Lendecke <vl@samba.org>
Autobuild-User(master): Simo Sorce <idra@samba.org>
Autobuild-Date(master): Sat Feb 16 17:13:32 CET 2013 on sn-devel-104
|
|
This was a horrible hybrid of duplicated epoll and select()
code.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Feb 14 22:40:30 CET 2013 on sn-devel-104
|
|
Falls back cleanly from epoll -> poll, or uses poll if
epoll not available.
Signed-off-by: Jeremy Allison <jra@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
Not yet used, but will be called by the "standard"
fallback from epoll -> poll backends.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
|
|
Signed-off-by: Stefan Metzmacher <metze@samba.org>
|