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.
It's not needed, and may lead to unexpected side effects.
grant_fsp_oplock_type() is the only place to touch this.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Pair-Programmed-With: Stefan Metzmacher <metze@samba.org>
Pair-Programmed-With: Jeremy Allison <jra@samba.org>
Signed-off-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
As we atomically create using O_CREAT|O_EXCL,
then if new_file_created is true, then
file_existed *MUST* have been false (even
if the file was previously detected as being
there.
We use the variable file_existed again in logic
below this statement, so we must set file_existed = false,
if new_file_created returns are true from open_file().
Based on a fix from Michael Adam.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=10809
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Autobuild-User(master): Michael Adam <obnox@samba.org>
Autobuild-Date(master): Thu Sep 11 22:29:22 CEST 2014 on sn-devel-104
The race is when a file vanishes between
existence check and acl check.
In this case, open_file_ncreate() returns
OBJECT_NAME_NOT_FOUND even if the create
was called with disposition OPEN_IF.
But in this case, the file should be created.
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
Fix from Volker, really - just tidied up a little.
The S_ISFIFO check may not be strictly neccessary,
but doesn't hurt (might make the code a bit more complex
than it needs to be).
Fixes bug #10671 - Samba file corruption as a result of failed lock check.
https://bugzilla.samba.org/show_bug.cgi?id=10671
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
Bug 10593 is a panic that happens if we get an oplock break reply via
dbwrap_watch for which we can't find the SMB request anymore. This
error condition can legally happen when a client cancels the create
request before the oplock break response comes in. This patch drops the
dbwrap_watch_send request waiting for the oplock break when the request
is cancelled. Yet another talloc hierarchy problem, but if done right,
talloc hierarchies can make rundown of state easy :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
We exit if any of these if-statement fails, so a simple swap should not
make a difference.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
The main point is to get a talloc parent that will go away when the
request is cancelled
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
No longer used (hurrah!).
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>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Fri May 2 23:47:38 CEST 2014 on sn-devel-104
This causes deadlocks which cause smbd to crash if the locking
database has already been locked for a compound operation we
need to be atomic (as in the file rename case).
Ensure INTERNAL_OPEN_ONLY opens are synonymous with req==NULL.
INTERNAL_OPEN_ONLY opens leave a NO_OPLOCK record in
the share mode database, so they can be detected by other
processes for share mode violation purposes (because
they're doing an operation on the file that may include
reads or writes they need to have real state inside the
locking database) but have an fnum of FNUM_FIELD_INVALID
and a local share_file_id of zero, as they will never be
seen on the wire.
Ensure validate_my_share_entries() ignores
INTERNAL_OPEN_ONLY records (share_file_id == 0).
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>
Signed-off-by: Volker Lendecke <vl@samba.org>
Signed-off-by: Abhidnya Joshi <achirmul@in.ibm.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Christof Schmitt <cs@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Thu Jan 23 03:56:35 CET 2014 on sn-devel-104
If we set a non-null 'old timestamp' in the share mode database
when creating a directory handle, this prevents mtime (write time)
updates from being seen by clients, as we will always return the
timestamp stored in the database whilst the handle is open.
For files this is ok, as we update the stored timestamp
ourselves when we write to the handle. For directories
we should just rely on the mtime value from the underlying
filesystem.
Torture test to follow.
BUG: https://bugzilla.samba.org/show_bug.cgi?id=9870
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
This is a performance improvement for heavily contended files, in
particular in a cluster. The separate call to get_file_infos makes us
pull the locking.tdb record twice per open. For a contended file this
can be a performance penalty, this gets the # of record accesses for
the open/close cycle down from 3 to 2.
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 Nov 23 00:40:49 CET 2013 on sn-devel-104
Files and directories created with FILE_NO_COMPRESSION should not
inherit the compression attribute from their parent directory.
Signed-off-by: David Disseldorp <ddiss@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
https://bugzilla.samba.org/show_bug.cgi?id=10229
We need to check if the requested access mask
could be used to open the underlying file (if
it existed), as we're passing in zero for the
access mask to the base filename.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: David Disseldorp <ddiss@suse.de>
This reverts commit 7b70fa1873.
This is a change in behaviour which needs much further investigation
and testing.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
Autobuild-User(master): David Disseldorp <ddiss@samba.org>
Autobuild-Date(master): Fri Oct 25 14:22:20 CEST 2013 on sn-devel-104
This reverts commit e689b7d51e.
This is a change in behaviour which needs much further investigation
and testing.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
Later on we will have all the oplock/sharemode operations in one routine.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
This makes the is_stat_open special case in grant_fsp_oplock_type
redundant because in open_file_ntcreate further up we have already set
oplock_request to NO_OPLOCK for stat opens.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
This avoids the question where it could happen that something else but
fsp->oplock_type might be useful as an argument here.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
The level we have to break to depend on the breakers create_disposition:
If we overwrite, we have to break to none.
This patch overloads the "op_type" field in the break message we send
across to the smbd holding the oplock with the oplock level we want to
break to. Because it depends on the create_disposition in the breaking
open, only the breaker can make that decision. We might want to use
a different mechanism for this in the future, but for now using the
op_type field seems acceptable to me.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
There's no reason why we should not do this. This has turned into a pure
internal consistency check that should apply fine every time.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
While refactoring find_oplock_types to validate_oplock_types I forgot
that stat opens will end up in locking.tdb. So even with a batch oplock
around we can have more than one entry. This means the consistency check
in validate_oplock_types was wrong and too strict.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Only one call to delay_for_oplocks left. Metze showed me the new logic:
BATCH is broken if we have a sharing violation. Exclusive is broken
otherwise. That's it.
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): Wed Oct 16 02:51:53 CEST 2013 on sn-devel-104
No clue what this does. In open_directory, "dir_existed" is not used after
open_mode_check. In open_file_ntcreate it's used, but I can't think of a case
right now where we would find a formerly nonexisting file to exist suddenly.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
This makes grant_fsp_oplock_type independent from the values computed
in validate_oplock_types. It *might* make oplock calculation a bit
slower for heavily shared files, as we are walking the share mode array
twice. But we are doing so much stuff in open that I doubt the difference
is measurable. It clears up the code for me however, and I think that's
worth it.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
This removes two variables in open_file_ntcreate based on the observation
that for exclusive and batch oplocks there can only be one entry. So
in these cases we don't need to keep pointers from find_oplock_types to
delay_for_oplocks. We can just reference the only share mode entry around.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
smb_panic() does not take a printf style argument. This improves debug
output by easily printing the index that we fell over. Also, doing
smb_panic deep down is bad style IMHO.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
share_mode_data.num_share_modes is a uint32.
48 bytes less in .o text size for -O3 :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
Define a variable to dereference lck->data just once. Believe it or not,
this saves a few bytes .o with -O3 :-)
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@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>
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 13 14:35:26 CEST 2013 on sn-devel-104
This came from delete_on_close handling which was factored out.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
We need to check for DELETE_PENDING before the first oplock break
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 Oct 12 01:56:18 CEST 2013 on sn-devel-104
In open_file_ntcreate we do the del_share_mode on error. We should do
it here as well.
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 9 01:58:55 CEST 2013 on sn-devel-104
3.6 and earlier allowed open for execution when execute permissions are
not present on a file. This has been fixed in Samba 4.0.
This patch changes smbd to skip the execute bit from the ACL check
in the open code if "acl allow execute always = yes", hence
re-establishing the old behaviour in this case.
Signed-off-by: Michael Adam <obnox@samba.org>
Reviewed-by: Volker Lendecke <vl@samba.org>
Reviewed-by: David Disseldorp <ddiss@samba.org>
We don't have lck allocated yet at these points. Remove the TALLOC_FREE
calls that triggered me looking for the get_share_mode_lock calls.
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
This flag existed to break an exclusive or batch oplock in just one
instead of two steps down to "no oplock" when we did an allocation or file
size change. Running raw.oplock against W2k12 differs in this respect
from W2k3: W2k12 takes two steps (via level2) to break to none. This
removes the special flag that we only had for compatibility with systems
older than W2k12...
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 Sep 6 00:47:07 CEST 2013 on sn-devel-104
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 Apr 27 02:06:10 CEST 2013 on sn-devel-104
We do this if the idmap layer resolves Builtin_Administrators
as ID_TYPE_BOTH and if the current token has the
Builtin_Administrators SID or it's SYSTEM.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
If we can access the path to this file, by
default we have FILE_READ_ATTRIBUTES from the
containing directory. See the section.
"Algorithm to Check Access to an Existing File"
in MS-FSA.pdf.
Signed-off-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Michael Adam <obnox@samba.org>
according to coding guidelines
Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Jeremy Allison <jra@samba.org>
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Wed Oct 17 14:37:58 CEST 2012 on sn-devel-104
This makes it clear which context the returned SD is allocated on, as
a number of callers do not want it on talloc_tos().
As the ACL transformation allocates and then no longer needs a great
deal of memory, a talloc_stackframe() call is used to contain the
memory that is not returned further up the stack.
Andrew Bartlett
This reverts commit c251a6b044.
Remove this as we're planning to remove the security mask,
directory security mask parameters and only use create mask/directory mask.
Currently we call FSET_NT_ACL to inherit any ACLs on create. However
FSET_NT_ACL uses the security mask/directory security mask parameters
instead of the create mask/directory mask parameters.
Swap them temporarily when creating to ensure the correct masks
are applied.
Autobuild-User(master): Jeremy Allison <jra@samba.org>
Autobuild-Date(master): Tue Oct 2 22:27:17 CEST 2012 on sn-devel-104
With the prior code we assumed that we do not have kernel oplocks around
when we open a file because we handled samba-internal oplock breaks
before the open attempt.
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Sat Sep 29 19:29:36 CEST 2012 on sn-devel-104
This is a significant behaviour change: We do not open the file under
the share mode lock anymore. This might lead to more open calls in case
of oplock breaks or sharing violations, but those are slow error paths
and as such might be not too performance sensitive. The benefit of this
patch is a significant reduction of complexity of open_file_ntcreate()
With the new behaviour, we call fcb_or_dos_open after open_file(). It
is open_file() that sets up the fsp so that fcb_or_dos_open can find it
in the list of fsps. Avoid finding the fsp we are just setting up.
This is a 1:1 copy&paste of the oplock/sharemode code that we do before
an existing file is opened. It is a prerequiste for a patch that removes
all of that handling before we open the file.
The "else" is not necessary. In the if-branch we just returned.
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Wed Sep 26 18:08:03 CEST 2012 on sn-devel-104
The "access_bits" clause is redundant. is_stat_open says that exactly
at least one of the stat_open_bits must be set and none else.
Autobuild-User(master): Volker Lendecke <vl@samba.org>
Autobuild-Date(master): Tue Sep 25 20:05:06 CEST 2012 on sn-devel-104
