+++ /dev/null
-Readme for Samba 4
-==================
-
-Samba 4 is the ambitious next version of the Samba suite that is being
-developed in parallel to the stable 3.0 series. The main emphasis in
-this branch is support for the Active Directory logon protocols used
-by Windows 2000 and above.
-
-While we welcome your interest in Samba 4, we don't want you to run your
-network with it quite yet. Please note the WARNINGS below, and the
-STATUS file, which aims to document what should and should not work.
-
-With 4 years of development under our belt since Tridge first proposed
-a new Virtual File System (VFS) layer for Samba3 (a project which
-eventually lead to our Active Directory efforts), we felt that we
-should create something we could 'show off' to our users. This is a
-Technology Preview (TP), aimed at allowing you, our users, managers and
-developers to see how we have progressed, and to invite your feedback and
-support.
-
-Warnings
---------
-
-Samba4 is currently at alpha stage. That is more a reference to
-Samba4's lack of the features we expect you will need than a statement
-of code quality, but clearly it hasn't seen a broad deployment yet. If
-you were to upgrade Samba3 (or indeed Windows) to Samba4, you would find
-many things work, but that other key features you may have relied on
-simply are not there yet.
-
-For example, while Samba 3.0 is an excellent member of a Active
-Directory domain, Samba4 is happier as a domain controller: (This is
-where we have done most of the research and development).
-
-While Samba4 is subjected to an awesome battery of tests on an
-automated basis, and we have found Samba4 to be very stable in it's
-behaviour, we have to recommend against upgrading production servers
-from Samba 3 to Samba 4 at this stage. If you are upgrading an
-experimental server, or looking to develop and test Samba, you should
-backup all configuration and data.
-
-As we research the needs of Active Directory integration more closely,
-we may need to change the format of the user database, in particular
-as we begin to understand how the attributes are generated and stored.
-At a worst case, we expect users will be able to extract the stored
-data as LDIF and hand munge it, but until we make an alpha release, we
-won't do this automatically. Indeed, many module changes are simply
-easier to cope with if you just re-provision after the upgrade.
-
-We value the security of your computers, and so we must warn you that
-Samba 4 Technology Preview includes basic Access Control List (ACL)
-protection on the main user database, but due to time constraints,
-none on the registry at this stage. We also do not currently have
-ACLs on the SWAT web-based management tool. This means that Samba 4
-Technology Preview is not secure, and should not be exposed to
-untrusted networks.
-
-Within the above proviso, file system access should occur as the
-logged in user, much as Samba3 does.
-
-As such, we must strongly recommend against using Samba4 in a
-production environment at this stage.
-
-New Features
-------------
-
-Samba4 supports the server-side of the Active Directory logon environment
-used by Windows 2000 and later, so we can do full domain join
-and domain logon operations with these clients.
-
-Our Domain Controller (DC) implementation includes our own built-in
-LDAP server and Kerberos Key Distribution Centre (KDC) as well as the
-Samba3-like logon services provided over CIFS. We correctly generate
-the infamous Kerberos PAC, and include it with the Kerberos tickets we
-issue.
-
-SWAT is now integrated into Samba 4 as the user-friendly interface to
-Samba server management. SWAT provides easy access to our
-setup and migration tools. Using SWAT, you can migrate windows
-domains in Samba 4, allowing easy setup of initial user databases, and
-upgrades from Samba 3.
-
-The new VFS features in Samba 4 adapts the file-system on the server to
-match the Windows client semantics, allowing Samba 4 to better match
-windows behaviour and application expectations. This includes file
-annotation information (in streams) and NT ACLs in particular. The
-VFS is backed with an extensive automated test suite.
-
-A new scripting interface has been added to Samba 4, allowing
-JavaScript programs to interface to Samba's internals.
-
-The Samba 4 architecture is based around an LDAP-like database that
-can use a range of modular backends. One of the backends supports
-standards compliant LDAP servers (including OpenLDAP), and we are
-working on modules to map between AD-like behaviours and this back-end.
-We are aiming for Samba 4 to be powerful front-end to large
-directories.
-
-Changes
--------
-
-Those familiar with Samba 3 can find a list of user-visible changes
-since that release series in the NEWS file.
-
-- An optional password is no longer supported as the second argument to
- smbclient.
-
-- The default location of smb.conf in non-FHS builds has changed from the
- PREFIX/lib directory to the PREFIX/etc directory.
-
-Known issues
-------------
-
-- Standalone server and domain member roles are not currently
- supported. While we have much of the infrastructure required, we
- have not collected these pieces together.
-
-- There is no printing support in the current release.
-
-- SWAT can be painful with <TAB> and forms. Just use the mouse, as
- the JavaScript layer doing this will change.
-
-Running Samba4
---------------
-
-A short guide to setting up Samba 4 can be found in the howto.txt file
-in root of the tarball.
-
-Development and feedback
-------------------------
-
-Bugs can be filed at https://bugzilla.samba.org/. Please
-look at the STATUS file before filing a bug to see if a particular
-is supposed to work yet.
-
-Development and general discussion about Samba 4 happens mainly on
-the #samba-technical IRC channel (on irc.freenode.net) and
-the samba-technical mailing list (see http://lists.samba.org/ for
-details).
-
-What's new in Samba 4 alpha3
+What's new in Samba 4 alpha4
============================
Samba 4 is the ambitious next version of the Samba suite that is being
production environments. Note the WARNINGS below, and the STATUS file,
which aims to document what should and should not work.
-Samba4 alpha3 follows on from our second alpha release (made in
-December), the first alpha release (made in September), and the
-Technology Preview series we have offered for some time now.
+Samba4 alpha4 follows on from the alpha release series we have been
+publishing since September last year.
WARNINGS
========
-Samba4 alpha3 is not a final Samba release. That is more a reference
+Samba4 alpha4 is not a final Samba release. That is more a reference
to Samba4's lack of the features we expect you will need than a
statement of code quality, but clearly it hasn't seen a broad
deployment yet. If you were to upgrade Samba3 (or indeed Windows) to
We are aiming for Samba 4 to be powerful frontend to large
directories.
-CHANGES SINCE Alpha2
+CHANGES SINCE Alpha3
=====================
In the time since Samba4 Alpha2 was released in December 2007, Samba has
continued to evolve, but you may particularly notice these areas:
- Python Bindings: Bindings for Python are now in place, and used for
- Samba's provision script, slowly displacing EJS as the embedded
- scripting language. With its increased use, Python is no longer
- optional, and configure will generate an error if it cannot locate
- an appropriate Python installation.
+ Python Bindings: Bindings for Python are now used for all internal
+ scripting, and the system python installation is used to run all
+ Samba python scripts (in place of smbpython found in the previous
+ alpha).
- SWAT Disabled: Due to a lack of developer time and without a
- long-term web developer to maintain it, the SWAT web UI has been
- disabled.
+ As such Python is no longer optional, and configure will generate an
+ error if it cannot locate an appropriate Python installation.
- Oplock support: Samba4's file server now supports oplocks
+ SWAT Remains Disabled: Due to a lack of developer time and without a
+ long-term web developer to maintain it, the SWAT web UI remains been
+ disabled (and would need to be rewritten in python in any case).
GNU Make: To try and simplfy our build system, we rely on GNU Make
to avoid autogenerating a massive single makefile.
- Account Expiry: Samba4 now better handles installations over 30 days
- old (thanks to our long-suffering testers for keeping installations
- around that long!)
+ Registry: Samba4's registry library has continued to improve.
- Registry: Samba4 registry interoperability has been improved in
- both the client utilities and in the registry service exposed by
- the Samba4 server itself.
+ ID mapping: Samba4 uses the internal ID mapping in winbind for all
+ but a few core users. Samba users should not appear in /etc/passwd,
+ as Samba will generate new user and group IDs regradless.
- Administrative Tools: Many enhancements have been made that allow
- better integration with Windows administrative tools, especially
- Active Directory Users and Computers.
+ NTP: Samba4 can act as a signing server for the ntp.org NTP deamon,
+ allowing NTPd to reply using Microsoft's non-standard signing
+ scheme. A patch to make NTPd talk to Samba for this purpose has
+ been submitted to the ntp.org project.
- ID mapping: Samba4 now handles ID mapping via winbind. The mappings
- are stored in a central ldb that could be shared across multiple
- machines using LDAP. Internal callers access this interface via a new
- wbclient library.
+ CLDAP: Users should experience less arbitary delays and more success with
+ group policy, domain joins and logons due to an improved
+ implementation of CLDAP and the 'netlogon' mailslot datagrams.
+
+ SMB2: The Samba4 SMB2 server and testsuite have been greatly
+ improved, but the SMB2 server remains off by default.
+
+ Secure DNS update: Configuration for GSS-TSIG updates of DNS records
+ is now generated by the provision script.
These are just some of the highlights of the work done in the past few
months. More details can be found in our GIT history.
- There is no printing support in the current release.
+- There is no netbios browsing support in the current release
+
- The Samba4 port of the CTDB clustering support is not yet complete
- Clock Synchronisation is critical. Many 'wrong password' errors are
actually due to Kerberos objecting to a clock skew between client
- and server.
+ and server. (The NTP work is partly to assist with this problem).
+
+- Samba4 alpha4 is currently only portable to recent Linux
+ distributions. Work to return support for other Unix varients is
+ expected during the next alpha cycle
+- Samba4 alpha4 is incompatible with GnuTLS 2.0, found in Fedora 9 and
+ recent Ubuntu releases. Please remove the
+ gnutls-devel/libgnutls-dev package before compiling (otherwise 'make
+ test' and LDAPS operations will hang).
RUNNING Samba4
==============
# e.g. SAMBA_VERSION_ALPHA_RELEASE=1 #
# -> "4.0.0alpha1" #
########################################################
-SAMBA_VERSION_ALPHA_RELEASE=4
+SAMBA_VERSION_ALPHA_RELEASE=5
########################################################
# For 'pre' releases the version will be #
*/
static NTSTATUS odb_ctdb_open_file(struct odb_lock *lck,
void *file_handle, const char *path,
- int *fd, bool allow_level_II_oplock,
+ int *fd, NTTIME open_write_time,
+ bool allow_level_II_oplock,
uint32_t oplock_level, uint32_t *oplock_granted)
{
return odb_push_record(lck, &file);
}
+static NTSTATUS odb_ctdb_set_write_time(struct odb_lock *lck,
+ NTTIME write_time, bool force)
+{
+ /*
+ * as this file will went away and isn't used yet,
+ * copy the implementation from the tdb backend
+ * --metze
+ */
+ return NT_STATUS_FOOBAR;
+}
+
/*
return the current value of the delete_on_close bit, and how many
people still have the file open
*/
-static NTSTATUS odb_ctdb_get_delete_on_close(struct odb_context *odb,
- DATA_BLOB *key, bool *del_on_close)
+static NTSTATUS odb_ctdb_get_file_infos(struct odb_context *odb, DATA_BLOB *key,
+ bool *del_on_close, NTTIME *write_time)
{
- NTSTATUS status;
- struct opendb_file file;
- struct odb_lock *lck;
-
- (*del_on_close) = false;
-
- lck = odb_lock(odb, odb, key);
- NT_STATUS_HAVE_NO_MEMORY(lck);
-
- status = odb_pull_record(lck, &file);
- if (NT_STATUS_EQUAL(NT_STATUS_OBJECT_NAME_NOT_FOUND, status)) {
- talloc_free(lck);
- return NT_STATUS_OK;
- }
- if (!NT_STATUS_IS_OK(status)) {
- talloc_free(lck);
- return status;
- }
-
- (*del_on_close) = file.delete_on_close;
-
- talloc_free(lck);
-
- return NT_STATUS_OK;
+ /*
+ * as this file will went away and isn't used yet,
+ * copy the implementation from the tdb backend
+ * --metze
+ */
+ return NT_STATUS_FOOBAR;
}
.odb_rename = odb_ctdb_rename,
.odb_get_path = odb_ctdb_get_path,
.odb_set_delete_on_close = odb_ctdb_set_delete_on_close,
- .odb_get_delete_on_close = odb_ctdb_get_delete_on_close,
+ .odb_set_write_time = odb_ctdb_set_write_time,
+ .odb_get_file_infos = odb_ctdb_get_file_infos,
.odb_can_open = odb_ctdb_can_open,
.odb_update_oplock = odb_ctdb_update_oplock,
.odb_break_oplocks = odb_ctdb_break_oplocks
goto failure;
}
- ret = _gsskrb5_encapsulate (minor_status, &outbuf, output_token,
- (u_char *)"\x01\x00", GSS_KRB5_MECHANISM);
- if (ret)
- goto failure;
+ if (flags & GSS_C_DCE_STYLE) {
+ output_token->value = outbuf.data;
+ output_token->length = outbuf.length;
+ } else {
+ ret = _gsskrb5_encapsulate (minor_status, &outbuf, output_token,
+ (u_char *)"\x01\x00", GSS_KRB5_MECHANISM);
+ if (ret)
+ goto failure;
+
+ krb5_data_free (&outbuf);
+ }
- krb5_data_free (&outbuf);
krb5_free_creds(context, kcred);
free_Checksum(&cksum);
if (cred == NULL)
from ldb import (SCOPE_SUBTREE, SCOPE_ONELEVEL, SCOPE_BASE, LdbError,
LDB_ERR_NO_SUCH_OBJECT, LDB_ERR_ATTRIBUTE_OR_VALUE_EXISTS,
LDB_ERR_ENTRY_ALREADY_EXISTS, LDB_ERR_UNWILLING_TO_PERFORM,
- LDB_ERR_NOT_ALLOWED_ON_NON_LEAF, LDB_ERR_OTHER)
+ LDB_ERR_NOT_ALLOWED_ON_NON_LEAF, LDB_ERR_OTHER, LDB_ERR_INVALID_DN_SYNTAX)
from samba import Ldb
from subunit import SubunitTestRunner
from samba import param
"userAccountControl": "4096",
"displayname": "ldap testy"})
+ self.delete_force(self.ldb, "cn=ldaptestcomputer3,cn=computers," + self.base_dn)
+ try:
+ ldb.add({"dn": "cn=ldaptestcomputer3,cn=computers," + self.base_dn,
+ "objectClass": "computer",
+ "cn": "LDAPtest2COMPUTER"
+ })
+ self.fail()
+ except LdbError, (num, _):
+ self.assertEquals(num, LDB_ERR_INVALID_DN_SYNTAX)
+
+ self.delete_force(self.ldb, "cn=ldaptestcomputer3,cn=computers," + self.base_dn)
+ try:
+ ldb.add({"dn": "cn=ldaptestcomputer3,cn=computers," + self.base_dn,
+ "objectClass": "computer",
+ "cn": "ldaptestcomputer3",
+ "sAMAccountType": "805306368"
+ })
+ self.fail()
+ except LdbError, (num, _):
+ self.assertEquals(num, LDB_ERR_UNWILLING_TO_PERFORM)
+
+ self.delete_force(self.ldb, "cn=ldaptestcomputer3,cn=computers," + self.base_dn)
+ try:
+ ldb.add({"dn": "cn=ldaptestcomputer3,cn=computers," + self.base_dn,
+ "objectClass": "computer",
+ "cn": "ldaptestcomputer3",
+ "userAccountControl": "0"
+ })
+ self.fail()
+ except LdbError, (num, _):
+ self.assertEquals(num, LDB_ERR_UNWILLING_TO_PERFORM)
+
+ self.delete_force(self.ldb, "cn=ldaptestuser7,cn=users," + self.base_dn)
+ try:
+ ldb.add({"dn": "cn=ldaptestuser7,cn=users," + self.base_dn,
+ "objectClass": "user",
+ "cn": "LDAPtestuser7",
+ "userAccountControl": "0"
+ })
+ self.fail()
+ except LdbError, (num, _):
+ self.assertEquals(num, LDB_ERR_UNWILLING_TO_PERFORM)
+
+ self.delete_force(self.ldb, "cn=ldaptestuser7,cn=users," + self.base_dn)
+
+ ldb.add({"dn": "cn=ldaptestuser7,cn=users," + self.base_dn,
+ "objectClass": "user",
+ "cn": "LDAPtestuser7",
+ "userAccountControl": "2"
+ })
+
+ self.delete_force(self.ldb, "cn=ldaptestuser7,cn=users," + self.base_dn)
+
+ self.delete_force(self.ldb, "cn=ldaptestcomputer3,cn=computers," + self.base_dn)
+ ldb.add({"dn": "cn=ldaptestcomputer3,cn=computers," + self.base_dn,
+ "objectClass": "computer",
+ "cn": "LDAPtestCOMPUTER3"
+ })
+
+ print "Testing ldb.search for (&(cn=ldaptestcomputer3)(objectClass=user))";
+ res = ldb.search(self.base_dn, expression="(&(cn=ldaptestcomputer3)(objectClass=user))");
+ self.assertEquals(len(res), 1, "Found only %d for (&(cn=ldaptestcomputer3)(objectClass=user))" % len(res))
+
+ self.assertEquals(str(res[0].dn), ("CN=ldaptestcomputer3,CN=Computers," + self.base_dn));
+ self.assertEquals(res[0]["cn"][0], "ldaptestcomputer3");
+ self.assertEquals(res[0]["name"][0], "ldaptestcomputer3");
+ self.assertEquals(res[0]["objectClass"][0], "top");
+ self.assertEquals(res[0]["objectClass"][1], "person");
+ self.assertEquals(res[0]["objectClass"][2], "organizationalPerson");
+ self.assertEquals(res[0]["objectClass"][3], "user");
+ self.assertEquals(res[0]["objectClass"][4], "computer");
+ self.assertTrue("objectGUID" in res[0])
+ self.assertTrue("whenCreated" in res[0])
+ self.assertEquals(res[0]["objectCategory"][0], ("CN=Computer,CN=Schema,CN=Configuration," + self.base_dn));
+ self.assertEquals(int(res[0]["primaryGroupID"][0]), 513);
+ self.assertEquals(int(res[0]["sAMAccountType"][0]), 805306368);
+ self.assertEquals(int(res[0]["userAccountControl"][0]), 546);
+
+ self.delete_force(self.ldb, "cn=ldaptestcomputer3,cn=computers," + self.base_dn)
+
print "Testing attribute or value exists behaviour"
try:
ldb.modify_ldif("""
servicePrincipalName: host/ldaptest2computer
servicePrincipalName: cifs/ldaptest2computer
""")
+ self.fail()
except LdbError, (num, msg):
self.assertEquals(num, LDB_ERR_ATTRIBUTE_OR_VALUE_EXISTS)
- ldb.modify_ldif("""
+ ldb.modify_ldif("""
dn: cn=ldaptest2computer,cn=computers,""" + self.base_dn + """
changetype: modify
replace: servicePrincipalName
servicePrincipalName: host/ldaptest2computer
servicePrincipalName: cifs/ldaptest2computer
""")
- try:
- ldb.modify_ldif("""
+ try:
+ ldb.modify_ldif("""
dn: cn=ldaptest2computer,cn=computers,""" + self.base_dn + """
changetype: modify
add: servicePrincipalName
servicePrincipalName: host/ldaptest2computer
""")
- except LdbError, (num, msg):
- self.assertEquals(num, LDB_ERR_ATTRIBUTE_OR_VALUE_EXISTS)
-
- print "Testing ranged results"
- ldb.modify_ldif("""
+ self.fail()
+ except LdbError, (num, msg):
+ self.assertEquals(num, LDB_ERR_ATTRIBUTE_OR_VALUE_EXISTS)
+
+ print "Testing ranged results"
+ ldb.modify_ldif("""
dn: cn=ldaptest2computer,cn=computers,""" + self.base_dn + """
changetype: modify
replace: servicePrincipalName
""")
- ldb.modify_ldif("""
+ ldb.modify_ldif("""
dn: cn=ldaptest2computer,cn=computers,""" + self.base_dn + """
changetype: modify
add: servicePrincipalName
servicePrincipalName: host/ldaptest2computer29
""")
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE,
- attrs=["servicePrincipalName;range=0-*"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- #print len(res[0]["servicePrincipalName;range=0-*"])
- self.assertEquals(len(res[0]["servicePrincipalName;range=0-*"]), 30)
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE,
+ attrs=["servicePrincipalName;range=0-*"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ #print len(res[0]["servicePrincipalName;range=0-*"])
+ self.assertEquals(len(res[0]["servicePrincipalName;range=0-*"]), 30)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=0-19"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=0-19"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
# print res[0]["servicePrincipalName;range=0-19"].length
- self.assertEquals(len(res[0]["servicePrincipalName;range=0-19"]), 20)
+ self.assertEquals(len(res[0]["servicePrincipalName;range=0-19"]), 20)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=0-30"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- self.assertEquals(len(res[0]["servicePrincipalName;range=0-*"]), 30)
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=0-30"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ self.assertEquals(len(res[0]["servicePrincipalName;range=0-*"]), 30)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=0-40"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- self.assertEquals(len(res[0]["servicePrincipalName;range=0-*"]), 30)
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=0-40"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ self.assertEquals(len(res[0]["servicePrincipalName;range=0-*"]), 30)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=30-40"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- self.assertEquals(len(res[0]["servicePrincipalName;range=30-*"]), 0)
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=30-40"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ self.assertEquals(len(res[0]["servicePrincipalName;range=30-*"]), 0)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=10-40"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- self.assertEquals(len(res[0]["servicePrincipalName;range=10-*"]), 20)
- # pos_11 = res[0]["servicePrincipalName;range=10-*"][18]
-
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=11-40"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- self.assertEquals(len(res[0]["servicePrincipalName;range=11-*"]), 19)
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=10-40"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ self.assertEquals(len(res[0]["servicePrincipalName;range=10-*"]), 20)
+ # pos_11 = res[0]["servicePrincipalName;range=10-*"][18]
+
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=11-40"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ self.assertEquals(len(res[0]["servicePrincipalName;range=11-*"]), 19)
# print res[0]["servicePrincipalName;range=11-*"][18]
# print pos_11
# self.assertEquals((res[0]["servicePrincipalName;range=11-*"][18]), pos_11)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=11-15"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
- self.assertEquals(len(res[0]["servicePrincipalName;range=11-15"]), 5)
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName;range=11-15"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ self.assertEquals(len(res[0]["servicePrincipalName;range=11-15"]), 5)
# self.assertEquals(res[0]["servicePrincipalName;range=11-15"][4], pos_11)
- res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName"])
- self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
+ res = ldb.search(self.base_dn, expression="(cn=ldaptest2computer))", scope=SCOPE_SUBTREE, attrs=["servicePrincipalName"])
+ self.assertEquals(len(res), 1, "Could not find (cn=ldaptest2computer)")
# print res[0]["servicePrincipalName"][18]
# print pos_11
- self.assertEquals(len(res[0]["servicePrincipalName"]), 30)
+ self.assertEquals(len(res[0]["servicePrincipalName"]), 30)
# self.assertEquals(res[0]["servicePrincipalName"][18], pos_11)
self.delete_force(self.ldb, "cn=ldaptestuser2,cn=users," + self.base_dn)
res = ldb.search(expression="(&(anr=not ldap user2)(objectClass=user))")
self.assertEquals(len(res), 0, "Must not find (&(anr=not ldap user2)(objectClass=user))")
+ # Testing ldb.search for (&(anr="testy ldap")(objectClass=user)) (ie, with quotes)
+ res = ldb.search(expression="(&(anr==\"testy ldap\")(objectClass=user))")
+ self.assertEquals(len(res), 0, "Found (&(anr==\"testy ldap\")(objectClass=user))")
+
print "Testing Group Modifies"
ldb.modify_ldif("""
dn: cn=ldaptestgroup,cn=users,""" + self.base_dn + """
self.assertEquals(res[0]["cn"], "ldaptestUSER3")
self.assertEquals(res[0]["name"], "ldaptestUSER3")
+ #"Testing ldb.search for (&(&(cn=ldaptestuser3)(userAccountControl=*))(objectClass=user))"
+ res = ldb.search(expression="(&(&(cn=ldaptestuser3)(userAccountControl=*))(objectClass=user))")
+ self.assertEquals(len(res), 1, "(&(&(cn=ldaptestuser3)(userAccountControl=*))(objectClass=user))")
+
+ self.assertEquals(str(res[0].dn), ("CN=ldaptestUSER3,CN=Users," + self.base_dn))
+ self.assertEquals(res[0]["cn"], "ldaptestUSER3")
+ self.assertEquals(res[0]["name"], "ldaptestUSER3")
+
+ #"Testing ldb.search for (&(&(cn=ldaptestuser3)(userAccountControl=546))(objectClass=user))"
+ res = ldb.search(expression="(&(&(cn=ldaptestuser3)(userAccountControl=546))(objectClass=user))")
+ self.assertEquals(len(res), 1, "(&(&(cn=ldaptestuser3)(userAccountControl=546))(objectClass=user))")
+
+ self.assertEquals(str(res[0].dn), ("CN=ldaptestUSER3,CN=Users," + self.base_dn))
+ self.assertEquals(res[0]["cn"], "ldaptestUSER3")
+ self.assertEquals(res[0]["name"], "ldaptestUSER3")
+
+ #"Testing ldb.search for (&(&(cn=ldaptestuser3)(userAccountControl=547))(objectClass=user))"
+ res = ldb.search(expression="(&(&(cn=ldaptestuser3)(userAccountControl=547))(objectClass=user))")
+ self.assertEquals(len(res), 0, "(&(&(cn=ldaptestuser3)(userAccountControl=547))(objectClass=user))")
+
# This is a Samba special, and does not exist in real AD
# print "Testing ldb.search for (dn=CN=ldaptestUSER3,CN=Users," + self.base_dn + ")"
# res = ldb.search("(dn=CN=ldaptestUSER3,CN=Users," + self.base_dn + ")")
self.assertTrue("whenCreated" in res[0])
self.assertEquals(res[0]["objectCategory"], ("CN=Person,CN=Schema,CN=Configuration," + self.base_dn))
self.assertEquals(int(res[0]["sAMAccountType"][0]), 805306368)
- # self.assertEquals(res[0].userAccountControl, 546)
+ self.assertEquals(int(res[0]["userAccountControl"][0]), 546)
self.assertEquals(res[0]["memberOf"][0], ("CN=ldaptestgroup2,CN=Users," + self.base_dn))
self.assertEquals(len(res[0]["memberOf"]), 1)
self.assertTrue("whenCreated" in res[0])
self.assertEquals(res[0]["objectCategory"], ("CN=Computer,CN=Schema,CN=Configuration," + self.base_dn))
self.assertEquals(int(res[0]["primaryGroupID"][0]), 513)
- # self.assertEquals(res[0].sAMAccountType, 805306368)
- # self.assertEquals(res[0].userAccountControl, 546)
+ self.assertEquals(int(res[0]["sAMAccountType"][0]), 805306368)
+ self.assertEquals(int(res[0]["userAccountControl"][0]), 546)
self.assertEquals(res[0]["memberOf"][0], "CN=ldaptestgroup2,CN=Users," + self.base_dn)
self.assertEquals(len(res[0]["memberOf"]), 1)
self.assertTrue("whenCreated" in res[0])
self.assertEquals(res[0]["objectCategory"][0], "CN=Computer,CN=Schema,CN=Configuration," + self.base_dn)
self.assertEquals(int(res[0]["sAMAccountType"][0]), 805306369)
- # self.assertEquals(res[0].userAccountControl, 4098)
+ self.assertEquals(int(res[0]["userAccountControl"][0]), 4096)
ldb.delete(res[0].dn)
typedef [public] struct {
boolean8 delete_on_close;
+ NTTIME open_write_time;
+ NTTIME changed_write_time;
utf8string path;
uint32 num_entries;
opendb_entry entries[num_entries];
*/
NTSTATUS odb_open_file(struct odb_lock *lck,
void *file_handle, const char *path,
- int *fd, bool allow_level_II_oplock,
+ int *fd, NTTIME open_write_time,
+ bool allow_level_II_oplock,
uint32_t oplock_level, uint32_t *oplock_granted)
{
return ops->odb_open_file(lck, file_handle, path,
- fd, allow_level_II_oplock,
+ fd, open_write_time,
+ allow_level_II_oplock,
oplock_level, oplock_granted);
}
}
/*
- return the current value of the delete_on_close bit, and how many
- people still have the file open
+ update the write time on an open file
*/
-NTSTATUS odb_get_delete_on_close(struct odb_context *odb,
- DATA_BLOB *key, bool *del_on_close)
+NTSTATUS odb_set_write_time(struct odb_lock *lck,
+ NTTIME write_time, bool force)
{
- return ops->odb_get_delete_on_close(odb, key, del_on_close);
+ return ops->odb_set_write_time(lck, write_time, force);
}
+/*
+ return the current value of the delete_on_close bit,
+ and the current write time.
+*/
+NTSTATUS odb_get_file_infos(struct odb_context *odb, DATA_BLOB *key,
+ bool *del_on_close, NTTIME *write_time)
+{
+ return ops->odb_get_file_infos(odb, key, del_on_close, write_time);
+}
/*
determine if a file can be opened with the given share_access,
DATA_BLOB (*odb_get_key)(TALLOC_CTX *mem_ctx, struct odb_lock *lck);
NTSTATUS (*odb_open_file)(struct odb_lock *lck,
void *file_handle, const char *path,
- int *fd, bool allow_level_II_oplock,
+ int *fd, NTTIME open_write_time,
+ bool allow_level_II_oplock,
uint32_t oplock_level, uint32_t *oplock_granted);
NTSTATUS (*odb_open_file_pending)(struct odb_lock *lck, void *private);
NTSTATUS (*odb_close_file)(struct odb_lock *lck, void *file_handle,
NTSTATUS (*odb_rename)(struct odb_lock *lck, const char *path);
NTSTATUS (*odb_get_path)(struct odb_lock *lck, const char **path);
NTSTATUS (*odb_set_delete_on_close)(struct odb_lock *lck, bool del_on_close);
- NTSTATUS (*odb_get_delete_on_close)(struct odb_context *odb,
- DATA_BLOB *key, bool *del_on_close);
+ NTSTATUS (*odb_set_write_time)(struct odb_lock *lck,
+ NTTIME write_time, bool force);
+ NTSTATUS (*odb_get_file_infos)(struct odb_context *odb, DATA_BLOB *key,
+ bool *del_on_close, NTTIME *write_time);
NTSTATUS (*odb_can_open)(struct odb_lock *lck,
uint32_t stream_id, uint32_t share_access,
uint32_t access_mask, bool delete_on_close,
*/
static NTSTATUS odb_tdb_open_file(struct odb_lock *lck,
void *file_handle, const char *path,
- int *fd, bool allow_level_II_oplock,
+ int *fd, NTTIME open_write_time,
+ bool allow_level_II_oplock,
uint32_t oplock_level, uint32_t *oplock_granted)
{
struct odb_context *odb = lck->odb;
NT_STATUS_HAVE_NO_MEMORY(lck->file.path);
}
+ if (lck->file.open_write_time == 0) {
+ lck->file.open_write_time = open_write_time;
+ }
+
/*
possibly grant an exclusive, batch or level2 oplock
*/
return odb_push_record(lck, &lck->file);
}
+/*
+ update the write time on an open file
+*/
+static NTSTATUS odb_tdb_set_write_time(struct odb_lock *lck,
+ NTTIME write_time, bool force)
+{
+ if (lck->file.path == NULL) {
+ return NT_STATUS_OBJECT_NAME_NOT_FOUND;
+ }
+
+ if (lck->file.changed_write_time != 0 && !force) {
+ return NT_STATUS_OK;
+ }
+
+ lck->file.changed_write_time = write_time;
+
+ return odb_push_record(lck, &lck->file);
+}
+
/*
return the current value of the delete_on_close bit, and how many
people still have the file open
*/
-static NTSTATUS odb_tdb_get_delete_on_close(struct odb_context *odb,
- DATA_BLOB *key, bool *del_on_close)
+static NTSTATUS odb_tdb_get_file_infos(struct odb_context *odb, DATA_BLOB *key,
+ bool *del_on_close, NTTIME *write_time)
{
struct odb_lock *lck;
- (*del_on_close) = false;
+ if (del_on_close) {
+ *del_on_close = false;
+ }
+ if (write_time) {
+ *write_time = 0;
+ }
lck = odb_lock(odb, odb, key);
NT_STATUS_HAVE_NO_MEMORY(lck);
- (*del_on_close) = lck->file.delete_on_close;
+ if (del_on_close) {
+ *del_on_close = lck->file.delete_on_close;
+ }
+ if (write_time) {
+ if (lck->file.changed_write_time == 0) {
+ *write_time = lck->file.open_write_time;
+ } else {
+ *write_time = lck->file.changed_write_time;
+ }
+ }
talloc_free(lck);
.odb_rename = odb_tdb_rename,
.odb_get_path = odb_tdb_get_path,
.odb_set_delete_on_close = odb_tdb_set_delete_on_close,
- .odb_get_delete_on_close = odb_tdb_get_delete_on_close,
+ .odb_set_write_time = odb_tdb_set_write_time,
+ .odb_get_file_infos = odb_tdb_get_file_infos,
.odb_can_open = odb_tdb_can_open,
.odb_update_oplock = odb_tdb_update_oplock,
.odb_break_oplocks = odb_tdb_break_oplocks
/*
fill in the dos file attributes for a file
*/
-NTSTATUS pvfs_fill_dos_info(struct pvfs_state *pvfs, struct pvfs_filename *name, int fd)
+NTSTATUS pvfs_fill_dos_info(struct pvfs_state *pvfs, struct pvfs_filename *name,
+ uint_t flags, int fd)
{
+ NTSTATUS status;
+ DATA_BLOB lkey;
+ NTTIME write_time;
+
/* make directories appear as size 0 with 1 link */
if (S_ISDIR(name->st.st_mode)) {
name->st.st_size = 0;
name->dos.file_id = (((uint64_t)name->st.st_dev)<<32) | name->st.st_ino;
name->dos.flags = 0;
- return pvfs_dosattrib_load(pvfs, name, fd);
+ status = pvfs_dosattrib_load(pvfs, name, fd);
+ NT_STATUS_NOT_OK_RETURN(status);
+
+ if (flags & PVFS_RESOLVE_NO_OPENDB) {
+ return NT_STATUS_OK;
+ }
+
+ status = pvfs_locking_key(name, name, &lkey);
+ NT_STATUS_NOT_OK_RETURN(status);
+
+ status = odb_get_file_infos(pvfs->odb_context, &lkey,
+ NULL, &write_time);
+ data_blob_free(&lkey);
+ if (!NT_STATUS_IS_OK(status)) {
+ DEBUG(1,("WARNING: odb_get_file_infos: %s\n", nt_errstr(status)));
+ return status;
+ }
+
+ if (!null_time(write_time)) {
+ name->dos.write_time = write_time;
+ }
+
+ return NT_STATUS_OK;
}
f->handle->position = 0;
f->handle->mode = 0;
f->handle->oplock = NULL;
+ ZERO_STRUCT(f->handle->write_time);
f->handle->open_completed = false;
if ((create_options & NTCREATEX_OPTIONS_DELETE_ON_CLOSE) &&
/* now really mark the file as open */
status = odb_open_file(lck, f->handle, name->full_name,
- NULL, false, OPLOCK_NONE, NULL);
+ NULL, name->dos.write_time,
+ false, OPLOCK_NONE, NULL);
if (!NT_STATUS_IS_OK(status)) {
talloc_free(lck);
}
status = odb_open_file(lck, f->handle, name->full_name,
- NULL, false, OPLOCK_NONE, NULL);
+ NULL, name->dos.write_time,
+ false, OPLOCK_NONE, NULL);
if (!NT_STATUS_IS_OK(status)) {
goto cleanup_delete;
*/
static int pvfs_handle_destructor(struct pvfs_file_handle *h)
{
+ talloc_free(h->write_time.update_event);
+ h->write_time.update_event = NULL;
+
if ((h->create_options & NTCREATEX_OPTIONS_DELETE_ON_CLOSE) &&
h->name->stream_name) {
NTSTATUS status;
h->fd = -1;
}
+ if (!h->write_time.update_forced &&
+ h->write_time.update_on_close &&
+ h->write_time.close_time == 0) {
+ struct timeval tv;
+ tv = timeval_current();
+ h->write_time.close_time = timeval_to_nttime(&tv);
+ }
+
if (h->have_opendb_entry) {
struct odb_lock *lck;
NTSTATUS status;
return 0;
}
+ if (h->write_time.update_forced) {
+ status = odb_get_file_infos(h->pvfs->odb_context,
+ &h->odb_locking_key,
+ NULL,
+ &h->write_time.close_time);
+ if (!NT_STATUS_IS_OK(status)) {
+ DEBUG(0,("Unable get write time for '%s' - %s\n",
+ h->name->full_name, nt_errstr(status)));
+ }
+
+ h->write_time.update_forced = false;
+ h->write_time.update_on_close = true;
+ } else if (h->write_time.update_on_close) {
+ status = odb_set_write_time(lck, h->write_time.close_time, true);
+ if (!NT_STATUS_IS_OK(status)) {
+ DEBUG(0,("Unable set write time for '%s' - %s\n",
+ h->name->full_name, nt_errstr(status)));
+ }
+ }
+
status = odb_close_file(lck, h, &delete_path);
if (!NT_STATUS_IS_OK(status)) {
DEBUG(0,("Unable to remove opendb entry for '%s' - %s\n",
FILE_NOTIFY_CHANGE_FILE_NAME,
delete_path);
}
+ h->write_time.update_on_close = false;
}
talloc_free(lck);
}
+ if (h->write_time.update_on_close) {
+ struct timeval tv[2];
+
+ nttime_to_timeval(&tv[0], h->name->dos.access_time);
+ nttime_to_timeval(&tv[1], h->write_time.close_time);
+
+ if (!timeval_is_zero(&tv[0]) || !timeval_is_zero(&tv[1])) {
+ if (utimes(h->name->full_name, tv) == -1) {
+ DEBUG(0,("pvfs_handle_destructor: utimes() failed '%s' - %s\n",
+ h->name->full_name, strerror(errno)));
+ }
+ }
+ }
+
return 0;
}
DATA_BLOB locking_key;
status = pvfs_locking_key(parent, req, &locking_key);
NT_STATUS_NOT_OK_RETURN(status);
- status = odb_get_delete_on_close(pvfs->odb_context, &locking_key,
- &del_on_close);
+ status = odb_get_file_infos(pvfs->odb_context, &locking_key,
+ &del_on_close, NULL);
NT_STATUS_NOT_OK_RETURN(status);
if (del_on_close) {
return NT_STATUS_DELETE_PENDING;
}
/* re-resolve the open fd */
- status = pvfs_resolve_name_fd(pvfs, fd, name);
+ status = pvfs_resolve_name_fd(pvfs, fd, name, 0);
if (!NT_STATUS_IS_OK(status)) {
close(fd);
return status;
f->handle->mode = 0;
f->handle->oplock = NULL;
f->handle->have_opendb_entry = true;
+ ZERO_STRUCT(f->handle->write_time);
f->handle->open_completed = false;
status = odb_open_file(lck, f->handle, name->full_name,
- &f->handle->fd, allow_level_II_oplock,
+ &f->handle->fd, name->dos.write_time,
+ allow_level_II_oplock,
oplock_level, &oplock_granted);
talloc_free(lck);
if (!NT_STATUS_IS_OK(status)) {
f->handle->mode = 0;
f->handle->oplock = NULL;
f->handle->have_opendb_entry = false;
+ ZERO_STRUCT(f->handle->write_time);
f->handle->open_completed = false;
/* form the lock context used for byte range locking and
/* now really mark the file as open */
status = odb_open_file(lck, f->handle, name->full_name,
- &f->handle->fd, allow_level_II_oplock,
+ &f->handle->fd, name->dos.write_time,
+ allow_level_II_oplock,
oplock_level, &oplock_granted);
if (!NT_STATUS_IS_OK(status)) {
}
/* re-resolve the open fd */
- status = pvfs_resolve_name_fd(f->pvfs, fd, f->handle->name);
+ status = pvfs_resolve_name_fd(f->pvfs, fd, f->handle->name, PVFS_RESOLVE_NO_OPENDB);
if (!NT_STATUS_IS_OK(status)) {
talloc_free(lck);
return status;
{
struct pvfs_state *pvfs = ntvfs->private_data;
struct pvfs_file *f;
- struct utimbuf unix_times;
if (io->generic.level == RAW_CLOSE_SPLCLOSE) {
return NT_STATUS_DOS(ERRSRV, ERRerror);
}
if (!null_time(io->generic.in.write_time)) {
- unix_times.actime = 0;
- unix_times.modtime = io->close.in.write_time;
- utime(f->handle->name->full_name, &unix_times);
+ f->handle->write_time.update_forced = false;
+ f->handle->write_time.update_on_close = true;
+ unix_to_nt_time(&f->handle->write_time.close_time, io->generic.in.write_time);
}
if (io->generic.in.flags & SMB2_CLOSE_FLAGS_FULL_INFORMATION) {
NTSTATUS status;
bool del_on_close;
- status = odb_get_delete_on_close(pvfs->odb_context, &h->odb_locking_key,
- &del_on_close);
+ status = odb_get_file_infos(pvfs->odb_context, &h->odb_locking_key,
+ &del_on_close, NULL);
if (!NT_STATUS_IS_OK(status)) {
DEBUG(1,("WARNING: unable to determine delete on close status for open file\n"));
return false;
/* get a pvfs_filename source object */
status = pvfs_resolve_partial(pvfs, mem_ctx,
- dir_path, fname1, &name1);
+ dir_path, fname1,
+ PVFS_RESOLVE_NO_OPENDB,
+ &name1);
if (!NT_STATUS_IS_OK(status)) {
goto failed;
}
/* get a pvfs_filename dest object */
status = pvfs_resolve_partial(pvfs, mem_ctx,
- dir_path, fname2, &name2);
+ dir_path, fname2,
+ PVFS_RESOLVE_NO_OPENDB,
+ &name2);
if (NT_STATUS_IS_OK(status)) {
status = pvfs_can_delete(pvfs, req, name2, NULL);
if (!NT_STATUS_IS_OK(status)) {
TODO: add a cache for previously resolved case-insensitive names
TODO: add mangled name support
*/
-static NTSTATUS pvfs_case_search(struct pvfs_state *pvfs, struct pvfs_filename *name)
+static NTSTATUS pvfs_case_search(struct pvfs_state *pvfs,
+ struct pvfs_filename *name,
+ uint_t flags)
{
/* break into a series of components */
int num_components;
name->full_name = partial_name;
if (name->exists) {
- return pvfs_fill_dos_info(pvfs, name, -1);
+ return pvfs_fill_dos_info(pvfs, name, flags, -1);
}
return NT_STATUS_OK;
/* we need to search for a matching name */
saved_name = (*name)->full_name;
(*name)->full_name = dir_name;
- status = pvfs_case_search(pvfs, *name);
+ status = pvfs_case_search(pvfs, *name, flags);
if (!NT_STATUS_IS_OK(status)) {
/* the directory doesn't exist */
(*name)->full_name = saved_name;
/* if we can stat() the full name now then we are done */
if (stat((*name)->full_name, &(*name)->st) == 0) {
(*name)->exists = true;
- return pvfs_fill_dos_info(pvfs, *name, -1);
+ return pvfs_fill_dos_info(pvfs, *name, flags, -1);
}
/* search for a matching filename */
- status = pvfs_case_search(pvfs, *name);
+ status = pvfs_case_search(pvfs, *name, flags);
return status;
}
*/
NTSTATUS pvfs_resolve_partial(struct pvfs_state *pvfs, TALLOC_CTX *mem_ctx,
const char *unix_dir, const char *fname,
- struct pvfs_filename **name)
+ uint_t flags, struct pvfs_filename **name)
{
NTSTATUS status;
(*name)->stream_name = NULL;
(*name)->stream_id = 0;
- status = pvfs_fill_dos_info(pvfs, *name, -1);
+ status = pvfs_fill_dos_info(pvfs, *name, flags, -1);
return status;
}
to update the pvfs_filename stat information, and by pvfs_open()
*/
NTSTATUS pvfs_resolve_name_fd(struct pvfs_state *pvfs, int fd,
- struct pvfs_filename *name)
+ struct pvfs_filename *name, uint_t flags)
{
dev_t device = (dev_t)0;
ino_t inode = 0;
name->exists = true;
- return pvfs_fill_dos_info(pvfs, name, fd);
+ return pvfs_fill_dos_info(pvfs, name, flags, fd);
}
/*
talloc_free(lck);
}
- status = pvfs_resolve_name_fd(pvfs, h->fd, h->name);
+ /*
+ * TODO: pass PVFS_RESOLVE_NO_OPENDB and get
+ * the write time from odb_lock() above.
+ */
+ status = pvfs_resolve_name_fd(pvfs, h->fd, h->name, 0);
NT_STATUS_NOT_OK_RETURN(status);
+ if (!null_nttime(h->write_time.close_time)) {
+ h->name->dos.write_time = h->write_time.close_time;
+ }
+
return NT_STATUS_OK;
}
(*name)->stream_name = NULL;
(*name)->stream_id = 0;
- status = pvfs_fill_dos_info(pvfs, *name, -1);
+ status = pvfs_fill_dos_info(pvfs, *name, PVFS_RESOLVE_NO_OPENDB, -1);
return status;
}
in pvfs_list_seek_ofs() for
how we cope with this */
- status = pvfs_resolve_partial(pvfs, file, unix_path, fname, &name);
+ status = pvfs_resolve_partial(pvfs, file, unix_path, fname, 0, &name);
if (!NT_STATUS_IS_OK(status)) {
return status;
}
break;
case SEEK_MODE_END:
- status = pvfs_resolve_name_fd(pvfs, h->fd, h->name);
+ status = pvfs_resolve_name_fd(pvfs, h->fd, h->name, PVFS_RESOLVE_NO_OPENDB);
h->seek_offset = h->name->st.st_size + io->lseek.in.offset;
break;
}
union smb_setfileinfo *info)
{
struct pvfs_state *pvfs = ntvfs->private_data;
- struct utimbuf unix_times;
struct pvfs_file *f;
struct pvfs_file_handle *h;
struct pvfs_filename newstats;
}
/* possibly change the file timestamps */
- ZERO_STRUCT(unix_times);
if (newstats.dos.create_time != h->name->dos.create_time) {
change_mask |= FILE_NOTIFY_CHANGE_CREATION;
}
if (newstats.dos.access_time != h->name->dos.access_time) {
- unix_times.actime = nt_time_to_unix(newstats.dos.access_time);
change_mask |= FILE_NOTIFY_CHANGE_LAST_ACCESS;
}
if (newstats.dos.write_time != h->name->dos.write_time) {
- unix_times.modtime = nt_time_to_unix(newstats.dos.write_time);
change_mask |= FILE_NOTIFY_CHANGE_LAST_WRITE;
}
- if (unix_times.actime != 0 || unix_times.modtime != 0) {
- if (utime(h->name->full_name, &unix_times) == -1) {
- return pvfs_map_errno(pvfs, errno);
+ if ((change_mask & FILE_NOTIFY_CHANGE_LAST_ACCESS) ||
+ (change_mask & FILE_NOTIFY_CHANGE_LAST_WRITE)) {
+ struct timeval tv[2];
+
+ nttime_to_timeval(&tv[0], newstats.dos.access_time);
+ nttime_to_timeval(&tv[1], newstats.dos.write_time);
+
+ if (!timeval_is_zero(&tv[0]) || !timeval_is_zero(&tv[1])) {
+ if (utimes(h->name->full_name, tv) == -1) {
+ DEBUG(0,("pvfs_setfileinfo: utimes() failed '%s' - %s\n",
+ h->name->full_name, strerror(errno)));
+ return pvfs_map_errno(pvfs, errno);
+ }
}
}
+ if (change_mask & FILE_NOTIFY_CHANGE_LAST_WRITE) {
+ struct odb_lock *lck;
+
+ lck = odb_lock(req, h->pvfs->odb_context, &h->odb_locking_key);
+ if (lck == NULL) {
+ DEBUG(0,("Unable to lock opendb for write time update\n"));
+ return NT_STATUS_INTERNAL_ERROR;
+ }
+
+ status = odb_set_write_time(lck, newstats.dos.write_time, true);
+ if (!NT_STATUS_IS_OK(status)) {
+ DEBUG(0,("Unable to update write time: %s\n",
+ nt_errstr(status)));
+ talloc_free(lck);
+ return status;
+ }
+
+ talloc_free(lck);
+
+ h->write_time.update_forced = true;
+ h->write_time.update_on_close = false;
+ talloc_free(h->write_time.update_event);
+ h->write_time.update_event = NULL;
+ }
/* possibly change the attribute */
if (newstats.dos.attrib != h->name->dos.attrib) {
struct pvfs_filename *name;
struct pvfs_filename newstats;
NTSTATUS status;
- struct utimbuf unix_times;
uint32_t access_needed;
uint32_t change_mask = 0;
struct odb_lock *lck = NULL;
}
/* possibly change the file timestamps */
- ZERO_STRUCT(unix_times);
if (newstats.dos.create_time != name->dos.create_time) {
change_mask |= FILE_NOTIFY_CHANGE_CREATION;
}
if (newstats.dos.access_time != name->dos.access_time) {
- unix_times.actime = nt_time_to_unix(newstats.dos.access_time);
change_mask |= FILE_NOTIFY_CHANGE_LAST_ACCESS;
}
if (newstats.dos.write_time != name->dos.write_time) {
- unix_times.modtime = nt_time_to_unix(newstats.dos.write_time);
change_mask |= FILE_NOTIFY_CHANGE_LAST_WRITE;
}
- if (unix_times.actime != 0 || unix_times.modtime != 0) {
- if (utime(name->full_name, &unix_times) == -1) {
- return pvfs_map_errno(pvfs, errno);
+ if ((change_mask & FILE_NOTIFY_CHANGE_LAST_ACCESS) ||
+ (change_mask & FILE_NOTIFY_CHANGE_LAST_WRITE)) {
+ struct timeval tv[2];
+
+ nttime_to_timeval(&tv[0], newstats.dos.access_time);
+ nttime_to_timeval(&tv[1], newstats.dos.write_time);
+
+ if (!timeval_is_zero(&tv[0]) || !timeval_is_zero(&tv[1])) {
+ if (utimes(name->full_name, tv) == -1) {
+ DEBUG(0,("pvfs_setpathinfo: utimes() failed '%s' - %s\n",
+ name->full_name, strerror(errno)));
+ return pvfs_map_errno(pvfs, errno);
+ }
+ }
+ }
+ if (change_mask & FILE_NOTIFY_CHANGE_LAST_WRITE) {
+ if (lck == NULL) {
+ DATA_BLOB lkey;
+ status = pvfs_locking_key(name, name, &lkey);
+ NT_STATUS_NOT_OK_RETURN(status);
+
+ lck = odb_lock(req, pvfs->odb_context, &lkey);
+ data_blob_free(&lkey);
+ if (lck == NULL) {
+ DEBUG(0,("Unable to lock opendb for write time update\n"));
+ return NT_STATUS_INTERNAL_ERROR;
+ }
+ }
+
+ status = odb_set_write_time(lck, newstats.dos.write_time, true);
+ if (NT_STATUS_EQUAL(status, NT_STATUS_OBJECT_NAME_NOT_FOUND)) {
+ /* it could be that nobody has opened the file */
+ } else if (!NT_STATUS_IS_OK(status)) {
+ DEBUG(0,("Unable to update write time: %s\n",
+ nt_errstr(status)));
+ return status;
}
}
/* resolve the cifs name to a posix name */
status = pvfs_resolve_name(pvfs, req, unl->unlink.in.pattern,
- PVFS_RESOLVE_WILDCARD | PVFS_RESOLVE_STREAMS, &name);
+ PVFS_RESOLVE_WILDCARD |
+ PVFS_RESOLVE_STREAMS |
+ PVFS_RESOLVE_NO_OPENDB,
+ &name);
if (!NT_STATUS_IS_OK(status)) {
return status;
}
/* get a pvfs_filename object */
status = pvfs_resolve_partial(pvfs, req,
pvfs_list_unix_path(dir),
- fname, &name);
+ fname,
+ PVFS_RESOLVE_NO_OPENDB,
+ &name);
if (!NT_STATUS_IS_OK(status)) {
return status;
}
#include "includes.h"
#include "vfs_posix.h"
#include "librpc/gen_ndr/security.h"
+#include "lib/events/events.h"
+static void pvfs_write_time_update_handler(struct event_context *ev,
+ struct timed_event *te,
+ struct timeval tv,
+ void *private_data)
+{
+ struct pvfs_file_handle *h = talloc_get_type(private_data,
+ struct pvfs_file_handle);
+ struct odb_lock *lck;
+ NTSTATUS status;
+ NTTIME write_time;
+
+ lck = odb_lock(h, h->pvfs->odb_context, &h->odb_locking_key);
+ if (lck == NULL) {
+ DEBUG(0,("Unable to lock opendb for write time update\n"));
+ return;
+ }
+
+ write_time = timeval_to_nttime(&tv);
+
+ status = odb_set_write_time(lck, write_time, false);
+ if (!NT_STATUS_IS_OK(status)) {
+ DEBUG(0,("Unable to update write time: %s\n",
+ nt_errstr(status)));
+ return;
+ }
+
+ talloc_free(lck);
+
+ h->write_time.update_event = NULL;
+}
+
+static void pvfs_trigger_write_time_update(struct pvfs_file_handle *h)
+{
+ struct pvfs_state *pvfs = h->pvfs;
+ struct timeval tv;
+
+ if (h->write_time.update_triggered) {
+ return;
+ }
+
+ tv = timeval_current_ofs(0, pvfs->writetime_delay);
+
+ h->write_time.update_triggered = true;
+ h->write_time.update_on_close = true;
+ h->write_time.update_event = event_add_timed(pvfs->ntvfs->ctx->event_ctx,
+ h, tv,
+ pvfs_write_time_update_handler,
+ h);
+ if (!h->write_time.update_event) {
+ DEBUG(0,("Failed event_add_timed\n"));
+ }
+}
/*
write to a file
status = pvfs_break_level2_oplocks(f);
NT_STATUS_NOT_OK_RETURN(status);
+ pvfs_trigger_write_time_update(f->handle);
+
if (f->handle->name->stream_name) {
ret = pvfs_stream_write(pvfs,
f->handle,
PVFS_OPLOCK_TIMEOUT,
PVFS_OPLOCK_TIMEOUT_DEFAULT);
+ pvfs->writetime_delay = share_int_option(scfg,
+ PVFS_WRITETIME_DELAY,
+ PVFS_WRITETIME_DELAY_DEFAULT);
+
pvfs->share_name = talloc_strdup(pvfs, scfg->name);
pvfs->fs_attribs =
/* the oplock break timeout (secs) */
uint_t oplock_break_timeout;
+ /* the write time update delay (nsecs) */
+ uint_t writetime_delay;
+
/* filesystem attributes (see FS_ATTR_*) */
uint32_t fs_attribs;
/* we need this hook back to our parent for lock destruction */
struct pvfs_state *pvfs;
+ struct {
+ bool update_triggered;
+ struct timed_event *update_event;
+ bool update_on_close;
+ NTTIME close_time;
+ bool update_forced;
+ } write_time;
+
/* the open went through to completion */
bool open_completed;
};
/* flags to pvfs_resolve_name() */
#define PVFS_RESOLVE_WILDCARD (1<<0)
#define PVFS_RESOLVE_STREAMS (1<<1)
+#define PVFS_RESOLVE_NO_OPENDB (1<<2)
/* flags in pvfs->flags */
#define PVFS_FLAG_CI_FILESYSTEM (1<<0) /* the filesystem is case insensitive */
#define PVFS_FAKE_OPLOCKS "posix:fakeoplocks"
#define PVFS_SHARE_DELAY "posix:sharedelay"
#define PVFS_OPLOCK_TIMEOUT "posix:oplocktimeout"
+#define PVFS_WRITETIME_DELAY "posix:writetimeupdatedelay"
#define PVFS_ALLOCATION_ROUNDING "posix:allocationrounding"
#define PVFS_SEARCH_INACTIVITY "posix:searchinactivity"
#define PVFS_ACL "posix:acl"
#define PVFS_FAKE_OPLOCKS_DEFAULT false
#define PVFS_SHARE_DELAY_DEFAULT 1000000 /* nsecs */
#define PVFS_OPLOCK_TIMEOUT_DEFAULT 30 /* secs */
+#define PVFS_WRITETIME_DELAY_DEFAULT 2000000 /* nsecs */
#define PVFS_ALLOCATION_ROUNDING_DEFAULT 512
#define PVFS_SEARCH_INACTIVITY_DEFAULT 300
# a successful run for any of these tests an error.
local.resolve.*.async
local.iconv.*.next_codepoint()
-base.delaywrite.finfo update on close
base.delete.*.deltest20a
base.delete.*.deltest20b
rpc.winreg.*security
#
# Please add a comment for each testsuite you disable explaining why
# it is being skipped.
-base.delaywrite
raw.composite
base.iometer
base.casetable
smb2.notify
smb2.scan
ntvfs.cifs.base.charset
-ntvfs.cifs.base.delaywrite
ntvfs.cifs.base.iometer
ntvfs.cifs.base.casetable
ntvfs.cifs.base.nttrans
plantest "rpc.echo on ncacn_np over smb2" dc $smb4torture ncacn_np:"\$SERVER[smb2]" -U"\$USERNAME"%"\$PASSWORD" -W \$DOMAIN RPC-ECHO "$*"
# Tests against the NTVFS POSIX backend
-NTVFSARGS="--option=torture:sharedelay=100000 --option=torture:oplocktimeout=3"
+NTVFSARGS=""
+NTVFSARGS="${NTVFSARGS} --option=torture:sharedelay=100000"
+NTVFSARGS="${NTVFSARGS} --option=torture:oplocktimeout=3"
+NTVFSARGS="${NTVFSARGS} --option=torture:writetimeupdatedelay=500000"
+
smb2=`$smb4torture --list | grep "^SMB2-" | xargs`
#The QFILEINFO-IPC test needs to be on ipc$
raw=`$smb4torture --list | grep "^RAW-" | grep -v "RAW-QFILEINFO-IPC"| xargs`
posix:sharedelay = 100000
posix:eadb = $lockdir/eadb.tdb
posix:oplocktimeout = 3
+ posix:writetimeupdatedelay = 500000
[test1]
path = $tmpdir/test1
posix:sharedelay = 100000
posix:eadb = $lockdir/eadb.tdb
posix:oplocktimeout = 3
+ posix:writetimeupdatedelay = 500000
[test2]
path = $tmpdir/test2
posix:sharedelay = 100000
posix:eadb = $lockdir/eadb.tdb
posix:oplocktimeout = 3
+ posix:writetimeupdatedelay = 500000
[cifs]
read only = no
bool err = false; \
if (strict && (g cmp c)) { \
err = true; \
- } else if (gr cmp cr) { \
+ } else if ((g cmp c) && (gr cmp cr)) { \
/* handle filesystem without high resolution timestamps */ \
err = true; \
} \
}
GET_INFO_BOTH(finfo1,pinfo1);
+ COMPARE_WRITE_TIME_GREATER(pinfo1, pinfo0);
/* sure any further write doesn't update the write time */
start = timeval_current();
}
GET_INFO_BOTH(finfo1,pinfo1);
+ COMPARE_WRITE_TIME_GREATER(pinfo1, pinfo0);
/* sure any further write doesn't update the write time */
start = timeval_current();
assert(res.msgs.length == 2);
}
- var res = ldb.search("(&(anr=testy ldap)(objectClass=user))");
- if (res.error != 0 || res.msgs.length != 2) {
- println("Found only " + res.msgs.length + " for (&(anr=\"testy ldap\")(objectClass=user))");
+ var res = ldb.search("(&(anr=\"testy ldap\")(objectClass=user))");
+ if (res.error != 0 || res.msgs.length != 0) {
+ println("Found " + res.msgs.length + " for (&(anr=\"testy ldap\")(objectClass=user))");
assert(res.error == 0);
- assert(res.msgs.length == 2);
+ assert(res.msgs.length == 0);
}
// Testing ldb.search for (&(anr=ldap)(objectClass=user))