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.
-
We ran across a bug joining our Samba server to a Win2K domain with LDAP
signing turned on. Upon investigation I discovered that there is a bug
in Win2K server which returns a duplicated responseToken in the LDAP
bindResponse packet. This blob is placed in the optional mechListMIC
field which is unsupported in both Win2K and Win2K3. You can see RFC
2478 for the proper packet construction. I've worked with metze on this
to confirm all these finding.
This patch properly parses then discards the mechListMIC field if it
exists in the packet, so we don't produce a malformed packet error,
causing LDAP signed joins to fail. Also attached is a sniff of the
domain join, exposing Win2Ks bad behavior (packet 21).
-
(I've just changed the scope of the DATA_BLOB mechList)
metze
(This used to be commit 200b5bfb8180af09446762e915eac63d14c6c7b0)
to make this change for ages, but now with the issue of "open" requiring it,
this is the time to just do all of them.
Derrell
(This used to be commit e746aaaf4db7099252ef048da7857bd488cb681f)
device. The device resets a NBT connection on port 139 when it receives a
NetBIOS keepalive request. That request should be supported when NetBIOS is
in use; Windows is behaving badly.
libsmbclient needs a way to determine if a connection is still alive, and
was using a NetBIOS keepalive request if port 139 was in use (on the
assumption that it was probably NBT), and getpeername() when port 139 was
not being used (assuming naked transport).
This patch simplifies the code by exclusively using getpeername() to check
whether a connection is still alive. The NetBIOS keepalive request is
optional anyway (with preference being given to using TCP mechanisms for the
same purpose), so this should be both simpler and more reliable.
Derrell
(This used to be commit 1f122352b02e3f4be9ac2d638b18807dafd05429)
Windows Explorer doesn't complain about the order (and so that they get
interpreted properly).
Derrell
(This used to be commit 8f371e2ea97a3b58d1c7c3aa1368a0904295f681)
the request, presumably due to the PROTECTED flag not being set. Setting
that flag (in make_sec_desc()) has much wider implications than just to
libsmbclient, so instead of modifying that, we'll remove security
descriptors by setting the number of ACEs to zero. At some point, we might
want to look into whether we should actually be setting the PROTECTED flag
in the DACL.
Reference http://www.microsoft.com/technet/prodtechnol/windows2000serv/reskit/distrib/dsce_ctl_qxju.mspx?mfr=true
Derrell
(This used to be commit 319df380e579fd860348a8f08a584e13161dde9d)
prs_mem_free() is not the function to be called to free memory allocated by
prs_alloc_mem(). I've added a comment so others may not get bitten too.
- Remove incorrect memory free calls added yesterday to replace SAFE_FREE.
The memory is actually now on a talloc context, so gets freed by the caller
when that context is freed. We don't need to free it iternally.
Derrell
(This used to be commit 2fde343150c17959fc970b18e1eb4efde800b4db)
incremented too far in some circumstances. In these cases, only the first
of multiple concatenated strings would be seen.
- Working on bug 4649 pertaining to delete an ACL, this fixes the reported
crash. It appears to have been an incomplete switchover from malloc to
talloc, as the memory was still being freed with SAFE_FREE.
Deleting ACLs still doesn't work. Although a valid request is sent to the
server and a SUCCESS response is returned, the method that's used in
libsmbclient for deleting ACLs seems to be incorrect. In looking at the
samba4 torture tests, it appears that we should be turning on the INHERIT
flag if we want to delete the ACL. (I could use some assistance on the
proper flags to send, from anyone familiar with this stuff.)
- Apply patch from SATOH Fumiyasu to fix bug 4750. smbc_telldir_ctx() was not
returning a value useful to smbc_lseekdir_ctx().
Derrell
(This used to be commit 2ac502e29bd8390252fe4ae8344faab49ca01ff5)
left as nonzero as returned by the failed cli_session_setup_spnego. When we then try
to authenticate as the user in cli_session_setup this returns an
error "Bad userid" (as seen in wireshark).
"We should only leave cli->vuid != 0 on success. Looks like it's
getting set in the cli_session_setup_blob_receive() call and not
cleared again on error."
Jeremy.
(This used to be commit fa8e66dd8d2c68b91b27169c3c43820989f58758)
checkin will pull this up to srvstr_get_path. At that point we can get more
independent of the inbuf, the base_ptr in pull_string will only be used
to satisfy UCS2 alignment constraints.
(This used to be commit 836782b07bf133e9b2598c4a089f1c810e4c7754)
When renaming a file across 2 filesystem a samba server returns
NT_STATUS_NOT_SAME_DEVICE but thius is not translated to EXDEV,
and the generic EINVAL is returned instead.
This should fix it, Jeremy or Derrel please check if this is ok.
(This used to be commit b35038fa4e3e69f1397758497a46dc0d37edee79)
We were incorrectly using the renew_till timestamp instead of the renewed
ticket's endtime to calculate the next refreshing date.
Guenther
(This used to be commit aa3511a5b5e6a96a02110a7ad0ab1d43e6d25766)
This replaces the internal explicit dev/ino file id representation by a
"struct file_id". This is necessary as cluster file systems and NFS
don't necessarily assign the same device number to the shared file
system. With this structure in place we can now easily add different
schemes to map a file to a unique 64-bit device node.
Jeremy, you might note that I did not change the external interface of
smb_share_modes.c.
Volker
(This used to be commit 9b10dbbd5de8813fc15ebbb6be9b18010ffe8139)
