Age | Commit message (Collapse) | Author | Files | Lines |
|
metze
|
|
This patch introduces
struct stat_ex {
dev_t st_ex_dev;
ino_t st_ex_ino;
mode_t st_ex_mode;
nlink_t st_ex_nlink;
uid_t st_ex_uid;
gid_t st_ex_gid;
dev_t st_ex_rdev;
off_t st_ex_size;
struct timespec st_ex_atime;
struct timespec st_ex_mtime;
struct timespec st_ex_ctime;
struct timespec st_ex_btime; /* birthtime */
blksize_t st_ex_blksize;
blkcnt_t st_ex_blocks;
};
typedef struct stat_ex SMB_STRUCT_STAT;
It is really large because due to the friendly libc headers playing macro
tricks with fields like st_ino, so I renamed them to st_ex_xxx.
Why this change? To support birthtime, we already have quite a few #ifdef's at
places where it does not really belong. With a stat struct that we control, we
can consolidate the nanosecond timestamps and the birthtime deep in the VFS
stat calls.
At this moment it is triggered by a request to support the birthtime field for
GPFS. GPFS does not extend the system level struct stat, but instead has a
separate call that gets us the additional information beyond posix. Without
being able to do that within the VFS stat calls, that support would have to be
scattered around the main smbd code.
It will very likely break all the onefs modules, but I think the changes will
be reasonably easy to do.
|
|
Thanks,
Volker
|
|
This is the first of a series of patches that change path based
operations to operate on a struct smb_filename instead of a char *.
This same concept already exists in source4.
My goals for this series of patches are to eventually:
1) Solve the stream vs. posix filename that contains a colon ambiguity
that currently exists.
2) Make unix_convert the only function that parses the stream name.
3) Clean up the unix_convert API.
4) Change all path based vfs operation to take a struct smb_filename.
5) Make is_ntfs_stream_name() a constant operation that can simply
check the state of struct smb_filename rather than re-parse the
filename.
6) Eliminate the need for split_ntfs_stream_name() to exist.
My strategy is to start from the inside at unix_convert() and work my
way out through the vfs layer, call by call. This first patch does
just that, by changing unix_convert and all of its callers to operate
on struct smb_filename. Since this is such a large change, I plan on
pushing the patches in phases, where each phase keeps full
compatibility and passes make test.
The API of unix_convert has been simplified from:
NTSTATUS unix_convert(TALLOC_CTX *ctx,
connection_struct *conn,
const char *orig_path,
bool allow_wcard_last_component,
char **pp_conv_path,
char **pp_saved_last_component,
SMB_STRUCT_STAT *pst)
to:
NTSTATUS unix_convert(TALLOC_CTX *ctx,
connection_struct *conn,
const char *orig_path,
struct smb_filename *smb_fname,
uint32_t ucf_flags)
Currently the smb_filename struct looks like:
struct smb_filename {
char *base_name;
char *stream_name;
char *original_lcomp;
SMB_STRUCT_STAT st;
};
One key point here is the decision to break up the base_name and
stream_name. I have introduced a helper function called
get_full_smb_filename() that takes an smb_filename struct and
allocates the full_name. I changed the callers of unix_convert() to
subsequently call get_full_smb_filename() for the time being, but I
plan to eventually eliminate get_full_smb_filename().
|
|
sbuf as invalid (if not already read via stat()). Still trying
to find the build farm RAW-STREAM errors and it's happening
in a openX call....
Jeremy.
|
|
We keep the seqnum/mid mapping in the smb_request structure.
This also moves one global variable into the
smbd_server_connection struct.
metze
|
|
[MS-SMB] 3.3.5.1 Receiving Any Message says that the seqnum
is incremented by only for ntcancel requests for any other
request it's by incremented by 2, even if it doesn't expect
a response.
metze
|
|
This extends the file_id struct to add an additional generic uint64_t
field: extid. For backwards compatibility with dev/inodes stored in
xattr_tdbs and acl_tdbs, the ext id is ignored for these databases.
This patch should cause no functional change on systems that don't use
SMB_VFS_FILE_ID_CREATE to set the extid.
Existing code that uses the smb_share_mode library will need to be
updated to be compatibile with the new extid.
|
|
This changelist allows for the addition of custom performance
monitoring modules through smb.conf. Entrypoints in the main message
processing code have been added to capture the command, subop, ioctl,
identity and message size statistics.
|
|
This allows module implementors to customize what allocation size is
returned to the client.
|
|
np_open/read/write don't have to know about files_struct
|
|
It is possible for a posix file created locally or over nfs to have a
":" in the name. Since ":" is a reserved character in windows,
filenames containing a colon must be mangled in a directory listing.
Right now files containing colons will not even be displayed in
directory listings if streams modules are in use. During the
directory listing the file will be detected as a stream because of the
colon, but the streams module will fail to find the stream since it
doesn't exist. This fix adds a step to is_ntfs_stream_name that stats
the filename to differentiate between actual streams and files
containing colons.
While this is an improvement, it isn't perfect. Consider the case
where there is a file on disk called "a.txt:s1" and also a file called
"a.txt" that has a stream called "s1". This patch will always
preference "a.txt:s1" over a.txt's s1 stream.
The real issue is that at the vfs level, the vfs modules have no way
to tell between a demangled name with a colon and an actual stream. A
more invasive, but better, long-term fix would be to add all paths
that come over the wire into a struct containing metadata about the
path. This metadata could include a flag to indicate whether the path
came over the wire with a colon ":" (guaranteeing that the client is
requesting a stream). Passing this struct down to the lower levels,
including all path-based vfs calls, would allow the above case to be
handled correctly in all cases.
|
|
The goal is to move all this variables into a big context structure.
metze
|
|
Jeremy.
|
|
|
|
The test_streams_rename2 test in RAW-STREAMS verifies these changes
|
|
|
|
|
|
|
|
This replaces the is_dos_path bool with a more future-proof argument.
The next step is to plumb INTERNAL_OPEN_ONLY through this flag instead
of overridding the oplock_request.
|
|
SMB_VFS_CREATE_FILE
|
|
Now unix paths can be differentiated from windows paths so the
underlying create_file implementations can convert paths correctly.
|
|
Modify all callers of create_file to go through SMB_VFS_CREATE_FILE
|
|
This could also be handled inside each ACL VFS module, by setting the PRESENT
flag when a NULL [SD]ACL is created.
|
|
|
|
This is the one where I found the problem that led to 3.2.5. So if there is one
checkin in the last year that I would like others to review and *understand*,
it is this one :-)
Volker
|
|
This converts the range checks for the setup[] array to rely on req->wct being
set correctly in init_smb_request. As that already verifies the vwv array to be
in the range of the smb_request inbuf, we don't have to do overflow checks here
anymore.
Jeremy, please check thoroughly! :-)
Thanks,
Volker
|
|
This fixes a potential crash bug, a client can make us read memory we
should not read. Luckily I got the disp checks right...
Volker
(cherry picked from commit 64a1d80851da5b05e70ec6c96f6e9bd473748369)
(cherry picked from commit f04c5650a3aeca23591ddc781c4b297caaf9bb3f)
|
|
This is a modification of Jeremy's 7522ef15aca2429ef57c75d8297dd8121e79c9da
commit.
If no DACL/SACL is present in the packet, the SEC_INFO field should still be
passed down as is to the VFS layer to signal the creation of a NULL DACL/SACL.
As seen in metze RAW-ACL test_nttrans_create_null_dacl(), a NULL DACL is set
regardless of the SEC_DESC_DACL_PRESENT bit being set.
|
|
|
|
Jeremy.
|
|
Instead, fix up the outbuf in send_xx_reply. In those routines, we know
what we are returning.
|
|
|
|
|
|
right. Only inheritance left to fix.
Jeremy.
|
|
|
|
|
|
Now that "req" is available everywhere, use it. Rename srvstr_pull_buf_talloc()
to srvstr_pull_req()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
on the way to get rid of chain_fsp
|
|
The goal is to get rid of the chain_fsp global variable
|
|
|
|
to specific bits every time a security descriptor is set. The S4 torture suite proves
that generic bits are not returned when querying an ACL set using them (ie. only
the specific bits are stored on disk).
Jeremy.
|
|
(This used to be commit 23e9fed28cf9998534b3c55f4b20a562da507552)
|
|
When alignment was in place, we pretended to send more data/params according to
the param_offset/param_length and data_offset/data_length parameters than would
actually fit into the SMB according to the NBSS length field.
(This used to be commit ef3c132b8455c6fe4d0bb9f0be881040a806a4ed)
|