24 May, 2009
1 commit
-
Small change (mostly formatting) to limit lookup based open calls to
file create only.After discussion yesteday on samba-technical about the posix lookup
regression, and looking at a problem with cifs posix open to one
particular Samba version, Jeff and JRA realized that Samba server's
behavior changed in this area (posix open behavior on files vs.
directories). To make this behavior consistent, JRA just made a
fix to Samba server to alter how it handles open of directories (now
returning the equivalent of EISDIR instead of success). Since we don't
know at lookup time whether the inode is a directory or file (and
thus whether posix open will succeed with most current Samba server),
this change avoids the posix open code on lookup open (just issues
posix open on creates). This gets the semantic benefits we want
(atomicity, posix byte range locks, improved write semantics on newly
created files) and file create still is fast, and we avoid the problem
that Jeff noticed yesterday with "openat" (and some open directory
calls) of non-cached directories to one version of Samba server, and
will work with future Samba versions (which include the fix jra just
pushed into Samba server). I confirmed this approach with jra
yesterday and with Shirish today.Posix open is only called (at lookup time) for file create now.
For opens (rather than creates), because we do not know if it
is a file or directory yet, and current Samba no longer allows
us to do posix open on dirs, we could end up wasting an open call
on what turns out to be a dir. For file opens, we wait to call posix
open till cifs_open. It could be added here (lookup) in the future
but the performance tradeoff of the extra network request when EISDIR
or EACCES is returned would have to be weighed against the 50%
reduction in network traffic in the other paths.Reviewed-by: Shirish Pargaonkar
Tested-by: Jeff Layton
CC: Jeremy Allison
Signed-off-by: Steve French
22 May, 2009
1 commit
-
Posix open code was not properly adding the file to the
list of open files. Fix allocating cifsFileInfo
more than once, and adding twice to flist and tlist.
Also fix mode setting to be done in one place in these
paths.Signed-off-by: Steve French
Reviewed-by: Shirish Pargaonkar
Tested-by: Jeff Layton
Tested-by: Luca Tettamanti
19 May, 2009
1 commit
-
This is the third respin of the patch posted yesterday to fix the error
handling in cifs_follow_symlink. It also includes a fix for a bogus NULL
pointer check in CIFSSMBQueryUnixSymLink that Jeff Moyer spotted.It's possible for CIFSSMBQueryUnixSymLink to return without setting
target_path to a valid pointer. If that happens then the current value
to which we're initializing this pointer could cause an oops when it's
kfree'd.This patch is a little more comprehensive than the last patches. It
reorganizes cifs_follow_link a bit for (hopefully) better readability.
It should also eliminate the uneeded allocation of full_path on servers
without unix extensions (assuming they can get to this point anyway, of
which I'm not convinced).On a side note, I'm not sure I agree with the logic of enabling this
query even when unix extensions are disabled on the client. It seems
like that should disable this as well. But, changing that is outside the
scope of this fix, so I've left it alone for now.Reported-by: Jeff Moyer
Signed-off-by: Jeff Layton
Reviewed-by: Jeff Moyer
Reviewed-by: Christoph Hellwig
Signed-off-by: Steve French
14 May, 2009
1 commit
-
cifs_strndup_from_ucs returns NULL on error, not an ERR_PTR
Signed-off-by: Jeff Layton
Signed-off-by: Steve French
09 May, 2009
2 commits
-
Signed-off-by: Al Viro
-
Push BKL down into ->umount_begin()
Signed-off-by: Alessio Igor Bogani
Signed-off-by: Al Viro
08 May, 2009
1 commit
-
Remove adding open file entry twice to lists in the file
Do not fill file info twice in case of posix opens and createsSigned-off-by: Shirish Pargaonkar
Signed-off-by: Steve French
06 May, 2009
2 commits
-
On mount, "sec=ntlmssp" can now be specified to allow
"rawntlmssp" security to be enabled during
CIFS session establishment/authentication (ntlmssp used to
require specifying krb5 which was counterintuitive).Signed-off-by: Steve French
-
We were not setting the SMB uid in NTLMSSP authenticate
request which could lead to INVALID_PARAMETER error
on 2nd session setup.Signed-off-by: Steve French
04 May, 2009
1 commit
-
The NTLMSSP code was removed from fs/cifs/connect.c and merged
(75% smaller, cleaner) into fs/cifs/sess.cAs with the old code it requires that cifs be built with
CONFIG_CIFS_EXPERIMENTAL, the /proc/fs/cifs/Experimental flag
must be set to 2, and mount must turn on extended security
(e.g. with sec=krb5).Although NTLMSSP encapsulated in SPNEGO is not enabled yet,
"raw" ntlmssp is common and useful in some cases since it
offers more complete security negotiation, and is the
default way of negotiating security for many Windows systems.
SPNEGO encapsulated NTLMSSP will be able to reuse the same
code.Signed-off-by: Steve French
02 May, 2009
7 commits
-
Signed-off-by: Steve French
-
Signed-off-by: Steve French
-
Signed-off-by: Steve French
-
Signed-off-by: Jeff Layton
Signed-off-by: Steve French -
Remove dead NTLMSSP support from connect.c prior to addition of
the new code to replace it.Signed-off-by: Steve French
-
When multiply mounting from the same client to the same server, with
different userids, we create a vcnum which should be unique if
possible (this is not the same as the smb uid, which is the handle
to the security context). We were not endian converting additional
(beyond the first which is zero) vcnum properly.CC: Stable
Acked-by: Shirish Pargaonkar
Acked-by: Jeff Layton
Signed-off-by: Steve French -
Signed-off-by: Steve French
01 May, 2009
9 commits
-
Removes two sparse CHECK_ENDIAN warnings from Jeffs earlier patch,
and removes the dead readlink code (after noting where in
findfirst we will need to add something like that in the future
to handle the newly discovered unexpected error on FindFirst of NTFS symlinks.Signed-off-by: Steve French
-
Signed-off-by: Steve French
-
Signed-off-by: Steve French
-
The earlier patch to move this code to use the new unicode helpers
assumed that the filename strings would be null terminated. That's not
always the case.Instead of passing "max_len" to the string converter, pass "min(len,
max_len)", which makes it do the right thing while still keeping the
parser confined to the response. Also fix up the prototypes of this
function and the callers so that max_len is unsigned (like len is).Signed-off-by: Jeff Layton
Signed-off-by: Steve French -
Signed-off-by: Steve French
-
In most cases, cifs_strndup is converting from Unicode (UCS2 / UTF-32) to
the configured local code page for the Linux mount (usually UTF8), so
Jeff suggested that to make it more clear that cifs_strndup is doing
a conversion not just memory allocation and copy, rename the function
to including "from_ucs" (ie Unicode)Signed-off-by: Steve French
-
Added loop check when mounting DFS tree. mount will fail with
ELOOP if referral walks exceed MAX_NESTED_LINK count.Signed-off-by: Igor Mammedov
Acked-by: Jeff Layton
Signed-off-by: Steve French -
Having remote dfs root support in cifs_mount, we can
afford to pass into it UNC that is remote.Signed-off-by: Igor Mammedov
Acked-by: Jeff Layton
Signed-off-by: Steve French -
Two years ago, when the session setup code in cifs was rewritten and moved
to fs/cifs/sess.c, we were asked to keep the old code for a release or so
(which could be reenabled at runtime) since it was such a large change and
because the asn (SPNEGO) and NTLMSSP code was not rewritten and needed to
be. This was useful to avoid regressions, but is long overdue to be removed.
Now that the Kerberos (asn/spnego) code is working in fs/cifs/sess.c,
and the NTLMSSP code moved (NTLMSSP blob setup be rewritten with the
next patch in this series) quite a bit of dead code from fs/cifs/connect.c
now can be removed.This old code should have been removed last year, but the earlier krb5
patches did not move/remove the NTLMSSP code which we had asked to
be done first. Since no one else volunteered, I am doing it now.It is extremely important that we continue to examine the documentation
for this area, to make sure our code continues to be uptodate with
changes since Windows 2003.Signed-off-by: Steve French
30 Apr, 2009
8 commits
-
...and remove cifs_convertUCSpath. There are no more callers. Also add a
#define for the buffer used in the readdir path so that we don't have so
many magic numbers floating around.Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
Change CIFSSMBUnixQuerySymLink to use the new unicode helper functions.
Also change the calling conventions so that the allocation of the target
name buffer is done in CIFSSMBUnixQuerySymLink rather than by the caller.Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
...and change decode_unicode_ssetup to be a void function. It never
returns an actual error anyway.Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
Rename cifs_strlcpy_to_host to cifs_strndup since that better describes
what this function really does. Then, convert it to use the new string
conversion and measurement functions that work in units of bytes rather
than wide chars.Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
Working in units of words means we do a lot of unnecessary conversion back
and forth. Standardize on bytes instead since that's more useful for
allocating buffers and such. Also, remove hostlen_fromUCS since the new
function has a similar purpose.Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
Add a replacement function for cifs_strtoUCS_le. cifs_from_ucs2
takes args for the source and destination length so that we can ensure
that the function is confined within the intended buffers.Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French -
Signed-off-by: Jeff Layton
Acked-by: Suresh Jayaraman
Signed-off-by: Steve French
21 Apr, 2009
3 commits
-
Increase size of tmp_buf to possible maximum to avoid potential
overflows.Pointed-out-by: Jeff Layton
Signed-off-by: Suresh Jayaraman
Acked-by: Jeff Layton
Signed-off-by: Steve French -
There is a possibility for the path_name and node_name buffers to
overflow if they contain charcters that are >2 bytes in the local
charset. Resize the buffer allocation so to avoid this possibility.Also, as pointed out by Jeff Layton, it would be appropriate to
rename the function to cifs_strlcpy_to_host to reflect the fact
that the copied string is always NULL terminated.Signed-off-by: Suresh Jayaraman
Acked-by: Jeff Layton
Signed-off-by: Steve French -
Signed-off-by: Steve French
18 Apr, 2009
1 commit
-
When attempting to rename a file on a read-only share, the kernel can
call cifs_unlink on a negative dentry, which causes an oops. Only try
to unlink the file if it's a positive dentry.Signed-off-by: Jeff Layton
Tested-by: Shirish Pargaonkar
CC: Stable
Signed-off-by: Steve French
17 Apr, 2009
2 commits
-
This pointer isn't used again after this point. It's also not updated in
the ascii case, so there's no need to update it here.Pointed-out-by: Dave Kleikamp
Signed-off-by: Jeff Layton
Signed-off-by: Steve French -
...to make it easier to find problems in this area in the future.
Signed-off-by: Jeff Layton
Signed-off-by: Steve French