Will fix the style comment - but on the fall back to lowest common
denominator ... it won't work.
For this server you have to fall back to 260 (then old servers fall
back to 1). Interestingly this server seemed to have problems with
the default level and with level 1 ... so I don't see a choice to
having two fall backs (but could remove one of the three obviously).
On Wed, Aug 24, 2011 at 2:08 PM, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
> On Tue, 23 Aug 2011 22:16:54 -0500
> Steve French <smfrench@xxxxxxxxx> wrote:
>
>> From 9b7b518633fd3187c7efe260fe20a0961e4a5f0b Mon Sep 17 00:00:00 2001
>> From: Steve French <sfrench@xxxxxxxxxx>
>> Date: Tue, 23 Aug 2011 22:00:26 -0500
>> Subject: [CIFS] ls on mounts to WindowsCE failing with invalid level message
>>
>> Mount succeeds to WindowsCE but when user tries
>> to access the contents of the directory gets "NT Status:
>> STATUS_INVALID_LEVEL" according to wireshark.
>>
>> This server does not support the common windows
>> readdir (Findfirst) level 257 nor level 258, so "fall back"
>> to level 260 if get an EOPNOTSUPP on level FIND_FILE_DIRECTORY_INFO
>> and while we are at it fall back to level 1 (FILE_INFO_STANDARD)
>> for very old servers if we get an EOPNOTSUPP on level 260.
>> (Note that the server returned errors on queries for inode
>> numbers via QueryPathInfo so we don't use the infolevel
>> SMB_FIND_FILE_ID_FULL_DIR_INFO)
>>
>> Reported-and-tested-by: Bryan James <Bryan.James.Krone-EXT@xxxxxxx>
>> Signed-off-by: Steve French <smfrench@xxxxxxxxx>
>> ---
>> fs/cifs/readdir.c | 17 +++++++++++++++++
>> 1 files changed, 17 insertions(+), 0 deletions(-)
>>
>> diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c
>> index 5de03ec..ed29706 100644
>> --- a/fs/cifs/readdir.c
>> +++ b/fs/cifs/readdir.c
>> @@ -270,6 +270,7 @@ ffirst_retry:
>> cifsFile->srch_inf.info_level = SMB_FIND_FILE_DIRECTORY_INFO;
>> }
>>
>> +ffirst_retry_with_std_level:
>> rc = CIFSFindFirst(xid, pTcon, full_path, cifs_sb->local_nls,
>> &cifsFile->netfid, &cifsFile->srch_inf,
>> cifs_sb->mnt_cifs_flags &
>> @@ -283,7 +284,23 @@ ffirst_retry:
>> (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_SERVER_INUM)) {
>> cifs_sb->mnt_cifs_flags &= ~CIFS_MOUNT_SERVER_INUM;
>> goto ffirst_retry;
>> + } else if ((rc == -EOPNOTSUPP) && (cifsFile->srch_inf.info_level ==
>> + SMB_FIND_FILE_DIRECTORY_INFO)) {
>
> ^^^^ there's no need for the extra sets of parenthesis
> here. That violates kernel coding conventions and makes this
> harder to read.
>
>> + /* Windows CE can return EOPNOTSUPP on FILE_DIRECTORY_INFO */
>> + cifsFile->srch_inf.info_level =
>> + SMB_FIND_FILE_BOTH_DIRECTORY_INFO;
> ^^^^
> The code doesn't currently use
> BOTH_DIRECTORY_INFO anywhere. Would it be
> better to skip this step?
>
>> + goto ffirst_retry_with_std_level;
>> + } else if ((rc == -EOPNOTSUPP) && (cifsFile->srch_inf.info_level ==
>> + SMB_FIND_FILE_BOTH_DIRECTORY_INFO)) {
>> + cifsFile->srch_inf.info_level =
>> + SMB_FIND_FILE_FULL_DIRECTORY_INFO;
>
> ^^^^
> Ditto here. Would it be best to just fall back
> to the lowest common denominator here? I'm not sure
> I see a lot of benefit in using any of these
> other infolevels. We'll basically get the same
> info out of SMB_FIND_FILE_INFO_STANDARD and it's
> more widely supported.
>
>> + goto ffirst_retry_with_std_level;
>> + } else if ((rc == -EOPNOTSUPP) && (cifsFile->srch_inf.info_level ==
>> + SMB_FIND_FILE_FULL_DIRECTORY_INFO)) {
>> + cifsFile->srch_inf.info_level = SMB_FIND_FILE_INFO_STANDARD;
>> + goto ffirst_retry_with_std_level;
>> }
>> +
>> error_exit:
>> kfree(full_path);
>> cifs_put_tlink(tlink);
>
>
> --
> Jeff Layton <jlayton@xxxxxxxxxx>
>
--
Thanks,
Steve
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
