I tried out 2.1.70 + your smbfs_2.1.70 patches against the WfW box. The
only difference I see is that the application doing the copying no longer
complains about being unable to setattr. However, the target file still
is not picking up mtime :-(.
To make sure that I'm not going senile, I mounted the WfW share from a box
running 2.0.32. Under that environment, the mtimes _are_ getting set
properly. We're still not doing exactly what it must be...
Here are the logs:
Dec 3 21:25:37 cy kernel: smb_newconn: state valid, pid=234
Dec 3 21:25:42 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:25:52 cy last message repeated 2 times
Dec 3 21:28:39 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:28:44 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:28:49 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:28:49 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:28:49 cy kernel: smb_proc_setattr: BENCH/ingo, ignoring ERRDOS code 50
Dec 3 21:28:49 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:29:02 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:29:09 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:29:09 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:29:09 cy kernel: smb_proc_setattr: BENCH/ingo, ignoring ERRDOS code 50
Dec 3 21:29:10 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:29:14 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:29:44 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:29:44 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:29:44 cy kernel: smb_proc_setattr: BENCH/INGO, ignoring ERRDOS code 50
Dec 3 21:29:44 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:29:53 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:30:05 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:30:05 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:30:05 cy kernel: smb_proc_setattr: BENCH/INGO, ignoring ERRDOS code 50
Dec 3 21:30:05 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:30:09 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:30:20 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:30:20 cy kernel: smb_errno: class ERRDOS, code 50 from command 9
Dec 3 21:30:20 cy kernel: smb_proc_setattr: BENCH/INGO, ignoring ERRDOS code 50
Dec 3 21:30:21 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
Dec 3 21:30:24 cy kernel: smb_request_ok: rcls=1, err=18 mapped to 0
I guess the "ignoring.." message means that you are just swallowing the
error status at the filesystem level and not reporting back, yes?
I'm willing to keep at this if you are.
Also, I keep forgetting to mention this, but I found that 2.0.32 smbfs
exhibited the best performance when the share was mounted with:
smbmount //via/c /via -m 1024
Certain things seemed sensitive to the maximum xmit block size (at least I
_think_ that is what the -m flag does on older smbmount binaries). At
any rate, I'm using that setting under 2.0.32 and timestamps work
properly.
How can we enforce a similar setting under the new smbclient mount method?
Steve