Re: First snapshot of a proposed set of fixes to make 2.0.34

Bill Hawes (
Mon, 19 Jan 1998 16:11:31 -0500

Marty Leisner wrote:

> Timestamping has been a problem ever since I remember with smbfs.
> Its also a problem that win95 and NT sends the reverse order of
> dates/times in one of the queries. I haven't done an analysis recently
> of 2.1.* and 2.0.* against NT and win95...

The main reason it's been so difficult to get the timestamp problems
fixed in smbfs is that all of the MS SMB servers implement different
subsets of the SMB protocol. That coupled with the incomplete or
misleading documentation makes it a trial-and-error process to find
working messages to get timestamps. Problems like the reversed order of
date/time have to be discovered and worked around.

> The fact there has to be changes is one of the reasons I think having these
> vendor specific file systems decoupled from the kernel not a bad idea...
> Until recently, ntfs was decoupled, so I could update ntfs independently
> fromn the kernel.

As long as the code is reasonably maintained I don't see a problem with
having it in the kernel. SMB servers may be buggy, but once the problems
are fixed thay stay fixed (at least until the next MS release). And you
could still maintain NTFS separately from the kernel if you want to ..