Smbmount broken?

[Urban Widmark]

On Sat, 5 Jan 2002, Andrew Bartlett wrote:

> What I did for my ERRMAPEXTRACT smbtorture test was to set
> cli->use_spnego = False, which makes things sane again.

Ok. smbmount already does a 'c->force_dos_errors = True;' so it would just
be another flag.

> > And HEAD as a smb server always seems to talk unicode (no matter what the
> > client says). I don't know if it is an intentional move to drop support
> > for older clients or if the plan is to add that back later on. Or if it is
> > just something with the new session setup code.
> 
> We *should* only be doing unicode if negotiated...

I am pretty sure it ignored what the client said. But that could have had
something to do with cli->use_spnego or something else I didn't do. I'll
check again, next time I play "get samba-HEAD to behave" :)

However,
smbd/process.c:construct_reply_common() sets FLAGS2_UNICODE_STRINGS
unconditionally. As well as FLAGS2_32_BIT_ERROR_CODES.

And,
lib/charcnv.c:push_string() looks at smb_flg2 when deciding if it should 
use unicode or ascii.

To me that doesn't look like it tries to follow what was negotiated.

/Urban