Question about smbtorture/NEGNOWAIT

[Andrew Bartlett]

paul popelka wrote:
> 
> I've been running the NEGNOWAIT test that is part of smbtorture.
> 
> I've got some questions about the test.
> 
> It tries to send 50000 SMB_COM_NEGOTIATE requests to a
> server all with the same mid (1) and pid.  The test does not wait for
> responses.
> So, there can be multiple negotiate requests outstanding at a given
> time all with the same mid and pid.  This would seem like a violation of
> 
> the cifs protocol.
> 
> What is this test trying to test?

What the server does when the client violates the CIFS protocol :-)

> Is it checking to see if the server rejects the 49999 negotiates after
> the first one?
> Is it checking to see if the server can cope which several concurrent
> requests with the same mid and pid?

Mainly its just a plain, booring 'crash NT' test.   I think MS fixed it
- I understand they actually have an internal port of smbtorture to
win32 that they use to avoid these embarassing issues ;-)

> Also, I noticed that the test always fails because it calls
> close_connection()
> which tries to do a tree disconnect.  Since the test uses
> open_nbt_connection()
> there is no tree connected.  So, the tree disconnect fails causing
> close_connection()
> to fail causing the test to fail.
> 
> Should the test call cli_shutdown() instead?

Probably.  Feel free to provide a patch.

Andrew Bartlett

-- 
Andrew Bartlett                                 abartlet at pcug.org.au
Manager, Authentication Subsystems, Samba Team  abartlet at samba.org
Student Network Administrator, Hawker College   abartlet at hawkerc.net
http://samba.org     http://build.samba.org     http://hawkerc.net