> >I have not understood you here. Could you please explain it more detailed, > >please? > >There is no open() call in modules (like the authentication module), only in > >drivers. > Well, SeOS is doing it for different reason - it suppose to make Unix or Windows more secure. So they replaced standard system open() call with their own. > So whatever userland application tries to open it has to go through authentication module of SeOS. Don't get fooled by the name 'SeOS' - it's not an OS it's just a secure subsystem for Unix and Windows. Replacement of existing security schema with more secure. And it's commercial product. I think other commercial security products are using the same idea. My point was leave the authentication to userland application (as I think you planned) and don't even create special hooks for it in your module - standard open() call is enough. Is this an authentication method for applications that want to access a driver/module? What I wanted to do is to route the incoming connection tries to a userland daemon that authenticates the incoming connection try, not the driver access. I hope I have understood you correctly. > Are you talking about PPP daemon that accepts connections from modem pool ? It means at least 1000 modem lines or to be more realistic 10,000 modem lines. > I guess you know that there are hardware PPP implementations - old DEC terminal server we throw away 10 years ago was capable of this. And they are not that expensive - don't try to compete with hardware implementations. So, how much connection traffic does a server have then? What about PPPoE servers? Are there hardware implementations for them or is this a normal Linux server? Do you think that there will never be more than 30 incoming connection per second to a big university or to a small ISP? Then, a small authentication daemon that creates 30 threads for authentication requests and tells one unused thread to authenticate when a new connection is incoming should be enough. It should not be too complicated to implement this. > >Oh, having 1000 requests at the same time means having 1000 threads at the > >same time. I heard that our limit is 1600 threads or so. Or was it the Zeta > >limit? Or is this completely wrong? I just remembered the number 1600 in the > >context of a kernel limitation. > >This might become a problem, so blocking is useful at some point anyway. > > > >Waldemar > > > I think the simpler you make it the faster it will be. That is correct. :) I will do my best (which can be not enough ;). In this week and the beginning of the next week I will not have much time to code, but I will release some headers soon. Is PPP a high-priority task? Maybe I am too slow, so if you want to make it going faster, just help me. If there are more important tasks at the moment I can work on my own in my free time. Waldemar