Thread: N900 as a Wifi Hotspot
View Single Post
rambo is offline rambo
2010-01-31 , 16:45
Posts: 540 | Thanked: 288 times | Joined on Sep 2009
#126
Originally Posted by hcm View Post
I just tried it out and would like to share my experience:
Many thanks for testing

Originally Posted by hcm View Post
Code:
ERROR:dbus.proxies:Introspect error on :1.79:/com/nokia/icd: dbus.exceptions.DBusException: org.freedesktop.DBus.Error.NotSupported: Unsupported interface or method
IcD[1,2] does not provide DBUS introspection, this sucks a lot in more ways than just getting this cosmetic problem, it's safe to ignore this error.

Originally Posted by hcm View Post
When connecting to the network with another device, I experienced the problem that *sometimes* the DNS did not work… Sometimes I had to give a manual DNS (had to try several servers), sometimes it worked with autoassigning…
Sometimes I have problem with not getting IP from the DHCP server in dnsmasq, though then my laptop refuses to work with the n900 as router even if I assign ip etc manually.

Originally Posted by hcm View Post
Is it possible to connect multiple devices at the same time? The network shows up as ad-hoc, but apparently I was able to connect with two devices at the same time (but with DNS-problems, so I'm not 100% sure about it…)
WiFi Ad-Hoc mode itself has no artificial limit on how many devices may talk to each other, dnsmasq DHCP range in this case is 10-100 (which is way overkill but costs nothing).

Now there will be some practical issues when number of devices using the N900 as router goes above X (such as bandwidth, processor and other resource constraints), determining X will take some experimentation.

Originally Posted by hcm View Post
What changes are made in the custom kernel, apart from NAT-stuff and framebuffer? Is it likely that some functionality of the original kernel is missing or that power-consumption is higher (when not using the hotspot)?
The configuration patch (apply this to the .config generated via "make rx51_defconfig") is in svn, but basically I have added:
  • Framebuffer
  • NAT related modules
  • QoS modules

If all modules are unloaded correctly (big if...) on hotspot stop then in theory it should not consume (noticeably) more power, if the connection tracking stuff is loaded in kernel while the hotspot is no longer active I would guess network activity would take slightly more processing power than it normally would.

The QoS stuff might be removed, it adds size to the base kernel too and I'm not sure if we're going to use it for anything (if there was a good way to know the actual speed of the GPRS connection we could automagically run wondershaper to get nicer network performance)

In other news I have made some improvements to the backend and frontend, see updated instructions on project home page. NOTE: for debugging any backend issues it's best to run the hotspot_backend.py manually.
 

The Following User Says Thank You to rambo For This Useful Post: