proprietary firmware

joerg joerg.twinklephone at gmx.de
Fri Feb 8 23:07:03 CET 2008


> I like the idea of having total control over my electronic devices - 
> especially if they are able to collect everything about my life like a 
> mobile phone. Thats why I'm currently living without any mobil.
> If I am able to look into what runs on my device, I can trust that 
> stuff. so I'm one of those guys saying doing everything open source is 
> way better than gaining a little cpu-speed. 

Yep, but who shall do this coding, if not the manufacturer of the gsm-modem, 
the gps-chip, the wlan-chip... etc.?

> and by the way I don't think  
> that the cpu-speed is too limited on that device. usually cpus don't 
> have to do anything. 

That's why they sleep and don't consume a lot of power.

> and a driver doesnt need too much. This smal gap  
> could be closed esysly by optimizing things for the hardware.

And by waking the powerful main cpu 200/sec for such a silly thing as decoding 
GSM-CCH packets to detect a call, your battery will drain like a... well, you 
wouldn't like it ;-) And it even can't be done, because on GSM-modem there is 
no external interrupt line to wake the main CPU every time a packet comes in. 
The modem is designed to use internal uC driven by firmware to do this job, 
better than the main CPU ever might.

btw: you *never* know exactly what's going on on your device, because you 
never can trust in (LSI-)chip design, which usually isn't "open source". If 
you want to know, you have to observe from outside the system.  Do you know 
what your BIOS is doing? Sure not! Do you demand it to be coded "userspace" 
FOSS therefore? It can't be done. You always have a hardware API somewhere, 
and you may not see what's behind. So you have to do a tradeoff between total 
transparency and potential risk arising from an opaque subsystem. I don't 
want to know how 802.11b protocol is handled in the wlan-chip. I want to have 
a powerfull bugfree API for the subsystem. Complete control is an illusion.
An anecdote to illustrate: The german FETAP711 (IIRC) telephone had a units 
counter built around a uA709 -opamp. By no means of analyzing the circuit and 
the specs of the used parts you could see: this device was oscillating at 
about 100KHz and made a fine eavesdrop-transmitter whenever the handset was 
offhook. In *all* of those phones the POST telco produced back in the '80s! 
And this was in pre-firmware times, with simple parts like transistors etc.

j




More information about the community mailing list