3G modem and USB, old & new

[Nick Hibma]

Poul-Henning,

I have had many reports of devices working. I've also had several people 
report that the device failed miserably with similar problems like you are 
perceiving.

I am talking oldusb here, as that is what I am familiar with.

I've not been able to reproduce the problems reliably. But looking at the 
symptoms somehow buffering goes pear-shaped somewhere. There is no 
buffering being done in the u3g code.That's all handled by ucom, but to me 
that looks like cut&paste from other code. So I presume (wildly pointing 
fingers at code I do not yet understand) that the problem is somewhere in 
the combination of ucom and tty layer, or perhaps even in the TTY layer.

Perhaps you have a clue as to where in the TTY layer we could look for 
problems? The usage patterns for the u3g devices is much different from 
other serial devices, as a) the speeds are much higher than other serial 
(USB) devices, and b) data arrives in large chunks of several kb in some 
cases.

Any pointers would be appreciated.

Nick


> I tried using my 3g modem (Huawei E196) yesterday, with both the
> old and the new USB stack, and it fails in slightly different
> ways.
>
> With the old USB stack, it works until I actually try to get a packet
> of more than approx 1024 bytes through, at which point it hangs with
> 	ucom0: ucomreadcb: IOERROR
> And I need to stop and start ppp(1) to get it working until the next
> big packet comes around.
>
> It does not help to reduce the MRU because two small packets back to
> back will also trigger this error.
>
> With the new USB stack, I am not able to talk to the modem at all
> using the cuaU* devices.