[Asterisk-Users] app_rxfax creating bad faxes? (StripOffsets)

[Rich Adamson]

Inline...

> > BTW, spandsp-pre9 is the most current.

That's what I get for only looking at the bottom of the list. :( 
Missed the top one.

> Rich, spandsp-0.0.2pre10 has been on Steve Underwood's site for some 
> time.  There are also "test" app_rxfax and app_txfax modules available.

What do the "test" modules do? (Obviously I've not downloaded as yet.)

> I have two Canon fax machines that consistently fail on the third page. 
>   From the traces, I get a page full of "slow carrier up, slow carrier 
> down" message pairs and then a T4 timeout.  I get the exact result from 
> both machines.  This is on a TDM22B card.  I've tried the machine 
> directly connected to the FXS module as well as dialed-up on the FXO module.
> 
> I know you've read this before (I'm trying to get Digium's attention): 
> on all my machines I have CPU spiking.  The time between spikes varys, 
> depending on the motherboard, from four to ten seconds.  The spike 
> consumes about 30% of the CPU in "system" mode (not user) for one or two 
> seconds.  You can try it on your machine by watching 'vmstat 1'. During 
> the spike, the number of interrupts does not increase, meaning a 
> particular interrupt is being held for 20-30ms.  When I 'modprobe -r 
> wcfxs' the spikes stop and the system goes to 100% idle.

Yes, I'm well aware of the vmstat thing. In fact, I was playing around
with that again early this morning to reconfirm previous research.
It's obviously coming from either wctdm or zaptel, and not wcusb.

> I demonstrated this to Matt O. at Digium, and he confirmed that his 
> (Digium) machine was spiking.
> 
> Rich, thank you for your "clarification" post this morning--a very good 

Out of about six fax attempts in the past 48 hours, none have come
through reliably with spandsp (using a five year old fax machine).
Most received at least some part of the original fax before failing,
and it seems the failure point is not the same each time. So, best
guess remains that wctdm and/or zaptel is at the root of the problem.

I then tried the faxdetect=incoming option and directed the incoming
fax to that same five year old machine via an ata186 (on the same
wire), and it failed as well further suggesting a wctdm/zaptel problem.

Sure wish we could get this resolved....