[asterisk-users] tdm loosing interrupts and latency

Alex Samad alex at samad.com.au
Mon Jun 15 20:00:11 CDT 2009


Hi

I have come across a problem, with my tdp410 and soekris board
(basically pc on a chip amd geode cpu).

I am using the box as a firewall/asterisk box. The problem occurs when I
drop ppp and I get dead loop dectiotn going, I seem to lose interrupts
and get lots of messages in syslog from wctdm24xx saying missed
interrupt increasing latency

its out lined here
(http://forums.digium.com/viewtopic.php?p=126997&highlight=&sid=9de59f41f1a93ee8701b28fdd0cf6073)

Seems like the driver (and this is in zaptel & dadhi code), increases
latency by +1 until 30. and then the card seems to not work. In my case
I have seen latency increase from 8m (I have this as a starting point in
the module load) up to 17ms usually around here the fxs and fxo ports
stop working ..... I have to unload and then reload the module. bummer.


I can think of a couple of solutions 

1) build some intelligence to bring down the number when things are okay
2) build logic to say if a number is provided on module load to fix it
to that
3) add a sysfs (/proc) interface to allow changing this value on the fly

I could also try and solve my problem with the dead loop detection

 cat /proc/interrupts 
           CPU0       
  0:   23809265    XT-PIC-XT        timer
  1:          0    XT-PIC-XT        i8042
  2:          0    XT-PIC-XT        cascade
  4:        255    XT-PIC-XT        serial
  5:     459544    XT-PIC-XT        eth1
  8:          0    XT-PIC-XT        rtc0
 10:   95177163    XT-PIC-XT        wctdm24xxp0
 11:   28938443    XT-PIC-XT        eth0
 12:   28938632    XT-PIC-XT        eth3
 14:    3624228    XT-PIC-XT        ide0
 15:          1    XT-PIC-XT        ehci_hcd:usb1, ohci_hcd:usb2
NMI:          0   Non-maskable interrupts
LOC:          0   Local timer interrupts
TRM:          0   Thermal event interrupts
SPU:          0   Spurious interrupts
ERR:          0
MIS:          0


as you can see with the interrupts the wctdm24xxp0 is above eth0 (local
lan) and eth3 (my adsl)

eth1 is wireless and not heavily used


So any one had this problems, any other possible solution to this ?


How to engage digium to providing a fix for this ?

Alex

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: Digital signature
Url : http://lists.digium.com/pipermail/asterisk-users/attachments/20090616/f7de0312/attachment.pgp 


More information about the asterisk-users mailing list