IP0X Bugs

The IP0X hardware and software is very reliable. There are thousands of IP0X devices in the field and over 10,000 analog ports deployed. However there are a few issues with a small number of IP0X installations. As the software and hardware is very open, the IP0X community is working together to close out the few remaining issues below. You are welcome to contribute.

Outstanding Issues

  1. MMC card support is still experimental and not enabled on the IP04s shipped from the Free Telephony Project Store. However with 256 MByte of internal flash this is not a practical problem for most applications. See the IP0X How To for more information.
  2. The IP08 USB port currently does not work. We are investigating this issue. Probably a CPLD firmware bug – USB worked fine on earlier IP08s.
  3. On a small number of IP04s there have been reports of FXO ports locking up after 1-14 days. In this state incoming calls and outgoing calls are not possible, and the IP04 needs to be rebooted or the wcfxs driver unloaded and loaded. We are investigating this bug. The work around is a cron job to unload/load the wcfxs module every day (see forum for scripts). The vast majority of IP04 installations do not experience this problem, which seems to be site related, perhaps due to noisey power or phone line noise. For more information please contact Rowetel.
  4. There have been a few popped FXS modules reported. This seems to be linked with ringing multiple FXS ports simultaneously, and/or non-standard power supplies.

There are two known network issues that affect a small number of IP0X devices:

  1. Sometimes the DM9000 Ethernet chip will fail on the 10/100M auto-negotiation with certain switches. In this case the network link will not come up until you unplug/replug the RJ45 cable or reboot the device. Atcom have developed a fix based on an upgrade to the DM9000 driver. You can test this fix by installing the Atcom firmware.
  2. Keep alive issue: The IP0X will stop responding to the network, however if you send a ping from the IP0X (e.g. using the serial console), the network will come up again. This appears to be a uClinux issue, as it also occurs on Blackfin IP-PBX devices that do not use the DM9000 Ethernet chip.

    On some networks this problem can be induced by unplugging the Ethernet cable for 15 minutes, then reconnecting the cable.

    A reliable work around for this problem is to ping your gateway every few seconds. A ping command running in the background can be started using a line in your /etc/init.d network start-up script.

Gotchas

Some subtle problems that people have experienced in the field, and how to fix them:

  1. Your IP0X works fine for several months, then one day the network won’t come up and no phone functions work. I have seen this problem caused twice due to the file system filling up. In the first case it was due to very verbose asterisk logging, in the second case a large audio file.
  2. Connect the RS232 serial cable and wait for the root prompt. Then check to see if the file system is full:
    root:~> df -h
    Filesystem Size Used Available Use% Mounted on
    /dev/mtdblock2 248.0M 20.4M 227.6M 8% /

    You can track down and delete the large files using “du -h”

  3. Your IP0X doesn’t boot so you connect the RS232 serial cable and it boots OK. You disconnect the RS232 cable and it doesn’t boot any more. Not sure why this one happens but in the three times I have seen it the fix is to reflash your IP0X with the firmware of your choice.
  4. When your IP0X boots you see messages like this in the dmesg boot log or serial console:
    --snip--
    block 791 is bad
    --snip--

    i.e. one of more bad blocks reported as the Yaffs file system is mounted. This is completely normal – the NAND flash chips often have a few bad blocks when they are manufactured and the file system works around them.