[asterisk-users] DAHDI tor2 Trouble on Intel Atom (D510) ?

[Rusty Dekema]

Greetings,

I have a quad port Tormenta 2 PCI card (uses the tor2 driver), which I have
been successfully using for some time on an older, power-hungry Intel
machine. I recently moved it to a newer machine with an Intel Atom D510 CPU
in an effort to save on energy costs, but I cannot seem to get it to work
there.

Both machines are running Ubuntu 13.10 x64, with DAHDI support coming from
the Ubuntu 'dahdi', 'dahdi-dkms', and 'dahdi-linux' packages, versions
1:2.7.0-1ubuntu1, 1:2.5.0.1+dfsg-1ubuntu3, and 1:2.5.0.1+dfsg-1ubuntu3,
respectively. These package versions are identical on both machines.

On the new machine, the T1 card shows up in lspci and dahdi_hardware, and
the tor2 driver appears to load successfully, but none of the other DAHDI
tools appear to "see" any of the 4 T1 spans on the card.

Here is the 'lspci -vvv' output relevant to the card on the non-working
server:

    https://gist.github.com/DrDeke/9436688

Here is the output of dahdi_hardware on the non-working server:

pci:0000:03:00.0     tor2+        10b5:d00d Tormenta 2 Quad T1/PRI or E1/PRA

Here are the lines that show up in dmesg when loading the tor2 module on
the non-working server:

[ 1789.517727] Detected Tormenta 2 Quad T1/PRI or E1/PRA at
0xf6bff000/0xf6bfe800 irq 21
[ 1789.688024] Xilinx Chip successfully loaded, configured and started!!
[ 1789.688185] Tormenta 2 Quad E1/PRA Card
[ 1789.688463] Detected Card number: 255
[ 1789.688573] Registered Tormenta2 PCI

This differs from the output on the working server. On the working server,
you get the following:

[  133.359963] Detected Tormenta 2 Quad T1/PRI or E1/PRA at
0xd1000800/0xd1000000 irq 22
[  133.512015] Xilinx Chip successfully loaded, configured and started!!
[  133.512113] Tormenta 2 Quad T1/PRI Card
[  133.512270] Detected Card number: 0
[  133.512274] Tor2: Launching card: 0
[  133.515699] Registered Tormenta2 PCI

The fact that I do not get a "Tor2: Launching card" message on the
non-working server seems like a likely (at least intermediate) cause of the
problem. Does anyone have any idea what might be causing the card to get
detected as card 255 instead of card 0, and then not start? There are no
other telephony cards (or expansion cards of any kind) installed in the
non-working server.

If I run dahdi_genconf -vvv, I get:

Default parameters from /etc/dahdi/genconf_parameters
Generating /etc/dahdi/system.conf
Generating /etc/asterisk/dahdi-channels.conf

Subsequently running dahdi_cfg -vvv gives the following:

    https://gist.github.com/DrDeke/9436856

Examining the /etc/dahdi/system.conf that gets generated confirms that no
cards or spans were configured:

    https://gist.github.com/DrDeke/9436874

Running dahdi_tool shows only one span; "DAHDI_DUMMY/1 (source: HRtimer)",
in the "UNCONFIGURED" state.

If anyone has any ideas for me to try, or would like to see any further
debugging information, I would be very grateful if you let me know.

Thanks,
Rusty Dekema
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-users/attachments/20140308/097444fc/attachment.html>