Summary:ASTERISK-03940: [patch] Conflict ztdummy with hardware modules
Reporter:Tilghman Lesher (tilghman)Labels:
Date Opened:2005-04-14 13:16:29Date Closed:2011-06-07 14:05:27
Versions:Frequency of
Environment:Attachments:( 0) 20050414__conflict_ztdummy.diff.txt
Description:This change to modules will permit hardware modules to refuse to load if ztdummy is already providing timing, and vice-versa, if a hardware device is already providing timing.


Disclaimer on file.  Needs a lot more testing prior to being committed.
Comments:By: Paul Cadach (pcadach) 2005-04-14 13:26:44

Which problems leads to load ztdummy together with other hardware timing sources?

By: ewieling (ewieling) 2005-04-14 18:09:51

For one thing ztdummy would generate an additional 1000 interrupts per second that is not needed.

By: Mark Spencer (markster) 2005-04-15 00:35:35

Why not just not load it?

By: Paul Cadach (pcadach) 2005-04-15 00:40:13

As I understand there is $50 bounty as announced at -dev list.

By: Serge Vecher (serge-v) 2005-04-15 08:29:13

Sorry if this is trivial, markster, but how do you not load ztdummy?
I have 8 E&M wink lines provisioned on a wct1xxp.

I start asterisk as follows after a system restart:
modprobe zaptel
modprobe wct1xxp
asterisk -Tvvvvvvvvc

When asterisk starts up, it will report that it configures lines 1-8 as E&M wink signalling _and_ loads ztdummy after that.

By: Matthew Fredrickson (mattf) 2005-04-15 11:27:20

Asterisk does not load kernel modules AT ALL.  It will not automatically load modules when it begins unless you have some sort of wrapper script that you're running around asterisk or before it is run (such as zaptel.init or something of that nature).

Aside from all that, I think that there was a fix applied to the init scripts (in HEAD at least) so that the init scripts do not load ztdummy if real zaptel hardware is found.

By: Serge Vecher (serge-v) 2005-04-15 23:19:46

Sorry about that; I've just restarted the server and got the log below. I was referring to the "pseudo zap channel", not ztdummy. They are not the same thing.

[Apr 16 00:09:47]   == Parsing '/etc/asterisk/zapata.conf': [Apr 16 00:09:47] Found
[Apr 16 00:09:47]     -- Registered channel 1, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 2, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 3, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 4, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 5, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 6, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 7, E & M Wink signalling
[Apr 16 00:09:47]     -- Registered channel 8, E & M Wink signalling
[Apr 16 00:09:47]     -- Automatically generated pseudo channel

By: ewieling (ewieling) 2005-04-15 23:26:13

The bounty is $25.

I created this bounty just to reduce the number is silly things that new users might do to cause issues.  It's not a BIG deal, but if you have Digium card drivers loaded you don't need ztdummy.  I've seen several posts to the mailing lists recently that have both ztdummy and Digium cards loaded.

By: Paul Cadach (pcadach) 2005-04-16 01:07:51

Ok, you have loaded Zaptel driver but not configure any spans on it, so you will not have interrupts from such card(s) (at least for TE410P). There is another big issue to manipulate timing circuits more carefully and have "global" clocks coming from most preferable span (including ztdummy if no other is available). I.e. such clocking module like ztdummy should have options to enable or to disable timing "on the fly", when span configuration changes.

By: Serge Vecher (serge-v) 2005-04-18 17:42:20

Paul, I assume your response was to my posting on 04/15.
Actually, zaptel is configured -- here is the section from zaptel.conf:


edited on: 04-18-05 17:46

By: Tilghman Lesher (tilghman) 2005-04-18 17:57:58

vechers:  please take questions of this nature to the -users mailing list.  It has nothing to do with this patchset.

By: Kevin P. Fleming (kpfleming) 2005-04-20 11:11:15

There is already support in the Zaptel init script in CVS HEAD to automatically load ztdummy _only_ if not Zap channels come up after loading the hardware drivers. Is there some reason why this is not adequate?

By: Matthew Fredrickson (mattf) 2005-04-21 15:45:20



By: twisted (twisted) 2005-05-10 23:41:55

I agree that the init script should be sufficient for this.