[asterisk-dev] Why was bug 5126 closed as a "Tech Support issue"?

Tony Mountifield tony at softins.clara.co.uk
Fri Feb 10 03:01:39 MST 2006


Why was bug 5126 closed by drumkilla as a "Tech Support issue"?

I am currently investigating the same issue, and it is clearly an
incompatibility between the 2.6 kernel and the way ztdynamic/ztd-eth
are written. It's definitely not just a user configuration issue.

So I don't understand the response to that bug. Please could someone explain?

The problem is that dev_queue_xmit(skb) is called while interrupts
are disabled, due to "spin_lock_irqsave(&dlock, flags);". This spinlock
is in effect to protect the traversal of the span list.

Fabio Ferarri posted a suggested solution to the list at
http://lists.digium.com/pipermail/asterisk-dev/2005-November/016692.html
but the problem with that solution is that it moves ethernet specific
code up into the generic ztdynamic layer.

I'm still trying to work out a better solution, perhaps by adding a
"flush" method to the driver, but still can't get away from the need
to traverse the span list. I'm wondering whether the read and write
spinlocks might be appropriate, since I might be able to avoid having
to disable interrupts. Any kernel experts here?

I can't re-open bug 5126 because it is closed and I was not the reporter.
If I raised another bug about it, would it too get summarily closed and me
awarded -2 karma?

Cheers
Tony

-- 
Tony Mountifield
Work: tony at softins.co.uk - http://www.softins.co.uk
Play: tony at mountifield.org - http://tony.mountifield.org



More information about the asterisk-dev mailing list