asterisk users - Feb 2005 - TDMOE + kernel badness

Anybody have any issues running tdmoe on kernel 2.6+?
I've got  Suse 9.1 + 9.2 running  2.6.5 and 2.6.8 respectively, and when I
enable dynamic spans between them, both boxes dump something similar to:

Badness in local_bh_enable at kernel/softirq.c:141
 [<c0120768>] local_bh_enable+0x48/0x60
 [<c02952b0>] dev_queue_xmit+0x230/0x240
 [<c02a0980>] eth_header+0x0/0x140
 [<f8f12301>] ztdeth_transmit+0x131/0x200 [ztd_eth]
 [<f8ef17ff>] __ztdynamic_run+0x20f/0x240 [ztdynamic]
 [<f8ef15e4>] ztd_tasklet+0x34/0x40 [ztdynamic]
 [<c0120928>] tasklet_hi_action+0x38/0x70
 [<c0120693>] __do_softirq+0x43/0xa0
 [<c0120716>] do_softirq+0x26/0x30
 [<c0108955>] do_IRQ+0xe5/0x120
 [<c0106cd8>] common_interrupt+0x18/0x20
 [<c02330ff>] lf+0x4f/0x60
 [<c0234265>] do_con_trol+0x95/0xbd0
 [<c02350c8>] do_con_write+0x328/0x760
 [<c0235aef>] con_put_char+0x2f/0x40
 [<c0229195>] opost+0xa5/0x1b0
 [<c022b171>] write_chan+0x1a1/0x230
 [<c0119c70>] default_wake_function+0x0/0x10
 [<c0119c70>] default_wake_function+0x0/0x10
 [<c02266a8>] tty_write+0xd8/0x1a0
 [<c022afd0>] write_chan+0x0/0x230
 [<c0151319>] do_readv_writev+0x169/0x210
 [<c02265d0>] tty_write+0x0/0x1a0
 [<c0151454>] vfs_writev+0x44/0x50
 [<c015150c>] sys_writev+0x3c/0x70
 [<c0105c69>] sysenter_past_esp+0x52/0x79

Both boxes are running current asterisk 1_0 stable cvs, Suse 9.2 has a
T100P hooked up to a Definity over PRI which works fine on its own.
Both boxes are also on the same ethernet segment and switch.
All googling has lead to similar kernel dumps mostly dealing with ppp.

On a different but similar note, I've set up an iax trunk between the two
machines which works fine except for error:
WARNING[24339]: chan_iax2.c:3044 iax2_trunk_queue: Maximum trunk data space
exceeded to x.x.x.x:4569
x.x.x.x being the ip address of the second asterisk server.
The iax trunk still works when the error pops up, but floods the console
repeatedly.


Googling results in the source code which seem to indicate some sort of out
of memory condition, yet the machine has plenty of memory and no swap used.

Any clues?

Rob

Robert.Kelly@ebimed.com wrote:
>Anybody have any issues running tdmoe on kernel 2.6+?
>I've got  Suse 9.1 + 9.2 running  2.6.5 and 2.6.8 respectively, and when
I
>enable dynamic spans between them, both boxes dump something similar to:
>
>Badness in local_bh_enable at kernel/softirq.c:141
> [<c0120768>] local_bh_enable+0x48/0x60
> [<c02952b0>] dev_queue_xmit+0x230/0x240
>  
>
Did you make any progress since your last post?

I am experiencing the same problem on Sarge with 2.6.10. When I ztcfg a 
dynamic span while asterisk is running, I get a segmentation fault and 
network access freezes completely. If I configure the spans before 
loading * I get the same errors as you do and my log files grow to 10GB 
in a matter of minutes. While that happens T1 spans lose synchronization 
and channels restart everywhere.

-- 
cheers,

Bartek Bulzak

RDE Solutions Inc. - Montreal, Canada

t 514-489-0620 x222
f 514-221-3106

"Badness in local_bh_enable at kernel/softirq.c on 2.6.X"

I'm seeing this on Kernel 2.6.+ implementations, namely Centos 4.1
machines while trying to do TDMoE trunks between two machines. 
2.4 Kernel operates fine on the same hardware....

I'm compiling zaptel-1.0.9.2 as per instructions in README.Linux26 +
README.udev

Here are some references where the issue has been reported before but
I've yet to find a documented solution; Any tips?

http://lists.digium.com/pipermail/asterisk-users/2005-February/091867.html

http://bugs.digium.com/view.php?id=5126


Thanks.