[mythtv-users] Playback hung

[Michael T. Dean]

On 03/18/2014 08:17 AM, Will Dormann wrote:
> On 3/18/14, 12:53 AM, Dick Steffens wrote:
>> I was just trying to be polite and not overload the list with a long
>> log. The entire collection of events logged at 19:45:03 is included
>> below. I've started with the entry before it and ended with the entry
>> after it.
>>
>> I do appreciate your willingness to wade through this collection.
>
>
> Not a problem.   And your post wasn't too much for the list, IMO.   In
> cases where the full log might be required, I think the recommended
> route is to post to pastebin.
>
> In the case of a kernel panic (or other crash, really) the call trace is
> what tells you how you got there.  In your case:
>
>> Call Trace:
>> [<ffffffff811a8e4a>] ? traverse+0x5a/0x220
>> [<ffffffff811a9830>] ? seq_put_decimal_ll+0x60/0x60
>> [<ffffffff811a9b85>] seq_read+0x355/0x400
>> [<ffffffff811a9830>] ? seq_put_decimal_ll+0x60/0x60
>> [<ffffffff811e8202>] proc_reg_read+0x82/0xc0
>> [<ffffffff81187950>] vfs_read+0xb0/0x180
>> [<ffffffff81187a6a>] sys_read+0x4a/0x90
>> [<ffffffff816a6a69>] system_call_fastpath+0x16/0x1b
> The other things like register values aren't quite as useful.  The way
> that I interpret the above is that something failed filesystem-related.

However this:

>> Mar 16 19:45:03 ThinkCentre-M58p kernel: [1648447.398228] Process 
>> irqbalance (pid: 1080, threadinfo ffff880079474000, task 
>> ffff8800363bae00) 

seems to indicate it's something related to hardware device access 
(possibly even BIOS-level/ACPI type stuff)--though I'll be the first to 
admit I'm not a kernel/low-level-system guru.  It may be a bug in 
irqbalance, or it may just be your system/firmware does things in a way 
that gave irqbalance problems.  (Oh, and FWIW, during playback of video, 
you'll have lots of IRQ's in play, so you may just be stressing 
irqbalance more than any typical tester.)

BTW, this isn't running in a Xen environment, is it?  Just asking 
because a simple (meaning I didn't spend much time on it, rather than 
saying it's something you should have done) seems there are lots of 
irqbalance NX-bit kernel oopses when using Xen.

Mike