Re: something's wrong

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

I've got similar behaviour on my 6.99.25 amd64 box - complete freeze
during building of Firefox 25 - easily repeatable. The system was
stable until 6.99.24, something happened later. Unmodified GENERIC
kernel, I couldn't break and get a dump, only hard reset brings it
back. I'll try the blind trigger suggestion tomorrow.

On the other hand, my build box (XEN3_DOMU under XenServer) gets
updated at the same time as the above box and continues to work as
before.

Chavdar

On 2 November 2013 14:43, Mindaugas Rasiukevicius <rmind%netbsd.org@localhost> 
wrote:
> Thomas Klausner <wiz%NetBSD.org@localhost> wrote:
>> <...>
>>
>> Please note that the machine doesn't panic -- otherwise I'd have seen
>> reboots instead.
>>
>> I'm regularly updating, so I think this problem is rather new (last
>> month or so). I was usually running with clang; using libc++ is new,
>> no idea if this is related or not (I guess not).
>>
>> Does anyone have any ideas about this?
>
> Just a few..
>
> Can you get into DDB during a lockup (blindly triggering it and typing
> call ddb_vgapost)?  If you can, inspecting the LWPs (with ps/l) to see
> what are they doing would be useful.
>
> During the glitches, you can try to inspect the LWPs using crash(8).
> Also, it would be useful to monitor vmstat -e counters, perhaps some
> interrupt rates increase to unusual levels.  When glitches start to
> happen, you might try to run constant snapshots of lockstat sleep 15,
> perhaps some lock contention would indicate what is going on.
>
>>
>> Thanks,
>>  Thomas
>
> --
> Mindaugas



-- 
----