Re: automatic stacktrace on panic, was: Re: Boot fail as kvm guest

[Marc Balmer <mbalmer%NetBSD.org@localhost>]

Am 17.02.12 14:37, schrieb Paul Goyette:
> On Fri, 17 Feb 2012, Marc Balmer wrote:
> 
>> Am 17.02.12 14:16, schrieb Hubert Feyrer:
>>> On Fri, 17 Feb 2012, Marc Balmer wrote:
>>>>> I think NetBSD should add a stack trace on panics automatically...
>>>>
>>>> Can't that be achieved using ddb.commandonenter ?
>>>> Or should a ddb.stacktrace sysctl variable be introduced for this?
>>>
>>> Whatever one, it should be on by default.
>>> Each time we have to ask people to give us information, this is
>>> suboptimal.
>>
>> So what is the best way (I am now collecting opinions)?
>>
>> 1) Introduce a new sysctl ddb.stacktrace, which is on by default
>> 2) Set ddb.commandonenter to 'bt' by default
>> 3) Do it like OpenBSD and yell at the users in an all-capslock message
> 
> Can we also get the per-CPU data by default?

Well, the output will get quite massive.  In the end we would need a
backtrace of each cpu, plus the output of ps.

Would just the backtrace be an acceptable compromise or do we need to go
further, like ps etc?

Maybe we need a ddb.fullinfo sysctl, if set to 1 (the default) it will
read you the whole bible on entering ddb...