Re: How how latent should non-preemptive scheduling be?

[Sitsofe Wheeler <sitsofe@yahoo.com>]

Ingo Molnar wrote:
> * Sitsofe Wheeler <sitsofe@yahoo.com> wrote:
> 
>> Ingo Molnar wrote:
>>> well, since they went away after you enabled CONFIG_PREEMPT=y, they are 
>>> definitely in-kernel latencies, not any external SMM latencies.
>>>
>>> I.e. they are inherently fixable. Could you enable:
>>>
>>>   CONFIG_DYNAMIC_FTRACE=y
>>>   CONFIG_FTRACE_MCOUNT_RECORD=y
>>>
>>> that should make the traces a lot more verbose - every kernel function  
>>> executed in the latency path will be logged. That way we'll be able to  
>>> say which one takes that long.
>> I do not appear to have the CONFIG_FTRACE_MCOUNT_RECORD option in 
>> 2.6.27rc7. Is it an option that is only in -tip ?
> 
> yeah - it's a new ftrace feature queued up for v2.6.28.

I've been struggling to boot -tip/master - currently it blows up just 
after printing SLUB information saying:

BUG: unable to handle kernel NULL pointer dereference at 00000004
IP: [<c0120078>] account_system_time+0x48/0x120
*pde = 00000000
Thread overran stack, or stack corrupted
Oops: 0002 [#1] PREEMPT

-- 
Sitsofe | http://sucs.org/~sits/