Unified tracing buffer

[Masami Hiramatsu]

Hi Martin,

Martin Bligh wrote:
>>            log event c
>> hit event b
>> log event b
>>            log event d
>>
>> so, I think if we really need to order events, we have to stop
>> irq right after hitting an event.
> 
> How could you fix that in any practical way?

In that case, I think it's easy to know that event c is irq
related event by checking source code. :-)
And practically, in most cases, I think we can presume what
happened by event arguments and subsequent events.

Anyway, since there must be a delay between hitting event(and
as linus said, happening event) and logging it, I think buffering
mechanism itself should focus on ensuring only writing and
reading order.

Thank you,

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com