Re: [RFC PATCH 1/3] Unified trace buffer

[Steven Rostedt]

On Wed, 24 Sep 2008, Martin Bligh wrote:

> > On Wed, 24 Sep 2008, Martin Bligh wrote:
> >>
> >> If we use 32 bits instead of 27, then the timestamp events are only
> >> about once per second, which is probably fine for overhead ... ?
> >
> > You'd have them ONCE IN A BLUE MOON.
> >
> > If there is nothing going on, you don't need the timestamps at all.
> 
> Yeah, you're right - we can just mark it dirty, and 'pre-log' the timestamp
> events when someone calls a reserve and we haven't logged anything
> for more time than we can store. Did not think of that. Was only 5 bits
> for us, not an extra 37, but still, is much better.
> 
> Is a 5-bit event id generic enough though?
> 

Actually, I was keeping the event id completely out of the ring buffer and 
let a higher layer deal with that. For padding, I just made the length 
field zero.

For overflows of the timestamp, we can reserve the -1 timestamp as a 
trigger to read the tsc again and put the full 64 bits into the record.

Just an idea.

-- Steve

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/