Measure the Accept Queueing Time

[Andi Kleen]

On Friday 17 February 2006 17:15, Frank Ch. Eigler wrote:
> Hi -
> 
> ak wrote:
> > [...]
> > > But the two probes must have some connection in order for that time
> > > difference to make sense.  tid() ?
> > 
> > Doesn't work for nested interrupts. 
> 
> Could you outline a problematic scenario more fully?


interrupt 1
calls function to be measured
startprobe 
... function runs....
  interrupt 2 from a different device 
      calls function to be measured again
      startprobe nested
      ... function runs nested ... 
      endprobe nested
  interrupt 2 ends
... function still runs ...
endprobe

x86-64 has a per cpu interrupt nesting count, but even with that
and using (tid, count) as index there are some problems.

I hit this when trying to measure dma_map_sg()

-Andi