[PATCH 19/27] habanalabs: capture interrupt timestamp in handler

[Ofir Bitton]

On Thu, Feb 16, 2023 16:39 PM, Stanislaw Gruszka <stanislaw.gruszka at linux.intel.com> wrote:
> On Sun, Feb 12, 2023 at 10:44:46PM +0200, Oded Gabbay wrote:
> > From: Ofir Bitton <obitton at habana.ai>
> >
> > In order for interrupt timestamp to be more accurate we should capture
> > it during the interrupt handling rather than in threaded irq context.
> 
> Why this is important to have this timestamp more accurate ?

I agree that the time diff between taking the timestamp in the interrupt handler vs taking it in the
threaded irq context is negligible.

Having said that it is still important as we would like to have the same timestamp for events that finished together,
rather than having different timestamps when we process the events in the threaded irq handler.

> What actually 'more accurate' mean in this context ?
> 
> Regards
> Stanislaw

By 'more accurate' we mean closest to the MSIX interrupt.

Ofir.