RE: [PATCH 3/3] v4l: Set proper timestamp type in selected drivers which use videobuf2

[Kamil Debski <k.debski@samsung.com>]

Hi,

One more comment. Also adding videobuf2 maintainers to CC, as they were not
added in the patch "v4l: Tell user space we're using monotonic timestamps".

> From: Sakari Ailus [mailto:sakari.ailus@iki.fi]
> Sent: Saturday, January 19, 2013 6:43 PM
> 
> Hi Kamil,
> 
> Thanks for the patch.
> 
> On Mon, Jan 14, 2013 at 10:36:04AM +0100, Kamil Debski wrote:
> > Set proper timestamp type in drivers that I am sure that use either
> > MONOTONIC or COPY timestamps. Other drivers will correctly report
> > UNKNOWN timestamp type instead of assuming that all drivers use
> > monotonic timestamps.
> 
> What other kind of timestamps there can be? All drivers (at least those
> not
> mem-to-mem) that do obtain timestamps using system clock use monotonic
> ones.
> 
> I'd think that there should no longer be any drivers using the UNKNOWN
> timestamp type: UNKNOWN is either from monotonic or realtime clock, and
> we just replaced all of them with the monotonic ones. No driver uses
> realtime timestamps anymore.
> 
> How about making MONOTONIC timestamps default instead, or at least
> assigning all drivers something else than UNKNOWN?

In addition to my previous comment - making MONOTONIC default in videobuf2
would be inconsistent. For non videobuf2 drivers the default is UNKNOWN, so
making
MONOTONIC default for videobuf2 drivers is, in my opinion, wrong.

Best wishes,
Kamil Debski