| Message ID | 20210617171611.80542-1-jacopo+renesas@jmondi.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | media: Introduce post_register() subdev operation | expand |
On 17/06/2021 19:16, Jacopo Mondi wrote: > Hello Hans, > this is the result of the discussion we had yesterday, sent out just to > have a taste of how it looks like. > > I won't pretend I like the outcome: it feels a bit hackish and meant to support > this precise use case. > > Compared to the proposal to resurect 'init()' it indeed has the advantage that > the subdevice driver works in both deffered and non-deferred mode, but the > notifier flags seems really custom. > > Also, being the new flag part of the notifier it won't be available for i2c > subdevs. > > What do you think ? Does the result match your understanding ? That's what I came up with, yes. I think some of the names can be improved, but otherwise the mechanism is what I had in mind. Regards, Hans > > Thanks > j > > Jacopo Mondi (4): > media: v4l2-subdev: Introduce post_register() core op > media: v4l2-async: Add notifier flags > media: v4l2-async: Call post_register() subdev op > media: i2c: gmsl: Defer camera intialization > > drivers/media/i2c/max9286.c | 21 ++++++-- > drivers/media/i2c/rdacm20.c | 81 ++++++++++++++++------------ > drivers/media/v4l2-core/v4l2-async.c | 11 ++++ > include/media/v4l2-async.h | 10 ++++ > include/media/v4l2-subdev.h | 3 ++ > 5 files changed, 89 insertions(+), 37 deletions(-) > > -- > 2.31.1 >
Hi Hans, On Fri, Jun 18, 2021 at 12:21:43PM +0200, Hans Verkuil wrote: > On 17/06/2021 19:16, Jacopo Mondi wrote: > > Hello Hans, > > this is the result of the discussion we had yesterday, sent out just to > > have a taste of how it looks like. > > > > I won't pretend I like the outcome: it feels a bit hackish and meant to support > > this precise use case. > > > > Compared to the proposal to resurect 'init()' it indeed has the advantage that > > the subdevice driver works in both deffered and non-deferred mode, but the > > notifier flags seems really custom. > > > > Also, being the new flag part of the notifier it won't be available for i2c > > subdevs. > > > > What do you think ? Does the result match your understanding ? > > That's what I came up with, yes. I think some of the names can be improved, > but otherwise the mechanism is what I had in mind. Great then! Do you have any suggestions on names that can help moving forward ? Thanks j > > Regards, > > Hans > > > > > Thanks > > j > > > > Jacopo Mondi (4): > > media: v4l2-subdev: Introduce post_register() core op > > media: v4l2-async: Add notifier flags > > media: v4l2-async: Call post_register() subdev op > > media: i2c: gmsl: Defer camera intialization > > > > drivers/media/i2c/max9286.c | 21 ++++++-- > > drivers/media/i2c/rdacm20.c | 81 ++++++++++++++++------------ > > drivers/media/v4l2-core/v4l2-async.c | 11 ++++ > > include/media/v4l2-async.h | 10 ++++ > > include/media/v4l2-subdev.h | 3 ++ > > 5 files changed, 89 insertions(+), 37 deletions(-) > > > > -- > > 2.31.1 > > >