* soc-camera: status, roadmap
@ 2009-06-10 16:45 Guennadi Liakhovetski
2009-06-10 20:09 ` Hans Verkuil
2009-06-17 13:05 ` Guennadi Liakhovetski
0 siblings, 2 replies; 8+ messages in thread
From: Guennadi Liakhovetski @ 2009-06-10 16:45 UTC (permalink / raw)
To: Linux Media Mailing List; +Cc: Hans Verkuil, Mauro Carvalho Chehab
Hi all
for those interested here's a (not so) short status report and a proposed
roadmap for general soc-camera development, and, of course, its ongoing
conversion to v4l2-subdev API.
1. v4l2-subdev conversion. I have posted several versions of the
conversion patch series to the list, of which the last takes an IMHO
correct approach of a graduate conversion, avoiding mega-patches,
modifying multiple platforms and drivers at once. With this approach the
roadmap consists of the following steps:
1.1. preparatory patch to soc-camera core, allowing parallel existence of
"legacy" (all in the mainline) platforms and converted platforms (pcm037
i.MX31 platform so far) by introducing some backwards compatibility code.
This patch is currently in v4l next and in linux-next, i.e., it is going
in with 2.6.31.
1.2. convert all (around 7) mainline platforms to the new layout. This
step is necessary for further conversions, but it depends on 1.1.
Therefore this can only be done later in 2.6.31 merge window, when 1.1. is
in the mainline.
1.3. convert soc-camera core and drivers to an intermediate state, with
which all cameras are registered by platforms as platform devices, later
soc-camera core probes them and dynamically registers respective i2c (or
other, as in soc_camera_platform.c case) devices. This patch depends on
1.2., and it is hard to expect to be able to push these three steps within
the 2.6.31 merge window. It could be possible, we could request to accept
this patch after -rc1, maybe we would be allowed to do this, but
1.4. this is the actual conversion to v4l2-subdev. It depends on some bits
and pieces in the v4l2-subdev framework, which are still in progress
(e.g. v4l2_i2c_new_subdev_board), I believe (Hans, am I right? or what's
the outcome of Mauro's last reply to you in the "[PULL]
http://www.linuxtv.org/hg/~hverkuil/v4l-dvb-subdev" thread?), so, it
becomes practically impossible to also pull it for 2.6.31.
Now, I do not want to have soc-camera in the intermediate 1.3. state for a
whole 2.6.31 kernel, which means, we have to postpone 1.3. and 1.4. until
2.6.32.
2. The above means, I'll have to maintain and update my patches for a
whole 2.6.31 development cycle. In this time I'll try to update and upload
them as a quilt patch series and announce it on the list a couple of
times.
3. This also means, development will become more difficult, new features
and drivers will only be accepted on the top of my patch stack, bugfixes
will have to be accpeted against the mainline, which then will mean extra
porting work for me.
4. In a message I posted a few minutes ago
http://www.mail-archive.com/linux-media@vger.kernel.org/msg06294.html
I'm asking about a correct interpretation of S_CROP and S_FMT operations.
I suspect, what soc-camera framework and all drivers thereunder are doing
is wrong, and have to be fixed rather sooner than later. However, I'd be
very much against fixing this in the present stack, because that would
mean a _lot_ of porting. So, this will remain standard-non-compliant until
2.6.32 too.
5. The conversion described in (1) is only partial, in its current form it
does not replace the existing soc-camera API between sensor drivers and
the soc-camera core with v4l2-subdev operations completely. Partly because
many of the current soc-camera methods are still missing in v4l2-subdev,
partly because it just makes more sense to first push the principle
conversion in the mainline, which at least removes soc-camera device
registration and switches to i2c driver autoloading and replaces some
trivially replaceable methods, like [gs]_fmt, [gs]_register, [gs]_control.
Some of the missing methods, like [gs]_crop should be easy to add, others,
like pixel format and bus parameter negotiations would require some
thinking and substantial work. Which makes this all some 2.6.33
material... but who wants to think that far...
6. As you see, this all looks like a lot of work, and I so far have been
doing all of this in my free time. But it would become difficult with
these amounts of work. So, I would welcome if either someone would step
forward to help with this work, or if some company would volunteer to
support this work financially.
Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-06-10 16:45 soc-camera: status, roadmap Guennadi Liakhovetski
@ 2009-06-10 20:09 ` Hans Verkuil
2009-06-10 21:49 ` Guennadi Liakhovetski
2009-06-11 0:31 ` Mauro Carvalho Chehab
2009-06-17 13:05 ` Guennadi Liakhovetski
1 sibling, 2 replies; 8+ messages in thread
From: Hans Verkuil @ 2009-06-10 20:09 UTC (permalink / raw)
To: Guennadi Liakhovetski; +Cc: Linux Media Mailing List, Mauro Carvalho Chehab
Hi Guennadi,
Let me start with a big Thank You! for all your hard work on this! Much
appreciated!
On Wednesday 10 June 2009 18:45:36 Guennadi Liakhovetski wrote:
> Hi all
>
> for those interested here's a (not so) short status report and a proposed
> roadmap for general soc-camera development, and, of course, its ongoing
> conversion to v4l2-subdev API.
>
> 1. v4l2-subdev conversion. I have posted several versions of the
> conversion patch series to the list, of which the last takes an IMHO
> correct approach of a graduate conversion, avoiding mega-patches,
> modifying multiple platforms and drivers at once. With this approach the
> roadmap consists of the following steps:
>
> 1.1. preparatory patch to soc-camera core, allowing parallel existence of
> "legacy" (all in the mainline) platforms and converted platforms (pcm037
> i.MX31 platform so far) by introducing some backwards compatibility code.
> This patch is currently in v4l next and in linux-next, i.e., it is going
> in with 2.6.31.
>
> 1.2. convert all (around 7) mainline platforms to the new layout. This
> step is necessary for further conversions, but it depends on 1.1.
> Therefore this can only be done later in 2.6.31 merge window, when 1.1.
> is in the mainline.
>
> 1.3. convert soc-camera core and drivers to an intermediate state, with
> which all cameras are registered by platforms as platform devices, later
> soc-camera core probes them and dynamically registers respective i2c (or
> other, as in soc_camera_platform.c case) devices. This patch depends on
> 1.2., and it is hard to expect to be able to push these three steps
> within the 2.6.31 merge window. It could be possible, we could request to
> accept this patch after -rc1, maybe we would be allowed to do this, but
>
> 1.4. this is the actual conversion to v4l2-subdev. It depends on some
> bits and pieces in the v4l2-subdev framework, which are still in progress
> (e.g. v4l2_i2c_new_subdev_board), I believe (Hans, am I right? or what's
> the outcome of Mauro's last reply to you in the "[PULL]
> http://www.linuxtv.org/hg/~hverkuil/v4l-dvb-subdev" thread?), so, it
> becomes practically impossible to also pull it for 2.6.31.
I haven't seen a reaction yet from Mauro regarding my latest pull request: I
think it addresses all his concerns regarding existing functionality so I
actually hope this can be merged. It would help a lot with this and similar
efforts.
> Now, I do not want to have soc-camera in the intermediate 1.3. state for
> a whole 2.6.31 kernel, which means, we have to postpone 1.3. and 1.4.
> until 2.6.32.
>
> 2. The above means, I'll have to maintain and update my patches for a
> whole 2.6.31 development cycle. In this time I'll try to update and
> upload them as a quilt patch series and announce it on the list a couple
> of times.
>
> 3. This also means, development will become more difficult, new features
> and drivers will only be accepted on the top of my patch stack, bugfixes
> will have to be accpeted against the mainline, which then will mean extra
> porting work for me.
If there is anything I can do to help this along, please let me know. In
particular: what else besides the v4l2_i2c_new_subdev_board do you need? I
didn't have much time in the past few weeks, but things are more relaxed
now and I expect to be able to do a lot more in the coming weeks (fingers
crossed :-) ).
Regards,
Hans
> 4. In a message I posted a few minutes ago
>
> http://www.mail-archive.com/linux-media@vger.kernel.org/msg06294.html
>
> I'm asking about a correct interpretation of S_CROP and S_FMT operations.
> I suspect, what soc-camera framework and all drivers thereunder are doing
> is wrong, and have to be fixed rather sooner than later. However, I'd be
> very much against fixing this in the present stack, because that would
> mean a _lot_ of porting. So, this will remain standard-non-compliant
> until 2.6.32 too.
>
> 5. The conversion described in (1) is only partial, in its current form
> it does not replace the existing soc-camera API between sensor drivers
> and the soc-camera core with v4l2-subdev operations completely. Partly
> because many of the current soc-camera methods are still missing in
> v4l2-subdev, partly because it just makes more sense to first push the
> principle conversion in the mainline, which at least removes soc-camera
> device registration and switches to i2c driver autoloading and replaces
> some trivially replaceable methods, like [gs]_fmt, [gs]_register,
> [gs]_control. Some of the missing methods, like [gs]_crop should be easy
> to add, others, like pixel format and bus parameter negotiations would
> require some thinking and substantial work. Which makes this all some
> 2.6.33
> material... but who wants to think that far...
>
> 6. As you see, this all looks like a lot of work, and I so far have been
> doing all of this in my free time. But it would become difficult with
> these amounts of work. So, I would welcome if either someone would step
> forward to help with this work, or if some company would volunteer to
> support this work financially.
>
> Thanks
> Guennadi
> ---
> Guennadi Liakhovetski, Ph.D.
> Freelance Open-Source Software Developer
> http://www.open-technology.de/
--
Hans Verkuil - video4linux developer - sponsored by TANDBERG Telecom
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-06-10 20:09 ` Hans Verkuil
@ 2009-06-10 21:49 ` Guennadi Liakhovetski
2009-06-10 22:00 ` Hans Verkuil
2009-06-11 0:31 ` Mauro Carvalho Chehab
1 sibling, 1 reply; 8+ messages in thread
From: Guennadi Liakhovetski @ 2009-06-10 21:49 UTC (permalink / raw)
To: Hans Verkuil; +Cc: Linux Media Mailing List, Mauro Carvalho Chehab
On Wed, 10 Jun 2009, Hans Verkuil wrote:
> On Wednesday 10 June 2009 18:45:36 Guennadi Liakhovetski wrote:
[snip]
> > 3. This also means, development will become more difficult, new features
> > and drivers will only be accepted on the top of my patch stack, bugfixes
> > will have to be accpeted against the mainline, which then will mean extra
> > porting work for me.
>
> If there is anything I can do to help this along, please let me know. In
> particular: what else besides the v4l2_i2c_new_subdev_board do you need? I
> didn't have much time in the past few weeks, but things are more relaxed
> now and I expect to be able to do a lot more in the coming weeks (fingers
> crossed :-) ).
Thanks for your offer!
Well, yes, at least we could start with these three things:
1. s_crop, g_crop, cropcap. Would be nice if you could add them just
following the standard API.
2. bus parameter negotiation. I appreciate Murali's effort, but I don't
like re-inventing the wheel. So, we should now either bring his
implementation to support _at least_ all features so far supported in
soc-camera, or you could just copy the soc-camera implementation over,
just renaming functions and macros as appropriate.
3. pixel format negotiation. The easiest would be to also just copy it
over.
These 3 points would make the porting much easier. I'll certainly be happy
to answer any questions you might get working with soc-camera code.
Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-06-10 21:49 ` Guennadi Liakhovetski
@ 2009-06-10 22:00 ` Hans Verkuil
0 siblings, 0 replies; 8+ messages in thread
From: Hans Verkuil @ 2009-06-10 22:00 UTC (permalink / raw)
To: Guennadi Liakhovetski; +Cc: Linux Media Mailing List, Mauro Carvalho Chehab
On Wednesday 10 June 2009 23:49:23 Guennadi Liakhovetski wrote:
> On Wed, 10 Jun 2009, Hans Verkuil wrote:
> > On Wednesday 10 June 2009 18:45:36 Guennadi Liakhovetski wrote:
>
> [snip]
>
> > > 3. This also means, development will become more difficult, new
> > > features and drivers will only be accepted on the top of my patch
> > > stack, bugfixes will have to be accpeted against the mainline, which
> > > then will mean extra porting work for me.
> >
> > If there is anything I can do to help this along, please let me know.
> > In particular: what else besides the v4l2_i2c_new_subdev_board do you
> > need? I didn't have much time in the past few weeks, but things are
> > more relaxed now and I expect to be able to do a lot more in the coming
> > weeks (fingers crossed :-) ).
>
> Thanks for your offer!
>
> Well, yes, at least we could start with these three things:
>
> 1. s_crop, g_crop, cropcap. Would be nice if you could add them just
> following the standard API.
No problem.
> 2. bus parameter negotiation. I appreciate Murali's effort, but I don't
> like re-inventing the wheel. So, we should now either bring his
> implementation to support _at least_ all features so far supported in
> soc-camera, or you could just copy the soc-camera implementation over,
> just renaming functions and macros as appropriate.
Obviously all features should be supported, but as you can no doubt tell I'm
not sold (yet?) on the autonegotiation part of the soc-camera
implementation.
> 3. pixel format negotiation. The easiest would be to also just copy it
> over.
I haven't looked at this yet, but I will do so soon.
> These 3 points would make the porting much easier. I'll certainly be
> happy to answer any questions you might get working with soc-camera code.
Thanks,
Hans
--
Hans Verkuil - video4linux developer - sponsored by TANDBERG Telecom
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-06-10 20:09 ` Hans Verkuil
2009-06-10 21:49 ` Guennadi Liakhovetski
@ 2009-06-11 0:31 ` Mauro Carvalho Chehab
1 sibling, 0 replies; 8+ messages in thread
From: Mauro Carvalho Chehab @ 2009-06-11 0:31 UTC (permalink / raw)
To: Hans Verkuil; +Cc: Guennadi Liakhovetski, Linux Media Mailing List
Em Wed, 10 Jun 2009 22:09:07 +0200
Hans Verkuil <hverkuil@xs4all.nl> escreveu:
> > 1.4. this is the actual conversion to v4l2-subdev. It depends on some
> > bits and pieces in the v4l2-subdev framework, which are still in progress
> > (e.g. v4l2_i2c_new_subdev_board), I believe (Hans, am I right? or what's
> > the outcome of Mauro's last reply to you in the "[PULL]
> > http://www.linuxtv.org/hg/~hverkuil/v4l-dvb-subdev" thread?), so, it
> > becomes practically impossible to also pull it for 2.6.31.
>
> I haven't seen a reaction yet from Mauro regarding my latest pull request: I
> think it addresses all his concerns regarding existing functionality so I
> actually hope this can be merged. It would help a lot with this and similar
> efforts.
As Guennadi pointed, it seems that those patches were not yet properly
discussed at the ML.
Since they touch at the core of the subsystem, could you please submit they as
RFC, at linux-media?
Let's give people some time to review the newly added functions. After that,
you can send me V3 of your pull request.
Cheers,
Mauro
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-06-10 16:45 soc-camera: status, roadmap Guennadi Liakhovetski
2009-06-10 20:09 ` Hans Verkuil
@ 2009-06-17 13:05 ` Guennadi Liakhovetski
2009-07-03 18:59 ` Jonathan Cameron
1 sibling, 1 reply; 8+ messages in thread
From: Guennadi Liakhovetski @ 2009-06-17 13:05 UTC (permalink / raw)
To: Linux Media Mailing List; +Cc: Hans Verkuil, Mauro Carvalho Chehab
On Wed, 10 Jun 2009, Guennadi Liakhovetski wrote:
> 2. The above means, I'll have to maintain and update my patches for a
> whole 2.6.31 development cycle. In this time I'll try to update and upload
> them as a quilt patch series and announce it on the list a couple of
> times.
As promised, I just uploaded my current tree snapshot at
http://download.open-technology.de/soc-camera/20090617/
This is nothing remarkable, just my current patch-stack for those working
with the soc-camera framework. It is still based on a linux-next snapshot
of 07.05.2009 "history" branch. The exact commit on which the stack is
based is, as usual, in 0000-base. This is still based off 2.6.30-rc4, and
I expect to upgrade next time after 2.6.31-rc1.
Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-06-17 13:05 ` Guennadi Liakhovetski
@ 2009-07-03 18:59 ` Jonathan Cameron
2009-07-03 19:20 ` Guennadi Liakhovetski
0 siblings, 1 reply; 8+ messages in thread
From: Jonathan Cameron @ 2009-07-03 18:59 UTC (permalink / raw)
To: Guennadi Liakhovetski
Cc: Linux Media Mailing List, Hans Verkuil, Mauro Carvalho Chehab
Guennadi Liakhovetski wrote:
> On Wed, 10 Jun 2009, Guennadi Liakhovetski wrote:
>
>> 2. The above means, I'll have to maintain and update my patches for a
>> whole 2.6.31 development cycle. In this time I'll try to update and upload
>> them as a quilt patch series and announce it on the list a couple of
>> times.
>
> As promised, I just uploaded my current tree snapshot at
> http://download.open-technology.de/soc-camera/20090617/
> This is nothing remarkable, just my current patch-stack for those working
> with the soc-camera framework. It is still based on a linux-next snapshot
> of 07.05.2009 "history" branch. The exact commit on which the stack is
> based is, as usual, in 0000-base. This is still based off 2.6.30-rc4, and
> I expect to upgrade next time after 2.6.31-rc1.
>
Hi Guennadi,
I notice you've posted newer version of these patches on your website. 20090701/
I was wondering what tree these are based on?
I can't seem to track down the base commit.
Thanks,
Jonathan
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: soc-camera: status, roadmap
2009-07-03 18:59 ` Jonathan Cameron
@ 2009-07-03 19:20 ` Guennadi Liakhovetski
0 siblings, 0 replies; 8+ messages in thread
From: Guennadi Liakhovetski @ 2009-07-03 19:20 UTC (permalink / raw)
To: Jonathan Cameron
Cc: Linux Media Mailing List, Hans Verkuil, Mauro Carvalho Chehab
On Fri, 3 Jul 2009, Jonathan Cameron wrote:
> Guennadi Liakhovetski wrote:
> > On Wed, 10 Jun 2009, Guennadi Liakhovetski wrote:
> >
> >> 2. The above means, I'll have to maintain and update my patches for a
> >> whole 2.6.31 development cycle. In this time I'll try to update and upload
> >> them as a quilt patch series and announce it on the list a couple of
> >> times.
> >
> > As promised, I just uploaded my current tree snapshot at
> > http://download.open-technology.de/soc-camera/20090617/
> > This is nothing remarkable, just my current patch-stack for those working
> > with the soc-camera framework. It is still based on a linux-next snapshot
> > of 07.05.2009 "history" branch. The exact commit on which the stack is
> > based is, as usual, in 0000-base. This is still based off 2.6.30-rc4, and
> > I expect to upgrade next time after 2.6.31-rc1.
> >
> Hi Guennadi,
>
> I notice you've posted newer version of these patches on your website. 20090701/
> I was wondering what tree these are based on?
> I can't seem to track down the base commit.
It has been announced here:
http://thread.gmane.org/gmane.linux.drivers.video-input-infrastructure/7492/focus=7534
Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2009-07-03 19:20 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-06-10 16:45 soc-camera: status, roadmap Guennadi Liakhovetski
2009-06-10 20:09 ` Hans Verkuil
2009-06-10 21:49 ` Guennadi Liakhovetski
2009-06-10 22:00 ` Hans Verkuil
2009-06-11 0:31 ` Mauro Carvalho Chehab
2009-06-17 13:05 ` Guennadi Liakhovetski
2009-07-03 18:59 ` Jonathan Cameron
2009-07-03 19:20 ` Guennadi Liakhovetski
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.