OMAP3 ISP and camera drivers (update 2)

OMAP3 ISP and camera drivers (update 2)

[Sakari Ailus]

Hi,

I've again updated the patchset in Gitorious after a long break. It's
here. The base is fairly recent linux-omap (May) but I wouldn't expect
problems in rebasing on top of newer updates either.

<URL:http://www.gitorious.org/projects/omap3camera>

The amount of changes is more or less huge but I'll try to summarise
them. The base branch is no longer needed, the patch has been integrated
to linux-omap. The v4l2_subdev transition hasn't begun yet, however.

- Many ISP subdrivers have been rewritten or refactored. The new code
should be easier to understand.

- VIDIOC_TRY_FMT has no longer have side effects except perhaps to the
resizer. This is being worked on.

- Crop has been mostly rewritten.

- Locking has been corrected, although probably not definitely fixed.

- A separate ispstat module for handling the H3A, AF and HIST 
statistics. H3A and AF are using it already.

- Lots of redundant code has been removed.

- Most busy-locked register are should be no longer updated when 
corresponding modules are busy. There are still some cases this is 
happening, though.

- Configuration of the modules in the interrupt handler is done so that 
the module is disabled first or used in oneshot mode.

- Lots of things I can't remember now. The individual changes can be 
seen in the omap3isp and omap34xxcam branches. The branches just contain 
the patches in order so git diff doesn't help, unfortunately.

I won't be available for questions for a month or so (holidays). In the 
meantime you can contact Tuukka Toivonen, David Cohen and Sergio Aguirre 
for questions.

-- 
Sakari Ailus
sakari.ailus@maxwell.research.nokia.com

Re: OMAP3 ISP and camera drivers (update 2)

[Dongsoo, Nathaniel Kim]

Hello Sakari,

2009/6/19 Sakari Ailus <sakari.ailus@maxwell.research.nokia.com>:
> Hi,
>
> I've again updated the patchset in Gitorious after a long break. It's
> here. The base is fairly recent linux-omap (May) but I wouldn't expect
> problems in rebasing on top of newer updates either.
>
> <URL:http://www.gitorious.org/projects/omap3camera>
>
> The amount of changes is more or less huge but I'll try to summarise
> them. The base branch is no longer needed, the patch has been integrated
> to linux-omap. The v4l2_subdev transition hasn't begun yet, however.
>
> - Many ISP subdrivers have been rewritten or refactored. The new code
> should be easier to understand.
>
> - VIDIOC_TRY_FMT has no longer have side effects except perhaps to the
> resizer. This is being worked on.
>
> - Crop has been mostly rewritten.
>
> - Locking has been corrected, although probably not definitely fixed.
>
> - A separate ispstat module for handling the H3A, AF and HIST statistics.
> H3A and AF are using it already.
>
> - Lots of redundant code has been removed.
>
> - Most busy-locked register are should be no longer updated when
> corresponding modules are busy. There are still some cases this is
> happening, though.
>
> - Configuration of the modules in the interrupt handler is done so that the
> module is disabled first or used in oneshot mode.
>
> - Lots of things I can't remember now. The individual changes can be seen in
> the omap3isp and omap34xxcam branches. The branches just contain the patches
> in order so git diff doesn't help, unfortunately.
>
> I won't be available for questions for a month or so (holidays). In the
> meantime you can contact Tuukka Toivonen, David Cohen and Sergio Aguirre for
> questions.
>
> --
> Sakari Ailus
> sakari.ailus@maxwell.research.nokia.com
>
>
>

By the way, it's quite tough doing a code review without a patch
in-lined with e-mail.
Anyway, I took a quick look at the gitorious repository and found
something strange.
Following patch.
http://www.gitorious.org/omap3camera/mainline/commit/d92c96406296310a977b00f45b209523929b15b5
What happens to the capability when the int device is dummy? (does it
mean that there is no int device?)

And the most thing that makes me cautious to review the patch is all
about v4l2 subdev thing. Because most device drivers in V4L2
repository already got started moving to subdev framework.

And one more thing. If I want to test how the "ISP" driver is working,
is there any target board that I can buy also a sensor device already
attached on it? If anybody knows that, please let me know.
Cheers,

Nate




-- 
=
DongSoo, Nathaniel Kim
Engineer
Mobile S/W Platform Lab.
Digital Media & Communications R&D Centre
Samsung Electronics CO., LTD.
e-mail : dongsoo.kim@gmail.com
          dongsoo45.kim@samsung.com

Re: OMAP3 ISP and camera drivers (update 2)

[Tuukka.O Toivonen]

On Saturday 20 June 2009 12:05:13 ext Dongsoo, Nathaniel Kim wrote:
> Following patch.
> http://www.gitorious.org/omap3camera/mainline/commit/d92c96406296310a977b00f45b209523929b15b5
> What happens to the capability when the int device is dummy? (does it
> mean that there is no int device?)

Yes, when the int device is dummy, there is no such a device.
For example, when vdev->vdev_sensor == v4l2_int_device_dummy()
it means that the device has no sensor.

In that case, obviously, the device is not capable of capturing
or streaming.

> And one more thing. If I want to test how the "ISP" driver is working,
> is there any target board that I can buy also a sensor device already
> attached on it? 

I think that TI probably has some boards for sale, you
could take a look at their web pages.

- Tuukka

Re: OMAP3 ISP and camera drivers (update 2)

[Dongsoo Kim]

2009. 06. 22, 오후 10:52, Tuukka.O Toivonen 작성:

> On Saturday 20 June 2009 12:05:13 ext Dongsoo, Nathaniel Kim wrote:
>> Following patch.
>> http://www.gitorious.org/omap3camera/mainline/commit/d92c96406296310a977b00f45b209523929b15b5
>> What happens to the capability when the int device is dummy? (does it
>> mean that there is no int device?)
>
> Yes, when the int device is dummy, there is no such a device.
> For example, when vdev->vdev_sensor == v4l2_int_device_dummy()
> it means that the device has no sensor.
>
> In that case, obviously, the device is not capable of capturing
> or streaming.

OK, what I'm afraid is that even though the device could be opened and  
recognized as a v4l2 device but has no capability should be weird.  
Actually I'm not sure about this case is spec-in or not.
In my opinion it should be better when the camera interface (or ISP)  
has no int device (or subdev) attahced on it, no device node mounted  
in /dev or returning ENODEV. But before that, I'm very curious about  
why you made in that way.

>
>> And one more thing. If I want to test how the "ISP" driver is  
>> working,
>> is there any target board that I can buy also a sensor device already
>> attached on it?
>
> I think that TI probably has some boards for sale, you
> could take a look at their web pages.
>
> - Tuukka

Thank you I'll try to find on their web site :-)
Cheers,

Nate

==========================
Dong Soo, Kim
Engineer
Mobile S/W Platform Lab.
Telecommunication R&D Centre
Samsung Electronics CO., LTD.
e-mail : dongsoo.kim@gmail.com
            dongsoo45.kim@samsung.com
==========================

RE: OMAP3 ISP and camera drivers (update 2)

[Aguirre Rodriguez, Sergio Alberto]

> -----Original Message-----
> From: linux-media-owner@vger.kernel.org [mailto:linux-media-
> owner@vger.kernel.org] On Behalf Of Tuukka.O Toivonen
> Sent: Monday, June 22, 2009 8:52 AM
> To: ext Dongsoo, Nathaniel Kim
> Cc: Sakari Ailus; linux-media@vger.kernel.org; Aguirre Rodriguez, Sergio
> Alberto; Hiremath, Vaibhav; Koskipaa Antti (Nokia-D/Helsinki); Cohen
> David.A (Nokia-D/Helsinki); Alexey Klimov; gary@mlbassoc.com
> Subject: Re: OMAP3 ISP and camera drivers (update 2)
> 
> On Saturday 20 June 2009 12:05:13 ext Dongsoo, Nathaniel Kim wrote:
> > Following patch.
> >
> http://www.gitorious.org/omap3camera/mainline/commit/d92c96406296310a977b0
> 0f45b209523929b15b5
> > What happens to the capability when the int device is dummy? (does it
> > mean that there is no int device?)
> 
> Yes, when the int device is dummy, there is no such a device.
> For example, when vdev->vdev_sensor == v4l2_int_device_dummy()
> it means that the device has no sensor.
> 
> In that case, obviously, the device is not capable of capturing
> or streaming.
> 
> > And one more thing. If I want to test how the "ISP" driver is working,
> > is there any target board that I can buy also a sensor device already
> > attached on it?
> 
> I think that TI probably has some boards for sale, you
> could take a look at their web pages.

Hi Nate,

I'm currently rebasing these patches on top of latest Kevin's PM tree, and trying to make 3430SDP (MT9P012 and OV3640), Zoom1 and Zoom2 (not there yet, but in the works) sensors to work in there.

You can find this tree on:

http://dev.omapzoom.org/?p=saaguirre/linux-omap-camera.git;a=summary

Checkout devel branch.

That's my latest progress.

Regards,
Sergio
> 
> - Tuukka
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: OMAP3 ISP and camera drivers (update 2)

[Dongsoo Kim]

2009. 06. 22, 오후 11:01, Aguirre Rodriguez, Sergio Alberto 작성:

>> -----Original Message-----
>> From: linux-media-owner@vger.kernel.org [mailto:linux-media-
>> owner@vger.kernel.org] On Behalf Of Tuukka.O Toivonen
>> Sent: Monday, June 22, 2009 8:52 AM
>> To: ext Dongsoo, Nathaniel Kim
>> Cc: Sakari Ailus; linux-media@vger.kernel.org; Aguirre Rodriguez,  
>> Sergio
>> Alberto; Hiremath, Vaibhav; Koskipaa Antti (Nokia-D/Helsinki); Cohen
>> David.A (Nokia-D/Helsinki); Alexey Klimov; gary@mlbassoc.com
>> Subject: Re: OMAP3 ISP and camera drivers (update 2)
>>
>> On Saturday 20 June 2009 12:05:13 ext Dongsoo, Nathaniel Kim wrote:
>>> Following patch.
>>>
>> http://www.gitorious.org/omap3camera/mainline/commit/d92c96406296310a977b0
>> 0f45b209523929b15b5
>>> What happens to the capability when the int device is dummy? (does  
>>> it
>>> mean that there is no int device?)
>>
>> Yes, when the int device is dummy, there is no such a device.
>> For example, when vdev->vdev_sensor == v4l2_int_device_dummy()
>> it means that the device has no sensor.
>>
>> In that case, obviously, the device is not capable of capturing
>> or streaming.
>>
>>> And one more thing. If I want to test how the "ISP" driver is  
>>> working,
>>> is there any target board that I can buy also a sensor device  
>>> already
>>> attached on it?
>>
>> I think that TI probably has some boards for sale, you
>> could take a look at their web pages.
>
> Hi Nate,
>
> I'm currently rebasing these patches on top of latest Kevin's PM  
> tree, and trying to make 3430SDP (MT9P012 and OV3640), Zoom1 and  
> Zoom2 (not there yet, but in the works) sensors to work in there.

Thank you Sergio. So you mean that I can buy OMAP Zoom target board  
with MT or OV sensor on it sooner or later? cool!

>
> You can find this tree on:
>
> http://dev.omapzoom.org/?p=saaguirre/linux-omap-camera.git;a=summary
>
> Checkout devel branch.
>
> That's my latest progress.

OK I'll try to look at the devel branch.
Cheers,

Nate

>
> Regards,
> Sergio
>>
>> - Tuukka
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux- 
>> media" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

RE: OMAP3 ISP and camera drivers (update 2)

[Aguirre Rodriguez, Sergio Alberto]

> -----Original Message-----
> From: Dongsoo Kim [mailto:dongsoo.kim@gmail.com]
> Sent: Monday, June 22, 2009 9:05 AM
> To: Aguirre Rodriguez, Sergio Alberto
> Cc: Tuukka.O Toivonen; Sakari Ailus; linux-media@vger.kernel.org;
> Hiremath, Vaibhav; Koskipaa Antti (Nokia-D/Helsinki); Cohen David.A
> (Nokia-D/Helsinki); Alexey Klimov; gary@mlbassoc.com
> Subject: Re: OMAP3 ISP and camera drivers (update 2)
> 
> 
> 2009. 06. 22, 오후 11:01, Aguirre Rodriguez, Sergio Alberto 작성:
> 
> >> -----Original Message-----
> >> From: linux-media-owner@vger.kernel.org [mailto:linux-media-
> >> owner@vger.kernel.org] On Behalf Of Tuukka.O Toivonen
> >> Sent: Monday, June 22, 2009 8:52 AM
> >> To: ext Dongsoo, Nathaniel Kim
> >> Cc: Sakari Ailus; linux-media@vger.kernel.org; Aguirre Rodriguez,
> >> Sergio
> >> Alberto; Hiremath, Vaibhav; Koskipaa Antti (Nokia-D/Helsinki); Cohen
> >> David.A (Nokia-D/Helsinki); Alexey Klimov; gary@mlbassoc.com
> >> Subject: Re: OMAP3 ISP and camera drivers (update 2)
> >>
> >> On Saturday 20 June 2009 12:05:13 ext Dongsoo, Nathaniel Kim wrote:
> >>> Following patch.
> >>>
> >>
> http://www.gitorious.org/omap3camera/mainline/commit/d92c96406296310a977b0
> >> 0f45b209523929b15b5
> >>> What happens to the capability when the int device is dummy? (does
> >>> it
> >>> mean that there is no int device?)
> >>
> >> Yes, when the int device is dummy, there is no such a device.
> >> For example, when vdev->vdev_sensor == v4l2_int_device_dummy()
> >> it means that the device has no sensor.
> >>
> >> In that case, obviously, the device is not capable of capturing
> >> or streaming.
> >>
> >>> And one more thing. If I want to test how the "ISP" driver is
> >>> working,
> >>> is there any target board that I can buy also a sensor device
> >>> already
> >>> attached on it?
> >>
> >> I think that TI probably has some boards for sale, you
> >> could take a look at their web pages.
> >
> > Hi Nate,
> >
> > I'm currently rebasing these patches on top of latest Kevin's PM
> > tree, and trying to make 3430SDP (MT9P012 and OV3640), Zoom1 and
> > Zoom2 (not there yet, but in the works) sensors to work in there.
> 
> Thank you Sergio. So you mean that I can buy OMAP Zoom target board
> with MT or OV sensor on it sooner or later? cool!

AFAIK, when you buy the Zoom Target platform, you can only have OV3640 sensor. BUT you could hack the board to include another sensor (Maybe consulting Logic people could clarify this).

In Zoom1, I'll be able just to test the OV3640 sensor, which is the one I have available here.

On 3430SDP, is where I do have MT9P012 sensor (5MP RAW sensor) connected in parallel, and an OV3640 (Smart sensor, but driver is using it as RAW sensor currently only) in CSI2 interface.

> 
> >
> > You can find this tree on:
> >
> > http://dev.omapzoom.org/?p=saaguirre/linux-omap-camera.git;a=summary
> >
> > Checkout devel branch.
> >
> > That's my latest progress.
> 
> OK I'll try to look at the devel branch.
> Cheers,
> 
> Nate
> 
> >
> > Regards,
> > Sergio
> >>
> >> - Tuukka
> >> --
> >> To unsubscribe from this list: send the line "unsubscribe linux-
> >> media" in
> >> the body of a message to majordomo@vger.kernel.org
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
> 

RE: OMAP3 ISP and camera drivers (update 2)

[Menon, Nishanth]

> -----Original Message-----
> From: linux-media-owner@vger.kernel.org [mailto:linux-media-
> owner@vger.kernel.org] On Behalf Of Aguirre Rodriguez, Sergio Alberto
> Sent: Monday, June 22, 2009 5:22 PM
> > Thank you Sergio. So you mean that I can buy OMAP Zoom target board
> > with MT or OV sensor on it sooner or later? cool!
> 
> AFAIK, when you buy the Zoom Target platform, you can only have OV3640
> sensor. BUT you could hack the board to include another sensor (Maybe
> consulting Logic people could clarify this).
> 
> In Zoom1, I'll be able just to test the OV3640 sensor, which is the one I
> have available here.
> 
> On 3430SDP, is where I do have MT9P012 sensor (5MP RAW sensor) connected
> in parallel, and an OV3640 (Smart sensor, but driver is using it as RAW
> sensor currently only) in CSI2 interface.
> 

Curious: Thought we had two sensors: OV3640[1] on zoom1 and a 8MP sensor on zoom2[2] -> am I wrong in saying that the connectors are compatible since both are CSI2[3]?

SDP3430[4] supports MT9p012(CPI) and ov3640(CSI2).. as long as someone can put a sensor with the right connectors and voltage checks, they should be "plug and play" - at least from a h/w perspective ;)

Regards,
Nishanth Menon
Ref:
[1] http://www.ovt.com/products/part_detail.php?id=26
[2] https://www.omapzoom.org/gf/project/omapzoom/wiki/?pagename=WhatIsZoom2 
[3] https://www.omapzoom.com/gf/project/omapandroid/mailman/?_forum_action=ForumMessageBrowse&thread_id=1912&action=ListThreads&mailman_id=22 
[4] http://focus.ti.com/general/docs/wtbu/wtbugencontent.tsp?templateId=6123&navigationId=12013&contentId=28741#sdp 

RE: OMAP3 ISP and camera drivers (update 2)

[Aguirre Rodriguez, Sergio Alberto]

> -----Original Message-----
> From: Menon, Nishanth
> Sent: Monday, June 22, 2009 9:50 AM
> To: Aguirre Rodriguez, Sergio Alberto; Dongsoo Kim
> Cc: Tuukka.O Toivonen; Sakari Ailus; linux-media@vger.kernel.org;
> Hiremath, Vaibhav; Koskipaa Antti (Nokia-D/Helsinki); Cohen David.A
> (Nokia-D/Helsinki); Alexey Klimov; gary@mlbassoc.com
> Subject: RE: OMAP3 ISP and camera drivers (update 2)
> 
> > -----Original Message-----
> > From: linux-media-owner@vger.kernel.org [mailto:linux-media-
> > owner@vger.kernel.org] On Behalf Of Aguirre Rodriguez, Sergio Alberto
> > Sent: Monday, June 22, 2009 5:22 PM
> > > Thank you Sergio. So you mean that I can buy OMAP Zoom target board
> > > with MT or OV sensor on it sooner or later? cool!
> >
> > AFAIK, when you buy the Zoom Target platform, you can only have OV3640
> > sensor. BUT you could hack the board to include another sensor (Maybe
> > consulting Logic people could clarify this).
> >
> > In Zoom1, I'll be able just to test the OV3640 sensor, which is the one
> I
> > have available here.
> >
> > On 3430SDP, is where I do have MT9P012 sensor (5MP RAW sensor) connected
> > in parallel, and an OV3640 (Smart sensor, but driver is using it as RAW
> > sensor currently only) in CSI2 interface.
> >
> 
> Curious: Thought we had two sensors: OV3640[1] on zoom1 and a 8MP sensor
> on zoom2[2] -> am I wrong in saying that the connectors are compatible
> since both are CSI2[3]?

You're not wrong. The connectors are compatible, it'll be a matter of the driver part located in board specific code to request the necessary datalanes for the transmission to be done adequately. But both should follow CSI2 standard spec.

> 
> SDP3430[4] supports MT9p012(CPI) and ov3640(CSI2).. as long as someone can
> put a sensor with the right connectors and voltage checks, they should be
> "plug and play" - at least from a h/w perspective ;)

Yes, that's correct.

Only proper sensor driver will be required of course...

> 
> Regards,
> Nishanth Menon
> Ref:
> [1] http://www.ovt.com/products/part_detail.php?id=26
> [2]
> https://www.omapzoom.org/gf/project/omapzoom/wiki/?pagename=WhatIsZoom2
> [3]
> https://www.omapzoom.com/gf/project/omapandroid/mailman/?_forum_action=For
> umMessageBrowse&thread_id=1912&action=ListThreads&mailman_id=22
> [4]
> http://focus.ti.com/general/docs/wtbu/wtbugencontent.tsp?templateId=6123&n
> avigationId=12013&contentId=28741#sdp

Re: OMAP3 ISP and camera drivers (update 2)

[Tuukka.O Toivonen]

On Monday 22 June 2009 17:00:53 ext Dongsoo Kim wrote:
> OK, what I'm afraid is that even though the device could be opened and  
> recognized as a v4l2 device but has no capability should be weird.  
> Actually I'm not sure about this case is spec-in or not.
> In my opinion it should be better when the camera interface (or ISP)  
> has no int device (or subdev) attahced on it, no device node mounted  
> in /dev or returning ENODEV. But before that, I'm very curious about  
> why you made in that way.

We had to be able to use other slave devices (eg. flash)
before attaching the actual camera module.

- Tuukka