Re: [PATCH 2/2 v3] pxa-camera: pixel format negotiation

[Robert Jarzmik <robert.jarzmik@free.fr>]

Guennadi Liakhovetski <g.liakhovetski@gmx.de> writes:

> I know this code repeats, and it is not nice. But as I was writing it I 
> didn't see another possibility. Or more precisely, I did see it, but I 
> couldn't compare the two versions well without having at least one of them 
> in code in front of my eyes:-) Now that I see it, I think, yes, there is a 
> way to do this only once by using a translation struct similar to what you 
> have proposed. Now _this_ would be a possibly important advantage, so it 
> is useful not _only_ for debugging:-) But we would have to extend it with 
> at least a buswidth. Like
>
> 	const struct soc_camera_data_format *cam_fmt;
> 	const struct soc_camera_data_format *host_fmt;
> 	unsigned char buswidth;
>
> Now this _seems_ to provide the complete information so far... In 
> pxa_camera_get_formats() we would
>
> 1. compute camera- and host-formats and buswidth
> 2. call pxa_camera_try_bus_param() to check bus-parameter compatibility
>
> and then in try_fmt() and set_fmt() just traverse the list of translation 
> structs and adjust geometry?

Hi Guennadi,

I began the work. I have a pending question here. Do you want to have the
translation structure fully contained into pxa_camera (in host_priv for
example), or do you want to replace the user formats by translation structure
(ie. soc_camera_init_user_formats() would generate a list of
soc_camera_format_translate instead of a list of soc_camera_data format) ?

I'm asking because in pxa_camera, there is no easy way to "guess" the size of
the array of translations. And as vmalloc() is done in
soc_camera_init_user_formats(), and allocates only soc_camera_data_format
structures, I see no easy way to generate the list of translations in
pxa_camera.c.

I thought I would modify soc_camera.c in this way :
static int soc_camera_init_user_formats(struct soc_camera_device *icd)
{
<snip>
-       icd->user_formats = vmalloc(sizeof(struct soc_camera_data_format *) *
-                                   fmts);
+       icd->user_formats =
+               vmalloc(sizeof(struct soc_camera_format_translate *) * fmts);
<snip>

Is that what you had in mind ? If not, how do you allocate the
soc_camera_format_translate array in pxa_camera.c ?

--
Robert

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request@redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list