* RFC: soc_camera, rcar_vin, and adv7604 @ 2014-07-09 16:42 Ian Molton 2014-07-09 20:34 ` Guennadi Liakhovetski 0 siblings, 1 reply; 4+ messages in thread From: Ian Molton @ 2014-07-09 16:42 UTC (permalink / raw) To: linux-media Cc: William Towle, mchehab, hans.verkuil, sylvester.nawrocki, g.liakhovetski, vladimir.barinov Hi folks, My colleague and I are trying to work out what to do to support the following combination: soc_camera + rcar_vin for capture, and the mainline adv7604 driver (which we have modified to successfully drive the adv7612). The problem we face is that the 7604 driver uses the new "pads" API, but soc_camera based drivers like rcar_vin do not. Obviously, there are a few approaches we could take, but we could use some guidance on this. One approach would be to bodge some non-pads older API support into the 7604 driver. This would probably be the easiest solution. A better approach might be to add pad API support to soc_camera, but it seems to me that the soc_camera API does not abstract away all of the areas that might need to be touched, which would lead to much pad-related churn in all the other soc_camera drivers. The codebase is rather large, and we're struggling to see a clear path through this. Whatever we do, we would like to be acceptable upstream, so we'd like to open a discussion. Perhaps a soc_camera2 with pads support? -- Ian Molton <ian.molton@codethink.co.uk> ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: RFC: soc_camera, rcar_vin, and adv7604 2014-07-09 16:42 RFC: soc_camera, rcar_vin, and adv7604 Ian Molton @ 2014-07-09 20:34 ` Guennadi Liakhovetski 2014-07-10 10:12 ` Ian Molton 0 siblings, 1 reply; 4+ messages in thread From: Guennadi Liakhovetski @ 2014-07-09 20:34 UTC (permalink / raw) To: Ian Molton Cc: linux-media, William Towle, mchehab, hans.verkuil, sylvester.nawrocki, vladimir.barinov Hi all, (disclaimer: I was away for the last two weeks with no proper access to my email, so, will be slowly fighting through my mail backlog, I know there have been a few mails re: soc-camera + DT etc., please, bear with me) On Wed, 9 Jul 2014, Ian Molton wrote: > Hi folks, > > My colleague and I are trying to work out what to do to support the > following combination: > > soc_camera + rcar_vin for capture, and the mainline adv7604 driver > (which we have modified to successfully drive the adv7612). > > The problem we face is that the 7604 driver uses the new "pads" API, but > soc_camera based drivers like rcar_vin do not. I'm not a huge expert in the pad-level API, but from what I understand there are at least two aspects to it: 1. subdevice pad-level API, which is just a newer, than the used by soc-camera subdevice API. 2. pad-level ioctl()s, exported to the user space, IIUC, as a part of the media controller API. For (2) - yes, soc-camera core support would need to be added, possibly with larger changes. Whereas for (1) it's only the host that has to use the pad-level API to access subdevice drivers. Interaction between camera host and subdevice drivers in soc-camera is direct, so, no soc-camera core modifications would be needed. Maybe we dould add some support, say, to help with (fake) file handles just to aid the transition. We might also add wrappers for subdev drivers, not implementing the pad-level API to avoid having host drivers support both APIs, but those are just convenience modifications. I might well be wrong. Please, correct me if so. Thanks Guennadi > Obviously, there are a few approaches we could take, but we could use > some guidance on this. > > One approach would be to bodge some non-pads older API support into the > 7604 driver. This would probably be the easiest solution. > > A better approach might be to add pad API support to soc_camera, but it > seems to me that the soc_camera API does not abstract away all of the > areas that might need to be touched, which would lead to much > pad-related churn in all the other soc_camera drivers. > > The codebase is rather large, and we're struggling to see a clear path > through this. Whatever we do, we would like to be acceptable upstream, > so we'd like to open a discussion. > > Perhaps a soc_camera2 with pads support? > > -- > Ian Molton <ian.molton@codethink.co.uk> > ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: RFC: soc_camera, rcar_vin, and adv7604 2014-07-09 20:34 ` Guennadi Liakhovetski @ 2014-07-10 10:12 ` Ian Molton 2014-10-15 17:03 ` William Towle 0 siblings, 1 reply; 4+ messages in thread From: Ian Molton @ 2014-07-10 10:12 UTC (permalink / raw) To: Guennadi Liakhovetski Cc: linux-media, William Towle, mchehab, hans.verkuil, sylvester.nawrocki, vladimir.barinov On Wed, 9 Jul 2014 22:34:07 +0200 (CEST) Guennadi Liakhovetski <g.liakhovetski@gmx.de> wrote: > Maybe we dould add some support, say, to > help with (fake) file handles just to aid the transition. Indeed - the filehandles are probably the biggest sticking point. I already have the soc_camera code able to select mutually acceptable data formats, but the calls to get/set resolution seem to use fh's I will persist with approach 2 then for now. -- Ian Molton <ian.molton@codethink.co.uk> ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: RFC: soc_camera, rcar_vin, and adv7604 2014-07-10 10:12 ` Ian Molton @ 2014-10-15 17:03 ` William Towle 0 siblings, 0 replies; 4+ messages in thread From: William Towle @ 2014-10-15 17:03 UTC (permalink / raw) To: g.liakhovetski Cc: ian.molton, Guennadi Liakhovetski, linux-media, mchehab, hans.verkuil, sylvester.nawrocki, vladimir.barinov Earlier this year, a colleague and I sought advice on the combination of "soc_camera + rcar_vin for capture, and the mainline adv7604 driver (which we have modified to successfully drive the adv7612).", with reasonable results, noting that we might move towards "a soc_camera2 with pads support?". For the next stage of work, we have created a test branch that contains an alternative version of rcar_vin.c, based on v4l2-pci-skeleton.c [from the Documentation tree] and with device tree initialisation code transplanted back into it. This presently creates device nodes happily, and if code from soc_camera.c is transplanted, we observe appropriate low-level interaction with hardware prompted by code in adv7604.c. I have not yet progressed to seeing calls to rcar_vin_irq() succeed, however, and wondered if you could shed light onto the places I might have overlooked? Cheers, Wills. (William Towle; william.towle@codethink.co.uk) ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2014-10-15 17:03 UTC | newest] Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2014-07-09 16:42 RFC: soc_camera, rcar_vin, and adv7604 Ian Molton 2014-07-09 20:34 ` Guennadi Liakhovetski 2014-07-10 10:12 ` Ian Molton 2014-10-15 17:03 ` William Towle
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.