From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from hermes.mlbassoc.com ([64.234.241.98]:43699 "EHLO mail.chez-thomas.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754705Ab1KYLu2 (ORCPT ); Fri, 25 Nov 2011 06:50:28 -0500 Message-ID: <4ECF8101.7050603@mlbassoc.com> Date: Fri, 25 Nov 2011 04:50:25 -0700 From: Gary Thomas MIME-Version: 1.0 To: Laurent Pinchart CC: Javier Martinez Canillas , Linux Media Mailing List Subject: Re: Using MT9P031 digital sensor References: <4EB04001.9050803@mlbassoc.com> <201111160226.34977.laurent.pinchart@ideasonboard.com> <4EC3A67F.2010901@mlbassoc.com> <201111241228.38082.laurent.pinchart@ideasonboard.com> In-Reply-To: <201111241228.38082.laurent.pinchart@ideasonboard.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-media-owner@vger.kernel.org List-ID: On 2011-11-24 04:28, Laurent Pinchart wrote: > Hi Gary, > > On Wednesday 16 November 2011 13:03:11 Gary Thomas wrote: >> On 2011-11-15 18:26, Laurent Pinchart wrote: >>> On Monday 14 November 2011 12:42:54 Gary Thomas wrote: >>>> On 2011-11-11 07:26, Laurent Pinchart wrote: >>>>> On Wednesday 09 November 2011 17:24:26 Gary Thomas wrote: >>>>>> On 2011-11-09 09:18, Laurent Pinchart wrote: >>>>>>> On Wednesday 09 November 2011 12:01:34 Gary Thomas wrote: >>>>>>>> On 2011-11-08 17:54, Laurent Pinchart wrote: >>>>>>>>> On Tuesday 08 November 2011 14:38:55 Gary Thomas wrote: >>>>>>>>>> On 2011-11-08 06:06, Laurent Pinchart wrote: >>>>>>>>>>> On Tuesday 08 November 2011 13:52:25 Gary Thomas wrote: >>>>>>>>>>>> On 2011-11-08 05:30, Javier Martinez Canillas wrote: >>>>>>>>>>>>> On Tue, Nov 8, 2011 at 1:20 PM, Gary Thomas wrote: >>>>>>>>>>>>>> On 2011-11-04 04:37, Laurent Pinchart wrote: >>>>>>>>>>>>>>> On Tuesday 01 November 2011 19:52:49 Gary Thomas wrote: >>>>>>>>>>>>>>>> I'm trying to use the MT9P031 digital sensor with the Media >>>>>>>>>>>>>>>> Controller Framework. media-ctl tells me that the sensor is >>>>>>>>>>>>>>>> set to capture using SGRBG12 2592x1944 >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> Questions: >>>>>>>>>>>>>>>> * What pixel format in ffmpeg does this correspond to? >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> I don't know if ffmpeg supports Bayer formats. The >>>>>>>>>>>>>>> corresponding fourcc in V4L2 is BA12. >>>>>>>>>>>>>> >>>>>>>>>>>>>> ffmpeg doesn't seem to support these formats >>>>>>>>>>>>>> >>>>>>>>>>>>>>> If your sensor is hooked up to the OMAP3 ISP, you can then >>>>>>>>>>>>>>> configure the pipeline to include the preview engine and the >>>>>>>>>>>>>>> resizer, and capture YUV data >>>>>>>>>>>>>>> at the resizer output. >>>>>>>>>>>>>> >>>>>>>>>>>>>> I am using the OMAP3 ISP, but it's a bit unclear to me how to >>>>>>>>>>>>>> set up the pipeline >>>>>>>>>>>>> >>>>>>>>>>>>> Hi Gary, >>>>>>>>>>>>> >>>>>>>>>>>>> I'm also using another sensor mtv9034 with OMAP3 ISP, so maybe >>>>>>>>>>>>> I can help you. >>>>>>>>>>>>> >>>>>>>>>>>>>> using media-ctl (I looked for documentation on this tool, but >>>>>>>>>>>>>> came up dry - is there any?) >>>>>>>>>>>>>> >>>>>>>>>>>>>> Do you have an example of how to configure this using the >>>>>>>>>>>>>> OMAP3 ISP? >>>>>>>>>>>>> >>>>>>>>>>>>> This is how I configure the pipeline to connect the CCDC with >>>>>>>>>>>>> the Previewer and Resizer: >>>>>>>>>>>>> >>>>>>>>>>>>> ./media-ctl -l '"mt9v032 3-005c":0->"OMAP3 ISP CCDC":0[1]' >>>>>>>>>>>>> ./media-ctl -l '"OMAP3 ISP CCDC":2->"OMAP3 ISP preview":0[1]' >>>>>>>>>>>>> ./media-ctl -l '"OMAP3 ISP preview":1->"OMAP3 ISP >>>>>>>>>>>>> resizer":0[1]' ./media-ctl -l '"OMAP3 ISP resizer":1->"OMAP3 >>>>>>>>>>>>> ISP resizer output":0[1]' ./media-ctl -f '"mt9v032 >>>>>>>>>>>>> 3-005c":0[SGRBG10 752x480]' ./media-ctl -f '"OMAP3 ISP >>>>>>>>>>>>> CCDC":0 [SGRBG10 752x480]' ./media-ctl -f '"OMAP3 ISP CCDC":1 >>>>>>>>>>>>> [SGRBG10 752x480]' ./media-ctl -f '"OMAP3 ISP preview":0 >>>>>>>>>>>>> [SGRBG10 752x479]' ./media-ctl -f '"OMAP3 ISP resizer":0 >>>>>>>>>>>>> [YUYV 734x471]' ./media-ctl -f '"OMAP3 ISP resizer":1 [YUYV >>>>>>>>>>>>> 640x480]' >>>>>>>>>>>>> >>>>>>>>>>>>> Hope it helps, >>>>>>>>>>>> >>>>>>>>>>>> Thanks, I'll give this a try. >>>>>>>>>>>> >>>>>>>>>>>> I assume that your sensor is probably larger than 752x480 (the >>>>>>>>>>>> mt9p031 is 2592x1944 raw) and that setting the smaller frame >>>>>>>>>>>> size enables some scaling and/or cropping in the driver? >>>>>>>>>>> >>>>>>>>>>> The mt9v034 is a wide VGA 752x480 sensor if I'm not mistaken. You >>>>>>>>>>> should modify the resolutions in the above commands according to >>>>>>>>>>> your sensor. Note that the CCDC crops online line when outputting >>>>>>>>>>> data to the preview engine, and that the preview engine crops 18 >>>>>>>>>>> columsn and 8 lines. You can then scale the image by modifying >>>>>>>>>>> the resizer output size. >>>>>>>>>> >>>>>>>>>> Thanks. After much trial and error (and some kernel printks to >>>>>>>>>> >>>>>>>>>> understand what parameters were failing), I came up with this >>> >>> sequence: >>>>>>>>>> media-ctl -r >>>>>>>>>> media-ctl -l '"mt9p031 3-005d":0->"OMAP3 ISP CCDC":0[1]' >>>>>>>>>> media-ctl -l '"OMAP3 ISP CCDC":2->"OMAP3 ISP preview":0[1]' >>>>>>>>>> media-ctl -l '"OMAP3 ISP preview":1->"OMAP3 ISP >>>>>>>>>> resizer":0[1]' media-ctl -l '"OMAP3 ISP resizer":1->"OMAP3 >>>>>>>>>> ISP resizer output":0[1]' media-ctl -f '"mt9p031 >>>>>>>>>> 3-005d":0[SGRBG12 2592x1944]' media-ctl -f '"OMAP3 ISP >>>>>>>>>> CCDC":0 [SGRBG12 2592x1944]' >>>>>>>>>> media-ctl -f '"OMAP3 ISP CCDC":1 [SGRBG12 2592x1944]' >>>>>>>>>> media-ctl -f '"OMAP3 ISP preview":0 [SGRBG12 2592x1943]' >>>>>>>>>> media-ctl -f '"OMAP3 ISP resizer":0 [YUYV 2574x1935]' >>>>>>>>>> media-ctl -f '"OMAP3 ISP resizer":1 [YUYV 642x483]' >>>>>>>>>> >>>>>>>>>> When I tried to grab though, I got this: >>>>>>>>>> >>>>>>>>>> # yavta --capture=4 -f YUYV -s 642x483 -F /dev/video6 >>>>>>>>>> Device /dev/video6 opened. >>>>>>>>>> Device `OMAP3 ISP resizer output' on `media' is a video capture >>>>>>>>>> device. Video format set: YUYV (56595559) 642x483 buffer size >>>>>>>>>> 633696 Video format: YUYV (56595559) 642x483 buffer size 633696 >>>>>>>>>> 8 buffers requested. >>>>>>>>>> length: 633696 offset: 0 >>>>>>>>>> Buffer 0 mapped at address 0x4028c000. >>>>>>>>>> length: 633696 offset: 634880 >>>>>>>>>> Buffer 1 mapped at address 0x403d0000. >>>>>>>>>> length: 633696 offset: 1269760 >>>>>>>>>> Buffer 2 mapped at address 0x404b3000. >>>>>>>>>> length: 633696 offset: 1904640 >>>>>>>>>> Buffer 3 mapped at address 0x4062b000. >>>>>>>>>> length: 633696 offset: 2539520 >>>>>>>>>> Buffer 4 mapped at address 0x406d6000. >>>>>>>>>> length: 633696 offset: 3174400 >>>>>>>>>> Buffer 5 mapped at address 0x40821000. >>>>>>>>>> length: 633696 offset: 3809280 >>>>>>>>>> Buffer 6 mapped at address 0x4097c000. >>>>>>>>>> length: 633696 offset: 4444160 >>>>>>>>>> Buffer 7 mapped at address 0x40adf000. >>>>>>>>>> >>>>>>>>>> Unable to handle kernel NULL pointer dereference at virtual >>>>>>>>>> address 00000018 >>>>>>>>> >>>>>>>>> Ouch :-( >>>>>>>>> >>>>>>>>> Could you please verify that arch/arm/mach-omap2/board-overo.c >>>>>>>>> includes the following code, and that CONFIG_OMAP_MUX is enabled ? >>>>>>>> >>>>>>>> I'm not using an Overo board - rather one of our own internal >>>>>>>> designs. >>>>>>> >>>>>>> My bad, sorry. >>>>>>> >>>>>>>> I have verified that the pull up/down on those pins is disabled. >>>>>>>> >>>>>>>> The failure is coming from this code in ispccdc.c >>>>>>>> >>>>>>>> static void ccdc_hs_vs_isr(struct isp_ccdc_device *ccdc) >>>>>>>> { >>>>>>>> >>>>>>>> struct isp_pipeline *pipe = >>>>>>>> >>>>>>>> to_isp_pipeline(&ccdc->video_out.video.entity); >>>>>>>> >>>>>>>> The value of pipe is NULL which leads to the failure. >>>>>>>> >>>>>>>> Questions: >>>>>>>> * I assume that getting HS/VS interrupts is correct in this mode? >>>>>>>> * Why is the statement not written (as all others are) >>>>>>>> >>>>>>>> struct isp_pipeline *pipe = to_isp_pipeline(&ccdc->subdev.entity); >>>>>>>> >>>>>>>> I tried this change and the kernel doesn't crash. >>>>>>>> >>>>>>>> I've found that I can get raw frames out of CCDC, but I don't get >>>>>>>> anything at all when the output continues through the preview and/or >>>>>>>> resize nodes. >>>>>>>> >>>>>>>> Ideas? >>>>>>> >>>>>>> I'm really puzzled, this should have been caught much earlier :-) >>>>>>> >>>>>>> Your analysis makes sense. Would you like to submit a patch yourself >>>>>>> ? If not I can do it. >>>>>> >>>>>> Sure, I can submit a patch. I would like to figure out why it's not >>>>>> working first. >>>>> >>>>> Oops, I've overlooked that, sorry. >>>>> >>>>>> Any ideas how I can debug this? I can't seem to get anything past the >>>>>> CCDC, e.g. into the preview or resize units. Is there some way to >>>>>> trace packets/data through the various stages? Any ideas what might >>>>>> cause it to stall? >>>>> >>>>> How have you configured your pipeline ? Can you try tracing the preview >>>>> engine and/or resizer interrupts ? >>>> >>>> Here's my pipeline: >>>> media-ctl -r >>>> media-ctl -l '"mt9p031 3-005d":0->"OMAP3 ISP CCDC":0[1]' >>>> media-ctl -l '"OMAP3 ISP CCDC":2->"OMAP3 ISP preview":0[1]' >>>> media-ctl -l '"OMAP3 ISP preview":1->"OMAP3 ISP resizer":0[1]' >>>> media-ctl -l '"OMAP3 ISP resizer":1->"OMAP3 ISP resizer >>>> output":0[1]' media-ctl -f '"mt9p031 3-005d":0[SGRBG12 2592x1944]' >>>> media-ctl -f '"OMAP3 ISP CCDC":0 [SGRBG10 2592x1944]' >>>> media-ctl -f '"OMAP3 ISP CCDC":1 [SGRBG10 2592x1944]' >>>> media-ctl -f '"OMAP3 ISP preview":0 [SGRBG10 2592x1943]' >>>> media-ctl -f '"OMAP3 ISP resizer":0 [YUYV 2574x1935]' >>>> media-ctl -f '"OMAP3 ISP resizer":1 [YUYV 642x483]' >>>> >>>> The full media-ctl dump is at http://www.mlbassoc.com/misc/pipeline.out >>>> >>>> When I try to grab from /dev/video6 (output node of resizer), I see >>>> only previewer interrupts, no resizer interrrupts. I added a simple >>>> printk at each of the previewer/resizer *_isr functions, and I only >>>> >>>> ever see this one: >>>> omap3isp_preview_isr_frame_sync.1373 >>>> >>>> Can you give me an overview of what events/interrupts should occur so >>>> I can try to trace through the ISP to see where it is failing? >>> >>> The CCDC generates VD0, VD1 and HS/VS interrupts regardless of whether it >>> processes video or not, as long as it receives a video stream at its >>> input. The preview engine and resizer will only generate an interrupt >>> after writing an image to memory. With your above configuration VD0, >>> VD1, HS/VS and resizer interrupts should be generated. >>> >>> Your pipeline configuration looks correct, except that the downscaling >>> factor is slightly larger than 4. Could you try to setup the resizer to >>> output a 2574x1935 image instead of 642x483 ? If that works, try to >>> downscale to 660x496. If that works as well, the driver should be fixed >>> to disallow resolutions that won't work. >> >> No change. I also tried using only the previewer like this: >> media-ctl -r >> media-ctl -l '"mt9p031 3-005d":0->"OMAP3 ISP CCDC":0[1]' >> media-ctl -l '"OMAP3 ISP CCDC":2->"OMAP3 ISP preview":0[1]' >> media-ctl -l '"OMAP3 ISP preview":1->"OMAP3 ISP preview output":0[1]' >> media-ctl -f '"mt9p031 3-005d":0[SGRBG12 2592x1944]' >> media-ctl -f '"OMAP3 ISP CCDC":0 [SGRBG12 2592x1944]' >> media-ctl -f '"OMAP3 ISP CCDC":1 [SGRBG10 2592x1944]' >> media-ctl -f '"OMAP3 ISP preview":0 [SGRBG10 2592x1943]' >> media-ctl -f '"OMAP3 ISP preview":1 [YUYV 2574x1935]' >> >> yavta --capture=4 -f YUYV -s 2574x1935 -F /dev/video4 >> >> I still only get the frame sync interrupts in the previewer, no buffer >> interrupts, hence no data flowing to my application. What else can I >> look at? > > Do you get VD0 and VD1 interrupts ? Yes, the CCDC is working correctly, but nothing moves through the previewer. Here's a trace of the interrupt sequence I get, repeated over and over. These are printed as __FUNCTION__.__LINE__ --- ccdc_vd0_isr.1615 --- ccdc_hs_vs_isr.1482 --- ccdc_vd1_isr.1664 --- omap3isp_preview_isr_frame_sync.1373 What's the best tree to try this against? 3.2-rc2 doesn't have the BT656 stuff in it yet, so I've been still using my older tree (3.0.0 + drivers/media from your tree) Thanks -- ------------------------------------------------------------ Gary Thomas | Consulting for the MLB Associates | Embedded world ------------------------------------------------------------