mt9p031 error

anyone has any ideas?

the reply is blank

I guess there's something wrong with the system, I tried to post a couple of times, but it kept complaining the content is empty.

Anyway, for your issue, first make sure you can run the encode demo properly before you try gstreamer. According to TI developer, you need to patch v4l2src to make camera input work with gstreamer.

Thanks Hongfend, any reference of what files I have to patch and what should I look for? I'm using the leopard board dm365 with the mt9p031 5MP camera module.

has anyone encounter this problem?

Hie Andre

I have got the same problem I have speak with TI  engineers and it's still have no solution to solve this problem. You can't use gstreamer with LI-5M02 camera :(

I have already succeed with gstreamer, dm368evm and an analog camera. But I can't use the provided camera (LI-5M02) with gstreamer. After search and discussion with TI engineer, it's seems impossible to use this camera with gstreamer.

MY QUESTION: What way should I follow to develop my application ? Is there an other possibility to stream video over network who don't use Gstreamer ?


Regards,


Maxime Suire

Gstreamer works with MP4 over network, I'm been trying to figure out how to configure the camera so it would work with gstreamer, I email TI regarding this problem but they havent got back to me yet.

anyone know why I get this "vpfe-capture vpfe-capture" error, but if I select on the kernel capture SoC camera and compile the kernel with that selection all the "vpfe-capture vpfe-capture" errors dissapear and I get different errors.

was anyone able to get the leopard board DM365 and  the MT9P031 camera board working with the DVSDK 4?

anyone?

Is the patch for this available?

I am able to run the encodedecode demo program with the LI-5M02 camera just fine, and I am able to run gstreamer loop-back pipelines using an NTSC camera just fine.  However, I am unable to run a loopback pipeline using the LI-5M02 camera.  The pipeline I a trying to use is:

gst-launch -v v4l2src always-copy=FALSE input-src=Camera num-buffers=400 ! video/x-raw-bayer, bpp=1, width=1280, height=720, bytesperline=1280 ! bayer2rgb ! ffmpegcolorspace ! tidisplaysink2 video-standard=auto display-output=component

I get the following error:

ERROR: Pipeline doesn't want to pause.
ERROR: from element /GstPipeline:pipeline0/GstV4l2Src:v4l2src0: Could not negotiate format
Additional debug info:
gstbasesrc.c(2755): gst_base_src_start (): /GstPipeline:pipeline0/GstV4l2Src:v4l2src0:
Check your filtered caps, if any

Any word on the patch that is mentioned?

Doug

With a hack I got it working.

First of all, it turns out that I don't want video/x-raw-bayer caps, but this pipeline:

gst-launch -v v4l2src always-copy=FALSE input-src=Camera num-buffers=800 ! video/x-raw-yuv, format=\(fourcc\)NV12, width=1280, height=720 ! queue ! tidisplaysink2 video-standard=720p display-output=component

The hack, however, is in the mt9p031.c file.  In the mt9p031_querystd function, change:

 *id = V4L2_STD_MT9P031_STD_ALL;

to:

*id = V4L2_STD_720P_60;

This hack tells gstreamer that the detected standard is 720P_60, overriding the "all" value returned.

This is not the "correct" solution, but it's a place to start.

Hi Doug,

you said, you got the encodedecode demo working with the li-5m02.

I'm trying to get the encode demo working on my dm365evm with the li-5m02,

but I always get the error:

Encode demo started.                                                           
CMEMK Error: Failed to find a pool which fits 28672                            
CMEM Error: getPCMEMK Error: get_phys: Unable to find phys addr for 0x00000000 
ool: Failed to gCMEMK Error: get_phys: get_user_pages() failed: -14            
et a pool fittinCMEMK Error: GETPHYS: Failed to convert virtual 0x0 to physical.
g a size 28672                                                                 
CMEMK Error: get_phys: Unable to find phys addr for 0x00000000                 
CMEM Error: getPCMEMK Error: get_phys: get_user_pages() failed: -14            
hys: Failed to gCMEMK Error: FREE: Failed to convert virtual 0x0 to physical   
et physical address of 0                                                       
CMEM Error: free: failed to free 0                                             
Error: Failed to create attribute window device

The capture_prev_rsz_onthe_fly_bayer app works with the li-5m02 and the encode demo works with the component input.

Can you give me a hint, what I'm doing wrong and how you got the demo working?

BR,

Simon

Simon your error is here

CMEMK Error: Failed to find a pool which fits 28672   

you can find the solution here:

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/100/p/116107/412296.aspx

you need to increase the pool size

Hi Andre,

thank you for your quick response.

I added an additional pool with the size the error complains about (1x28672) in my "loadmodules_hd.sh" (also tried "loadmodules.sh"),

but there is still the same problem, that it can't find a pool which fits 28672.



I don't know if it's a problem, that I'm using the DM365, as the encode demo says:

Video inputs available: ... 4 Imager/Camera - for DM368

I'm trying to use Input 4, as I try to encode videos from my Li-5M02-camera.



Would be nice, if anyone could help me.



BR,

Simon

Could anyone tell me if you are able to use MT9P031 with gstreamer using DMA-COPY=true?  When I try to do it I get a warning

0:00:02.001614220  2215    0xfa258 WARN          tidisplaysink2 gsttidisplaysink2.c:340:xcopy:<tidisplaysink20> DMA copy is not possible on non contiguous buffer, defaulting to slow copy

Here it is gstreamer command

gst-launch --gst-debug=2 -v v4l2src always-copy=true input-src
=Camera ! video/x-raw-yuv, format=\(fourcc\)NV12, width=1280, height=720 ! queue
 max-size-buffers=3 ! dmaiperf ! tidisplaysink2 video-standard=720p display-out
put=component

When I use mmap-buffer=true and always-copy=true I can achieve about 24fps with CPU usage over 90%. I don't know where is mistake. dma-copy=true works well when i play some mp4 file.

Is there any solution for such problem?

Simon,

In my loadmodules.sh i have:

 modprobe cmemk phys_start=0x85800000 phys_end=0x88000000 allowOverlap=1 phys_start_1=0x00001000 phys_end_1=0x00008000 pools_1=1x28672 useHeapIfPoolUnavailable=1

and it works fine

You need to change phys_start and phys_end to match it to your memory size

Hi Karol,

I finally got time to test your suggestion, but it unfortunately doesn't work for me.

I tried to play a little bit with module insertion of cmemk and finally I tried:

insmod cmemk.ko phys_start=0x83C08000 phys_end=0x88000000 allowOverlap=1 phys_start_1=0x83C01000 phys_end_1=0x83C08000 pools_1=1x28672 useHeapIfPoolUnavailable=1

but if I try to start any of the demos, I get the error:

:/~ encode -I 4 -y 3 -v test.mpeg4  
Encode demo started.
Error: Failed to create attribute window device

Any idea, where my failure is?

BR,

Simon

You should run a debug to v4l2 and to DMAI module using

DMAI_DEBUG=2 encode...

also you can turn on debug on all dm3x modules using

echo 1 > /sys/module/../parameters/debug

It would much easier to detect problem

Karol

I tried the DMAI-Debug and the output is:

root@dm365:/opt/dvsdk/dm365# DMAI_DEBUG=2 encode -I 4 -y 3 -v test.mpeg4
Encode demo started.
@0x000634c2:[T:0x4001f070] ti.sdo.dmai - [Dmai] Dmai log level set to '2'. Note
that calling CERuntime_init after this point may cause unexpected change to DMAI
tracing behavior.
@0x0007bdc3:[T:0x4001f070] ti.sdo.dmai - [Display] Failed to open fb device /dev
/fb2 (No such file or directory)
Error: Failed to create attribute window device

After this, I tried the capture_prev_rsz_onthe_fly_bayer_mew, which worked for me before, but now I get there an error, too,

although I tried it with my original loadmodules_hd.sh:

root@dm365:/opt/dvsdk/dm365# ./loadmodules_hd.sh
cmemk unregistered
irqk unregistered
CMEMK module: built on Nov 4 2011 at 12:54:40
Reference Linux version 2.6.32
File /home/simmal/development/ti-dvsdk_dm365-evm_4_01_00_09/linuxutils_2_25_05
_11/packages/ti/sdo/linuxutils/cmem/src/module/cmemk.c
allocated heap buffer 0xc8000000 of size 0x4400000
cmemk initialized
IRQK module: built on Nov 4 2011 at 12:54:42
Reference Linux version 2.6.32
File /home/simmal/development/ti-dvsdk_dm365-evm_4_01_00_09/linuxutils_2_25_05
_11/packages/ti/sdo/linuxutils/irq/src/module/irqk.c
irqk initialized
EDMAK module: built on Nov 4 2011 at 12:54:44
Reference Linux version 2.6.32
File /home/simmal/development/ti-dvsdk_dm365-evm_4_01_00_09/linuxutils_2_25_05
_11/packages/ti/sdo/linuxutils/edma/src/module/edmak.c
root@simmals_board:/opt/dvsdk/dm365# ./capture_prev_rsz_onthe_fly_bayer_mew -i1
-m4 -h60
input_std_paramsCMEMK Error: ALLOC: invalid pool (0) passed.
: name = V4L2_STD_720P_30
input_std_params: output size = 1280 * 720
calling cmem utilities for allocating frame buffers
CMEM Error: allocPool: ioctl CMEM_IOCALLOC failed from pool 0: -1
Failed to allocate cmem pool
Unable to Allocate user buffers

I didn't change anything, so I'm really confused, why I get now an error here.

Have you setup your bootargs?

If you didn't make any changes besides cmem module configuration, it should not make any other change to system structure. Have you rebuild your kernel or make any changes inside it? You are using framebuffer, maybe try with v4l2 /dev/video. I don't know your board configuration so I can't tell more

Karol

As bootargs, I use the recommened ones from TI:

bootargs=console=ttyS0,115200n8 noinitrd rw ip=dhcp root=/dev/nfs nfsroot=192.168.178.40:/home/simmal/development/nfs,nolock mem=60M video=davincifb:vid0=OFF:vid1=OFF:osd0=1920x1080x4,32400K dm365_imp.oper_mode=0 vpfe_capture.interface=1 vpfe_capture.bufsize=4147200 davinci_enc_mngr.ch0_mode=1080I-30

I changed my Kernel a few weeks ago, but I used the camera after this kernel-upgrade successfully. I didn't make any further changes of the kernel or inside the kernel.

Do you mean the hardware config of my board? At the moment I'm still using the original DM365-EVM.

Is there a guide or sth like that, how I can change from framebuffer to v4l2, which you can recommend to me? I'm quite new to this topic, as you can guess ;)

To enable video you have to change line

video=davincifb:vid0=OFF:vid1=OFF:osd0=1920x1080x4

to

video=davincifb:vid0=0,4050K:vid1=OFF:osd0=1920x1080x4,32400K davinci_display.cont2_bufsize=4147200

that is my commandline works well although I got board based on DM368 designed by our company  not DVEVM

In /dev/ you should get /dev/video2

All memory settings for output you can find in /sys/module/davinci_display/parameters

You got mt9p031 camera? Was it working well with gstreamer using DMA? I still got problem with DMA and this cam with gstreamer, and I cannot check if it is runing correct on DVEVM.

Thank you for this hint. I hope I can try it soon, maybe tomorrow. For today I have a lot of other work, which is waiting (next week are midterm exams).

I never worked with gstreamer, yet. At the moment, I just try to run the TI-Demos with the camera. When I have this working, I probably will try gstreamer.

I can't figure out gstreamer's advantages at the moment, as I'm not deeply involved in this topic, yet.

Hi Karol,

I tried your proposed bootargs, but that did't change anything, the camera is still not working. Maybe I'll try gstreamer, but I'm not sure if this is usefull, before I didn't get the demos running.

Are your using the original DVSDK? Which version?

BR,

Simon