can no dequeue display buffer

Klaus Schwarzkopf,

Going through the description, looks like LCD supports 320x240p60.

Klaus Schwarzkopf said:

I think, the problem is the framebuffer size bootargs. If i use only the bootargs below, the LCD works, but i have the descripted problem.

If i add "video=vpbe_fb:vid0=640x480x16,1800K@0,0:vid1=off:osd0=640x480x16,1800K@0,0:osd1=640x480x4,600K@0,0", the LCD don't work and i get some errors:

I believe you cant set 640x480x16 if the display supports only 320x240p60.

------------------------------------------------------------------------------------------------------

For the issue of sensor to LCD loop-back, the driver has a limitation due to shared resource between Digital output controller and Sensor, hence you are seeing the issue. This is the reason why no sensor-LCD loop-back application in arago.

Hi,

Please check below things as well

1, See if the enough buffers are enqueued in the display driver as per the requirement of the driver

2, See if the pixel clock is configured correctly. If it is not, it may not be able to work properly.

3, Check the /proc/interrupts entry in to see if the interrupts are occuring for the display. If it is, display is sending out the data, but driver is not allowing to dequeue buffers.

Thanks,

Brijesh Jadav

Hi,

thanks for your helpful hints!

@Brijesh Jadav

I check your hints, see my next post!

@Nagabhushana

You are right, my mistake!

Buffer size: (320x240x2)/1024 = 150K * 3 buffer = 450K

Configuration 1:

With this boot parameters: video=vpbe_fb:vid0=320x240x16,450K@0,0:vid1=320x240x16,450K@0,0

* no error at boot time

vpbe-fb vpbe-fb: dm_osd0_fb: 320x240x16@0,0 with framebuffer size 150KB

vpbe-fb vpbe-fb: dm_vid0_fb: 320x240x16@0,0 with framebuffer size 450KB

vpbe-fb vpbe-fb: dm_osd1_fb: 320x240x4@0,0 with framebuffer size 150KB

vpbe-fb vpbe-fb: dm_vid1_fb: 320x240x16@0,0 with framebuffer size 450KB

* LCD shows boot logo

* can not open /dev/video5 or /dev/video6

# cat /dev/video6 

vpbe-v4l2 vpbe-v4l2: Display Manager failed to allocate layer

vpbe-v4l2 vpbe-v4l2: Unable to configure video layer for id = 1

cat: can't open '/dev/video6': Invalid argument

Configuration 2:

With this boot parameters: video=vpbe_fb:vid0=320x240x16,450K@0,0:vid1=320x240x16,450K@0,0:osd0=320x240x16,150K@0,0:osd1=320x240x4,150K@0,0

* no error at boot time

* no boot logo

* can not open /dev/video5 or /dev/video6

Configuration 3:

Without boot parameters:

* no error at boot time

vpbe-fb vpbe-fb: dm_osd0_fb: 320x240x16@0,0 with framebuffer size 150KB

vpbe-fb vpbe-fb: dm_vid0_fb: 0x0x16@0,0 with framebuffer size 225KB

vpbe-fb vpbe-fb: dm_osd1_fb: 320x240x4@0,0 with framebuffer size 150KB

vpbe-fb vpbe-fb: dm_vid1_fb: 0x0x16@0,0 with framebuffer size 225KB

* LCD shows boot logo

* can open /dev/video5 or /dev/video6

# /make_media_dev.sh 

# /mt9v126_ccdc_loopback_mmap 

media device opened

4.enumerating media entities

[1]:81 0 mt9v126

[2]:81 1 DAVINCI CCDC

[3]:81 2 DAVINCI VIDEO CCDC output

[4]:81 3 DAVINCI PREVIEWER

[5]:81 4 DAVINCI VIDEO PRV input

[6]:81 5 DAVINCI VIDEO PRV output

[7]:81 6 DAVINCI RESIZER

[8]:81 7 DLeopard: switch to VGA imager-MT9V126 video input

AVINCI VIDEO RSZ input

[9]:81 8 DAVINCI VIDEO RSZ output

Leopard: switch to VGA imager-MT9V126 video input

[b]:81 10 DAVINCI AF

total number of entities: b

**********************************************

6. ENABLEing link [MT9V126]----------->[ccdc]

[MT9V126]----------->[ccdc] ENABLED

7. ENABLEing link [ccdc]----------->[video_node]

[ccdc]----------->[video_node] ENABLED

**********************************************

5.enumerating links/pads for entities

pads for entity 1=(0, OUTPUT) 

[1:0]-------------->[2:0] ACTIVE

pads for entity 2=(0, INPUT) (1, OUTPUT) 

[2:1]-------------->[3:0] ACTIVE

[2:1]-------------->[a:0] INACTIVE 

[2:1]-------------->[b:0] INACTIVE 

[2:1]-------------->[4:0] INACTIVE 

pads for entity 3=(0, INPUT) 

pads for entity 4=(0, INPUT) (1, OUTPUT) 

[4:1]-------------->[6:0] INACTIVE 

[4:1]-------------->[7:0] INACTIVE 

pads for entity 5=(0, OUTPUT) 

[5:0]-------------->[4:0] INACTIVE 

pads for entity 6=(0, INPUT) 

pads for entity 7=(0, INPUT) (1, OUTPUT) 

[7:1]-------------->[9:0] INACTIVE 

pads for entity 8=(0, OUTPUT) 

[8:0]-------------->[7:0] INACTIVE 

pads for entity 9=(0, INPUT) 

pads for entity a=(0, INPUT) 

pads for entity b=(0, INPUT) 

**********************************************

ccdc device opened

display device opened

8. setting format on pad of MT9V126 entity. . .

successfully format is set on pad 0

closed mt9v126 sub-device

9. setting format on sink-pad of ccdc entity. . .

successfully format is set on pad 0

10. setting format on OF-pad of ccdc entity. . . 

successfully format is set on pad 1

**********************************************

closed ccdc sub-device

ccdc_output is video capture device

ccdc_output does support streaming i/o

enumerating INPUTS

[0].Camera

**********************************************

setting camera input. . .

successfully set camera input

**********************************************

Available image formats at the capture driver :-

fmt_desc.index = 0

fmt_desc.type = 1

fmt_desc.description = 

fmt_desc.pixelformat = 59565955

Force Format: 59565955

Selected image format at the capture driver :-

fmt.fmt.pix.bytesperline = 1280

fmt.fmt.pix.width = 640

fmt.fmt.pix.height = 480

fmt.fmt.pix.pixelformat = 59565955

VIDIOC_CROPCAP

setting format V4L2_PIX_FMT_UYVY

successfully set the format on capture device

capture_pitch: 500

**********************************************

3 buffers are supported for streaming (capture)

**********************************************

Querying capture buffers and MMAPing capture

buffer 0 queried 

buf length : 614400

Capture Buffer 0 mapped at address 0x40336000.

buffer 1 queried 

buf length : 614400

Capture Buffer 1 mapped at address 0x4044b000.

buffer 2 queried 

buf length : 614400

Capture Buffer 2 mapped at address 0x4056a000.

**********************************************

setting Display output. . . 

successfully set Display display

Available image formats at the display driver :-

fmt_desc.index = 0

fmt_desc.type = 2

fmt_desc.description = YUV 4:2:2 - UYVY

fmt_desc.pixelformat = 59565955

Test S_FMT display

display S_FMT successful

size image : 153600

Test G_FMT display

display G_FMT successful

3. Test GetFormat

dispheight = 240

disppitch = 640

dispwidth = 320

imagesize = 153600, field 4

Test REQBUF display

3 buffers are supported for streaming (display)

**********************************************

Querying display buffers and MMAPing 

buffer 0 queried 

buf length : 153600

Display Buffer 0 mapped at address 0x40048000.

buffer 1 queried 

buf length : 153600

Display Buffer 1 mapped at address 0x402c3000.

buffer 2 queried 

buf length : 153600

Display Buffer 2 mapped at address 0x403cc000.

**********************************************

Q capture BUF 0---40336000---96000

Q capture BUF 1---4044b000---96000

Q capture BUF 2---4056a000---96000

Q display BUF 0---40048000---25800

Q display BUF 1---402c3000---25800

Q display BUF 2---403cc000---25800

streaming started successfully

Test STREAMON display

display STREAMON successful

Start while...

Start while... 1

DQ buffer from capture device

 

 I2C write failed

quit program with ctrl+c

# cat /proc/interrupts 

           CPU0

  0:        484       AINTC  vpfe_capture0

  1:        484       AINTC  vpfe_capture1

  5:          0       AINTC  Imp_Sdram_Irq

  8:      15037       AINTC  vpbe-v4l2

 16:          0       AINTC  edma

 17:          0       AINTC  edma_error

 26:       1280       AINTC  mmc0

 32:       4042       AINTC  clockevent

 33:          0       AINTC  free-run counter

 39:       4461       AINTC  i2c_davinci

 40:        370       AINTC  serial

 52:          0       AINTC  eth0

 53:      13395       AINTC  eth0

 54:      13467       AINTC  eth0

 55:          0       AINTC  eth0

Err:          0

I am not sure about Configuration 1 and Configuration 2 failing.

Regarding sensor to LCD loop-back, as I mentioned in earlier reply, there is a issue in driver.  I need to work on patch to fix the same. I don't know when I get time.

Brijesh Jadav said:

1, See if the enough buffers are enqueued in the display driver as per the requirement of the driver

I think, the display driver in drivers/media/video/davinci/vpbe_display.c use 3 buffers and my application use 3 buffers.

Brijesh Jadav said:

2, See if the pixel clock is configured correctly. If it is not, it may not be able to work properly.

What pixel clock do you mean? Which registers?

Brijesh Jadav said:

3, Check the /proc/interrupts entry in to see if the interrupts are occuring for the display. If it is, display is sending out the data, but driver is not allowing to dequeue buffers.

# cat /proc/interrupts 

           CPU0

  0:        484       AINTC  vpfe_capture0

  1:        484       AINTC  vpfe_capture1

  5:          0       AINTC  Imp_Sdram_Irq

  8:      15037       AINTC  vpbe-v4l2

 16:          0       AINTC  edma

 17:          0       AINTC  edma_error

 26:       1280       AINTC  mmc0

 32:       4042       AINTC  clockevent

 33:          0       AINTC  free-run counter

 39:       4461       AINTC  i2c_davinci

 40:        370       AINTC  serial

 52:          0       AINTC  eth0

 53:      13395       AINTC  eth0

 54:      13467       AINTC  eth0

 55:          0       AINTC  eth0

Err:          0

The interrupts looks ok.

Thanks.

Regards,

Klaus

Hi,

From the interrupts, it looks like it is outputting data. Can you see if the number of interrupts in /proc/interrupt matches with your fps?

Thanks,

Brijesh Jadav

Brijesh Jadav said:

Can you see if the number of interrupts in /proc/interrupt matches with your fps?

vpbe-v4l2 has 250 interrupts per sec. I think, this is to fast.

Regards,

Klaus

Klaus Schwarzkopf,

Can you please try this patch? Since my setup is busy and I don't have application, I could not test it. Let me know what you observe.

5305.0001-dm36x-fix-cpld-resets-corruption-during-lcd-enabling.patch.zip

Hi,

This clearly tells that pixel clock is configured to almost four times the required pixel clock. Please look into the display driver and see where it is configuring pixel clock. Please configure it as required for your mode.

But this should not cause hang while dequeuing the buffer. Are you able to dequeue at-least one buffer from the display driver?

thanks,

Brijesh Jadav

Nagabhushana said:

Can you please try this patch? Since my setup is busy and I don't have application, I could not test it. Let me know what you observe.

I use the dm365 processor. The patch is for dm368, right?

Brijesh Jadav said:

But this should not cause hang while dequeuing the buffer. Are you able to dequeue at-least one buffer from the display driver?

I can no buffer dequeue.

Which(dm365) revision? I believe rev E and above is called dm368.

I'm sorry, for not responding for a long time.

I use the leopardboard 365 and this board has no CPLD. So your patch can not help me and i can not test it.