TDA2HG: If we enable DSP1 & DSP2, it will no output display in our usecase

Hi terence,

The kernel logs you shared are not giving any details on why display is not observed. 

Can you share some information like what algorithms are running on DSP2?

If you remove one of the chains (either Capture_dsswb or capture) do you see the display?

Can you also get the statistics with option 'p' ?

Thanks

RamPrasad

Hi RamPrasad，

1.We had capture the newest log in the attachment, include  all cores print log, and get the statistics with option 'p'.

2.Nothing is running on DSP1&DSP2, except to the default thread of VSDK.   Also you can check in the log. 

3.If we remove the chains of Capture_dsswb, we can see the display.

Thanks

Terence

enable_dsp1&dps2.log

Hi Terence,

3.If we remove the chains of Capture_dsswb, we can see the display.

Does it mean if you remove this chain, issue is no more reproduced with both DSP1 and DSP2 inclusion?

Can you please add prints in display Link to confirm if the buffers are reaching display link?

Thanks

RamPrasad

Hi RamPrasad,

Where can we confirm if the buffers are reaching in display link? Which function of displayLink_drv.c or displayLink_tsk.c should we add prints?

Thanks

Terence

Hi Terence,

Please add some prints in displayLink_tsk.c for the case SYSTEM_CMD_NEW_DATA.

This confirms data is reaching from previous link.

Previous link sends this command to displayLink .

Thanks

RamPrasad

Hi RamPrasad,

According to your suggestion, we add print of "step into case SYSTEM_CMD_NEW_DATA in the function of DisplayLink_tskRun" in displayLink_tsk.c for the case SYSTEM_CMD_NEW_DATA.  Is this confirms  that data reached?

Please find the log in the attachment. 

Thanks

Terence

7115.enable_dsp1&dps2.log

Hi RamPrasad,

Is there any good news for this issue?

Thanks

Terence

Hi Terence,

From the log, I am seeing NEW_DATA command coming to display link. But there are two display links in your usecase.

Display_M4 and Display_yuy422. so I could not distinguish .

Can you please add some more prints in display_link like prevLinkID  and linkID etc to get more information ?

Thanks

RamPrasad

Hi RamPrasad,

According to your suggestion, We added prevLinkID print and reproduced it again. Please find the log in the attchment.

Thanks

Terence

5074.enable_dsp1&dps2.log

Hi Terence,

The log shows there are two NEW_DATA commands received on both the display links. 

Can you tell me which display is not seen?

DispDstSrcweston--> Display_M4 is a weston usecase. I guess this display is seen always. Are you able to launch weston clients? 

Second chain Capture_dssWb --> Display_yuy422 . Is this being expected on a second display? and is this is not observed when DSP2 is enabled?

Thanks

RamPrasad

Hi  RamPrasad,

Yes, you are right.

DispDstSrcweston--> Display_M4 is seen always. We can not see the dispaly of Capture_dssWb --> Display_yuv422, when DSP2 is enabled. While, only enable DSP1, the Display_yuy422 can output normal.

Hi Terence,

Is Capture_dssWb -->Display_yuv422 a weston application?

This link is running on IPU but not getting any clue why enabling DSP2 makes no display .

The log says that buffers are being sent to displayLink for displaying the output.

Thanks

RamPrasad

Ramprasad said:

Is Capture_dssWb -->Display_yuv422 a weston application?

 as the weston output is ARGB, the display to IHU should be the YUV422 that we use the Capture writeback to convert ARGB to YUV422

Hi  RamPrasad,

1.For "Is Capture_dssWb -->Display_yuv422 a weston application?"

My colleague had replied to you.

2.For "The log says that buffers are being sent to displayLink for displaying the output."

Now that the buffers be sent, but no display. Why?

Thanks

Terence

Hi RamPrasad,

Yes,it is normal when we replaced it with M2M WB link.

Thanks