No DSI MIPI data on custom carrier board

Yes, this sounds like should be noted in hardware design guide. I will check with some hardware folks about this.

Please do share your full dts to me later. Thanks.

1 Like

Also want to know that , for your case, do you need to pull up both DP1 and DP0 HPD? Or only one of them is needed to be pulled up?

At the beginning we pulled up both pins. Now we checked and it seems that only DP1_HPD needs to be pulled up for the DSI to work.

But could be it dependent of the nvdisplay address? I mean, if the DSI display was on nvdisplay@15210000, maybe DP0_HPD would need to be pulled up?


Only the SORx is mapping to the hardware, which means DP1_HPD is for SOR1 and DP0_HPD is for SOR0.

I am wondering the same thing as what you are talking about. Since nvdisplay@15200000 you are using now was for SOR1 to use, if you try to use nvdisplay@15210000, which was for SOR0, then you will need to pulled up DP0_HPD.

But it would be better to let me check your dts first because it still has chance to be software driver problem. I mean it should not care about the hpd pin in DSI driver flow, but it still uses. I have to to prevent such case first.

1 Like

Okay, I will give you the full DTS at the end of the work day


Are you able to share the DT? It shall only take less than few mins to convert the dtb back to dts.

1 Like


Yes, sorry for doing it on friday. I PMed you the DTS.


Could you remove that "nvidia,hpd-gpio = <0x21 0x79 0x1>; " from your sor1 DTS and see if this issue would be gone?

I mean removing this nvidia,hpd-gpio and see if it still needs to pull up the HPD pin.

Okay, will do


Sorry for the late response - we tried removing the HPD pin from DeviceTree but issue still persisted.

Ok. We will check this.

1 Like


Just in case, could you double confirm /proc/device-tree really has no hpd-gpio info there?

Yes, no hpd-gpio in proc device tree


In the meanwhile our hardware folks are checking, could you share the dts file that converted back from your dtb by using the dtc tool?

Doesn’t the one I sent you in PM work?


Since you just changed something, it would be better to share the latest one.

Okay, will do at the end of the workday. Promise :)

Also, once you have a new dtb file, please use full flash instead of flash kernel-dtb partition only to make sure some potential error.

But there are many things on eMMC that are on my device? How can I do full flash without eMMC format?


I can only say everyone has their own method for working.
Generally, if this is still doing hardware debug, then it is kind of weird that you already put lots of files over the board, because frequently re-flashing is expected. At least for me this is a little bit weird.

You can also try this when you have other available board. Of course the whole debug process will be slow down. That depends on you.