iMX6SX PCIe Clock Switching Off on Linux Boot

[simonlocke]

Hi all,

I've been looking around and I can't seem to find any answer for this.

Does any one know any reasons why the PCIe clock would switch off on kernel boot?

I've seen someone had a similar post: I.MX6 Solo/DL PCIE clock issue but this was never answered. I know that other people must be using PCIe to talk to PCI devices with their SABRE dev boards.

I've configured it to be a Root Complex in the kernel menuconfig, and although I don't have a driver, I get a clock briefly during kernel startup, and then it switches off. It does basically find the device, because lspci shows the correct device and vendor IDs. Is it down to not having a device driver for the end point in the kernel at boot (as yet)? Or is there possibly some other fundamental code switch that I need to enable?

[art]

You should have a valid device driver to keep the PCIe clock running and bus communicating when a valid PCIe device is found.

Have a great day,
Artur

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[simonlocke]

Thanks Arthur,

You're kind of confirming my suspicions. It was my gut feeling that it had switched off because there was no valid device driver installed.

However, this to me seems like a bit of a bug to me, as even if I then install a driver after boot, it does not recover. I get an error:

Refused to change power state, currently in D3

And I can see from my hardware that the clock is still off. I didn't initially expect the clock to go off. When I've done vxWorks stuff before, the bus stays alive even if there's is no driver, and you can peek the PCI config space. That was what I wanted to be able to do. But now I have to do a lot more work to just do some simple testing.

I am working on getting an image with the driver in, but this is proving complicated...