FreeBSD on Layerscape/QorIQ LX2160X

greg at unrelenting.technology greg at unrelenting.technology
Wed Jun 3 23:40:56 UTC 2020


June 3, 2020 7:09 PM, "myfreeweb" <greg at unrelenting.technology> wrote:

> On June 3, 2020 1:37:34 PM UTC, Dan Kotowski <dan.kotowski at a9development.com> wrote:
> 
>>>>> I've sent a link to a known firmware build before:
>>>>> https://drive.google.com/file/d/1yXSS1O1U8CmtwaIPfxNDkzhAClJGvErK/view
>>>>> Have you tried it? Any difference in FreeBSD/NetBSD, with NVMe?
>>>> 
>>>> I decided to go back to the UEFI sources and have found some differences that I think need to be
>>> reconciled before moving forward. That said, I'm not an ACPI wizard by any means - for me it's
>>> low-level mage spells at best...
>>>> In https://github.com/SolidRun/edk2-platforms we have 2 different branches that SolidRun seems to
>>> use:
>>>> 
>>>> 1. LSDK-19.09-sr
>>>> 2. master-lx2160a
>>>> 
>>>> I've been building from the latter branch, but found some significant differences in the former
>>> that I think may be important to merge in.
>>> 
>>> To me it seems like 19.09 is just outdated and doesn't have any benefits. Ask the solidrun people
>>> to be sure.
>>> 
>>> Either way, nothing here would fix the interrupt bug. It's our bug since NetBSD works fine :(
>> 
>> Any chance I can get a new test kernel without PCIe quirks? I just got a much more recent image
>> from jnettlet with the following comments:
>> 
>> BEGIN QUOTE
>> If you are using that recent uefi firmware I posted then you shouldn't be using the quirks for
>> pcie. That has an ecam shift setup where it should behave....relatively to SBSA standards.
>> it definitely won't work with the quirk enabled though. I have to add an interface to edk2 to turn
>> the mode on or off depending if you want access to the root bus and have a kernel with the quirk
>> applied, or you want it to work with just the devices exposed but in a more compliant manner
>> without quirks
>> END QUOTE
> 
> In the last couple kernels I posted, you should be able to set debug.acpi.disabled=pci_layerscape
> to skip the quirk.
> 
> I'll build the next one soon though, I guess with more interrupt debug prints lol

https://send.firefox.com/download/ae38fa35246497c1/#AVSGMsnrM0YB2MSL7rRJRQ

- customized pcie driver
+ https://reviews.freebsd.org/D25121
+ more interrupt debugging (stray interrupts, all GIC config writes)


More information about the freebsd-arm mailing list