precedence of built-in vs. platform trees?
Hey guys,
I apologize if I should have RTFM. If a platform provides a device tree at boot time, and the kernel also has a tree appended, what behavior is supposed to happen? i.e. what is the standard that is anticipated here?
Jon.
_______________________________________________
arm mailing list
[email protected]
https://admin.fedoraproject.org/mailman/listinfo/arm
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 5 lines · Jon Masters
Hey guys,
I apologize if I should have RTFM. If a platform provides a device tree at boot time, and the kernel also has a tree appended, what behavior is supposed to happen? i.e. what is the standard that is anticipated here?
From the kernel config help on CONFIG_ARM_APPENDED_DTB:
<<<
This is meant as a backward compatibility convenience for those systems with a bootloader that can't be upgraded to accommodate the documented boot protocol using a device tree.
Quoted 1 line
Quoted 1 line
Quoted 1 line
and
<<<
Do not leave this option active in a production kernel if you don't intend to always append a DTB.
Quoted 1 line
Quoted 1 line
Quoted 1 line
Meaning that if the loader supports passing DTB, the kernel shouldn't have an appended one.
Apart from that, if you use an appended DTB, this completely overrides any DTB passed by the loader, so (unless you also enable the unholy CONFIG_ARM_ATAG_DTB_COMPAT) you also lose the ability to pass kernel command line options from the loader.
/
Leif
_______________________________________________
arm mailing list
[email protected]
https://admin.fedoraproject.org/mailman/listinfo/arm
Quoted 5 lines · Jon Masters
Hey guys,
I apologize if I should have RTFM. If a platform provides a device tree at boot time, and the kernel also has a tree appended, what behavior is supposed to happen? i.e. what is the standard that is anticipated here?
Hmmm, nobody has asked that before. I don't think it is really defined :-)
I presume that the built-in DT will end up getting used from what I remember about the code.
g.
A message that is not in the archive (sent off-list or not yet received): CACxGe6uc-E2kHuQeKbJNDgEWLa_DLqDRzC4eRyKReeZTVdQVeA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org
Quoted 11 lines · Grant Likely
Quoted 5 lines · Jon Masters
Hey guys,
I apologize if I should have RTFM. If a platform provides a device tree at boot time, and the kernel also has a tree appended, what behavior is supposed to happen? i.e. what is the standard that is anticipated here?
Hmmm, nobody has asked that before. I don't think it is really defined :-)
I presume that the built-in DT will end up getting used from what I remember about the code.
That sounds like the right default, but it would be nice if one could override it with a cmdline option. It's *usually* easier to change the kernel than the platform firmware or bootloader, but there are exceptions. In our (OLPC's) case, DT mods can be made trivially by editing a script (and the firmware includes an editor). We use this feature all the time in development, and sometimes even for customers, for testing bug fixes.
Quoted 7 lines
g.
_______________________________________________
devicetree-discuss mailing list devicetree-discuss-uLR06cmDAlY/[email protected]
https://lists.ozlabs.org/listinfo/devicetree-discuss