For those of you not on the OLPC devel list...
We have the unified device tree code for all arch's close to ready. Its in /proc for now but I think it will be moved into /sys in the next week or 2.
Look for it in an olpc kernel in the near future.
I think this will help us out in linuxbios quite a bit.
Here's a snippit of the thread.
On Mon, 2006-12-25 at 00:42 -1000, Mitch Bradley wrote:
-bash-3.1# cd /proc/device-tree/ -bash-3.1# ls #address-cells chosen memory@0 null-nvram pci aliases cpus mmu openprom architecture dma-ranges model options banner-name flash@fff20000 nandflash@20000000 packages
Neat! What's left to do, and what's the schedule for integration?
Dan
Left to do:
a) Audit the device tree and correct any errors and omissions.
b) Release an Open Firmware version that uses virtual addressing so that Linux doesn't clobber it upon startup. The coding is already done, so this is just a release and test step. This version is compatible with existing kernels so its release doesn't require a simultaneous release of a new kernel.
c) Decide whether or not we want to go ahead with the /proc/device-tree version, or convert to /sys first. If the latter, we have to develop the sysfs version of the code. I think I understand how to convert to sysfs for the bulk of the code, but there is still an issue that I don't grok, namely the kobjects tie-in.
d) Do the administrative mechanics of expressing the changes as a patch.
e) Start converting OLPC userland code to use the device tree info instead of the various ad-hoc detections schemes currently in place.
f) Do whatever needs to be done to announce the patch, submit it upstream, or whatever we want to do on that front.
Richard, if this goes in the Linux mainstream then our life just got a whole lot easier!
Let me know when the discussion starts, I don't generally read LKML but would like to add words of support if possible.
The we can work on *BSD to parse this stuff to :-)
Thanks
ron
Let me know when the discussion starts, I don't generally read LKML but would like to add words of support if possible.
Even better -- just cross-post to here.
The we can work on *BSD to parse this stuff to :-)
Some of the BSDs have OF support for some architectures already, just like Linux. But the more and the more generic, the better :-)
Segher
Segher Boessenkool wrote:
Let me know when the discussion starts, I don't generally read LKML but would like to add words of support if possible.
Even better -- just cross-post to here.
We want the patch to get reviewed by jon corbet first and then it will probably get submitted by someone with lmkl name recognition. Cross posting probably won't happen but I'll certainly post a note when the patch is available in the olpc tree and when sent upstream.
-- Richard.
I googled around and read a bit about the OpenFirmware device tree, but I'm still a bit clueless.
Can you briefly summarize the purpose of the OpenFirmware device tree? How does/would Linux use the information in the tree? Does it replace or supplement things like the MP table, IRQ routing table, ACPI tables, etc.?
--Ed
On 12/26/06, Richard Smith smithbone@gmail.com wrote:
For those of you not on the OLPC devel list...
Can you briefly summarize the purpose of the OpenFirmware device tree?
The OF device tree describes all hardware devices in the system, and the relations between those devices.
How does/would Linux use the information in the tree?
Linux would use this information to detect and configure devices.
Does it replace or supplement things like the MP table, IRQ routing table, ACPI tables, etc.?
It can replace all of these (and actually get it right :-) )
Segher
On 12/27/06, Segher Boessenkool segher@kernel.crashing.org wrote:
Does it replace or supplement things like the MP table, IRQ routing table, ACPI tables, etc.?
It can replace all of these (and actually get it right :-) )
That's the good news. The bad news is that when Stefan and I have brought this up before, we've hit resistance. But I am hoping this time, given the code is here, the linux core guys are willing to accept OF support in all trees.
ron