svn commit: r244899 - head/sys/mips/beri

[Nathan Whitehorn]

On 01/02/13 16:02, Andrew Turner wrote:
> On Wed, 2 Jan 2013 11:01:27 +0000
> "Robert N. M. Watson" <rwatson at FreeBSD.org> wrote:
>
>> On 1 Jan 2013, at 22:08, Andrew Turner wrote:
>>
>>>> On a semi-related note: the current obstacle to moving more devices
>>>> over to using FDT on BERI is that our FDT implementation appears to
>>>> require a PIC to be configured. We're not actually using a PIC on
>>>> BERI currently. It works fine attached to nexus, as the implied
>>>> fallback for not having a PIC is to simply use the suitably
>>>> numbered interrupt wires direct into the MIPS. However, trying the
>>>> same setup described using FDT leads to an interrupt-related
>>>> warning at boot, and no interrupt being provided to the driver. I
>>>> suspect I need to provide a PIC-alike software component as a
>>>> fall-back for the non-PIC case. From a brief e-mail exchange with
>>>> JC, it sounds like the XLP FDT setup is actually not using
>>>> interrupts either, currently.
>>> I'm not sure if a PIC is required. From my reading of the code it
>>> appears not. By the look of it if you are using the nexus to
>>> handle interrupts you need to put an interrupts property in the
>>> device and "#interrupt-cells = <1>;" in the soc node.
>>>
>>> You will also need to implement an interrupt decode function and
>>> add it to the fdt_pic_table array. ARM has a number of almost
>>> identical copies of this function you can use for inspration.
>>
>> This seemed to do the trick; what do you think of the attached? This
>> isn't a board-specific change, so I dropped it into the common
>> fdt_mips.c code. On the other hand, this left it a bit open as to
>> what the right compatible= line to use was, so feedback there most
>> welcome.
> The patch looks good. From my reading of [1] the compatible value
> should be something like "mips,mips4k" as it's value is of the form
> "<manufacturer>,<model>".
>
> I have been thinking the best way of merging these almost identical
> decode functions. Linux appears to do it by providing a per-controller
> function that can translate between the interrupt spec and the
> configuration allowing them to have a generic parsing function that
> doesn't need to check if the controller is compatible. I would like us
> to have something similar as it will remove the duplicate function.
>
> Andrew
>
> [1] http://www.devicetree.org/Device_Tree_Usage
PowerPC does something very much like this currently in the real OF
case. Looking at the contents of /sys/dev/fdt, it seems like there is a
lot of code duplication in particular with interrupts and PCI handling
compared to the equivalent parts of /sys/dev/ofw (and a few extra things
that are actually MI like /sys/powerpc/ofw_pci*). It would be very nice
if we could make FDT follow the normal OF codepaths instead of
(partially) reinventing the wheel.
-Nathan