[RFC PATCH for Juno 0/2] Drivers for Juno to boot from ACPI
Jon Masters
jcm at redhat.com
Mon Sep 15 16:00:43 PDT 2014
On 09/15/2014 06:57 PM, Grant Likely wrote:
> On Tue, 09 Sep 2014 12:55:09 +0200, Arnd Bergmann <arnd at arndb.de> wrote:
>> On Tuesday 09 September 2014 02:55:02 Jon Masters wrote:
>>> On 09/01/2014 11:29 AM, Arnd Bergmann wrote:
>>>> On Monday 01 September 2014 23:05:59 Hanjun Guo wrote:
>>>>> This patch set is example of the sort of driver changes needed to boot
>>>>> Juno using ACPI tables, which using the ACPI tables devloped for MS
>>>>> Windows and published by ARM [1].
>>>>
>>>> What about platform support?
>>>
>>> I think Juno is just serving as an example here since it's from ARM and
>>> thus various people have access to it already, etc. It's not the best
>>> example of a server platform out there, since that wasn't its original
>>> design purpose (but it's a wonderful platform nonetheless). There is
>>> work currently in flight to offer other reference ACPI examples.
>>
>> Ok, I see. This is a bit disappointing because it's actually the third
>> attempt to come up with a reference platform (after Exynos and X-Gene),
>> and it always seems to come down to deciding that this particular
>> platform doesn't really work with ACPI beyond basic booting.
>>
>> Which platform(s) are you talking about? Are you sure it will be better
>> suited than the first three?
>
> Juno is actually working out as a good reference platform. It works and
> there aren't horrible things that need to be done to get the system into
> a working state. AMD Seattle is a second system that will boot with
> Hanjun's ACPI series with a couple of device driver patches (the SBSA
> driver which is common with Juno and an Ethernet driver IIRC).
...and further there are also patches for Mustang that (this time) are
"done right" as another reference (relies upon updated UEFI firmware
that contains ACPI tables that don't have the kinds of hacks seen last
time - no clocks being exposed and handled through hacks or the kinds of
things that made Arnd uncomfortable). But perhaps let's take Seattle as
the second example and see what people think. There is a plan here at
Connect to have a Seattle example posted out within a day or so.
Jon.
More information about the linux-arm-kernel
mailing list