Re: [Xen-devel] [PATCH 01/10] xen/arm: Implement hip04-d01 platform

[Frediano Ziglio <freddy77@xxxxxxxxx>]

2015-01-13 15:48 GMT+00:00 Frediano Ziglio <freddy77@xxxxxxxxx>:
> 2015-01-13 15:23 GMT+00:00 Ian Campbell <Ian.Campbell@xxxxxxxxxx>:
>> On Tue, 2015-01-13 at 15:11 +0000, Frediano Ziglio wrote:
>>> 2015-01-13 14:42 GMT+00:00 Ian Campbell <Ian.Campbell@xxxxxxxxxx>:
>>> > On Tue, 2015-01-13 at 14:09 +0000, Frediano Ziglio wrote:
>>> >> 2015-01-13 11:58 GMT+00:00 Ian Campbell <Ian.Campbell@xxxxxxxxxx>:
>>> >> > On Mon, 2014-11-03 at 10:11 +0000, Frediano Ziglio wrote:
>>> >> >> Add this new platform to Xen.
>>> >> >> This platform require specific code to initialize CPUs.
>>> >> >
>>> >> > What is the "bootwrapper"? Are you running this on real silicon or on 
>>> >> > an
>>> >> > emulator? Can the platform be made to do PSCI instead?
>>> >> >
>>> >>
>>> >> Very real. It's actually on my desk and I'm not in Matrix :-)
>>> >
>>> > OK. The choice of bootwrapper as a name is a bit unfortunate, since it
>>> > is already used for something else, but oh well.
>>> >
>>> >> Has no PSCI support. Would be honestly very great. As we (as company)
>>> >> write the firmware could be technically doable. There is no plan. This
>>> >> piece of software is meant to bring the CPU from Secure mode to
>>> >> Unsecure Hypervisor mode before calling kernel/hypervisor code and
>>> >> provide supervisor calls.
>>> >
>>> > Sounds a lot like PSCI to me, except non-standard ;-)
>>> >
>>> >> >> +    np_fab = dt_find_compatible_node(NULL, NULL, 
>>> >> >> "hisilicon,hip04-fabric");
>>> >> >
>>> >> > Please add a reference to the DT bindings document for these values.
>>> >> >
>>> >> > linux/Documentation/devicetree/bindings/arm/hisilicon/hisilicon.txt
>>> >> > seems related but doesn't talk about most of these fields.
>>> >> >
>>> >>
>>> >> There are documentation in the Linaro kernel, see
>>> >> https://git.linaro.org/kernel/linux-linaro-tracking.git/blob/HEAD:/Documentation/devicetree/bindings/arm/hisilicon/hisilicon.txt.
>>> >> I hope it will be merged soon.
>>> >
>>> > Thanks, but this doesn't seem to cover many of the properties used by
>>> > the code you are adding, e.g. bootwrapper-{size,magic},
>>> > relocation-{entry,size} (in fact it suggests they are part of a
>>> > boot-method array).
>>> >
>>> > I get the feeling these might be legacy/deprecated. Perhaps we could get
>>> > away without supporting such things?
>>> >
>>> > Ian.
>>> >
>>>
>>> No, this is the newest code.
>>
>> So in this:
>>     if ( !dt_property_read_u32(np, "bootwrapper-phys",
>>                                &hip04_boot.bootwrapper_phys) ) {
>> [...read boot-method array...]
>>     }
>>     else
>>     {
>> [... read a bunch of separate properties...]
>>     }
>>
>> Both sides of the if are used by non-legacy code? That's,...
>> unfortunate.
>>
>> I note that only the if case, not the else case, is covered by the
>> hisilicon.txt bindings...
>>
>> Ian.
>>
>
> Yes, you are right, there are actually two versions of these settings.
> Actually there has been a lot of bouncing between the two versions. We
> are also having some discussion on Linaro ML about implementing PSCI.
>

Hi,
   I finally finish to do a PSCI implementation in the boot wrapper
for this platform. You can find it at
https://github.com/freddy77/bootwrapper/tree/psci. It works fine with
Xen which is able to setup CPUs and reboot correctly without using
platform code.

I actually wondering why the branch at
http://xenbits.xen.org/gitweb/?p=people/julieng/xen-unstable.git;a=shortlog;h=refs/heads/hisilicon-d01-v3
was not merged into the master branch after the release. Actually
using the above boot wrapper I was able to remove two patches from the
patchset. See https://github.com/freddy77/xen/commits/frediano/use_psci.

Why this patch was not integrated?
http://xenbits.xen.org/gitweb/?p=people/julieng/xen-unstable.git;a=commitdiff;h=612816078b81e3b244d61843f0c27830bbf3da69.
It was not related specifically to board and already discussed.

Frediano

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel