qemu-ppc
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH 4/4] xics: Merge TYPE_ICS_BASE and TYPE_ICS_SIMPLE classes


From: Cédric Le Goater
Subject: Re: [PATCH 4/4] xics: Merge TYPE_ICS_BASE and TYPE_ICS_SIMPLE classes
Date: Wed, 25 Sep 2019 08:04:30 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.1.0

On 25/09/2019 03:46, David Gibson wrote:
> On Tue, Sep 24, 2019 at 04:06:02PM +0200, Cédric Le Goater wrote:
>> On 24/09/2019 13:41, David Gibson wrote:
>>> On Tue, Sep 24, 2019 at 07:31:44AM +0200, Cédric Le Goater wrote:
>>>> On 24/09/2019 06:59, David Gibson wrote:
>>>>> TYPE_ICS_SIMPLE is the only subtype of TYPE_ICS_BASE that's ever
>>>>> instantiated, and the only one we're ever likely to want.  The
>>>>> existence of different classes is just a hang over from when we
>>>>> (misguidedly) had separate subtypes for the KVM and non-KVM version of
>>>>> the device.
>>>>>
>>>>> So, collapse the two classes together into just TYPE_ICS.
>>>>
>>>>
>>>> Well, I have been maintaining another subclass for the PHB3 MSI 
>>>> but it has never been merged and it will require some rework.
>>>
>>> Well, if you did do this again, is there an actual need for it to be a
>>> subclass of ICS_BASE, and not ICS_SIMPLE?  AFAICT the merged ICS class
>>> should be fine for pnv as well.
>>
>> the reject resend handlers might be an issue. Anyhow, let's merge this 
>> cleanup. PHB3 has been out of tree for too long.
> 
> Hrm, are you sure.  I remember thinking the other day "whatever
> happened to that PHB3 patchset?".  Is it actually broken,

It is not broken. 

PowerNV machines can boot rather complex PCI layouts on P8 (XICS) 
and P9 (XIVE). See the complex configuration examples here :

  https://github.com/legoater/qemu/wiki/PowerNV

> or has it just been a long time since it was posted, and therefore 
> been forgotten by me.

It hasn't been posted in a long time (+1 year). Here are the latest
exchanges we had in November: 

  http://patchwork.ozlabs.org/patch/951227/

Nothing really worrying : 

  - some XICS infrastructure are needed (currently being removed)
  - check if we need a separate source model for the MSI/LSI 
  - rework the PBCQ modeling to some extent.
  - misc cleanups.

I have been quite busy this last year. I would appreciate if someone 
could take ownership of the PHB part. 

C.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]