Re: [PATCH v2 3/4] watchdog: iTCO_wdt: Fix PMC GCR memory mapping failure

[sathyanarayanan kuppuswamy <sathyanarayanan.kuppuswamy@linux.intel.com>]

Hi Andy,


On 03/17/2017 11:38 AM, Andy Shevchenko wrote:
> On Fri, Mar 17, 2017 at 7:37 PM, sathyanarayanan kuppuswamy
> <sathyanarayanan.kuppuswamy@linux.intel.com> wrote:
>> On 03/17/2017 07:25 AM, Andy Shevchenko wrote:
>>> On Fri, Mar 17, 2017 at 3:40 PM, Guenter Roeck <linux@roeck-us.net> wrote:
>>>> On 03/17/2017 04:43 AM, Rajneesh Bhardwaj wrote:
>>>>> On Thu, Mar 16, 2017 at 05:41:35PM -0700, Kuppuswamy Sathyanarayanan
>>>>> wrote:
>>> I already asked once [1] to fix up the mess we have in PDx86 regarding SCU
>>> IPC.
>>> (PMC IPC how it's called is actually just a [main] part of SCU in newer
>>> SoCs).
>>>
>>> Rajneesh, Kuppuswamy,
>>> please pay attention on the below.
>>>
>>> We have two libraries doing almost the same (basics) one for old
>>> platforms, one for new.
>>>
>>> My vision what should be done before we go further is:
>>> 1. Split out common part from intel_scu_ipc and intel_pmc_ipc to some
>>> library.
>> I think we should create MFD driver for PMC and remove the redundant
>> resource and platform device creation codes.
>> Yes, there is common code in IPC implementation between scu_ipc and pmc_ipc
>> code. This needs be modularized.
>>
>> I can work on it and send a RFC patch for this cleanup. But it could take
>> more time for merging this cleanup patch.
>> So I think, in the mean time, we should merge this watchdog fix first to
>> remove iTCO watchdog device probe issue.
> I have heard already such excuses. Let's consider this as a "Last
> Chinese Warning".
>
> So, we consider reviewing applying *already floating around* patches
> in exchange to looking forward for clean up next.
> Do we have a deal?
Deal.
>
> Before you are going to implement anything in the code, please, share
> a document (architectural point of view) how you would see things
> should be done.
> Also consider to address PMC (Atom drivers) and P-Unit drivers which
> are related to SCU / IPC to have some structure.
Will send out the design document summarizing the issues we want to 
solve and a proposed design model.
>
>>> 2. Move headers to linux/platform_data/x86 for sharing with drivers
>>> that are supporting non-Intel / not-newest-Intel hardware.
>>> 3. Fix the mess inside the intel_pmc_ipc code (like use devm_()
>>> helpers where it makes sense, no use of global variables, etc)
>> Agreed.
>>>
>>> On top of that
>>> 4. Fix up Whiskey Cove PMIC code (See Hans' message [2] for the details)
>>>
>>> [1] Oops, it happened on internal mailing list Jan 27. And mentioned
>>> publicly after in a review on some patch here.
>>> [2] http://lkml.iu.edu/hypermail/linux/kernel/1702.3/01408.html

-- 
Sathyanarayanan Kuppuswamy
Android kernel developer