Re: [PATCH v2 14/16] mfd: intel-lpss: Pass SDA hold time to I2C host controller driver

[Andy Shevchenko]

On Mon, Feb 8, 2016 at 12:29 PM, Mika Westerberg
<mika.westerberg@xxxxxxxxxxxxxxx> wrote:
> On Mon, Feb 08, 2016 at 11:09:43AM +0100, Wolfram Sang wrote:
>> On Mon, Nov 30, 2015 at 05:11:42PM +0200, Andy Shevchenko wrote:

>> There has been a request to get this into stable:
>>
>> https://bugzilla.kernel.org/show_bug.cgi?id=108581
>>
>> Since we probably don't want to backport the whole device_property
>> series, is there a simpler fix to backport for stable?
>
> Basically all it needs is to have:
>
>         dev->sda_hold_time = 28
>
> in dw_i2c_acpi_configure(). However, we cannot just unconditionally set
> it as it will break existing non-SKL systems so there should be some way
> to check if this is SKL or not. Perhaps we could look for DMI entries of
> the affected machines?
>
> BTW, this is not a regression in a sense as SKL I2C never worked before
> this series.

I could add that "backporting" something which is drastically
different to the original mainstream may probably prevent a smooth
backporting the fixes come later if any. Thus, I would try to port as
much closer as we can to the existing upstream code.

-- 
With Best Regards,
Andy Shevchenko