Date:   Sat, 11 Jan 2020 18:32:04 +0300
From:   Dmitry Osipenko <digetx@...il.com>
To:     Sowjanya Komatineni <skomatineni@...dia.com>,
        Sameer Pujar <spujar@...dia.com>, thierry.reding@...il.com,
        jonathanh@...dia.com, broonie@...nel.org, lgirdwood@...il.com,
        perex@...ex.cz, tiwai@...e.com, mperttunen@...dia.com,
        gregkh@...uxfoundation.org, sboyd@...nel.org, robh+dt@...nel.org,
        mark.rutland@....com
Cc:     pdeschrijver@...dia.com, pgaikwad@...dia.com, josephl@...dia.com,
        daniel.lezcano@...aro.org, mmaddireddy@...dia.com,
        markz@...dia.com, devicetree@...r.kernel.org,
        linux-clk@...r.kernel.org, linux-tegra@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v7 15/21] ASoC: tegra: Add fallback implementation for
 audio mclk

11.01.2020 02:14, Sowjanya Komatineni пишет:
> 
> On 1/10/20 3:02 PM, Dmitry Osipenko wrote:
>> External email: Use caution opening links or attachments
>>
>>
>> 11.01.2020 01:13, Sowjanya Komatineni пишет:
>>> On 1/10/20 2:05 PM, Dmitry Osipenko wrote:
>>>> External email: Use caution opening links or attachments
>>>>
>>>>
>>>> 10.01.2020 20:04, Sowjanya Komatineni пишет:
>>>>> On 1/9/20 10:52 AM, Sowjanya Komatineni wrote:
>>>>>> On 1/7/20 10:28 PM, Sameer Pujar wrote:
>>>>>>> On 1/8/2020 11:18 AM, Sowjanya Komatineni wrote:
>>>>>>>> On 1/7/20 9:34 PM, Sameer Pujar wrote:
>>>>>>>>> On 1/8/2020 9:55 AM, Sowjanya Komatineni wrote:
>>>>>>>>>> mclk is from clk_out_1 which is part of Tegra PMC block and pmc
>>>>>>>>>> clocks
>>>>>>>>>> are moved to Tegra PMC driver with pmc as clock provider and
>>>>>>>>>> using
>>>>>>>>>> pmc
>>>>>>>>>> clock ids.
>>>>>>>>>>
>>>>>>>>>> New device tree uses clk_out_1 from pmc clock provider.
>>>>>>>>>>
>>>>>>>>>> So, this patch adds implementation for mclk fallback to extern1
>>>>>>>>>> when
>>>>>>>>>> retrieving mclk returns -ENOENT to be backward compatible of new
>>>>>>>>>> device
>>>>>>>>>> tree with older kernels.
>>>>>>>>>>
>>>>>>>>>> Tested-by: Dmitry Osipenko <digetx@...il.com>
>>>>>>>>>> Reviewed-by: Dmitry Osipenko <digetx@...il.com>
>>>>>>>>>> Signed-off-by: Sowjanya Komatineni <skomatineni@...dia.com>
>>>>>>>>>> ---
>>>>>>>>>>     sound/soc/tegra/tegra_asoc_utils.c | 11 ++++++++++-
>>>>>>>>>>     1 file changed, 10 insertions(+), 1 deletion(-)
>>>>>>>>>>
>>>>>>>>>> diff --git a/sound/soc/tegra/tegra_asoc_utils.c
>>>>>>>>>> b/sound/soc/tegra/tegra_asoc_utils.c
>>>>>>>>>> index 9cfebef74870..9a5f81039491 100644
>>>>>>>>>> --- a/sound/soc/tegra/tegra_asoc_utils.c
>>>>>>>>>> +++ b/sound/soc/tegra/tegra_asoc_utils.c
>>>>>>>>>> @@ -183,7 +183,16 @@ int tegra_asoc_utils_init(struct
>>>>>>>>>> tegra_asoc_utils_data *data,
>>>>>>>>>>         data->clk_cdev1 = devm_clk_get(dev, "mclk");
>>>>>>>>>>         if (IS_ERR(data->clk_cdev1)) {
>>>>>>>>>>             dev_err(data->dev, "Can't retrieve clk cdev1\n");
>>>>>>>>> This error print can be moved inside below if, when this actually
>>>>>>>>> meant to be an error condition.
>>>>>>>>>
>>>>>>>> Want to show error even if mclk retrieval returns ENOENT to clearly
>>>>>>>> indicate mclk does not exist along with message of falling back to
>>>>>>>> extern1.
>>>>>>> Yes, but falling back essentially means 'mclk' is not available and
>>>>>>> fallback print is not an error.
>>>>>>> Not a major issue though, you can consider updating. Otherwise LGTM.
>>>>>>>
>>>>>> Will update
>>>>>>>>>> -        return PTR_ERR(data->clk_cdev1);
>>>>>>>>>> +        if (PTR_ERR(data->clk_cdev1) != -ENOENT)
>>>>>>>>>> +            return PTR_ERR(data->clk_cdev1);
>>>>>>>>>> +        /* Fall back to extern1 */
>>>>>>>>>> +        data->clk_cdev1 = devm_clk_get(dev, "extern1");
>>>>>>>>>> +        if (IS_ERR(data->clk_cdev1)) {
>>>>>>>>>> +            dev_err(data->dev, "Can't retrieve clk extern1\n");
>>>>>>>>>> +            return PTR_ERR(data->clk_cdev1);
>>>>>>>>>> +        }
>>>>>>>>>> +
>>>>>>>>>> +        dev_err(data->dev, "Falling back to extern1\n");
>>>>>>>>> This can be a info print?
>>>>>> Will use dev_info
>>>>>>>>>>         }
>>>>>>>>>>           /*
>>>>>> Dmitry/Rob, there was discussion in v3 regarding backporting mclk
>>>>>> fallback.
>>>>>>
>>>>>> Dmitry wanted Rob to confirm on this
>>>>>>
>>>>>> I think openSUSE Leap could be one of those distros that use LTS
>>>>>> kernel
>>>>>> with newer device-trees, but that's not 100%. Maybe Rob could help
>>>>>> clarifying that.
>>>>>>
>>>>>> Dmitry/Rob, Can you please confirm if mclk fallback patch need
>>>>>> backport to have new device tree work with old kernels?
>>>>>>
>>>>> Dmitry,
>>>>>
>>>>> Can you please confirm if we need to backport this mclk fallback
>>>>> patch?
>>>>>
>>>> Seems only T210 was making use of the CaR's TEGRA*_CLK_CLK_OUT_*, thus
>>>> the backporting isn't needed.
>>> Thanks Dmitry
>>>> Also, please use 'git rebase --exec make ...' to make sure that all
>>>> patches are compiling without problems. The removal of the legacy clock
>>>> IDs should be done after the device-trees changes, otherwise it looks
>>>> like DTBs compilation will fail. It's possible that the order of the
>>>> patches could be changed if Thierry will chose to split this series
>>>> into
>>>> several pull requests, nevertheless all patches should compile and work
>>>> in the original order.
>>> OK, Will move patches of device tree updates to use new DT ID prior to
>>> removal of old ID.
>> Oh, wait. But then the newer device-trees won't work with the stable
>> kernels, so it actually won't hurt to backport this change.
> ok will add Fixes tag to have this mclk fallback patch backported.
>>
>> Secondly, please move the "Use device managed resource APIs to get the
>> clock" after the ASoC patches with the stable tags, such that the stable
>> patches could be applied cleanly.
> OK
>>
>> Lastly, please separate the assigned-clocks change from the the audio
>> mclk enable/disable into a standalone patch. These changes are not
>> interdependent, unless I'm missing something.
> 
> But parent configuration when assigned-clock-parents are not in DT are
> needed along with mclk enable
> 
> as we are removing audio clocks parent configuration and enabling them
> together from clock driver.
> 
> So doesn't both parent configuration and enabling mclk together need to
> be in same patch to match what we are removing from clock driver?
> 

All current stable kernels happen to work without any visible problems
because of the non-critical clk-enable refcounting bug that masks the
problem. Thus the mclk will be enabled in stable kernels without any
extra changes and the assigned-clock-parents shouldn't affect that.

Please make sure that every patch in this series:

1) Compiles without any errors and warnings.

2) Works, i.e. you should be able to checkout any commit and kernel
should boot/work without any regressions.

3) Stable patches could be cherry-picked into stable kernels without
merge conflicts.

To achieve that you'll need to sort patches in the correct order and do
some basic testing.

Powered by blists - more mailing lists