Re: [PATCH v4 13/19] ASoC: tegra: Add fallback implementation for audio mclk

From: Dmitry Osipenko
Date: Wed Dec 18 2019 - 11:29:19 EST


18.12.2019 18:43, Sowjanya Komatineni ÐÐÑÐÑ:
>
> On 12/17/19 11:31 PM, Dmitry Osipenko wrote:
>> 18.12.2019 10:22, Dmitry Osipenko ÐÐÑÐÑ:
>>> 18.12.2019 10:14, Sowjanya Komatineni ÐÐÑÐÑ:
>>>> On 12/17/19 11:01 PM, Dmitry Osipenko wrote:
>>>>> 18.12.2019 09:59, Dmitry Osipenko ÐÐÑÐÑ:
>>>>>> 17.12.2019 23:04, Sowjanya Komatineni ÐÐÑÐÑ:
>>>>>>> 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.
>>>>>>>
>>>>>>> Signed-off-by: Sowjanya Komatineni <skomatineni@xxxxxxxxxx>
>>>>>>> ---
>>>>>>> ÂÂ sound/soc/tegra/tegra_asoc_utils.c | 11 ++++++++++-
>>>>>>> ÂÂ 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 fe9ca8acd0fb..1b88c6043082 100644
>>>>>>> --- a/sound/soc/tegra/tegra_asoc_utils.c
>>>>>>> +++ b/sound/soc/tegra/tegra_asoc_utils.c
>>>>>>> @@ -191,7 +191,16 @@ int tegra_asoc_utils_init(struct
>>>>>>> tegra_asoc_utils_data *data,
>>>>>>> ÂÂÂÂÂÂ data->clk_cdev1 = devm_clk_get(dev, "mclk");
>>>>>>> ÂÂÂÂÂÂ if (IS_ERR(data->clk_cdev1)) {
>>>>>>> ÂÂÂÂÂÂÂÂÂÂ dev_err(data->dev, "Can't retrieve clk cdev1\n");
>>>>>>> -ÂÂÂÂÂÂÂ return PTR_ERR(data->clk_cdev1);
>>>>>>> +ÂÂÂÂÂÂÂ if (PTR_ERR(data->clk_cdev1) != -ENOENT)
>>>>>>> +ÂÂÂÂÂÂÂÂÂÂÂ return PTR_ERR(data->clk_cdev1);
>>>>>>> +ÂÂÂÂÂÂÂ /* Fall back to extern1 */
>>>>>>> +ÂÂÂÂÂÂÂ data->clk_cdev1 = devm_clk_get(dev, "extern1");
>>>>>>> +ÂÂÂÂÂÂÂ if (IS_ERR(data->clk_cdev1)) {
>>>>>>> +ÂÂÂÂÂÂÂÂÂÂÂ dev_err(data->dev, "Can't retrieve clk extern1\n");
>>>>>>> +ÂÂÂÂÂÂÂÂÂÂÂ return PTR_ERR(data->clk_cdev1);
>>>>>>> +ÂÂÂÂÂÂÂ }
>>>>>>> +
>>>>>>> +ÂÂÂÂÂÂÂ dev_err(data->dev, "Falling back to extern1\n");
>>>>>>> ÂÂÂÂÂÂ }
>>>>>>> ÂÂ ÂÂÂÂÂ /*
>>>>>>>
>>>>>> [ÂÂÂ 1.769091] ------------[ cut here ]------------
>>>>>> [ÂÂÂ 1.769249] WARNING: CPU: 2 PID: 1 at drivers/clk/clk.c:954
>>>>>> clk_core_disable+0xa5/0x1d4
>>>>>> [ÂÂÂ 1.769330] clk_out_1 already disabled
>>>>>> [ÂÂÂ 1.769459] Modules linked in:
>>>>>> [ÂÂÂ 1.769541] CPU: 2 PID: 1 Comm: swapper/0 Not tainted
>>>>>> 5.5.0-rc1-next-20191213-00167-g6b9fbcdac8f3-dirty #266
>>>>>> [ÂÂÂ 1.769676] Hardware name: NVIDIA Tegra SoC (Flattened Device
>>>>>> Tree)
>>>>>> [ÂÂÂ 1.769775] [<c010e4bd>] (unwind_backtrace) from [<c010a0fd>]
>>>>>> (show_stack+0x11/0x14)
>>>>>> [ÂÂÂ 1.769918] [<c010a0fd>] (show_stack) from [<c09a37b1>]
>>>>>> (dump_stack+0x85/0x94)
>>>>>> [ÂÂÂ 1.770061] [<c09a37b1>] (dump_stack) from [<c011f3d1>]
>>>>>> (__warn+0xc1/0xc4)
>>>>>> [ÂÂÂ 1.770144] [<c011f3d1>] (__warn) from [<c011f691>]
>>>>>> (warn_slowpath_fmt+0x61/0x78)
>>>>>> [ÂÂÂ 1.770285] [<c011f691>] (warn_slowpath_fmt) from [<c04a0e7d>]
>>>>>> (clk_core_disable+0xa5/0x1d4)
>>>>>> [ÂÂÂ 1.770427] [<c04a0e7d>] (clk_core_disable) from [<c04a0fc3>]
>>>>>> (clk_core_disable_lock+0x17/0x20)
>>>>>> [ÂÂÂ 1.770516] [<c04a0fc3>] (clk_core_disable_lock) from [<c07792bb>]
>>>>>> (tegra_asoc_utils_set_rate+0x53/0x208)
>>>>>> [ÂÂÂ 1.770662] [<c07792bb>] (tegra_asoc_utils_set_rate) from
>>>>>> [<c077b8c5>] (tegra_rt5640_probe+0xd5/0x128)
>>>>>> [ÂÂÂ 1.770808] [<c077b8c5>] (tegra_rt5640_probe) from [<c0555eb7>]
>>>>>> (platform_drv_probe+0x33/0x68)
>>>>>> [ÂÂÂ 1.770958] [<c0555eb7>] (platform_drv_probe) from [<c055471d>]
>>>>>> (really_probe+0x14d/0x240)
>>>>>> [ÂÂÂ 1.771099] [<c055471d>] (really_probe) from [<c055493f>]
>>>>>> (driver_probe_device+0x43/0x11c)
>>>>>> [ÂÂÂ 1.771187] [<c055493f>] (driver_probe_device) from [<c0554b25>]
>>>>>> (device_driver_attach+0x3d/0x40)
>>>>>> [ÂÂÂ 1.771328] [<c0554b25>] (device_driver_attach) from [<c0554b5f>]
>>>>>> (__driver_attach+0x37/0x78)
>>>>>> [ÂÂÂ 1.771469] [<c0554b5f>] (__driver_attach) from [<c05532fb>]
>>>>>> (bus_for_each_dev+0x43/0x6c)
>>>>>> [ÂÂÂ 1.771609] [<c05532fb>] (bus_for_each_dev) from [<c0553e0f>]
>>>>>> (bus_add_driver+0xe3/0x148)
>>>>>> [ÂÂÂ 1.771692] [<c0553e0f>] (bus_add_driver) from [<c055531d>]
>>>>>> (driver_register+0x39/0xa0)
>>>>>> [ÂÂÂ 1.771833] [<c055531d>] (driver_register) from [<c0102c2f>]
>>>>>> (do_one_initcall+0x43/0x1bc)
>>>>>> [ÂÂÂ 1.771979] [<c0102c2f>] (do_one_initcall) from [<c1000ce5>]
>>>>>> (kernel_init_freeable+0x121/0x194)
>>>>>> [ÂÂÂ 1.772129] [<c1000ce5>] (kernel_init_freeable) from [<c09b40e9>]
>>>>>> (kernel_init+0xd/0xd0)
>>>>>> [ÂÂÂ 1.772215] [<c09b40e9>] (kernel_init) from [<c01010bd>]
>>>>>> (ret_from_fork+0x11/0x34)
>>>>>> [ÂÂÂ 1.772349] Exception stack(0xde907fb0 to 0xde907ff8)
>>>>>>
>>>>> Although, that's probably related to the "ASoC: tegra: Add initial
>>>>> parent configuration for audio mclk".
>>>>>
>>>> Actually I see these warnings of already unprepared and already
>>>> disabled
>>>> for pll_a, pll_a_out0, and clk_out_1 even without this whole patch
>>>> series as well.
>>>>
>>>> I think its from tegra_asoc_utils_set_rate() doing
>>>> clk_disable_unprepare
>>>> and these clocks are already unprepared and disabled so its just
>>>> warning
>>>> from clk_core_unprepare and clk_core_disable.
>>> Doesn't happen for me without this series.
>
> I looked at wrong log, right earlier clock driver keeps them enabled so
> asoc_utils_set_rate() disables the clock fine but now enabling audio
> clock should be done in asoc_utils_init() to let the
> clk_disable_unprepare from asoc_utils_set_rate not to show this warning.
>
> But actually we don't need to have clock enabled in asoc_utils_init
> prior to invoking asoc_utils_set_rate from utils_init and its just warns
> during sound driver probe because clock is already in disabled state. At
> same time it doesn't harm to have it kept enabled in utils_init. So will
> keep it enabled in asoc_utils_init to prevent this warning to show up.
>
>> But sound works with both old/new device-trees.

The rule of thumb: don't enable anything when it shouldn't be enabled.
If clocks are disabled at the time of drivers probe, then drivers
shouldn't disable the clocks.