mbox series

[0/4] clk: qcom : add sm8250 LPASS GFM drivers

Message ID 20200917132850.7730-1-srinivas.kandagatla@linaro.org
Headers show
Series clk: qcom : add sm8250 LPASS GFM drivers | expand

Message

Srinivas Kandagatla Sept. 17, 2020, 1:28 p.m. UTC
This patchset adds support for GFM Muxes found in LPASS
(Low Power Audio SubSystem) IP in Audio Clock Controller
and Always ON clock controller.

Clocks derived from these muxes are consumed by LPASS Digital Codec.
Currently the driver for Audio and Always ON clock controller only
supports GFM Muxes, however it should be easy to add more clock
support when required.

Srinivas Kandagatla (4):
  dt-bindings: clock: Add support for LPASS Audio Clock Controller
  dt-bindings: clock: Add support for LPASS Always ON Controller
  clk: qcom: Add support to LPASS AUDIO_CC Glitch Free Mux clocks
  clk: qcom: Add support to LPASS AON_CC Glitch Free Mux clocks

 .../bindings/clock/qcom,aoncc-sm8250.yaml     |  58 ++++
 .../bindings/clock/qcom,audiocc-sm8250.yaml   |  58 ++++
 drivers/clk/qcom/Kconfig                      |   7 +
 drivers/clk/qcom/Makefile                     |   1 +
 drivers/clk/qcom/lpass-gfm-sm8250.c           | 296 ++++++++++++++++++
 .../clock/qcom,sm8250-lpass-aoncc.h           |  11 +
 .../clock/qcom,sm8250-lpass-audiocc.h         |  13 +
 7 files changed, 444 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/clock/qcom,aoncc-sm8250.yaml
 create mode 100644 Documentation/devicetree/bindings/clock/qcom,audiocc-sm8250.yaml
 create mode 100644 drivers/clk/qcom/lpass-gfm-sm8250.c
 create mode 100644 include/dt-bindings/clock/qcom,sm8250-lpass-aoncc.h
 create mode 100644 include/dt-bindings/clock/qcom,sm8250-lpass-audiocc.h

Comments

Rob Herring (Arm) Sept. 18, 2020, 5:29 p.m. UTC | #1
On Thu, 17 Sep 2020 14:28:48 +0100, Srinivas Kandagatla wrote:
> Always ON Clock controller is a block inside LPASS which controls
> 1 Glitch free muxes to LPASS codec Macros.
> 
> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
> ---
>  .../bindings/clock/qcom,aoncc-sm8250.yaml     | 58 +++++++++++++++++++
>  .../clock/qcom,sm8250-lpass-aoncc.h           | 11 ++++
>  2 files changed, 69 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/clock/qcom,aoncc-sm8250.yaml
>  create mode 100644 include/dt-bindings/clock/qcom,sm8250-lpass-aoncc.h
> 


My bot found errors running 'make dt_binding_check' on your patch:

Error: Documentation/devicetree/bindings/clock/qcom,aoncc-sm8250.example.dts:25.30-31 syntax error
FATAL ERROR: Unable to parse input tree
make[1]: *** [scripts/Makefile.lib:342: Documentation/devicetree/bindings/clock/qcom,aoncc-sm8250.example.dt.yaml] Error 1
make[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:1366: dt_binding_check] Error 2


See https://patchwork.ozlabs.org/patch/1366126

If you already ran 'make dt_binding_check' and didn't see the above
error(s), then make sure dt-schema is up to date:

pip3 install git+https://github.com/devicetree-org/dt-schema.git@master --upgrade

Please check and re-submit.
Stephen Boyd Sept. 22, 2020, 6:41 p.m. UTC | #2
Quoting Srinivas Kandagatla (2020-09-17 06:28:49)
> GFM Muxes in AUDIO_CC control clocks to LPASS WSA and RX Codec Macros.
> This patch adds support to these muxes.
> 
> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
> ---
>  drivers/clk/qcom/Kconfig            |   7 +
>  drivers/clk/qcom/Makefile           |   1 +
>  drivers/clk/qcom/lpass-gfm-sm8250.c | 235 ++++++++++++++++++++++++++++
>  3 files changed, 243 insertions(+)
>  create mode 100644 drivers/clk/qcom/lpass-gfm-sm8250.c
> 
> diff --git a/drivers/clk/qcom/Kconfig b/drivers/clk/qcom/Kconfig
> index 2eda63b7c46c..4e46fd339e62 100644
> --- a/drivers/clk/qcom/Kconfig
> +++ b/drivers/clk/qcom/Kconfig
> @@ -494,4 +494,11 @@ config KRAITCC
>           Support for the Krait CPU clocks on Qualcomm devices.
>           Say Y if you want to support CPU frequency scaling.
>  
> +config CLK_GFM_LPASS_SM8250
> +       tristate "GFM LPASS Clocks"
> +       depends on SND_SOC_QDSP6_COMMON

What is the depends for?

> +       help
> +         Support for the GFM Glitch Free Mux LPASS clock. Say Y
> +         if you want to support GFM Clocks on LPASS for SM8250 SoC.
> +
>  endif
> diff --git a/drivers/clk/qcom/Makefile b/drivers/clk/qcom/Makefile
> index 8eb395d02a32..c18e4ead6c9e 100644
> --- a/drivers/clk/qcom/Makefile
> +++ b/drivers/clk/qcom/Makefile
> @@ -74,3 +74,4 @@ obj-$(CONFIG_SPMI_PMIC_CLKDIV) += clk-spmi-pmic-div.o
>  obj-$(CONFIG_KPSS_XCC) += kpss-xcc.o
>  obj-$(CONFIG_QCOM_HFPLL) += hfpll.o
>  obj-$(CONFIG_KRAITCC) += krait-cc.o
> +obj-$(CONFIG_CLK_GFM_LPASS_SM8250) += lpass-gfm-sm8250.o

I don't know but probably this should be sorted by Kconfig instead of
tacked onto the end here.

> diff --git a/drivers/clk/qcom/lpass-gfm-sm8250.c b/drivers/clk/qcom/lpass-gfm-sm8250.c
> new file mode 100644
> index 000000000000..2d5c41ae4969
> --- /dev/null
> +++ b/drivers/clk/qcom/lpass-gfm-sm8250.c
> @@ -0,0 +1,235 @@
> +// SPDX-License-Identifier: GPL-2.0
> +
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include <linux/clk-provider.h>
> +#include <linux/io.h>
> +#include <linux/clk.h>
> +#include <linux/slab.h>
> +#include <linux/err.h>
> +#include <linux/notifier.h>
> +#include <linux/device.h>
> +#include <linux/platform_device.h>
> +#include <linux/of_device.h>
> +#include <dt-bindings/clock/qcom,sm8250-lpass-audiocc.h>
> +
> +struct lpass_gfm {
> +       struct device *dev;
> +       void __iomem *base;
> +       struct clk *core_vote;
> +       struct clk *bus_clk;
> +};
> +
> +struct clk_gfm {
> +       unsigned int mux_reg;
> +       unsigned int mux_mask;
> +       struct clk_hw   hw;
> +       struct lpass_gfm *priv;
> +       void __iomem *gfm_mux;
> +};
> +
> +#define GFM_MASK       BIT(1)
> +#define to_clk_gfm(_hw) container_of(_hw, struct clk_gfm, hw)
> +
> +static u8 clk_gfm_get_parent(struct clk_hw *hw)
> +{
> +       struct clk_gfm *clk = to_clk_gfm(hw);
> +
> +       return readl(clk->gfm_mux) & GFM_MASK;
> +}
> +
> +static int clk_gfm_set_parent(struct clk_hw *hw, u8 index)
> +{
> +       struct clk_gfm *clk = to_clk_gfm(hw);
> +       unsigned int val;
> +
> +       val = readl(clk->gfm_mux);
> +
> +       if (index)
> +               val |= GFM_MASK;
> +       else
> +               val &= ~GFM_MASK;
> +
> +       writel(val, clk->gfm_mux);
> +
> +       return 0;
> +}
> +
> +static const struct clk_ops clk_gfm_ops = {
> +       .get_parent = clk_gfm_get_parent,
> +       .set_parent = clk_gfm_set_parent,
> +       .determine_rate = __clk_mux_determine_rate,
> +};
> +
> +static struct clk_gfm lpass_gfm_wsa_mclk = {
> +       .mux_reg = 0x220d8,
> +       .mux_mask = BIT(0),
> +       .hw.init = &(struct clk_init_data) {
> +               .name = "WSA_MCLK",
> +               .ops = &clk_gfm_ops,
> +               .flags = CLK_SET_RATE_PARENT | CLK_OPS_PARENT_ENABLE,
> +               .parent_names = (const char *[]){
> +                       "LPASS_CLK_ID_TX_CORE_MCLK",
> +                       "LPASS_CLK_ID_WSA_CORE_MCLK",
> +               },
> +               .parent_data = (const struct clk_parent_data[]){
> +                               { .index = 0 },
> +                               { .index = 1 },
> +               },
> +               .num_parents = 2,
> +       },
> +};
> +
> +static struct clk_gfm lpass_gfm_wsa_npl = {
> +       .mux_reg = 0x220d8,
> +       .mux_mask = BIT(0),
> +       .hw.init = &(struct clk_init_data) {
> +               .name = "WSA_NPL",
> +               .ops = &clk_gfm_ops,
> +               .flags = CLK_SET_RATE_PARENT | CLK_OPS_PARENT_ENABLE,
> +               .parent_names = (const char *[]){
> +                       "LPASS_CLK_ID_TX_CORE_NPL_MCLK",
> +                       "LPASS_CLK_ID_WSA_CORE_NPL_MCLK",
> +               },
> +               .parent_data = (const struct clk_parent_data[]){
> +                               { .index = 0 },
> +                               { .index = 1 },
> +               },
> +               .num_parents = 2,
> +       },
> +};
> +
> +static struct clk_gfm lpass_gfm_rx_mclk_mclk2 = {
> +       .mux_reg = 0x240d8,
> +       .mux_mask = BIT(0),
> +       .hw.init = &(struct clk_init_data) {
> +               .name = "RX_MCLK_MCLK2",
> +               .ops = &clk_gfm_ops,
> +               .flags = CLK_SET_RATE_PARENT | CLK_OPS_PARENT_ENABLE,
> +               .parent_names = (const char *[]){
> +                       "LPASS_CLK_ID_TX_CORE_MCLK",
> +                       "LPASS_CLK_ID_RX_CORE_MCLK",
> +               },
> +               .parent_data = (const struct clk_parent_data[]){
> +                               { .index = 0 },
> +                               { .index = 1 },
> +               },
> +               .num_parents = 2,
> +       },
> +};
> +
> +static struct clk_gfm lpass_gfm_rx_npl = {
> +       .mux_reg = 0x240d8,
> +       .mux_mask = BIT(0),
> +       .hw.init = &(struct clk_init_data) {
> +               .name = "RX_NPL",
> +               .ops = &clk_gfm_ops,
> +               .flags = CLK_SET_RATE_PARENT | CLK_OPS_PARENT_ENABLE,
> +               .parent_names = (const char *[]){
> +                       "LPASS_CLK_ID_TX_CORE_NPL_MCLK",
> +                       "LPASS_CLK_ID_RX_CORE_NPL_MCLK",
> +               },
> +               .parent_data = (const struct clk_parent_data[]){
> +                               { .index = 0 },
> +                               { .index = 1 },
> +               },

Please use parent_data instead of parent_names. The name can go in the
parent_data struct.

> +               .num_parents = 2,
> +       },
> +};
> +
> +static struct clk_gfm *audiocc_gfm_clks[] = {
> +       [LPASS_CDC_WSA_NPL]             = &lpass_gfm_wsa_npl,
> +       [LPASS_CDC_WSA_MCLK]            = &lpass_gfm_wsa_mclk,
> +       [LPASS_CDC_RX_NPL]              = &lpass_gfm_rx_npl,
> +       [LPASS_CDC_RX_MCLK_MCLK2]       = &lpass_gfm_rx_mclk_mclk2,
> +};
> +
> +static struct clk_hw_onecell_data audiocc_hw_onecell_data = {
> +       .hws = {
> +               [LPASS_CDC_WSA_NPL]     = &lpass_gfm_wsa_npl.hw,
> +               [LPASS_CDC_WSA_MCLK]    = &lpass_gfm_wsa_mclk.hw,
> +               [LPASS_CDC_RX_NPL]      = &lpass_gfm_rx_npl.hw,
> +               [LPASS_CDC_RX_MCLK_MCLK2] = &lpass_gfm_rx_mclk_mclk2.hw,
> +       },
> +       .num = ARRAY_SIZE(audiocc_gfm_clks),
> +};
> +
> +struct lpass_gfm_data {
> +       struct clk_hw_onecell_data *onecell_data;
> +       struct clk_gfm **gfm_clks;
> +};
> +
> +static struct lpass_gfm_data audiocc_data = {
> +       .onecell_data = &audiocc_hw_onecell_data,
> +       .gfm_clks = audiocc_gfm_clks,
> +};
> +
> +static int lpass_gfm_clk_driver_probe(struct platform_device *pdev)
> +{
> +       const struct lpass_gfm_data *data;
> +       struct device *dev = &pdev->dev;
> +       struct resource *res;
> +       struct clk_gfm *gfm;
> +       struct lpass_gfm *cc;
> +       int err, i;
> +
> +       cc = devm_kzalloc(dev, sizeof(*cc), GFP_KERNEL);
> +       if (!cc)
> +               return -ENOMEM;
> +
> +       cc->core_vote = devm_clk_get(&pdev->dev, "core");
> +       if (IS_ERR(cc->core_vote)) {
> +               dev_dbg(dev, "Failed to get lpass core clk\n");
> +               return PTR_ERR(cc->core_vote);
> +       }

Can this use the pm_clk stuff?

> +
> +       data = of_device_get_match_data(dev);

What if data is NULL?

> +       cc->bus_clk = devm_clk_get(&pdev->dev, "bus");
> +       if (IS_ERR(cc->bus_clk)) {
> +               dev_dbg(dev, "Failed to get lpass bus clk\n");
> +               return PTR_ERR(cc->bus_clk);
> +       }
> +
> +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> +       cc->base = devm_ioremap_resource(dev, res);
> +       if (IS_ERR(cc->base))
> +               return PTR_ERR(cc->base);
> +
> +       clk_prepare_enable(cc->core_vote);
> +       clk_prepare_enable(cc->bus_clk);

These will always be left enabled if the probe function bails out after here.

> +
> +       for (i = 0; i < data->onecell_data->num; i++) {
> +               if (!data->gfm_clks[i])
> +                       continue;
> +
> +               gfm = data->gfm_clks[i];
> +               gfm->priv = cc;
> +               gfm->gfm_mux = cc->base;
> +               gfm->gfm_mux = gfm->gfm_mux + data->gfm_clks[i]->mux_reg;
> +
> +               err = devm_clk_hw_register(dev, &data->gfm_clks[i]->hw);
> +               if (err)
> +                       return err;
> +
> +       }
> +
> +       return devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
> +                                          data->onecell_data);
> +}
> +
> +static const struct of_device_id lpass_gfm_clk_match_table[] = {
> +       {
> +               .compatible = "qcom,sm8250-lpass-audiocc",
> +               .data = &audiocc_data,
> +       },
> +       { }
> +};
> +
> +static struct platform_driver lpass_gfm_clk_driver = {
> +       .probe          = lpass_gfm_clk_driver_probe,
> +       .driver         = {
> +               .name   = "lpass-gfm-clk",
> +               .of_match_table = lpass_gfm_clk_match_table,
> +       },
> +};
> +builtin_platform_driver(lpass_gfm_clk_driver);

Should be module_platform_driver()?
Srinivas Kandagatla Sept. 22, 2020, 9:45 p.m. UTC | #3
Thanks Stephen for review,

On 22/09/2020 19:41, Stephen Boyd wrote:
> Quoting Srinivas Kandagatla (2020-09-17 06:28:49)
>> GFM Muxes in AUDIO_CC control clocks to LPASS WSA and RX Codec Macros.
>> This patch adds support to these muxes.
>>
>> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
>> ---
>>   drivers/clk/qcom/Kconfig            |   7 +
>>   drivers/clk/qcom/Makefile           |   1 +
>>   drivers/clk/qcom/lpass-gfm-sm8250.c | 235 ++++++++++++++++++++++++++++
>>   3 files changed, 243 insertions(+)
>>   create mode 100644 drivers/clk/qcom/lpass-gfm-sm8250.c
>>
>> diff --git a/drivers/clk/qcom/Kconfig b/drivers/clk/qcom/Kconfig
>> index 2eda63b7c46c..4e46fd339e62 100644
>> --- a/drivers/clk/qcom/Kconfig
>> +++ b/drivers/clk/qcom/Kconfig
>> @@ -494,4 +494,11 @@ config KRAITCC
>>            Support for the Krait CPU clocks on Qualcomm devices.
>>            Say Y if you want to support CPU frequency scaling.
>>   
>> +config CLK_GFM_LPASS_SM8250
>> +       tristate "GFM LPASS Clocks"
>> +       depends on SND_SOC_QDSP6_COMMON
> 
> What is the depends for?
I think I forgot to remove this!

> 
>> +       help
>> +         Support for the GFM Glitch Free Mux LPASS clock. Say Y
>> +         if you want to support GFM Clocks on LPASS for SM8250 SoC.
>> +
>>   endif
>> diff --git a/drivers/clk/qcom/Makefile b/drivers/clk/qcom/Makefile
>> index 8eb395d02a32..c18e4ead6c9e 100644
>> --- a/drivers/clk/qcom/Makefile
>> +++ b/drivers/clk/qcom/Makefile
>> @@ -74,3 +74,4 @@ obj-$(CONFIG_SPMI_PMIC_CLKDIV) += clk-spmi-pmic-div.o
>>   obj-$(CONFIG_KPSS_XCC) += kpss-xcc.o
>>   obj-$(CONFIG_QCOM_HFPLL) += hfpll.o
>>   obj-$(CONFIG_KRAITCC) += krait-cc.o
>> +obj-$(CONFIG_CLK_GFM_LPASS_SM8250) += lpass-gfm-sm8250.o
> 
> I don't know but probably this should be sorted by Kconfig instead of
> tacked onto the end here.
> 
>> diff --git a/drivers/clk/qcom/lpass-gfm-sm8250.c b/drivers/clk/qcom/lpass-gfm-sm8250.c
>> new file mode 100644
>> index 000000000000..2d5c41ae4969
>> --- /dev/null
>> +++ b/drivers/clk/qcom/lpass-gfm-sm8250.c
>> @@ -0,0 +1,235 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +
>> +#include <linux/kernel.h>
>> +#include <linux/module.h>
>> +#include <linux/clk-provider.h>
>> +#include <linux/io.h>
>> +#include <linux/clk.h>
>> +#include <linux/slab.h>
>> +#include <linux/err.h>
>> +#include <linux/notifier.h>
>> +#include <linux/device.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/of_device.h>
>> +#include <dt-bindings/clock/qcom,sm8250-lpass-audiocc.h>
>> +
>> +struct lpass_gfm {
>> +       struct device *dev;
>> +       void __iomem *base;
>> +       struct clk *core_vote;
>> +       struct clk *bus_clk;
>> +};
>> +
>> +struct clk_gfm {
>> +       unsigned int mux_reg;
>> +       unsigned int mux_mask;
>> +       struct clk_hw   hw;
>> +       struct lpass_gfm *priv;
>> +       void __iomem *gfm_mux;
>> +};
>> +
...
>> +static struct clk_gfm lpass_gfm_rx_npl = {
>> +       .mux_reg = 0x240d8,
>> +       .mux_mask = BIT(0),
>> +       .hw.init = &(struct clk_init_data) {
>> +               .name = "RX_NPL",
>> +               .ops = &clk_gfm_ops,
>> +               .flags = CLK_SET_RATE_PARENT | CLK_OPS_PARENT_ENABLE,
>> +               .parent_names = (const char *[]){
>> +                       "LPASS_CLK_ID_TX_CORE_NPL_MCLK",
>> +                       "LPASS_CLK_ID_RX_CORE_NPL_MCLK",
>> +               },
>> +               .parent_data = (const struct clk_parent_data[]){
>> +                               { .index = 0 },
>> +                               { .index = 1 },
>> +               },
> 
> Please use parent_data instead of parent_names. The name can go in the
> parent_data struct.

Ah.. I missed that, will fix that in next version!

> 
>> +               .num_parents = 2,
>> +       },
>> +};
>> +
>> +static struct clk_gfm *audiocc_gfm_clks[] = {
>> +       [LPASS_CDC_WSA_NPL]             = &lpass_gfm_wsa_npl,
>> +       [LPASS_CDC_WSA_MCLK]            = &lpass_gfm_wsa_mclk,
>> +       [LPASS_CDC_RX_NPL]              = &lpass_gfm_rx_npl,
>> +       [LPASS_CDC_RX_MCLK_MCLK2]       = &lpass_gfm_rx_mclk_mclk2,
>> +};
>> +
>> +static struct clk_hw_onecell_data audiocc_hw_onecell_data = {
>> +       .hws = {
>> +               [LPASS_CDC_WSA_NPL]     = &lpass_gfm_wsa_npl.hw,
>> +               [LPASS_CDC_WSA_MCLK]    = &lpass_gfm_wsa_mclk.hw,
>> +               [LPASS_CDC_RX_NPL]      = &lpass_gfm_rx_npl.hw,
>> +               [LPASS_CDC_RX_MCLK_MCLK2] = &lpass_gfm_rx_mclk_mclk2.hw,
>> +       },
>> +       .num = ARRAY_SIZE(audiocc_gfm_clks),
>> +};
>> +
>> +struct lpass_gfm_data {
>> +       struct clk_hw_onecell_data *onecell_data;
>> +       struct clk_gfm **gfm_clks;
>> +};
>> +
>> +static struct lpass_gfm_data audiocc_data = {
>> +       .onecell_data = &audiocc_hw_onecell_data,
>> +       .gfm_clks = audiocc_gfm_clks,
>> +};
>> +
>> +static int lpass_gfm_clk_driver_probe(struct platform_device *pdev)
>> +{
>> +       const struct lpass_gfm_data *data;
>> +       struct device *dev = &pdev->dev;
>> +       struct resource *res;
>> +       struct clk_gfm *gfm;
>> +       struct lpass_gfm *cc;
>> +       int err, i;
>> +
>> +       cc = devm_kzalloc(dev, sizeof(*cc), GFP_KERNEL);
>> +       if (!cc)
>> +               return -ENOMEM;
>> +
>> +       cc->core_vote = devm_clk_get(&pdev->dev, "core");
>> +       if (IS_ERR(cc->core_vote)) {
>> +               dev_dbg(dev, "Failed to get lpass core clk\n");
>> +               return PTR_ERR(cc->core_vote);
>> +       }
> 
> Can this use the pm_clk stuff?

you mean add runtime pm support or something else?

I can give it a go and see!


> 
>> +
>> +       data = of_device_get_match_data(dev);
> 
> What if data is NULL?
It should not be here if there is no match of compatible string, so data 
should not be NULL!

> 
>> +       cc->bus_clk = devm_clk_get(&pdev->dev, "bus");
>> +       if (IS_ERR(cc->bus_clk)) {
>> +               dev_dbg(dev, "Failed to get lpass bus clk\n");
>> +               return PTR_ERR(cc->bus_clk);
>> +       }
>> +
>> +       res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>> +       cc->base = devm_ioremap_resource(dev, res);
>> +       if (IS_ERR(cc->base))
>> +               return PTR_ERR(cc->base);
>> +
>> +       clk_prepare_enable(cc->core_vote);
>> +       clk_prepare_enable(cc->bus_clk);
> 
> These will always be left enabled if the probe function bails out after here.
> 
I agree, Will add some error handling
>> +
>> +       for (i = 0; i < data->onecell_data->num; i++) {
>> +               if (!data->gfm_clks[i])
>> +                       continue;
>> +
>> +               gfm = data->gfm_clks[i];
>> +               gfm->priv = cc;
>> +               gfm->gfm_mux = cc->base;
>> +               gfm->gfm_mux = gfm->gfm_mux + data->gfm_clks[i]->mux_reg;
>> +
>> +               err = devm_clk_hw_register(dev, &data->gfm_clks[i]->hw);
>> +               if (err)
>> +                       return err;
>> +
>> +       }
>> +
>> +       return devm_of_clk_add_hw_provider(dev, of_clk_hw_onecell_get,
>> +                                          data->onecell_data);
>> +}
>> +
>> +static const struct of_device_id lpass_gfm_clk_match_table[] = {
>> +       {
>> +               .compatible = "qcom,sm8250-lpass-audiocc",
>> +               .data = &audiocc_data,
>> +       },
>> +       { }
>> +};
>> +
>> +static struct platform_driver lpass_gfm_clk_driver = {
>> +       .probe          = lpass_gfm_clk_driver_probe,
>> +       .driver         = {
>> +               .name   = "lpass-gfm-clk",
>> +               .of_match_table = lpass_gfm_clk_match_table,
>> +       },
>> +};
>> +builtin_platform_driver(lpass_gfm_clk_driver);
> 
> Should be module_platform_driver()?
sure, will do that!

>
Dmitry Baryshkov Sept. 22, 2020, 9:49 p.m. UTC | #4
On 17/09/2020 16:28, Srinivas Kandagatla wrote:
> GFM Muxes in AUDIO_CC control clocks to LPASS WSA and RX Codec Macros.
> This patch adds support to these muxes.
> 
> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
> ---
>   drivers/clk/qcom/Kconfig            |   7 +
>   drivers/clk/qcom/Makefile           |   1 +
>   drivers/clk/qcom/lpass-gfm-sm8250.c | 235 ++++++++++++++++++++++++++++
>   3 files changed, 243 insertions(+)
>   create mode 100644 drivers/clk/qcom/lpass-gfm-sm8250.c
> 
> diff --git a/drivers/clk/qcom/Kconfig b/drivers/clk/qcom/Kconfig
> index 2eda63b7c46c..4e46fd339e62 100644
> --- a/drivers/clk/qcom/Kconfig
> +++ b/drivers/clk/qcom/Kconfig
> @@ -494,4 +494,11 @@ config KRAITCC
>   	  Support for the Krait CPU clocks on Qualcomm devices.
>   	  Say Y if you want to support CPU frequency scaling.
>   
> +config CLK_GFM_LPASS_SM8250
> +	tristate "GFM LPASS Clocks"

Since this driver can be built as module, few things are missing (see 
below).

> +	depends on SND_SOC_QDSP6_COMMON
> +	help
> +	  Support for the GFM Glitch Free Mux LPASS clock. Say Y
> +	  if you want to support GFM Clocks on LPASS for SM8250 SoC.
> +

[skipped]

> +
> +static const struct of_device_id lpass_gfm_clk_match_table[] = {
> +	{
> +		.compatible = "qcom,sm8250-lpass-audiocc",
> +		.data = &audiocc_data,
> +	},
> +	{ }
> +};

MODULE_DEVICE_TABLE(of, lpass_gfm_clk_match_table);

> +
> +static struct platform_driver lpass_gfm_clk_driver = {
> +	.probe		= lpass_gfm_clk_driver_probe,
> +	.driver		= {
> +		.name	= "lpass-gfm-clk",
> +		.of_match_table = lpass_gfm_clk_match_table,
> +	},
> +};
> +builtin_platform_driver(lpass_gfm_clk_driver);


Wouldn't you like to use module_platform_driver() here, like other LPASS 
CC drivers do?

Also MODULE_LICENSE is missing.
Stephen Boyd Sept. 23, 2020, 11:28 p.m. UTC | #5
Quoting Srinivas Kandagatla (2020-09-22 14:45:07)
> Thanks Stephen for review,
> 
> On 22/09/2020 19:41, Stephen Boyd wrote:
> > Quoting Srinivas Kandagatla (2020-09-17 06:28:49)
> >> diff --git a/drivers/clk/qcom/lpass-gfm-sm8250.c b/drivers/clk/qcom/lpass-gfm-sm8250.c
> >> new file mode 100644
> >> index 000000000000..2d5c41ae4969
> >> --- /dev/null
> >> +++ b/drivers/clk/qcom/lpass-gfm-sm8250.c
> >> @@ -0,0 +1,235 @@
[...]
> >> +};
> >> +
> >> +static int lpass_gfm_clk_driver_probe(struct platform_device *pdev)
> >> +{
> >> +       const struct lpass_gfm_data *data;
> >> +       struct device *dev = &pdev->dev;
> >> +       struct resource *res;
> >> +       struct clk_gfm *gfm;
> >> +       struct lpass_gfm *cc;
> >> +       int err, i;
> >> +
> >> +       cc = devm_kzalloc(dev, sizeof(*cc), GFP_KERNEL);
> >> +       if (!cc)
> >> +               return -ENOMEM;
> >> +
> >> +       cc->core_vote = devm_clk_get(&pdev->dev, "core");
> >> +       if (IS_ERR(cc->core_vote)) {
> >> +               dev_dbg(dev, "Failed to get lpass core clk\n");
> >> +               return PTR_ERR(cc->core_vote);
> >> +       }
> > 
> > Can this use the pm_clk stuff?
> 
> you mean add runtime pm support or something else?
> 
> I can give it a go and see!
> 

Yes use the runtime PM support and pm_clk APIs. There are some examples
in this directory already.

> 
> > 
> >> +
> >> +       data = of_device_get_match_data(dev);
> > 
> > What if data is NULL?
> It should not be here if there is no match of compatible string, so data 
> should not be NULL!

Ok sure but it always makes me feel better if we check for a valid
pointer with this API.