| Message ID | 20230818093028.7807-6-kkartik@nvidia.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | soc/tegra: fuse: Add ACPI support | expand |
On Fri, Aug 18, 2023 at 03:00:27PM +0530, Kartik wrote: > Add tegra_fuse_acpi_probe() to initialize Tegra fuse while using ACPI. > Also, drop '__init' keyword for tegra_soc_device_register() as this is also > used by tegra_fuse_acpi_probe(). > > Note that as ACPI subsystem initialize at subsys init, function > tegra_fuse_acpi_probe() also contains the necessary initialization > that we are currently doing for device-tree boot as a part of > early init. ... > +#include <linux/acpi.h> You meed mod_devicetable.h and possibly property.h, not this header (see below). ... > +static const struct acpi_device_id tegra_fuse_acpi_match[] = { > + { > + .id = "NVDA200F", > + }, Single line, no inner comma. > + { /* sentinel */ }, The idea of sentinel is to guard, the trailing comma ruins this contract. > +}; ... > +static int tegra_fuse_acpi_probe(struct platform_device *pdev) > +{ Why you need a separate function? > + struct resource *res; > + u8 chip; > + int err; > + > + tegra_acpi_init_apbmisc(); > + > + chip = tegra_get_chip_id(); > + switch (chip) { > +#if defined(CONFIG_ARCH_TEGRA_194_SOC) Can we avoid ugly ifdeffery? > + case TEGRA194: > + fuse->soc = &tegra194_fuse_soc; > + break; > +#endif > +#if defined(CONFIG_ARCH_TEGRA_234_SOC) Ditto. > + case TEGRA234: > + fuse->soc = &tegra234_fuse_soc; > + break; > +#endif > + default: > + dev_err(&pdev->dev, "Unsupported SoC: %02x\n", chip); > + return -EINVAL; return dev_err_probe(...); > + } > + > + fuse->base = devm_platform_get_and_ioremap_resource(pdev, 0, &res); > + if (IS_ERR(fuse->base)) > + return PTR_ERR(fuse->base); > + fuse->phys = res->start; > + platform_set_drvdata(pdev, fuse); Is it being used? > + fuse->dev = &pdev->dev; > + > + err = tegra_fuse_nvmem_register(fuse, &pdev->dev); > + if (err) > + return err; > + > + fuse->soc->init(fuse); > + tegra_soc_device_register(); > + tegra_fuse_pr_sku_info(&tegra_sku_info); > + > + err = tegra_fuse_add_lookups(fuse); > + if (err) { > + dev_err(&pdev->dev, "failed to add FUSE lookups\n"); > + return err; return dev_err_probe(...); > + } > + > + return 0; > +} ... > + if (has_acpi_companion(&pdev->dev)) > + return tegra_fuse_acpi_probe(pdev); Why is the ACPI so special here? Why you can't go same flow? ... > + /* fuse->clk is not required when ACPI is used. */ > + if (!fuse->read || (!fuse->clk && !has_acpi_companion(fuse->dev))) No, just make CLK optional and that's it. > return -EPROBE_DEFER;
On Fri, 2023-08-18 at 18:07 +0300, Andy Shevchenko wrote: >On Fri, Aug 18, 2023 at 03:00:27PM +0530, Kartik wrote: >> Add tegra_fuse_acpi_probe() to initialize Tegra fuse while using ACPI. >> Also, drop '__init' keyword for tegra_soc_device_register() as this is also >> used by tegra_fuse_acpi_probe(). >> >> Note that as ACPI subsystem initialize at subsys init, function >> tegra_fuse_acpi_probe() also contains the necessary initialization >> that we are currently doing for device-tree boot as a part of >> early init. > >... > >> +#include <linux/acpi.h> > >You meed mod_devicetable.h and possibly property.h, not this header >(see below). > >... > >> +static const struct acpi_device_id tegra_fuse_acpi_match[] = { >> + { >> + .id = "NVDA200F", >> + }, > >Single line, no inner comma. ACK. > >> + { /* sentinel */ }, > >The idea of sentinel is to guard, the trailing comma ruins this contract. > >> +}; ACK. > >... > >> +static int tegra_fuse_acpi_probe(struct platform_device *pdev) >> +{ > >Why you need a separate function? > >> + struct resource *res; >> + u8 chip; >> + int err; >> + >> + tegra_acpi_init_apbmisc(); >> + >> + chip = tegra_get_chip_id(); >> + switch (chip) { >> +#if defined(CONFIG_ARCH_TEGRA_194_SOC) > >Can we avoid ugly ifdeffery? > No, the SoC data is defined only when the SoC specific config is enabled. So, guarding these with ifdef's is required here. >> + case TEGRA194: >> + fuse->soc = &tegra194_fuse_soc; >> + break; >> +#endif >> +#if defined(CONFIG_ARCH_TEGRA_234_SOC) > >Ditto. > >> + case TEGRA234: >> + fuse->soc = &tegra234_fuse_soc; >> + break; >> +#endif >> + default: >> + dev_err(&pdev->dev, "Unsupported SoC: %02x\n", chip); >> + return -EINVAL; > > return dev_err_probe(...); > ACK. >> + } >> + >> + fuse->base = devm_platform_get_and_ioremap_resource(pdev, 0, &res); >> + if (IS_ERR(fuse->base)) >> + return PTR_ERR(fuse->base); >> + fuse->phys = res->start; > >> + platform_set_drvdata(pdev, fuse); > >Is it being used? > Looks like this is not being used. >> + fuse->dev = &pdev->dev; >> + >> + err = tegra_fuse_nvmem_register(fuse, &pdev->dev); >> + if (err) >> + return err; >> + >> + fuse->soc->init(fuse); >> + tegra_soc_device_register(); >> + tegra_fuse_pr_sku_info(&tegra_sku_info); >> + >> + err = tegra_fuse_add_lookups(fuse); >> + if (err) { > >> + dev_err(&pdev->dev, "failed to add FUSE lookups\n"); >> + return err; > > return dev_err_probe(...); > >> + } >> + >> + return 0; >> +} > >... > >> + if (has_acpi_companion(&pdev->dev)) >> + return tegra_fuse_acpi_probe(pdev); > >Why is the ACPI so special here? Why you can't go same flow? > >... > We need to initialize the soc data before we continue the probe. I guess we can have a conditional here for this initialization. and re-use the same function. I will update this in the next patch. >> + /* fuse->clk is not required when ACPI is used. */ >> + if (!fuse->read || (!fuse->clk && !has_acpi_companion(fuse->dev))) > >No, just make CLK optional and that's it. > >> return -EPROBE_DEFER; > >-- If the Fuse driver is probed using device-tree. Then we need to make sure that fuse->clk has been initilaized. >With Best Regards, >Andy Shevchenko Regards, Kartik
On Mon, Aug 21, 2023 at 05:08:19PM +0530, Kartik wrote: > On Fri, 2023-08-18 at 18:07 +0300, Andy Shevchenko wrote: > >On Fri, Aug 18, 2023 at 03:00:27PM +0530, Kartik wrote: ... > >> + /* fuse->clk is not required when ACPI is used. */ > >> + if (!fuse->read || (!fuse->clk && !has_acpi_companion(fuse->dev))) > > > >No, just make CLK optional and that's it. > > > >> return -EPROBE_DEFER; > > > If the Fuse driver is probed using device-tree. Then we need to make > sure that fuse->clk has been initilaized. So, I recommend to think about it more, maybe you can find better solution than above. And actually don't use has_acpi_companion() in this case. What you need is is_acpi_node() or is_of_node() depending on the case you want to check for.
diff --git a/drivers/soc/tegra/fuse/fuse-tegra.c b/drivers/soc/tegra/fuse/fuse-tegra.c index 70e8eeddcbd9..9c5596d968a2 100644 --- a/drivers/soc/tegra/fuse/fuse-tegra.c +++ b/drivers/soc/tegra/fuse/fuse-tegra.c @@ -3,6 +3,7 @@ * Copyright (c) 2013-2023, NVIDIA CORPORATION. All rights reserved. */ +#include <linux/acpi.h> #include <linux/clk.h> #include <linux/device.h> #include <linux/kobject.h> @@ -94,6 +95,13 @@ static const struct of_device_id tegra_fuse_match[] = { { /* sentinel */ } }; +static const struct acpi_device_id tegra_fuse_acpi_match[] = { + { + .id = "NVDA200F", + }, + { /* sentinel */ }, +}; + static int tegra_fuse_read(void *priv, unsigned int offset, void *value, size_t bytes) { @@ -176,12 +184,65 @@ static void tegra_fuse_pr_sku_info(struct tegra_sku_info *tegra_sku_info) tegra_sku_info->cpu_speedo_id, tegra_sku_info->soc_speedo_id); } +static int tegra_fuse_acpi_probe(struct platform_device *pdev) +{ + struct resource *res; + u8 chip; + int err; + + tegra_acpi_init_apbmisc(); + + chip = tegra_get_chip_id(); + switch (chip) { +#if defined(CONFIG_ARCH_TEGRA_194_SOC) + case TEGRA194: + fuse->soc = &tegra194_fuse_soc; + break; +#endif +#if defined(CONFIG_ARCH_TEGRA_234_SOC) + case TEGRA234: + fuse->soc = &tegra234_fuse_soc; + break; +#endif + default: + dev_err(&pdev->dev, "Unsupported SoC: %02x\n", chip); + return -EINVAL; + } + + fuse->base = devm_platform_get_and_ioremap_resource(pdev, 0, &res); + if (IS_ERR(fuse->base)) + return PTR_ERR(fuse->base); + fuse->phys = res->start; + + platform_set_drvdata(pdev, fuse); + fuse->dev = &pdev->dev; + + err = tegra_fuse_nvmem_register(fuse, &pdev->dev); + if (err) + return err; + + fuse->soc->init(fuse); + tegra_soc_device_register(); + tegra_fuse_pr_sku_info(&tegra_sku_info); + + err = tegra_fuse_add_lookups(fuse); + if (err) { + dev_err(&pdev->dev, "failed to add FUSE lookups\n"); + return err; + } + + return 0; +} + static int tegra_fuse_probe(struct platform_device *pdev) { void __iomem *base = fuse->base; struct resource *res; int err; + if (has_acpi_companion(&pdev->dev)) + return tegra_fuse_acpi_probe(pdev); + err = devm_add_action(&pdev->dev, tegra_fuse_restore, (void __force *)base); if (err) return err; @@ -306,6 +367,7 @@ static struct platform_driver tegra_fuse_driver = { .driver = { .name = "tegra-fuse", .of_match_table = tegra_fuse_match, + .acpi_match_table = tegra_fuse_acpi_match, .pm = &tegra_fuse_pm, .suppress_bind_attrs = true, }, @@ -327,7 +389,8 @@ u32 __init tegra_fuse_read_early(unsigned int offset) int tegra_fuse_readl(unsigned long offset, u32 *value) { - if (!fuse->read || !fuse->clk) + /* fuse->clk is not required when ACPI is used. */ + if (!fuse->read || (!fuse->clk && !has_acpi_companion(fuse->dev))) return -EPROBE_DEFER; if (IS_ERR(fuse->clk)) @@ -410,7 +473,7 @@ const struct attribute_group tegra194_soc_attr_group = { }; #endif -struct device * __init tegra_soc_device_register(void) +struct device *tegra_soc_device_register(void) { struct soc_device_attribute *attr; struct soc_device *dev;
Add tegra_fuse_acpi_probe() to initialize Tegra fuse while using ACPI. Also, drop '__init' keyword for tegra_soc_device_register() as this is also used by tegra_fuse_acpi_probe(). Note that as ACPI subsystem initialize at subsys init, function tegra_fuse_acpi_probe() also contains the necessary initialization that we are currently doing for device-tree boot as a part of early init. Signed-off-by: Kartik <kkartik@nvidia.com> --- drivers/soc/tegra/fuse/fuse-tegra.c | 67 ++++++++++++++++++++++++++++- 1 file changed, 65 insertions(+), 2 deletions(-)