Message ID | 62280d24a7cf7c0be849c1186c909c850f4d2cc6.1495119548.git.jan.kiszka@siemens.com |
---|---|
State | New |
Headers | show |
On Thu, May 18, 2017 at 5:59 PM, Jan Kiszka <jan.kiszka@siemens.com> wrote: > This fixes reloading of the GPIO driver for the same platform device > instance as created by the exar UART driver: First of all, the driver > sets drvdata to its own value during probing and does not restore the > original value on exit. But this won't help anyway as the core clears > drvdata after the driver left. > > Use stable platform_data instead. Okay, basically what we are trying to do here is to reinvent part of MFD framework. I'd like to hear Linus' and others opinions if it worth to use it instead.
On 2017-05-18 19:14, Andy Shevchenko wrote: > On Thu, May 18, 2017 at 5:59 PM, Jan Kiszka <jan.kiszka@siemens.com> wrote: >> This fixes reloading of the GPIO driver for the same platform device >> instance as created by the exar UART driver: First of all, the driver >> sets drvdata to its own value during probing and does not restore the >> original value on exit. But this won't help anyway as the core clears >> drvdata after the driver left. >> >> Use stable platform_data instead. > > Okay, basically what we are trying to do here is to reinvent part of > MFD framework. > > I'd like to hear Linus' and others opinions if it worth to use it instead. > I've looked into MFD modeling, but it would only make sense if we break up the exar driver, change its xr17v35x part into a platform device and create a dual-cell MFD for the PCI device. I don't think that would be beneficial here. There are also dependencies between the UART part and the MPIOs, specifically during init. All that would create a lot of churn to the existing exar code. I'm now passing the parent reference via device.parent instead of using platform data. Jan
On Sun, May 21, 2017 at 1:44 PM, Jan Kiszka <jan.kiszka@siemens.com> wrote: > On 2017-05-18 19:14, Andy Shevchenko wrote: >> On Thu, May 18, 2017 at 5:59 PM, Jan Kiszka <jan.kiszka@siemens.com> wrote: >>> This fixes reloading of the GPIO driver for the same platform device >>> instance as created by the exar UART driver: First of all, the driver >>> sets drvdata to its own value during probing and does not restore the >>> original value on exit. But this won't help anyway as the core clears >>> drvdata after the driver left. >>> >>> Use stable platform_data instead. >> >> Okay, basically what we are trying to do here is to reinvent part of >> MFD framework. >> >> I'd like to hear Linus' and others opinions if it worth to use it instead. > > I've looked into MFD modeling, but it would only make sense if we break > up the exar driver, change its xr17v35x part into a platform device and > create a dual-cell MFD for the PCI device. I don't think that would be > beneficial here. There are also dependencies between the UART part and > the MPIOs, specifically during init. All that would create a lot of > churn to the existing exar code. > > I'm now passing the parent reference via device.parent instead of using > platform data. Actually I am pretty much OK with either, there are gray areas in the device model and so it has to be sometimes. I'd just like Greg's ACK on this so I can merge the whole series through the GPIO tree. Incidentally, he is the device model maintainer so he might have some comments. Or be as tolerant as me. I don't know. Greg? Yours, Linus Walleij -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/gpio/gpio-exar.c b/drivers/gpio/gpio-exar.c index 081076771217..d62e57513144 100644 --- a/drivers/gpio/gpio-exar.c +++ b/drivers/gpio/gpio-exar.c @@ -14,6 +14,7 @@ #include <linux/kernel.h> #include <linux/module.h> #include <linux/pci.h> +#include <linux/platform_data/gpio-exar.h> #include <linux/platform_device.h> #define EXAR_OFFSET_MPIOLVL_LO 0x90 @@ -119,7 +120,8 @@ static int exar_direction_input(struct gpio_chip *chip, unsigned int offset) static int gpio_exar_probe(struct platform_device *pdev) { - struct pci_dev *pcidev = platform_get_drvdata(pdev); + struct gpio_exar_pdata *pdata = pdev->dev.platform_data; + struct pci_dev *pcidev = pdata->parent; struct exar_gpio_chip *exar_gpio; void __iomem *p; int index, ret; diff --git a/drivers/tty/serial/8250/8250_exar.c b/drivers/tty/serial/8250/8250_exar.c index b4fa585156c7..0a806daaff8b 100644 --- a/drivers/tty/serial/8250/8250_exar.c +++ b/drivers/tty/serial/8250/8250_exar.c @@ -13,6 +13,7 @@ #include <linux/kernel.h> #include <linux/module.h> #include <linux/pci.h> +#include <linux/platform_data/gpio-exar.h> #include <linux/serial_core.h> #include <linux/serial_reg.h> #include <linux/slab.h> @@ -191,13 +192,16 @@ static void * xr17v35x_register_gpio(struct pci_dev *pcidev) { struct platform_device *pdev; + struct gpio_exar_pdata pdata; pdev = platform_device_alloc("gpio_exar", PLATFORM_DEVID_AUTO); if (!pdev) return NULL; - platform_set_drvdata(pdev, pcidev); - if (platform_device_add(pdev) < 0) { + pdata.parent = pcidev; + + if (platform_device_add_data(pdev, &pdata, sizeof(pdata)) < 0 || + platform_device_add(pdev) < 0) { platform_device_put(pdev); return NULL; } diff --git a/include/linux/platform_data/gpio-exar.h b/include/linux/platform_data/gpio-exar.h new file mode 100644 index 000000000000..1a13e9ddaf7d --- /dev/null +++ b/include/linux/platform_data/gpio-exar.h @@ -0,0 +1,22 @@ +/* + * GPIO handling for Exar XR17V35X chip + * + * Copyright (c) 2017 Siemens AG + * + * Written by Jan Kiszka <jan.kiszka@siemens.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#ifndef __GPIO_EXAR_PDATA_H +#define __GPIO_EXAR_PDATA_H + +struct pci_dev; + +struct gpio_exar_pdata { + struct pci_dev *parent; +}; + +#endif /* __GPIO_EXAR_PDATA_H */
This fixes reloading of the GPIO driver for the same platform device instance as created by the exar UART driver: First of all, the driver sets drvdata to its own value during probing and does not restore the original value on exit. But this won't help anyway as the core clears drvdata after the driver left. Use stable platform_data instead. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> --- drivers/gpio/gpio-exar.c | 4 +++- drivers/tty/serial/8250/8250_exar.c | 8 ++++++-- include/linux/platform_data/gpio-exar.h | 22 ++++++++++++++++++++++ 3 files changed, 31 insertions(+), 3 deletions(-) create mode 100644 include/linux/platform_data/gpio-exar.h