kernel-ark/drivers/gpio/gpiolib.c

2167 lines
53 KiB
C
Raw Normal View History

gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/interrupt.h>
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#include <linux/irq.h>
#include <linux/spinlock.h>
#include <linux/list.h>
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
#include <linux/device.h>
#include <linux/err.h>
#include <linux/debugfs.h>
#include <linux/seq_file.h>
#include <linux/gpio.h>
#include <linux/of_gpio.h>
#include <linux/idr.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#define CREATE_TRACE_POINTS
#include <trace/events/gpio.h>
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* Optional implementation infrastructure for GPIO interfaces.
*
* Platforms may want to use this if they tend to use very many GPIOs
* that aren't part of a System-On-Chip core; or across I2C/SPI/etc.
*
* When kernel footprint or instruction count is an issue, simpler
* implementations may be preferred. The GPIO programming interface
* allows for inlining speed-critical get/set operations for common
* cases, so that access to SOC-integrated GPIOs can sometimes cost
* only an instruction or two per bit.
*/
/* When debugging, extend minimal trust to callers and platform code.
* Also emit diagnostic messages that may help initial bringup, when
* board setup or driver bugs are most common.
*
* Otherwise, minimize overhead in what may be bitbanging codepaths.
*/
#ifdef DEBUG
#define extra_checks 1
#else
#define extra_checks 0
#endif
/* gpio_lock prevents conflicts during gpio_desc[] table updates.
* While any GPIO is requested, its gpio_chip is not removable;
* each GPIO's "requested" flag serves as a lock and refcount.
*/
static DEFINE_SPINLOCK(gpio_lock);
struct gpio_desc {
struct gpio_chip *chip;
unsigned long flags;
/* flag symbols are bit numbers */
#define FLAG_REQUESTED 0
#define FLAG_IS_OUT 1
#define FLAG_EXPORT 2 /* protected by sysfs_lock */
#define FLAG_SYSFS 3 /* exported via /sys/class/gpio/control */
#define FLAG_TRIG_FALL 4 /* trigger on falling edge */
#define FLAG_TRIG_RISE 5 /* trigger on rising edge */
#define FLAG_ACTIVE_LOW 6 /* sysfs value has active low */
#define FLAG_OPEN_DRAIN 7 /* Gpio is open drain type */
#define FLAG_OPEN_SOURCE 8 /* Gpio is open source type */
#define ID_SHIFT 16 /* add new flags before this one */
#define GPIO_FLAGS_MASK ((1 << ID_SHIFT) - 1)
#define GPIO_TRIGGER_MASK (BIT(FLAG_TRIG_FALL) | BIT(FLAG_TRIG_RISE))
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#ifdef CONFIG_DEBUG_FS
const char *label;
#endif
};
static struct gpio_desc gpio_desc[ARCH_NR_GPIOS];
#define GPIO_OFFSET_VALID(chip, offset) (offset >= 0 && offset < chip->ngpio)
static LIST_HEAD(gpio_chips);
#ifdef CONFIG_GPIO_SYSFS
static DEFINE_IDR(dirent_idr);
#endif
/*
* Internal gpiod_* API using descriptors instead of the integer namespace.
* Most of this should eventually go public.
*/
static int gpiod_request(struct gpio_desc *desc, const char *label);
static void gpiod_free(struct gpio_desc *desc);
static int gpiod_direction_input(struct gpio_desc *desc);
static int gpiod_direction_output(struct gpio_desc *desc, int value);
static int gpiod_get_direction(const struct gpio_desc *desc);
static int gpiod_set_debounce(struct gpio_desc *desc, unsigned debounce);
static int gpiod_get_value_cansleep(const struct gpio_desc *desc);
static void gpiod_set_value_cansleep(struct gpio_desc *desc, int value);
static int gpiod_get_value(const struct gpio_desc *desc);
static void gpiod_set_value(struct gpio_desc *desc, int value);
static int gpiod_cansleep(const struct gpio_desc *desc);
static int gpiod_to_irq(const struct gpio_desc *desc);
static int gpiod_export(struct gpio_desc *desc, bool direction_may_change);
static int gpiod_export_link(struct device *dev, const char *name,
struct gpio_desc *desc);
static int gpiod_sysfs_set_active_low(struct gpio_desc *desc, int value);
static void gpiod_unexport(struct gpio_desc *desc);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
static inline void desc_set_label(struct gpio_desc *d, const char *label)
{
#ifdef CONFIG_DEBUG_FS
d->label = label;
#endif
}
/*
* Return the GPIO number of the passed descriptor relative to its chip
*/
static int gpio_chip_hwgpio(const struct gpio_desc *desc)
{
return desc - &desc->chip->desc[0];
}
/**
* Convert a GPIO number to its descriptor
*/
static struct gpio_desc *gpio_to_desc(unsigned gpio)
{
if (WARN(!gpio_is_valid(gpio), "invalid GPIO %d\n", gpio))
return NULL;
else
return &gpio_desc[gpio];
}
/**
* Convert a GPIO descriptor to the integer namespace.
* This should disappear in the future but is needed since we still
* use GPIO numbers for error messages and sysfs nodes
*/
static int desc_to_gpio(const struct gpio_desc *desc)
{
return desc->chip->base + gpio_chip_hwgpio(desc);
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* Warn when drivers omit gpio_request() calls -- legal but ill-advised
* when setting direction, and otherwise illegal. Until board setup code
* and drivers use explicit requests everywhere (which won't happen when
* those calls have no teeth) we can't avoid autorequesting. This nag
* message should motivate switching to explicit requests... so should
* the weaker cleanup after faults, compared to gpio_request().
*
* NOTE: the autorequest mechanism is going away; at this point it's
* only "legal" in the sense that (old) code using it won't break yet,
* but instead only triggers a WARN() stack dump.
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
*/
static int gpio_ensure_requested(struct gpio_desc *desc)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
const struct gpio_chip *chip = desc->chip;
const int gpio = desc_to_gpio(desc);
if (WARN(test_and_set_bit(FLAG_REQUESTED, &desc->flags) == 0,
"autorequest GPIO-%d\n", gpio)) {
if (!try_module_get(chip->owner)) {
pr_err("GPIO-%d: module can't be gotten \n", gpio);
clear_bit(FLAG_REQUESTED, &desc->flags);
/* lose */
return -EIO;
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
desc_set_label(desc, "[auto]");
/* caller must chip->request() w/o spinlock */
if (chip->request)
return 1;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
return 0;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
static struct gpio_chip *gpiod_to_chip(const struct gpio_desc *desc)
{
return desc ? desc->chip : NULL;
}
/* caller holds gpio_lock *OR* gpio is marked as requested */
struct gpio_chip *gpio_to_chip(unsigned gpio)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
return gpiod_to_chip(gpio_to_desc(gpio));
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
/* dynamic allocation of GPIOs, e.g. on a hotplugged device */
static int gpiochip_find_base(int ngpio)
{
struct gpio_chip *chip;
int base = ARCH_NR_GPIOS - ngpio;
list_for_each_entry_reverse(chip, &gpio_chips, list) {
/* found a free space? */
if (chip->base + chip->ngpio <= base)
break;
else
/* nope, check the space right before the chip */
base = chip->base - ngpio;
}
if (gpio_is_valid(base)) {
pr_debug("%s: found new base at %d\n", __func__, base);
return base;
} else {
pr_err("%s: cannot find free range\n", __func__);
return -ENOSPC;
}
}
/* caller ensures gpio is valid and requested, chip->get_direction may sleep */
static int gpiod_get_direction(const struct gpio_desc *desc)
{
struct gpio_chip *chip;
unsigned offset;
int status = -EINVAL;
chip = gpiod_to_chip(desc);
offset = gpio_chip_hwgpio(desc);
if (!chip->get_direction)
return status;
status = chip->get_direction(chip, offset);
if (status > 0) {
/* GPIOF_DIR_IN, or other positive */
status = 1;
/* FLAG_IS_OUT is just a cache of the result of get_direction(),
* so it does not affect constness per se */
clear_bit(FLAG_IS_OUT, &((struct gpio_desc *)desc)->flags);
}
if (status == 0) {
/* GPIOF_DIR_OUT */
set_bit(FLAG_IS_OUT, &((struct gpio_desc *)desc)->flags);
}
return status;
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
#ifdef CONFIG_GPIO_SYSFS
/* lock protects against unexport_gpio() being called while
* sysfs files are active.
*/
static DEFINE_MUTEX(sysfs_lock);
/*
* /sys/class/gpio/gpioN... only for GPIOs that are exported
* /direction
* * MAY BE OMITTED if kernel won't allow direction changes
* * is read/write as "in" or "out"
* * may also be written as "high" or "low", initializing
* output value as specified ("out" implies "low")
* /value
* * always readable, subject to hardware behavior
* * may be writable, as zero/nonzero
* /edge
* * configures behavior of poll(2) on /value
* * available only if pin can generate IRQs on input
* * is read/write as "none", "falling", "rising", or "both"
* /active_low
* * configures polarity of /value
* * is read/write as zero/nonzero
* * also affects existing and subsequent "falling" and "rising"
* /edge configuration
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
*/
static ssize_t gpio_direction_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const struct gpio_desc *desc = dev_get_drvdata(dev);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags)) {
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = -EIO;
} else {
gpiod_get_direction(desc);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = sprintf(buf, "%s\n",
test_bit(FLAG_IS_OUT, &desc->flags)
? "out" : "in");
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
mutex_unlock(&sysfs_lock);
return status;
}
static ssize_t gpio_direction_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t size)
{
struct gpio_desc *desc = dev_get_drvdata(dev);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags))
status = -EIO;
else if (sysfs_streq(buf, "high"))
status = gpiod_direction_output(desc, 1);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
else if (sysfs_streq(buf, "out") || sysfs_streq(buf, "low"))
status = gpiod_direction_output(desc, 0);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
else if (sysfs_streq(buf, "in"))
status = gpiod_direction_input(desc);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
else
status = -EINVAL;
mutex_unlock(&sysfs_lock);
return status ? : size;
}
static /* const */ DEVICE_ATTR(direction, 0644,
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
gpio_direction_show, gpio_direction_store);
static ssize_t gpio_value_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct gpio_desc *desc = dev_get_drvdata(dev);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags)) {
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = -EIO;
} else {
int value;
value = !!gpiod_get_value_cansleep(desc);
if (test_bit(FLAG_ACTIVE_LOW, &desc->flags))
value = !value;
status = sprintf(buf, "%d\n", value);
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
mutex_unlock(&sysfs_lock);
return status;
}
static ssize_t gpio_value_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t size)
{
struct gpio_desc *desc = dev_get_drvdata(dev);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags))
status = -EIO;
else if (!test_bit(FLAG_IS_OUT, &desc->flags))
status = -EPERM;
else {
long value;
status = strict_strtol(buf, 0, &value);
if (status == 0) {
if (test_bit(FLAG_ACTIVE_LOW, &desc->flags))
value = !value;
gpiod_set_value_cansleep(desc, value != 0);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = size;
}
}
mutex_unlock(&sysfs_lock);
return status;
}
static const DEVICE_ATTR(value, 0644,
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
gpio_value_show, gpio_value_store);
static irqreturn_t gpio_sysfs_irq(int irq, void *priv)
{
struct sysfs_dirent *value_sd = priv;
sysfs_notify_dirent(value_sd);
return IRQ_HANDLED;
}
static int gpio_setup_irq(struct gpio_desc *desc, struct device *dev,
unsigned long gpio_flags)
{
struct sysfs_dirent *value_sd;
unsigned long irq_flags;
int ret, irq, id;
if ((desc->flags & GPIO_TRIGGER_MASK) == gpio_flags)
return 0;
irq = gpiod_to_irq(desc);
if (irq < 0)
return -EIO;
id = desc->flags >> ID_SHIFT;
value_sd = idr_find(&dirent_idr, id);
if (value_sd)
free_irq(irq, value_sd);
desc->flags &= ~GPIO_TRIGGER_MASK;
if (!gpio_flags) {
ret = 0;
goto free_id;
}
irq_flags = IRQF_SHARED;
if (test_bit(FLAG_TRIG_FALL, &gpio_flags))
irq_flags |= test_bit(FLAG_ACTIVE_LOW, &desc->flags) ?
IRQF_TRIGGER_RISING : IRQF_TRIGGER_FALLING;
if (test_bit(FLAG_TRIG_RISE, &gpio_flags))
irq_flags |= test_bit(FLAG_ACTIVE_LOW, &desc->flags) ?
IRQF_TRIGGER_FALLING : IRQF_TRIGGER_RISING;
if (!value_sd) {
value_sd = sysfs_get_dirent(dev->kobj.sd, NULL, "value");
if (!value_sd) {
ret = -ENODEV;
goto err_out;
}
ret = idr_alloc(&dirent_idr, value_sd, 1, 0, GFP_KERNEL);
if (ret < 0)
goto free_sd;
id = ret;
desc->flags &= GPIO_FLAGS_MASK;
desc->flags |= (unsigned long)id << ID_SHIFT;
if (desc->flags >> ID_SHIFT != id) {
ret = -ERANGE;
goto free_id;
}
}
ret = request_any_context_irq(irq, gpio_sysfs_irq, irq_flags,
"gpiolib", value_sd);
if (ret < 0)
goto free_id;
desc->flags |= gpio_flags;
return 0;
free_id:
idr_remove(&dirent_idr, id);
desc->flags &= GPIO_FLAGS_MASK;
free_sd:
if (value_sd)
sysfs_put(value_sd);
err_out:
return ret;
}
static const struct {
const char *name;
unsigned long flags;
} trigger_types[] = {
{ "none", 0 },
{ "falling", BIT(FLAG_TRIG_FALL) },
{ "rising", BIT(FLAG_TRIG_RISE) },
{ "both", BIT(FLAG_TRIG_FALL) | BIT(FLAG_TRIG_RISE) },
};
static ssize_t gpio_edge_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const struct gpio_desc *desc = dev_get_drvdata(dev);
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags))
status = -EIO;
else {
int i;
status = 0;
for (i = 0; i < ARRAY_SIZE(trigger_types); i++)
if ((desc->flags & GPIO_TRIGGER_MASK)
== trigger_types[i].flags) {
status = sprintf(buf, "%s\n",
trigger_types[i].name);
break;
}
}
mutex_unlock(&sysfs_lock);
return status;
}
static ssize_t gpio_edge_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t size)
{
struct gpio_desc *desc = dev_get_drvdata(dev);
ssize_t status;
int i;
for (i = 0; i < ARRAY_SIZE(trigger_types); i++)
if (sysfs_streq(trigger_types[i].name, buf))
goto found;
return -EINVAL;
found:
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags))
status = -EIO;
else {
status = gpio_setup_irq(desc, dev, trigger_types[i].flags);
if (!status)
status = size;
}
mutex_unlock(&sysfs_lock);
return status;
}
static DEVICE_ATTR(edge, 0644, gpio_edge_show, gpio_edge_store);
static int sysfs_set_active_low(struct gpio_desc *desc, struct device *dev,
int value)
{
int status = 0;
if (!!test_bit(FLAG_ACTIVE_LOW, &desc->flags) == !!value)
return 0;
if (value)
set_bit(FLAG_ACTIVE_LOW, &desc->flags);
else
clear_bit(FLAG_ACTIVE_LOW, &desc->flags);
/* reconfigure poll(2) support if enabled on one edge only */
if (dev != NULL && (!!test_bit(FLAG_TRIG_RISE, &desc->flags) ^
!!test_bit(FLAG_TRIG_FALL, &desc->flags))) {
unsigned long trigger_flags = desc->flags & GPIO_TRIGGER_MASK;
gpio_setup_irq(desc, dev, 0);
status = gpio_setup_irq(desc, dev, trigger_flags);
}
return status;
}
static ssize_t gpio_active_low_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const struct gpio_desc *desc = dev_get_drvdata(dev);
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags))
status = -EIO;
else
status = sprintf(buf, "%d\n",
!!test_bit(FLAG_ACTIVE_LOW, &desc->flags));
mutex_unlock(&sysfs_lock);
return status;
}
static ssize_t gpio_active_low_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t size)
{
struct gpio_desc *desc = dev_get_drvdata(dev);
ssize_t status;
mutex_lock(&sysfs_lock);
if (!test_bit(FLAG_EXPORT, &desc->flags)) {
status = -EIO;
} else {
long value;
status = strict_strtol(buf, 0, &value);
if (status == 0)
status = sysfs_set_active_low(desc, dev, value != 0);
}
mutex_unlock(&sysfs_lock);
return status ? : size;
}
static const DEVICE_ATTR(active_low, 0644,
gpio_active_low_show, gpio_active_low_store);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
static const struct attribute *gpio_attrs[] = {
&dev_attr_value.attr,
&dev_attr_active_low.attr,
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
NULL,
};
static const struct attribute_group gpio_attr_group = {
.attrs = (struct attribute **) gpio_attrs,
};
/*
* /sys/class/gpio/gpiochipN/
* /base ... matching gpio_chip.base (N)
* /label ... matching gpio_chip.label
* /ngpio ... matching gpio_chip.ngpio
*/
static ssize_t chip_base_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const struct gpio_chip *chip = dev_get_drvdata(dev);
return sprintf(buf, "%d\n", chip->base);
}
static DEVICE_ATTR(base, 0444, chip_base_show, NULL);
static ssize_t chip_label_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const struct gpio_chip *chip = dev_get_drvdata(dev);
return sprintf(buf, "%s\n", chip->label ? : "");
}
static DEVICE_ATTR(label, 0444, chip_label_show, NULL);
static ssize_t chip_ngpio_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const struct gpio_chip *chip = dev_get_drvdata(dev);
return sprintf(buf, "%u\n", chip->ngpio);
}
static DEVICE_ATTR(ngpio, 0444, chip_ngpio_show, NULL);
static const struct attribute *gpiochip_attrs[] = {
&dev_attr_base.attr,
&dev_attr_label.attr,
&dev_attr_ngpio.attr,
NULL,
};
static const struct attribute_group gpiochip_attr_group = {
.attrs = (struct attribute **) gpiochip_attrs,
};
/*
* /sys/class/gpio/export ... write-only
* integer N ... number of GPIO to export (full access)
* /sys/class/gpio/unexport ... write-only
* integer N ... number of GPIO to unexport
*/
static ssize_t export_store(struct class *class,
struct class_attribute *attr,
const char *buf, size_t len)
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
{
long gpio;
struct gpio_desc *desc;
int status;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = strict_strtol(buf, 0, &gpio);
if (status < 0)
goto done;
desc = gpio_to_desc(gpio);
/* reject invalid GPIOs */
if (!desc) {
pr_warn("%s: invalid GPIO %ld\n", __func__, gpio);
return -EINVAL;
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
/* No extra locking here; FLAG_SYSFS just signifies that the
* request and export were done by on behalf of userspace, so
* they may be undone on its behalf too.
*/
status = gpiod_request(desc, "sysfs");
if (status < 0) {
if (status == -EPROBE_DEFER)
status = -ENODEV;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
goto done;
}
status = gpiod_export(desc, true);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (status < 0)
gpiod_free(desc);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
else
set_bit(FLAG_SYSFS, &desc->flags);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
done:
if (status)
pr_debug("%s: status %d\n", __func__, status);
return status ? : len;
}
static ssize_t unexport_store(struct class *class,
struct class_attribute *attr,
const char *buf, size_t len)
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
{
long gpio;
struct gpio_desc *desc;
int status;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = strict_strtol(buf, 0, &gpio);
if (status < 0)
goto done;
desc = gpio_to_desc(gpio);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
/* reject bogus commands (gpio_unexport ignores them) */
if (!desc) {
pr_warn("%s: invalid GPIO %ld\n", __func__, gpio);
return -EINVAL;
}
status = -EINVAL;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
/* No extra locking here; FLAG_SYSFS just signifies that the
* request and export were done by on behalf of userspace, so
* they may be undone on its behalf too.
*/
if (test_and_clear_bit(FLAG_SYSFS, &desc->flags)) {
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = 0;
gpiod_free(desc);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
}
done:
if (status)
pr_debug("%s: status %d\n", __func__, status);
return status ? : len;
}
static struct class_attribute gpio_class_attrs[] = {
__ATTR(export, 0200, NULL, export_store),
__ATTR(unexport, 0200, NULL, unexport_store),
__ATTR_NULL,
};
static struct class gpio_class = {
.name = "gpio",
.owner = THIS_MODULE,
.class_attrs = gpio_class_attrs,
};
/**
* gpio_export - export a GPIO through sysfs
* @gpio: gpio to make available, already requested
* @direction_may_change: true if userspace may change gpio direction
* Context: arch_initcall or later
*
* When drivers want to make a GPIO accessible to userspace after they
* have requested it -- perhaps while debugging, or as part of their
* public interface -- they may use this routine. If the GPIO can
* change direction (some can't) and the caller allows it, userspace
* will see "direction" sysfs attribute which may be used to change
* the gpio's direction. A "value" attribute will always be provided.
*
* Returns zero on success, else an error.
*/
static int gpiod_export(struct gpio_desc *desc, bool direction_may_change)
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
{
unsigned long flags;
int status;
const char *ioname = NULL;
struct device *dev;
int offset;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
/* can't export until sysfs is available ... */
if (!gpio_class.p) {
pr_debug("%s: called too early!\n", __func__);
return -ENOENT;
}
if (!desc) {
pr_debug("%s: invalid gpio descriptor\n", __func__);
return -EINVAL;
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
mutex_lock(&sysfs_lock);
spin_lock_irqsave(&gpio_lock, flags);
if (!test_bit(FLAG_REQUESTED, &desc->flags) ||
test_bit(FLAG_EXPORT, &desc->flags)) {
spin_unlock_irqrestore(&gpio_lock, flags);
pr_debug("%s: gpio %d unavailable (requested=%d, exported=%d)\n",
__func__, desc_to_gpio(desc),
test_bit(FLAG_REQUESTED, &desc->flags),
test_bit(FLAG_EXPORT, &desc->flags));
status = -EPERM;
goto fail_unlock;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
}
if (!desc->chip->direction_input || !desc->chip->direction_output)
direction_may_change = false;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
spin_unlock_irqrestore(&gpio_lock, flags);
offset = gpio_chip_hwgpio(desc);
if (desc->chip->names && desc->chip->names[offset])
ioname = desc->chip->names[offset];
dev = device_create(&gpio_class, desc->chip->dev, MKDEV(0, 0),
desc, ioname ? ioname : "gpio%u",
desc_to_gpio(desc));
if (IS_ERR(dev)) {
status = PTR_ERR(dev);
goto fail_unlock;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
}
status = sysfs_create_group(&dev->kobj, &gpio_attr_group);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (status)
goto fail_unregister_device;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (direction_may_change) {
status = device_create_file(dev, &dev_attr_direction);
if (status)
goto fail_unregister_device;
}
if (gpiod_to_irq(desc) >= 0 && (direction_may_change ||
!test_bit(FLAG_IS_OUT, &desc->flags))) {
status = device_create_file(dev, &dev_attr_edge);
if (status)
goto fail_unregister_device;
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
set_bit(FLAG_EXPORT, &desc->flags);
mutex_unlock(&sysfs_lock);
return 0;
fail_unregister_device:
device_unregister(dev);
fail_unlock:
mutex_unlock(&sysfs_lock);
pr_debug("%s: gpio%d status %d\n", __func__, desc_to_gpio(desc),
status);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
return status;
}
int gpio_export(unsigned gpio, bool direction_may_change)
{
return gpiod_export(gpio_to_desc(gpio), direction_may_change);
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
EXPORT_SYMBOL_GPL(gpio_export);
static int match_export(struct device *dev, const void *data)
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
{
return dev_get_drvdata(dev) == data;
}
/**
* gpio_export_link - create a sysfs link to an exported GPIO node
* @dev: device under which to create symlink
* @name: name of the symlink
* @gpio: gpio to create symlink to, already exported
*
* Set up a symlink from /sys/.../dev/name to /sys/class/gpio/gpioN
* node. Caller is responsible for unlinking.
*
* Returns zero on success, else an error.
*/
static int gpiod_export_link(struct device *dev, const char *name,
struct gpio_desc *desc)
{
int status = -EINVAL;
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return -EINVAL;
}
mutex_lock(&sysfs_lock);
if (test_bit(FLAG_EXPORT, &desc->flags)) {
struct device *tdev;
tdev = class_find_device(&gpio_class, NULL, desc, match_export);
if (tdev != NULL) {
status = sysfs_create_link(&dev->kobj, &tdev->kobj,
name);
} else {
status = -ENODEV;
}
}
mutex_unlock(&sysfs_lock);
if (status)
pr_debug("%s: gpio%d status %d\n", __func__, desc_to_gpio(desc),
status);
return status;
}
int gpio_export_link(struct device *dev, const char *name, unsigned gpio)
{
return gpiod_export_link(dev, name, gpio_to_desc(gpio));
}
EXPORT_SYMBOL_GPL(gpio_export_link);
/**
* gpio_sysfs_set_active_low - set the polarity of gpio sysfs value
* @gpio: gpio to change
* @value: non-zero to use active low, i.e. inverted values
*
* Set the polarity of /sys/class/gpio/gpioN/value sysfs attribute.
* The GPIO does not have to be exported yet. If poll(2) support has
* been enabled for either rising or falling edge, it will be
* reconfigured to follow the new polarity.
*
* Returns zero on success, else an error.
*/
static int gpiod_sysfs_set_active_low(struct gpio_desc *desc, int value)
{
struct device *dev = NULL;
int status = -EINVAL;
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return -EINVAL;
}
mutex_lock(&sysfs_lock);
if (test_bit(FLAG_EXPORT, &desc->flags)) {
dev = class_find_device(&gpio_class, NULL, desc, match_export);
if (dev == NULL) {
status = -ENODEV;
goto unlock;
}
}
status = sysfs_set_active_low(desc, dev, value);
unlock:
mutex_unlock(&sysfs_lock);
if (status)
pr_debug("%s: gpio%d status %d\n", __func__, desc_to_gpio(desc),
status);
return status;
}
int gpio_sysfs_set_active_low(unsigned gpio, int value)
{
return gpiod_sysfs_set_active_low(gpio_to_desc(gpio), value);
}
EXPORT_SYMBOL_GPL(gpio_sysfs_set_active_low);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
/**
* gpio_unexport - reverse effect of gpio_export()
* @gpio: gpio to make unavailable
*
* This is implicit on gpio_free().
*/
static void gpiod_unexport(struct gpio_desc *desc)
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
{
int status = 0;
Fix circular locking dependency (3.3-rc2) Hi, On Wed, Feb 8, 2012 at 8:41 PM, Felipe Balbi <balbi@ti.com> wrote: > Hi guys, > > I have just triggered the folllowing: > > [   84.860321] ====================================================== > [   84.860321] [ INFO: possible circular locking dependency detected ] > [   84.860321] 3.3.0-rc2-00026-ge4e8a39 #474 Not tainted > [   84.860321] ------------------------------------------------------- > [   84.860321] bash/949 is trying to acquire lock: > [   84.860321]  (sysfs_lock){+.+.+.}, at: [<c0275358>] gpio_value_store+0x24/0xcc > [   84.860321] > [   84.860321] but task is already holding lock: > [   84.860321]  (s_active#22){++++.+}, at: [<c016996c>] sysfs_write_file+0xdc/0x184 > [   84.911468] > [   84.911468] which lock already depends on the new lock. > [   84.911468] > [   84.920043] > [   84.920043] the existing dependency chain (in reverse order) is: > [   84.920043] > [   84.927886] -> #1 (s_active#22){++++.+}: > [   84.927886]        [<c008f640>] check_prevs_add+0xdc/0x150 > [   84.927886]        [<c008fc18>] validate_chain.clone.24+0x564/0x694 > [   84.927886]        [<c0090cdc>] __lock_acquire+0x49c/0x980 > [   84.951660]        [<c0091838>] lock_acquire+0x98/0x100 > [   84.951660]        [<c016a8e8>] sysfs_deactivate+0xb0/0x100 > [   84.962982]        [<c016b1b4>] sysfs_addrm_finish+0x2c/0x6c > [   84.962982]        [<c016b8bc>] sysfs_remove_dir+0x84/0x98 > [   84.962982]        [<c02590d8>] kobject_del+0x10/0x78 > [   84.974670]        [<c02c29e8>] device_del+0x140/0x170 > [   84.974670]        [<c02c2a24>] device_unregister+0xc/0x18 > [   84.985382]        [<c0276894>] gpio_unexport+0xbc/0xdc > [   84.985382]        [<c02768c8>] gpio_free+0x14/0xfc > [   85.001708]        [<c0276a28>] unexport_store+0x78/0x8c > [   85.001708]        [<c02c5af8>] class_attr_store+0x18/0x24 > [   85.007293]        [<c0169990>] sysfs_write_file+0x100/0x184 > [   85.018981]        [<c0109d48>] vfs_write+0xb4/0x148 > [   85.018981]        [<c0109fd0>] sys_write+0x40/0x70 > [   85.018981]        [<c0013cc0>] ret_fast_syscall+0x0/0x3c > [   85.035003] > [   85.035003] -> #0 (sysfs_lock){+.+.+.}: > [   85.035003]        [<c008f54c>] check_prev_add+0x680/0x698 > [   85.035003]        [<c008f640>] check_prevs_add+0xdc/0x150 > [   85.052093]        [<c008fc18>] validate_chain.clone.24+0x564/0x694 > [   85.052093]        [<c0090cdc>] __lock_acquire+0x49c/0x980 > [   85.052093]        [<c0091838>] lock_acquire+0x98/0x100 > [   85.069885]        [<c047e280>] mutex_lock_nested+0x3c/0x2f4 > [   85.069885]        [<c0275358>] gpio_value_store+0x24/0xcc > [   85.069885]        [<c02c18dc>] dev_attr_store+0x18/0x24 > [   85.087158]        [<c0169990>] sysfs_write_file+0x100/0x184 > [   85.087158]        [<c0109d48>] vfs_write+0xb4/0x148 > [   85.098297]        [<c0109fd0>] sys_write+0x40/0x70 > [   85.098297]        [<c0013cc0>] ret_fast_syscall+0x0/0x3c > [   85.109069] > [   85.109069] other info that might help us debug this: > [   85.109069] > [   85.117462]  Possible unsafe locking scenario: > [   85.117462] > [   85.117462]        CPU0                    CPU1 > [   85.128417]        ----                    ---- > [   85.128417]   lock(s_active#22); > [   85.128417]                                lock(sysfs_lock); > [   85.128417]                                lock(s_active#22); > [   85.142486]   lock(sysfs_lock); > [   85.151794] > [   85.151794]  *** DEADLOCK *** > [   85.151794] > [   85.151794] 2 locks held by bash/949: > [   85.158020]  #0:  (&buffer->mutex){+.+.+.}, at: [<c01698b8>] sysfs_write_file+0x28/0x184 > [   85.170349]  #1:  (s_active#22){++++.+}, at: [<c016996c>] sysfs_write_file+0xdc/0x184 > [   85.170349] > [   85.178588] stack backtrace: > [   85.178588] [<c001b824>] (unwind_backtrace+0x0/0xf0) from [<c008de64>] (print_circular_bug+0x100/0x114) > [   85.193023] [<c008de64>] (print_circular_bug+0x100/0x114) from [<c008f54c>] (check_prev_add+0x680/0x698) > [   85.193023] [<c008f54c>] (check_prev_add+0x680/0x698) from [<c008f640>] (check_prevs_add+0xdc/0x150) > [   85.212524] [<c008f640>] (check_prevs_add+0xdc/0x150) from [<c008fc18>] (validate_chain.clone.24+0x564/0x694) > [   85.212524] [<c008fc18>] (validate_chain.clone.24+0x564/0x694) from [<c0090cdc>] (__lock_acquire+0x49c/0x980) > [   85.233306] [<c0090cdc>] (__lock_acquire+0x49c/0x980) from [<c0091838>] (lock_acquire+0x98/0x100) > [   85.233306] [<c0091838>] (lock_acquire+0x98/0x100) from [<c047e280>] (mutex_lock_nested+0x3c/0x2f4) > [   85.242614] [<c047e280>] (mutex_lock_nested+0x3c/0x2f4) from [<c0275358>] (gpio_value_store+0x24/0xcc) > [   85.261840] [<c0275358>] (gpio_value_store+0x24/0xcc) from [<c02c18dc>] (dev_attr_store+0x18/0x24) > [   85.261840] [<c02c18dc>] (dev_attr_store+0x18/0x24) from [<c0169990>] (sysfs_write_file+0x100/0x184) > [   85.271240] [<c0169990>] (sysfs_write_file+0x100/0x184) from [<c0109d48>] (vfs_write+0xb4/0x148) > [   85.290008] [<c0109d48>] (vfs_write+0xb4/0x148) from [<c0109fd0>] (sys_write+0x40/0x70) > [   85.298400] [<c0109fd0>] (sys_write+0x40/0x70) from [<c0013cc0>] (ret_fast_syscall+0x0/0x3c) > -bash: echo: write error: Operation not permitted > > the way to trigger is: > > root@legolas:~# cd /sys/class/gpio/ > root@legolas:/sys/class/gpio# echo 2 > export > root@legolas:/sys/class/gpio# echo 2 > unexport > root@legolas:/sys/class/gpio# echo 2 > export > root@legolas:/sys/class/gpio# cd gpio2/ > root@legolas:/sys/class/gpio/gpio2# echo 1 > value Looks 'sysfs_lock' needn't to be held for unregister, so the patch below may fix the problem. Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
2012-02-13 14:53:20 +00:00
struct device *dev = NULL;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return;
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
mutex_lock(&sysfs_lock);
if (test_bit(FLAG_EXPORT, &desc->flags)) {
dev = class_find_device(&gpio_class, NULL, desc, match_export);
if (dev) {
gpio_setup_irq(desc, dev, 0);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
clear_bit(FLAG_EXPORT, &desc->flags);
} else
status = -ENODEV;
}
mutex_unlock(&sysfs_lock);
Fix circular locking dependency (3.3-rc2) Hi, On Wed, Feb 8, 2012 at 8:41 PM, Felipe Balbi <balbi@ti.com> wrote: > Hi guys, > > I have just triggered the folllowing: > > [   84.860321] ====================================================== > [   84.860321] [ INFO: possible circular locking dependency detected ] > [   84.860321] 3.3.0-rc2-00026-ge4e8a39 #474 Not tainted > [   84.860321] ------------------------------------------------------- > [   84.860321] bash/949 is trying to acquire lock: > [   84.860321]  (sysfs_lock){+.+.+.}, at: [<c0275358>] gpio_value_store+0x24/0xcc > [   84.860321] > [   84.860321] but task is already holding lock: > [   84.860321]  (s_active#22){++++.+}, at: [<c016996c>] sysfs_write_file+0xdc/0x184 > [   84.911468] > [   84.911468] which lock already depends on the new lock. > [   84.911468] > [   84.920043] > [   84.920043] the existing dependency chain (in reverse order) is: > [   84.920043] > [   84.927886] -> #1 (s_active#22){++++.+}: > [   84.927886]        [<c008f640>] check_prevs_add+0xdc/0x150 > [   84.927886]        [<c008fc18>] validate_chain.clone.24+0x564/0x694 > [   84.927886]        [<c0090cdc>] __lock_acquire+0x49c/0x980 > [   84.951660]        [<c0091838>] lock_acquire+0x98/0x100 > [   84.951660]        [<c016a8e8>] sysfs_deactivate+0xb0/0x100 > [   84.962982]        [<c016b1b4>] sysfs_addrm_finish+0x2c/0x6c > [   84.962982]        [<c016b8bc>] sysfs_remove_dir+0x84/0x98 > [   84.962982]        [<c02590d8>] kobject_del+0x10/0x78 > [   84.974670]        [<c02c29e8>] device_del+0x140/0x170 > [   84.974670]        [<c02c2a24>] device_unregister+0xc/0x18 > [   84.985382]        [<c0276894>] gpio_unexport+0xbc/0xdc > [   84.985382]        [<c02768c8>] gpio_free+0x14/0xfc > [   85.001708]        [<c0276a28>] unexport_store+0x78/0x8c > [   85.001708]        [<c02c5af8>] class_attr_store+0x18/0x24 > [   85.007293]        [<c0169990>] sysfs_write_file+0x100/0x184 > [   85.018981]        [<c0109d48>] vfs_write+0xb4/0x148 > [   85.018981]        [<c0109fd0>] sys_write+0x40/0x70 > [   85.018981]        [<c0013cc0>] ret_fast_syscall+0x0/0x3c > [   85.035003] > [   85.035003] -> #0 (sysfs_lock){+.+.+.}: > [   85.035003]        [<c008f54c>] check_prev_add+0x680/0x698 > [   85.035003]        [<c008f640>] check_prevs_add+0xdc/0x150 > [   85.052093]        [<c008fc18>] validate_chain.clone.24+0x564/0x694 > [   85.052093]        [<c0090cdc>] __lock_acquire+0x49c/0x980 > [   85.052093]        [<c0091838>] lock_acquire+0x98/0x100 > [   85.069885]        [<c047e280>] mutex_lock_nested+0x3c/0x2f4 > [   85.069885]        [<c0275358>] gpio_value_store+0x24/0xcc > [   85.069885]        [<c02c18dc>] dev_attr_store+0x18/0x24 > [   85.087158]        [<c0169990>] sysfs_write_file+0x100/0x184 > [   85.087158]        [<c0109d48>] vfs_write+0xb4/0x148 > [   85.098297]        [<c0109fd0>] sys_write+0x40/0x70 > [   85.098297]        [<c0013cc0>] ret_fast_syscall+0x0/0x3c > [   85.109069] > [   85.109069] other info that might help us debug this: > [   85.109069] > [   85.117462]  Possible unsafe locking scenario: > [   85.117462] > [   85.117462]        CPU0                    CPU1 > [   85.128417]        ----                    ---- > [   85.128417]   lock(s_active#22); > [   85.128417]                                lock(sysfs_lock); > [   85.128417]                                lock(s_active#22); > [   85.142486]   lock(sysfs_lock); > [   85.151794] > [   85.151794]  *** DEADLOCK *** > [   85.151794] > [   85.151794] 2 locks held by bash/949: > [   85.158020]  #0:  (&buffer->mutex){+.+.+.}, at: [<c01698b8>] sysfs_write_file+0x28/0x184 > [   85.170349]  #1:  (s_active#22){++++.+}, at: [<c016996c>] sysfs_write_file+0xdc/0x184 > [   85.170349] > [   85.178588] stack backtrace: > [   85.178588] [<c001b824>] (unwind_backtrace+0x0/0xf0) from [<c008de64>] (print_circular_bug+0x100/0x114) > [   85.193023] [<c008de64>] (print_circular_bug+0x100/0x114) from [<c008f54c>] (check_prev_add+0x680/0x698) > [   85.193023] [<c008f54c>] (check_prev_add+0x680/0x698) from [<c008f640>] (check_prevs_add+0xdc/0x150) > [   85.212524] [<c008f640>] (check_prevs_add+0xdc/0x150) from [<c008fc18>] (validate_chain.clone.24+0x564/0x694) > [   85.212524] [<c008fc18>] (validate_chain.clone.24+0x564/0x694) from [<c0090cdc>] (__lock_acquire+0x49c/0x980) > [   85.233306] [<c0090cdc>] (__lock_acquire+0x49c/0x980) from [<c0091838>] (lock_acquire+0x98/0x100) > [   85.233306] [<c0091838>] (lock_acquire+0x98/0x100) from [<c047e280>] (mutex_lock_nested+0x3c/0x2f4) > [   85.242614] [<c047e280>] (mutex_lock_nested+0x3c/0x2f4) from [<c0275358>] (gpio_value_store+0x24/0xcc) > [   85.261840] [<c0275358>] (gpio_value_store+0x24/0xcc) from [<c02c18dc>] (dev_attr_store+0x18/0x24) > [   85.261840] [<c02c18dc>] (dev_attr_store+0x18/0x24) from [<c0169990>] (sysfs_write_file+0x100/0x184) > [   85.271240] [<c0169990>] (sysfs_write_file+0x100/0x184) from [<c0109d48>] (vfs_write+0xb4/0x148) > [   85.290008] [<c0109d48>] (vfs_write+0xb4/0x148) from [<c0109fd0>] (sys_write+0x40/0x70) > [   85.298400] [<c0109fd0>] (sys_write+0x40/0x70) from [<c0013cc0>] (ret_fast_syscall+0x0/0x3c) > -bash: echo: write error: Operation not permitted > > the way to trigger is: > > root@legolas:~# cd /sys/class/gpio/ > root@legolas:/sys/class/gpio# echo 2 > export > root@legolas:/sys/class/gpio# echo 2 > unexport > root@legolas:/sys/class/gpio# echo 2 > export > root@legolas:/sys/class/gpio# cd gpio2/ > root@legolas:/sys/class/gpio/gpio2# echo 1 > value Looks 'sysfs_lock' needn't to be held for unregister, so the patch below may fix the problem. Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
2012-02-13 14:53:20 +00:00
if (dev) {
device_unregister(dev);
put_device(dev);
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (status)
pr_debug("%s: gpio%d status %d\n", __func__, desc_to_gpio(desc),
status);
}
void gpio_unexport(unsigned gpio)
{
gpiod_unexport(gpio_to_desc(gpio));
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
}
EXPORT_SYMBOL_GPL(gpio_unexport);
static int gpiochip_export(struct gpio_chip *chip)
{
int status;
struct device *dev;
/* Many systems register gpio chips for SOC support very early,
* before driver model support is available. In those cases we
* export this later, in gpiolib_sysfs_init() ... here we just
* verify that _some_ field of gpio_class got initialized.
*/
if (!gpio_class.p)
return 0;
/* use chip->base for the ID; it's already known to be unique */
mutex_lock(&sysfs_lock);
dev = device_create(&gpio_class, chip->dev, MKDEV(0, 0), chip,
"gpiochip%d", chip->base);
if (!IS_ERR(dev)) {
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = sysfs_create_group(&dev->kobj,
&gpiochip_attr_group);
} else
status = PTR_ERR(dev);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
chip->exported = (status == 0);
mutex_unlock(&sysfs_lock);
if (status) {
unsigned long flags;
unsigned gpio;
spin_lock_irqsave(&gpio_lock, flags);
gpio = 0;
while (gpio < chip->ngpio)
chip->desc[gpio++].chip = NULL;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
spin_unlock_irqrestore(&gpio_lock, flags);
pr_debug("%s: chip %s status %d\n", __func__,
chip->label, status);
}
return status;
}
static void gpiochip_unexport(struct gpio_chip *chip)
{
int status;
struct device *dev;
mutex_lock(&sysfs_lock);
dev = class_find_device(&gpio_class, NULL, chip, match_export);
if (dev) {
put_device(dev);
device_unregister(dev);
chip->exported = 0;
status = 0;
} else
status = -ENODEV;
mutex_unlock(&sysfs_lock);
if (status)
pr_debug("%s: chip %s status %d\n", __func__,
chip->label, status);
}
static int __init gpiolib_sysfs_init(void)
{
int status;
unsigned long flags;
struct gpio_chip *chip;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
status = class_register(&gpio_class);
if (status < 0)
return status;
/* Scan and register the gpio_chips which registered very
* early (e.g. before the class_register above was called).
*
* We run before arch_initcall() so chip->dev nodes can have
* registered, and so arch_initcall() can always gpio_export().
*/
spin_lock_irqsave(&gpio_lock, flags);
list_for_each_entry(chip, &gpio_chips, list) {
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (!chip || chip->exported)
continue;
spin_unlock_irqrestore(&gpio_lock, flags);
status = gpiochip_export(chip);
spin_lock_irqsave(&gpio_lock, flags);
}
spin_unlock_irqrestore(&gpio_lock, flags);
return status;
}
postcore_initcall(gpiolib_sysfs_init);
#else
static inline int gpiochip_export(struct gpio_chip *chip)
{
return 0;
}
static inline void gpiochip_unexport(struct gpio_chip *chip)
{
}
static inline int gpiod_export(struct gpio_desc *desc,
bool direction_may_change)
{
return -ENOSYS;
}
static inline int gpiod_export_link(struct device *dev, const char *name,
struct gpio_desc *desc)
{
return -ENOSYS;
}
static inline int gpiod_sysfs_set_active_low(struct gpio_desc *desc, int value)
{
return -ENOSYS;
}
static inline void gpiod_unexport(struct gpio_desc *desc)
{
}
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
#endif /* CONFIG_GPIO_SYSFS */
/*
* Add a new chip to the global chips list, keeping the list of chips sorted
* by base order.
*
* Return -EBUSY if the new chip overlaps with some other chip's integer
* space.
*/
static int gpiochip_add_to_list(struct gpio_chip *chip)
{
struct list_head *pos = &gpio_chips;
struct gpio_chip *_chip;
int err = 0;
/* find where to insert our chip */
list_for_each(pos, &gpio_chips) {
_chip = list_entry(pos, struct gpio_chip, list);
/* shall we insert before _chip? */
if (_chip->base >= chip->base + chip->ngpio)
break;
}
/* are we stepping on the chip right before? */
if (pos != &gpio_chips && pos->prev != &gpio_chips) {
_chip = list_entry(pos->prev, struct gpio_chip, list);
if (_chip->base + _chip->ngpio > chip->base) {
dev_err(chip->dev,
"GPIO integer space overlap, cannot add chip\n");
err = -EBUSY;
}
}
if (!err)
list_add_tail(&chip->list, pos);
return err;
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/**
* gpiochip_add() - register a gpio_chip
* @chip: the chip to register, with chip->base initialized
* Context: potentially before irqs or kmalloc will work
*
* Returns a negative errno if the chip can't be registered, such as
* because the chip->base is invalid or already associated with a
* different chip. Otherwise it returns zero as a success code.
*
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
* When gpiochip_add() is called very early during boot, so that GPIOs
* can be freely used, the chip->dev device must be registered before
* the gpio framework's arch_initcall(). Otherwise sysfs initialization
* for GPIOs will fail rudely.
*
* If chip->base is negative, this requests dynamic assignment of
* a range of valid GPIOs.
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
*/
int gpiochip_add(struct gpio_chip *chip)
{
unsigned long flags;
int status = 0;
unsigned id;
int base = chip->base;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if ((!gpio_is_valid(base) || !gpio_is_valid(base + chip->ngpio - 1))
&& base >= 0) {
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
status = -EINVAL;
goto fail;
}
spin_lock_irqsave(&gpio_lock, flags);
if (base < 0) {
base = gpiochip_find_base(chip->ngpio);
if (base < 0) {
status = base;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
goto unlock;
}
chip->base = base;
}
status = gpiochip_add_to_list(chip);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (status == 0) {
chip->desc = &gpio_desc[chip->base];
for (id = 0; id < chip->ngpio; id++) {
struct gpio_desc *desc = &chip->desc[id];
desc->chip = chip;
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
/* REVISIT: most hardware initializes GPIOs as
* inputs (often with pullups enabled) so power
* usage is minimized. Linux code should set the
* gpio direction first thing; but until it does,
* and in case chip->get_direction is not set,
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
* we may expose the wrong direction in sysfs.
*/
desc->flags = !chip->direction_input
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
? (1 << FLAG_IS_OUT)
: 0;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
}
#ifdef CONFIG_PINCTRL
INIT_LIST_HEAD(&chip->pin_ranges);
#endif
of_gpiochip_add(chip);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
unlock:
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
spin_unlock_irqrestore(&gpio_lock, flags);
if (status)
goto fail;
status = gpiochip_export(chip);
if (status)
goto fail;
pr_debug("gpiochip_add: registered GPIOs %d to %d on device: %s\n",
chip->base, chip->base + chip->ngpio - 1,
chip->label ? : "generic");
return 0;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
fail:
/* failures here can mean systems won't boot... */
pr_err("gpiochip_add: gpios %d..%d (%s) failed to register\n",
chip->base, chip->base + chip->ngpio - 1,
chip->label ? : "generic");
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return status;
}
EXPORT_SYMBOL_GPL(gpiochip_add);
/**
* gpiochip_remove() - unregister a gpio_chip
* @chip: the chip to unregister
*
* A gpio_chip with any GPIOs still requested may not be removed.
*/
int gpiochip_remove(struct gpio_chip *chip)
{
unsigned long flags;
int status = 0;
unsigned id;
spin_lock_irqsave(&gpio_lock, flags);
gpiochip_remove_pin_ranges(chip);
of_gpiochip_remove(chip);
for (id = 0; id < chip->ngpio; id++) {
if (test_bit(FLAG_REQUESTED, &chip->desc[id].flags)) {
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
status = -EBUSY;
break;
}
}
if (status == 0) {
for (id = 0; id < chip->ngpio; id++)
chip->desc[id].chip = NULL;
list_del(&chip->list);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
spin_unlock_irqrestore(&gpio_lock, flags);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
if (status == 0)
gpiochip_unexport(chip);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return status;
}
EXPORT_SYMBOL_GPL(gpiochip_remove);
/**
* gpiochip_find() - iterator for locating a specific gpio_chip
* @data: data to pass to match function
* @callback: Callback function to check gpio_chip
*
* Similar to bus_find_device. It returns a reference to a gpio_chip as
* determined by a user supplied @match callback. The callback should return
* 0 if the device doesn't match and non-zero if it does. If the callback is
* non-zero, this function will return to the caller and not iterate over any
* more gpio_chips.
*/
struct gpio_chip *gpiochip_find(void *data,
int (*match)(struct gpio_chip *chip,
void *data))
{
struct gpio_chip *chip;
unsigned long flags;
spin_lock_irqsave(&gpio_lock, flags);
list_for_each_entry(chip, &gpio_chips, list)
if (match(chip, data))
break;
/* No match? */
if (&chip->list == &gpio_chips)
chip = NULL;
spin_unlock_irqrestore(&gpio_lock, flags);
return chip;
}
EXPORT_SYMBOL_GPL(gpiochip_find);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#ifdef CONFIG_PINCTRL
/**
* gpiochip_add_pin_range() - add a range for GPIO <-> pin mapping
* @chip: the gpiochip to add the range for
* @pinctrl_name: the dev_name() of the pin controller to map to
* @gpio_offset: the start offset in the current gpio_chip number space
* @pin_offset: the start offset in the pin controller number space
* @npins: the number of pins from the offset of each pin space (GPIO and
* pin controller) to accumulate in this range
*/
int gpiochip_add_pin_range(struct gpio_chip *chip, const char *pinctl_name,
unsigned int gpio_offset, unsigned int pin_offset,
unsigned int npins)
{
struct gpio_pin_range *pin_range;
int ret;
pin_range = kzalloc(sizeof(*pin_range), GFP_KERNEL);
if (!pin_range) {
pr_err("%s: GPIO chip: failed to allocate pin ranges\n",
chip->label);
return -ENOMEM;
}
/* Use local offset as range ID */
pin_range->range.id = gpio_offset;
pin_range->range.gc = chip;
pin_range->range.name = chip->label;
pin_range->range.base = chip->base + gpio_offset;
pin_range->range.pin_base = pin_offset;
pin_range->range.npins = npins;
pin_range->pctldev = pinctrl_find_and_add_gpio_range(pinctl_name,
&pin_range->range);
if (IS_ERR(pin_range->pctldev)) {
ret = PTR_ERR(pin_range->pctldev);
pr_err("%s: GPIO chip: could not create pin range\n",
chip->label);
kfree(pin_range);
return ret;
}
pr_debug("GPIO chip %s: created GPIO range %d->%d ==> %s PIN %d->%d\n",
chip->label, gpio_offset, gpio_offset + npins - 1,
pinctl_name,
pin_offset, pin_offset + npins - 1);
list_add_tail(&pin_range->node, &chip->pin_ranges);
return 0;
}
EXPORT_SYMBOL_GPL(gpiochip_add_pin_range);
/**
* gpiochip_remove_pin_ranges() - remove all the GPIO <-> pin mappings
* @chip: the chip to remove all the mappings for
*/
void gpiochip_remove_pin_ranges(struct gpio_chip *chip)
{
struct gpio_pin_range *pin_range, *tmp;
list_for_each_entry_safe(pin_range, tmp, &chip->pin_ranges, node) {
list_del(&pin_range->node);
pinctrl_remove_gpio_range(pin_range->pctldev,
&pin_range->range);
kfree(pin_range);
}
}
EXPORT_SYMBOL_GPL(gpiochip_remove_pin_ranges);
#endif /* CONFIG_PINCTRL */
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* These "optional" allocation calls help prevent drivers from stomping
* on each other, and help provide better diagnostics in debugfs.
* They're called even less than the "set direction" calls.
*/
static int gpiod_request(struct gpio_desc *desc, const char *label)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
struct gpio_chip *chip;
int status = -EPROBE_DEFER;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
unsigned long flags;
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return -EINVAL;
}
spin_lock_irqsave(&gpio_lock, flags);
chip = desc->chip;
if (chip == NULL)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
goto done;
if (!try_module_get(chip->owner))
goto done;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* NOTE: gpio_request() can be called in early boot,
* before IRQs are enabled, for non-sleeping (SOC) GPIOs.
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
*/
if (test_and_set_bit(FLAG_REQUESTED, &desc->flags) == 0) {
desc_set_label(desc, label ? : "?");
status = 0;
} else {
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
status = -EBUSY;
module_put(chip->owner);
goto done;
}
if (chip->request) {
/* chip->request may sleep */
spin_unlock_irqrestore(&gpio_lock, flags);
status = chip->request(chip, gpio_chip_hwgpio(desc));
spin_lock_irqsave(&gpio_lock, flags);
if (status < 0) {
desc_set_label(desc, NULL);
module_put(chip->owner);
clear_bit(FLAG_REQUESTED, &desc->flags);
goto done;
}
}
if (chip->get_direction) {
/* chip->get_direction may sleep */
spin_unlock_irqrestore(&gpio_lock, flags);
gpiod_get_direction(desc);
spin_lock_irqsave(&gpio_lock, flags);
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
done:
if (status)
pr_debug("_gpio_request: gpio-%d (%s) status %d\n",
desc_to_gpio(desc), label ? : "?", status);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
spin_unlock_irqrestore(&gpio_lock, flags);
return status;
}
int gpio_request(unsigned gpio, const char *label)
{
return gpiod_request(gpio_to_desc(gpio), label);
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
EXPORT_SYMBOL_GPL(gpio_request);
static void gpiod_free(struct gpio_desc *desc)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
unsigned long flags;
struct gpio_chip *chip;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
might_sleep();
if (!desc) {
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
WARN_ON(extra_checks);
return;
}
gpiod_unexport(desc);
gpio: sysfs interface This adds a simple sysfs interface for GPIOs. /sys/class/gpio /export ... asks the kernel to export a GPIO to userspace /unexport ... to return a GPIO to the kernel /gpioN ... for each exported GPIO #N /value ... always readable, writes fail for input GPIOs /direction ... r/w as: in, out (default low); write high, low /gpiochipN ... for each gpiochip; #N is its first GPIO /base ... (r/o) same as N /label ... (r/o) descriptive, not necessarily unique /ngpio ... (r/o) number of GPIOs; numbered N .. N+(ngpio - 1) GPIOs claimed by kernel code may be exported by its owner using a new gpio_export() call, which should be most useful for driver debugging. Such exports may optionally be done without a "direction" attribute. Userspace may ask to take over a GPIO by writing to a sysfs control file, helping to cope with incomplete board support or other "one-off" requirements that don't merit full kernel support: echo 23 > /sys/class/gpio/export ... will gpio_request(23, "sysfs") and gpio_export(23); use /sys/class/gpio/gpio-23/direction to (re)configure it, when that GPIO can be used as both input and output. echo 23 > /sys/class/gpio/unexport ... will gpio_free(23), when it was exported as above The extra D-space footprint is a few hundred bytes, except for the sysfs resources associated with each exported GPIO. The additional I-space footprint is about two thirds of the current size of gpiolib (!). Since no /dev node creation is involved, no "udev" support is needed. Related changes: * This adds a device pointer to "struct gpio_chip". When GPIO providers initialize that, sysfs gpio class devices become children of that device instead of being "virtual" devices. * The (few) gpio_chip providers which have such a device node have been updated. * Some gpio_chip drivers also needed to update their module "owner" field ... for which missing kerneldoc was added. * Some gpio_chips don't support input GPIOs. Those GPIOs are now flagged appropriately when the chip is registered. Based on previous patches, and discussion both on and off LKML. A Documentation/ABI/testing/sysfs-gpio update is ready to submit once this merges to mainline. [akpm@linux-foundation.org: a few maintenance build fixes] Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de> Cc: Greg KH <greg@kroah.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-25 08:46:07 +00:00
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
spin_lock_irqsave(&gpio_lock, flags);
chip = desc->chip;
if (chip && test_bit(FLAG_REQUESTED, &desc->flags)) {
if (chip->free) {
spin_unlock_irqrestore(&gpio_lock, flags);
might_sleep_if(chip->can_sleep);
chip->free(chip, gpio_chip_hwgpio(desc));
spin_lock_irqsave(&gpio_lock, flags);
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
desc_set_label(desc, NULL);
module_put(desc->chip->owner);
clear_bit(FLAG_ACTIVE_LOW, &desc->flags);
clear_bit(FLAG_REQUESTED, &desc->flags);
clear_bit(FLAG_OPEN_DRAIN, &desc->flags);
clear_bit(FLAG_OPEN_SOURCE, &desc->flags);
} else
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
WARN_ON(extra_checks);
spin_unlock_irqrestore(&gpio_lock, flags);
}
void gpio_free(unsigned gpio)
{
gpiod_free(gpio_to_desc(gpio));
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
EXPORT_SYMBOL_GPL(gpio_free);
/**
* gpio_request_one - request a single GPIO with initial configuration
* @gpio: the GPIO number
* @flags: GPIO configuration as specified by GPIOF_*
* @label: a literal description string of this GPIO
*/
int gpio_request_one(unsigned gpio, unsigned long flags, const char *label)
{
struct gpio_desc *desc;
int err;
desc = gpio_to_desc(gpio);
err = gpiod_request(desc, label);
if (err)
return err;
if (flags & GPIOF_OPEN_DRAIN)
set_bit(FLAG_OPEN_DRAIN, &desc->flags);
if (flags & GPIOF_OPEN_SOURCE)
set_bit(FLAG_OPEN_SOURCE, &desc->flags);
if (flags & GPIOF_DIR_IN)
err = gpiod_direction_input(desc);
else
err = gpiod_direction_output(desc,
(flags & GPIOF_INIT_HIGH) ? 1 : 0);
if (err)
goto free_gpio;
if (flags & GPIOF_EXPORT) {
err = gpiod_export(desc, flags & GPIOF_EXPORT_CHANGEABLE);
if (err)
goto free_gpio;
}
return 0;
free_gpio:
gpiod_free(desc);
return err;
}
EXPORT_SYMBOL_GPL(gpio_request_one);
/**
* gpio_request_array - request multiple GPIOs in a single call
* @array: array of the 'struct gpio'
* @num: how many GPIOs in the array
*/
int gpio_request_array(const struct gpio *array, size_t num)
{
int i, err;
for (i = 0; i < num; i++, array++) {
err = gpio_request_one(array->gpio, array->flags, array->label);
if (err)
goto err_free;
}
return 0;
err_free:
while (i--)
gpio_free((--array)->gpio);
return err;
}
EXPORT_SYMBOL_GPL(gpio_request_array);
/**
* gpio_free_array - release multiple GPIOs in a single call
* @array: array of the 'struct gpio'
* @num: how many GPIOs in the array
*/
void gpio_free_array(const struct gpio *array, size_t num)
{
while (num--)
gpio_free((array++)->gpio);
}
EXPORT_SYMBOL_GPL(gpio_free_array);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/**
* gpiochip_is_requested - return string iff signal was requested
* @chip: controller managing the signal
* @offset: of signal within controller's 0..(ngpio - 1) range
*
* Returns NULL if the GPIO is not currently requested, else a string.
* If debugfs support is enabled, the string returned is the label passed
* to gpio_request(); otherwise it is a meaningless constant.
*
* This function is for use by GPIO controller drivers. The label can
* help with diagnostics, and knowing that the signal is used as a GPIO
* can help avoid accidentally multiplexing it to another controller.
*/
const char *gpiochip_is_requested(struct gpio_chip *chip, unsigned offset)
{
struct gpio_desc *desc;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (!GPIO_OFFSET_VALID(chip, offset))
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return NULL;
desc = &chip->desc[offset];
if (test_bit(FLAG_REQUESTED, &desc->flags) == 0)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return NULL;
#ifdef CONFIG_DEBUG_FS
return desc->label;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#else
return "?";
#endif
}
EXPORT_SYMBOL_GPL(gpiochip_is_requested);
/* Drivers MUST set GPIO direction before making get/set calls. In
* some cases this is done in early boot, before IRQs are enabled.
*
* As a rule these aren't called more than once (except for drivers
* using the open-drain emulation idiom) so these are natural places
* to accumulate extra debugging checks. Note that we can't (yet)
* rely on gpio_request() having been called beforehand.
*/
static int gpiod_direction_input(struct gpio_desc *desc)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
unsigned long flags;
struct gpio_chip *chip;
int status = -EINVAL;
int offset;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return -EINVAL;
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
spin_lock_irqsave(&gpio_lock, flags);
chip = desc->chip;
if (!chip || !chip->get || !chip->direction_input)
goto fail;
status = gpio_ensure_requested(desc);
if (status < 0)
goto fail;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* now we know the gpio is valid and chip won't vanish */
spin_unlock_irqrestore(&gpio_lock, flags);
might_sleep_if(chip->can_sleep);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
offset = gpio_chip_hwgpio(desc);
if (status) {
status = chip->request(chip, offset);
if (status < 0) {
pr_debug("GPIO-%d: chip request fail, %d\n",
desc_to_gpio(desc), status);
/* and it's not available to anyone else ...
* gpio_request() is the fully clean solution.
*/
goto lose;
}
}
status = chip->direction_input(chip, offset);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (status == 0)
clear_bit(FLAG_IS_OUT, &desc->flags);
trace_gpio_direction(desc_to_gpio(desc), 1, status);
lose:
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return status;
fail:
spin_unlock_irqrestore(&gpio_lock, flags);
if (status)
pr_debug("%s: gpio-%d status %d\n", __func__,
desc_to_gpio(desc), status);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return status;
}
int gpio_direction_input(unsigned gpio)
{
return gpiod_direction_input(gpio_to_desc(gpio));
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
EXPORT_SYMBOL_GPL(gpio_direction_input);
static int gpiod_direction_output(struct gpio_desc *desc, int value)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
unsigned long flags;
struct gpio_chip *chip;
int status = -EINVAL;
int offset;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return -EINVAL;
}
/* Open drain pin should not be driven to 1 */
if (value && test_bit(FLAG_OPEN_DRAIN, &desc->flags))
return gpiod_direction_input(desc);
/* Open source pin should not be driven to 0 */
if (!value && test_bit(FLAG_OPEN_SOURCE, &desc->flags))
return gpiod_direction_input(desc);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
spin_lock_irqsave(&gpio_lock, flags);
chip = desc->chip;
if (!chip || !chip->set || !chip->direction_output)
goto fail;
status = gpio_ensure_requested(desc);
if (status < 0)
goto fail;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* now we know the gpio is valid and chip won't vanish */
spin_unlock_irqrestore(&gpio_lock, flags);
might_sleep_if(chip->can_sleep);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
offset = gpio_chip_hwgpio(desc);
if (status) {
status = chip->request(chip, offset);
if (status < 0) {
pr_debug("GPIO-%d: chip request fail, %d\n",
desc_to_gpio(desc), status);
/* and it's not available to anyone else ...
* gpio_request() is the fully clean solution.
*/
goto lose;
}
}
status = chip->direction_output(chip, offset, value);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (status == 0)
set_bit(FLAG_IS_OUT, &desc->flags);
trace_gpio_value(desc_to_gpio(desc), 0, value);
trace_gpio_direction(desc_to_gpio(desc), 0, status);
lose:
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return status;
fail:
spin_unlock_irqrestore(&gpio_lock, flags);
if (status)
pr_debug("%s: gpio-%d status %d\n", __func__,
desc_to_gpio(desc), status);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return status;
}
int gpio_direction_output(unsigned gpio, int value)
{
return gpiod_direction_output(gpio_to_desc(gpio), value);
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
EXPORT_SYMBOL_GPL(gpio_direction_output);
2010-05-26 21:42:23 +00:00
/**
* gpio_set_debounce - sets @debounce time for a @gpio
* @gpio: the gpio to set debounce time
* @debounce: debounce time is microseconds
*/
static int gpiod_set_debounce(struct gpio_desc *desc, unsigned debounce)
2010-05-26 21:42:23 +00:00
{
unsigned long flags;
struct gpio_chip *chip;
int status = -EINVAL;
int offset;
2010-05-26 21:42:23 +00:00
if (!desc) {
pr_warn("%s: invalid GPIO\n", __func__);
return -EINVAL;
}
2010-05-26 21:42:23 +00:00
spin_lock_irqsave(&gpio_lock, flags);
chip = desc->chip;
if (!chip || !chip->set || !chip->set_debounce)
goto fail;
status = gpio_ensure_requested(desc);
2010-05-26 21:42:23 +00:00
if (status < 0)
goto fail;
/* now we know the gpio is valid and chip won't vanish */
spin_unlock_irqrestore(&gpio_lock, flags);
might_sleep_if(chip->can_sleep);
2010-05-26 21:42:23 +00:00
offset = gpio_chip_hwgpio(desc);
return chip->set_debounce(chip, offset, debounce);
2010-05-26 21:42:23 +00:00
fail:
spin_unlock_irqrestore(&gpio_lock, flags);
if (status)
pr_debug("%s: gpio-%d status %d\n", __func__,
desc_to_gpio(desc), status);
2010-05-26 21:42:23 +00:00
return status;
}
int gpio_set_debounce(unsigned gpio, unsigned debounce)
{
return gpiod_set_debounce(gpio_to_desc(gpio), debounce);
}
2010-05-26 21:42:23 +00:00
EXPORT_SYMBOL_GPL(gpio_set_debounce);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* I/O calls are only valid after configuration completed; the relevant
* "is this a valid GPIO" error checks should already have been done.
*
* "Get" operations are often inlinable as reading a pin value register,
* and masking the relevant bit in that register.
*
* When "set" operations are inlinable, they involve writing that mask to
* one register to set a low value, or a different register to set it high.
* Otherwise locking is needed, so there may be little value to inlining.
*
*------------------------------------------------------------------------
*
* IMPORTANT!!! The hot paths -- get/set value -- assume that callers
* have requested the GPIO. That can include implicit requesting by
* a direction setting call. Marking a gpio as requested locks its chip
* in memory, guaranteeing that these table lookups need no more locking
* and that gpiochip_remove() will fail.
*
* REVISIT when debugging, consider adding some instrumentation to ensure
* that the GPIO was actually requested.
*/
/**
* __gpio_get_value() - return a gpio's value
* @gpio: gpio whose value will be returned
* Context: any
*
* This is used directly or indirectly to implement gpio_get_value().
* It returns the zero or nonzero value provided by the associated
* gpio_chip.get() method; or zero if no such method is provided.
*/
static int gpiod_get_value(const struct gpio_desc *desc)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
struct gpio_chip *chip;
int value;
int offset;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
if (!desc)
return 0;
chip = desc->chip;
offset = gpio_chip_hwgpio(desc);
/* Should be using gpio_get_value_cansleep() */
WARN_ON(chip->can_sleep);
value = chip->get ? chip->get(chip, offset) : 0;
trace_gpio_value(desc_to_gpio(desc), 1, value);
return value;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
int __gpio_get_value(unsigned gpio)
{
return gpiod_get_value(gpio_to_desc(gpio));
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
EXPORT_SYMBOL_GPL(__gpio_get_value);
/*
* _gpio_set_open_drain_value() - Set the open drain gpio's value.
* @gpio: Gpio whose state need to be set.
* @chip: Gpio chip.
* @value: Non-zero for setting it HIGH otherise it will set to LOW.
*/
static void _gpio_set_open_drain_value(struct gpio_desc *desc, int value)
{
int err = 0;
struct gpio_chip *chip = desc->chip;
int offset = gpio_chip_hwgpio(desc);
if (value) {
err = chip->direction_input(chip, offset);
if (!err)
clear_bit(FLAG_IS_OUT, &desc->flags);
} else {
err = chip->direction_output(chip, offset, 0);
if (!err)
set_bit(FLAG_IS_OUT, &desc->flags);
}
trace_gpio_direction(desc_to_gpio(desc), value, err);
if (err < 0)
pr_err("%s: Error in set_value for open drain gpio%d err %d\n",
__func__, desc_to_gpio(desc), err);
}
/*
* _gpio_set_open_source() - Set the open source gpio's value.
* @gpio: Gpio whose state need to be set.
* @chip: Gpio chip.
* @value: Non-zero for setting it HIGH otherise it will set to LOW.
*/
static void _gpio_set_open_source_value(struct gpio_desc *desc, int value)
{
int err = 0;
struct gpio_chip *chip = desc->chip;
int offset = gpio_chip_hwgpio(desc);
if (value) {
err = chip->direction_output(chip, offset, 1);
if (!err)
set_bit(FLAG_IS_OUT, &desc->flags);
} else {
err = chip->direction_input(chip, offset);
if (!err)
clear_bit(FLAG_IS_OUT, &desc->flags);
}
trace_gpio_direction(desc_to_gpio(desc), !value, err);
if (err < 0)
pr_err("%s: Error in set_value for open source gpio%d err %d\n",
__func__, desc_to_gpio(desc), err);
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/**
* __gpio_set_value() - assign a gpio's value
* @gpio: gpio whose value will be assigned
* @value: value to assign
* Context: any
*
* This is used directly or indirectly to implement gpio_set_value().
* It invokes the associated gpio_chip.set() method.
*/
static void gpiod_set_value(struct gpio_desc *desc, int value)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
struct gpio_chip *chip;
if (!desc)
return;
chip = desc->chip;
/* Should be using gpio_set_value_cansleep() */
WARN_ON(chip->can_sleep);
trace_gpio_value(desc_to_gpio(desc), 0, value);
if (test_bit(FLAG_OPEN_DRAIN, &desc->flags))
_gpio_set_open_drain_value(desc, value);
else if (test_bit(FLAG_OPEN_SOURCE, &desc->flags))
_gpio_set_open_source_value(desc, value);
else
chip->set(chip, gpio_chip_hwgpio(desc), value);
}
void __gpio_set_value(unsigned gpio, int value)
{
return gpiod_set_value(gpio_to_desc(gpio), value);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
EXPORT_SYMBOL_GPL(__gpio_set_value);
/**
* __gpio_cansleep() - report whether gpio value access will sleep
* @gpio: gpio in question
* Context: any
*
* This is used directly or indirectly to implement gpio_cansleep(). It
* returns nonzero if access reading or writing the GPIO value can sleep.
*/
static int gpiod_cansleep(const struct gpio_desc *desc)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
if (!desc)
return 0;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* only call this on GPIOs that are valid! */
return desc->chip->can_sleep;
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
int __gpio_cansleep(unsigned gpio)
{
return gpiod_cansleep(gpio_to_desc(gpio));
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
EXPORT_SYMBOL_GPL(__gpio_cansleep);
/**
* __gpio_to_irq() - return the IRQ corresponding to a GPIO
* @gpio: gpio whose IRQ will be returned (already requested)
* Context: any
*
* This is used directly or indirectly to implement gpio_to_irq().
* It returns the number of the IRQ signaled by this (input) GPIO,
* or a negative errno.
*/
static int gpiod_to_irq(const struct gpio_desc *desc)
{
struct gpio_chip *chip;
int offset;
if (!desc)
return -EINVAL;
chip = desc->chip;
offset = gpio_chip_hwgpio(desc);
return chip->to_irq ? chip->to_irq(chip, offset) : -ENXIO;
}
int __gpio_to_irq(unsigned gpio)
{
return gpiod_to_irq(gpio_to_desc(gpio));
}
EXPORT_SYMBOL_GPL(__gpio_to_irq);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
/* There's no value in making it easy to inline GPIO calls that may sleep.
* Common examples include ones connected to I2C or SPI chips.
*/
static int gpiod_get_value_cansleep(const struct gpio_desc *desc)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
struct gpio_chip *chip;
int value;
int offset;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
might_sleep_if(extra_checks);
if (!desc)
return 0;
chip = desc->chip;
offset = gpio_chip_hwgpio(desc);
value = chip->get ? chip->get(chip, offset) : 0;
trace_gpio_value(desc_to_gpio(desc), 1, value);
return value;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
int gpio_get_value_cansleep(unsigned gpio)
{
return gpiod_get_value_cansleep(gpio_to_desc(gpio));
}
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
EXPORT_SYMBOL_GPL(gpio_get_value_cansleep);
static void gpiod_set_value_cansleep(struct gpio_desc *desc, int value)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
struct gpio_chip *chip;
might_sleep_if(extra_checks);
if (!desc)
return;
chip = desc->chip;
trace_gpio_value(desc_to_gpio(desc), 0, value);
if (test_bit(FLAG_OPEN_DRAIN, &desc->flags))
_gpio_set_open_drain_value(desc, value);
else if (test_bit(FLAG_OPEN_SOURCE, &desc->flags))
_gpio_set_open_source_value(desc, value);
else
chip->set(chip, gpio_chip_hwgpio(desc), value);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
void gpio_set_value_cansleep(unsigned gpio, int value)
{
return gpiod_set_value_cansleep(gpio_to_desc(gpio), value);
}
EXPORT_SYMBOL_GPL(gpio_set_value_cansleep);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
#ifdef CONFIG_DEBUG_FS
static void gpiolib_dbg_show(struct seq_file *s, struct gpio_chip *chip)
{
unsigned i;
unsigned gpio = chip->base;
struct gpio_desc *gdesc = &chip->desc[0];
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
int is_out;
for (i = 0; i < chip->ngpio; i++, gpio++, gdesc++) {
if (!test_bit(FLAG_REQUESTED, &gdesc->flags))
continue;
gpiod_get_direction(gdesc);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
is_out = test_bit(FLAG_IS_OUT, &gdesc->flags);
seq_printf(s, " gpio-%-3d (%-20.20s) %s %s",
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
gpio, gdesc->label,
is_out ? "out" : "in ",
chip->get
? (chip->get(chip, i) ? "hi" : "lo")
: "? ");
seq_printf(s, "\n");
}
}
static void *gpiolib_seq_start(struct seq_file *s, loff_t *pos)
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
{
unsigned long flags;
struct gpio_chip *chip = NULL;
loff_t index = *pos;
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
s->private = "";
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
spin_lock_irqsave(&gpio_lock, flags);
list_for_each_entry(chip, &gpio_chips, list)
if (index-- == 0) {
spin_unlock_irqrestore(&gpio_lock, flags);
return chip;
}
spin_unlock_irqrestore(&gpio_lock, flags);
return NULL;
}
static void *gpiolib_seq_next(struct seq_file *s, void *v, loff_t *pos)
{
unsigned long flags;
struct gpio_chip *chip = v;
void *ret = NULL;
spin_lock_irqsave(&gpio_lock, flags);
if (list_is_last(&chip->list, &gpio_chips))
ret = NULL;
else
ret = list_entry(chip->list.next, struct gpio_chip, list);
spin_unlock_irqrestore(&gpio_lock, flags);
s->private = "\n";
++*pos;
return ret;
}
static void gpiolib_seq_stop(struct seq_file *s, void *v)
{
}
static int gpiolib_seq_show(struct seq_file *s, void *v)
{
struct gpio_chip *chip = v;
struct device *dev;
seq_printf(s, "%sGPIOs %d-%d", (char *)s->private,
chip->base, chip->base + chip->ngpio - 1);
dev = chip->dev;
if (dev)
seq_printf(s, ", %s/%s", dev->bus ? dev->bus->name : "no-bus",
dev_name(dev));
if (chip->label)
seq_printf(s, ", %s", chip->label);
if (chip->can_sleep)
seq_printf(s, ", can sleep");
seq_printf(s, ":\n");
if (chip->dbg_show)
chip->dbg_show(s, chip);
else
gpiolib_dbg_show(s, chip);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
return 0;
}
static const struct seq_operations gpiolib_seq_ops = {
.start = gpiolib_seq_start,
.next = gpiolib_seq_next,
.stop = gpiolib_seq_stop,
.show = gpiolib_seq_show,
};
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
static int gpiolib_open(struct inode *inode, struct file *file)
{
return seq_open(file, &gpiolib_seq_ops);
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
}
static const struct file_operations gpiolib_operations = {
.owner = THIS_MODULE,
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
.open = gpiolib_open,
.read = seq_read,
.llseek = seq_lseek,
.release = seq_release,
gpiolib: add gpio provider infrastructure Provide new implementation infrastructure that platforms may choose to use when implementing the GPIO programming interface. Platforms can update their GPIO support to use this. In many cases the incremental cost to access a non-inlined GPIO should be less than a dozen instructions, with the memory cost being about a page (total) of extra data and code. The upside is: * Providing two features which were "want to have (but OK to defer)" when GPIO interfaces were first discussed in November 2006: - A "struct gpio_chip" to plug in GPIOs that aren't directly supported by SOC platforms, but come from FPGAs or other multifunction devices using conventional device registers (like UCB-1x00 or SM501 GPIOs, and southbridges in PCs with more open specs than usual). - Full support for message-based GPIO expanders, where registers are accessed through sleeping I/O calls. Previous support for these "cansleep" calls was just stubs. (One example: the widely used pcf8574 I2C chips, with 8 GPIOs each.) * Including a non-stub implementation of the gpio_{request,free}() calls, making those calls much more useful. The diagnostic labels are also recorded given DEBUG_FS, so /sys/kernel/debug/gpio can show a snapshot of all GPIOs known to this infrastructure. The driver programming interfaces introduced in 2.6.21 do not change at all; this infrastructure is entirely below those covers. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Eric Miao <eric.miao@marvell.com> Cc: Haavard Skinnemoen <hskinnemoen@atmel.com> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Ben Gardner <bgardner@wabtec.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 06:28:20 +00:00
};
static int __init gpiolib_debugfs_init(void)
{
/* /sys/kernel/debug/gpio */
(void) debugfs_create_file("gpio", S_IFREG | S_IRUGO,
NULL, NULL, &gpiolib_operations);
return 0;
}
subsys_initcall(gpiolib_debugfs_init);
#endif /* DEBUG_FS */