LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* [patch 2.6.25-rc1 1/3] gpiolib: better rmmod infrastructure
@ 2008-02-11  0:22 David Brownell
  0 siblings, 0 replies; only message in thread
From: David Brownell @ 2008-02-11  0:22 UTC (permalink / raw)
  To: Andrew Morton; +Cc: lkml, Guennadi Liakhovetski

From: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de>

As long as one or more GPIOs on a gpio chip are used its driver should not 
be unloaded.  The existing mechanism (gpiochip_remove failure) doesn't
address that, since rmmod can no longer be made to fail by having the
cleanup code report errors.  Module usecounts are the solution.

Assuming standard "initialize struct to zero" policies, this change won't
affect SOC platform drivers.  However, drivers for external chips (on I2C
and SPI busses) should be updated if they can be built as modules.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@pengutronix.de>
[ gpio_ensure_requested() needs to update module usecounts too ]
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
---
 drivers/gpio/gpiolib.c     |   15 ++++++++++++---
 include/asm-generic/gpio.h |    2 ++
 2 files changed, 14 insertions(+), 3 deletions(-)

--- g26.orig/drivers/gpio/gpiolib.c	2008-02-10 16:06:30.000000000 -0800
+++ g26/drivers/gpio/gpiolib.c	2008-02-10 16:18:35.000000000 -0800
@@ -68,6 +68,9 @@ static void gpio_ensure_requested(struct
 	if (test_and_set_bit(FLAG_REQUESTED, &desc->flags) == 0) {
 		pr_warning("GPIO-%d autorequested\n", (int)(desc - gpio_desc));
 		desc_set_label(desc, "[auto]");
+		if (!try_module_get(desc->chip->owner))
+			pr_err("GPIO-%d: module can't be gotten \n",
+					(int)(desc - gpio_desc));
 	}
 }
 
@@ -177,6 +180,9 @@ int gpio_request(unsigned gpio, const ch
 	if (desc->chip == NULL)
 		goto done;
 
+	if (!try_module_get(desc->chip->owner))
+		goto done;
+
 	/* NOTE:  gpio_request() can be called in early boot,
 	 * before IRQs are enabled.
 	 */
@@ -184,8 +190,10 @@ int gpio_request(unsigned gpio, const ch
 	if (test_and_set_bit(FLAG_REQUESTED, &desc->flags) == 0) {
 		desc_set_label(desc, label ? : "?");
 		status = 0;
-	} else
+	} else {
 		status = -EBUSY;
+		module_put(desc->chip->owner);
+	}
 
 done:
 	if (status)
@@ -209,9 +217,10 @@ void gpio_free(unsigned gpio)
 	spin_lock_irqsave(&gpio_lock, flags);
 
 	desc = &gpio_desc[gpio];
-	if (desc->chip && test_and_clear_bit(FLAG_REQUESTED, &desc->flags))
+	if (desc->chip && test_and_clear_bit(FLAG_REQUESTED, &desc->flags)) {
 		desc_set_label(desc, NULL);
-	else
+		module_put(desc->chip->owner);
+	} else
 		WARN_ON(extra_checks);
 
 	spin_unlock_irqrestore(&gpio_lock, flags);
--- g26.orig/include/asm-generic/gpio.h	2008-02-10 16:06:30.000000000 -0800
+++ g26/include/asm-generic/gpio.h	2008-02-10 16:18:30.000000000 -0800
@@ -17,6 +17,7 @@
 #endif
 
 struct seq_file;
+struct module;
 
 /**
  * struct gpio_chip - abstract a GPIO controller
@@ -48,6 +49,7 @@ struct seq_file;
  */
 struct gpio_chip {
 	char			*label;
+	struct module		*owner;
 
 	int			(*direction_input)(struct gpio_chip *chip,
 						unsigned offset);

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2008-02-11  0:23 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-02-11  0:22 [patch 2.6.25-rc1 1/3] gpiolib: better rmmod infrastructure David Brownell

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).