Skip to contentSkip to conversation

PM: /sys/devices/.../power/state updates

PATCH In review David Brownell 2 participants 2 messages Jul 19, 2006 – Aug 16, 2006

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 2/2 PM: /sys/devices/.../power/state updates

Two changes here: (a) make writes to these files fail on devices that may need the new operations in Linus' patch; (b) update in-source docs to match current behavior, including that change.

I hope nobody has issues with (a), but I can't see a good reason to do more than reject such writes since the notions behind that file are so broken.

- Dave

p.s. Greg, these two should IMO merge into your tree along with
Linus' patches, in addition to

http://lists.osdl.org/pipermail/linux-pm/2006-July/002968.html

Two updates to the /sys/devices/.../power/state code:

(a) Changes write path code so that it returns errors for devices using the new not-callable-here suspend_prepare(), suspend_late(), or resume_early() methods.

(a) Corrects some of the documentation about what's legal for writing to /sys/devices/.../power/state files to match the current code (including the minor changes for those new methods) below that documentation.

drivers/base/power/sysfs.c +21 -10
Index: linux/drivers/base/power/sysfs.c
===================================================================
@@ -11,18 +11,23 @@
  *	state - Control current power state of device
  *
  *	show() returns the current power state of the device. '0' indicates
- *	the device is on. Other values (1-3) indicate the device is in a low
+ *	the device is on. Other values (2) indicate the device is in some low
  *	power state.
  *
- *	store() sets the current power state, which is an integer value
- *	between 0-3. If the device is on ('0'), and the value written is
- *	greater than 0, then the device is placed directly into the low-power
- *	state (via its driver's ->suspend() method).
- *	If the device is currently in a low-power state, and the value is 0,
- *	the device is powered back on (via the ->resume() method).
- *	If the device is in a low-power state, and a different low-power state
- *	is requested, the device is first resumed, then suspended into the new
- *	low-power state.
+ *	store() sets the current power state, which is an integer valued
+ *	0, 2, or 3.  Devices with bus.suspend_prepare(), bus.suspend_late(),
+ *	or bus.resume_early() methods fail this operation; those methods
+ *	won't be called.  Otherwise,
+ *
+ *	- If the recorded dev->power.power_state.event matches the
+ *	  target value (PM_EVENT_ON or PM_EVENT_SUSPEND), nothing is done.
+ *	- If the recorded event code is nonzero, the device is reactivated
+ *	  by calling bus.resume() and/or class.resume().
+ *	- If the target value is nonzero, the device is suspended by
+ *	  calling class.suspend() and/or bus.suspend() with event code
+ *	  PM_EVENT_SUSPEND.
+ *
+ *	This mechanism is DEPRECATED and should only be used for testing.
  */
 
 static ssize_t state_show(struct device * dev, struct device_attribute *attr, char * buf)
@@ -38,6 +43,12 @@
 	pm_message_t state;
 	int error = -EINVAL;
 
+	/* disallow incomplete suspend sequences */
+	if (dev->bus && (dev->bus->suspend_prepare
+			|| dev->bus->suspend_late
+			|| dev->bus->resume_early))
+		return error;
+
 	state.event = PM_EVENT_SUSPEND;
 	/* Older apps expected to write "3" here - confused with PCI D3 */
 	if ((n == 1) && !strcmp(buf, "3"))
#1 lore raw
Quoted 7 lines · David Brownell

Two changes here: (a) make writes to these files fail on devices that may need the new operations in Linus' patch; (b) update in-source docs to match current behavior, including that change.

I hope nobody has issues with (a), but I can't see a good reason to do more than reject such writes since the notions behind that file are so broken.

Looks okay.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html