Thread (58 messages) 58 messages, 5 authors, 2008-09-22

Re: [PATCH] rfkill: clarify usage of rfkill_force_state() and rfkill->get_state()

From: Ivo van Doorn <hidden>
Date: 2008-09-18 15:24:21
Also in: lkml

On Thursday 18 September 2008, Henrique de Moraes Holschuh wrote:
rfkill_force_state() is used to update the rfkill core's idea of what is
really happening RIGHT NOW to the transmitter hardware in a PUSH model.

rfkill->get_state() does the same, in a PULL model.

Neither of them change the real hardware radio state through a call to
rfkill->toggle_radio() or anything of the sort, so they must deal with the
real, current state of the hardware.

Change some documentation to make that more clear (I hope).

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Ivo van Doorn <redacted>
See my other mail I just send out.
But I don't quite agree on this change of rfkill event interpretation.

Ivo
quoted hunk ↗ jump to hunk
Cc: Larry Finger <redacted>
---
 Documentation/rfkill.txt |   10 ++++++++++
 include/linux/rfkill.h   |    3 +++
 net/rfkill/rfkill.c      |    3 +++
 3 files changed, 16 insertions(+), 0 deletions(-)
diff --git a/Documentation/rfkill.txt b/Documentation/rfkill.txt
index b65f079..0143f1e 100644
--- a/Documentation/rfkill.txt
+++ b/Documentation/rfkill.txt
@@ -461,6 +461,16 @@ mandatory when the device has a hardware rfkill line, or when something else
 like the firmware could cause its state to be changed without going through the
 rfkill class.
 
+**** ATTENTION ****
+Both the state returned by the rfkill->get_state() hook, and the state
+passed to rfkill_force_state() MUST BE THE REAL, CURRENT STATE OF THE
+HARDWARE TRANSMITTER.  Never use for these functions the "desired" state,
+"user requested" state, or anything of the sort.
+
+Also, be warned that rfkill_force_state() cannot be called from atomic context,
+so interrupt handlers will have to make use of a workqueue to do it.
+**** ATTENTION ****
+
 Some hardware provides events when its status changes.  In these cases, it is
 best for the driver to not provide a get_state hook, and instead register the
 rfkill class *already* with the correct status, and keep it updated using
diff --git a/include/linux/rfkill.h b/include/linux/rfkill.h
index 4cd64b0..92b9c72 100644
--- a/include/linux/rfkill.h
+++ b/include/linux/rfkill.h
@@ -80,6 +80,9 @@ enum rfkill_state {
  *      may be called from atomic context, should return 0 on success.
  *      Either this handler OR judicious use of rfkill_force_state() is
  *      MANDATORY for any driver capable of RFKILL_STATE_HARD_BLOCKED.
+ *      THIS HOOK MUST RETURN THE CURRENT, REAL STATE OF THE HARDWARE
+ *      TRANSMITTER, and not the "desired state", "user requested state"
+ *      or anything of the sort.
  * @led_trigger: A LED trigger for this button's LED.
  * @dev: Device structure integrating the switch into device tree.
  * @node: Used to place switch into list of all switches known to the
diff --git a/net/rfkill/rfkill.c b/net/rfkill/rfkill.c
index f949a48..64a53f1 100644
--- a/net/rfkill/rfkill.c
+++ b/net/rfkill/rfkill.c
@@ -333,6 +333,9 @@ EXPORT_SYMBOL_GPL(rfkill_restore_states);
  * a notification by the firmware/hardware of the current *real*
  * state of the radio rfkill switch.
  *
+ * The state passed to this function MUST MATCH THE CURRENT,
+ * REAL HARDWARE STATE OF THE TRANSMITTER.
+ *
  * Devices which are subject to external changes on their rfkill
  * state (such as those caused by a hardware rfkill line) MUST
  * have their driver arrange to call rfkill_force_state() as soon
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help