[PATCH] Input: adbhid - fix capslock key state after suspend

[PATCH] Input: adbhid - fix capslock key state after suspend

[Rodney Lorrimar]

[-- Attachment #1: Type: text/plain, Size: 3055 bytes --]

If the adbhid module parameter restore_capslock_events is used,
sometimes capslock will get stuck down after resuming.

My fix is to remember the capslock state before suspend and then
ignore the first 'caps lock key down' message after resume if the
capslock LED was on before suspending.

Signed-off-by: Rodney Lorrimar <rodney@rodney.id.au>
---
 drivers/macintosh/adbhid.c |   43 +++++++++++++++++++++++++++++++++++++------
 1 files changed, 37 insertions(+), 6 deletions(-)

diff --git a/drivers/macintosh/adbhid.c b/drivers/macintosh/adbhid.c
index ef4c117..19d0984 100644
--- a/drivers/macintosh/adbhid.c
+++ b/drivers/macintosh/adbhid.c
@@ -219,11 +219,12 @@ struct adbhid {
 	int flags;
 };
 
-#define FLAG_FN_KEY_PRESSED	0x00000001
-#define FLAG_POWER_FROM_FN	0x00000002
-#define FLAG_EMU_FWDEL_DOWN	0x00000004
-#define FLAG_CAPSLOCK_TRANSLATE	0x00000008
-#define FLAG_CAPSLOCK_DOWN	0x00000010
+#define FLAG_FN_KEY_PRESSED		0x00000001
+#define FLAG_POWER_FROM_FN		0x00000002
+#define FLAG_EMU_FWDEL_DOWN		0x00000004
+#define FLAG_CAPSLOCK_TRANSLATE		0x00000008
+#define FLAG_CAPSLOCK_DOWN		0x00000010
+#define FLAG_CAPSLOCK_IGNORE_NEXT	0x00000020
 
 static struct adbhid *adbhid[16];
 
@@ -291,8 +292,15 @@ adbhid_input_keycode(int id, int scancode, int repeat)
 		if (keycode == ADB_KEY_CAPSLOCK && !up_flag) {
 			/* Key pressed, turning on the CapsLock LED.
 			 * The next 0xff will be interpreted as a release. */
-			ahid->flags |= FLAG_CAPSLOCK_TRANSLATE
+			if (ahid->flags & FLAG_CAPSLOCK_IGNORE_NEXT) {
+				/* Throw away this key event if it happens
+				 * just after resume. */
+				ahid->flags &= ~FLAG_CAPSLOCK_IGNORE_NEXT;
+				return;
+			} else {
+				ahid->flags |= FLAG_CAPSLOCK_TRANSLATE
 					| FLAG_CAPSLOCK_DOWN;
+			}
 		} else if (scancode == 0xff) {
 			/* Scancode 0xff usually signifies that the capslock
 			 * key was either pressed or released. */
@@ -681,6 +689,20 @@ static int adbhid_kbd_event(struct input_dev *dev, unsigned int type, unsigned i
 	return -1;
 }
 
+static void
+adbhid_kbd_capslock_remember(void)
+{
+	struct adbhid *ahid;
+	int i;
+	for (i = 1; i < 16; i++) {
+		ahid = adbhid[i];
+
+		if (ahid && ahid->id == ADB_KEYBOARD)
+			if (ahid->flags & FLAG_CAPSLOCK_TRANSLATE)
+				ahid->flags |= FLAG_CAPSLOCK_IGNORE_NEXT;
+	}
+}
+
 static int
 adb_message_handler(struct notifier_block *this, unsigned long code, void *x)
 {
@@ -699,6 +721,15 @@ adb_message_handler(struct notifier_block *this, unsigned long code, void *x)
 		/* Stop pending led requests */
 		while(leds_req_pending)
 			adb_poll();
+
+		/* After resume, and if the capslock LED is on, the PMU will
+		 * send a "capslock down" key event. This confuses the
+		 * restore_capslock_events logic. Remember if the capslock
+		 * LED was on before suspend so the unwanted key event can
+		 * be ignored after resume. */
+		if (restore_capslock_events)
+			adbhid_kbd_capslock_remember();
+
 		break;
 
 	case ADB_MSG_POST_RESET:
-- 
1.5.4.4


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH] Input: adbhid - fix capslock key state after suspend

[Pavel Machek]

On Fri 2008-03-28 01:15:56, Rodney Lorrimar wrote:
> If the adbhid module parameter restore_capslock_events is used,
> sometimes capslock will get stuck down after resuming.

Hmmm, why does that need to be a parameter? I mean... can we just do
the right thing?

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
pomozte zachranit klanovicky les:  http://www.ujezdskystrom.info/

Re: [PATCH] Input: adbhid - fix capslock key state after suspend

[Andrew McNabb]

[-- Attachment #1: Type: text/plain, Size: 654 bytes --]

On Fri, Mar 28, 2008 at 11:58:55AM +0100, Pavel Machek wrote:
> On Fri 2008-03-28 01:15:56, Rodney Lorrimar wrote:
> > If the adbhid module parameter restore_capslock_events is used,
> > sometimes capslock will get stuck down after resuming.
> 
> Hmmm, why does that need to be a parameter? I mean... can we just do
> the right thing?

I agree.  There really don't seem to be any drawbacks to keeping track
of capslock state.  The worst thing that can happen is that it somehow
occasionally fails to do the right thing.

-- 
Andrew McNabb
http://www.mcnabbs.org/andrew/
PGP Fingerprint: 8A17 B57C 6879 1863 DE55  8012 AB4D 6098 8826 6868

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH] Input: adbhid - fix capslock key state after suspend

[Rodney Lorrimar]

[-- Attachment #1: Type: text/plain, Size: 1186 bytes --]

On Fri, Mar 28, 2008 at 11:58:55AM +0100, Pavel Machek wrote:
> On Fri 2008-03-28 01:15:56, Rodney Lorrimar wrote:
> > If the adbhid module parameter restore_capslock_events is used,
> > sometimes capslock will get stuck down after resuming.
> 
> Hmmm, why does that need to be a parameter? I mean... can we just do
> the right thing?

Hi Pavel,

Sorry for taking so long to answer your questions. We can mostly do
the right thing. I suppose it was a parameter so the quirk fixing
didn't ruin the fun for the majority of users who don't remap their
capslock. I know of two issues with the code:

1. Under heavy capslock usage the capslock key will sometimes get
   stuck down.
2. If the capslock LED is on and the power button is pressed, a "power
   button release" key event won't happen until capslock is pressed
   again. So in some situations (e.g. with pbbuttonsd) the next time
   capslock is hit the machine will suspend.


I will send a patch for 2, and continue to investigate 1.

Regarding the original patch, I think it's worth applying. If 1 is
fixable then I will prepare two more patches to fix problem 1 and
remove the parameter.


Rodney

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

[PATCH] Input: adbhid - capslock and power button fix

[Rodney Lorrimar]

[-- Attachment #1: Type: text/plain, Size: 2308 bytes --]

If the adbhid module parameter restore_capslock_events is used,
pressing the power button may confuse the capslock state. This is
because the power button release scancode (0xff) is sometimes the same
as the capslock press/release scancode.

This fix adds yet another flag to track the state of the power button
so that it works independently of capslock.

Signed-off-by: Rodney Lorrimar <rodney@rodney.id.au>
---

This patch applies on top of the previous patch in the thread.

---
 drivers/macintosh/adbhid.c |   15 ++++++++++++---
 1 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/drivers/macintosh/adbhid.c b/drivers/macintosh/adbhid.c
index 19d0984..72b9011 100644
--- a/drivers/macintosh/adbhid.c
+++ b/drivers/macintosh/adbhid.c
@@ -225,6 +225,7 @@ struct adbhid {
 #define FLAG_CAPSLOCK_TRANSLATE		0x00000008
 #define FLAG_CAPSLOCK_DOWN		0x00000010
 #define FLAG_CAPSLOCK_IGNORE_NEXT	0x00000020
+#define FLAG_POWER_KEY_PRESSED		0x00000040
 
 static struct adbhid *adbhid[16];
 
@@ -301,9 +302,11 @@ adbhid_input_keycode(int id, int scancode, int repeat)
 				ahid->flags |= FLAG_CAPSLOCK_TRANSLATE
 					| FLAG_CAPSLOCK_DOWN;
 			}
-		} else if (scancode == 0xff) {
+		} else if (scancode == 0xff &&
+			   !(ahid->flags & FLAG_POWER_KEY_PRESSED)) {
 			/* Scancode 0xff usually signifies that the capslock
-			 * key was either pressed or released. */
+			 * key was either pressed or released, or that the
+			 * power button was released. */
 			if (ahid->flags & FLAG_CAPSLOCK_TRANSLATE) {
 				keycode = ADB_KEY_CAPSLOCK;
 				if (ahid->flags & FLAG_CAPSLOCK_DOWN) {
@@ -317,7 +320,7 @@ adbhid_input_keycode(int id, int scancode, int repeat)
 				}
 			} else {
 				printk(KERN_INFO "Spurious caps lock event "
-						"(scancode 0xff).");
+				                 "(scancode 0xff).\n");
 			}
 		}
 	}
@@ -344,6 +347,12 @@ adbhid_input_keycode(int id, int scancode, int repeat)
 		}
 		break;
 	case ADB_KEY_POWER:
+		/* Keep track of the power key state */
+		if (up_flag)
+			ahid->flags &= ~FLAG_POWER_KEY_PRESSED;
+		else
+			ahid->flags |= FLAG_POWER_KEY_PRESSED;
+
 		/* Fn + Command will produce a bogus "power" keycode */
 		if (ahid->flags & FLAG_FN_KEY_PRESSED) {
 			keycode = ADB_KEY_CMD;
-- 
1.5.4.5


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]