[RFC PATCH] vt: Fix deadlock on scroll-lock

[Alan Cox]

From: Alan Cox <alan@xxxxxxxxxxxxxxx>

Fixing the locking accidentally replaced a race in the scroll
lock handling with a deadlock. Turn it back into a race for
now.

The basic problem is that there are two paths into the tty
stop/start helpers. One via the tty layer ^S/^Q handling
where we need to take the kbd_event_lock and one via the
special keyboard handler for fn_hold where we already hold
it. Probably we need to split out into a separate LED lock
but for now just go back to the race as it's a bit close
to release.

Signed-off-by: Alan Cox <alan@xxxxxxxxxxxxxxx>
---

 drivers/tty/vt/keyboard.c |   22 +++++++++++++++++-----
 1 files changed, 17 insertions(+), 5 deletions(-)


diff --git a/drivers/tty/vt/keyboard.c b/drivers/tty/vt/keyboard.c
index 86dd1e3..6d54a6a 100644
--- a/drivers/tty/vt/keyboard.c
+++ b/drivers/tty/vt/keyboard.c
@@ -1085,15 +1085,21 @@ void vt_set_led_state(int console, int leds)
  *
  *	Handle console start. This is a wrapper for the VT layer
  *	so that we can keep kbd knowledge internal
+ *
+ *	FIXME: We eventually need to hold the kbd lock here to protect
+ *	the LED updating. We can't do it yet because fn_hold calls stop_tty
+ *	and start_tty under the kbd_event_lock, while normal tty paths
+ *	don't hold the lock. We probably need to split out an LED lock
+ *	but not during an -rc release!
  */
 void vt_kbd_con_start(int console)
 {
 	struct kbd_struct * kbd = kbd_table + console;
 	unsigned long flags;
-	spin_lock_irqsave(&kbd_event_lock, flags);
+/*	spin_lock_irqsave(&kbd_event_lock, flags); */
 	clr_vc_kbd_led(kbd, VC_SCROLLOCK);
 	set_leds();
-	spin_unlock_irqrestore(&kbd_event_lock, flags);
+/*	spin_unlock_irqrestore(&kbd_event_lock, flags); */
 }
 
 /**
@@ -1102,22 +1108,28 @@ void vt_kbd_con_start(int console)
  *
  *	Handle console stop. This is a wrapper for the VT layer
  *	so that we can keep kbd knowledge internal
+ *
+ *	FIXME: We eventually need to hold the kbd lock here to protect
+ *	the LED updating. We can't do it yet because fn_hold calls stop_tty
+ *	and start_tty under the kbd_event_lock, while normal tty paths
+ *	don't hold the lock. We probably need to split out an LED lock
+ *	but not during an -rc release!
  */
 void vt_kbd_con_stop(int console)
 {
 	struct kbd_struct * kbd = kbd_table + console;
 	unsigned long flags;
-	spin_lock_irqsave(&kbd_event_lock, flags);
+/*	spin_lock_irqsave(&kbd_event_lock, flags); */
 	set_vc_kbd_led(kbd, VC_SCROLLOCK);
 	set_leds();
-	spin_unlock_irqrestore(&kbd_event_lock, flags);
+/*	spin_unlock_irqrestore(&kbd_event_lock, flags); */
 }
 
 /*
  * This is the tasklet that updates LED state on all keyboards
  * attached to the box. The reason we use tasklet is that we
  * need to handle the scenario when keyboard handler is not
- * registered yet but we already getting updates form VT to
+ * registered yet but we already getting updates from the VT to
  * update led state.
  */
 static void kbd_bh(unsigned long dummy)

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/