[PATCH 5/9] UML - Add locking to xtime accesses

[Jeff Dike]

do_timer must be called with xtime_lock held. I'm not sure
boot_timer_handler needs this, however I don't think it hurts: it simply
disables irq and takes a spinlock.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@xxxxxxxx>

Index: linux-2.6.17/arch/um/kernel/time_kern.c
===================================================================
--- linux-2.6.17.orig/arch/um/kernel/time_kern.c	2006-06-29 10:37:16.000000000 -0400
+++ linux-2.6.17/arch/um/kernel/time_kern.c	2006-06-29 11:39:57.000000000 -0400
@@ -96,11 +96,15 @@ void time_init_kern(void)
 
 void do_boot_timer_handler(struct sigcontext * sc)
 {
+	unsigned long flags;
 	struct pt_regs regs;
 
 	CHOOSE_MODE((void) (UPT_SC(&regs.regs) = sc),
 		    (void) (regs.regs.skas.is_user = 0));
+
+	write_seqlock_irqsave(&xtime_lock, flags);
 	do_timer(&regs);
+	write_sequnlock_irqrestore(&xtime_lock, flags);
 }
 
 static DEFINE_SPINLOCK(timer_spinlock);
@@ -125,15 +129,17 @@ irqreturn_t um_timer(int irq, void *dev,
 	unsigned long long nsecs;
 	unsigned long flags;
 
+	write_seqlock_irqsave(&xtime_lock, flags);
+
 	do_timer(regs);
 
-	write_seqlock_irqsave(&xtime_lock, flags);
 	nsecs = get_time() + local_offset;
 	xtime.tv_sec = nsecs / NSEC_PER_SEC;
 	xtime.tv_nsec = nsecs - xtime.tv_sec * NSEC_PER_SEC;
+
 	write_sequnlock_irqrestore(&xtime_lock, flags);
 
-	return(IRQ_HANDLED);
+	return IRQ_HANDLED;
 }
 
 long um_time(int __user *tloc)

-
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/