Re: BKL still required for what functions?

[Richard Holden]

On Oct 7, 2008, at 10:35 AM, Alan Cox wrote:
> Its never so simple - a lot of the functions it depends what data  
> you are
> using or passing to them whether they are BKL safe.
>
> Which bits use the BKL and what do you need to know is safe ?

Actually, it looks like I was looking at 2 different places, the  
simple removal is in phonedev.c that you authored, as far as I can  
tell the only thing the BKL is protecting is a call to request_module,  
it may be some nesting rules that I don't understand but I think if it  
is still needed there we could push the lock_kernel call down to the  
site of the request_module call. See patch below.

Thanks,
Richard Holden

Signed-off-by: Richard Holden <aciddeath@xxxxxxxxx>

diff --git a/drivers/telephony/phonedev.c b/drivers/telephony/phonedev.c
index 4d74ba3..fd1424c 100644
--- a/drivers/telephony/phonedev.c
+++ b/drivers/telephony/phonedev.c
@@ -54,14 +54,16 @@ static int phone_open(struct inode *inode, struct  
file *file)
 	if (minor >= PHONE_NUM_DEVICES)
 		return -ENODEV;

-	lock_kernel();
+	
 	mutex_lock(&phone_lock);
 	p = phone_device[minor];
 	if (p)
 		new_fops = fops_get(p->f_op);
 	if (!new_fops) {
 		mutex_unlock(&phone_lock);
+		lock_kernel();
 		request_module("char-major-%d-%d", PHONE_MAJOR, minor);
+		unlock_kernel();
 		mutex_lock(&phone_lock);
 		p = phone_device[minor];
 		if (p == NULL || (new_fops = fops_get(p->f_op)) == NULL)
@@ -81,7 +83,6 @@ static int phone_open(struct inode *inode, struct  
file *file)
 	fops_put(old_fops);
 end:
 	mutex_unlock(&phone_lock);
-	unlock_kernel();
 	return err;
 }


--
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/