Re: USB: serial: kfifo_len locking

[Stefani Seibold <stefani@seibold.net>]

Am Dienstag, den 05.01.2010, 00:43 -0700 schrieb Pete Zaitcev:
> On Mon, 04 Jan 2010 20:20:04 +0100
> Stefani Seibold <stefani@seibold.net> wrote:
> > Am Montag, den 04.01.2010, 18:43 +0100 schrieb Johan Hovold:
> 
> > > I noticed that the locking that used to protect kfifo_len in
> > > usb_serial_generic_chars_in_buffer was removed when the kifo api changed
> > > to not use internal locking (c1e13f25674ed564948ecb7dfe5f83e578892896 -- 
> > > kfifo: move out spinlock). 
> > > 
> > > Was this intentional?
> > > 
> > 
> > Yes, the locking is not necessary until only one reader and one writer
> > is using the fifo. If you don't trust this you can apply this patch:
> > 
> > --- linux-2.6.33-rc2.orig/drivers/usb/serial/generic.c	2009-12-27 23:37:03.566060210 +0100
> > +++ linux-2.6.33-rc2.new/drivers/usb/serial/generic.c	2010-01-04 20:15:38.023351711 +0100
> > @@ -386,12 +386,12 @@ int usb_serial_generic_chars_in_buffer(s
> >  
> >  	dbg("%s - port %d", __func__, port->number);
> >  
> > -	if (serial->type->max_in_flight_urbs) {
> > -		spin_lock_irqsave(&port->lock, flags);
> > +	spin_lock_irqsave(&port->lock, flags);
> > +	if (serial->type->max_in_flight_urbs)
> >  		chars = port->tx_bytes_flight;
> > -		spin_unlock_irqrestore(&port->lock, flags);
> > -	} else if (serial->num_bulk_out)
> > +	else if (serial->num_bulk_out)
> >  		chars = kfifo_len(&port->write_fifo);
> > +	spin_unlock_irqrestore(&port->lock, flags);
> > 
> > Then everything kfifo_... access in the usb serial is handled with an
> > active spinlock.
> 
> This actually was a side effect of the "byte lost on close" patch
> that I submitted, it should be in Greg's tree. The relevant part goes
> like this:
> 
> diff --git a/drivers/usb/serial/generic.c b/drivers/usb/serial/generic.c
> index f1ea3a3..3372faa 100644
> --- a/drivers/usb/serial/generic.c
> +++ b/drivers/usb/serial/generic.c
> @@ -386,12 +386,15 @@ int usb_serial_generic_chars_in_buffer(struct tty_struct *tty)
>  
>  	dbg("%s - port %d", __func__, port->number);
>  
> +	spin_lock_irqsave(&port->lock, flags);
>  	if (serial->type->max_in_flight_urbs) {
> -		spin_lock_irqsave(&port->lock, flags);
>  		chars = port->tx_bytes_flight;
> -		spin_unlock_irqrestore(&port->lock, flags);
> -	} else if (serial->num_bulk_out)
> -		chars = kfifo_len(&port->write_fifo);
> +	} else if (serial->num_bulk_out) {
> +		/* This overcounts badly, but is good enough for drain wait. */
> +		chars = __kfifo_len(port->write_fifo);
> +		chars += port->write_urb_busy * port->bulk_out_size;
> +	}
> +	spin_unlock_irqrestore(&port->lock, flags);
>  
>  	dbg("%s - returns %d", __func__, chars);
>  	return chars;

This is the same patch as my. But __kfifo_len is renamed into kfifo_len.
Who should submit this patch? 

Stefani