Re: [PATCH v2] usb: gadget: u_serial: Add null pointer check in gserial_resume

[Alan Stern <stern@rowland.harvard.edu>]

On Sat, Feb 11, 2023 at 05:47:38PM +0530, Prashanth K wrote:
> Consider a case where gserial_disconnect has already cleared
> gser->ioport. And if a wakeup interrupt triggers afterwards,
> gserial_resume gets called, which will lead to accessing of
> gser->ioport and thus causing null pointer dereference.Add
> a null pointer check to prevent this.
> 
> Added a static spinlock to prevent gser->ioport from becoming
> null after the newly added check.
> 
> Fixes: aba3a8d01d62 ("usb: gadget: u_serial: add suspend resume callbacks")
> Signed-off-by: Prashanth K <quic_prashk@quicinc.com>
> ---
> v2: Added static spinlock and fixed Fixes tag.
> 
>  drivers/usb/gadget/function/u_serial.c | 16 ++++++++++++++--
>  1 file changed, 14 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/usb/gadget/function/u_serial.c b/drivers/usb/gadget/function/u_serial.c
> index 840626e..9ced0fa 100644
> --- a/drivers/usb/gadget/function/u_serial.c
> +++ b/drivers/usb/gadget/function/u_serial.c
> @@ -82,6 +82,8 @@
>  #define WRITE_BUF_SIZE		8192		/* TX only */
>  #define GS_CONSOLE_BUF_SIZE	8192
>  
> +static DEFINE_SPINLOCK(serial_port_lock);
> +
>  /* console info */
>  struct gs_console {
>  	struct console		console;
> @@ -1370,11 +1372,13 @@ EXPORT_SYMBOL_GPL(gserial_connect);
>  void gserial_disconnect(struct gserial *gser)
>  {
>  	struct gs_port	*port = gser->ioport;
> -	unsigned long	flags;
> +	unsigned long flags, serial_flag;

You don't need two separate flags here.  The fact that you wrote this 
indicates you don't understand how spin_lock_irqsave() and 
spin_lock_irqrestore() work.

When spin_lock_irqsave(&lock, flag) is called, it saves the current INT 
(interrupt-enable) setting in flag, disables interrupts, and acquires 
the lock.  When spin_unlock_irqrestore(&lock, flag) is called, it 
releases the lock and writes the value in flag back to the INT setting.

The end result is that if interrupts were enabled before 
spin_lock_irqsave() then they will be enabled after 
spin_unlock_irqrestore().  If interrupts were disabled beforehand, they 
will remain disabled afterward.  And either way, interrupts will be 
disabled between the two calls.

>  
>  	if (!port)
>  		return;
>  
> +	spin_lock_irqsave(&serial_port_lock, serial_flag);

So now interrupts are disabled.

> +
>  	/* tell the TTY glue not to do I/O here any more */
>  	spin_lock_irqsave(&port->port_lock, flags);

Hence there's no need for flag here.  You don't need to save the current 
INT setting because you already know what it is: interrupts are 
disabled.  You can simply call spin_lock(), which will acquire the lock 
without doing anything to the INT setting.

>  
> @@ -1392,6 +1396,7 @@ void gserial_disconnect(struct gserial *gser)
>  	}
>  	port->suspended = false;
>  	spin_unlock_irqrestore(&port->port_lock, flags);

Likewise, here you can call spin_unlock().

> +	spin_unlock_irqrestore(&serial_port_lock, serial_flag);
>  
>  	/* disable endpoints, aborting down any active I/O */
>  	usb_ep_disable(gser->out);
> @@ -1426,9 +1431,16 @@ EXPORT_SYMBOL_GPL(gserial_suspend);
>  void gserial_resume(struct gserial *gser)
>  {
>  	struct gs_port *port = gser->ioport;
> -	unsigned long	flags;
> +	unsigned long flags, serial_flag;
> +
> +	spin_lock_irqsave(&serial_port_lock, serial_flag);
> +	if (!port) {
> +		spin_unlock_irqrestore(&serial_port_lock, serial_flag);
> +		return;
> +	}

This is a little trickier, but the same principles apply.  Since 
spin_lock_irqsave() was called above, interrupts are now disabled.

>  
>  	spin_lock_irqsave(&port->port_lock, flags);

So there's no need for _irqsave here.

> +	spin_unlock_irqrestore(&serial_port_lock, serial_flag);

And here you must not use spin_unlock_irqrestore().  This will do the 
wrong thing, because it will enable interrupts if they were enabled at 
the start of the function.  Then you would be running with 
port->port_lock held and interrupts enabled, a bad combination.

>  	port->suspended = false;
>  	if (!port->start_delayed) {
>  		spin_unlock_irqrestore(&port->port_lock, flags);

Here, at the final unlock, is where you should restore the INT setting 
to the value it had at the start of the function.

Alan Stern