Re: [PATCH] char: lp: Fix NULL pointer dereference of cad

[Li Nan]

在 2025/12/30 10:10, Al Viro 写道:
> On Tue, Dec 30, 2025 at 09:51:43AM +0800, Li Nan wrote:
> > Friendly ping...
>
> > > @@ -569,10 +579,13 @@ static int lp_release(struct inode *inode, struct file *file)
> > >    {
> > >    	unsigned int minor = iminor(inode);
> > > +	if (mutex_lock_interruptible(&lp_table[minor].port_mutex))
> > > +		return -EINTR;
>
> ->release() return value is never checked, simply because there is nothing
> to do with it.  It will *not* leave file opened - it will simply leak,
> with no way to recover from that.
>
> If you need to report some errors on close, do that in ->flush().
> If you ever see ->release() returning a non-zero value, you are very
> likely looking at deeply confused code.
>
> Don't do that.  ->release() can't fail, period.  It should've been
> void (*release)(struct file *), but for historical reasons it returns
> int and there are too many instances to change that.

Thank you for your patient explanation.

Would it be acceptable to switch to mutex_lock() here? Looking at the code
and historical changes, I don't see a compelling reason for the interruptible
function here.

--
Thanks,
Nan