linux-kernel - Re: [PATCH 1/6] Input: db9 - use guard notation when acquiring mutex

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <Zz6ujWxy5Gyhx7oj@google.com>
Date: Thu, 21 Nov 2024 03:52:45 +0000
From: Dmitry Torokhov <dmitry.torokhov@...il.com>
To: David Lechner <dlechner@...libre.com>
Cc: linux-input@...r.kernel.org, Erick Archer <erick.archer@...look.com>,
	Christophe JAILLET <christophe.jaillet@...adoo.fr>,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/6] Input: db9 - use guard notation when acquiring mutex

On Wed, Nov 20, 2024 at 12:33:36PM -0600, David Lechner wrote:
> On 9/3/24 11:30 PM, Dmitry Torokhov wrote:
> > Using guard notation makes the code more compact and error handling
> > more robust by ensuring that mutexes are released in all code paths
> > when control leaves critical section.
> > 
> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@...il.com>
> > ---
> >  drivers/input/joystick/db9.c | 30 ++++++++++++++----------------
> >  1 file changed, 14 insertions(+), 16 deletions(-)
> > 
> > diff --git a/drivers/input/joystick/db9.c b/drivers/input/joystick/db9.c
> > index 682a29c27832..7ac0cfc3e786 100644
> > --- a/drivers/input/joystick/db9.c
> > +++ b/drivers/input/joystick/db9.c
> > @@ -505,24 +505,22 @@ static int db9_open(struct input_dev *dev)
> >  {
> >  	struct db9 *db9 = input_get_drvdata(dev);
> >  	struct parport *port = db9->pd->port;
> > -	int err;
> >  
> > -	err = mutex_lock_interruptible(&db9->mutex);
> > -	if (err)
> > -		return err;
> > -
> > -	if (!db9->used++) {
> > -		parport_claim(db9->pd);
> > -		parport_write_data(port, 0xff);
> > -		if (db9_modes[db9->mode].reverse) {
> > -			parport_data_reverse(port);
> > -			parport_write_control(port, DB9_NORMAL);
> > +	scoped_guard(mutex_intr, &db9->mutex) {
> > +		if (!db9->used++) {
> > +			parport_claim(db9->pd);
> > +			parport_write_data(port, 0xff);
> > +			if (db9_modes[db9->mode].reverse) {
> > +				parport_data_reverse(port);
> > +				parport_write_control(port, DB9_NORMAL);
> > +			}
> > +			mod_timer(&db9->timer, jiffies + DB9_REFRESH_TIME);
> >  		}
> > -		mod_timer(&db9->timer, jiffies + DB9_REFRESH_TIME);
> > +
> > +		return 0;
> >  	}
> >  
> > -	mutex_unlock(&db9->mutex);
> > -	return 0;
> > +	return -EINTR;
> 
> This patch and any others like it are potentially introducing a bug.
> 
> From inspecting the source code, it looks like
> mutex_lock_interruptible() can return -EINTR, -EALREADY, or -EDEADLK.
> 
> Before this patch, the return value of mutex_lock_interruptible() was
> passed to the caller. Now, the return value is reduced to pass/fail
> and only -EINTR is returned on failure when the reason could have
> been something else.

It is documented that mutex_lock_interruptible() only returns 0 or
-EINTR. These additional errors only returned from __mutex_lock_common()
for WW mutexes.

If there is another form of scoped_cond_guard() that would make
available error code returned by the constructor of the locking
primitive we can switch to it later.

Thanks.

-- 
Dmitry