Received-SPF: pass (google.com: domain of linux-kernel+bounces-194016-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Precedence: bulk
MIME-Version: 1.0
References: <20240527115419.92606-1-warthog618@gmail.com> <20240527115419.92606-4-warthog618@gmail.com>
In-Reply-To: <20240527115419.92606-4-warthog618@gmail.com>
From: Bartosz Golaszewski <brgl@bgdev.pl>
Date: Wed, 29 May 2024 13:24:45 +0200
Message-ID: <CAMRc=Me+M5PQfuOE=tqqxJF-Q_TVdFb=wh-=ApBO_2PvTV=ZJg@mail.gmail.com>
Subject: Re: [PATCH 3/3] gpiolib: cdev: Cleanup kfifo_out() error handling
To: Kent Gibson <warthog618@gmail.com>
Cc: linux-kernel@vger.kernel.org, linux-gpio@vger.kernel.org, 
	linus.walleij@linaro.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Mon, May 27, 2024 at 1:55=E2=80=AFPM Kent Gibson <warthog618@gmail.com> =
wrote:
>
> The handling of kfifo_out() errors in read functions obscures any error.
> The error condition should never occur but, while a ret is set to -EIO, i=
t
> is subsequently ignored and the read functions instead return the number
> of bytes copied to that point, potentially masking the fact that any erro=
r
> occurred.
>
> Return -EIO in the case of a kfifo_out() error to make it clear something
> very odd is going on here.
>
> Signed-off-by: Kent Gibson <warthog618@gmail.com>
> ---
>  drivers/gpio/gpiolib-cdev.c | 47 +++++++++++++++++--------------------
>  1 file changed, 21 insertions(+), 26 deletions(-)
>
> diff --git a/drivers/gpio/gpiolib-cdev.c b/drivers/gpio/gpiolib-cdev.c
> index c7218c9f2c5e..6a986d7f1f2f 100644
> --- a/drivers/gpio/gpiolib-cdev.c
> +++ b/drivers/gpio/gpiolib-cdev.c
> @@ -1642,16 +1642,13 @@ static ssize_t linereq_read(struct file *file, ch=
ar __user *buf,
>                                         return ret;
>                         }
>
> -                       ret =3D kfifo_out(&lr->events, &le, 1);
> -               }
> -               if (ret !=3D 1) {
> -                       /*
> -                        * This should never happen - we were holding the
> -                        * lock from the moment we learned the fifo is no
> -                        * longer empty until now.
> -                        */
> -                       ret =3D -EIO;
> -                       break;
> +                       if (kfifo_out(&lr->events, &le, 1) !=3D 1)
> +                               /*
> +                                * This should never happen - we hold the

I'm not a native speaker but this looks odd to me - shouldn't it be
"we held the lock from the moment..."?

> +                                * lock from the moment we learned the fi=
fo
> +                                * is no longer empty until now.
> +                                */
> +                               return -EIO;

Since this is so unlikely maybe a WARN() would be justified here too?

Bart

>                 }
>
>                 if (copy_to_user(buf + bytes_read, &le, sizeof(le)))
> @@ -1995,16 +1992,13 @@ static ssize_t lineevent_read(struct file *file, =
char __user *buf,
>                                         return ret;
>                         }
>
> -                       ret =3D kfifo_out(&le->events, &ge, 1);
> -               }
> -               if (ret !=3D 1) {
> -                       /*
> -                        * This should never happen - we were holding the=
 lock
> -                        * from the moment we learned the fifo is no long=
er
> -                        * empty until now.
> -                        */
> -                       ret =3D -EIO;
> -                       break;
> +                       if (kfifo_out(&le->events, &ge, 1) !=3D 1)
> +                               /*
> +                                * This should never happen - we hold the
> +                                * lock from the moment we learned the fi=
fo
> +                                * is no longer empty until now.
> +                                */
> +                               return -EIO;
>                 }
>
>                 if (copy_to_user(buf + bytes_read, &ge, ge_size))
> @@ -2707,12 +2701,13 @@ static ssize_t lineinfo_watch_read(struct file *f=
ile, char __user *buf,
>                         if (count < event_size)
>                                 return -EINVAL;
>  #endif
> -                       ret =3D kfifo_out(&cdev->events, &event, 1);
> -               }
> -               if (ret !=3D 1) {
> -                       ret =3D -EIO;
> -                       break;
> -                       /* We should never get here. See lineevent_read()=
 */
> +                       if (kfifo_out(&cdev->events, &event, 1) !=3D 1)
> +                               /*
> +                                * This should never happen - we hold the
> +                                * lock from the moment we learned the fi=
fo
> +                                * is no longer empty until now.
> +                                */
> +                               return -EIO;
>                 }
>
>  #ifdef CONFIG_GPIO_CDEV_V1
> --
> 2.39.2
>