Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:6 as permitted sender) client-ip=2620:137:e000::3:6;
MIME-Version: 1.0
References: <20230924010214.3700150-1-william@wkennington.com>
In-Reply-To: <20230924010214.3700150-1-william@wkennington.com>
From:   Tali Perry <tali.perry1@gmail.com>
Date:   Tue, 26 Sep 2023 08:41:23 +0300
Message-ID: <CAHb3i=uOMuUAz=zY_GFvBi-9JFaaPhN4F0Ve5i1f4WhyjgH8Bw@mail.gmail.com>
Subject: Re: [PATCH] i2c: npcm7xx: Fix callback completion ordering
To:     "William A. Kennington III" <william@wkennington.com>
Cc:     tmaimon77@gmail.com, avifishman70@gmail.com, wsa@kernel.org,
        joel@jms.id.au, linux-i2c@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Sun, Sep 24, 2023 at 4:02=E2=80=AFAM William A. Kennington III
<william@wkennington.com> wrote:
>
> Sometimes, our completions race with new master transfers and override
> the bus->operation and bus->master_or_slave variables. This causes
> transactions to timeout and kernel crashes less frequently.
>
> To remedy this, we re-order all completions to the very end of the
> function.
>
> Fixes: 56a1485b102e ("i2c: npcm7xx: Add Nuvoton NPCM I2C controller drive=
r")
> Signed-off-by: William A. Kennington III <william@wkennington.com>
> ---
>  drivers/i2c/busses/i2c-npcm7xx.c | 17 +++++++----------
>  1 file changed, 7 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/i2c/busses/i2c-npcm7xx.c b/drivers/i2c/busses/i2c-np=
cm7xx.c
> index 495a8b5f6a2b..ae4bae63ad4f 100644
> --- a/drivers/i2c/busses/i2c-npcm7xx.c
> +++ b/drivers/i2c/busses/i2c-npcm7xx.c
> @@ -694,6 +694,7 @@ static void npcm_i2c_callback(struct npcm_i2c *bus,
>  {
>         struct i2c_msg *msgs;
>         int msgs_num;
> +       bool do_complete =3D false;
>
>         msgs =3D bus->msgs;
>         msgs_num =3D bus->msgs_num;
> @@ -722,23 +723,17 @@ static void npcm_i2c_callback(struct npcm_i2c *bus,
>                                  msgs[1].flags & I2C_M_RD)
>                                 msgs[1].len =3D info;
>                 }
> -               if (completion_done(&bus->cmd_complete) =3D=3D false)
> -                       complete(&bus->cmd_complete);
> -       break;
> -
> +               do_complete =3D true;
> +               break;
>         case I2C_NACK_IND:
>                 /* MASTER transmit got a NACK before tx all bytes */
>                 bus->cmd_err =3D -ENXIO;
> -               if (bus->master_or_slave =3D=3D I2C_MASTER)
> -                       complete(&bus->cmd_complete);
> -
> +               do_complete =3D true;
>                 break;
>         case I2C_BUS_ERR_IND:
>                 /* Bus error */
>                 bus->cmd_err =3D -EAGAIN;
> -               if (bus->master_or_slave =3D=3D I2C_MASTER)
> -                       complete(&bus->cmd_complete);
> -
> +               do_complete =3D true;
>                 break;
>         case I2C_WAKE_UP_IND:
>                 /* I2C wake up */
> @@ -752,6 +747,8 @@ static void npcm_i2c_callback(struct npcm_i2c *bus,
>         if (bus->slave)
>                 bus->master_or_slave =3D I2C_SLAVE;
>  #endif
> +       if (do_complete)
> +               complete(&bus->cmd_complete);
>  }
>
>  static u8 npcm_i2c_fifo_usage(struct npcm_i2c *bus)
> --
> 2.42.0.515.g380fc7ccd1-goog
>

Thanks William for the fix!

Reviewed-by: Tali Perry <tali.perry1@gmail.com>