Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230607214102.2113-1-jason.gerecke@wacom.com>
 <20230608213828.2108-1-jason.gerecke@wacom.com> <CANRwn3R-XbfB+mP9AQ-J7R_19jLi4eS3MhswaWjL+LCEih-7pg@mail.gmail.com>
 <CAO-hwJJC12dRhmykE+P_LBcEJ2G0gHy3Nh1gvWULjdA=4qa-ZQ@mail.gmail.com>
In-Reply-To: <CAO-hwJJC12dRhmykE+P_LBcEJ2G0gHy3Nh1gvWULjdA=4qa-ZQ@mail.gmail.com>
From:   Jason Gerecke <killertofu@gmail.com>
Date:   Wed, 21 Jun 2023 13:52:10 -0700
Message-ID: <CANRwn3SwEOZWxkCAU8FvFZyLHNpq78bZQtN7tbGTQjU=sQ9iAA@mail.gmail.com>
Subject: Re: [PATCH v2] HID: wacom: Use ktime_t rather than int when dealing
 with timestamps
To:     Benjamin Tissoires <benjamin.tissoires@redhat.com>
Cc:     linux-input@vger.kernel.org, linux-kernel@vger.kernel.org,
        Jiri Kosina <jikos@kernel.org>,
        Ping Cheng <pinglinux@gmail.com>,
        Aaron Armstrong Skomra <skomra@gmail.com>,
        Joshua Dickens <Joshua@joshua-dickens.com>,
        Peter Hutterer <peter.hutterer@who-t.net>,
        Jason Gerecke <jason.gerecke@wacom.com>, stable@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Wed, Jun 21, 2023 at 9:04=E2=80=AFAM Benjamin Tissoires
<benjamin.tissoires@redhat.com> wrote:
>
> On Wed, Jun 21, 2023 at 5:18=E2=80=AFPM Jason Gerecke <killertofu@gmail.c=
om> wrote:
> >
> > Following up since no action seems to have been taken on this patch yet=
.
>
> Sorry, this went through the cracks (I seem to have a lot of cracks recen=
tly...)
>
> >
> > On Thu, Jun 8, 2023 at 2:38=E2=80=AFPM Jason Gerecke <killertofu@gmail.=
com> wrote:
> > >
> > > Code which interacts with timestamps needs to use the ktime_t type
> > > returned by functions like ktime_get. The int type does not offer
> > > enough space to store these values, and attempting to use it is a
> > > recipe for problems. In this particular case, overflows would occur
> > > when calculating/storing timestamps leading to incorrect values being
> > > reported to userspace. In some cases these bad timestamps cause input
> > > handling in userspace to appear hung.
>
> I have to ask, is this something we should consider writing a test for? :=
)
>

Very good point! I'm happy to work on such a test.

Are you open to merging this patch without delay while I write a
testcase? I don't like the idea of this (apparent) system freeze being
affecting users any longer than absolutely necessary.

> Also, you are missing the rev-by from Peter, not sure if the tools
> will pick it up automatically then.
>
> Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
>

Oof, good catch. Apologies to Peter :)

Jason

> Cheers,
> Benjamin
>
> > >
> > > Link: https://gitlab.freedesktop.org/libinput/libinput/-/issues/901
> > > Fixes: 17d793f3ed53 ("HID: wacom: insert timestamp to packed Bluetoot=
h (BT) events")
> > > CC: stable@vger.kernel.org
> > > Signed-off-by: Jason Gerecke <jason.gerecke@wacom.com>
> > > ---
> > > v2: Use div_u64 to perform division to deal with ARC and ARM architec=
tures
> > >     (as found by the kernel test robot)
> > >
> > >  drivers/hid/wacom_wac.c | 6 +++---
> > >  drivers/hid/wacom_wac.h | 2 +-
> > >  2 files changed, 4 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/drivers/hid/wacom_wac.c b/drivers/hid/wacom_wac.c
> > > index 2ccf838371343..174bf03908d7c 100644
> > > --- a/drivers/hid/wacom_wac.c
> > > +++ b/drivers/hid/wacom_wac.c
> > > @@ -1314,7 +1314,7 @@ static void wacom_intuos_pro2_bt_pen(struct wac=
om_wac *wacom)
> > >         struct input_dev *pen_input =3D wacom->pen_input;
> > >         unsigned char *data =3D wacom->data;
> > >         int number_of_valid_frames =3D 0;
> > > -       int time_interval =3D 15000000;
> > > +       ktime_t time_interval =3D 15000000;
> > >         ktime_t time_packet_received =3D ktime_get();
> > >         int i;
> > >
> > > @@ -1348,7 +1348,7 @@ static void wacom_intuos_pro2_bt_pen(struct wac=
om_wac *wacom)
> > >         if (number_of_valid_frames) {
> > >                 if (wacom->hid_data.time_delayed)
> > >                         time_interval =3D ktime_get() - wacom->hid_da=
ta.time_delayed;
> > > -               time_interval /=3D number_of_valid_frames;
> > > +               time_interval =3D div_u64(time_interval, number_of_va=
lid_frames);
> > >                 wacom->hid_data.time_delayed =3D time_packet_received=
;
> > >         }
> > >
> > > @@ -1359,7 +1359,7 @@ static void wacom_intuos_pro2_bt_pen(struct wac=
om_wac *wacom)
> > >                 bool range =3D frame[0] & 0x20;
> > >                 bool invert =3D frame[0] & 0x10;
> > >                 int frames_number_reversed =3D number_of_valid_frames=
 - i - 1;
> > > -               int event_timestamp =3D time_packet_received - frames=
_number_reversed * time_interval;
> > > +               ktime_t event_timestamp =3D time_packet_received - fr=
ames_number_reversed * time_interval;
> > >
> > >                 if (!valid)
> > >                         continue;
> > > diff --git a/drivers/hid/wacom_wac.h b/drivers/hid/wacom_wac.h
> > > index 1a40bb8c5810c..ee21bb260f22f 100644
> > > --- a/drivers/hid/wacom_wac.h
> > > +++ b/drivers/hid/wacom_wac.h
> > > @@ -324,7 +324,7 @@ struct hid_data {
> > >         int ps_connected;
> > >         bool pad_input_event_flag;
> > >         unsigned short sequence_number;
> > > -       int time_delayed;
> > > +       ktime_t time_delayed;
> > >  };
> > >
> > >  struct wacom_remote_data {
> > > --
> > > 2.41.0
> > >
> >
>