Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Message-ID: <cd62e9a6d317e106db5e5d6b5f36170524ed7ad9.camel@redhat.com>
Subject: Re: [PATCH] usb: typec: ucsi: Clear pending after acking connector
 change
From:   Benjamin Berg <bberg@redhat.com>
To:     Heikki Krogerus <heikki.krogerus@linux.intel.com>,
        Bjorn Andersson <bjorn.andersson@linaro.org>
Cc:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org
Date:   Mon, 17 May 2021 14:57:28 +0200
In-Reply-To: <YKJeYzIgvL/soGgw@kuha.fi.intel.com>
References: <20210516040953.622409-1-bjorn.andersson@linaro.org>
         <YKI/XT8qpZDjDuqs@kuha.fi.intel.com> <YKJeYzIgvL/soGgw@kuha.fi.intel.com>
Content-Type: multipart/signed; micalg="pgp-sha256";
        protocol="application/pgp-signature"; boundary="=-nvdWtIGDTaPoTjHWCKJp"
User-Agent: Evolution 3.40.0 (3.40.0-1.fc34) 
MIME-Version: 1.0
Precedence: bulk


--=-nvdWtIGDTaPoTjHWCKJp
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Heikki,


On Mon, 2021-05-17 at 15:15 +0300, Heikki Krogerus wrote:
> On Mon, May 17, 2021 at 01:03:12PM +0300, Heikki Krogerus wrote:
> > On Sat, May 15, 2021 at 09:09:53PM -0700, Bjorn Andersson wrote:
> > > It's possible that the interrupt handler for the UCSI driver
> > > signals a
> > > connector changes after the handler clears the PENDING bit, but
> > > before
> > > it has sent the acknowledge request. The result is that the handler
> > > is
> > > invoked yet again, to ack the same connector change.
> > >=20
> > > At least some versions of the Qualcomm UCSI firmware will not
> > > handle the
> > > second - "spurious" - acknowledgment gracefully. So make sure to
> > > not
> > > clear the pending flag until the change is acknowledged.
> > >=20
> > > Any connector changes coming in after the acknowledgment, that
> > > would
> > > have the pending flag incorrectly cleared, would afaict be covered
> > > by
> > > the subsequent connector status check.
> > >=20
> > > Fixes: 217504a05532 ("usb: typec: ucsi: Work around PPM losing
> > > change information")
> > > Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
> >=20
> > I'm OK with this if Bejamin does not see any problems with it. I'll
> > wait for his comments before giving my reviewed-by tag.
> >=20
> > That workaround (commit 217504a05532) is unfortunately too fragile.
> > I'm going to now separate the processing of the connector state from
> > the event handler (interrupt handler). That way we should be fairly
> > sure we don't loose any of the connector states even if an event is
> > generated while we are still in the middle of processing the previous
> > one(s), and at the same time be sure that we also don't confuse the
> > firmware.
> >=20
> > So the event handler shall after that only read the connector status,
> > schedule the unique job where it's processed and ACK the event.
> > Nothing else.
>=20
> Seems to be straightforward to implement. I'm attaching the patch I
> made for that. I think it should actually also remove the problem you
> are seeing. Can you test it?

Hmm, I am happy to try the patch tomorrow in the scenario where I
observed my race condition.

Unfortunately, I don't feel it'll work. The problem that I was seeing
looked like a race condition in the PPM itself, where the window is the
time between the UCSI_GET_CONNECTOR_STATUS command and the subsequent
ACK.
For such a firmware level bug in the PPM, we need a way to detect the
race condition when it happens (or get a fix for the firmware).

Note that there was also some code to always sent a
UCSI_GET_CAM_SUPPORTED command "to ensure the PPM does not get stuck in
case it assumes we do". I have no idea where this came from or what
PPMs might require this workaround. But I doubt it is a good idea to
drop it.

Benjamin

--=-nvdWtIGDTaPoTjHWCKJp
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: This is a digitally signed message part
Content-Transfer-Encoding: 7bit

-----BEGIN PGP SIGNATURE-----
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=dqmz
-----END PGP SIGNATURE-----

--=-nvdWtIGDTaPoTjHWCKJp--