From: Heikki Krogerus <heikki.krogerus@linux.intel.com>
To: Alexey Charkov <alchark@flipper.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Guenter Roeck <linux@roeck-us.net>,
Yueyao Zhu <yueyao@google.com>,
Sebastian Reichel <sebastian.reichel@collabora.com>,
linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] usb: typec: fusb302: Keep interrupt flags across a failed register read
Date: Fri, 20 Nov 2026 16:13:11 +0100 [thread overview]
Message-ID: <awBjhxTQmoj-MzHq@black.igk.intel.com> (raw)
In-Reply-To: <CAKTNdwES7sRRDjvgFdGttesY9dppOkfBjNhwsJuwUWN1T9DFvA@mail.gmail.com>
On Fri, Nov 20, 2026 at 05:52:45PM +0400, Alexey Charkov wrote:
> Hi Heikki,
>
> On Fri, Nov 20, 2026 at 5:04 PM Heikki Krogerus
> <heikki.krogerus@linux.intel.com> wrote:
> >
> > Hi Alexey,
> >
> > On Tue, Nov 17, 2026 at 12:25:31PM +0400, Alexey Charkov wrote:
> > > Reading INTERRUPT, INTERRUPTA or INTERRUPTB clears its flags in the chip,
> > > so bailing out on an I2C error in a later read loses the flags already
> > > read. On Flipper One, a NACK on INTERRUPTB just as a source raised VBUS
> > > lost VBUSOK. The driver kept treating VBUS as off, so TCPM never attached
> > > as a sink and left every Source_Capabilities unanswered.
> > >
> > > Keep the flags pending until they are handled, and retry the reads a few
> > > times, as such errors pass. Otherwise a flag left pending waits for the
> > > next interrupt.
> >
> > I did not quite understand this. What's the i2c error? Shouldn't the
> > i2c adapter driver take care of the retries?
>
> Any I2C transactions can fail - that's why the accessors return status
> codes. And they do, e.g. if another device on a shared bus misbehaves
> and hogs the bus down, or if there's any intermittent interference,
> etc. The I2C controller won't rerun the failed transaction and will
> propagate the error code to the caller instead. In my case it was
> -ENXIO when another device on the bus interfered with reading the
> interrupt status registers midway, causing the irq_work function to
> exit early and drop the flags it has already read without acting on
> them. The result was a permanently desynchronized TCPM state machine
> vs. what the partner saw on the wire.
Okay, got it.
> > > Fixes: c034a43e72dd ("staging: typec: Fairchild FUSB302 Type-c chip driver")
> > > Signed-off-by: Alexey Charkov <alchark@flipper.net>
> >
> > No stable?
>
> Cc: stable would make sense; my oversight. Thanks for spotting that.
> Shall I resend with it added?
Yes, please send v2 with it added.
Thanks,
--
heikki
prev parent reply other threads:[~2026-11-20 15:13 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-17 8:25 [PATCH] usb: typec: fusb302: Keep interrupt flags across a failed register read Alexey Charkov
2026-11-17 8:31 ` sashiko-bot
2026-11-20 13:04 ` Heikki Krogerus
2026-11-20 13:52 ` Alexey Charkov
2026-11-20 15:13 ` Heikki Krogerus [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=awBjhxTQmoj-MzHq@black.igk.intel.com \
--to=heikki.krogerus@linux.intel.com \
--cc=alchark@flipper.net \
--cc=gregkh@linuxfoundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=linux@roeck-us.net \
--cc=sebastian.reichel@collabora.com \
--cc=yueyao@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox