linux-usb.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Alexey Charkov" <alchark@flipper.net>
Cc: linux-usb@vger.kernel.org
Subject: Re: [PATCH v3 3/3] usb: gadget: udc: Decide on connecting in one place
Date: Tue, 17 Nov 2026 13:19:49 +0000	[thread overview]
Message-ID: <sashiko-outbox-218865@kernel.org> (raw)
In-Reply-To: <20261117-udc-connect-v3-3-9413b491d213@flipper.net>

Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] usb: gadget: udc: missing driver->disconnect() when
  !ops->pullup
- [Medium] usb: gadget: udc: -EOPNOTSUPP from pullup() breaks bind and
  activation
- [Medium] usb: gadget: udc: driver->disconnect() called when pullup
  disable fails
- [Low] usb: gadget: udc: unreachable !gadget->ops->pullup check in
  disconnect
--

commit f20c9b97618b74d0e27972a1d4cf419b12e02088
Author: Alexey Charkov <alchark@flipper.net>

usb: gadget: udc: Decide on connecting in one place

This commit centralizes UDC connection state evaluation in
usb_udc_update_connection_locked() so that pullup control depends on driver
binding, VBUS presence, activation state, and sysfs soft_connect.

> diff --git a/drivers/usb/gadget/udc/core.c b/drivers/usb/gadget/udc/core.c
> index 449b303c2f813..496083045a993 100644
> --- a/drivers/usb/gadget/udc/core.c
> +++ b/drivers/usb/gadget/udc/core.c
[ ... ]
> @@ -766,26 +773,20 @@ static int usb_gadget_disconnect_locked(struct usb_gadget *gadget)
>  {
>  	int ret = 0;
>  
> -	if (!gadget->ops->pullup) {
> -		ret = -EOPNOTSUPP;
> -		goto out;
> -	}
> -
>  	if (!gadget->connected)
>  		goto out;
>  
> -	if (gadget->deactivated || !gadget->udc->started) {
> -		/*
> -		 * If gadget is deactivated we only save new state.
> -		 * Gadget will stay disconnected after activation.
> -		 */
> -		gadget->connected = false;
> +	if (!gadget->ops->pullup) {
> +		ret = -EOPNOTSUPP;
>  		goto out;
>  	}

[Severity: Low]
Is this !gadget->ops->pullup check reachable?  gadget->connected is only set
in usb_gadget_connect_locked() after checking that gadget->ops->pullup is
non-NULL, so when ops->pullup is NULL, !gadget->connected always jumps to out
first and usb_gadget_disconnect() returns 0 instead of -EOPNOTSUPP.

>  
>  	ret = gadget->ops->pullup(gadget, 0);
>  	if (!ret)
>  		gadget->connected = 0;
> +	else if (ret != -EOPNOTSUPP)
> +		dev_warn(&gadget->udc->dev, "failed to disable the pullup: %d\n",
> +			 ret);
>  
>  	mutex_lock(&udc_lock);
>  	if (gadget->udc->driver)
>  		gadget->udc->driver->disconnect(gadget);
>  	mutex_unlock(&udc_lock);

[Severity: Medium]
Should gadget->udc->driver->disconnect(gadget) still run when
gadget->ops->pullup(gadget, 0) fails?

When disabling the pullup fails, gadget->connected stays 1 and
usb_gadget_deactivate() rolls back gadget->deactivated = false under the
assumption that the gadget remains connected and active.  However,
driver->disconnect() has already reset the composite configuration and
disabled the function endpoints while the hardware pullup is still enabled,
and any later retry (such as from soft_connect_store() or vbus_event_work())
will call driver->disconnect() again.

[ ... ]

> @@ -823,6 +828,34 @@ int usb_gadget_disconnect(struct usb_gadget *gadget)
>  }
>  EXPORT_SYMBOL_GPL(usb_gadget_disconnect);
>  
> +static void usb_udc_update_allow_connect(struct usb_udc *udc)
> +	__must_hold(&udc->connect_lock)
> +{
> +	udc->allow_connect = udc->bound && udc->vbus &&
> +			     !udc->gadget->deactivated &&
> +			     !udc->soft_disconnected;
> +}
> +
> +/*
> + * Decide whether the gadget may connect and connect or disconnect it to match.
> + * Call this after changing any of the state that
> + * usb_udc_update_allow_connect() checks.
> + */
> +static int usb_udc_update_connection_locked(struct usb_udc *udc)
> +	__must_hold(&udc->connect_lock)
> +{
> +	usb_udc_update_allow_connect(udc);
> +
> +	/* Without a pullup control there is nothing to switch */
> +	if (!udc->gadget->ops->pullup)
> +		return 0;
> +
> +	if (udc->allow_connect)
> +		return usb_gadget_connect_locked(udc->gadget);
> +
> +	return usb_gadget_disconnect_locked(udc->gadget);
> +}

[Severity: Medium]
What happens when a UDC driver defines ops->pullup, but that callback returns
-EOPNOTSUPP?

usb_gadget_connect_locked() and usb_gadget_disconnect_locked() suppress the
warning when ops->pullup() returns -EOPNOTSUPP, but still return -EOPNOTSUPP
to usb_udc_update_connection_locked().  UDC drivers such as pxa27x_udc (when
no pullup GPIO is configured) and fsl_udc_core (when VBUS is inactive, since
it does not use usb_udc_vbus_handler()) return -EOPNOTSUPP from their pullup
callbacks, which causes gadget_bind_driver(), usb_gadget_activate(), and
usb_gadget_deactivate() to fail with -EOPNOTSUPP.

[ ... ]

> @@ -1726,21 +1730,31 @@ static void gadget_unbind_driver(struct device *dev)
>  
>  	dev_dbg(&udc->dev, "unbinding gadget driver [%s]\n", driver->function);
>  
> -	udc->allow_connect = false;
> -	cancel_work_sync(&udc->vbus_work);
>  	mutex_lock(&udc->connect_lock);
> -	usb_gadget_disconnect_locked(gadget);
> +	udc->bound = false;
> +	usb_udc_update_connection_locked(udc);
> +	/*
> +	 * The UDC stops below even if the pullup refused to switch off (some
> +	 * UDCs refuse without VBUS), so treat the gadget as disconnected from
> +	 * now on. Neither a late VBUS event nor the next driver's bind should
> +	 * act on a connection that ends with this driver.
> +	 */
> +	gadget->connected = false;
>  	usb_gadget_disable_async_callbacks(udc);
>  	if (gadget->irq)
>  		synchronize_irq(gadget->irq);
>  	mutex_unlock(&udc->connect_lock);
>  
> +	/* A VBUS event can no longer connect the gadget, so just wait it out */
> +	cancel_work_sync(&udc->vbus_work);
> +
>  	udc->driver->unbind(gadget);

[Severity: High]
On a UDC without ops->pullup (such as max3420_udc or fotg210-udc), does
udc->driver->disconnect(gadget) ever get called before udc->driver->unbind()
or on VBUS loss?

With usb_udc_update_connection_locked() returning 0 early when
!udc->gadget->ops->pullup, gadget_bind_driver() now succeeds and starts the
UDC without setting gadget->connected.  Once the host enumerates and
configures the gadget, unbinding the driver (or losing VBUS in
vbus_event_work()) skips usb_gadget_disconnect_locked() and never invokes
udc->driver->disconnect(gadget).  udc->driver->unbind(gadget) then runs while
cdev->config is still active and function->disable() has never been called,
triggering WARN_ON(cdev->config) in __composite_unbind() or tearing down
active functions in configfs_composite_unbind() with enabled endpoints and
dangling function state (such as use-after-free on dev->port_usb in u_ether).

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117-udc-connect-v3-0-9413b491d213@flipper.net?part=3

      reply	other threads:[~2026-11-17 13:19 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 12:48 [PATCH v3 0/3] usb: gadget: udc: Rework when the gadget connects Alexey Charkov
2026-11-17 12:48 ` [PATCH v3 1/3] usb: gadget: udc: Clear a pending connection when starting the UDC Alexey Charkov
2026-11-17 13:03   ` sashiko-bot
2026-11-17 12:48 ` [PATCH v3 2/3] usb: gadget: udc: Forget a deactivation when the gadget driver unbinds Alexey Charkov
2026-11-17 13:06   ` sashiko-bot
2026-11-17 12:48 ` [PATCH v3 3/3] usb: gadget: udc: Decide on connecting in one place Alexey Charkov
2026-11-17 13:19   ` sashiko-bot [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=sashiko-outbox-218865@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=alchark@flipper.net \
    --cc=linux-usb@vger.kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /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;
as well as URLs for NNTP newsgroup(s).