Message-ID: <d7fd564775351ea8a60a6ada83a0368a99ea6b19.camel@esd.eu>
Date: Sun, 9 Feb 2025 18:53:43 +0000
From: Stefan Mätje <Stefan.Maetje@....eu>
To: "mkl@...gutronix.de" <mkl@...gutronix.de>
CC: "horms@...nel.org" <horms@...nel.org>, "linux-can@...r.kernel.org"
	<linux-can@...r.kernel.org>, Frank Jungclaus <frank.jungclaus@....eu>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"mailhol.vincent@...adoo.fr" <mailhol.vincent@...adoo.fr>
Subject: Re: [PATCH 1/1] can: esd_usb: Fix not detecting version reply in
 probe routine

Am Donnerstag, dem 06.02.2025 um 13:45 +0100 schrieb Marc Kleine-Budde:
> > On 03.02.2025 15:58:10, Stefan Mätje wrote:
> > > > This patch fixes some problems in the esd_usb_probe routine that render
> > > > the CAN interface unusable.
> > > > 
> > > > The probe routine sends a version request message to the USB device to
> > > > receive a version reply with the number of CAN ports and the hard-
> > > > & firmware versions. Then for each CAN port a CAN netdev is registered.
> > > > 
> > > > The previous code assumed that the version reply would be received
> > > > immediately. But if the driver was reloaded without power cycling the
> > > > USB device (i. e. on a reboot) there could already be other incoming
> > > > messages in the USB buffers. These would be in front of the version
> > > > reply and need to be skipped.
> > > > 
> > > > In the previous code these problems were present:
> > > > - Only one usb_bulk_msg() read was done into a buffer of
> > > >   sizeof(union esd_usb_msg) which is smaller than ESD_USB_RX_BUFFER_SIZE
> > > >   which could lead to an overflow error from the USB stack.
> > > > - The first bytes of the received data were taken without checking for
> > > >   the message type. This could lead to zero detected CAN interfaces.
> > > > - On kmalloc() fail for the "union esd_usb_msg msg" (i. e. msg == NULL)
> > > >   kfree() would be called with this NULL pointer.
> > > > 
> > > > To mitigate these problems:
> > > > - Use a transfer buffer "buf" with ESD_USB_RX_BUFFER_SIZE.
> > > > - Fix the error exit path taken after allocation failure for the
> > > >   transfer buffer.
> > > > - Added a function esd_usb_recv_version() that reads and skips incoming
> > > >   "esd_usb_msg" messages until a version reply message is found. This
> > > >   is evaluated to return the count of CAN ports and version information.
> > > > 
> > > > Fixes: 80662d943075 ("can: esd_usb: Add support for esd CAN-USB/3")
> > > > Signed-off-by: Stefan Mätje <stefan.maetje@....eu>
> > > > ---
> > > >  drivers/net/can/usb/esd_usb.c | 122 +++++++++++++++++++++++++---------
> > > >  1 file changed, 92 insertions(+), 30 deletions(-)
> > > > 
> > > > diff --git a/drivers/net/can/usb/esd_usb.c b/drivers/net/can/usb/esd_usb.c
> > > > index 03ad10b01867..a6b3b100f8ac 100644
> > > > --- a/drivers/net/can/usb/esd_usb.c
> > > > +++ b/drivers/net/can/usb/esd_usb.c
> > > > @@ -625,17 +625,86 @@ static int esd_usb_send_msg(struct esd_usb *dev, union esd_usb_msg *msg)
> > > >  			    1000);
> > > >  }
> > > >  
> > > > -static int esd_usb_wait_msg(struct esd_usb *dev,
> > > > -			    union esd_usb_msg *msg)
> > > > +static int esd_usb_req_version(struct esd_usb *dev, void *buf)
> > > > +{
> > > > +	union esd_usb_msg *msg = buf;
> > > > +
> > > > +	msg->hdr.cmd = ESD_USB_CMD_VERSION;
> > > > +	msg->hdr.len = sizeof(struct esd_usb_version_msg) / sizeof(u32); /* # of 32bit words */
> > > > +	msg->version.rsvd = 0;
> > > > +	msg->version.flags = 0;
> > > > +	msg->version.drv_version = 0;
> > > > +
> > > > +	return esd_usb_send_msg(dev, msg);
> > > > +}
> > > > +
> > > > +static int esd_usb_recv_version(struct esd_usb *dev,
> > > > +				void *rx_buf,
> > > > +				u32 *version,
> > > > +				int *net_count)
> > 
> > static int esd_usb_recv_version(struct esd_usb *dev, void *rx_buf,
> > 				u32 *version, int *net_count)
> > 
> > > >  {
> > > >  	int actual_length;
> > > > +	int cnt_other = 0;
> > > > +	int cnt_ts = 0;
> > > > +	int cnt_vs = 0;
> > > > +	int attempt;
> > > > +	int pos;
> > > > +	int err;
> > 
> > Try to reduce the scope of the variables and move them into the for-loop.

Changed this in V2.

> > > >  
> > > > -	return usb_bulk_msg(dev->udev,
> > > > -			    usb_rcvbulkpipe(dev->udev, 1),
> > > > -			    msg,
> > > > -			    sizeof(*msg),
> > > > -			    &actual_length,
> > > > -			    1000);
> > > > +	for (attempt = 0; attempt < 8 && cnt_vs == 0; ++attempt) {
> > 
> > Can you create a #define for the "8" to avoid a magic number here?

This value was found empirically. But I think I have to reconsider
the style of handling here and may need to adapt the solution.


> > > > +		err = usb_bulk_msg(dev->udev,
> > > > +				   usb_rcvbulkpipe(dev->udev, 1),
> > > > +				   rx_buf,
> > > > +				   ESD_USB_RX_BUFFER_SIZE,
> > > > +				   &actual_length,
> > > > +				   1000);
> > > > +		if (err)
> > > > +			break;
> > 
> > nitpick: I would make it explicit with "goto bail", should be the same?

Changed this in V2.


> > > > +
> > > > +		err = -ENOENT;
> > > > +		pos = 0;
> > > > +		while (pos < actual_length) {
> > > > +			union esd_usb_msg *p_msg;
> > > > +
> > > > +			p_msg = (union esd_usb_msg *)(rx_buf + pos);
> > > > +
> > > > +			switch (p_msg->hdr.cmd) {
> > > > +			case ESD_USB_CMD_VERSION:
> > > > +				++cnt_vs;
> > > > +				*net_count = (int)p_msg->version_reply.nets;
> > 
> > Cast not needed.

Changed this in V2.


> > What happens if nets is > 2? Please sanitize input from outside against
> > ESD_USB_MAX_NETS.

Will limit the net_count using min() macro in V2.

> > > > +				*version = le32_to_cpu(p_msg->version_reply.version);
> > > > +				err = 0;
> > > > +				dev_dbg(&dev->udev->dev, "TS 0x%08x, V 0x%08x, N %u, F 0x%02x, %.16s\n",
> > > > +					le32_to_cpu(p_msg->version_reply.ts),
> > > > +					le32_to_cpu(p_msg->version_reply.version),
> > > > +					p_msg->version_reply.nets,
> > > > +					p_msg->version_reply.features,
> > > > +					(char *)p_msg->version_reply.name
> > 
> > Is this cast needed?
> > What about using '"%.*s", sizeof(p_msg->version_reply.name)'?

Added a #define for the name attribute size that is also used in the
structure declaration and here as field precision.


> > > > +					);
> > 
> > Please move the closing ")" into the previous line.

Changed in V2.


> > > > +				break;
> > 
> > Why keep parsing after you've found the version?

The version message is assumed to be the last in the block of messages,
so actual_length should be exhausted and the while (pos < actual_length)
loop should stop anyway. If the unexpected happens that there is more
data left after the version message the loop continues to check at least
that the rest of the data forms valid message blocks and bails out if
not.

> > > > +			case ESD_USB_CMD_TS:
> > > > +				++cnt_ts;
> > > > +				dev_dbg(&dev->udev->dev, "TS 0x%08x\n",
> > > > +					le32_to_cpu(p_msg->rx.ts));
> > > > +				break;
> > > > +			default:
> > > > +				++cnt_other;
> > > > +				dev_dbg(&dev->udev->dev, "HDR %d\n", p_msg->hdr.cmd);
> > > > +				break;
> > > > +			}
> > > > +			pos += p_msg->hdr.len * sizeof(u32); /* convert to # of bytes */
> > > > +
> > > > +			if (pos > actual_length) {
> > > > +				dev_err(&dev->udev->dev, "format error\n");
> > > > +				err = -EPROTO;
> > > > +				goto bail;
> > > > +			}
> > > > +		}
> > > > +	}
> > > > +bail:
> > > > +	dev_dbg(&dev->udev->dev, "%s()->%d; ATT=%d, TS=%d, VS=%d, O=%d\n",
> > > > +		__func__, err, attempt, cnt_ts, cnt_vs, cnt_other);
> > > > +	return err;
> > > >  }
> > > >  
> > > >  static int esd_usb_setup_rx_urbs(struct esd_usb *dev)
> > > > @@ -1258,7 +1327,7 @@ static int esd_usb_probe_one_net(struct usb_interface *intf, int index)
> > > >  	}
> > > >  
> > > >  	dev->nets[index] = priv;
> > > > -	netdev_info(netdev, "device %s registered\n", netdev->name);
> > > > +	netdev_info(netdev, "registered\n");
> > > >  
> > > >  done:
> > > >  	return err;
> > > > @@ -1273,13 +1342,13 @@ static int esd_usb_probe(struct usb_interface *intf,
> > > >  			 const struct usb_device_id *id)
> > > >  {
> > > >  	struct esd_usb *dev;
> > > > -	union esd_usb_msg *msg;
> > > > +	void *buf;
> > > >  	int i, err;
> > > >  
> > > >  	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
> > > >  	if (!dev) {
> > > >  		err = -ENOMEM;
> > > > -		goto done;
> > > > +		goto bail;
> > > >  	}
> > > >  
> > > >  	dev->udev = interface_to_usbdev(intf);
> > > > @@ -1288,34 +1357,25 @@ static int esd_usb_probe(struct usb_interface *intf,
> > > >  
> > > >  	usb_set_intfdata(intf, dev);
> > > >  
> > > > -	msg = kmalloc(sizeof(*msg), GFP_KERNEL);
> > > > -	if (!msg) {
> > > > +	buf = kmalloc(ESD_USB_RX_BUFFER_SIZE, GFP_KERNEL);
> > > > +	if (!buf) {
> > > >  		err = -ENOMEM;
> > > > -		goto free_msg;
> > > > +		goto free_dev;
> > > >  	}
> > > >  
> > > >  	/* query number of CAN interfaces (nets) */
> > > > -	msg->hdr.cmd = ESD_USB_CMD_VERSION;
> > > > -	msg->hdr.len = sizeof(struct esd_usb_version_msg) / sizeof(u32); /* # of 32bit words */
> > > > -	msg->version.rsvd = 0;
> > > > -	msg->version.flags = 0;
> > > > -	msg->version.drv_version = 0;
> > > > -
> > > > -	err = esd_usb_send_msg(dev, msg);
> > > > +	err = esd_usb_req_version(dev, buf);
> > > >  	if (err < 0) {
> > > >  		dev_err(&intf->dev, "sending version message failed\n");
> > > > -		goto free_msg;
> > > > +		goto free_buf;
> > > >  	}
> > > >  
> > > > -	err = esd_usb_wait_msg(dev, msg);
> > > > +	err = esd_usb_recv_version(dev, buf, &dev->version, &dev->net_count);
> > 
> > Why pass the "&dev->version, &dev->net_count" pointers, if you already
> > pass dev?

Good point. Changed in V2.

> > > >  	if (err < 0) {
> > > >  		dev_err(&intf->dev, "no version message answer\n");
> > > > -		goto free_msg;
> > > > +		goto free_buf;
> > > >  	}
> > > >  
> > > > -	dev->net_count = (int)msg->version_reply.nets;
> > > > -	dev->version = le32_to_cpu(msg->version_reply.version);
> > > > -
> > > >  	if (device_create_file(&intf->dev, &dev_attr_firmware))
> > > >  		dev_err(&intf->dev,
> > > >  			"Couldn't create device file for firmware\n");
> > > > @@ -1332,11 +1392,12 @@ static int esd_usb_probe(struct usb_interface *intf,
> > > >  	for (i = 0; i < dev->net_count; i++)
> > > >  		esd_usb_probe_one_net(intf, i);
> > 
> > Return values are not checked here. :/

Yes, that is another flaw. This needs to be tackled in another patch.


> > > >  
> > > > -free_msg:
> > > > -	kfree(msg);
> > > > +free_buf:
> > > > +	kfree(buf);
> > > > +free_dev:
> > > >  	if (err)
> > > >  		kfree(dev);
> > > > -done:
> > > > +bail:
> > > >  	return err;
> > > >  }
> > > >  
> > > > @@ -1357,6 +1418,7 @@ static void esd_usb_disconnect(struct usb_interface *intf)
> > > >  		for (i = 0; i < dev->net_count; i++) {
> > > >  			if (dev->nets[i]) {
> > > >  				netdev = dev->nets[i]->netdev;
> > > > +				netdev_info(netdev, "unregister\n");
> > > >  				unregister_netdev(netdev);
> > > >  				free_candev(netdev);
> > > >  			}
> > > > -- 
> > > > 2.34.1
> > > > 
> > > > 
> > > > 
> > 
> > regards,
> > Marc
> > 

Thanks for looking at the patch.

Best regards,
   Stefan

Powered by blists - more mailing lists