From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ben Hutchings Subject: Re: [PATCHv3] drivers/net/usb: Add new driver ipheth Date: Thu, 01 Apr 2010 00:18:58 +0100 Message-ID: <1270077538.8653.484.camel@localhost> References: <1269984864-28159-1-git-send-email-agimenez@sysvalve.es> <1270064527-8485-1-git-send-email-agimenez@sysvalve.es> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: linux-kernel@vger.kernel.org, netdev@vger.kernel.org, linux-usb@vger.kernel.org, oliver@neukum.org, linville@tuxdriver.com, j.dumon@option.com, steve.glendinning@smsc.com, davem@davemloft.net, gregkh@suse.de, dgiagio@gmail.com, dborca@yahoo.com To: "L. Alberto" =?ISO-8859-1?Q?Gim=E9nez?= Return-path: In-Reply-To: <1270064527-8485-1-git-send-email-agimenez@sysvalve.es> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Wed, 2010-03-31 at 21:42 +0200, L. Alberto Gim=C3=A9nez wrote: [...] > --- /dev/null > +++ b/drivers/net/usb/ipheth.c [...] > +#include > +#include > +#include > +#include > +#include > +#include > +#include > +#include > +#include I don't think you need this header. > +#include > +#include > + > +#define USB_VENDOR_APPLE 0x05ac > +#define USB_PRODUCT_IPHETH 0x1290 > +#define USB_PRODUCT_IPHETH_3G 0x1292 > +#define USB_PRODUCT_IPHETH_3GS 0x1294 Apple doesn't assign device ids to ipheth so either the names are incorrect or you should get proper device ids. I believe the Linux =46oundation has a USB vendor id and could assign device ids under that= =2E > +struct ipheth_device { > + struct usb_device *udev; > + struct usb_interface *intf; > + struct net_device *net; > + struct net_device_stats stats; You can store stats in the net_device. > + struct sk_buff *tx_skb; > + struct urb *tx_urb; > + struct urb *rx_urb; > + unsigned char *tx_buf; > + unsigned char *rx_buf; > + unsigned char *ctrl_buf; > + __u8 bulk_in; > + __u8 bulk_out; No need for the double-underscores; that's a convention for use in definitions shared with user-space. > + struct delayed_work carrier_work; > +}; [...] > +static int ipheth_open(struct net_device *net) > +{ > + struct ipheth_device *dev =3D netdev_priv(net); > + struct usb_device *udev =3D dev->udev; > + int retval =3D 0; > + > + usb_set_interface(udev, IPHETH_INTFNUM, IPHETH_ALT_INTFNUM); > + usb_clear_halt(udev, usb_rcvbulkpipe(udev, dev->bulk_in)); > + usb_clear_halt(udev, usb_sndbulkpipe(udev, dev->bulk_out)); > + > + retval =3D ipheth_carrier_set(dev); > + if (retval) > + goto error; > + > + retval =3D ipheth_rx_submit(dev, GFP_KERNEL); > + if (retval) > + goto error; > + > + schedule_delayed_work(&dev->carrier_work, IPHETH_CARRIER_CHECK_TIME= OUT); > + netif_start_queue(net); > +error: > + return retval; > +} There is no cleanup here (and none appears to be needed) so you could take out the label and replace the gotos with returns. > +static int ipheth_tx(struct sk_buff *skb, struct net_device *net) > +{ > + struct ipheth_device *dev =3D netdev_priv(net); > + struct usb_device *udev =3D dev->udev; > + int retval; > + > + /* Paranoid */ > + if (skb->len > IPHETH_BUF_SIZE) { > + err("%s: skb too large: %d bytes", __func__, skb->len); > + dev->stats.tx_dropped++; > + dev_kfree_skb_irq(skb); > + goto exit; > + } Use WARN here as this would indicate a bug. Again the goto is unnecessary. > + memset(dev->tx_buf, 0, IPHETH_BUF_SIZE); > + memcpy(dev->tx_buf, skb->data, skb->len); This is wasteful. If you really must pad the buffer then do so, but don't clear the area that is going to be filled with real data. > + usb_fill_bulk_urb(dev->tx_urb, udev, > + usb_sndbulkpipe(udev, dev->bulk_out), > + dev->tx_buf, IPHETH_BUF_SIZE, > + ipheth_sndbulk_callback, > + dev); > + dev->tx_urb->transfer_flags |=3D URB_NO_TRANSFER_DMA_MAP; > + > + retval =3D usb_submit_urb(dev->tx_urb, GFP_ATOMIC); > + if (retval) { > + err("%s: usb_submit_urb: %d", __func__, retval); > + dev->stats.tx_errors++; > + dev_kfree_skb_irq(skb); > + } else { > + net->trans_start =3D jiffies; No longer needed. > + dev->tx_skb =3D skb; > + > + dev->stats.tx_packets++; > + dev->stats.tx_bytes +=3D skb->len; > + netif_stop_queue(net); > + } > +exit: > + return NETDEV_TX_OK; > +} [...] > +#ifdef HAVE_NET_DEVICE_OPS > +static const struct net_device_ops ipheth_netdev_ops =3D { > + .ndo_open =3D &ipheth_open, > + .ndo_stop =3D &ipheth_close, > + .ndo_start_xmit =3D &ipheth_tx, > + .ndo_tx_timeout =3D &ipheth_tx_timeout, > + .ndo_get_stats =3D &ipheth_stats, > +}; > +#endif Remove the #ifdef, there is no question whether we have net_device_ops. > +static int ipheth_probe(struct usb_interface *intf, > + const struct usb_device_id *id) > +{ [...] > +#ifdef HAVE_NET_DEVICE_OPS > + netdev->netdev_ops =3D &ipheth_netdev_ops; > +#else /* CONFIG_COMPAT_NET_DEV_OPS */ > + netdev->open =3D &ipheth_open; > + netdev->stop =3D &ipheth_close; > + netdev->hard_start_xmit =3D &ipheth_tx; > + netdev->tx_timeout =3D &ipheth_tx_timeout; > + netdev->get_stats =3D &ipheth_stats; > +#endif [...] Similarly here. As a general point, you should now be using netdev_err(), netdev_info()= , etc. for logging messages related to net devices. I have no idea about USB so I haven't checked the USB API usage at all. Ben. --=20 Ben Hutchings, Senior Software Engineer, Solarflare Communications Not speaking for my employer; that's the marketing department's job. They asked us to note that Solarflare product names are trademarked.