nuttx/drivers/usbdev/rndis.c

2906 lines
82 KiB
C
Raw Normal View History

/****************************************************************************
* drivers/usbdev/rndis.c
*
* Copyright (C) 2011-2018 Gregory Nutt. All rights reserved.
* Authors: Sakari Kapanen <sakari.m.kapanen@gmail.com>,
* Petteri Aimonen <jpa@git.mail.kapsi.fi>
*
* References:
* [MS-RNDIS]:
* Remote Network Driver Interface Specification (RNDIS) Protocol
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* 3. Neither the name NuttX nor the names of its contributors may be
* used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
* OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
* AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <queue.h>
#include <errno.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include <nuttx/net/net.h>
#include <nuttx/net/netdev.h>
#include <nuttx/net/arp.h>
#include <nuttx/kmalloc.h>
#include <nuttx/arch.h>
#include <nuttx/usb/usb.h>
#include <nuttx/usb/cdc.h>
#include <nuttx/usb/usbdev.h>
#include <nuttx/usb/usbdev_trace.h>
#include <nuttx/usb/rndis.h>
#include <nuttx/wdog.h>
#include <nuttx/wqueue.h>
#include "rndis_std.h"
/****************************************************************************
* Pre-processor definitions
****************************************************************************/
#define CONFIG_RNDIS_EP0MAXPACKET 64
#ifndef CONFIG_RNDIS_NWRREQS
# define CONFIG_RNDIS_NWRREQS (2)
#endif
#define RNDIS_PACKET_HDR_SIZE (sizeof(struct rndis_packet_msg))
#define CONFIG_RNDIS_BULKIN_REQLEN \
(CONFIG_NET_ETH_PKTSIZE + CONFIG_NET_GUARDSIZE + RNDIS_PACKET_HDR_SIZE)
#define CONFIG_RNDIS_BULKOUT_REQLEN CONFIG_RNDIS_BULKIN_REQLEN
#define RNDIS_NCONFIGS (1)
#define RNDIS_CONFIGID (1)
#define RNDIS_CONFIGIDNONE (0)
#define RNDIS_NINTERFACES (2)
#define RNDIS_EPINTIN_ADDR USB_EPIN(3)
#define RNDIS_EPBULKIN_ADDR USB_EPIN(1)
#define RNDIS_EPBULKOUT_ADDR USB_EPOUT(2)
#define RNDIS_NUM_EPS (3)
#define RNDIS_MANUFACTURERSTRID (1)
#define RNDIS_PRODUCTSTRID (2)
#define RNDIS_SERIALSTRID (3)
#define RNDIS_STR_LANGUAGE (0x0409) /* en-us */
#define RNDIS_MXDESCLEN (128)
#define RNDIS_MAXSTRLEN (RNDIS_MXDESCLEN-2)
#define RNDIS_CTRLREQ_LEN (512)
This commit attempts remove some long standard confusion in naming and some actual problems that result from the naming confusion. The basic problem is the standard MTU does not include the size of the Ethernet header. For clarity, I changed the naming of most things called MTU to PKTSIZE. For example, CONFIG_NET_ETH_MTU is now CONFIG_NET_ETH_PKTSIZE. This makes the user interface a little hostile. People thing of an MTU of 1500 bytes, but the corresponding packet is really 1514 bytes (including the 14 byte Ethernet header). A more friendly solution would configure the MTU (as before), but then derive the packet buffer size by adding the MAC header length. Instead, we define the packet buffer size then derive the MTU. The MTU is not common currency in networking. On the wire, the only real issue is the MSS which is derived from MTU by subtracting the IP header and TCP header sizes (for the case of TCP). Now it is derived for the PKTSIZE by subtracting the IP header, the TCP header, and the MAC header sizes. So we should be all good and without the recurring 14 byte error in MTU's and MSS's. Squashed commit of the following: Trivial update to fix some spacing issues. net/: Rename several macros containing _MTU to _PKTSIZE. net/: Rename CONFIG_NET_SLIP_MTU to CONFIG_NET_SLIP_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_6LOWPAN_MTU to CONFIG_NET_6LOWPAN_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_ETH_MTU to CONFIG_NET_ETH_PKTSIZE. This is not the MTU which does not include the size of the link layer header. This is the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename the file d_mtu in the network driver structure to d_pktsize. That value saved there is not the MTU. The packetsize is the memory large enough to hold the maximum packet PLUS the size of the link layer header. The MTU does not include the link layer header.
2018-07-04 22:10:40 +02:00
#define RNDIS_BUFFER_SIZE CONFIG_NET_ETH_PKTSIZE
#define RNDIS_BUFFER_COUNT 4
/* TX poll delay = 1 seconds. CLK_TCK is the number of clock ticks per second */
#define RNDIS_WDDELAY (1*CLK_TCK)
/* Work queue to use for network operations. LPWORK should be used here */
#define ETHWORK LPWORK
#ifndef min
# define min(a,b) ((a)<(b)?(a):(b))
#endif
#ifndef max
# define max(a,b) ((a)>(b)?(a):(b))
#endif
/****************************************************************************
* Private Types
****************************************************************************/
/* Container to support a list of requests */
struct rndis_req_s
{
FAR struct rndis_req_s *flink; /* Implements a singly linked list */
FAR struct usbdev_req_s *req; /* The contained request */
};
/* This structure describes the internal state of the driver */
struct rndis_dev_s
{
struct net_driver_s netdev; /* Network driver structure */
struct usbdev_devinfo_s devinfo;
FAR struct usbdev_s *usbdev; /* usbdev driver pointer */
FAR struct usbdev_ep_s *epintin; /* Interrupt IN endpoint structure */
FAR struct usbdev_ep_s *epbulkin; /* Bulk IN endpoint structure */
FAR struct usbdev_ep_s *epbulkout; /* Bulk OUT endpoint structure */
FAR struct usbdev_req_s *ctrlreq; /* Pointer to preallocated control request */
FAR struct usbdev_req_s *epintin_req; /* Pointer to preallocated interrupt in endpoint request */
FAR struct usbdev_req_s *rdreq; /* Pointer to Preallocated control endpoint read request */
struct sq_queue_s reqlist; /* List of free write request containers */
/* Preallocated USB request buffers */
struct rndis_req_s wrreqs[CONFIG_RNDIS_NWRREQS];
struct work_s rxwork; /* Worker for dispatching RX packets */
WDOG_ID txpoll; /* TX poll watchdog */
struct work_s pollwork; /* TX poll worker */
bool registered; /* Has netdev_register() been called */
uint8_t config; /* USB Configuration number */
FAR struct rndis_req_s *net_req; /* Pointer to request whose buffer is assigned to network */
FAR struct rndis_req_s *rx_req; /* Pointer request container that holds RX buffer */
size_t current_rx_received; /* Number of bytes of current RX datagram received over USB */
size_t current_rx_datagram_size; /* Total number of bytes of the current RX datagram */
size_t current_rx_datagram_offset; /* Offset of current RX datagram */
size_t current_rx_msglen; /* Length of the entire message to be received */
bool rdreq_submitted; /* Indicates if the read request is submitted */
bool rx_blocked; /* Indicates if we can receive packets on bulk in endpoint */
bool ctrlreq_has_encap_response; /* Indicates if ctrlreq buffer holds a response */
bool connected; /* Connection status indicator */
uint32_t rndis_packet_filter; /* RNDIS packet filter value */
uint32_t rndis_host_tx_count; /* TX packet counter */
uint32_t rndis_host_rx_count; /* RX packet counter */
uint8_t host_mac_address[6]; /* Host side MAC address */
};
/* The internal version of the class driver */
struct rndis_driver_s
{
struct usbdevclass_driver_s drvr;
FAR struct rndis_dev_s *dev;
};
/* This is what is allocated */
struct rndis_alloc_s
{
struct rndis_dev_s dev;
struct rndis_driver_s drvr;
};
/* RNDIS USB configuration descriptor */
struct rndis_cfgdesc_s
{
#ifndef CONFIG_RNDIS_COMPOSITE
struct usb_cfgdesc_s cfgdesc; /* Configuration descriptor */
#endif
struct usb_iaddesc_s assoc_desc; /* Interface association descriptor */
struct usb_ifdesc_s comm_ifdesc; /* Communication interface descriptor */
struct usb_epdesc_s epintindesc; /* Interrupt endpoint descriptor */
struct usb_ifdesc_s data_ifdesc; /* Data interface descriptor */
struct usb_epdesc_s epbulkindesc; /* Bulk in interface descriptor */
struct usb_epdesc_s epbulkoutdesc; /* Bulk out interface descriptor */
};
/* RNDIS object ID - value pair structure */
struct rndis_oid_value_s
{
uint32_t objid;
uint32_t length;
uint32_t value;
FAR const void *data; /* Data pointer overrides value if non-NULL. */
};
/****************************************************************************
* Private Function Prototypes
****************************************************************************/
/* Netdev driver callbacks */
static int rndis_ifup(FAR struct net_driver_s *dev);
static int rndis_ifdown(FAR struct net_driver_s *dev);
static int rndis_txavail(FAR struct net_driver_s *dev);
static int rndis_transmit(FAR struct rndis_dev_s *priv);
static int rndis_txpoll(FAR struct net_driver_s *dev);
static void rndis_polltimer(int argc, uint32_t arg, ...);
/* usbclass callbacks */
static int usbclass_setup(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev,
FAR const struct usb_ctrlreq_s *ctrl,
FAR uint8_t *dataout, size_t outlen);
static int usbclass_bind(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev);
static void usbclass_unbind(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev);
static void usbclass_disconnect(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev);
static int usbclass_setconfig(FAR struct rndis_dev_s *priv, uint8_t config);
static void usbclass_resetconfig(FAR struct rndis_dev_s *priv);
/* usbclass helpers */
static int usbclass_copy_epdesc(int epid, FAR struct usb_epdesc_s *epdesc,
FAR struct usbdev_devinfo_s *devinfo,
bool hispeed);
/****************************************************************************
* Private Data
****************************************************************************/
/* USB driver operations */
const static struct usbdevclass_driverops_s g_driverops =
{
&usbclass_bind,
&usbclass_unbind,
&usbclass_setup,
&usbclass_disconnect,
NULL,
NULL
};
#ifndef CONFIG_RNDIS_COMPOSITE
static const struct usb_devdesc_s g_devdesc =
{
USB_SIZEOF_DEVDESC, /* len */
USB_DESC_TYPE_DEVICE, /* type */
{LSBYTE(0x0200), MSBYTE(0x0200)}, /* usb */
0, /* classid */
0, /* subclass */
0, /* protocol */
CONFIG_RNDIS_EP0MAXPACKET, /* maxpacketsize */
{ LSBYTE(CONFIG_RNDIS_VENDORID), /* vendor */
MSBYTE(CONFIG_RNDIS_VENDORID) },
{ LSBYTE(CONFIG_RNDIS_PRODUCTID), /* product */
MSBYTE(CONFIG_RNDIS_PRODUCTID) },
{ LSBYTE(CONFIG_RNDIS_VERSIONNO), /* device */
MSBYTE(CONFIG_RNDIS_VERSIONNO) },
RNDIS_MANUFACTURERSTRID, /* imfgr */
RNDIS_PRODUCTSTRID, /* iproduct */
RNDIS_SERIALSTRID, /* serno */
RNDIS_NCONFIGS /* nconfigs */
};
#endif
const static struct rndis_cfgdesc_s g_rndis_cfgdesc =
{
#ifndef CONFIG_RNDIS_COMPOSITE
{
.len = USB_SIZEOF_CFGDESC,
.type = USB_DESC_TYPE_CONFIG,
.totallen = {0, 0},
.ninterfaces = RNDIS_NINTERFACES,
.cfgvalue = RNDIS_CONFIGID,
.icfg = 0,
.attr = USB_CONFIG_ATTR_ONE | USB_CONFIG_ATTR_SELFPOWER,
.mxpower = (CONFIG_USBDEV_MAXPOWER + 1) / 2
},
#endif
{
.len = USB_SIZEOF_IADDESC,
.type = USB_DESC_TYPE_INTERFACEASSOCIATION,
.firstif = 0,
.nifs = RNDIS_NINTERFACES,
.classid = 0xef,
.subclass = 0x04,
.protocol = 0x01,
.ifunction = 0
},
{
.len = USB_SIZEOF_IFDESC,
.type = USB_DESC_TYPE_INTERFACE,
.ifno = 0,
.alt = 0,
.neps = 1,
.classid = USB_CLASS_CDC,
.subclass = CDC_SUBCLASS_ACM,
.protocol = CDC_PROTO_VENDOR,
.iif = 0
},
{
.len = USB_SIZEOF_EPDESC,
.type = USB_DESC_TYPE_ENDPOINT,
.addr = RNDIS_EPINTIN_ADDR,
.attr = USB_EP_ATTR_XFER_INT,
.mxpacketsize = { LSBYTE(16), MSBYTE(16) },
.interval = 1
},
{
.len = USB_SIZEOF_IFDESC,
.type = USB_DESC_TYPE_INTERFACE,
.ifno = 1,
.alt = 0,
.neps = 2,
.classid = USB_CLASS_CDC_DATA,
.subclass = 0,
.protocol = 0,
.iif = 0
},
{
.len = USB_SIZEOF_EPDESC,
.type = USB_DESC_TYPE_ENDPOINT,
.addr = RNDIS_EPBULKIN_ADDR,
.attr = USB_EP_ATTR_XFER_BULK,
#ifdef CONFIG_USBDEV_DUALSPEED
.mxpacketsize = { LSBYTE(512), MSBYTE(512) },
.interval = 0
#else
.mxpacketsize = { LSBYTE(64), MSBYTE(64) },
.interval = 1
#endif
},
{
.len = USB_SIZEOF_EPDESC,
.type = USB_DESC_TYPE_ENDPOINT,
.addr = RNDIS_EPBULKOUT_ADDR,
.attr = USB_EP_ATTR_XFER_BULK,
#ifdef CONFIG_USBDEV_DUALSPEED
.mxpacketsize = { LSBYTE(512), MSBYTE(512) },
.interval = 0
#else
.mxpacketsize = { LSBYTE(64), MSBYTE(64) },
.interval = 1
#endif
}
};
/* Default MAC address given to the host side of the interface. */
static uint8_t g_rndis_default_mac_addr[6] =
{
0x02, 0x00, 0x00, 0x11, 0x22, 0x33
};
/* These lists give dummy responses to be returned to PC. The values are
* chosen so that Windows is happy - other operating systems don't really care
* much.
*/
static const uint32_t g_rndis_supported_oids[] =
{
RNDIS_OID_GEN_SUPPORTED_LIST,
RNDIS_OID_GEN_HARDWARE_STATUS,
RNDIS_OID_GEN_MEDIA_SUPPORTED,
RNDIS_OID_GEN_MEDIA_IN_USE,
RNDIS_OID_GEN_MAXIMUM_FRAME_SIZE,
RNDIS_OID_GEN_LINK_SPEED,
RNDIS_OID_GEN_TRANSMIT_BLOCK_SIZE,
RNDIS_OID_GEN_RECEIVE_BLOCK_SIZE,
RNDIS_OID_GEN_VENDOR_ID,
RNDIS_OID_GEN_VENDOR_DESCRIPTION,
RNDIS_OID_GEN_VENDOR_DRIVER_VERSION,
RNDIS_OID_GEN_CURRENT_PACKET_FILTER,
RNDIS_OID_GEN_MAXIMUM_TOTAL_SIZE,
RNDIS_OID_GEN_MEDIA_CONNECT_STATUS,
RNDIS_OID_GEN_PHYSICAL_MEDIUM,
RNDIS_OID_GEN_XMIT_OK,
RNDIS_OID_GEN_RCV_OK,
RNDIS_OID_GEN_XMIT_ERROR,
RNDIS_OID_GEN_RCV_ERROR,
RNDIS_OID_GEN_RCV_NO_BUFFER,
RNDIS_OID_802_3_PERMANENT_ADDRESS,
RNDIS_OID_802_3_CURRENT_ADDRESS,
RNDIS_OID_802_3_MULTICAST_LIST,
RNDIS_OID_802_3_MAC_OPTIONS,
RNDIS_OID_802_3_MAXIMUM_LIST_SIZE,
RNDIS_OID_802_3_RCV_ERROR_ALIGNMENT,
RNDIS_OID_802_3_XMIT_ONE_COLLISION,
RNDIS_OID_802_3_XMIT_MORE_COLLISION,
};
static const struct rndis_oid_value_s g_rndis_oid_values[] =
{
{RNDIS_OID_GEN_SUPPORTED_LIST, sizeof(g_rndis_supported_oids), 0, g_rndis_supported_oids},
This commit attempts remove some long standard confusion in naming and some actual problems that result from the naming confusion. The basic problem is the standard MTU does not include the size of the Ethernet header. For clarity, I changed the naming of most things called MTU to PKTSIZE. For example, CONFIG_NET_ETH_MTU is now CONFIG_NET_ETH_PKTSIZE. This makes the user interface a little hostile. People thing of an MTU of 1500 bytes, but the corresponding packet is really 1514 bytes (including the 14 byte Ethernet header). A more friendly solution would configure the MTU (as before), but then derive the packet buffer size by adding the MAC header length. Instead, we define the packet buffer size then derive the MTU. The MTU is not common currency in networking. On the wire, the only real issue is the MSS which is derived from MTU by subtracting the IP header and TCP header sizes (for the case of TCP). Now it is derived for the PKTSIZE by subtracting the IP header, the TCP header, and the MAC header sizes. So we should be all good and without the recurring 14 byte error in MTU's and MSS's. Squashed commit of the following: Trivial update to fix some spacing issues. net/: Rename several macros containing _MTU to _PKTSIZE. net/: Rename CONFIG_NET_SLIP_MTU to CONFIG_NET_SLIP_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_6LOWPAN_MTU to CONFIG_NET_6LOWPAN_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_ETH_MTU to CONFIG_NET_ETH_PKTSIZE. This is not the MTU which does not include the size of the link layer header. This is the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename the file d_mtu in the network driver structure to d_pktsize. That value saved there is not the MTU. The packetsize is the memory large enough to hold the maximum packet PLUS the size of the link layer header. The MTU does not include the link layer header.
2018-07-04 22:10:40 +02:00
{RNDIS_OID_GEN_MAXIMUM_FRAME_SIZE, 4, CONFIG_NET_ETH_PKTSIZE, NULL},
#ifdef CONFIG_USBDEV_DUALSPEED
{RNDIS_OID_GEN_LINK_SPEED, 4, 100000, NULL},
#else
{RNDIS_OID_GEN_LINK_SPEED, 4, 2000000, NULL},
#endif
This commit attempts remove some long standard confusion in naming and some actual problems that result from the naming confusion. The basic problem is the standard MTU does not include the size of the Ethernet header. For clarity, I changed the naming of most things called MTU to PKTSIZE. For example, CONFIG_NET_ETH_MTU is now CONFIG_NET_ETH_PKTSIZE. This makes the user interface a little hostile. People thing of an MTU of 1500 bytes, but the corresponding packet is really 1514 bytes (including the 14 byte Ethernet header). A more friendly solution would configure the MTU (as before), but then derive the packet buffer size by adding the MAC header length. Instead, we define the packet buffer size then derive the MTU. The MTU is not common currency in networking. On the wire, the only real issue is the MSS which is derived from MTU by subtracting the IP header and TCP header sizes (for the case of TCP). Now it is derived for the PKTSIZE by subtracting the IP header, the TCP header, and the MAC header sizes. So we should be all good and without the recurring 14 byte error in MTU's and MSS's. Squashed commit of the following: Trivial update to fix some spacing issues. net/: Rename several macros containing _MTU to _PKTSIZE. net/: Rename CONFIG_NET_SLIP_MTU to CONFIG_NET_SLIP_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_6LOWPAN_MTU to CONFIG_NET_6LOWPAN_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_ETH_MTU to CONFIG_NET_ETH_PKTSIZE. This is not the MTU which does not include the size of the link layer header. This is the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename the file d_mtu in the network driver structure to d_pktsize. That value saved there is not the MTU. The packetsize is the memory large enough to hold the maximum packet PLUS the size of the link layer header. The MTU does not include the link layer header.
2018-07-04 22:10:40 +02:00
{RNDIS_OID_GEN_TRANSMIT_BLOCK_SIZE, 4, CONFIG_NET_ETH_PKTSIZE, NULL},
{RNDIS_OID_GEN_RECEIVE_BLOCK_SIZE, 4, CONFIG_NET_ETH_PKTSIZE, NULL},
{RNDIS_OID_GEN_VENDOR_ID, 4, 0x00FFFFFF, NULL},
{RNDIS_OID_GEN_VENDOR_DESCRIPTION, 6, 0, "RNDIS"},
{RNDIS_OID_GEN_CURRENT_PACKET_FILTER, 4, 0, NULL},
{RNDIS_OID_GEN_MAXIMUM_TOTAL_SIZE, 4, 2048, NULL},
{RNDIS_OID_GEN_XMIT_OK, 4, 0, NULL},
{RNDIS_OID_GEN_RCV_OK, 4, 0, NULL},
{RNDIS_OID_802_3_PERMANENT_ADDRESS, 6, 0, NULL},
{RNDIS_OID_802_3_CURRENT_ADDRESS, 6, 0, NULL},
{RNDIS_OID_802_3_MULTICAST_LIST, 4, 0xE0000000, NULL},
{RNDIS_OID_802_3_MAXIMUM_LIST_SIZE, 4, 1, NULL},
{0x0, 4, 0, NULL}, /* Default fallback */
};
/****************************************************************************
* Private Data
****************************************************************************/
/****************************************************************************
* Buffering of data is implemented in the following manner:
*
* RNDIS driver holds a number of preallocated bulk IN endpoint write
* requests along with buffers large enough to hold an Ethernet packet and
* the corresponding RNDIS header.
*
* One of these is always reserved for packet reception - when data arrives
* on the bulk OUT endpoint, it is copied to the reserved request buffer.
* When the reception of an Ethernet packet is complete, a worker to process
* the packet is scheduled and bulk OUT endpoint is set to NAK.
*
* The processing worker passes the buffer to the network. When the network is
* done processing the packet, the buffer might contain data to be sent.
* If so, the corresponding write request is queued on the bulk IN endpoint.
* The NAK state on bulk OUT endpoint is cleared to allow new packets to
* arrive. If there's no data to send, the request is returned to the list of
* free requests.
*
* When a bulk IN write operation is complete, the request is added to the
* list of free requests.
*
****************************************************************************/
/****************************************************************************
* Name: rndis_submit_rdreq
*
* Description:
* Submits the bulk OUT read request. Takes care not to submit the request
* when the RX packet buffer is already in use.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Returned Value:
* The return value of the EP_SUBMIT operation
*
****************************************************************************/
static int rndis_submit_rdreq(FAR struct rndis_dev_s *priv)
{
irqstate_t flags = enter_critical_section();
int ret = OK;
if (!priv->rdreq_submitted && !priv->rx_blocked)
{
priv->rdreq->len = priv->epbulkout->maxpacket;
ret = EP_SUBMIT(priv->epbulkout, priv->rdreq);
if (ret != OK)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_RDSUBMIT),
(uint16_t)-priv->rdreq->result);
}
else
{
priv->rdreq_submitted = true;
}
}
leave_critical_section(flags);
return ret;
}
/****************************************************************************
* Name: rndis_cancel_rdreq
*
* Description:
* Cancels the bulk OUT endpoint read request.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
****************************************************************************/
static void rndis_cancel_rdreq(FAR struct rndis_dev_s *priv)
{
irqstate_t flags = enter_critical_section();
if (priv->rdreq_submitted)
{
EP_CANCEL(priv->epbulkout, priv->rdreq);
priv->rdreq_submitted = false;
}
leave_critical_section(flags);
}
/****************************************************************************
* Name: rndis_block_rx
*
* Description:
* Blocks reception of further bulk OUT endpoint data.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
****************************************************************************/
static void rndis_block_rx(FAR struct rndis_dev_s *priv)
{
irqstate_t flags = enter_critical_section();
priv->rx_blocked = true;
rndis_cancel_rdreq(priv);
leave_critical_section(flags);
}
/****************************************************************************
* Name: rndis_unblock_rx
*
* Description:
* Unblocks reception of bulk OUT endpoint data.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static void rndis_unblock_rx(FAR struct rndis_dev_s *priv)
{
priv->rx_blocked = false;
}
/****************************************************************************
* Name: rndis_allocwrreq
*
* Description:
* Allocates a bulk IN endpoint request from the list of free request
* buffers.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Returned Value:
* NULL if allocation failed; pointer to allocated request if succeeded
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static FAR struct rndis_req_s *rndis_allocwrreq(FAR struct rndis_dev_s *priv)
{
return (FAR struct rndis_req_s *)sq_remfirst(&priv->reqlist);
}
/****************************************************************************
* Name: rndis_hasfreereqs
*
* Description:
* Checks if there are free requests usable for TX data.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Returned Value:
* true if requests available; false if no requests available
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static bool rndis_hasfreereqs(FAR struct rndis_dev_s *priv)
{
return sq_count(&priv->reqlist) > 1;
}
/****************************************************************************
* Name: rndis_freewrreq
*
* Description:
* Returns a bulk IN endpoint write requests to the list of free requests.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
* req: pointer to the request
*
* Assumptions:
* Called with interrupts disabled.
*
****************************************************************************/
static void rndis_freewrreq(FAR struct rndis_dev_s *priv,
FAR struct rndis_req_s *req)
{
DEBUGASSERT(req != NULL);
sq_addlast((FAR sq_entry_t *)req, &priv->reqlist);
rndis_submit_rdreq(priv);
}
/****************************************************************************
* Name: rndis_allocnetreq
*
* Description:
* Allocates a request buffer to be used on the network.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Returned Value:
* true if succeeded; false if failed
*
* Assumptions:
* Caller holds the network lock
*
****************************************************************************/
static bool rndis_allocnetreq(FAR struct rndis_dev_s *priv)
{
irqstate_t flags = enter_critical_section();
DEBUGASSERT(priv->net_req == NULL);
if (!rndis_hasfreereqs(priv))
{
leave_critical_section(flags);
return false;
}
priv->net_req = rndis_allocwrreq(priv);
if (priv->net_req)
{
priv->netdev.d_buf = &priv->net_req->req->buf[RNDIS_PACKET_HDR_SIZE];
This commit attempts remove some long standard confusion in naming and some actual problems that result from the naming confusion. The basic problem is the standard MTU does not include the size of the Ethernet header. For clarity, I changed the naming of most things called MTU to PKTSIZE. For example, CONFIG_NET_ETH_MTU is now CONFIG_NET_ETH_PKTSIZE. This makes the user interface a little hostile. People thing of an MTU of 1500 bytes, but the corresponding packet is really 1514 bytes (including the 14 byte Ethernet header). A more friendly solution would configure the MTU (as before), but then derive the packet buffer size by adding the MAC header length. Instead, we define the packet buffer size then derive the MTU. The MTU is not common currency in networking. On the wire, the only real issue is the MSS which is derived from MTU by subtracting the IP header and TCP header sizes (for the case of TCP). Now it is derived for the PKTSIZE by subtracting the IP header, the TCP header, and the MAC header sizes. So we should be all good and without the recurring 14 byte error in MTU's and MSS's. Squashed commit of the following: Trivial update to fix some spacing issues. net/: Rename several macros containing _MTU to _PKTSIZE. net/: Rename CONFIG_NET_SLIP_MTU to CONFIG_NET_SLIP_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_6LOWPAN_MTU to CONFIG_NET_6LOWPAN_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_ETH_MTU to CONFIG_NET_ETH_PKTSIZE. This is not the MTU which does not include the size of the link layer header. This is the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename the file d_mtu in the network driver structure to d_pktsize. That value saved there is not the MTU. The packetsize is the memory large enough to hold the maximum packet PLUS the size of the link layer header. The MTU does not include the link layer header.
2018-07-04 22:10:40 +02:00
priv->netdev.d_len = CONFIG_NET_ETH_PKTSIZE;
}
leave_critical_section(flags);
return priv->net_req != NULL;
}
/****************************************************************************
* Name: rndis_sendnetreq
*
* Description:
* Submits the request buffer held by the network.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Caller holds the network lock
*
****************************************************************************/
static void rndis_sendnetreq(FAR struct rndis_dev_s *priv)
{
irqstate_t flags = enter_critical_section();
DEBUGASSERT(priv->net_req != NULL);
priv->net_req->req->priv = priv->net_req;
EP_SUBMIT(priv->epbulkin, priv->net_req->req);
priv->net_req = NULL;
priv->netdev.d_buf = NULL;
priv->netdev.d_len = 0;
leave_critical_section(flags);
}
/****************************************************************************
* Name: rndis_freenetreq
*
* Description:
* Frees the request buffer held by the network.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Caller holds the network lock
*
****************************************************************************/
static void rndis_freenetreq(FAR struct rndis_dev_s *priv)
{
irqstate_t flags = enter_critical_section();
rndis_freewrreq(priv, priv->net_req);
priv->net_req = NULL;
priv->netdev.d_buf = NULL;
priv->netdev.d_len = 0;
leave_critical_section(flags);
}
/****************************************************************************
* Name: rndis_allocrxreq
*
* Description:
* Allocates a buffer for packet reception if there already isn't one.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Returned Value:
* true if succeeded; false if failed
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static bool rndis_allocrxreq(FAR struct rndis_dev_s *priv)
{
if (priv->rx_req != NULL)
{
return true;
}
priv->rx_req = rndis_allocwrreq(priv);
return priv->rx_req != NULL;
}
/****************************************************************************
* Name: rndis_giverxreq
*
* Description:
* Passes the RX packet buffer to the network
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Caller holds the network lock
*
****************************************************************************/
static void rndis_giverxreq(FAR struct rndis_dev_s *priv)
{
DEBUGASSERT(priv->rx_req != NULL);
DEBUGASSERT(priv->net_req == NULL);
priv->net_req = priv->rx_req;
priv->netdev.d_buf = &priv->net_req->req->buf[RNDIS_PACKET_HDR_SIZE];
This commit attempts remove some long standard confusion in naming and some actual problems that result from the naming confusion. The basic problem is the standard MTU does not include the size of the Ethernet header. For clarity, I changed the naming of most things called MTU to PKTSIZE. For example, CONFIG_NET_ETH_MTU is now CONFIG_NET_ETH_PKTSIZE. This makes the user interface a little hostile. People thing of an MTU of 1500 bytes, but the corresponding packet is really 1514 bytes (including the 14 byte Ethernet header). A more friendly solution would configure the MTU (as before), but then derive the packet buffer size by adding the MAC header length. Instead, we define the packet buffer size then derive the MTU. The MTU is not common currency in networking. On the wire, the only real issue is the MSS which is derived from MTU by subtracting the IP header and TCP header sizes (for the case of TCP). Now it is derived for the PKTSIZE by subtracting the IP header, the TCP header, and the MAC header sizes. So we should be all good and without the recurring 14 byte error in MTU's and MSS's. Squashed commit of the following: Trivial update to fix some spacing issues. net/: Rename several macros containing _MTU to _PKTSIZE. net/: Rename CONFIG_NET_SLIP_MTU to CONFIG_NET_SLIP_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_6LOWPAN_MTU to CONFIG_NET_6LOWPAN_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_ETH_MTU to CONFIG_NET_ETH_PKTSIZE. This is not the MTU which does not include the size of the link layer header. This is the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename the file d_mtu in the network driver structure to d_pktsize. That value saved there is not the MTU. The packetsize is the memory large enough to hold the maximum packet PLUS the size of the link layer header. The MTU does not include the link layer header.
2018-07-04 22:10:40 +02:00
priv->netdev.d_len = CONFIG_NET_ETH_PKTSIZE;
priv->rx_req = NULL;
}
/****************************************************************************
* Name: rndis_fillrequest
*
* Description:
* Fills the RNDIS header to the request buffer
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
* req: the request whose buffer we should fill
*
* Returned Value:
* The total length of the request data
*
* Assumptions:
* Caller holds the network lock
*
****************************************************************************/
static uint16_t rndis_fillrequest(FAR struct rndis_dev_s *priv,
FAR struct usbdev_req_s *req)
{
size_t datalen;
req->len = 0;
datalen = min(priv->netdev.d_len,
CONFIG_RNDIS_BULKIN_REQLEN - RNDIS_PACKET_HDR_SIZE);
if (datalen > 0)
{
/* Send the required headers */
FAR struct rndis_packet_msg *msg = (FAR struct rndis_packet_msg *)req->buf;
memset(msg, 0, RNDIS_PACKET_HDR_SIZE);
msg->msgtype = RNDIS_PACKET_MSG;
msg->msglen = RNDIS_PACKET_HDR_SIZE + datalen;
msg->dataoffset = RNDIS_PACKET_HDR_SIZE - 8;
msg->datalen = datalen;
req->flags = USBDEV_REQFLAGS_NULLPKT;
req->len = datalen + RNDIS_PACKET_HDR_SIZE;
}
return req->len;
}
/****************************************************************************
* Name: rndis_rxdispatch
*
* Description:
* Processes the received Ethernet packet. Called from work queue.
*
* Input Parameters:
* arg: pointer to RNDIS device driver structure
*
****************************************************************************/
static void rndis_rxdispatch(FAR void *arg)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)arg;
FAR struct eth_hdr_s *hdr;
irqstate_t flags;
net_lock();
flags = enter_critical_section();
rndis_giverxreq(priv);
priv->netdev.d_len = priv->current_rx_datagram_size;
leave_critical_section(flags);
hdr = (FAR struct eth_hdr_s *)priv->netdev.d_buf;
/* We only accept IP packets of the configured type and ARP packets */
#ifdef CONFIG_NET_IPv4
if (hdr->type == HTONS(ETHTYPE_IP))
{
NETDEV_RXIPV4(&priv->netdev);
/* Handle ARP on input then give the IPv4 packet to the network
* layer
*/
arp_ipin(&priv->netdev);
ipv4_input(&priv->netdev);
if (priv->netdev.d_len > 0)
{
/* Update the Ethernet header with the correct MAC address */
#ifdef CONFIG_NET_IPv6
if (IFF_IS_IPv4(priv->netdev.d_flags))
#endif
{
arp_out(&priv->netdev);
}
#ifdef CONFIG_NET_IPv6
else
{
neighbor_out(&priv->netdev);
}
#endif
/* And send the packet */
rndis_transmit(priv);
}
}
else
#endif
#ifdef CONFIG_NET_IPv6
if (hdr->type == HTONS(ETHTYPE_IP6))
{
NETDEV_RXIPV6(&priv->netdev);
/* Give the IPv6 packet to the network layer */
arp_ipin(&priv->netdev);
ipv6_input(&priv->netdev);
if (priv->netdev.d_len > 0)
{
/* Update the Ethernet header with the correct MAC address */
#ifdef CONFIG_NET_IPv4
if (IFF_IS_IPv4(priv->netdev.d_flags))
{
arp_out(&priv->netdev);
}
else
#endif
#ifdef CONFIG_NET_IPv6
{
neighbor_out(&priv->netdev);
}
#endif
/* And send the packet */
rndis_transmit(priv);
}
}
else
#endif
#ifdef CONFIG_NET_ARP
if (hdr->type == htons(ETHTYPE_ARP))
{
NETDEV_RXARP(&priv->netdev);
arp_arpin(&priv->netdev);
if (priv->netdev.d_len > 0)
{
rndis_transmit(priv);
}
}
else
#endif
{
uerr("ERROR: Unsupported packet type dropped (%02x)\n", htons(hdr->type));
NETDEV_RXDROPPED(&priv->netdev);
priv->netdev.d_len = 0;
}
priv->current_rx_datagram_size = 0;
rndis_unblock_rx(priv);
if (priv->net_req != NULL)
{
rndis_freenetreq(priv);
}
net_unlock();
}
/****************************************************************************
* Name: rndis_txpoll
*
* Description:
* Sends the packet that is stored in the network packet buffer. Called
* from work queue by e.g. txavail and txpoll callbacks.
*
* Input Parameters:
* dev: pointer to network driver structure
*
* Assumptions:
* Caller holds the network lock
*
****************************************************************************/
static int rndis_txpoll(FAR struct net_driver_s *dev)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)dev->d_private;
int ret = OK;
if (!priv->connected)
{
return -EBUSY;
}
/* If the polling resulted in data that should be sent out on the network,
* the field d_len is set to a value > 0.
*/
ninfo("Poll result: d_len=%d\n", priv->netdev.d_len);
if (priv->netdev.d_len > 0)
{
/* Look up the destination MAC address and add it to the Ethernet
* header.
*/
#ifdef CONFIG_NET_IPv4
#ifdef CONFIG_NET_IPv6
if (IFF_IS_IPv4(priv->netdev.d_flags))
#endif
{
arp_out(&priv->netdev);
}
#endif /* CONFIG_NET_IPv4 */
#ifdef CONFIG_NET_IPv6
#ifdef CONFIG_NET_IPv4
else
#endif
{
neighbor_out(&priv->netdev);
}
#endif /* CONFIG_NET_IPv6 */
if (!devif_loopback(&priv->netdev))
{
ret = rndis_transmit(priv);
}
}
/* If zero is returned, the polling will continue until all connections have
* been examined.
*/
return ret;
}
/****************************************************************************
* Name: rndis_transmit
*
* Description:
* Start hardware transmission.
*
****************************************************************************/
static int rndis_transmit(FAR struct rndis_dev_s *priv)
{
int ret = OK;
/* Queue the packet */
rndis_fillrequest(priv, priv->net_req->req);
rndis_sendnetreq(priv);
if (!rndis_allocnetreq(priv))
{
ret = -EBUSY;
}
return ret;
}
/****************************************************************************
* Name: rndis_pollworker
*
* Description:
* Worker function called by txpoll worker.
*
****************************************************************************/
static void rndis_pollworker(FAR void *arg)
{
FAR struct rndis_dev_s *priv = (struct rndis_dev_s *)arg;
DEBUGASSERT(priv != NULL);
net_lock();
if (rndis_allocnetreq(priv))
{
devif_timer(&priv->netdev, rndis_txpoll);
if (priv->net_req != NULL)
{
rndis_freenetreq(priv);
}
}
net_unlock();
}
/****************************************************************************
* Name: rndis_polltimer
*
* Description:
* Network poll watchdog timer callback
*
****************************************************************************/
static void rndis_polltimer(int argc, uint32_t arg, ...)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)arg;
int ret;
if (work_available(&priv->pollwork))
{
ret = work_queue(ETHWORK, &priv->pollwork, rndis_pollworker,
(FAR void *)priv, 0);
DEBUGASSERT(ret == OK);
UNUSED(ret);
}
/* Setup the watchdog poll timer again */
(void)wd_start(priv->txpoll, RNDIS_WDDELAY, rndis_polltimer, 1,
(wdparm_t)arg);
}
/****************************************************************************
* Name: rndis_ifup
*
* Description:
* Network ifup callback
*
****************************************************************************/
static int rndis_ifup(FAR struct net_driver_s *dev)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)dev->d_private;
(void)wd_start(priv->txpoll, RNDIS_WDDELAY, rndis_polltimer,
1, (wdparm_t)priv);
return OK;
}
/****************************************************************************
* Name: rndis_ifdown
*
* Description:
* Network ifdown callback
*
****************************************************************************/
static int rndis_ifdown(FAR struct net_driver_s *dev)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)dev->d_private;
wd_cancel(priv->txpoll);
return OK;
}
/****************************************************************************
* Name: rndis_txavail_work
*
* Description:
* txavail worker function
*
****************************************************************************/
static void rndis_txavail_work(FAR void *arg)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)arg;
net_lock();
if (rndis_allocnetreq(priv))
{
devif_poll(&priv->netdev, rndis_txpoll);
if (priv->net_req != NULL)
{
rndis_freenetreq(priv);
}
}
net_unlock();
}
/****************************************************************************
* Name: rndis_txavail
*
* Description:
* Network txavail callback that's called when there are buffers available
* for sending data. May be called from an interrupt, so we must queue a
* worker to do the actual processing.
*
****************************************************************************/
static int rndis_txavail(FAR struct net_driver_s *dev)
{
FAR struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)dev->d_private;
if (work_available(&priv->pollwork))
{
work_queue(ETHWORK, &priv->pollwork, rndis_txavail_work, priv, 0);
}
return OK;
}
/************************************************************************************
* Name: rndis_recvpacket
*
* Description:
* Handles a USB packet arriving on the data bulk out endpoint.
*
* Assumptions:
* Called from the USB interrupt handler with interrupts disabled.
*
************************************************************************************/
static inline int rndis_recvpacket(FAR struct rndis_dev_s *priv,
FAR uint8_t *reqbuf, uint16_t reqlen)
{
if (!rndis_allocrxreq(priv))
{
return -ENOMEM;
}
if (!priv->connected)
{
return -EBUSY;
}
if (!priv->current_rx_datagram_size)
{
if (reqlen < 16)
{
/* Packet too small to contain a message header */
}
else
{
/* The packet contains a RNDIS packet message header */
FAR struct rndis_packet_msg *msg = (FAR struct rndis_packet_msg *)reqbuf;
if (msg->msgtype == RNDIS_PACKET_MSG)
{
priv->current_rx_received = reqlen;
priv->current_rx_datagram_size = msg->datalen;
priv->current_rx_msglen = msg->msglen;
/* According to RNDIS-over-USB send, if the message length is a
* multiple of endpoint max packet size, the host must send an
* additional single-byte zero packet. Take that in account here.
*/
if ((priv->current_rx_msglen % priv->epbulkout->maxpacket) == 0)
{
priv->current_rx_msglen += 1;
}
/* Data offset is defined as an offset from the beginning of the
* offset field itself
*/
priv->current_rx_datagram_offset = msg->dataoffset + 8;
if (priv->current_rx_datagram_offset < reqlen)
{
memcpy(&priv->rx_req->req->buf[RNDIS_PACKET_HDR_SIZE],
&reqbuf[priv->current_rx_datagram_offset],
reqlen - priv->current_rx_datagram_offset);
}
}
else
{
uerr("Unknown RNDIS message type %u\n", msg->msgtype);
}
}
}
else
{
if (priv->current_rx_received >= priv->current_rx_datagram_offset &&
priv->current_rx_received <= priv->current_rx_datagram_size +
priv->current_rx_datagram_offset)
{
size_t index = priv->current_rx_received - priv->current_rx_datagram_offset;
size_t copysize = min(reqlen, priv->current_rx_datagram_size - index);
/* Check if the received packet exceeds request buffer */
if ((index + copysize) <= CONFIG_NET_ETH_PKTSIZE)
{
memcpy(&priv->rx_req->req->buf[RNDIS_PACKET_HDR_SIZE + index], reqbuf,
copysize);
}
else
{
uerr("The packet exceeds request buffer (reqlen=%d) \n", reqlen);
}
}
priv->current_rx_received += reqlen;
}
if (priv->current_rx_received >= priv->current_rx_msglen)
{
/* Check for a usable packet length (4 added for the CRC) */
This commit attempts remove some long standard confusion in naming and some actual problems that result from the naming confusion. The basic problem is the standard MTU does not include the size of the Ethernet header. For clarity, I changed the naming of most things called MTU to PKTSIZE. For example, CONFIG_NET_ETH_MTU is now CONFIG_NET_ETH_PKTSIZE. This makes the user interface a little hostile. People thing of an MTU of 1500 bytes, but the corresponding packet is really 1514 bytes (including the 14 byte Ethernet header). A more friendly solution would configure the MTU (as before), but then derive the packet buffer size by adding the MAC header length. Instead, we define the packet buffer size then derive the MTU. The MTU is not common currency in networking. On the wire, the only real issue is the MSS which is derived from MTU by subtracting the IP header and TCP header sizes (for the case of TCP). Now it is derived for the PKTSIZE by subtracting the IP header, the TCP header, and the MAC header sizes. So we should be all good and without the recurring 14 byte error in MTU's and MSS's. Squashed commit of the following: Trivial update to fix some spacing issues. net/: Rename several macros containing _MTU to _PKTSIZE. net/: Rename CONFIG_NET_SLIP_MTU to CONFIG_NET_SLIP_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_6LOWPAN_MTU to CONFIG_NET_6LOWPAN_PKTSIZE and similarly for CONFIG_NET_TUN_MTU. These are not the MTU which does not include the size of the link layer header. These are the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename CONFIG_NET_ETH_MTU to CONFIG_NET_ETH_PKTSIZE. This is not the MTU which does not include the size of the link layer header. This is the full size of the packet buffer memory (minus any GUARD bytes). net/: Rename the file d_mtu in the network driver structure to d_pktsize. That value saved there is not the MTU. The packetsize is the memory large enough to hold the maximum packet PLUS the size of the link layer header. The MTU does not include the link layer header.
2018-07-04 22:10:40 +02:00
if (priv->current_rx_datagram_size > (CONFIG_NET_ETH_PKTSIZE + 4) ||
priv->current_rx_datagram_size <= (ETH_HDRLEN + 4))
{
uerr("ERROR: Bad packet size dropped (%d)\n",
priv->current_rx_datagram_size);
NETDEV_RXERRORS(&priv->netdev);
priv->current_rx_datagram_size = 0;
}
else
{
int ret;
DEBUGASSERT(work_available(&priv->rxwork));
ret = work_queue(ETHWORK, &priv->rxwork, rndis_rxdispatch,
priv, 0);
DEBUGASSERT(ret == 0);
UNUSED(ret);
rndis_block_rx(priv);
priv->rndis_host_tx_count++;
return -EBUSY;
}
}
return OK;
}
/****************************************************************************
* Name: rndis_prepare_response
*
* Description:
* Passes the RX packet buffer to the network
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static bool rndis_prepare_response(FAR struct rndis_dev_s *priv, size_t size,
FAR struct rndis_command_header *request_hdr)
{
FAR struct rndis_response_header *hdr =
(FAR struct rndis_response_header *)priv->ctrlreq->buf;
hdr->msgtype = request_hdr->msgtype | RNDIS_MSG_COMPLETE;
hdr->msglen = size;
hdr->reqid = request_hdr->reqid;
hdr->status = RNDIS_STATUS_SUCCESS;
priv->ctrlreq_has_encap_response = true;
return true;
}
/****************************************************************************
* Name: rndis_send_encapsulated_response
*
* Description:
* Give a notification to the host that there is an encapsulated response
* available.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static int rndis_send_encapsulated_response(FAR struct rndis_dev_s *priv)
{
FAR struct rndis_notification *notif =
(FAR struct rndis_notification *)priv->epintin_req->buf;
notif->notification = RNDIS_NOTIFICATION_RESPONSE_AVAILABLE;
notif->reserved = 0;
priv->epintin_req->len = sizeof(struct rndis_notification);
EP_CANCEL(priv->epintin, priv->epintin_req);
EP_SUBMIT(priv->epintin, priv->epintin_req);
return OK;
}
/****************************************************************************
* Name: rndis_handle_control_message
*
* Description:
* Handle a RNDIS control message.
*
* Input Parameters:
* priv: pointer to RNDIS device driver structure
*
* Assumptions:
* Called from critical section
*
****************************************************************************/
static int rndis_handle_control_message(FAR struct rndis_dev_s *priv,
FAR uint8_t *dataout, uint16_t outlen)
{
FAR struct rndis_command_header *cmd_hdr =
(FAR struct rndis_command_header *)dataout;
switch (cmd_hdr->msgtype)
{
case RNDIS_INITIALIZE_MSG:
{
FAR struct rndis_initialize_cmplt *resp;
rndis_prepare_response(priv, sizeof(struct rndis_initialize_cmplt), cmd_hdr);
resp = (FAR struct rndis_initialize_cmplt *)priv->ctrlreq->buf;
resp->major = RNDIS_MAJOR_VERSION;
resp->minor = RNDIS_MINOR_VERSION;
resp->devflags = RNDIS_DEVICEFLAGS;
resp->medium = RNDIS_MEDIUM_802_3;
resp->pktperxfer = 1;
resp->xfrsize = (4 + 44 + 22) + RNDIS_BUFFER_SIZE;
resp->pktalign = 2;
rndis_send_encapsulated_response(priv);
}
break;
case RNDIS_HALT_MSG:
{
priv->connected = false;
}
break;
case RNDIS_QUERY_MSG:
{
int i;
size_t max_reply_size = sizeof(struct rndis_query_cmplt) +
sizeof(g_rndis_supported_oids);
rndis_prepare_response(priv, max_reply_size, cmd_hdr);
FAR struct rndis_query_msg *req =
(FAR struct rndis_query_msg *)dataout;
FAR struct rndis_query_cmplt *resp =
(FAR struct rndis_query_cmplt *)priv->ctrlreq->buf;
resp->hdr.msglen = sizeof(struct rndis_query_cmplt);
resp->bufoffset = 0;
resp->buflen = 0;
resp->hdr.status = RNDIS_STATUS_NOT_SUPPORTED;
for (i = 0;
i < sizeof(g_rndis_oid_values)/sizeof(g_rndis_oid_values[0]);
i++)
{
bool match = (g_rndis_oid_values[i].objid == req->objid);
if (!match && g_rndis_oid_values[i].objid == 0)
{
int j;
/* Check whether to apply the fallback entry */
for (j = 0; j < sizeof(g_rndis_supported_oids)/sizeof(uint32_t); j++)
{
if (g_rndis_supported_oids[j] == req->objid)
{
match = true;
break;
}
}
}
if (match)
{
resp->hdr.status = RNDIS_STATUS_SUCCESS;
resp->bufoffset = 16;
resp->buflen = g_rndis_oid_values[i].length;
if (req->objid == RNDIS_OID_GEN_CURRENT_PACKET_FILTER)
{
resp->buffer[0] = priv->rndis_packet_filter;
}
else if (req->objid == RNDIS_OID_GEN_XMIT_OK)
{
resp->buffer[0] = priv->rndis_host_tx_count;
}
else if (req->objid == RNDIS_OID_GEN_RCV_OK)
{
resp->buffer[0] = priv->rndis_host_rx_count;
}
else if (req->objid == RNDIS_OID_802_3_CURRENT_ADDRESS ||
req->objid == RNDIS_OID_802_3_PERMANENT_ADDRESS)
{
memcpy(resp->buffer, priv->host_mac_address, 6);
}
else if (g_rndis_oid_values[i].data)
{
memcpy(resp->buffer, g_rndis_oid_values[i].data,
resp->buflen);
}
else
{
memcpy(resp->buffer, &g_rndis_oid_values[i].value,
resp->buflen);
}
break;
}
}
uinfo("RNDIS Query RID=%08x OID=%08x LEN=%d DAT=%08x",
(unsigned)req->hdr.reqid, (unsigned)req->objid,
(int)resp->buflen, (unsigned)resp->buffer[0]);
resp->hdr.msglen += resp->buflen;
rndis_send_encapsulated_response(priv);
}
break;
case RNDIS_SET_MSG:
{
FAR struct rndis_set_msg *req;
FAR struct rndis_response_header *resp;
rndis_prepare_response(priv, sizeof(struct rndis_response_header),
cmd_hdr);
req = (FAR struct rndis_set_msg *)dataout;
resp = (FAR struct rndis_response_header *)priv->ctrlreq->buf;
uinfo("RNDIS SET RID=%08x OID=%08x LEN=%d DAT=%08x",
(unsigned)req->hdr.reqid, (unsigned)req->objid,
(int)req->buflen, (unsigned)req->buffer[0]);
if (req->objid == RNDIS_OID_GEN_CURRENT_PACKET_FILTER)
{
priv->rndis_packet_filter = req->buffer[0];
if (req->buffer[0] == 0)
{
priv->connected = false;
}
else
{
uinfo("RNDIS is now connected");
priv->connected = true;
}
}
else if (req->objid == RNDIS_OID_802_3_MULTICAST_LIST)
{
uinfo("RNDIS multicast list ignored");
}
else
{
uinfo("RNDIS unsupported set %08x", (unsigned)req->objid);
resp->status = RNDIS_STATUS_NOT_SUPPORTED;
}
rndis_send_encapsulated_response(priv);
}
break;
case RNDIS_RESET_MSG:
{
FAR struct rndis_reset_cmplt *resp;
rndis_prepare_response(priv, sizeof(struct rndis_reset_cmplt),
cmd_hdr);
resp = (FAR struct rndis_reset_cmplt *)priv->ctrlreq->buf;
resp->addreset = 0;
priv->connected = false;
rndis_send_encapsulated_response(priv);
}
break;
case RNDIS_KEEPALIVE_MSG:
{
rndis_prepare_response(priv, sizeof(struct rndis_response_header),
cmd_hdr);
rndis_send_encapsulated_response(priv);
}
break;
default:
uwarn("Unsupported RNDIS control message: %u\n", cmd_hdr->msgtype);
}
return OK;
}
/****************************************************************************
* Name: rndis_rdcomplete
*
* Description:
* Handle completion of read request on the bulk OUT endpoint.
*
****************************************************************************/
static void rndis_rdcomplete(FAR struct usbdev_ep_s *ep,
FAR struct usbdev_req_s *req)
{
FAR struct rndis_dev_s *priv;
irqstate_t flags;
int ret;
/* Sanity check */
#ifdef CONFIG_DEBUG_FEATURES
if (!ep || !ep->priv || !req)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_INVALIDARG), 0);
return;
}
#endif
/* Extract references to private data */
priv = (FAR struct rndis_dev_s *)ep->priv;
/* Process the received data unless this is some unusual condition */
ret = OK;
flags = enter_critical_section();
priv->rdreq_submitted = false;
switch (req->result)
{
case 0: /* Normal completion */
ret = rndis_recvpacket(priv, req->buf, req->xfrd);
DEBUGASSERT(ret != -ENOMEM);
break;
case -ESHUTDOWN: /* Disconnection */
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_RDSHUTDOWN), 0);
leave_critical_section(flags);
return;
default: /* Some other error occurred */
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_RDUNEXPECTED), (uint16_t)-req->result);
break;
};
if (ret == OK)
{
rndis_submit_rdreq(priv);
}
leave_critical_section(flags);
}
/****************************************************************************
* Name: rndis_wrcomplete
*
* Description:
* Handle completion of write request. This function probably executes
* in the context of an interrupt handler.
*
****************************************************************************/
static void rndis_wrcomplete(FAR struct usbdev_ep_s *ep,
FAR struct usbdev_req_s *req)
{
FAR struct rndis_dev_s *priv;
FAR struct rndis_req_s *reqcontainer;
irqstate_t flags;
/* Sanity check */
#ifdef CONFIG_DEBUG_FEATURES
if (!ep || !ep->priv || !req || !req->priv)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_INVALIDARG), 0);
return;
}
#endif
/* Extract references to our private data */
priv = (FAR struct rndis_dev_s *)ep->priv;
reqcontainer = (FAR struct rndis_req_s *)req->priv;
/* Return the write request to the free list */
flags = enter_critical_section();
rndis_freewrreq(priv, reqcontainer);
if (rndis_hasfreereqs(priv))
{
rndis_txavail(&priv->netdev);
}
switch (req->result)
{
case OK: /* Normal completion */
priv->rndis_host_rx_count++;
break;
case -ESHUTDOWN: /* Disconnection */
break;
default: /* Some other error occurred */
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_WRUNEXPECTED),
(uint16_t)-req->result);
break;
}
leave_critical_section(flags);
}
/****************************************************************************
* Name: usbclass_ep0incomplete
*
* Description:
* Handle completion of EP0 control operations
*
****************************************************************************/
static void usbclass_ep0incomplete(FAR struct usbdev_ep_s *ep,
FAR struct usbdev_req_s *req)
{
if (req->result || req->xfrd != req->len)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_REQRESULT),
(uint16_t)-req->result);
}
else if (req->len > 0)
{
struct rndis_dev_s *priv = (FAR struct rndis_dev_s *)ep->priv;
priv->ctrlreq_has_encap_response = false;
}
}
/****************************************************************************
* Name: usbclass_ep0incomplete
*
* Description:
* Handle completion of interrupt IN endpoint operations
*
****************************************************************************/
static void usbclass_epintin_complete(FAR struct usbdev_ep_s *ep,
FAR struct usbdev_req_s *req)
{
if (req->result || req->xfrd != req->len)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_REQRESULT),
(uint16_t)-req->result);
}
}
/****************************************************************************
* Name: usbclass_freereq
*
* Description:
* Free a request instance along with its buffer
*
****************************************************************************/
static void usbclass_freereq(FAR struct usbdev_ep_s *ep,
FAR struct usbdev_req_s *req)
{
if (ep != NULL && req != NULL)
{
if (req->buf != NULL)
{
EP_FREEBUFFER(ep, req->buf);
}
EP_FREEREQ(ep, req);
}
}
/****************************************************************************
* Name: usbclass_allocreq
*
* Description:
* Allocate a request instance along with its buffer
*
****************************************************************************/
static FAR struct usbdev_req_s *usbclass_allocreq(FAR struct usbdev_ep_s *ep,
uint16_t len)
{
FAR struct usbdev_req_s *req;
req = EP_ALLOCREQ(ep);
if (req != NULL)
{
req->len = len;
req->buf = EP_ALLOCBUFFER(ep, len);
if (req->buf == NULL)
{
EP_FREEREQ(ep, req);
req = NULL;
}
}
return req;
}
/****************************************************************************
* Name: usbclass_mkstrdesc
*
* Description:
* Construct a string descriptor
*
****************************************************************************/
static int usbclass_mkstrdesc(uint8_t id, FAR struct usb_strdesc_s *strdesc)
{
FAR const char *str;
int len;
int ndata;
int i;
switch (id)
{
#ifndef CONFIG_RNDIS_COMPOSITE
case 0:
{
/* Descriptor 0 is the language id */
strdesc->len = 4;
strdesc->type = USB_DESC_TYPE_STRING;
strdesc->data[0] = LSBYTE(RNDIS_STR_LANGUAGE);
strdesc->data[1] = MSBYTE(RNDIS_STR_LANGUAGE);
return 4;
}
case RNDIS_MANUFACTURERSTRID:
str = CONFIG_RNDIS_VENDORSTR;
break;
case RNDIS_PRODUCTSTRID:
str = CONFIG_RNDIS_PRODUCTSTR;
break;
case RNDIS_SERIALSTRID:
str = CONFIG_RNDIS_SERIALSTR;
break;
#endif
default:
return -EINVAL;
}
/* The string is utf16-le. The poor man's utf-8 to utf16-le
* conversion below will only handle 7-bit en-us ascii
*/
len = strlen(str);
if (len > (RNDIS_MAXSTRLEN / 2))
{
len = (RNDIS_MAXSTRLEN / 2);
}
for (i = 0, ndata = 0; i < len; i++, ndata += 2)
{
strdesc->data[ndata] = str[i];
strdesc->data[ndata+1] = 0;
}
strdesc->len = ndata+2;
strdesc->type = USB_DESC_TYPE_STRING;
return strdesc->len;
}
/****************************************************************************
* Name: usbclass_copy_epdesc
*
* Description:
* Copies the requested Endpoint Description into the buffer given.
* Returns the number of Bytes filled in ( sizeof(struct usb_epdesc_s) ).
*
****************************************************************************/
static int usbclass_copy_epdesc(int epid, FAR struct usb_epdesc_s *epdesc,
FAR struct usbdev_devinfo_s *devinfo,
bool hispeed)
{
#ifndef CONFIG_USBDEV_DUALSPEED
UNUSED(hispeed);
#endif
switch (epid)
{
case RNDIS_EP_INTIN_IDX: /* Interrupt IN endpoint */
{
epdesc->len = USB_SIZEOF_EPDESC; /* Descriptor length */
epdesc->type = USB_DESC_TYPE_ENDPOINT; /* Descriptor type */
epdesc->addr = RNDIS_MKEPINTIN(devinfo); /* Endpoint address */
epdesc->attr = RNDIS_EPINTIN_ATTR; /* Endpoint attributes */
#ifdef CONFIG_USBDEV_DUALSPEED
if (hispeed)
{
/* Maximum packet size (high speed) */
epdesc->mxpacketsize[0] = LSBYTE(CONFIG_RNDIS_EPINTIN_HSSIZE);
epdesc->mxpacketsize[1] = MSBYTE(CONFIG_RNDIS_EPINTIN_HSSIZE);
}
else
#endif
{
/* Maximum packet size (full speed) */
epdesc->mxpacketsize[0] = LSBYTE(CONFIG_RNDIS_EPINTIN_FSSIZE);
epdesc->mxpacketsize[1] = MSBYTE(CONFIG_RNDIS_EPINTIN_FSSIZE);
}
epdesc->interval = 10; /* Interval */
}
break;
case RNDIS_EP_BULKOUT_IDX: /* Bulk OUT endpoint */
{
epdesc->len = USB_SIZEOF_EPDESC; /* Descriptor length */
epdesc->type = USB_DESC_TYPE_ENDPOINT; /* Descriptor type */
epdesc->addr = RNDIS_MKEPBULKOUT(devinfo); /* Endpoint address */
epdesc->attr = RNDIS_EPOUTBULK_ATTR; /* Endpoint attributes */
#ifdef CONFIG_USBDEV_DUALSPEED
if (hispeed)
{
/* Maximum packet size (high speed) */
epdesc->mxpacketsize[0] = LSBYTE(CONFIG_RNDIS_EPBULKOUT_HSSIZE);
epdesc->mxpacketsize[1] = MSBYTE(CONFIG_RNDIS_EPBULKOUT_HSSIZE);
}
else
#endif
{
/* Maximum packet size (full speed) */
epdesc->mxpacketsize[0] = LSBYTE(CONFIG_RNDIS_EPBULKOUT_FSSIZE);
epdesc->mxpacketsize[1] = MSBYTE(CONFIG_RNDIS_EPBULKOUT_FSSIZE);
}
epdesc->interval = 0; /* Interval */
}
break;
case RNDIS_EP_BULKIN_IDX: /* Bulk IN endpoint */
{
epdesc->len = USB_SIZEOF_EPDESC; /* Descriptor length */
epdesc->type = USB_DESC_TYPE_ENDPOINT; /* Descriptor type */
epdesc->addr = RNDIS_MKEPBULKIN(devinfo); /* Endpoint address */
epdesc->attr = RNDIS_EPINBULK_ATTR; /* Endpoint attributes */
#ifdef CONFIG_USBDEV_DUALSPEED
if (hispeed)
{
/* Maximum packet size (high speed) */
epdesc->mxpacketsize[0] = LSBYTE(CONFIG_RNDIS_EPBULKIN_HSSIZE);
epdesc->mxpacketsize[1] = MSBYTE(CONFIG_RNDIS_EPBULKIN_HSSIZE);
}
else
#endif
{
/* Maximum packet size (full speed) */
epdesc->mxpacketsize[0] = LSBYTE(CONFIG_RNDIS_EPBULKIN_FSSIZE);
epdesc->mxpacketsize[1] = MSBYTE(CONFIG_RNDIS_EPBULKIN_FSSIZE);
}
epdesc->interval = 0; /* Interval */
}
break;
default:
return 0;
}
return sizeof(struct usb_epdesc_s);
}
/****************************************************************************
* Name: usbclass_mkcfgdesc
*
* Description:
* Construct the configuration descriptor
*
****************************************************************************/
static int16_t usbclass_mkcfgdesc(FAR uint8_t *buf,
FAR struct usbdev_devinfo_s *devinfo)
{
FAR struct rndis_cfgdesc_s *dest = (FAR struct rndis_cfgdesc_s *)buf;
uint16_t totallen;
/* This is the total length of the configuration (not necessarily the
* size that we will be sending now).
*/
totallen = sizeof(g_rndis_cfgdesc);
memcpy(dest, &g_rndis_cfgdesc, totallen);
#ifndef CONFIG_RNDIS_COMPOSITE
/* For a stand-alone device, just fill in the total length */
dest->cfgdesc.totallen[0] = LSBYTE(totallen);
dest->cfgdesc.totallen[1] = MSBYTE(totallen);
#else
/* For composite device, apply possible offset to the interface numbers */
dest->assoc_desc.firstif += devinfo->ifnobase;
dest->comm_ifdesc.ifno += devinfo->ifnobase;
dest->epintindesc.addr = USB_EPIN(devinfo->epno[RNDIS_EP_INTIN_IDX]);
dest->data_ifdesc.ifno += devinfo->ifnobase;
dest->epbulkindesc.addr = USB_EPIN(devinfo->epno[RNDIS_EP_BULKIN_IDX]);
dest->epbulkoutdesc.addr = USB_EPOUT(devinfo->epno[RNDIS_EP_BULKOUT_IDX]);
#endif
return totallen;
}
/****************************************************************************
* Name: usbclass_bind
*
* Description:
* Invoked when the driver is bound to a USB device driver
*
****************************************************************************/
static int usbclass_bind(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev)
{
FAR struct rndis_dev_s *priv = ((FAR struct rndis_driver_s *)driver)->dev;
FAR struct rndis_req_s *reqcontainer;
irqstate_t flags;
uint16_t reqlen;
int ret;
int i;
usbtrace(TRACE_CLASSBIND, 0);
/* Bind the structures */
priv->usbdev = dev;
/* Save the reference to our private data structure in EP0 so that it
* can be recovered in ep0 completion events (Unless we are part of
* a composite device and, in that case, the composite device owns
* EP0).
*/
dev->ep0->priv = priv;
/* Preallocate control request */
priv->ctrlreq = usbclass_allocreq(dev->ep0, RNDIS_CTRLREQ_LEN);
if (priv->ctrlreq == NULL)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_ALLOCCTRLREQ), 0);
ret = -ENOMEM;
goto errout;
}
priv->ctrlreq->callback = usbclass_ep0incomplete;
/* Pre-allocate all endpoints... the endpoints will not be functional
* until the SET CONFIGURATION request is processed in usbclass_setconfig.
* This is done here because there may be calls to kmm_malloc and the SET
* CONFIGURATION processing probably occurs within interrupt handling
* logic where kmm_malloc calls will fail.
*/
/* Pre-allocate the IN interrupt endpoint */
priv->epintin = DEV_ALLOCEP(dev, USB_EPIN(priv->devinfo.epno[RNDIS_EP_INTIN_IDX]),
true, USB_EP_ATTR_XFER_INT);
if (!priv->epintin)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPINTINALLOCFAIL), 0);
ret = -ENODEV;
goto errout;
}
priv->epintin->priv = priv;
priv->epintin_req = usbclass_allocreq(priv->epintin, sizeof(struct rndis_notification));
if (priv->epintin_req == NULL)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_RDALLOCREQ), -ENOMEM);
ret = -ENOMEM;
goto errout;
}
priv->epintin_req->callback = usbclass_epintin_complete;
/* Pre-allocate the IN bulk endpoint */
priv->epbulkin = DEV_ALLOCEP(dev, USB_EPIN(priv->devinfo.epno[RNDIS_EP_BULKIN_IDX]),
true, USB_EP_ATTR_XFER_BULK);
if (!priv->epbulkin)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPBULKINALLOCFAIL), 0);
ret = -ENODEV;
goto errout;
}
priv->epbulkin->priv = priv;
/* Pre-allocate the OUT bulk endpoint */
priv->epbulkout = DEV_ALLOCEP(dev, USB_EPOUT(priv->devinfo.epno[RNDIS_EP_BULKOUT_IDX]),
false, USB_EP_ATTR_XFER_BULK);
if (!priv->epbulkout)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPBULKOUTALLOCFAIL), 0);
ret = -ENODEV;
goto errout;
}
priv->epbulkout->priv = priv;
/* Pre-allocate read requests. The buffer size is one full packet. */
reqlen = 64;
if (CONFIG_RNDIS_BULKOUT_REQLEN > reqlen)
{
reqlen = CONFIG_RNDIS_BULKOUT_REQLEN;
}
priv->rdreq = usbclass_allocreq(priv->epbulkout, reqlen);
if (priv->rdreq == NULL)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_RDALLOCREQ), -ENOMEM);
ret = -ENOMEM;
goto errout;
}
priv->rdreq->callback = rndis_rdcomplete;
/* Pre-allocate write request containers and put in a free list.
* The buffer size should be larger than a full packet. Otherwise,
* we will send a bogus null packet at the end of each packet.
*
* Pick the larger of the max packet size and the configured request
* size.
*/
reqlen = 64;
if (CONFIG_RNDIS_BULKIN_REQLEN > reqlen)
{
reqlen = CONFIG_RNDIS_BULKIN_REQLEN;
}
for (i = 0; i < CONFIG_RNDIS_NWRREQS; i++)
{
reqcontainer = &priv->wrreqs[i];
reqcontainer->req = usbclass_allocreq(priv->epbulkin, reqlen);
if (reqcontainer->req == NULL)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_WRALLOCREQ), -ENOMEM);
ret = -ENOMEM;
goto errout;
}
reqcontainer->req->priv = reqcontainer;
reqcontainer->req->callback = rndis_wrcomplete;
flags = enter_critical_section();
sq_addlast((FAR sq_entry_t *)reqcontainer, &priv->reqlist);
leave_critical_section(flags);
}
/* Report if we are selfpowered */
#ifndef CONFIG_RNDIS_COMPOSITE
#ifdef CONFIG_USBDEV_SELFPOWERED
DEV_SETSELFPOWERED(dev);
#endif
/* And pull-up the data line for the soft connect function */
DEV_CONNECT(dev);
#endif
return OK;
errout:
usbclass_unbind(driver, dev);
return ret;
}
/****************************************************************************
* Name: usbclass_unbind
*
* Description:
* Invoked when the driver is unbound from a USB device driver
*
****************************************************************************/
static void usbclass_unbind(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev)
{
FAR struct rndis_dev_s *priv;
FAR struct rndis_req_s *reqcontainer;
irqstate_t flags;
usbtrace(TRACE_CLASSUNBIND, 0);
#ifdef CONFIG_DEBUG_FEATURES
if (!driver || !dev || !dev->ep0)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_INVALIDARG), 0);
return;
}
#endif
/* Extract reference to private data */
priv = ((FAR struct rndis_driver_s *)driver)->dev;
#ifdef CONFIG_DEBUG_FEATURES
if (!priv)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EP0NOTBOUND), 0);
return;
}
#endif
/* Make sure that we are not already unbound */
if (priv != NULL)
{
/* Make sure that the endpoints have been unconfigured. If
* we were terminated gracefully, then the configuration should
* already have been reset. If not, then calling usbclass_resetconfig
* should cause the endpoints to immediately terminate all
* transfers and return the requests to us (with result == -ESHUTDOWN)
*/
usbclass_resetconfig(priv);
up_mdelay(50);
/* Free the interrupt IN endpoint */
if (priv->epintin)
{
DEV_FREEEP(dev, priv->epintin);
priv->epintin = NULL;
}
/* Free the bulk IN endpoint */
if (priv->epbulkin)
{
DEV_FREEEP(dev, priv->epbulkin);
priv->epbulkin = NULL;
}
/* Free the pre-allocated control request */
if (priv->ctrlreq != NULL)
{
usbclass_freereq(dev->ep0, priv->ctrlreq);
priv->ctrlreq = NULL;
}
if (priv->epintin_req != NULL)
{
usbclass_freereq(priv->epintin, priv->epintin_req);
priv->epintin_req = NULL;
}
/* Free pre-allocated read requests (which should all have
* been returned to the free list at this time -- we don't check)
*/
if (priv->rdreq)
{
usbclass_freereq(priv->epbulkout, priv->rdreq);
}
/* Free the bulk OUT endpoint */
if (priv->epbulkout)
{
DEV_FREEEP(dev, priv->epbulkout);
priv->epbulkout = NULL;
}
netdev_unregister(&priv->netdev);
/* Free write requests that are not in use (which should be all
* of them
*/
flags = enter_critical_section();
while (!sq_empty(&priv->reqlist))
{
reqcontainer = (struct rndis_req_s *)sq_remfirst(&priv->reqlist);
if (reqcontainer->req != NULL)
{
usbclass_freereq(priv->epbulkin, reqcontainer->req);
}
}
leave_critical_section(flags);
}
}
/****************************************************************************
* Name: usbclass_setup
*
* Description:
* Invoked for ep0 control requests. This function probably executes
* in the context of an interrupt handler.
*
****************************************************************************/
static int usbclass_setup(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev,
FAR const struct usb_ctrlreq_s *ctrl,
FAR uint8_t *dataout, size_t outlen)
{
FAR struct rndis_dev_s *priv;
FAR struct usbdev_req_s *ctrlreq;
uint16_t value;
uint16_t len;
int ret = -EOPNOTSUPP;
#ifdef CONFIG_DEBUG_FEATURES
if (!driver || !dev || !dev->ep0 || !ctrl)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_INVALIDARG), 0);
return -EIO;
}
#endif
/* Extract reference to private data */
usbtrace(TRACE_CLASSSETUP, ctrl->req);
priv = ((FAR struct rndis_driver_s *)driver)->dev;
#ifdef CONFIG_DEBUG_FEATURES
if (!priv || !priv->ctrlreq)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EP0NOTBOUND), 0);
return -ENODEV;
}
#endif
ctrlreq = priv->ctrlreq;
/* Extract the little-endian 16-bit values to host order */
value = GETUINT16(ctrl->value);
len = GETUINT16(ctrl->len);
uinfo("type=%02x req=%02x value=%04x len=%04x\n",
ctrl->type, ctrl->req, value, len);
switch (ctrl->type & USB_REQ_TYPE_MASK)
{
/***********************************************************************
* Standard Requests
***********************************************************************/
case USB_REQ_TYPE_STANDARD:
{
switch (ctrl->req)
{
case USB_REQ_GETDESCRIPTOR:
{
/* The value field specifies the descriptor type in the MS byte and the
* descriptor index in the LS byte (order is little endian)
*/
switch (ctrl->value[1])
{
#ifndef CONFIG_RNDIS_COMPOSITE
case USB_DESC_TYPE_DEVICE:
{
ret = USB_SIZEOF_DEVDESC;
memcpy(ctrlreq->buf, &g_devdesc, ret);
}
break;
#endif
case USB_DESC_TYPE_CONFIG:
{
ret = usbclass_mkcfgdesc(ctrlreq->buf, &priv->devinfo);
}
break;
case USB_DESC_TYPE_STRING:
{
/* index == language code. */
ret = usbclass_mkstrdesc(ctrl->value[0],
(FAR struct usb_strdesc_s *)ctrlreq->buf);
}
break;
default:
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_GETUNKNOWNDESC), value);
}
break;
}
}
break;
case USB_REQ_SETCONFIGURATION:
{
if (ctrl->type == 0)
{
ret = usbclass_setconfig(priv, value);
}
}
break;
case USB_REQ_GETCONFIGURATION:
{
if (ctrl->type == USB_DIR_IN)
{
*(FAR uint8_t *)ctrlreq->buf = priv->config;
ret = 1;
}
}
break;
default:
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_UNSUPPORTEDSTDREQ), ctrl->req);
break;
}
}
break;
/* Class requests */
case USB_REQ_TYPE_CLASS:
{
if ((ctrl->type & USB_REQ_RECIPIENT_MASK) == USB_REQ_RECIPIENT_INTERFACE)
{
if (ctrl->req == RNDIS_SEND_ENCAPSULATED_COMMAND)
{
ret = rndis_handle_control_message(priv, dataout, outlen);
}
else if (ctrl->req == RNDIS_GET_ENCAPSULATED_RESPONSE)
{
if (!priv->ctrlreq_has_encap_response)
{
ret = 1;
ctrlreq->buf[0] = 0;
}
else
{
/* There is data prepared in the ctrlreq buffer.
* Just assign the length.
*/
FAR struct rndis_response_header *hdr =
(struct rndis_response_header *)ctrlreq->buf;
ret = hdr->msglen;
}
}
}
}
break;
default:
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_UNSUPPORTEDTYPE), ctrl->type);
break;
}
/* Respond to the setup command if data was returned. On an error return
* value (ret < 0), the USB driver will stall.
*/
if (ret >= 0)
{
ctrlreq->len = min(len, ret);
ctrlreq->flags = USBDEV_REQFLAGS_NULLPKT;
ret = EP_SUBMIT(dev->ep0, ctrlreq);
if (ret < 0)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPRESPQ), (uint16_t)-ret);
ctrlreq->result = OK;
usbclass_ep0incomplete(dev->ep0, ctrlreq);
}
}
return ret;
}
/****************************************************************************
* Name: usbclass_disconnect
*
* Description:
* Invoked after all transfers have been stopped, when the host is
* disconnected. This function is probably called from the context of an
* interrupt handler.
*
****************************************************************************/
static void usbclass_disconnect(FAR struct usbdevclass_driver_s *driver,
FAR struct usbdev_s *dev)
{
FAR struct rndis_dev_s *priv;
irqstate_t flags;
usbtrace(TRACE_CLASSDISCONNECT, 0);
#ifdef CONFIG_DEBUG_FEATURES
if (!driver || !dev || !dev->ep0)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_INVALIDARG), 0);
return;
}
#endif
/* Extract reference to private data */
priv = ((FAR struct rndis_driver_s *)driver)->dev;
#ifdef CONFIG_DEBUG_FEATURES
if (!priv)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EP0NOTBOUND), 0);
return;
}
#endif
/* Inform the "upper half" network driver that we have lost the USB
* connection.
*/
priv->netdev.d_ifdown(&priv->netdev);
flags = enter_critical_section();
/* Reset the configuration */
usbclass_resetconfig(priv);
leave_critical_section(flags);
/* Perform the soft connect function so that we will we can be
* re-enumerated.
*/
DEV_CONNECT(dev);
}
/****************************************************************************
* Name: usbclass_resetconfig
*
* Description:
* Mark the device as not configured and disable all endpoints.
*
****************************************************************************/
static void usbclass_resetconfig(FAR struct rndis_dev_s *priv)
{
/* Are we configured? */
if (priv->config != RNDIS_CONFIGIDNONE)
{
/* Yes.. but not anymore */
priv->config = RNDIS_CONFIGIDNONE;
priv->netdev.d_ifdown(&priv->netdev);
/* Disable endpoints. This should force completion of all pending
* transfers.
*/
EP_DISABLE(priv->epintin);
EP_DISABLE(priv->epbulkin);
EP_DISABLE(priv->epbulkout);
}
}
/****************************************************************************
* Name: usbclass_setconfig
*
* Description:
* Set the device configuration by allocating and configuring endpoints and
* by allocating and queue read and write requests.
*
****************************************************************************/
static int usbclass_setconfig(FAR struct rndis_dev_s *priv, uint8_t config)
{
struct usb_epdesc_s epdesc;
bool hispeed = false;
int ret = 0;
#ifdef CONFIG_DEBUG_FEATURES
if (priv == NULL)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_INVALIDARG), 0);
return -EIO;
}
#endif
#ifdef CONFIG_USBDEV_DUALSPEED
hispeed = (priv->usbdev->speed == USB_SPEED_HIGH);
#endif
if (config == priv->config)
{
/* Already configured -- Do nothing */
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_ALREADYCONFIGURED), 0);
return 0;
}
/* Discard the previous configuration data */
usbclass_resetconfig(priv);
/* Was this a request to simply discard the current configuration? */
if (config == RNDIS_CONFIGIDNONE)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_CONFIGNONE), 0);
return 0;
}
/* We only accept one configuration */
if (config != RNDIS_CONFIGID)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_CONFIGIDBAD), 0);
return -EINVAL;
}
/* Configure the IN interrupt endpoint */
usbclass_copy_epdesc(RNDIS_EP_INTIN_IDX, &epdesc, &priv->devinfo, hispeed);
ret = EP_CONFIGURE(priv->epintin, &epdesc, false);
if (ret < 0)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPINTINCONFIGFAIL), 0);
goto errout;
}
priv->epintin->priv = priv;
/* Configure the IN bulk endpoint */
usbclass_copy_epdesc(RNDIS_EP_BULKIN_IDX, &epdesc, &priv->devinfo, hispeed);
ret = EP_CONFIGURE(priv->epbulkin, &epdesc, false);
if (ret < 0)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPBULKINCONFIGFAIL), 0);
goto errout;
}
priv->epbulkin->priv = priv;
/* Configure the OUT bulk endpoint */
usbclass_copy_epdesc(RNDIS_EP_BULKOUT_IDX, &epdesc, &priv->devinfo, hispeed);
ret = EP_CONFIGURE(priv->epbulkout, &epdesc, true);
if (ret < 0)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_EPBULKOUTCONFIGFAIL), 0);
goto errout;
}
priv->epbulkout->priv = priv;
/* Queue read requests in the bulk OUT endpoint */
priv->rdreq->callback = rndis_rdcomplete;
ret = rndis_submit_rdreq(priv);
if (ret != OK)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_RDSUBMIT), (uint16_t)-ret);
goto errout;
}
/* We are successfully configured */
priv->config = config;
if (priv->netdev.d_ifup(&priv->netdev) == OK)
{
priv->netdev.d_flags |= IFF_UP;
}
return OK;
errout:
usbclass_resetconfig(priv);
return ret;
}
/****************************************************************************
* Name: usbclass_classobject
*
* Description:
* Allocate memory for the RNDIS driver class object
*
* Returned Value:
* 0 on success, negative error code on failure.
*
****************************************************************************/
static int usbclass_classobject(int minor,
FAR struct usbdev_devinfo_s *devinfo,
FAR struct usbdevclass_driver_s **classdev)
{
FAR struct rndis_alloc_s *alloc;
FAR struct rndis_dev_s *priv;
FAR struct rndis_driver_s *drvr;
int ret;
/* Allocate the structures needed */
alloc = (FAR struct rndis_alloc_s *)kmm_zalloc(sizeof(struct rndis_alloc_s));
if (!alloc)
{
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_ALLOCDEVSTRUCT), 0);
return -ENOMEM;
}
/* Convenience pointers into the allocated blob */
priv = &alloc->dev;
drvr = &alloc->drvr;
*classdev = &drvr->drvr;
#ifdef CONFIG_RNDIS_COMPOSITE
priv->devinfo = *devinfo;
#else
priv->devinfo.epno[RNDIS_EP_INTIN_IDX] = USB_EPNO(RNDIS_EPINTIN_ADDR);
priv->devinfo.epno[RNDIS_EP_BULKIN_IDX] = USB_EPNO(RNDIS_EPBULKIN_ADDR);
priv->devinfo.epno[RNDIS_EP_BULKOUT_IDX] = USB_EPNO(RNDIS_EPBULKOUT_ADDR);
#endif
/* Initialize the USB ethernet driver structure */
sq_init(&priv->reqlist);
memcpy(priv->host_mac_address, g_rndis_default_mac_addr, 6);
priv->txpoll = wd_create();
priv->netdev.d_private = priv;
priv->netdev.d_ifup = &rndis_ifup;
priv->netdev.d_ifdown = &rndis_ifdown;
priv->netdev.d_txavail = &rndis_txavail;
/* MAC address filtering is purposefully left out of this driver. Since
* in the RNDIS USB scenario there are only two devices in the network
* (host and us), there shouldn't be any packets received that don't
* belong to us.
*/
/* Initialize the USB class driver structure */
drvr->drvr.speed = USB_SPEED_FULL;
drvr->drvr.ops = &g_driverops;
drvr->dev = priv;
ret = netdev_register(&priv->netdev, NET_LL_ETHERNET);
if (ret)
{
uerr("Failed to register net device");
return ret;
}
drvr->dev->registered = true;
return OK;
}
static void usbclass_uninitialize(FAR struct usbdevclass_driver_s *classdev)
{
FAR struct rndis_driver_s *drvr = (FAR struct rndis_driver_s *)classdev;
FAR struct rndis_alloc_s *alloc = (FAR struct rndis_alloc_s *)drvr->dev;
if (drvr->dev->registered)
{
netdev_unregister(&drvr->dev->netdev);
drvr->dev->registered = false;
}
else
{
kmm_free(alloc);
}
}
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: usbdev_rndis_initialize
*
* Description:
* Initialize the RNDIS USB device driver.
*
* Input Parameters:
* mac_address: pointer to an array of six octets which is the MAC address
* of the host side of the interface. May be NULL to use the
* default MAC address.
*
* Returned Value:
* 0 on success, -errno on failure
*
****************************************************************************/
#ifndef CONFIG_RNDIS_COMPOSITE
int usbdev_rndis_initialize(FAR const uint8_t *mac_address)
{
int ret;
FAR struct usbdevclass_driver_s *classdev;
FAR struct rndis_driver_s *drvr;
ret = usbclass_classobject(0, NULL, &classdev);
if (ret)
{
nerr("usbclass_classobject failed: %d\n", ret);
return ret;
}
drvr = (FAR struct rndis_driver_s *)classdev;
if (mac_address)
{
memcpy(drvr->dev->host_mac_address, mac_address, 6);
}
ret = usbdev_register(&drvr->drvr);
if (ret)
{
nerr("usbdev_register failed: %d\n", ret);
usbtrace(TRACE_CLSERROR(USBSER_TRACEERR_DEVREGISTER), (uint16_t)-ret);
usbclass_uninitialize(classdev);
return ret;
}
return OK;
}
#endif
/****************************************************************************
* Name: usbdev_rndis_set_host_mac_addr
*
* Description:
* Set host MAC address. Mainly for use with composite devices where
* the MAC cannot be given directly to usbdev_rndis_initialize().
*
* Input Parameters:
* netdev: pointer to the network interface. Can be obtained from
* e.g. netdev_findbyname().
*
* mac_address: pointer to an array of six octets which is the MAC address
* of the host side of the interface. May be NULL to use the
* default MAC address.
*
* Returned Value:
* 0 on success, -errno on failure
*
****************************************************************************/
int usbdev_rndis_set_host_mac_addr(FAR struct net_driver_s *netdev,
FAR const uint8_t *mac_address)
{
FAR struct rndis_dev_s *dev = (FAR struct rndis_dev_s *)netdev;
if (mac_address)
{
memcpy(dev->host_mac_address, mac_address, 6);
}
else
{
memcpy(dev->host_mac_address, g_rndis_default_mac_addr, 6);
}
return OK;
}
/****************************************************************************
* Name: usbdev_rndis_get_composite_devdesc
*
* Description:
* Helper function to fill in some constants into the composite
* configuration struct.
*
* Input Parameters:
* dev - Pointer to the configuration struct we should fill
*
* Returned Value:
* None
*
****************************************************************************/
#ifdef CONFIG_RNDIS_COMPOSITE
void usbdev_rndis_get_composite_devdesc(struct composite_devdesc_s *dev)
{
memset(dev, 0, sizeof(struct composite_devdesc_s));
dev->mkconfdesc = usbclass_mkcfgdesc;
dev->mkstrdesc = usbclass_mkstrdesc;
dev->classobject = usbclass_classobject;
dev->uninitialize = usbclass_uninitialize;
dev->nconfigs = RNDIS_NCONFIGS;
dev->configid = RNDIS_CONFIGID;
dev->cfgdescsize = sizeof(g_rndis_cfgdesc);
dev->devinfo.ninterfaces = RNDIS_NINTERFACES;
dev->devinfo.nstrings = 0;
dev->devinfo.nendpoints = RNDIS_NUM_EPS;
/* Default endpoint indexes, board-specific logic can override these */
dev->devinfo.epno[RNDIS_EP_INTIN_IDX] = USB_EPNO(RNDIS_EPINTIN_ADDR);
dev->devinfo.epno[RNDIS_EP_BULKIN_IDX] = USB_EPNO(RNDIS_EPBULKIN_ADDR);
dev->devinfo.epno[RNDIS_EP_BULKOUT_IDX] = USB_EPNO(RNDIS_EPBULKOUT_ADDR);
}
#endif