USB: ethernet gadget interop with MCCI Windows driver

It turns out that minor tweaks to the "CDC Subset" support in the Ethernet
gadget driver, just updating a config descriptor, let it be automagically
recognized by a Windows driver supported by MCCI.

This patch adds those descriptors, so systems using PXA 255 processors
(like Gumstix etc) can interop with those commercial MS-Windows drivers.
This is a Good Thing since Microsoft's RNDIS code has bugginess issues,
which are unfortunately compounded by "won't fix" issues as well as "the
published specs are incomplete and wrong" issues.  Being able to talk to
the MCCI driver gives Windows users another connectivity option.  (MCCI
also has CDC Ethernet drivers, which can help most non-PXA processors.)

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

1 files changed, 103 insertions, 39 deletions

diff --git a/drivers/usb/gadget/ether.c b/drivers/usb/gadget/ether.c
index ca8e0ebc79e8..72e2b65293c8 100644
--- a/drivers/usb/gadget/ether.c
+++ b/drivers/usb/gadget/ether.c
@@ -72,9 +72,18 @@
  *
  * There's some hardware that can't talk CDC.  We make that hardware
  * implement a "minimalist" vendor-agnostic CDC core:  same framing, but
- * link-level setup only requires activating the configuration.
- * Linux supports it, but other host operating systems may not.
- * (This is a subset of CDC Ethernet.)
+ * link-level setup only requires activating the configuration.  Only the
+ * endpoint descriptors, and product/vendor IDs, are relevant; no control
+ * operations are available.  Linux supports it, but other host operating
+ * systems may not.  (This is a subset of CDC Ethernet.)
+ *
+ * It turns out that if you add a few descriptors to that "CDC Subset",
+ * (Windows) host side drivers from MCCI can treat it as one submode of
+ * a proprietary scheme called "SAFE" ... without needing to know about
+ * specific product/vendor IDs.  So we do that, making it easier to use
+ * those MS-Windows drivers.  Those added descriptors make it resemble a
+ * CDC MDLM device, but they don't change device behavior at all.  (See
+ * MCCI Engineering report 950198 "SAFE Networking Functions".)
  *
  * A third option is also in use.  Rather than CDC Ethernet, or something
  * simpler, Microsoft pushes their own approach: RNDIS.  The published
@@ -254,6 +263,10 @@ MODULE_PARM_DESC(host_addr, "Host Ethernet Address");
 #define DEV_CONFIG_CDC
 #endif
 
+#ifdef CONFIG_USB_GADGET_S3C2410
+#define DEV_CONFIG_CDC
+#endif
+
 #ifdef CONFIG_USB_GADGET_AT91
 #define DEV_CONFIG_CDC
 #endif
@@ -283,9 +296,6 @@ MODULE_PARM_DESC(host_addr, "Host Ethernet Address");
 #define	DEV_CONFIG_SUBSET
 #endif
 
-#ifdef CONFIG_USB_GADGET_S3C2410
-#define DEV_CONFIG_CDC
-#endif
 
 /*-------------------------------------------------------------------------*/
 
@@ -487,8 +497,17 @@ rndis_config = {
  * endpoint.  Both have a "data" interface and two bulk endpoints.
  * There are also differences in how control requests are handled.
  *
- * RNDIS shares a lot with CDC-Ethernet, since it's a variant of
- * the CDC-ACM (modem) spec.
+ * RNDIS shares a lot with CDC-Ethernet, since it's a variant of the
+ * CDC-ACM (modem) spec.  Unfortunately MSFT's RNDIS driver is buggy; it
+ * may hang or oops.  Since bugfixes (or accurate specs, letting Linux
+ * work around those bugs) are unlikely to ever come from MSFT, you may
+ * wish to avoid using RNDIS.
+ *
+ * MCCI offers an alternative to RNDIS if you need to connect to Windows
+ * but have hardware that can't support CDC Ethernet.   We add descriptors
+ * to present the CDC Subset as a (nonconformant) CDC MDLM variant called
+ * "SAFE".  That borrows from both CDC Ethernet and CDC MDLM.  You can
+ * get those drivers from MCCI, or bundled with various products.
  */
 
 #ifdef	DEV_CONFIG_CDC
@@ -522,8 +541,6 @@ rndis_control_intf = {
 };
 #endif
 
-#if defined(DEV_CONFIG_CDC) || defined(CONFIG_USB_ETH_RNDIS)
-
 static const struct usb_cdc_header_desc header_desc = {
 	.bLength =		sizeof header_desc,
 	.bDescriptorType =	USB_DT_CS_INTERFACE,
@@ -532,6 +549,8 @@ static const struct usb_cdc_header_desc header_desc = {
 	.bcdCDC =		__constant_cpu_to_le16 (0x0110),
 };
 
+#if defined(DEV_CONFIG_CDC) || defined(CONFIG_USB_ETH_RNDIS)
+
 static const struct usb_cdc_union_desc union_desc = {
 	.bLength =		sizeof union_desc,
 	.bDescriptorType =	USB_DT_CS_INTERFACE,
@@ -564,7 +583,40 @@ static const struct usb_cdc_acm_descriptor acm_descriptor = {
 
 #endif
 
-#ifdef	DEV_CONFIG_CDC
+#ifndef DEV_CONFIG_CDC
+
+/* "SAFE" loosely follows CDC WMC MDLM, violating the spec in various
+ * ways:  data endpoints live in the control interface, there's no data
+ * interface, and it's not used to talk to a cell phone radio.
+ */
+
+static const struct usb_cdc_mdlm_desc mdlm_desc = {
+	.bLength =		sizeof mdlm_desc,
+	.bDescriptorType =	USB_DT_CS_INTERFACE,
+	.bDescriptorSubType =	USB_CDC_MDLM_TYPE,
+
+	.bcdVersion =		__constant_cpu_to_le16(0x0100),
+	.bGUID = {
+		0x5d, 0x34, 0xcf, 0x66, 0x11, 0x18, 0x11, 0xd6,
+		0xa2, 0x1a, 0x00, 0x01, 0x02, 0xca, 0x9a, 0x7f,
+	},
+};
+
+/* since "usb_cdc_mdlm_detail_desc" is a variable length structure, we
+ * can't really use its struct.  All we do here is say that we're using
+ * the submode of "SAFE" which directly matches the CDC Subset.
+ */
+static const u8 mdlm_detail_desc[] = {
+	6,
+	USB_DT_CS_INTERFACE,
+	USB_CDC_MDLM_DETAIL_TYPE,
+
+	0,	/* "SAFE" */
+	0,	/* network control capabilities (none) */
+	0,	/* network data capabilities ("raw" encapsulation) */
+};
+
+#endif
 
 static const struct usb_cdc_ether_desc ether_desc = {
 	.bLength =		sizeof ether_desc,
@@ -579,7 +631,6 @@ static const struct usb_cdc_ether_desc ether_desc = {
 	.bNumberPowerFilters =	0,
 };
 
-#endif
 
 #if defined(DEV_CONFIG_CDC) || defined(CONFIG_USB_ETH_RNDIS)
 
@@ -672,6 +723,9 @@ rndis_data_intf = {
 /*
  * "Simple" CDC-subset option is a simple vendor-neutral model that most
  * full speed controllers can handle:  one interface, two bulk endpoints.
+ *
+ * To assist host side drivers, we fancy it up a bit, and add descriptors
+ * so some host side drivers will understand it as a "SAFE" variant.
  */
 
 static const struct usb_interface_descriptor
@@ -682,8 +736,8 @@ subset_data_intf = {
 	.bInterfaceNumber =	0,
 	.bAlternateSetting =	0,
 	.bNumEndpoints =	2,
-	.bInterfaceClass =	USB_CLASS_VENDOR_SPEC,
-	.bInterfaceSubClass =	0,
+	.bInterfaceClass =      USB_CLASS_COMM,
+	.bInterfaceSubClass =	USB_CDC_SUBCLASS_MDLM,
 	.bInterfaceProtocol =	0,
 	.iInterface =		STRING_DATA,
 };
@@ -731,10 +785,15 @@ static const struct usb_descriptor_header *fs_eth_function [11] = {
 static inline void __init fs_subset_descriptors(void)
 {
 #ifdef DEV_CONFIG_SUBSET
+	/* behavior is "CDC Subset"; extra descriptors say "SAFE" */
 	fs_eth_function[1] = (struct usb_descriptor_header *) &subset_data_intf;
-	fs_eth_function[2] = (struct usb_descriptor_header *) &fs_source_desc;
-	fs_eth_function[3] = (struct usb_descriptor_header *) &fs_sink_desc;
-	fs_eth_function[4] = NULL;
+	fs_eth_function[2] = (struct usb_descriptor_header *) &header_desc;
+	fs_eth_function[3] = (struct usb_descriptor_header *) &mdlm_desc;
+	fs_eth_function[4] = (struct usb_descriptor_header *) &mdlm_detail_desc;
+	fs_eth_function[5] = (struct usb_descriptor_header *) &ether_desc;
+	fs_eth_function[6] = (struct usb_descriptor_header *) &fs_source_desc;
+	fs_eth_function[7] = (struct usb_descriptor_header *) &fs_sink_desc;
+	fs_eth_function[8] = NULL;
 #else
 	fs_eth_function[1] = NULL;
 #endif
@@ -828,10 +887,15 @@ static const struct usb_descriptor_header *hs_eth_function [11] = {
 static inline void __init hs_subset_descriptors(void)
 {
 #ifdef DEV_CONFIG_SUBSET
+	/* behavior is "CDC Subset"; extra descriptors say "SAFE" */
 	hs_eth_function[1] = (struct usb_descriptor_header *) &subset_data_intf;
-	hs_eth_function[2] = (struct usb_descriptor_header *) &fs_source_desc;
-	hs_eth_function[3] = (struct usb_descriptor_header *) &fs_sink_desc;
-	hs_eth_function[4] = NULL;
+	hs_eth_function[2] = (struct usb_descriptor_header *) &header_desc;
+	hs_eth_function[3] = (struct usb_descriptor_header *) &mdlm_desc;
+	hs_eth_function[4] = (struct usb_descriptor_header *) &mdlm_detail_desc;
+	hs_eth_function[5] = (struct usb_descriptor_header *) &ether_desc;
+	hs_eth_function[6] = (struct usb_descriptor_header *) &hs_source_desc;
+	hs_eth_function[7] = (struct usb_descriptor_header *) &hs_sink_desc;
+	hs_eth_function[8] = NULL;
 #else
 	hs_eth_function[1] = NULL;
 #endif
@@ -878,10 +942,8 @@ static char				manufacturer [50];
 static char				product_desc [40] = DRIVER_DESC;
 static char				serial_number [20];
 
-#ifdef	DEV_CONFIG_CDC
 /* address that the host will use ... usually assigned at random */
 static char				ethaddr [2 * ETH_ALEN + 1];
-#endif
 
 /* static strings, in UTF-8 */
 static struct usb_string		strings [] = {
@@ -889,9 +951,9 @@ static struct usb_string		strings [] = {
 	{ STRING_PRODUCT,	product_desc, },
 	{ STRING_SERIALNUMBER,	serial_number, },
 	{ STRING_DATA,		"Ethernet Data", },
+	{ STRING_ETHADDR,	ethaddr, },
 #ifdef	DEV_CONFIG_CDC
 	{ STRING_CDC,		"CDC Ethernet", },
-	{ STRING_ETHADDR,	ethaddr, },
 	{ STRING_CONTROL,	"CDC Communications Control", },
 #endif
 #ifdef	DEV_CONFIG_SUBSET
@@ -986,10 +1048,10 @@ set_ether_config (struct eth_dev *dev, gfp_t gfp_flags)
 	}
 #endif
 
-	dev->in = ep_desc (dev->gadget, &hs_source_desc, &fs_source_desc);
+	dev->in = ep_desc(gadget, &hs_source_desc, &fs_source_desc);
 	dev->in_ep->driver_data = dev;
 
-	dev->out = ep_desc (dev->gadget, &hs_sink_desc, &fs_sink_desc);
+	dev->out = ep_desc(gadget, &hs_sink_desc, &fs_sink_desc);
 	dev->out_ep->driver_data = dev;
 
 	/* With CDC,  the host isn't allowed to use these two data
@@ -2278,10 +2340,10 @@ eth_bind (struct usb_gadget *gadget)
 			"RNDIS/%s", driver_desc);
 
 	/* CDC subset ... recognized by Linux since 2.4.10, but Windows
-	 * drivers aren't widely available.
+	 * drivers aren't widely available.  (That may be improved by
+	 * supporting one submode of the "SAFE" variant of MDLM.)
 	 */
 	} else if (!cdc) {
-		device_desc.bDeviceClass = USB_CLASS_VENDOR_SPEC;
 		device_desc.idVendor =
 			__constant_cpu_to_le16(SIMPLE_VENDOR_NUM);
 		device_desc.idProduct =
@@ -2352,6 +2414,10 @@ autoconf_fail:
 	if (!cdc) {
 		eth_config.bNumInterfaces = 1;
 		eth_config.iConfiguration = STRING_SUBSET;
+
+		/* use functions to set these up, in case we're built to work
+		 * with multiple controllers and must override CDC Ethernet.
+		 */
 		fs_subset_descriptors();
 		hs_subset_descriptors();
 	}
@@ -2415,22 +2481,20 @@ autoconf_fail:
 
 	/* Module params for these addresses should come from ID proms.
 	 * The host side address is used with CDC and RNDIS, and commonly
-	 * ends up in a persistent config database.
+	 * ends up in a persistent config database.  It's not clear if
+	 * host side code for the SAFE thing cares -- its original BLAN
+	 * thing didn't, Sharp never assigned those addresses on Zaurii.
 	 */
 	if (get_ether_addr(dev_addr, net->dev_addr))
 		dev_warn(&gadget->dev,
 			"using random %s ethernet address\n", "self");
-	if (cdc || rndis) {
-		if (get_ether_addr(host_addr, dev->host_mac))
-			dev_warn(&gadget->dev,
-				"using random %s ethernet address\n", "host");
-#ifdef	DEV_CONFIG_CDC
-		snprintf (ethaddr, sizeof ethaddr, "%02X%02X%02X%02X%02X%02X",
-			dev->host_mac [0], dev->host_mac [1],
-			dev->host_mac [2], dev->host_mac [3],
-			dev->host_mac [4], dev->host_mac [5]);
-#endif
-	}
+	if (get_ether_addr(host_addr, dev->host_mac))
+		dev_warn(&gadget->dev,
+			"using random %s ethernet address\n", "host");
+	snprintf (ethaddr, sizeof ethaddr, "%02X%02X%02X%02X%02X%02X",
+		dev->host_mac [0], dev->host_mac [1],
+		dev->host_mac [2], dev->host_mac [3],
+		dev->host_mac [4], dev->host_mac [5]);
 
 	if (rndis) {
 		status = rndis_init();