| Message ID | 20220502110644.167179-1-marcan@marcan.st (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2] net: usb: ax88179_178a: Bind only to vendor-specific interface | expand |
On Mon, 2 May 2022 20:06:44 +0900 Hector Martin wrote: > The Anker PowerExpand USB-C to Gigabit Ethernet adapter uses this > chipset, but exposes CDC Ethernet configurations as well as the > vendor specific one. And we have reasons to believe all dongle vendors may have a similar problem? > This driver ends up binding first to both CDC > interfaces, tries to instantiate two Ethernet interfaces talking to > the same device, and the result is a nice fireworks show. > > Change all the ID matches to specifically match the vendor-specific > interface. By default the device comes up in CDC mode and is bound by > that driver (which works fine); users may switch it to the vendor > interface using sysfs to set bConfigurationValue, at which point the > device actually goes through a reconnect cycle and comes back as a > vendor specific only device, and then this driver binds and works too. > > v2: Fixed interface protocol match, commit message. > > Signed-off-by: Hector Martin <marcan@marcan.st> > --- > drivers/net/usb/ax88179_178a.c | 26 +++++++++++++------------- > 1 file changed, 13 insertions(+), 13 deletions(-) > > diff --git a/drivers/net/usb/ax88179_178a.c b/drivers/net/usb/ax88179_178a.c > index e2fa56b92685..7c7c2f31d9f1 100644 > --- a/drivers/net/usb/ax88179_178a.c > +++ b/drivers/net/usb/ax88179_178a.c > @@ -1914,55 +1914,55 @@ static const struct driver_info at_umc2000sp_info = { > static const struct usb_device_id products[] = { > { > /* ASIX AX88179 10/100/1000 */ > - USB_DEVICE(0x0b95, 0x1790), > + USB_DEVICE_AND_INTERFACE_INFO(0x0b95, 0x1790, 0xff, 0xff, 0), > .driver_info = (unsigned long)&ax88179_info, > }, Should we use USB_CLASS_VENDOR_SPEC and USB_SUBCLASS_VENDOR_SPEC ? Maybe define a local macro wrapper for USB_DEVICE_AND.. which will fill those in to avoid long lines?
On 05/05/2022 11.30, Jakub Kicinski wrote: > On Mon, 2 May 2022 20:06:44 +0900 Hector Martin wrote: >> The Anker PowerExpand USB-C to Gigabit Ethernet adapter uses this >> chipset, but exposes CDC Ethernet configurations as well as the >> vendor specific one. > > And we have reasons to believe all dongle vendors may have a similar > problem? Given this is a vendor-specific driver it seems correct to have it only bind to vendor-specific interfaces. That shouldn't break anything as long as nobody is implementing this interface with the wrong protocol IDs (which seems like quite a weird thing to do and fairly unlikely). FWIW, the one I have has the generic VID/PID, not a custom vendor one. If you prefer I can change just that one or both of the generic ones. >> diff --git a/drivers/net/usb/ax88179_178a.c b/drivers/net/usb/ax88179_178a.c >> index e2fa56b92685..7c7c2f31d9f1 100644 >> --- a/drivers/net/usb/ax88179_178a.c >> +++ b/drivers/net/usb/ax88179_178a.c >> @@ -1914,55 +1914,55 @@ static const struct driver_info at_umc2000sp_info = { >> static const struct usb_device_id products[] = { >> { >> /* ASIX AX88179 10/100/1000 */ >> - USB_DEVICE(0x0b95, 0x1790), >> + USB_DEVICE_AND_INTERFACE_INFO(0x0b95, 0x1790, 0xff, 0xff, 0), >> .driver_info = (unsigned long)&ax88179_info, >> }, > > Should we use USB_CLASS_VENDOR_SPEC and USB_SUBCLASS_VENDOR_SPEC ? > Maybe define a local macro wrapper for USB_DEVICE_AND.. which will > fill those in to avoid long lines? Sure, I'll do that!
On Thu, 5 May 2022 21:05:29 +0900 Hector Martin wrote: > On 05/05/2022 11.30, Jakub Kicinski wrote: > > On Mon, 2 May 2022 20:06:44 +0900 Hector Martin wrote: > >> The Anker PowerExpand USB-C to Gigabit Ethernet adapter uses this > >> chipset, but exposes CDC Ethernet configurations as well as the > >> vendor specific one. > > > > And we have reasons to believe all dongle vendors may have a similar > > problem? > > Given this is a vendor-specific driver it seems correct to have it only > bind to vendor-specific interfaces. That shouldn't break anything as > long as nobody is implementing this interface with the wrong protocol > IDs (which seems like quite a weird thing to do and fairly unlikely). > > FWIW, the one I have has the generic VID/PID, not a custom vendor one. > If you prefer I can change just that one or both of the generic ones. I have no strong preference, maybe folks more familiar with USB in general do and will chime in. Seems like a USB question more than a networking one. I know we have sort of the opposite problem with Realtek devices where they bind to the generic driver instead of the vendor one resulting in loss of functionality and higher power draw. But that's not 1:1 the problem you're solving. Let's just make sure the commit message is more explicit about which IDs are required to fix your system and which are just changed for a good measure, in case of regressions.
diff --git a/drivers/net/usb/ax88179_178a.c b/drivers/net/usb/ax88179_178a.c index e2fa56b92685..7c7c2f31d9f1 100644 --- a/drivers/net/usb/ax88179_178a.c +++ b/drivers/net/usb/ax88179_178a.c @@ -1914,55 +1914,55 @@ static const struct driver_info at_umc2000sp_info = { static const struct usb_device_id products[] = { { /* ASIX AX88179 10/100/1000 */ - USB_DEVICE(0x0b95, 0x1790), + USB_DEVICE_AND_INTERFACE_INFO(0x0b95, 0x1790, 0xff, 0xff, 0), .driver_info = (unsigned long)&ax88179_info, }, { /* ASIX AX88178A 10/100/1000 */ - USB_DEVICE(0x0b95, 0x178a), + USB_DEVICE_AND_INTERFACE_INFO(0x0b95, 0x178a, 0xff, 0xff, 0), .driver_info = (unsigned long)&ax88178a_info, }, { /* Cypress GX3 SuperSpeed to Gigabit Ethernet Bridge Controller */ - USB_DEVICE(0x04b4, 0x3610), + USB_DEVICE_AND_INTERFACE_INFO(0x04b4, 0x3610, 0xff, 0xff, 0), .driver_info = (unsigned long)&cypress_GX3_info, }, { /* D-Link DUB-1312 USB 3.0 to Gigabit Ethernet Adapter */ - USB_DEVICE(0x2001, 0x4a00), + USB_DEVICE_AND_INTERFACE_INFO(0x2001, 0x4a00, 0xff, 0xff, 0), .driver_info = (unsigned long)&dlink_dub1312_info, }, { /* Sitecom USB 3.0 to Gigabit Adapter */ - USB_DEVICE(0x0df6, 0x0072), + USB_DEVICE_AND_INTERFACE_INFO(0x0df6, 0x0072, 0xff, 0xff, 0), .driver_info = (unsigned long)&sitecom_info, }, { /* Samsung USB Ethernet Adapter */ - USB_DEVICE(0x04e8, 0xa100), + USB_DEVICE_AND_INTERFACE_INFO(0x04e8, 0xa100, 0xff, 0xff, 0), .driver_info = (unsigned long)&samsung_info, }, { /* Lenovo OneLinkDock Gigabit LAN */ - USB_DEVICE(0x17ef, 0x304b), + USB_DEVICE_AND_INTERFACE_INFO(0x17ef, 0x304b, 0xff, 0xff, 0), .driver_info = (unsigned long)&lenovo_info, }, { /* Belkin B2B128 USB 3.0 Hub + Gigabit Ethernet Adapter */ - USB_DEVICE(0x050d, 0x0128), + USB_DEVICE_AND_INTERFACE_INFO(0x050d, 0x0128, 0xff, 0xff, 0), .driver_info = (unsigned long)&belkin_info, }, { /* Toshiba USB 3.0 GBit Ethernet Adapter */ - USB_DEVICE(0x0930, 0x0a13), + USB_DEVICE_AND_INTERFACE_INFO(0x0930, 0x0a13, 0xff, 0xff, 0), .driver_info = (unsigned long)&toshiba_info, }, { /* Magic Control Technology U3-A9003 USB 3.0 Gigabit Ethernet Adapter */ - USB_DEVICE(0x0711, 0x0179), + USB_DEVICE_AND_INTERFACE_INFO(0x0711, 0x0179, 0xff, 0xff, 0), .driver_info = (unsigned long)&mct_info, }, { /* Allied Telesis AT-UMC2000 USB 3.0/USB 3.1 Gen 1 to Gigabit Ethernet Adapter */ - USB_DEVICE(0x07c9, 0x000e), + USB_DEVICE_AND_INTERFACE_INFO(0x07c9, 0x000e, 0xff, 0xff, 0), .driver_info = (unsigned long)&at_umc2000_info, }, { /* Allied Telesis AT-UMC200 USB 3.0/USB 3.1 Gen 1 to Fast Ethernet Adapter */ - USB_DEVICE(0x07c9, 0x000f), + USB_DEVICE_AND_INTERFACE_INFO(0x07c9, 0x000f, 0xff, 0xff, 0), .driver_info = (unsigned long)&at_umc200_info, }, { /* Allied Telesis AT-UMC2000/SP USB 3.0/USB 3.1 Gen 1 to Gigabit Ethernet Adapter */ - USB_DEVICE(0x07c9, 0x0010), + USB_DEVICE_AND_INTERFACE_INFO(0x07c9, 0x0010, 0xff, 0xff, 0), .driver_info = (unsigned long)&at_umc2000sp_info, }, { },
The Anker PowerExpand USB-C to Gigabit Ethernet adapter uses this chipset, but exposes CDC Ethernet configurations as well as the vendor specific one. This driver ends up binding first to both CDC interfaces, tries to instantiate two Ethernet interfaces talking to the same device, and the result is a nice fireworks show. Change all the ID matches to specifically match the vendor-specific interface. By default the device comes up in CDC mode and is bound by that driver (which works fine); users may switch it to the vendor interface using sysfs to set bConfigurationValue, at which point the device actually goes through a reconnect cycle and comes back as a vendor specific only device, and then this driver binds and works too. v2: Fixed interface protocol match, commit message. Signed-off-by: Hector Martin <marcan@marcan.st> --- drivers/net/usb/ax88179_178a.c | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-)