Re: Possible bug with cdc_ether, triggers NETDEV WATCHDOG
Am Mittwoch, den 09.10.2019, 09:27 -0400 schrieb Adam Bennett: > On 10/9/19 4:53 AM, Bjørn Mork wrote: > > This warning means that the gadget doesn't accept the packets we send > > it. There isn't much the host can do about that, except dropping > > packets on the floor. Which is why the warning is this loud. > > > > Would a firewall on either the linux host or the Pi Zero cause the same > problem (and message)? If a firewall discards a packet, it will not reach the transmit queue. Hence the answer to that question is negative. If other OSes get this device to work, we would need to know what they are doing differently. Can you get a trace from them and compare it to ours? Regards Oliver
Re: Possible bug with cdc_ether, triggers NETDEV WATCHDOG
On 10/9/19 4:53 AM, Bjørn Mork wrote: Adam Bennett writes: I've been messing around with a Raspberry Pi Zero, in its ethernet gadget mode. This possible bug report is not against the Pi Zero linux kernel, but rather the host computer's linux kernel. I've been able to reproduce the same host computer issue with my normal laptop, and an embedded board (buildroot-based). Both run a newish version of 4.19. The issue is reported on the host, but it's really a problem with the gadget. You'll probably have the same issues with any host, including hosts running something other than Linux. They just won't be as verbose about it. That was my original assumption also. But I tested the Pi Zero on Windows 10 (build 1903) and latest MacOS, and it works fine on those machines. The bug report is that most of the time I cannot ping through the local link, and I get a kernel debug message: sometimes I can ping the Pi Zero with no kernel message, most of the time I can't ping and the message comes up, and occasionally I get the message right when I plug in the Pi Zero, before I issue the ping command. Here is the dmesg on my normal laptop (I've included the plug-in sequence also): [11728.029900] usb 1-1: new high-speed USB device number 10 using xhci_hcd [11728.434200] usb 1-1: device descriptor read/64, error -71 First symptom of something wrong with the gadget... [11728.669543] usb 1-1: New USB device found, idVendor=0525, idProduct=a4a2, bcdDevice= 4.19 [11728.669548] usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=0 [11728.669551] usb 1-1: Product: RNDIS/Ethernet Gadget [11728.669554] usb 1-1: Manufacturer: Linux 4.19.75+ with 2098.usb [11728.674528] cdc_ether 1-1:1.0 usb0: register 'cdc_ether' at usb-:00:14.0-1, CDC Ethernet Device, 22:93:3a:1e:ac:5c [11730.725278] cdc_ether 1-1:1.0 enp0s20f0u1: renamed from usb0 [11768.174915] [ cut here ] [11768.174921] NETDEV WATCHDOG: enp0s20f0u1 (cdc_ether): transmit queue 0 timed out This warning means that the gadget doesn't accept the packets we send it. There isn't much the host can do about that, except dropping packets on the floor. Which is why the warning is this loud. Would a firewall on either the linux host or the Pi Zero cause the same problem (and message)? Bjørn
Re: Possible bug with cdc_ether, triggers NETDEV WATCHDOG
Adam Bennett writes: > I've been messing around with a Raspberry Pi Zero, in its ethernet > gadget mode. This possible bug report is not against the Pi Zero > linux kernel, but rather the host computer's linux kernel. I've been > able to reproduce the same host computer issue with my normal laptop, > and an embedded board (buildroot-based). Both run a newish version of > 4.19. The issue is reported on the host, but it's really a problem with the gadget. You'll probably have the same issues with any host, including hosts running something other than Linux. They just won't be as verbose about it. > The bug report is that most of the time I cannot ping through the > local link, and I get a kernel debug message: sometimes I can ping > the Pi Zero with no kernel message, most of the time I can't ping and > the message comes up, and occasionally I get the message right when I > plug in the Pi Zero, before I issue the ping command. > > Here is the dmesg on my normal laptop (I've included the plug-in > sequence also): > > [11728.029900] usb 1-1: new high-speed USB device number 10 using xhci_hcd > [11728.434200] usb 1-1: device descriptor read/64, error -71 First symptom of something wrong with the gadget... > [11728.669543] usb 1-1: New USB device found, idVendor=0525, > idProduct=a4a2, bcdDevice= 4.19 > [11728.669548] usb 1-1: New USB device strings: Mfr=1, Product=2, > SerialNumber=0 > [11728.669551] usb 1-1: Product: RNDIS/Ethernet Gadget > [11728.669554] usb 1-1: Manufacturer: Linux 4.19.75+ with 2098.usb > [11728.674528] cdc_ether 1-1:1.0 usb0: register 'cdc_ether' at > usb-:00:14.0-1, CDC Ethernet Device, 22:93:3a:1e:ac:5c > [11730.725278] cdc_ether 1-1:1.0 enp0s20f0u1: renamed from usb0 > [11768.174915] [ cut here ] > [11768.174921] NETDEV WATCHDOG: enp0s20f0u1 (cdc_ether): transmit > queue 0 timed out This warning means that the gadget doesn't accept the packets we send it. There isn't much the host can do about that, except dropping packets on the floor. Which is why the warning is this loud. Bjørn