Hi, On Fri, 2018-03-16 at 11:26 +0100, Jakob Unterwurzacher wrote: > On 15.03.18 23:30, John Fastabend wrote: > > > I have reproduced it using two USB network cards connected to each other. > > > The test tool sends UDP packets containing a counter and listens on the > > > other interface, it is available at > > > https://github.com/jakob-tsd/pfifo_stress/blob/master/pfifo_stress.py > > > > > > > Great thanks, can you also run this with taskset to bind to > > a single CPU, > > > > # taskset 0x1 ./pifof_stress.py > > > > And let me know if you still see the OOO. > > Interesting. Looks like it depends on which core it runs on. CPU0 is > clean, CPU1 is not. > > Clean: taskset --cpu-list 0 ./pfifo_stress.py > > Broken: taskset --cpu-list 1 ./pfifo_stress.py > > Maybe related: CPU0 is where USB interrupts are handled: > > > root@rk3399-q7:~# cat /proc/interrupts > > CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 > > 217: 2175353 0 0 0 0 0 > > GICv3 142 Level xhci-hcd:usb5
Is not clear to me if you can reproduce the bug with the vanilla kernel, or if you need some out-of-tree nic driver. Can you please clarify which NIC/driver are you using? Thanks, Paolo