[Bug 263283] Packet drops on rock64

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=263283

            Bug ID: 263283
           Summary: Packet drops on rock64
           Product: Base System
           Version: 13.0-RELEASE
          Hardware: arm64
                OS: Any
            Status: New
          Severity: Affects Some People
          Priority: ---
         Component: kern
          Assignee: bugs@FreeBSD.org
          Reporter: richard@kojedz.in

On a Rock64 board, I experience high packet drops:

# ping -c 1000 -q -f -s 1472 172.16.47.254
PING 172.16.47.254 (172.16.47.254): 1472 data bytes

--- 172.16.47.254 ping statistics ---
1000 packets transmitted, 845 packets received, 15.5% packet loss
round-trip min/avg/max/stddev = 0.289/0.301/0.518/0.017 ms

The errors are present in netstat:

# netstat -n -I dwc0
Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts Oerrs
 Coll
dwc0   1500 <Link#1>      9e:c8:cb:a2:e2:c0  1141478  2982     0  1700981     0
    0



This happens at gigabit speeds. If I limit the link speed to 100Mbit, no packet
loss is present. Have checked all the cables, switch ports, etc. The same board
works well with linux 5.15. Howewer, with linux 5.4 there were similar issues,
but then it got fixed somehow.

After some events (i.e. reboots, link up/down, port speed change) the situation
changes, the nic enters into a more stable state, when packet loss is much
lower, but still not zero. I cannot reliably reproduce this.

After a reboot:

# netstat -n -I dwc0
Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts Oerrs
 Coll
dwc0   1500# ping -c 1000 -q -f -s 1472 172.16.47.254
PING 172.16.47.254 (172.16.47.254): 1472 data bytes

--- 172.16.47.254 ping statistics ---
1000 packets transmitted, 1000 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.286/0.297/0.603/0.021 ms
 <Link#1>      9e:c8:cb:a2:e2:c0       82     0     0       83     0     0
# netstat -n -I dwc0
Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts Oerrs
 Coll
dwc0   1500 <Link#1>      9e:c8:cb:a2:e2:c0     2126     0     0     2113     0
    0


How should I proceed debugging this?

Richard

-- 
You are receiving this mail because:
You are the assignee for the bug.