Re: port-amd64/53155: OS wedges after <12h uptime when >2 bnx network interfaces in use

["J. Hannken-Illjes" <hannken%eis.cs.tu-bs.de@localhost>]

The following reply was made to PR port-amd64/53155; it has been noted by GNATS.

From: "J. Hannken-Illjes" <hannken%eis.cs.tu-bs.de@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: port-amd64/53155: OS wedges after <12h uptime when >2 bnx network
 interfaces in use
Date: Fri, 1 Jun 2018 12:40:37 +0200

 > db{0}>
 > 
 > db{0}> ps
 > PID    LID S CPU     FLAGS       STRUCT LWP *               NAME WAIT
 <snip>
 > 0    >   6 7   0       200   fffffe822f5eb440          softser/0
 > 0    >   5 7   0       200   fffffe822f5eb860          softclk/0
 > 0        4 1   0       200   fffffe822f607000          softbio/0
 > 0    >   3 7   0       200   fffffe822f607420          softnet/0
 > 0    >   2 7   0       201   fffffe822f607840             idle/0
 > 0        1 3   0       200   ffffffff81481b80            swapper uvm
 <snip>
 > db{0}> show all callout
 > hardclock_ticks now: 86821384
 >     ticks  wheel               arg  func
 >      -863 -1/-256                0  pffasttimo
 >      -861 -1/-256                0  pfslowtimo
 >      -861 -1/-256                0  tcp_slowtimo
 >      -851 -1/-256                0  sched_balance
 >      -849 -1/-256 fffffe812d614ad0  tcp_delack
 >      -848 -1/-256 fffffe810f382e10  uhci_poll_hub
 >      -833 -1/-256                0  key_timehandler
 >      -828 -1/-256 ffffffff81481b80  sleepq_timeout
 >      -822 -1/-256                0  nd6_timer
 >      -819 -1/-256                0  rt_timer_timer
 >      -811 -1/-256 ffff80000e836000  bnx_tick
 >      -811 -1/-256 ffff80000e830008  if_slowtimo
 >      -811 -1/-256 ffff80000e833008  if_slowtimo
 >      -811 -1/-256 ffff80000e836008  if_slowtimo
 >      -811 -1/-256 ffff80000e839008  if_slowtimo
 >      -811 -1/-256 ffff80000e830000  bnx_tick
 >      -798 -1/-256 ffff80000e839000  bnx_tick
 >      -784 -1/-256 fffffe822f5d0040  sleepq_timeout
 >      -561 -1/-256                0  vmem_rehash_all_kick
 >      -430 -1/-256 fffffe81f8e8f5d8  tcp_timer_rexmt
 >        13  0/107  fffffe810f372e10  uhci_poll_hub
 >        14  0/129  fffffe810f330e10  uhci_poll_hub
 >        43  0/163  fffffe810eb79aa0  sleepq_timeout
 >        13  0/182  fffffe810f351e10  uhci_poll_hub
 
 Suppose negative ticks means callout processing has stopped.
 
 Examined all softhandler on cpu0 and got:
 
 index             func  flags   func        ACTIVE,PENDING      LEVEL
 108 0xffffffff809c5cf0 0x3101   callout_softclock       AP      CLOCK
 113 0xffffffff809ee1ce 0x2102   xc__highpri_intr        A       SERIAL
 148 0xffffffff80a25ec8 0x1100   biointr                  P      BIO
 268 0xffffffff805f39e9 0x1002   comsoft                  P      SERIAL
 303 0xffffffff80705118 0x3303   ip6intr                 AP      NET
 
 If I get it right we have callout_softclock and ip6intr busy and with
 more work and we have xc__highpri_intr busy (there is a
 pserialize_perform running).
 
 Is it right the running ip6intr will block all other handlers as NET
 is the highest priority here?
 
 Any idea how to debug the current state of "softnet/0"
 
 --
 J. Hannken-Illjes - hannken%eis.cs.tu-bs.de@localhost - TU Braunschweig (Germany)