[Bug 209682] [panic] [netinet] arptimer race

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=209682

--- Comment #12 from Dave Baukus <daveb at spectralogic.com> ---
Back in March, 2018, I added hselasky's patch to our 11.1 base. I and have not
seen this issues since. I did not add the patch to our 11.0 release; this bug
reared its head in our 11.0 based release stream.

I do know what triggers this event; I can not reproduce at will, but its seems
that Hans Petter's patch is worthy.

#6  0xffffffff8081908c in trap (frame=0xfffffe07c4ff37b0) at
sys/amd64/amd64/trap.c:442
#7  0xffffffff807fd341 in calltrap () at sys/amd64/amd64/exception.S:236
#8  0xffffffff80625e64 in arptimer (arg=<value optimized out>) at
sys/netinet/if_ether.c:265
#9  0xffffffff8050dd0a in softclock_call_cc (c=<value optimized out>, cc=<value
optimized out>, direct=<value optimized out>)

Again the ifp from lle->lle_tbl->llt_ifp is invalid:

p *$lle->lle_tbl
$4 = {
  llt_link = {
    sle_next = 0x0
  }, 
  llt_af = 0, 
  llt_hsize = 0, 
  lle_head = 0xfffff800286a6000, 
  llt_ifp = 0x2, 
  llt_lookup = 0xc1, 
  llt_alloc_entry = 0, 
  llt_delete_entry = 0x9, 
  llt_prefix_free = 0xc0, 
  llt_dump_entry = 0x1667, 
  llt_hash = 0, 
  llt_match_prefix = 0x1d97, 
  llt_free_entry = 0x1a667b, 
  llt_foreach_entry = 0x1, 
  llt_link_entry = 0, 
  llt_unlink_entry = 0, 
  llt_fill_sa_entry = 0, 
  llt_free_tbl = 0
}

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