Re: Panic on multiuser boot after dhcpcd with msk(4)

[Jason Thorpe <thorpej%me.com@localhost>]

Actually, I just got confirmation from nick that my patch fixes the problem, so I’ll check it in shortly.

-- thorpej
Sent from my iPhone.

> On Feb 7, 2020, at 12:35 AM, Jason Thorpe <thorpej%me.com@localhost> wrote:
> 
> I have a potential fix for that that I can send you shortly ... the link state change handler needs to be a work queue instead of a soft int.  Can you test it?
> 
> -- thorpej
> Sent from my iPhone.
> 
>> On Feb 6, 2020, at 11:41 PM, Jaromír Doleček <jaromir.dolecek%gmail.com@localhost> wrote:
>> 
>> Hi,
>> 
>> I get repeatable panic with uptodate -current when system goes multiuser:
>> 
>> Adding interface aliases:.
>> Waiting for DAD to complete for statically configured addresses...
>> [  10.9328087] wm0: link state UP (was DOWN)
>> Starting dhcpcd.
>> [  12.8156398] msk0: link state DOWN (was UNKNOWN)
>> [  12.8156398] panic: assert_sleepable: softint caller=0xffffffff8051119e
>> [  12.8301938] cpu0: Begin traceback...
>> [  12.8301938] vpanic() at netbsd:vpanic+0x146
>> [  12.8361159] snprintf() at netbsd:snprintf
>> [  12.8361159] assert_sleepable() at netbsd:assert_sleepable+0xbf
>> [  12.8456876] percpu_foreach() at netbsd:percpu_foreach+0x1f
>> [  12.8456876] if_stats_to_if_data() at netbsd:if_stats_to_if_data+0x5f
>> [  12.8579148] if_export_if_data() at netbsd:if_export_if_data+0x15
>> [  12.8671028] rt_ifmsg() at netbsd:rt_ifmsg+0x68
>> [  12.8671028] if_link_state_change_softint() at
>> netbsd:if_link_state_change_softint+0xc0
>> [  12.8765414] if_link_state_change_si() at netbsd:if_link_state_change_si+0x73
>> [  12.8876237] softint_dispatch() at netbsd:softint_dispatch+0x345
>> address 0xffffce80ac72e0b8 is invalid
>> address 0xffffce80ac72e0b0 is invalid
>> address 0xffffce80ac72e0c0 is invalid
>> address 0xffffce80ac72e0b8 is invalid
>> address 0xffffce80ac72e0c8 is invalid
>> address 0xffffce80ac72e0c0 is invalid
>> address 0xffffce80ac72e0d0 is invalid
>> address 0xffffce80ac72e0c8 is invalid
>> [  12.9158149] DDB lost frame for netbsd:Xsoftintr+0x4f, trying
>> 0xffffce80ac72dff0
>> [  12.9258978] Xsoftintr() at netbsd:Xsoftintr+0x4f
>> [  12.9258978] --- interrupt ---
>> address 0xffffce80ac72e0c8 is invalid
>> address 0xffffce80ac72e080 is invalid
>> [  12.9382836] 0:
>> [  12.9468182] cpu0: End traceback...
>> [  12.9468182] fatal breakpoint trap in supervisor mode
>> [  12.9468182] trap type 1 code 0 rip 0xffffffff8021e3a5 cs 0x8 rflags
>> 0x202 cr2 0x7ea43f14f000 ilevel 0x6 rsp 0xffffce80ac72dc50
>> [  12.9676785] curlwp 0xffffef76c9615480 pid 0.3 lowest kstack
>> 0xffffce80ac7292c0
>> Stopped in pid 0.3 (system) at  netbsd:breakpoint+0x5:  leave
>> db{0}>
>> 
>> Any ideas what could be wrong? Maybe msk(4) is using some interface wrongly?
>> 
>> Jaromir