port-mac68k/57107: kernel panic on -current when configuring network with sn(4) on mac68k

[netbsd@linear.network]

>Number:         57107
>Category:       port-mac68k
>Synopsis:       kernel panic on -current when configuring network with sn(4) on mac68k
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-mac68k-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Dec 11 08:25:00 +0000 2022
>Originator:     linear cannon
>Release:        -current 202212102100
>Organization:
>Environment:
>Description:
on mac68k on a quadra 700 (sn(4) network driver), the following panic occurs when attempting to configure network during installation:

sn0: executing: /libexec/dhcpcd-rupool-run-hooks CARRIER
sn0: executing: /libexec/dhcpcd-rupool-ru
pool_redzone_check: [mclpl] 0x00 != 0x97
[  58.1755711] cpu0: Begin traceback...
[  58.1755711] ?(?)
[  58.2357605] db_panic(a54040,ffffffff,ffffffff,9e748c,59ade64) at 0
[  58.3024361] vpanic(225620,59ade70,59ade84,149730,225620) + 162
[  58.4357886] panic(225620,22673b,229da9,0,97) + c
[  58.5691411] pool_redzone_check.part.0(?)
[  58.6524809] eventfd_filt_read(9e748c, a54040) + 74
[  58.7191602] pool_cache_put_paddr(9e748c,a54040,ffffffff) + 48
[  58.8691743] m_ext_free(?)
[  58.9525255] unp_drop(a52124) + 17a
[  59.0191998] m_free(a52124) + 3c
[  59.0859379] m_freem(a52124,2000,148d80,148b74,38ea) + 16
[  59.2025842] ether_input (a98a40,a52124,a98a40,a52124,a9c400,0) + 506
[  59.3025867] if_percpuq_softint(a14de0,fbb70,40,59a0170,5bdea0) + 9c
[  59.3859762] softint_thread(59a0170) + 264
[  59.4525970] lwp_trampoline() + e
[  59.4525970] cpu0: End traceback...
Stopped in pid 0.3 (system) at netbsd:cpu_Debugger+0x6:    unlk    a6
db> 

(panic was OCRd and hand-edited, there may be minor mistakes i missed)
>How-To-Repeat:
on a 68k mac with a network device supported by the sn(4) driver, boot the installation kernel, and attempt to configure the network.

the panic happens very reliably during configuration, and the stack trace output is, as far as i can tell, identical each time, save for timestamps.


>Fix: