Re: kern/54908: kernel semi-lock-up

[Andrew Doran <ad%netbsd.org@localhost>]

The following reply was made to PR kern/54908; it has been noted by GNATS.

From: Andrew Doran <ad%netbsd.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/54908: kernel semi-lock-up
Date: Wed, 29 Jan 2020 22:12:42 +0000

 > db{0}> bt/a 92e914c0
 > trace: pid 0 lid 67 at 0x9bdcae54
 > 0x9bdcae54: netbsd:dosoftints+0xc
 > 0x9bdcaed4: netbsd:irq_entry+0x88
 > 0x9bdcafac: netbsd:uvm_pageout+0x178
 > db{0}> bt/a 929f7940
 > trace: pid 0 lid 20 at 0x80b16f3c
 > 0x80b16f3c: netbsd:cpu_idle+0x48
 > 0x80b16f54: netbsd:sched_curcpu_runnable_p+0x24
 > 0x80b16f94: netbsd:idle_loop+0x144
 
 > --- tf 0x80b16fb0 ---
 > db{0}> bt/a 928d8880
 > trace: pid 0 lid 5 at 0x9ba5beac
 > 0x9ba5beac: 92e72050
 > Bad frame pointer: 0x92b8f948
 
 I dunno about arm but on x86 your best chance of getting a backtrace from a
 running thread is straight "bt", assuming you are on the same CPU.
 
 I see it's in softclk/0.  If the same happens again, is there any movement
 from the callout evcnts in "show event"?  Would be interested to see what
 commonality there is if it happens again.
 
 A CPU hanging up in a soft interrupt often seems to make the machine half
 deadlocked in my recent experience..
 
 Andrew