[Bug 224479] kernel panic in reboot+swapoff sys call

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=224479

            Bug ID: 224479
           Summary: kernel panic in reboot+swapoff sys call
           Product: Base System
           Version: CURRENT
          Hardware: Any
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: kern
          Assignee: freebsd-bugs at FreeBSD.org
          Reporter: wosch at FreeBSD.org

I built a new kernel on -current. The build was successfully, and I rebooted
the machine. To my surprise the machine hangs in a kernel panic during swapoff
sys call.

It turns out that if the machine used some swap, it will panic at the next
reboot. This is 100% reproducible.

How to repeat:

# make sure that the machine used some swap. Here I’m using a perl script which
will 
# grow at least a 1GB RAM big
$ perl -e '$a=`man tcsh`; for(0..3000) { $b.=$a}'

# check the swap usage, e.g.
$  top -b 2 | egrep ^Swap
Swap: 2500M Total, 18M Used, 2482M Free

# reboot the machine
$ reboot

on the console I get this:
swap_pager: I/O error - pagein failed; blkno 280694, size 4096, error 5
panic: swap_pager_force_pagein: read from swap failed

I saw this on two machines, one use a 1GB swap device and the other a 2GB swap
file
cat /etc/fstab 
/dev/vtbd0p3    /               ufs     rw      1       1

cat /etc/fstab 
md99    none    swap    sw,file=/var/swap/swap.0,late   0       0

This problem is new to me, I didn’t saw this 3 weeks ago in November. Using
swapoff/swapon on the command line works fine.

Workaround: before a reboot, disable swap with the swapoff command:
$ swapoff  -a
swapoff: removing /dev/md99 as swap device
$ reboot

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