Date: Sat, 09 Dec 2017 14:08:49 -0700 From: vagabond <vagabond@blackfoot.net> To: Freebsd Questions <questions@freebsd.org> Subject: thunderbird causing system crash, need guidance Message-ID: <714b3ea002f0ec1419e3e48340906d84@blackfoot.net>
next in thread | raw e-mail | index | archive | help
Hi all,
Looking for guidance diagnosing a system crash caused by
attempting to start Thunderbird.
10.3-RELEASE-p20 FreeBSD 10.3-RELEASE-p20 #0: Wed Jul 12 03:13:07
UTC 2017 root@amd64-builder.daemonology.net:/usr/obj/usr
/src/sys/GENERIC amd64
After running flawlessly for over a month, I started having
sluggish behavior. Since this is a known problem with firefox
56, I exited and restarted it several times over the course of a
few days. Then yesterday (2017-12-08) the system hung and
crashed.
I have narrowed the cause down to Thunderbird 52.4.0, or at least
something associated with it.
The system + X seem to still run fine; openoffice, firefox, gimp.
When I attempt to start t-bird the cursor disappears almost
immediately, followed by a long wait with the display apparently
frozen, and then results in a crash and reboot.
It seems t-bird should crash/dump core without crashing the
system if it was just a t-bird problem, even if it's a bad binary
image?
I originally had crash dumps disabled, so changed
dumpdev="NO" to "AUTO" in rc.conf
but still no dump in /var/crash
only thing in /var/crash is minfree, which says "2048"
Normally 16GB RAM but I have removed 3 of the 4 sticks so only
4G now and still crashes.
An issue for getting dumps may be my sys config, as /tmp and swap
are memdisks, but I'm not sure as I don't know how it really
works.
The disk has no swap or tmp partition; I'm not sure how or if I
can modify fstab or the config to get swap on disk for a crash
dump save.
From fstab:
/dev/ufs/hd250G1root / ufs rw,noatime 1 1
/dev/ufs/hd250G1var /var ufs rw,noatime 2 2
/dev/ufs/hd250G1usr /usr ufs rw,noatime 7 3
tmpfs /tmp tmpfs rw,mode=01777 0 0
md99 none swap sw,file=/usr/swap/swap,late 0
0
/var is 16G
It seems like it may be corrupted disk data, but I'm wondering if
there's a good way to diagnose that; thought about rebuilding
t-bird but wanted to try to isolate the problem first.
Before rebuilding t-bird would it be wise to rename the t-bird
binary and associated libs so any potentially bad disk sectors
are still held?
When the system started being particularly unresponsive, I exited
firefox and restarted it (several times over the course of a few
days). Then t-bird was slow, so I exited and restarted it.
I *think* that's when the system first crashed.
In the ~.thunderbird/xxx.default (profile) directory, the last
date on any files is the "lock" symlink, from (I think) the first
crash.
Other files show times 27 min earlier, which may be the last time
t-bird semi-successfully started up.
Interestingly, deleting the "lock" symlink and attempting to
restart t-bird results in the "lock" symlink being recreated with
the same (old!) timestamp, Dec 8 21:08. The latest file before
that is panacea.dat at Dec 8 20:50.
There's a chance this is caused by an incompatible library;
I've rebuilt and updated several ports over the past month, and I
don't know if I restarted t-bird during that period.
But again, I would expect t-bird to crash but not the system.
Any and all thoughts welcome.
Thanks,
Gary
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?714b3ea002f0ec1419e3e48340906d84>