[Solved] Firefox and Thunderbird bus errors

[mark9117]

As of tonight, Firefox and Thunderbird have bot failed to start. They are failing on a "bus error".

Information on the internet says to uninstall xulrunner, nspluginwrapper, and firefox. Of these 3 packages only firefox is installed on my system. Reinstalling firefox has not resolved the issue. I also tried moving ~/.mozilla/firefox. The directory was recreated when I tried to start firefox, but it failed on the same bus error.

Some information:

Code: Select all

rpm -qa *xulrun* *firefox* *mozilla* *flash* *plugin*
mozilla-filesystem-1.9-6.mga5
firefox-45.3.0-1.mga5
flash-imagewriter-2.4-4.mga5
flash-player-plugin-kde-11.2.202.635-1.mga5.nonfree
flash-player-plugin-11.2.202.635-1.mga5.nonfree
kmozillahelper-0.6.4-5.mga5


Any assistance out there?

Thanks.


Mark

[mark9117]

Quick addition here: Installed Geary and it suffers the same bus error. I'm assuming something in the basic structure of these applications is the same and it trips the bus error. What update broke these things?

Additional: Installed Evolution and it fails to load on a bus error. Any idea what is going on?

[mark9117]

I was able to resolve the issue with Firefox by downloading and installing Firefox-49. It seems to have no trouble.

Thunderbird, on the other hand, seems to be irredeemable. I downloaded the archive and unpacked it into a directory under /opt. It tries to run, but I get the same failure I had with the rpm version of Thunderbird I installed from repos:

Code: Select all

$ thunderbird
[calBackendLoader] Using libical backend at /usr/lib64/mozilla/extensions/{3550f703-e582-4d05-9a08-453d09bdfdc6}/{e2fda1a4-762b-4020-b5ad-a41df1933103}/components/libical-manifest
Bus error


I did some poking around at the Mozilla Bugzilla site. No actual help there. Similar bug is listed but it's several years old.

This is soaking up a lot of time and energy. I'm also spending a lot of time configuring Claws into a mail reader that I can live with. It's pretty unsatisfying.

Anybody have any ideas out there?

[doktor5000]

Did you reboot after the update, and what packages were installed before this started to occur? rpm -qa --last should help with the latter ...

[mark9117]

Did you reboot after the update, and what packages were installed before this started to occur? rpm -qa --last should help with the latter ...

I did reboot soon after I realized there was trouble.

A partial listing of the output of "rpm -qa --last" is here. It's only partial because it overran the scrollback buffer. I'd be curious as to what you make of it.

Thanks for the attention, BTW. I'd really love to know what's broken in all this.

Mark

[mark9117]

Also, and I meant to include this with the initial information on this situation, this is the output of the Thunderbird executable when invoked in a konsole window as root:

Code: Select all

# thunderbird

(thunderbird:22105): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken.

(thunderbird:22105): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken.

(thunderbird:22105): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken.

(thunderbird:22105): GConf-WARNING **: Client failed to connect to the D-BUS daemon:
Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken.
[calBackendLoader] Using libical backend at /usr/lib64/mozilla/extensions/{3550f703-e582-4d05-9a08-453d09bdfdc6}/{e2fda1a4-762b-4020-b5ad-a41df1933103}/components/libical-manifest
Bus error


No idea what that actually means.

Mark

[doktor5000]

That means there's no dbus session bus running for root, hence you get some dbus errors due to that - and FWIW you should never run regular GUI programs as root.

To be clear, your're running this locally on your box, not inside some ssh session or so? Have you tried renaming ~/.mozilla and also running in safe mode, disabling all addons and theme via thunderbird --safe-mode ?

Can you also please show the output of

Code: Select all

urpmq --list-media active --list-url
urpmq --not-available


[mark9117]

That means there's no dbus session bus running for root, hence you get some dbus errors due to that - and FWIW you should never run regular GUI programs as root.

Yeah, I'm just desperately looking for something that will get T-bird working. As little as I use these thing, I'm struggling to survive without it.

To be clear, your're running this locally on your box, not inside some ssh session or so? Have you tried renaming ~/.mozilla and also running in safe mode, disabling all addons and theme via thunderbird --safe-mode ?

Right. I'm running it on my own desktop. It's nearly impossible to troubleshoot because I can't get it to start at all. I've tried renaming ~/.mozilla to get it out of the way. It makes no difference. Starting in safe mode isn't an option. It just fails out immediately:

Code: Select all

$ thunderbird -safe-mode
Bus error


Can you also please show the output of

Code: Select all

urpmq --list-media active --list-url
urpmq --not-available


But of course:

Code: Select all

# urpmq --list-media active --list-url
Core Release2 http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/core/release
Core Updates http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/core/updates
Nonfree Release2 http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/nonfree/release
Nonfree Updates http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/nonfree/updates
Tainted Release http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/tainted/release
Tainted Updates http://mirrors.kernel.org/mageia/distrib/5/x86_64/media/tainted/updates
Core 32bit Release http://mirrors.kernel.org/mageia/distrib/5/i586/media/core/release
Core 32bit Updates http://mirrors.kernel.org/mageia/distrib/5/i586/media/core/updates
Nonfree 32bit Release http://mirrors.kernel.org/mageia/distrib/5/i586/media/nonfree/release
Nonfree 32bit Updates http://mirrors.kernel.org/mageia/distrib/5/i586/media/nonfree/updates
chrome_x86_64 http://dl.google.com/linux/chrome/rpm/stable/x86_64


Code: Select all

# urpmq --not-available
SpiderOak-5.1.6-1.x86_64

Still looking for solutions. I appreciate your assistance.

Thank you sir!

Mark

[mark9117]

Finally found a work around for this. At least I think it's a work around.

I went back to Mozilla and downloaded a fresh copy of Thunderbird-45.3. This also happens to be the current version of Thunderbird downloaded from repo's.

Code: Select all

# rpm -q thunderbird
thunderbird-45.3.0-2.mga5

I wrestled with it enough to get it to read the correct profile under my home directory (~/.thunderbird/profile.ini) and it functioned as expected.

Not sure what's going on with the version installed from repo's but it's still throwing the Buss Error.

Just wanted to get that out there. I'll mark this resolved because I am using my preferred email client in my preferred Linux distro. I'm assuming it will be resolved in a more genuine way in the next Mageia release.

Mark

[doktor5000]

I've marked it [DONE] as it's in no way resolved, only workarounded ...

[mark9117]

I've marked it [DONE] as it's in no way resolved, only workarounded ...

Right. That's why I elected for "Resolved" rather than "Solved". But tomato - tomahto, I'm good with "Done".

[wintpe]

just as an addon Mark, i run thunderbird, from the repos, and mine updated and worked fine the other day.

so as Doktor suggested , its not firefox/thunderbird its something perhaps wrong with dbus on your host,

more info required otherwise any advise would just be a guess.

regards peter

[doktor5000]

Actually a "bus error" is caused by the program getting SIGBUS signal, totally unrelated to dbus, check
http://stackoverflow.com/questions/2089 ... -x86-linux or http://stackoverflow.com/questions/2124 ... -bus-error
That being said, I've only seen this reported one time so far and it was not really reproducible:
https://bugs.mageia.org/show_bug.cgi?id=14704

And without a full backtrace chances to successfully submit this upstream at mozilla are near to zero as it would be quite useless without.

[jiml8]

SIGBUS is a pretty serious error, and usually occurs because the program tries to access a nonexistent address, or access beyond the end of a mmap region. It can also happen sometimes if the kernel is under memory pressure and the OOM task killer is disabled. Might also occur due to a failed I/O request.

How much memory does your system have? Take a look at the usage of your swapfile, and how your virtual memory handler is configured (look in /etc/sysctl.conf, and try as root the command sysctl -a.

For a program with the maturity of thunderbird to fail that way implies there is some other problem; I would be surprised if it was an upstream problem. Most likely there is some feature of your system that is getting you into trouble.

Tweaking the virtual memory manager is definitely an advanced topic, but can yield huge benefits in the behavior of your system - and I strongly suspect your problem is here somewhere.

edit:

If your system has a reasonably fast drive and more than 8 GB of RAM, you will probably gain both performance and reliability by adding these lines to /etc/sysctl.conf :

Code: Select all

vm.dirty_background_ratio=2
vm.dirty_ratio=4
vm.overcommit_ratio=80
vm.overcommit_memory=2


and then rebooting.

I would give good odds that this will fix your problem with thunderbird. I do suggest you study these values a bit to understand what they do.

[mark9117]

SIGBUS is a pretty serious error, and usually occurs because the program tries to access a nonexistent address, or access beyond the end of a mmap region. It can also happen sometimes if the kernel is under memory pressure and the OOM task killer is disabled. Might also occur due to a failed I/O request.How much memory does your system have? Take a look at the usage of your swapfile, and how your virtual memory handler is configured (look in /etc/sysctl.conf, and try as root the command sysctl -a.

For a program with the maturity of thunderbird to fail that way implies there is some other problem; I would be surprised if it was an upstream problem. Most likely there is some feature of your system that is getting you into trouble.

Memory looks a little like this:

Code: Select all

$ free
             total       used       free     shared    buffers     cached
Mem:       8046788    7672840     373948     293968     321960    4731204
-/+ buffers/cache:    2619676    5427112
Swap:      2040248     681416    1358832


I only have 8Gb of memory installed and the system HD is an older Western Digital WD740ADFD-00. It's a 10000 RPM drive with 16MB cache. Not sure how that rates against a modern standard for "fast".

sysctl -a

Code: Select all

.
.
.
vm.admin_reserve_kbytes = 8192
vm.block_dump = 0
vm.compact_unevictable_allowed = 1
vm.dirty_background_bytes = 0
vm.dirty_background_ratio = 10
vm.dirty_bytes = 0
vm.dirty_expire_centisecs = 3000
vm.dirty_ratio = 20
vm.dirty_writeback_centisecs = 500
vm.dirtytime_expire_seconds = 43200
vm.drop_caches = 0
vm.extfrag_threshold = 500
vm.hugepages_treat_as_movable = 0
vm.hugetlb_shm_group = 0
vm.laptop_mode = 0
vm.legacy_va_layout = 0
vm.lowmem_reserve_ratio = 256   256   32
vm.max_map_count = 65530
vm.memory_failure_early_kill = 0
vm.memory_failure_recovery = 1
vm.min_free_kbytes = 67584
vm.min_slab_ratio = 5
vm.min_unmapped_ratio = 1
vm.mmap_min_addr = 4096
vm.nr_hugepages = 0
vm.nr_hugepages_mempolicy = 0
vm.nr_overcommit_hugepages = 0
vm.nr_pdflush_threads = 0
vm.numa_zonelist_order = default
vm.oom_dump_tasks = 1
vm.oom_kill_allocating_task = 0
vm.overcommit_kbytes = 0
vm.overcommit_memory = 0
vm.overcommit_ratio = 50
vm.page-cluster = 3
vm.panic_on_oom = 0
vm.percpu_pagelist_fraction = 0
vm.stat_interval = 1
vm.swappiness = 60
vm.user_reserve_kbytes = 131072
vm.vfs_cache_pressure = 100
vm.zone_reclaim_mode = 0

Code: Select all

# System default settings live in /usr/lib/sysctl.d/00-system.conf.
# To override those settings, enter new settings here, or in an /etc/sysctl.d/<name>.conf file
#
# For more information, see sysctl.conf(5) and sysctl.d(5).
net.ipv4.icmp_ignore_bogus_error_responses=1
net.ipv4.icmp_echo_ignore_broadcasts=0
net.ipv4.icmp_echo_ignore_all=0
net.ipv4.conf.all.rp_filter=1
net.ipv4.conf.all.log_martians=1

That's all network related gibberish to me.

Tweaking the virtual memory manager is definitely an advanced topic, but can yield huge benefits in the behavior of your system - and I strongly suspect your problem is here somewhere.

edit:

If your system has a reasonably fast drive and more than 8 GB of RAM, you will probably gain both performance and reliability by adding these lines to /etc/sysctl.conf :

Code: Select all

vm.dirty_background_ratio=2
vm.dirty_ratio=4
vm.overcommit_ratio=80
vm.overcommit_memory=2


and then rebooting.

I would give good odds that this will fix your problem with thunderbird. I do suggest you study these values a bit to understand what they do.

I appreciate the time and attention jiml8. It's my bedtime and I need to be fresh for work tomorrow. I'll see if I can find some time to sort this out when consciousness is not at such a premium. If you think of anything else, please feel free to let me know.

Mark

[jiml8]

Your hard drive is quite decently fast.

That system does not appear to be under memory pressure...at least, not at the time that you executed those commands. You have too many I/O buffers specified IMO...but this is the default setting in Mageia Linux. I will note that I have 32 GB of RAM on my workstation, and with my settings I have fewer buffers than you do (you show 321960 while I have 304096). This is set using the vm.dirty sysctls, however it should not be a big deal for you.

You are actually only using about 1/3 of your RAM; the rest is easily fungible if needed. Your swap is about 1/3 full also.

I will note that these values can change VERY quickly if your system is busy, and without profiling it you can't tell what your peaks are like. Dramatic changes in small fractions of a second are not uncommon.

However, the default setup is usually adequate for a machine with 8 GB of memory, and your setup (based on the output of sysctl -a) is default. Also, your system looks perfectly normal at the time you executed those commands.

Nonetheless, a SIGBUS error in that program strongly suggests there is an issue with your system. Maybe not with VM...possibly some I/O error. Are you experiencing any other discrepancies?

[jiml8]

I should also point out that running the command: "killall -SIGBUS <programname>" will cause <programname> to terminate with a SIGBUS error, unless the program is specifically configured to ignore that signal. This is another possibility, if something in the system is using either the kill or killall. Who, and why? I have no idea. However this is another way to encounter the problem.

You might get a clue by using the strace program. (man strace to learn about it).

[jiml8]

I guess there is on other point I should make about the "free" command.

Free does not always properly identify the memory that is in use, and the memory that is cached. It took me awhile to learn this, but it is so.

I run a lot of virtual machines on my workstation. At this writing, I have the following VMs up and running in my Mageia host: Windows 2000, Windows 7 Pro, Windows 10, Linux Mint 17.3, OpenSuse 13.2, FreeBSD 8.4, and two copies of FreeBSD 10.2. Yet, in spite of this load, free on my system reports the following:

Code: Select all

jiml@dadsbox:jiml> free
             total       used       free     shared    buffers     cached
Mem:      32863968   32248756     615212   21216404     310552   23772856
-/+ buffers/cache:    8165348   24698620
Swap:     36909400    2783628   34125772


Thus, free shows that I have only about 8 GB in use, and 23.7 GB cached. This is wrong. In fact, around 21 GB of that cached RAM is actually assigned to the various VMs. For instance, Win7 gets about 3.2 GB, Win10 gets about 3.2 GB, the Linux systems each get 4GB, and the rest goes to the other VMs.

So, in fact, my workstation does frequently experience memory pressure, which is why I have 35 GB of swap defined for it, and which is why I have made those tweaks I mentioned earlier. The combination I show makes this system rock solid; it never crashes (unless my NAS goes down, but that's another story...).

So, the point is that though your system does not appear to be under memory pressure based on the commands you ran, it still could be under pressure, just depending on the details. If you run virtual machines, then the output from free cannot be trusted. it could be there are other programs or systems that also make free unreliable, but offhand I don't know what they might be.

[mark9117]

Nonetheless, a SIGBUS error in that program strongly suggests there is an issue with your system. Maybe not with VM...possibly some I/O error. Are you experiencing any other discrepancies?

The only other things I've got going on here, aside from failure for the repo version Thunderbird to run, is that the repo version of all Mozilla programs halt on a bus error. I am also seeing Geary fail in the same way. I assume my Mozilla apps share a framework with Geary.

Also, I can't start Mandriva Control Center, but that seems to be an established bug: https://bugs.mageia.org/show_bug.cgi?id=18499. You can also read about that adventure here: viewtopic.php?f=7&t=10999.

Beyond that, this desktop is incident free.

Mark

[mark9117]

I should also point out that running the command: "killall -SIGBUS <programname>" will cause <programname> to terminate with a SIGBUS error, unless the program is specifically configured to ignore that signal. This is another possibility, if something in the system is using either the kill or killall. Who, and why? I have no idea. However this is another way to encounter the problem.

You might get a clue by using the strace program. (man strace to learn about it).

Ran strace on Thunderbird and Geary. The output of the Thunderbird strace is here: http://pastebin.com/9qeJyxAs . I noticed that both T-bird and Geary seem to grind to a halt after a number of references to the flashplayer plugin. I haven't seen anything that points to flashplayer plugin as an obvious culprit, but it seems to be implicated. And I wouldn't know an obvious culprit in strace if it fell on me.

So threre's that.

[mark9117]

So, the point is that though your system does not appear to be under memory pressure based on the commands you ran, it still could be under pressure, just depending on the details. If you run virtual machines, then the output from free cannot be trusted. it could be there are other programs or systems that also make free unreliable, but offhand I don't know what they might be.

This machine is hardly a development platform. I don't even claim to code. I should mention that I have Gkrellm running on this machine at all times along with the three other machines on my network that provide services here (at this point in my life, these machines basically support my HTPC). At any rate I tend to watch CPU load, memory, load on the network interfaces and I/O on all these machines. It helps to give me an idea about what's going on with my network, which as I said above, really isn't all that much. I do not run virtual machines at all due to the abundance of computing resources I have available to allow me to watch MythTV and the odd movie.

And the HTPC has no trouble with Firefox or anything else.

But the point is I'm not seeing any odd or unexpected spikes in memory usage or anything else on this machine.

[mark9117]

Was poking around in my programs this evening and discovered that the things not working previously are now working as near as I can tell. The stock versions of Firefox, Thunderbird, Geary, etc. are all opening and functioning as expected.

Marking this "Solved" though I can't say what solved it.


Mark