Tor 0.4.6.10 crashed with Update 165

smithchart · 30 March 2022 09:23

Hi,

I did update today my Ipfire from release 164 to 165, including tor to 0.4.6.10. After a reboot Tor didn’t start anymore.
Following message is in /var/log/messages from today:

Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Mar 30 10:57:37 Tor[17367]: Bug: Starting with guard context “default”
tor_bug_occurred_(): Bug: src/lib/evloop/workqueue.c:353: workerthread_new: This line should not have been reached. (on Tor 0.4.6.10 )
Bug: Tor 0.4.6.10: Line unexpectedly reached at workerthread_new at src/lib/evloop/workqueue.c:353. Stack trace: (on Tor 0.4.6.10 )
/usr/bin/tor(log_backtrace_impl+0x5b) [0x6335471dcb7b] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_bug_occurred_+0x16f) [0x6335471e7f6f] (on Tor 0.4.6.10 )
/usr/bin/tor(threadpool_new+0x207) [0x633547311ea7] (on Tor 0.4.6.10 )
/usr/bin/tor(cpu_init+0x8d) [0x6335472a7bed] (on Tor 0.4.6.10 )
/usr/bin/tor(run_tor_main_loop+0xe9) [0x63354715f5f9] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_run_main+0x1e5) [0x63354715fb45] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_main+0x4d) [0x63354715c0ad] (on Tor 0.4.6.10 )
/usr/bin/tor(main+0x1d) [0x63354715bbad] (on Tor 0.4.6.10 )
/lib/libc.so.6(+0x29590) [0x77abd0f9e590] (on Tor 0.4.6.10 )
/lib/libc.so.6(_libc_start_main+0x80) [0x77abd0f9e640] (on Tor 0.4.6.10 )
/usr/bin/tor(start+0x25) [0x63354715bcb5] (on Tor 0.4.6.10 )
Can’t launch worker thread.
tor_bug_occurred(): Bug: src/lib/evloop/workqueue.c:519: threadpool_start_threads: This line should not have been reached. (on Tor 0.4.6.10 )
Bug: Tor 0.4.6.10: Line unexpectedly reached at threadpool_start_threads at src/lib/evloop/workqueue.c:519. Stack trace: (on Tor 0.4.6.10 )
/usr/bin/tor(log_backtrace_impl+0x5b) [0x6335471dcb7b] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_bug_occurred+0x16f) [0x6335471e7f6f] (on Tor 0.4.6.10 )
/usr/bin/tor(threadpool_new+0x24c) [0x633547311eec] (on Tor 0.4.6.10 )
/usr/bin/tor(cpu_init+0x8d) [0x6335472a7bed] (on Tor 0.4.6.10 )
/usr/bin/tor(run_tor_main_loop+0xe9) [0x63354715f5f9] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_run_main+0x1e5) [0x63354715fb45] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_main+0x4d) [0x63354715c0ad] (on Tor 0.4.6.10 )
/usr/bin/tor(main+0x1d) [0x63354715bbad] (on Tor 0.4.6.10 )
/lib/libc.so.6(+0x29590) [0x77abd0f9e590] (on Tor 0.4.6.10 )
/lib/libc.so.6(_libc_start_main+0x80) [0x77abd0f9e640] (on Tor 0.4.6.10 )
/usr/bin/tor(start+0x25) [0x63354715bcb5] (on Tor 0.4.6.10 )
tor_bug_occurred(): Bug: src/lib/evloop/workqueue.c:563: threadpool_new: This line should not have been reached. (on Tor 0.4.6.10 )
Bug: Tor 0.4.6.10: Line unexpectedly reached at threadpool_new at src/lib/evloop/workqueue.c:563. Stack trace: (on Tor 0.4.6.10 )
/usr/bin/tor(log_backtrace_impl+0x5b) [0x6335471dcb7b] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_bug_occurred+0x16f) [0x6335471e7f6f] (on Tor 0.4.6.10 )
/usr/bin/tor(threadpool_new+0x27f) [0x633547311f1f] (on Tor 0.4.6.10 )
/usr/bin/tor(cpu_init+0x8d) [0x6335472a7bed] (on Tor 0.4.6.10 )
/usr/bin/tor(run_tor_main_loop+0xe9) [0x63354715f5f9] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_run_main+0x1e5) [0x63354715fb45] (on Tor 0.4.6.10 )
/usr/bin/tor(tor_main+0x4d) [0x63354715c0ad] (on Tor 0.4.6.10 )
/usr/bin/tor(main+0x1d) [0x63354715bbad] (on Tor 0.4.6.10 )
/lib/libc.so.6(+0x29590) [0x77abd0f9e590] (on Tor 0.4.6.10 )
/lib/libc.so.6(__libc_start_main+0x80) [0x77abd0f9e640] (on Tor 0.4.6.10 )
/usr/bin/tor(_start+0x25) [0x63354715bcb5] (on Tor 0.4.6.10 )

What is the issue here?

BR,

Thorsten

pmueller · 31 March 2022 07:31

Hi,

ah, zut alors, that was my fault again.

This is an issue with the Tor sandbox, which we enabled a while ago to increase security. With the toolchain changes in Core Update 165, Tor apparently uses an additional syscall, which it did not declare to use beforehand, so the sandbox denies it.

A patch is already available, and this issue is also being tracked as bug #12807.

For the moment, commenting out the

Sandbox 1

line in /var/ipfire/tor/torrc makes Tor work again. Please be aware that any change to the Tor configuration made via the web interface overwrites this.

Thanks, and best regards,
Peter Müller

smithchart · 31 March 2022 18:58

Hi Peter,
merci for the quick reply

Best regards
Thorsten

mumpitz · 31 March 2022 22:57

Why did the patch not find its way into today’s update?
Because after the update today to 166 it still does not work. I have now chosen the manual method.

jon · 31 March 2022 23:23

today’s update was an urgent update. There were only two items updated.

see this blog:

legwood · 1 April 2022 06:54

Thanks, for the fix.
i’ve the same problem

rainmaker · 1 April 2022 19:02

I also had this issue, and found the solution in this thread. Thank you! I haven’t visited here for a while as I’ve been using OpenBSD on my network edge for a couple of years. I see you have a new forum, I had to sign up again.

One thing I noticed today, in relation to the tor package in update 165: It preserved my existing (modified) torrc, which has tweaks for MyFamily amongst others. However what I didn’t notice is that, somehow, it also reset my advertised bandwidth to a mere few KB (note the ‘K’), from its previous 20Mbps. My advertised bandwidth score on the Metrics page had fallen to 0.

I only realised by chance when my relay probability and bandwidth consensus dropped significantly on the Tor Metrics page. After changing it back and reloading the config (again) we’re back in business. Check your configs just in case, folks!

pmueller · 3 April 2022 09:46

Hi,

for the records: It was discussed to include this fix and other into Core Update 166 as well, but in the end we decided not to delay the release of Core Update 166 any longer.

Upcoming Core Update 167 will be closed early next week, since it also contains security-relevant updates to some packages, so it should be available for testing soon.

Sorry for the delay, and best regards,
Peter Müller

pmueller · 10 April 2022 16:33

Hi all,

just a technical update on that issue: Apparently, Tor needs the rseq syscall as well, which the upstream patch did not cover, yet.

For IPFire, that has been fixed in this commit, and I will try to upstream it as soon as possible.

Core Update 167 is going to be available for testing within the next few days, so there should finally be an end in sight for this issue.

Thanks, and best regards,
Peter Müller

pmueller · 13 April 2022 21:06

pmueller · 25 April 2022 18:13

A post was split to a new topic: Tor does not connect