[tor-bugs] #6029 [Tor Relay]: relay crash in libcrypto (tor_tls_handshake)
Tor Bug Tracker & Wiki
torproject-admin at torproject.org
Fri Jun 1 14:30:34 UTC 2012
#6029: relay crash in libcrypto (tor_tls_handshake)
-----------------------+----------------------------------------------------
Reporter: ln5 | Owner:
Type: defect | Status: new
Priority: major | Milestone: Tor: 0.2.3.x-final
Component: Tor Relay | Version: Tor: 0.2.3.15-alpha
Keywords: | Parent:
Points: | Actualpoints:
-----------------------+----------------------------------------------------
Comment(by ln5):
Replying to [comment:2 nickm]:
> If the crash isn't always in the same place, I'd suspect a memory
corruption issue.
Forgot to mention that.
First two crashes, before rebuilding with symbols and w/o gcc hardening
options:
May 31 04:24:44 tor kernel: [9547180.260104] tor[32318]: segfault at
7f328fff8000 ip 00007f32b876e5d8 sp 00007fffd3eaac60 error 4 in
libcrypto.so.1.0.0[7f32b86f8000+1b2000]
Jun 1 02:24:13 tor kernel: [9626147.616888] tor[20728]: segfault at
7fc521325000 ip 00007fc53f4fd5dd sp 00007fffc4c89130 error 4 in
libcrypto.so.1.0.0[7fc53f487000+1b2000]
Same offset in libcrypto.so.
In the last crash, I unfortunately don't know where libcrypto was
loaded (when running tor in gdb). Strangely enough, nm -g
libcrypto.so shows nothing with a higher offset than '0000000000179320
R RSA_version'. I don't understand this.
I will 'continue' in gdb after the next crash and hopefully get the
kernel segfault printout.
I just realised that this machine is a bit unhappy about something
related to TCP. The kernel said
WARNING: at /build/buildd/linux-2.6.32/net/ipv4/tcp_input.c:2919
tcp_fastretrans_alert+0x3f9/0xd90()
yesterday (not at the time of a tor crash). Maybe I should i) check
RAM and ii) benchmark TCP throughput excluding tor before we spend
more time on this.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6029#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
More information about the tor-bugs
mailing list