discuss segfault on re-adding a meeting

[geofft]

I saw the following on my Squeeze amd64 server:

$ catchsegv discuss
Discuss version 1.7.  Type '?' for a list of commands.

discuss:  am charon.mit.edu:/var/spool/discuss/cfs
Meeting The Cryptographic File System users list already exists.
Do you wish to delete the old one and add the new one? yes
Segmentation fault

...

Backtrace:
/lib/libc.so.6(+0x10d35a)[0x7f028384d35a]
discuss[0x40c154]
discuss[0x406b03]
discuss[0x406df5]
discuss[0x407657]
/lib/libss.so.2(+0x2eda)[0x7f0283cbbeda]
/lib/libss.so.2(ss_execute_line+0x95)[0x7f0283cbbfc5]
/lib/libss.so.2(ss_listen+0x18f)[0x7f0283cbc3ff]
discuss[0x402fb8]
/lib/libc.so.6(__libc_start_main+0xfd)[0x7f028375ec4d]
discuss[0x402a39]

This doesn't happen when re-adding, say, the debian-devel meeting.

[jdreed]

So, uh, this has one of the longer "long names". Is it dependent on the length of the meeting name?

For maximum awesomeness, on Natty:

discuss:  am charon:/usr/spool/discuss/cfs
discuss:  am charon:/var/spool/discuss/cfs
*** buffer overflow detected ***: discuss terminated
======= Backtrace: =========
/lib/x86_64-linux-gnu/libc.so.6(__fortify_fail+0x37)[0x7ffa385661d7]
/lib/x86_64-linux-gnu/libc.so.6(+0xfd0f0)[0x7ffa385650f0]
/lib/x86_64-linux-gnu/libc.so.6(+0xfc569)[0x7ffa38564569]
/lib/x86_64-linux-gnu/libc.so.6(_IO_default_xsputn+0xd8)[0x7ffa384dcb98]
/lib/x86_64-linux-gnu/libc.so.6(_IO_vfprintf+0x1a12)[0x7ffa384af272]
/lib/x86_64-linux-gnu/libc.so.6(__vsprintf_chk+0x94)[0x7ffa38564604]
/lib/x86_64-linux-gnu/libc.so.6(__sprintf_chk+0x7a)[0x7ffa3856454a]
discuss[0x406e9e]
discuss[0x407134]
/lib/x86_64-linux-gnu/libss.so.2(+0x2ada)[0x7ffa38a16ada]
/lib/x86_64-linux-gnu/libss.so.2(ss_execute_line+0x95)[0x7ffa38a16c55]
/lib/x86_64-linux-gnu/libss.so.2(ss_listen+0x1a8)[0x7ffa38a16ff8]
discuss[0x402c28]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xff)[0x7ffa38486eff]
discuss[0x402889]

[mitchb]

Is it dependent on the length of the meeting name?

I hate the world. Yes. See clients/addmtg.c:371,385
The question about whether to punt the old meeting and add the new one is being sprintf'd into a 100 character array. ... ... ... *glare*

[jdreed]

r25764