I've filed this bug at bugreport.apple.com - #5614373. This is a
security bug outside of our code, but X11 users should be aware of it
because it's somewhat related...
---
Some users on the X11 mailing list reported that starting an xterm can
cause that process to not display and consume 100% cpu. I haven't
reliably encountered this problem until today.
Here's a backtrace of the problem when it occurred... it seems to be
looping in ioctl() after ptytty_unix::get()
Attaching to program: `/opt/local/bin/urxvt', process 39730.
Reading symbols for shared libraries ++++++++++....... done
0x90c51100 in ioctl ()
(gdb) bt
#0 0x90c51100 in ioctl ()
#1 0x000154a4 in ptytty_unix::get ()
#2 0x0000f1b8 in rxvt_term::run_command ()
#3 0x00013588 in rxvt_term::init ()
#4 0x000027a0 in main ()
(gdb) n
Single stepping until exit from function ioctl,
which has no line number information.
<never leaves ioctl()>
Now, this is interesting... I noticed a timed out ssh connect to my
leopard box a few times and then noticed sshd consuming 100% as well!
I'm fairly certain these events are related (as they both are most
certainly doing the same process to get a tty).
This bug technically has a workaround (kill the process and retry),
but I think it should be classified as security for the following
reason:
A user with access to connect via ssh can DoS the server by spawning
multiple ssh connections thus spawning multiple processes which can
deluge the OS's process wait list.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
X11-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/x11-users/email@hidden
