Comment 156 for bug 585765

I've had this problem and I've reliably strace'd the offending thread (usually via SSH).

It seems to be locked on the general pattern of an futex with a NULL timeout:
futex( ... , FUTEX_WAIT_PRIVATE, NULL , ... ,

Backtracing the process shows from gdb:
0: __kernel_vsyscall ()
1: ?? () in libc6.so.6
Backtrace stopped: Corrupt stack.

I've been able to stroke the bug by compiling Enlightenment (DR17) with the Eweather module, setting its provider to "google", clicking "apply" and then "close". This will lock the system fairly reliably.

It seems that if init gets this futex call down the line, it will /stall/ the system.

I suspected it was bad RAM, so I ran Memtest86 and it came up with bad RAM at first, so I swapped out and I'm now running on 256MB of RAM (max on this box is 385MB) and the bug still persists, so RAM corruption is not something thats happening on hardware.

I will see if installing the 2.6.32.11 kernel helps -- Currently I'm on a 2.6.32-15 compiled for SMP and Preemptive and optimized for Coppermine chips (Pentium 3 M here)

<sarcasm> 5 years of this, is going to be a little insane.</sarcasm>

Booooooooooggggg.