i am using windows vista business edition 32 bit with the BF252_r27589 blender release along with an nvidia GeForce7300GT intel core 2 Duo (E4500 2.20ghz) 3.00GB ram i tried to replicate the crash no luck what system were you using (by the way i cant program but if the bug can be replicated it might get noticed by someone that can)

Cannot reproduce this as well, using latest SVN. Could you try with a more recent build on graphicall.org

My specs:
Ubuntu 9.04 x64 (kernel 2.6.28-17 generic) with latest NVIDA
drivers
NVIDIA GeForce 9600M GS

Ubuntu 9.10, intel xeon, nvidia

Tried it with newest on graphicall and it still crashes. But I have to admit that it crashed yesterday almost on every try.
Rendering and then connecting and fooling with the compositor around (just connecting and reconnecting stuff, playing around with rgb curve..) helps

here is output:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffd32fe910 (LWP 24178)]
0x0000000000c082ef in multitex_nodes (tex=0x3ea6a6a73ea6a6a7,
texvec=0x7fffd32fdac0, dxt=0x0, dyt=0x0, osatex=0, texres=0x7fffd32fda80,
thread=0, which_output=0, shi=0x0, mtex=0x0)
at /media/data/blender_soenke/blender/source/blender/render/intern/source/texture_stack.c:2180
2180 if(tex->type==TEX_IMAGE) {
#0 0x0000000000c082ef in multitex_nodes (tex=0x3ea6a6a73ea6a6a7,
texvec=0x7fffd32fdac0, dxt=0x0, dyt=0x0, osatex=0, texres=0x7fffd32fda80,
thread=0, which_output=0, shi=0x0, mtex=0x0)
at /media/data/blender_soenke/blender/source/blender/render/intern/source/texture_stack.c:2180
#1 0x0000000000c0869c in multitex_ext (tex=0x3ea6a6a73ea6a6a7,
texvec=0x7fffd32fdac0, dxt=0x0, dyt=0x0, osatex=0, texres=0x7fffd32fda80)
at /media/data/blender_soenke/blender/source/blender/render/intern/source/texture_stack.c:2230
#2 0x0000000000e56b5a in texture_procedural (cbuf=0x3f2c2f0, out=0x4198980,
xco=0.429618776, yco=-1)
at /media/data/blender_soenke/blender/source/blender/nodes/intern/CMP_nodes/CMP_texture.c:58
#3 0x0000000000e450bd in compbuf_get_pixel (cbuf=0x3f2c2f0, rectf=0x4198980,
---Type <return> to continue, or q <return> to quit---
x=879, y=-872, xrad=2046, yrad=872)
at /media/data/blender_soenke/blender/source/blender/nodes/intern/CMP_util.c:416
#4 0x0000000000e45217 in composit1_pixel_processor (node=0x7fffe015fb70,
out=0x41bac30, src_buf=0x308f830, src_col=0x4198980,
func=0xe51c0e <do_curves>, src_type=4)
at /media/data/blender_soenke/blender/source/blender/nodes/intern/CMP_util.c:453
#5 0x0000000000e51f06 in node_composit_exec_curve_rgb (data=0x44779a8,
node=0x7fffe015fb70, in=0x7fffd32fde50, out=0x7fffd32fdc50)
at /media/data/blender_soenke/blender/source/blender/nodes/intern/CMP_nodes/CMP_curves.c:181
#6 0x0000000000d66d5a in exec_composite_node (node_v=0x7fffe015fb70)
at /media/data/blender_soenke/blender/source/blender/blenkernel/intern/node.c:2263
#7 0x0000000000f012c5 in tslot_thread_start (tslot_p=0x41b9870)
at /media/data/blender_soenke/blender/source/blender/blenlib/intern/threads.c:205
#8 0x00007ffff68f5a04 in start_thread () from /lib/libpthread.so.0
#9 0x00007ffff6e6e80d in clone () from /lib/libc.so.6
#10 0x0000000000000000 in ?? ()
(gdb)

Ok, after a couple more tries I was able to reproduce this. I also notice you're using the rendering branch and not trunk? Although I reproduced this with trunk.
It looks a lot like a threading problem (especially the random factor makes it more likely).

Confirmed in trunk, I ran into this bug and actually saved the identical backtrace earlier this week.

I've committed a fix, which seems to be theoretically right, and also has fixed the crashes for me. Please give it a try! I'll close this report, but let me know if there are still probs.