Re: Help debugging hang with epoll_engine

[Daniel Miller <bonsaiviking () gmail com>]

Ok, I've narrowed it down to a loop in function "run" in nse_main.lua. 
The call to l_loop (socket.loop in Lua) happens on line 937. I ran 
nse_main.lua through luac -l to generate opcodes, so here's my best 
guess at the execution trace:
>   OP_GETUPVAL OP_MOVE OP_CALL -- pairs(pending)
>   OP_JMP OP_TFORCALL OP_TFORLOOP -- for co, thread in pairs(pending) do
>     -- pending must be empty, because we move past it to:
>   OP_GETUPVAL OP_LOADK OP_CALL -- collectgarbage "step"
> OP_JMP -- end -- loop back to line 850, while
>   OP_GETUPVAL OP_MOVE OP_CALL OP_TEST -- next running
>     OP_GETUPVAL OP_MOVE OP_CALL OP_TEST -- next waiting
>     OP_TEST -- threads_iter
>     --false, so fal through to line 863
>   OP_GETUPVAL OP_MOVE OP_CALL -- table_size
>     --function table_size
>     OP_LOADK -- local n=0
>     OP_GETUPVAL OP_MOVE OP_CALL -- pairs(t)
>     OP_JMP OP_TFORCALL OP_TFORLOOP -- for _ in pairs(t) do
>     OP_RETURN -- return 0, since t was empty
>   OP_GETUPVAL OP_MOVE OP_CALL -- table_size
>     --function table_size
>     OP_LOADK -- local n=0
>     OP_GETUPVAL OP_MOVE OP_CALL -- pairs(t)
>     OP_JMP OP_TFORCALL OP_TFORLOOP -- for _ in pairs(t) do
>       OP_ADD -- n = n + 1
>       OP_TFORCALL OP_TFORLOOP -- for _ in pairs(t) do
>     OP_RETURN -- return 1
>   -- back in run() at line 864
>   OP_GETTABUP OP_CALL -- cnse.key_was_pressed()
>     OP_TEST -- false, go to line 878
>   OP_MOVE OP_LOADK OP_CALL -- elseif progress "mayBePrinted" then
>     OP_TEST -- false, go to line 887
>   OP_GETUPVAL OP_MOVE OP_CALL -- pairs(waiting)
>   OP_JMP OP_TFORCALL OP_TFORLOOP -- for co, thread in pairs(waiting) do
>     OP_SELF OP_CALL -- thread:timed_out() -- line 300
>       OP_GETTABLE OP_EQ OP_GETTABLE OP_EQ -- if self.type == 
> "hostrule" or self.type == "portrule" then
>         OP_GETTABUP OP_GETTABLE -- return cnse.timedOut(self.host);
>         OP_TAILCALL
>         OP_RETURN
>     -- back in run() line 888
>     OP_TEST -- false (not timed out)
>     OP_TFORCALL OP_TFORLOOP -- for co, thread in pairs(waiting) do
>   OP_GETUPVAL OP_MOVE OP_CALL -- pairs(running)
>   OP_JMP OP_TFORCALL OP_TFORLOOP -- for co, thread in pairs(running) do
>     -- No running threads
>   OP_GETUPVAL OP_LOADK OP_CALL -- loop(50)
>   Breakpoint 4, l_loop (L=0x8c56ad8) at nse_nsock.cc:401


So there is a problem where the thread is stuck in waiting state, and 
the socket won't timeout. I can't figure it out, and I need to move on 
to other things right now. I hope someone else can see what I can't.

Dan
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/