Subject: Re: Walking an inpcb.inp_queue
To: None <>
From: Manuel Bouyer <>
List: tech-net
Date: 07/30/2001 21:35:18
On Sun, Jul 29, 2001 at 06:10:28PM -0400, Scott Barron wrote:
> Hello,
> I am using kvm to access and then walk an inpcb.inp_queue (as netstat does)
> to get to various members of struct socket.  I am wondering if there is a
> danger of the kernel pulling one of the pointers out from under me (example:
> if another process got scheduled and closed down that socket while I held
> the pointer, etc).  I've minimized my accesses to inp_queue but am wondering
> if I need to be worried about this and if so, is there anyway I can prevent
> it?  (I wasn't sure wheather to post this to tech-net or tech-kern, please
> let me know if its better suited to tech-kern.)

Yes, you may have problems like this. But the kvm routines should be able to
fail gracefully this case, and if your program checks the return value this
should work well.

Manuel Bouyer <>