On Mon, May 04, 2009 at 05:37:46PM +0200, Casper.Dik at Sun.COM wrote: > > (Why was your reply not send to psarc*?)
Your questions weren't sent to PSARC, so I didn't send my reply to PSARC either. > >I could always extend /proc, but it seems unnecessary. > > I think it is pretty much required. The missing context, for PSARC readers, is that Casper says that to modify ptools one ought to modify /proc as well. In this case the new system calls provide enough observability that /proc changes would be redundant, but modifying pcred would still be useful for observability, and anyways, holding a proc(4) handle to a process to prevent PID reuse while examining it is also useful. If there's a hard and fast rule that ptools cannot use facilities outside proc(4) for observing targets then we could easily extend proc(4) to make CPG information available through proc(4). But as I said, given that the CPG syscalls provide sufficient observability it seems unnecessary to extend proc(4). Nico --
