Subject: Re: proc_stop/proc_unstop
To: Emmanuel Dreyfus <firstname.lastname@example.org>
From: Stephan Uphoff <email@example.com>
Date: 12/20/2003 21:19:15
> Some thoughts about proc_stop and proc_unstop:
> - In src/sys/kern/kern_sig.c the proc_stop comments says the process
> must not be on the run queue. Isn't that obsolete? There is a case for
> threads in LSRUN state in that function.
There is a case for LSRUN ... but it is wrong ;-)
The lwp yanked from the runqueue might be active in the
kernel and hold an important lock or other resources.
( Example: lock on the root vnode)
Unfortunately sys__lwp_suspend and suspendsched
do the same thing.
I am not sure if anyone else is using LWPs outside of SA.
There are probably some nasty surprises waiting for you.
I really hate being so negative - but unfortunately I don't
have the time to do more than pointing fingers.