Skip site navigation (1)Skip section navigation (2) 
Date:      Wed, 28 Aug 2002 12:31:48 -0700
From:      Mike Makonnen <makonnen@pacbell.net>
To:        smp@FreeBSD.ORG
Subject:   Locking questions
Message-ID:  <20020828123148.7717f871.makonnen@pacbell.net>
next in thread | raw e-mail | index | archive | help 
Hi folks,

I resumed working on my resource limit lock patch, and I have a couple of
questions.

1. I get "duplicate lock of same type" witness warnings because of this:

	limhold(plimit) { lock(plimit), bump refcount, unlock(plimit) }

        newlim = limget();
        PROC_LOCK(p2);
        PROC_LOCK(p1);
        LIM_LOCK(p1->p_limit);
        if (p1->p_limit->p_lflags & PL_SHAREMOD) {
                limcopy(newlim, p1->p_limit);
                p2->p_limit = limhold(newlim);
        } else {
                p2->p_limit = limhold(p1->p_limit);
        }
        LIM_UNLOCK(p1->p_limit);
        PROC_UNLOCK(p1);
        PROC_UNLOCK(p2);
        limfree(newlim);
        newlim = NULL;

I get this message in limhold() because of:
	p2->p_limit = limhold(p1->p_limit);

Why is this bad? 

Is it safe to drop the p1->p_plimit lock
right before I do the hold? It should be because you need
both a PROC_LOCK and a LIM_LOCK to modify the plimit.

2. Should I add the plimit lock to 
	static struct witness_order_list_entry order_lists[] ?
    What is the rationale behind the ordering of the list?


Cheers,
Mike Makonnen
    

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020828123148.7717f871.makonnen>