[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 10 Oct 2019 11:36:30 -0700
From: Cong Wang <xiyou.wangcong@...il.com>
To: Maciej Żenczykowski <zenczykowski@...il.com>
Cc: Florian Westphal <fw@...len.de>,
"David S . Miller" <davem@...emloft.net>,
Linux NetDev <netdev@...r.kernel.org>,
Eric Dumazet <edumazet@...gle.com>,
Pablo Neira Ayuso <pablo@...filter.org>
Subject: Re: [PATCH 2/2] netfilter: revert "conntrack: silent a memory leak warning"
On Tue, Oct 8, 2019 at 12:10 AM Maciej Żenczykowski
<zenczykowski@...il.com> wrote:
>
> Here's my reasoning:
>
> old = ct->ext;
>
> //... stuff that doesn't change old.
>
> alloc = max(newlen, NF_CT_EXT_PREALLOC); <-- will be >= 128,
> so not zero
> kmemleak_not_leak(old);
> new = __krealloc(old, alloc, gfp);
> if (!new)
> return NULL; <--- if we return here, ct->ext still
> holds old, so no leak.
>
> if (!old) {
> memset(new->offset, 0, sizeof(new->offset));
> ct->ext = new; <--- old is NULL so can't leak
> } else if (new != old) {
> kfree_rcu(old, rcu); <-- we free old, so doesn't leak
> rcu_assign_pointer(ct->ext, new);
> } <--- else new == old && it's still in ct->ext, so it doesn't leak
>
So you conclude as it is not leak too? Then what are you trying to
fix?
I am becoming more confused after this. :-/
> Basically AFAICT our use of __krealloc() is exactly like krealloc()
> except instead of kfree() we do kfree_rcu().
>
> And thus I don't understand the need for kmemleak_not_leak(old).
kfree_rcu() is a callback deferred after a grace period, so if we
allocate the memory again before that callback, it is reported to
kmemleak as a memory leak unless we mark it as not, right?
Or kfree_rcu() works nicely with kmemleak which I am not aware
of?
Thanks.
Powered by blists - more mailing lists