From: Pablo Neira Ayuso <pablo@netfilter.org>
To: Florian Westphal <fw@strlen.de>
Cc: netfilter-devel@vger.kernel.org,
Fernando Fernandez Mancera <ffmancera@riseup.net>
Subject: Re: [PATCH v2 nf] netfilter: nf_tables: fix memory leak during stateful obj update
Date: Mon, 21 Feb 2022 12:39:27 +0100 [thread overview]
Message-ID: <YhN575l459m+oTtK@salvia> (raw)
In-Reply-To: <20220221105922.GC18967@breakpoint.cc>
On Mon, Feb 21, 2022 at 11:59:22AM +0100, Florian Westphal wrote:
> Florian Westphal <fw@strlen.de> wrote:
> > Pablo Neira Ayuso <pablo@netfilter.org> wrote:
> > > On Sun, Feb 20, 2022 at 12:18:50PM +0100, Florian Westphal wrote:
> > > > stateful objects can be updated from the control plane.
> > > > The transaction logic allocates a temporary object for this purpose.
> > > >
> > > > This object has to be released via nft_obj_destroy, not kfree, since
> > > > the ->init function was called and it can have side effects beyond
> > > > memory allocation.
> > > >
> > > > Unlike normal NEWOBJ path, the objects module refcount isn't
> > > > incremented, so add nft_newobj_destroy and use that.
> > >
> > > Probably this? .udata and .key is NULL for the update path so kfree
> > > should be fine.
> >
> > Yes, that works too.
> >
> > We could also ...
> >
> > > - module_put(obj->ops->type->owner);
> > > + /* nf_tables_updobj does not increment module refcount */
> > > + if (!update)
> > > + module_put(obj->ops->type->owner);
> > > +
> >
> > Increment the refcount for update case as well to avoid the special
> > case?
Yes, I also though of this one. I prefer this approach indeed to
consolidate this path.
Would you test and submit v3?
Thanks!
> Untested:
>
> diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
> index 3081c4399f10..49060f281342 100644
> --- a/net/netfilter/nf_tables_api.c
> +++ b/net/netfilter/nf_tables_api.c
> @@ -6553,10 +6553,13 @@ static int nf_tables_updobj(const struct nft_ctx *ctx,
> struct nft_trans *trans;
> int err;
>
> + if (!try_module_get(type->owner))
> + return -ENOENT;
> +
> trans = nft_trans_alloc(ctx, NFT_MSG_NEWOBJ,
> sizeof(struct nft_trans_obj));
> if (!trans)
> - return -ENOMEM;
> + goto err_trans;
>
> newobj = nft_obj_init(ctx, type, attr);
> if (IS_ERR(newobj)) {
> @@ -6573,6 +6576,8 @@ static int nf_tables_updobj(const struct nft_ctx *ctx,
>
> err_free_trans:
> kfree(trans);
> +err_trans:
> + module_put(type->owner);
> return err;
> }
>
> @@ -8185,7 +8190,7 @@ static void nft_obj_commit_update(struct nft_trans *trans)
> if (obj->ops->update)
> obj->ops->update(obj, newobj);
>
> - kfree(newobj);
> + nft_obj_destroy(&trans->ctx, newobj);
> }
>
> static void nft_commit_release(struct nft_trans *trans)
> @@ -8976,7 +8981,7 @@ static int __nf_tables_abort(struct net *net, enum nfnl_abort_action action)
> break;
> case NFT_MSG_NEWOBJ:
> if (nft_trans_obj_update(trans)) {
> - kfree(nft_trans_obj_newobj(trans));
> + nft_obj_destroy(&trans->ctx, nft_trans_obj_newobj(trans));
> nft_trans_destroy(trans);
> } else {
> trans->ctx.table->use--;
prev parent reply other threads:[~2022-02-21 11:39 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-20 11:18 [PATCH v2 nf] netfilter: nf_tables: fix memory leak during stateful obj update Florian Westphal
2022-02-21 10:34 ` Pablo Neira Ayuso
2022-02-21 10:46 ` Florian Westphal
2022-02-21 10:59 ` Florian Westphal
2022-02-21 11:39 ` Pablo Neira Ayuso [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YhN575l459m+oTtK@salvia \
--to=pablo@netfilter.org \
--cc=ffmancera@riseup.net \
--cc=fw@strlen.de \
--cc=netfilter-devel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).