lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening PHC | |
Open Source and information security mailing list archives
| ||
|
Date: Fri, 14 Aug 2020 15:45:58 +0200 From: Willem de Bruijn <willemdebruijn.kernel@...il.com> To: linmiaohe <linmiaohe@...wei.com> Cc: Willem de Bruijn <willemdebruijn.kernel@...il.com>, David Miller <davem@...emloft.net>, Jakub Kicinski <kuba@...nel.org>, Florian Westphal <fw@...len.de>, "martin.varghese@...ia.com" <martin.varghese@...ia.com>, "pshelar@....org" <pshelar@....org>, "dcaratti@...hat.com" <dcaratti@...hat.com>, Eric Dumazet <edumazet@...gle.com>, Steffen Klassert <steffen.klassert@...unet.com>, Paolo Abeni <pabeni@...hat.com>, Shmulik Ladkani <shmulik@...anetworks.com>, Yadu Kishore <kyk.segfault@...il.com>, "sowmini.varadhan@...cle.com" <sowmini.varadhan@...cle.com>, Network Development <netdev@...r.kernel.org>, linux-kernel <linux-kernel@...r.kernel.org> Subject: Re: [PATCH] net: add missing skb_uarg refcount increment in pskb_carve_inside_header() On Fri, Aug 14, 2020 at 9:20 AM linmiaohe <linmiaohe@...wei.com> wrote: > > Willem de Bruijn <willemdebruijn.kernel@...il.com> wrote: > >On Thu, Aug 13, 2020 at 2:16 PM Miaohe Lin <linmiaohe@...wei.com> wrote: > >> > >> If the skb is zcopied, we should increase the skb_uarg refcount before > >> we involve skb_release_data(). See pskb_expand_head() as a reference. > > > >Did you manage to observe a bug through this datapath in practice? > > > >pskb_carve_inside_header is called > > from pskb_carve > > from pskb_extract > > from rds_tcp_data_recv > > > >That receive path should not see any packets with zerocopy state associated. > > > > This works fine yet as its caller is limited. But we should take care of the skb_uarg refcount for future use. If a new application of this interface is proposed, the author will have to make sure that it is exercised correctly. > On the other hand, because this codepath should not see any packets with zerocopy state associated, then we > should not call skb_orphan_frags here. I'm also not convinced that the skb_orphan_frags here are needed, given the only path is from tcp_read_sock.
Powered by blists - more mailing lists