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  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CALW65jZe3JQGNcWsZtqU-B4-V-JZ6ocninxvoqMGeusMaU7C=A@mail.gmail.com>
Date: Mon, 17 Mar 2025 17:41:39 +0800
From: Qingfang Deng <dqfext@...il.com>
To: Antonio Quartulli <antonio@...nvpn.net>
Cc: andrew+netdev@...n.ch, donald.hunter@...il.com, edumazet@...gle.com, 
	horms@...nel.org, kuba@...nel.org, linux-kernel@...r.kernel.org, 
	linux-kselftest@...r.kernel.org, netdev@...r.kernel.org, pabeni@...hat.com, 
	ryazanov.s.a@...il.com, sd@...asysnail.net, shaw.leon@...il.com, 
	shuah@...nel.org, "Jason A. Donenfeld" <Jason@...c4.com>
Subject: Re: [PATCH net-next v23 03/23] ovpn: add basic interface
 creation/destruction/management routines

Hi Antonio,

On Mon, Mar 17, 2025 at 5:23 PM Antonio Quartulli <antonio@...nvpn.net> wrote:
> >> +static void ovpn_setup(struct net_device *dev)
> >> +{
> >> +    netdev_features_t feat = NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_RXCSUM |
> >
> > Do not advertise NETIF_F_HW_CSUM or NETIF_F_RXCSUM, as TX/RX checksum is
> > not handled in hardware.
>
> The idea behind these flags was that the OpenVPN protocol will take care
> of authenticating packets, thus substituting what the CSUM would do here.
> For this I wanted to avoid the stack to spend time computing the CSUM in
> software.

For the RX part (NETIF_F_RXCSUM), you might be correct, but in patch
08 you wrote:
> /* we can't guarantee the packet wasn't corrupted before entering the
> * VPN, therefore we give other layers a chance to check that
> */
> skb->ip_summed = CHECKSUM_NONE;

So NETIF_F_RXCSUM has no effect.

For the TX part (NETIF_F_HW_CSUM) however, I believe wireguard made
the same mistake.
Your code both contains the pattern:

if (skb->ip_summed == CHECKSUM_PARTIAL && skb_checksum_help(skb)) // ...

NETIF_F_HW_CSUM causes the upper layers to send packets with
CHECKSUM_PARTIAL, assuming hardware offload will complete the
checksum, but if skb_checksum_help(skb) is invoked, the checksum is
still computed in software. This means there's no real benefit unless
there's an actual hardware offload mechanism.

+Cc: zx2c4

>
> I believe wireguard sets those flags for the same reason.
>
> Does it make sense to you?
>
> >
> >> +                             NETIF_F_GSO | NETIF_F_GSO_SOFTWARE |
> >> +                             NETIF_F_HIGHDMA;
>
>
> Regards,
>
> --
> Antonio Quartulli
> OpenVPN Inc.
>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ