netdev - Re: [PATCH v3 net-next] r8169: improve WoL handling

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] [day] [month] [year] [list]

Message-Id: <20190131.125321.967538547103372593.davem@davemloft.net>
Date:   Thu, 31 Jan 2019 12:53:21 -0800 (PST)
From:   David Miller <davem@...emloft.net>
To:     hkallweit1@...il.com
Cc:     nic_swsd@...ltek.com, netdev@...r.kernel.org
Subject: Re: [PATCH v3 net-next] r8169: improve WoL handling

From: Heiner Kallweit <hkallweit1@...il.com>
Date: Thu, 31 Jan 2019 19:57:26 +0100

> WoL handling for the RTL8168 family is a little bit tricky because of
> different types of broken BIOS and/or chip quirks.
> 
> Two known issues:
> 1. Network properly resumes from suspend only if WoL is enabled in the chip.
> 2. Some notebooks wake up immediately if system is suspended and network
>    device is wakeup-enabled.
> 
> Few patches tried to deal with this:
> 7edf6d314cd0 ("r8169: disable WOL per default")
> 18041b523692 ("r8169: restore previous behavior to accept BIOS WoL
> settings")
> 
> Currently we have the situation that the chip WoL settings as set by
> the BIOS are respected (to prevent issue 1), but the device doesn't get
> wakeup-enabled (to prevent issue 2).
> 
> This leads to another issue:
> If systemd is told to set WoL it first checks whether the requested
> settings are active already (and does nothing if yes). Due to the chip
> WoL flags being set properly systemd assumes that WoL is configured
> properly in our case. Result is that device doesn't get wakeup-enabled
> and WoL doesn't work (until it's set e.g. by ethtool).
> 
> This patch now:
> - leaves the chip WoL settings as is (to prevent issue 1)
> - keeps the behavior to not wakeup-enable the device initially
>   (to prevent issue 2)
> - In addition we report WoL as being disabled in get_wol, matching
>   that device isn't wakeup-enabled. If systemd is told to enable WoL,
>   it will therefore detect that it has to do something and will
>   call set_wol.
> 
> Of course the user still has the option to override this with
> e.g. ethtool.
> 
> v2:
> - Don't just exclude __rtl8169_get_wol() from compiling, remove it.
> v3:
> - adjust commit message
> 
> Signed-off-by: Heiner Kallweit <hkallweit1@...il.com>

Applied, thanks.