netdev - Re: [PATCH net 5/6] net: hns3: fix wrong print link down up

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-next>] [day] [month] [year] [list]

Message-ID: <e7219114-774f-49d0-8985-8875fd351b60@lunn.ch>
Date: Sat, 29 Jul 2023 20:23:59 +0200
From: Andrew Lunn <andrew@...n.ch>
To: Jijie Shao <shaojijie@...wei.com>
Cc: yisen.zhuang@...wei.com, salil.mehta@...wei.com, davem@...emloft.net,
	edumazet@...gle.com, kuba@...nel.org, pabeni@...hat.com,
	shenjian15@...wei.com, wangjie125@...wei.com,
	liuyonglong@...wei.com, wangpeiyang1@...wei.com,
	netdev@...r.kernel.org, stable@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH net 5/6] net: hns3: fix wrong print link down up

>     Now i wounder if you are fixing the wrong thing. Maybe you should be
>     fixing the PHY so it does not report up and then down? You say 'very
>     snall intervals', which should in fact be 1 second. So is the PHY
>     reporting link for a number of poll intervals? 1min to 10 minutes?
> 
>               Andrew
> 
> Yes, according to the log records, the phy polls every second,
> but the link status changes take time.
> Generally, it takes 10 seconds for the phy to detect link down,
> but occasionally it takes several minutes to detect link down,

What PHY driver is this?

It is not so clear what should actually happen with auto-neg turned
off. With it on, and the link going down, the PHY should react after
about 1 second. It is not supposed to react faster than that, although
some PHYs allow fast link down notification to be configured.

Have you checked 802.3 to see what it says about auto-neg off and link
down detection?

I personally would not suppress this behaviour in the MAC
driver. Otherwise you are going to have funny combinations of special
cases of a feature which very few people actually use, making your
maintenance costs higher.

	    Andrew