netdev - Re: [PATCH v5] netns: optimize netns cleaning by batching unhash

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: <20260204075347.3508631-1-realwujing@gmail.com>
Date: Wed,  4 Feb 2026 02:53:42 -0500
From: Qiliang Yuan <realwujing@...il.com>
To: kuniyu@...gle.com
Cc: brauner@...nel.org,
	davem@...emloft.net,
	edumazet@...gle.com,
	horms@...nel.org,
	jack@...e.cz,
	jlayton@...nel.org,
	kuba@...nel.org,
	linux-kernel@...r.kernel.org,
	netdev@...r.kernel.org,
	pabeni@...hat.com,
	realwujing@...il.com,
	sd@...asysnail.net,
	yuanql9@...natelecom.cn
Subject: Re: [PATCH v5] netns: optimize netns cleaning by batching unhash_nsid calls

Hi Kuniyuki,

Thanks for your review and suggestions!

On Tue, 3 Feb 2026 16:20:18 -0800 Kuniyuki Iwashima <kuniyu@...gle.com> wrote:
> batch and system is a bit unclear, maybe M_dying_net,
> N_alive_net or something ?

Good point. I've updated the terminology in the commit message to 
M_dying_net and N_alive_net for better clarity.

> Also, isn't __peernet2id() also O(N_ids) ?

Yes, you are right. __peernet2id() performs a linear search in the IDR, 
so it is indeed O(N_id). I've corrected the complexity analysis in the 
commit log. This actually makes the batching optimization even more 
valuable.

> Please put this in a hole, maybe after after hash_mix :
>
> $ pahole -C net vmlinux

Updated. I've moved the 'is_dying' flag to the hole after 'hash_mix' 
to avoid increasing the size of struct net.

> No need to move id up.

Fixed. I've scoped the 'id' variable locally within the traversal loop 
to keep it cleaner.

> This can be done unconditionally as the id will never
> be assigned to dying nets.
>
> while (...) {
>     id++;
>
>     if (!net->dying)
>         continue;
>
>     /* do cleanup */
> }

Simplified as suggested. The loop logic is much cleaner now with the 
unconditional increment.

I've sent out v6 with these changes.

v6 link: https://lore.kernel.org/all/20260204074854.3506916-1-realwujing@gmail.com/

Best regards,
Qiliang