netdev - [PATCH v2 5.10] net: defer final 'struct net' free in netns dismantle

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: <20250127134248.25731-1-abuehaze@amazon.com>
Date: Mon, 27 Jan 2025 13:42:48 +0000
From: Hazem Mohamed Abuelfotoh <abuehaze@...zon.com>
To: <kovalev@...linux.org>
CC: <edumazet@...gle.com>, <i.maximets@....org>, <kuba@...nel.org>,
	<kuniyu@...zon.com>, <lvc-project@...uxtesting.org>,
	<netdev@...r.kernel.org>, <pabeni@...hat.com>, <stable@...r.kernel.org>
Subject: [PATCH v2 5.10] net: defer final 'struct net' free in netns dismantle

I think a better approach will be to backport 41467d2ff4df 
("net: net_namespace: Optimize the code") as a prerequisite for
0f6ede9fbc74 ("net: defer final 'struct net' free in netns 
dismantle") otherwise we may endup with refcount underflow whenever
"net_free" & "net_drop_ns" called sequentially as with the current 
version of the backport both of them apparently call 
refcount_dec_and_test(&ns->passive). In 41467d2ff4df
("net: net_namespace: Optimize the code") there are two
hunks that aren't directly related to this backport however 
these are removing duplicate code into a separate function 
which will likely help with future backports. I have backported
both commits cleanly on 5.10.y and didn't see any issues and I 
am currently testing the same on 5.4 as it's also missing both 
commits.