netdev - Re: [PATCH net 1/2] net: ethernet: ftgmac100: prevent use after free on unregister when using NCSI

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: <fe5630d4-1502-45eb-a6fb-6b5bc33506a9@lunn.ch>
Date: Mon, 28 Oct 2024 21:15:24 +0100
From: Andrew Lunn <andrew@...n.ch>
To: Jeremy Kerr <jk@...econstruct.com.au>
Cc: Andrew Lunn <andrew+netdev@...n.ch>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
	Joel Stanley <joel@....id.au>,
	Jacky Chou <jacky_chou@...eedtech.com>,
	Jacob Keller <jacob.e.keller@...el.com>, netdev@...r.kernel.org
Subject: Re: [PATCH net 1/2] net: ethernet: ftgmac100: prevent use after free
 on unregister when using NCSI

On Mon, Oct 28, 2024 at 12:54:10PM +0800, Jeremy Kerr wrote:
> When removing an active ftgmac100 netdev that is configured for NCSI, we
> have a double free of the ncsi device: We currently unregister the ncsi
> device (freeing it), then unregister the netdev itself. If the netdev is
> running, the netdev_unregister() path performs a ->ndo_stop(), which
> calls ncsi_stop_dev() on the now-free ncsi pointer.
> 
> Instead, modify ftgmac100_stop() to check the ncsi pointer before
> freeing (rather than use_ncsi, which reflects configuration intent), and
> clear the pointer once we have done the ncsi_unregister().

This seems like the wrong fix. I would expect ftgmac100_stop() to be a
mirror of ftgmac100_open(). So unregistering in ndo_stop is correct,
and the real double free bug is in ftgmac100_remove().
ftgmac100_remove() should be a mirror of ftgmac100_probe() which does
not register the ncsi device....

    Andrew

---
pw-bot: cr