linux-kernel - Re: [PATCH] serial: core: Fix atomicity violation in uart

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]

Date: Fri, 12 Jan 2024 19:23:17 +0800
From: Gui-Dong Han <2045gemini@...il.com>
To: Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>,
 John Ogness <john.ogness@...utronix.de>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
 Jiri Slaby <jirislaby@...nel.org>, Tony Lindgren <tony@...mide.com>,
 l.sanfilippo@...bus.com, tglx@...utronix.de,
 Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
 LKML <linux-kernel@...r.kernel.org>,
 linux-serial <linux-serial@...r.kernel.org>, baijiaju1990@...look.com,
 stable@...r.kernel.org
Subject: Re: [PATCH] serial: core: Fix atomicity violation in uart_tiocmget

Hi,

You are correct about the 'Fixes' tag. It should indeed be c5f4644e6c8b 
("[PATCH] Serial: Adjust serial locking"). I will update this in the 
patch v2.

Regarding the issue found in Linux 5.17, I mistakenly used git blame 
which led to the incorrect identification of commit 559c7ff4e324. The 
issue indeed exists in Linux 5.17 and I acknowledge the error in tracing 
the commit.

In uart_tiocmget(), the result variable is stable. However, there's a 
risk of inconsistency due to the updates in uart_update_mctrl(). 
Consider a scenario where uart_tiocmget() reads uport->mctrl into result 
before entering the critical section. If uart_update_mctrl() updates 
port->mctrl and calls set_mctrl concurrently, the subsequent execution 
of result |= uport->ops->get_mctrl(uport); in uart_tiocmget() might 
yield an inaccurate result. This happens because result contains the old 
value of port->mctrl, which no longer matches the updated state 
retrieved by get_mctrl.

Thanks,

Han