linux-kernel - Re: [PATCH] Driver core: fix race in dev_driver

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: <Pine.LNX.4.44L0.0912041148150.3070-100000@iolanthe.rowland.org>
Date:	Fri, 4 Dec 2009 11:57:57 -0500 (EST)
From:	Alan Stern <stern@...land.harvard.edu>
To:	Oliver Neukum <oliver@...kum.org>
cc:	Greg KH <gregkh@...e.de>, <stable@...nel.org>,
	Rickard Bellini <rickard.bellini@...csson.com>,
	"linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>,
	Torgny Johansson <torgny.johansson@...csson.com>,
	Kernel development list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] Driver core: fix race in dev_driver_string

On Fri, 4 Dec 2009, Oliver Neukum wrote:

> Am Freitag, 4. Dezember 2009 17:06:57 schrieb Alan Stern:
> > Oliver:
> > 
> > We don't have to worry about the device structure being deallocated 
> > while the routine is running.  If that happens it's a bug in the 
> > caller: improper refcounting.
> 
> That raises two points
> 
> 1. am I supposed to get a reference just so that I can use dev_err?

You're supposed to hold a reference before using dev at all.  If the 
only use you make of dev is to call dev_err(), then yes.

That's how refcounting is meant to work.

> 2. what happens if this is a soft disconnect and the device is reconnected?
> It seems to me that you'd print the wrong driver's name.

You mean the device is unbound from driver A and then bound to driver 
B, after which a thread running in driver A calls dev_err()?  Then yes, 
the "wrong" name will be printed.

This is part of the idea behind dev_err() and friends.  They print the 
name of the currently bound driver, not the name of the calling source 
file.  That's why, for example, the name "usb-storage" shows up in a 
message printed from within usbcore rather than "hub.c".

Alan Stern

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/