All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
@ 2010-09-15 21:43 Dan Rosenberg
  2010-09-16 15:57 ` Steven Rostedt
  2010-09-17  4:55 ` David Miller
  0 siblings, 2 replies; 7+ messages in thread
From: Dan Rosenberg @ 2010-09-15 21:43 UTC (permalink / raw)
  To: j.dumon; +Cc: linux-kernel, netdev, security, davem, stable

Fixed formatting (tabs and line breaks).

The TIOCGICOUNT device ioctl allows unprivileged users to read
uninitialized stack memory, because the "reserved" member of the
serial_icounter_struct struct declared on the stack in hso_get_count()
is not altered or zeroed before being copied back to the user.  This
patch takes care of it.

Signed-off-by: Dan Rosenberg <dan.j.rosenberg@gmail.com>

--- linux-2.6.35.4.orig/drivers/net/usb/hso.c	2010-08-26 19:47:12.000000000 -0400
+++ linux-2.6.35.4/drivers/net/usb/hso.c	2010-09-14 21:26:18.477585183 -0400
@@ -1653,6 +1653,8 @@ static int hso_get_count(struct hso_seri
 	struct uart_icount cnow;
 	struct hso_tiocmget  *tiocmget = serial->tiocmget;
 
+	memset(&icount, 0, sizeof(struct serial_icounter_struct));
+
 	if (!tiocmget)
 		 return -ENOENT;
 	spin_lock_irq(&serial->serial_lock);







^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
  2010-09-15 21:43 [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory Dan Rosenberg
@ 2010-09-16 15:57 ` Steven Rostedt
  2010-09-16 17:07   ` Alan Cox
  2010-09-17  4:55 ` David Miller
  1 sibling, 1 reply; 7+ messages in thread
From: Steven Rostedt @ 2010-09-16 15:57 UTC (permalink / raw)
  To: Dan Rosenberg; +Cc: j.dumon, linux-kernel, netdev, security, davem, stable

On Wed, Sep 15, 2010 at 05:43:28PM -0400, Dan Rosenberg wrote:
> Fixed formatting (tabs and line breaks).
> 
> The TIOCGICOUNT device ioctl allows unprivileged users to read
> uninitialized stack memory, because the "reserved" member of the
> serial_icounter_struct struct declared on the stack in hso_get_count()
> is not altered or zeroed before being copied back to the user.  This
> patch takes care of it.
> 
> Signed-off-by: Dan Rosenberg <dan.j.rosenberg@gmail.com>
> 
> --- linux-2.6.35.4.orig/drivers/net/usb/hso.c	2010-08-26 19:47:12.000000000 -0400
> +++ linux-2.6.35.4/drivers/net/usb/hso.c	2010-09-14 21:26:18.477585183 -0400
> @@ -1653,6 +1653,8 @@ static int hso_get_count(struct hso_seri
>  	struct uart_icount cnow;
>  	struct hso_tiocmget  *tiocmget = serial->tiocmget;
>  
> +	memset(&icount, 0, sizeof(struct serial_icounter_struct));
> +

Move the above to after the spinlocks.

-- Steve

>  	if (!tiocmget)
>  		 return -ENOENT;
>  	spin_lock_irq(&serial->serial_lock);
> 

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
  2010-09-16 17:07   ` Alan Cox
@ 2010-09-16 16:52     ` Steven Rostedt
  2010-09-16 17:47       ` Alan Cox
  0 siblings, 1 reply; 7+ messages in thread
From: Steven Rostedt @ 2010-09-16 16:52 UTC (permalink / raw)
  To: Alan Cox
  Cc: Dan Rosenberg, j.dumon, linux-kernel, netdev, security, davem, stable

On Thu, 2010-09-16 at 18:07 +0100, Alan Cox wrote:
> ->tiocmget;
> > >  
> > > +	memset(&icount, 0, sizeof(struct serial_icounter_struct));
> > > +
> > 
> > Move the above to after the spinlocks.
> 
> Why - its a local variable ?
> 
> Easier to write
> 
> 	struct serial_icounter_struct icount = { 0 };
> 
> though

True, but if we want to micro-optimize, moving it after the spinlocks
means we do no zero initialization in the tiocmget == 0. Although gcc
may be smart enough to realize that too.

-- Steve



^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
  2010-09-16 15:57 ` Steven Rostedt
@ 2010-09-16 17:07   ` Alan Cox
  2010-09-16 16:52     ` Steven Rostedt
  0 siblings, 1 reply; 7+ messages in thread
From: Alan Cox @ 2010-09-16 17:07 UTC (permalink / raw)
  To: Steven Rostedt
  Cc: Dan Rosenberg, j.dumon, linux-kernel, netdev, security, davem, stable

->tiocmget;
> >  
> > +	memset(&icount, 0, sizeof(struct serial_icounter_struct));
> > +
> 
> Move the above to after the spinlocks.

Why - its a local variable ?

Easier to write

	struct serial_icounter_struct icount = { 0 };

though

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
  2010-09-16 17:47       ` Alan Cox
@ 2010-09-16 17:36         ` Steven Rostedt
  0 siblings, 0 replies; 7+ messages in thread
From: Steven Rostedt @ 2010-09-16 17:36 UTC (permalink / raw)
  To: Alan Cox
  Cc: Dan Rosenberg, j.dumon, linux-kernel, netdev, security, davem, stable

On Thu, 2010-09-16 at 18:47 +0100, Alan Cox wrote:

> If you want to micro-optimise it you can enumerate the cases that
> tiocmget can be NULL in that driver..

Heh, I think that's going a bit too far. Of course mico-optimizations
usually make the system go faster, but not enough to show out of the
noise, and micro-optimizations most of the time just make the code
harder to understand and more error prone.

A lot of the patches in this patch set could just do the:

	struct foo bar = { 0 };

This would let gcc optimize what needs to be done, and avoids a function
call to memset.

-- Steve



^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
  2010-09-16 16:52     ` Steven Rostedt
@ 2010-09-16 17:47       ` Alan Cox
  2010-09-16 17:36         ` Steven Rostedt
  0 siblings, 1 reply; 7+ messages in thread
From: Alan Cox @ 2010-09-16 17:47 UTC (permalink / raw)
  To: Steven Rostedt
  Cc: Dan Rosenberg, j.dumon, linux-kernel, netdev, security, davem, stable

On Thu, 16 Sep 2010 12:52:40 -0400
Steven Rostedt <rostedt@goodmis.org> wrote:

> On Thu, 2010-09-16 at 18:07 +0100, Alan Cox wrote:
> > ->tiocmget;
> > > >  
> > > > +	memset(&icount, 0, sizeof(struct serial_icounter_struct));
> > > > +
> > > 
> > > Move the above to after the spinlocks.
> > 
> > Why - its a local variable ?
> > 
> > Easier to write
> > 
> > 	struct serial_icounter_struct icount = { 0 };
> > 
> > though
> 
> True, but if we want to micro-optimize, moving it after the spinlocks
> means we do no zero initialization in the tiocmget == 0. Although gcc
> may be smart enough to realize that too.

If you want to micro-optimise it you can enumerate the cases that
tiocmget can be NULL in that driver..

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory
  2010-09-15 21:43 [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory Dan Rosenberg
  2010-09-16 15:57 ` Steven Rostedt
@ 2010-09-17  4:55 ` David Miller
  1 sibling, 0 replies; 7+ messages in thread
From: David Miller @ 2010-09-17  4:55 UTC (permalink / raw)
  To: drosenberg; +Cc: j.dumon, linux-kernel, netdev, security, stable

From: Dan Rosenberg <drosenberg@vsecurity.com>
Date: Wed, 15 Sep 2010 17:43:28 -0400

> Fixed formatting (tabs and line breaks).
> 
> The TIOCGICOUNT device ioctl allows unprivileged users to read
> uninitialized stack memory, because the "reserved" member of the
> serial_icounter_struct struct declared on the stack in hso_get_count()
> is not altered or zeroed before being copied back to the user.  This
> patch takes care of it.
> 
> Signed-off-by: Dan Rosenberg <dan.j.rosenberg@gmail.com>

Applied.

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2010-09-17  4:55 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-09-15 21:43 [PATCH v2] drivers/net/usb/hso.c: prevent reading uninitialized memory Dan Rosenberg
2010-09-16 15:57 ` Steven Rostedt
2010-09-16 17:07   ` Alan Cox
2010-09-16 16:52     ` Steven Rostedt
2010-09-16 17:47       ` Alan Cox
2010-09-16 17:36         ` Steven Rostedt
2010-09-17  4:55 ` David Miller

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.