[PATCH driver-core-linus] sysfs: make sure read buffer is zeroed

[PATCH driver-core-linus] sysfs: make sure read buffer is zeroed

[Tejun Heo]

13c589d5b0ac ("sysfs: use seq_file when reading regular files")
switched sysfs from custom read implementation to seq_file to enable
later transition to kernfs.  After the change, the buffer passed to
->show() is acquired through seq_get_buf(); unfortunately, this
introduces a subtle behavior change.  Before the commit, the buffer
passed to ->show() was always zero as it was allocated using
get_zeroed_page().  Because seq_file doesn't clear buffers on
allocation and neither does seq_get_buf(), after the commit, depending
on the behavior of ->show(), we may end up exposing uninitialized data
to userland thus possibly altering userland visible behavior and
leaking information.

Fix it by explicitly clearing the buffer.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Ron <ron@debian.org> 
Fixes: 13c589d5b0ac ("sysfs: use seq_file when reading regular files")
---
 fs/sysfs/file.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
index 28cc1acd..e9ef59b 100644
--- a/fs/sysfs/file.c
+++ b/fs/sysfs/file.c
@@ -47,12 +47,13 @@ static int sysfs_kf_seq_show(struct seq_file *sf, void *v)
 	ssize_t count;
 	char *buf;
 
-	/* acquire buffer and ensure that it's >= PAGE_SIZE */
+	/* acquire buffer and ensure that it's >= PAGE_SIZE and clear */
 	count = seq_get_buf(sf, &buf);
 	if (count < PAGE_SIZE) {
 		seq_commit(sf, -1);
 		return 0;
 	}
+	memset(buf, 0, PAGE_SIZE);
 
 	/*
 	 * Invoke show().  Control may reach here via seq file lseek even

Re: [PATCH driver-core-linus] sysfs: make sure read buffer is zeroed

[Greg Kroah-Hartman]

On Mon, May 19, 2014 at 03:52:10PM -0400, Tejun Heo wrote:
> 13c589d5b0ac ("sysfs: use seq_file when reading regular files")
> switched sysfs from custom read implementation to seq_file to enable
> later transition to kernfs.  After the change, the buffer passed to
> ->show() is acquired through seq_get_buf(); unfortunately, this
> introduces a subtle behavior change.  Before the commit, the buffer
> passed to ->show() was always zero as it was allocated using
> get_zeroed_page().  Because seq_file doesn't clear buffers on
> allocation and neither does seq_get_buf(), after the commit, depending
> on the behavior of ->show(), we may end up exposing uninitialized data
> to userland thus possibly altering userland visible behavior and
> leaking information.
> 
> Fix it by explicitly clearing the buffer.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>
> Reported-by: Ron <ron@debian.org> 
> Fixes: 13c589d5b0ac ("sysfs: use seq_file when reading regular files")
> ---
>  fs/sysfs/file.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
> index 28cc1acd..e9ef59b 100644
> --- a/fs/sysfs/file.c
> +++ b/fs/sysfs/file.c
> @@ -47,12 +47,13 @@ static int sysfs_kf_seq_show(struct seq_file *sf, void *v)
>  	ssize_t count;
>  	char *buf;
>  
> -	/* acquire buffer and ensure that it's >= PAGE_SIZE */
> +	/* acquire buffer and ensure that it's >= PAGE_SIZE and clear */
>  	count = seq_get_buf(sf, &buf);
>  	if (count < PAGE_SIZE) {
>  		seq_commit(sf, -1);
>  		return 0;
>  	}
> +	memset(buf, 0, PAGE_SIZE);
>  
>  	/*
>  	 * Invoke show().  Control may reach here via seq file lseek even

Thanks, I'll go queue this up.

greg k-h

Re: [PATCH driver-core-linus] sysfs: make sure read buffer is zeroed

[Joe Perches]

On Mon, 2014-05-19 at 15:25 -0700, Greg Kroah-Hartman wrote:
> On Mon, May 19, 2014 at 03:52:10PM -0400, Tejun Heo wrote:
> > 13c589d5b0ac ("sysfs: use seq_file when reading regular files")
> > switched sysfs from custom read implementation to seq_file to enable
> > later transition to kernfs.  After the change, the buffer passed to
> > ->show() is acquired through seq_get_buf(); unfortunately, this
> > introduces a subtle behavior change.  Before the commit, the buffer
> > passed to ->show() was always zero as it was allocated using
> > get_zeroed_page().  Because seq_file doesn't clear buffers on
> > allocation and neither does seq_get_buf(), after the commit, depending
> > on the behavior of ->show(), we may end up exposing uninitialized data
> > to userland thus possibly altering userland visible behavior and
> > leaking information.
> > 
> > Fix it by explicitly clearing the buffer.
[]
> > diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
[]
> > @@ -47,12 +47,13 @@ static int sysfs_kf_seq_show(struct seq_file *sf, void *v)
> >  	ssize_t count;
> >  	char *buf;
> >  
> > -	/* acquire buffer and ensure that it's >= PAGE_SIZE */
> > +	/* acquire buffer and ensure that it's >= PAGE_SIZE and clear */
> >  	count = seq_get_buf(sf, &buf);
> >  	if (count < PAGE_SIZE) {
> >  		seq_commit(sf, -1);
> >  		return 0;
> >  	}
> > +	memset(buf, 0, PAGE_SIZE);

	memset(buf, 0, count)?

> Thanks, I'll go queue this up.

Re: [PATCH driver-core-linus] sysfs: make sure read buffer is zeroed

[Tejun Heo]

On Mon, May 19, 2014 at 03:32:22PM -0700, Joe Perches wrote:
> > > +	memset(buf, 0, PAGE_SIZE);
> 
> 	memset(buf, 0, count)?

It's always PAGE_SIZE.  Outputs longer than that are truncated.

Thanks.

-- 
tejun

Re: [PATCH driver-core-linus] sysfs: make sure read buffer is zeroed

[Joe Perches]

On Mon, 2014-05-19 at 20:49 -0400, Tejun Heo wrote:
> On Mon, May 19, 2014 at 03:32:22PM -0700, Joe Perches wrote:
> > > > +	memset(buf, 0, PAGE_SIZE);
> > 
> > 	memset(buf, 0, count)?
> 
> It's always PAGE_SIZE.  Outputs longer than that are truncated.

Yeah, I know, but the code and comment above it don't match.

/* acquire buffer and ensure that it's >= PAGE_SIZE and clear */