[PATCH] net: dsa: vsc73xxx: Fix uninitalized 'val' in vsc73xx_adjust_link

[PATCH] net: dsa: vsc73xxx: Fix uninitalized 'val' in vsc73xx_adjust_link

[Danila Chernetsov]

Using uninitialized variable after calls vsc73xx_read 
without error checking may cause incorrect driver behavior.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: 95711cd5f0b4 ("net: dsa: vsc73xx: Split vsc73xx driver")
Signed-off-by: Danila Chernetsov <listdansp@mail.ru>
---
 drivers/net/dsa/vitesse-vsc73xx-core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/dsa/vitesse-vsc73xx-core.c b/drivers/net/dsa/vitesse-vsc73xx-core.c
index ae55167ce0a6..729005d6cb7e 100644
--- a/drivers/net/dsa/vitesse-vsc73xx-core.c
+++ b/drivers/net/dsa/vitesse-vsc73xx-core.c
@@ -758,7 +758,7 @@ static void vsc73xx_adjust_link(struct dsa_switch *ds, int port,
 				struct phy_device *phydev)
 {
 	struct vsc73xx *vsc = ds->priv;
-	u32 val;
+	u32 val = 0;
 
 	/* Special handling of the CPU-facing port */
 	if (port == CPU_PORT) {
-- 
2.25.1

Re: [PATCH] net: dsa: vsc73xxx: Fix uninitalized 'val' in vsc73xx_adjust_link

[Simon Horman]

On Sun, Mar 12, 2023 at 03:50:08PM +0000, Danila Chernetsov wrote:
> Using uninitialized variable after calls vsc73xx_read 
> without error checking may cause incorrect driver behavior.

I wonder if it is:
a) intentional that these calls are not checked for errors
b) errors can occur in these call paths

> Found by Linux Verification Center (linuxtesting.org) with SVACE.
> 
> Fixes: 95711cd5f0b4 ("net: dsa: vsc73xx: Split vsc73xx driver")
> Signed-off-by: Danila Chernetsov <listdansp@mail.ru>
> ---
>  drivers/net/dsa/vitesse-vsc73xx-core.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/dsa/vitesse-vsc73xx-core.c b/drivers/net/dsa/vitesse-vsc73xx-core.c
> index ae55167ce0a6..729005d6cb7e 100644
> --- a/drivers/net/dsa/vitesse-vsc73xx-core.c
> +++ b/drivers/net/dsa/vitesse-vsc73xx-core.c
> @@ -758,7 +758,7 @@ static void vsc73xx_adjust_link(struct dsa_switch *ds, int port,
>  				struct phy_device *phydev)
>  {
>  	struct vsc73xx *vsc = ds->priv;
> -	u32 val;
> +	u32 val = 0;
>  
>  	/* Special handling of the CPU-facing port */
>  	if (port == CPU_PORT) {
> -- 
> 2.25.1
> 

Re: [PATCH] net: dsa: vsc73xxx: Fix uninitalized 'val' in vsc73xx_adjust_link

[Jakub Kicinski]

On Mon, 13 Mar 2023 19:59:27 +0100 Simon Horman wrote:
> On Sun, Mar 12, 2023 at 03:50:08PM +0000, Danila Chernetsov wrote:
> > Using uninitialized variable after calls vsc73xx_read 
> > without error checking may cause incorrect driver behavior.  
> 
> I wonder if it is:
> a) intentional that these calls are not checked for errors
> b) errors can occur in these call paths

At the very least we should handle the error rather than silencing 
the static checker complain by picking a semi-random init value for 
the entire function.

Re: [PATCH] net: dsa: vsc73xxx: Fix uninitalized 'val' in vsc73xx_adjust_link

[Vladimir Oltean]

On Mon, Mar 13, 2023 at 07:59:27PM +0100, Simon Horman wrote:
> On Sun, Mar 12, 2023 at 03:50:08PM +0000, Danila Chernetsov wrote:
> > Using uninitialized variable after calls vsc73xx_read 
> > without error checking may cause incorrect driver behavior.
> 
> I wonder if it is:
> a) intentional that these calls are not checked for errors

probably no; this is precisely the only vsc73xx_read() call whose return
code is ignores. I'd say it partly has to do with the fact that vsc73xx_adjust_link()
returns void, so the author was thinking there'd be no point in checking
for errors, but there clearly is

> b) errors can occur in these call paths

probably yes; one of the instantiations of vsc73xx is over SPI, where
the controller can time out, etc.