[PATCH] usb: hub: avoid possible division by zero

[PATCH] usb: hub: avoid possible division by zero

[Johannes Thumshirn]

Avoid possible division by zero in led_work, if hdev->maxchild is 0.

See also:
http://buildbot.llvm.linuxfoundation.org/checker/scan-build-latest/report-b65939.html#EndPath

Signed-off-by: Johannes Thumshirn <morbidrsa@gmail.com>
---
 drivers/usb/core/hub.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
index a7c04e2..8c7aa4e 100644
--- a/drivers/usb/core/hub.c
+++ b/drivers/usb/core/hub.c
@@ -453,7 +453,8 @@ static void led_work (struct work_struct *work)
 	unsigned		changed = 0;
 	int			cursor = -1;
 
-	if (hdev->state != USB_STATE_CONFIGURED || hub->quiescing)
+	if (hdev->state != USB_STATE_CONFIGURED || hub->quiescing
+		|| hdev->maxchild == 0)
 		return;
 
 	for (i = 0; i < hdev->maxchild; i++) {
-- 
1.8.4.3

Re: [PATCH] usb: hub: avoid possible division by zero

[Greg Kroah-Hartman]

On Sat, Nov 23, 2013 at 05:31:34PM +0100, Johannes Thumshirn wrote:
> Avoid possible division by zero in led_work, if hdev->maxchild is 0.
> 
> See also:
> http://buildbot.llvm.linuxfoundation.org/checker/scan-build-latest/report-b65939.html#EndPath
> 
> Signed-off-by: Johannes Thumshirn <morbidrsa@gmail.com>
> ---
>  drivers/usb/core/hub.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> index a7c04e2..8c7aa4e 100644
> --- a/drivers/usb/core/hub.c
> +++ b/drivers/usb/core/hub.c
> @@ -453,7 +453,8 @@ static void led_work (struct work_struct *work)
>  	unsigned		changed = 0;
>  	int			cursor = -1;
>  
> -	if (hdev->state != USB_STATE_CONFIGURED || hub->quiescing)
> +	if (hdev->state != USB_STATE_CONFIGURED || hub->quiescing
> +		|| hdev->maxchild == 0)

And how can maxchild ever be equal to 0?

Look at where it is assigned (it comes from bNbrPorts in the
descriptor), in hub_configure() and if it is set to 0, the creation of
the structure fails and then this code will never run, right?

So I don't think this is needed, and your checker needs to be fixed.

sorry,

greg k-h

Re: [PATCH] usb: hub: avoid possible division by zero

[Johannes Thumshirn]

On Sat, Nov 23, 2013 at 09:20:10AM -0800, Greg Kroah-Hartman wrote:
> On Sat, Nov 23, 2013 at 05:31:34PM +0100, Johannes Thumshirn wrote:
> > Avoid possible division by zero in led_work, if hdev->maxchild is 0.
> >
> > See also:
> > http://buildbot.llvm.linuxfoundation.org/checker/scan-build-latest/report-b65939.html#EndPath
> >
> > Signed-off-by: Johannes Thumshirn <morbidrsa@gmail.com>
> > ---
> >  drivers/usb/core/hub.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> > index a7c04e2..8c7aa4e 100644
> > --- a/drivers/usb/core/hub.c
> > +++ b/drivers/usb/core/hub.c
> > @@ -453,7 +453,8 @@ static void led_work (struct work_struct *work)
> >  	unsigned		changed = 0;
> >  	int			cursor = -1;
> >
> > -	if (hdev->state != USB_STATE_CONFIGURED || hub->quiescing)
> > +	if (hdev->state != USB_STATE_CONFIGURED || hub->quiescing
> > +		|| hdev->maxchild == 0)
>
> And how can maxchild ever be equal to 0?
>
> Look at where it is assigned (it comes from bNbrPorts in the
> descriptor), in hub_configure() and if it is set to 0, the creation of
> the structure fails and then this code will never run, right?
>
> So I don't think this is needed, and your checker needs to be fixed.
>
> sorry,

Args, sorry for the noise. Should've checked that better.