In ravb_probe, priv->work was bound with ravb_tx_timeout_work.
If timeout occurs, it will start the work. And if we call
ravb_remove without finishing the work, ther may be a use
after free bug on ndev.
Fix it by finishing the job before cleanup in ravb_remove.
Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
Signed-off-by: Zheng Wang <[email protected]>
---
drivers/net/ethernet/renesas/ravb_main.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 0f54849a3823..07a08e72f440 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -2892,6 +2892,7 @@ static int ravb_remove(struct platform_device *pdev)
struct ravb_private *priv = netdev_priv(ndev);
const struct ravb_hw_info *info = priv->info;
+ cancel_work_sync(&priv->work);
/* Stop PTP Clock driver */
if (info->ccc_gac)
ravb_ptp_stop(ndev);
--
2.25.1
Hello!
On 3/9/23 1:02 PM, Zheng Wang wrote:
> In ravb_probe, priv->work was bound with ravb_tx_timeout_work.
> If timeout occurs, it will start the work. And if we call
> ravb_remove without finishing the work, ther may be a use
> after free bug on ndev.
Have you actually encountered it?
> Fix it by finishing the job before cleanup in ravb_remove.
>
> Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
Hm... sorry about that. :-)
> Signed-off-by: Zheng Wang <[email protected]>
Reviewed-by: Sergey Shtylyov <[email protected]>
[...]
> diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
> index 0f54849a3823..07a08e72f440 100644
> --- a/drivers/net/ethernet/renesas/ravb_main.c
> +++ b/drivers/net/ethernet/renesas/ravb_main.c
> @@ -2892,6 +2892,7 @@ static int ravb_remove(struct platform_device *pdev)
> struct ravb_private *priv = netdev_priv(ndev);
> const struct ravb_hw_info *info = priv->info;
>
> + cancel_work_sync(&priv->work);
I think we need an empty line here...
> /* Stop PTP Clock driver */
> if (info->ccc_gac)
> ravb_ptp_stop(ndev);
MBR, Sergey
On 2023/3/9 18:02, Zheng Wang wrote:
> In ravb_probe, priv->work was bound with ravb_tx_timeout_work.
> If timeout occurs, it will start the work. And if we call
> ravb_remove without finishing the work, ther may be a use
ther -> there
> after free bug on ndev.
>
> Fix it by finishing the job before cleanup in ravb_remove.
>
> Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
> Signed-off-by: Zheng Wang <[email protected]>
> ---
> drivers/net/ethernet/renesas/ravb_main.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
> index 0f54849a3823..07a08e72f440 100644
> --- a/drivers/net/ethernet/renesas/ravb_main.c
> +++ b/drivers/net/ethernet/renesas/ravb_main.c
> @@ -2892,6 +2892,7 @@ static int ravb_remove(struct platform_device *pdev)
> struct ravb_private *priv = netdev_priv(ndev);
> const struct ravb_hw_info *info = priv->info;
>
> + cancel_work_sync(&priv->work);
As your previous patch, I still do not see anything stopping
dev_watchdog() from calling dev->netdev_ops->ndo_tx_timeout
after cancel_work_sync(), maybe I missed something obvious
here?
> /* Stop PTP Clock driver */
> if (info->ccc_gac)
> ravb_ptp_stop(ndev);
>
Sergey Shtylyov <[email protected]> 于2023年3月9日周四 23:47写道：
>
> Hello!
>
> On 3/9/23 1:02 PM, Zheng Wang wrote:
>
> > In ravb_probe, priv->work was bound with ravb_tx_timeout_work.
> > If timeout occurs, it will start the work. And if we call
> > ravb_remove without finishing the work, ther may be a use
> > after free bug on ndev.
>
> Have you actually encountered it?
Sorry, I haven't encountered it. All of the analysis is based on
static analysis.
>
> > Fix it by finishing the job before cleanup in ravb_remove.
> >
> > Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
>
> Hm... sorry about that. :-)
>
> > Signed-off-by: Zheng Wang <[email protected]>
>
> Reviewed-by: Sergey Shtylyov <[email protected]>
>
> [...]
> > diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
> > index 0f54849a3823..07a08e72f440 100644
> > --- a/drivers/net/ethernet/renesas/ravb_main.c
> > +++ b/drivers/net/ethernet/renesas/ravb_main.c
> > @@ -2892,6 +2892,7 @@ static int ravb_remove(struct platform_device *pdev)
> > struct ravb_private *priv = netdev_priv(ndev);
> > const struct ravb_hw_info *info = priv->info;
> >
> > + cancel_work_sync(&priv->work);
>
> I think we need an empty line here...
Yes, will add it in the next version of patch.
Best regards,
Zheng
>
> > /* Stop PTP Clock driver */
> > if (info->ccc_gac)
> > ravb_ptp_stop(ndev);
>
Yunsheng Lin <[email protected]> 于2023年3月10日周五 09:12写道：
>
> On 2023/3/9 18:02, Zheng Wang wrote:
> > In ravb_probe, priv->work was bound with ravb_tx_timeout_work.
> > If timeout occurs, it will start the work. And if we call
> > ravb_remove without finishing the work, ther may be a use
>
> ther -> there
>
Sorry about the typo, will correct it in the next version.
> > after free bug on ndev.
> >
> > Fix it by finishing the job before cleanup in ravb_remove.
> >
> > Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
> > Signed-off-by: Zheng Wang <[email protected]>
> > ---
> > drivers/net/ethernet/renesas/ravb_main.c | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
> > index 0f54849a3823..07a08e72f440 100644
> > --- a/drivers/net/ethernet/renesas/ravb_main.c
> > +++ b/drivers/net/ethernet/renesas/ravb_main.c
> > @@ -2892,6 +2892,7 @@ static int ravb_remove(struct platform_device *pdev)
> > struct ravb_private *priv = netdev_priv(ndev);
> > const struct ravb_hw_info *info = priv->info;
> >
> > + cancel_work_sync(&priv->work);
>
> As your previous patch, I still do not see anything stopping
> dev_watchdog() from calling dev->netdev_ops->ndo_tx_timeout
> after cancel_work_sync(), maybe I missed something obvious
> here?
>
Yes, that's a keyed suggestion. I was hurry to report the issue today
so wrote with many mistakes.
Thanks agagin for the advice. I will review other patch carefully.
Best regards,
Zheng
>
> > /* Stop PTP Clock driver */
> > if (info->ccc_gac)
> > ravb_ptp_stop(ndev);
> >