2022-11-16 08:52:30

by Yang Yingliang

Subject: [PATCH] mfd: core: fix UAF while using device of node

I got the following UAF report:

refcount_t: underflow; use-after-free.
WARNING: CPU: 1 PID: 270 at lib/refcount.c:29 refcount_warn_saturate+0x121/0x180
...
OF: ERROR: memory leak, expected refcount 1 instead of -1073741824,
of_node_get()/of_node_put() unbalanced - destroy cset entry:
attach overlay node /i2c/pmic@62/powerkey

The of_node of device assigned in mfd_match_of_node_to_dev() need be
get, and it will be put in platform_device_release().

Fixes: 002be8114007 ("mfd: core: Add missing of_node_put for loop iteration")
Signed-off-by: Yang Yingliang <[email protected]>
---
drivers/mfd/mfd-core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mfd/mfd-core.c b/drivers/mfd/mfd-core.c
index 16d1861e9682..8e57f67719cf 100644
--- a/drivers/mfd/mfd-core.c
+++ b/drivers/mfd/mfd-core.c
@@ -161,7 +161,7 @@ static int mfd_match_of_node_to_dev(struct platform_device *pdev,
of_entry->np = np;
list_add_tail(&of_entry->list, &mfd_of_node_list);

- pdev->dev.of_node = np;
+ pdev->dev.of_node = of_node_get(np);
pdev->dev.fwnode = &np->fwnode;
#endif
return 0;
--
2.25.1

2022-12-08 14:01:30

by Yang Yingliang

Subject: Re: [PATCH] mfd: core: fix UAF while using device of node

Hi,

On 2022/11/16 15:41, Yang Yingliang wrote:
> I got the following UAF report:
>
> refcount_t: underflow; use-after-free.
> WARNING: CPU: 1 PID: 270 at lib/refcount.c:29 refcount_warn_saturate+0x121/0x180
> ...
> OF: ERROR: memory leak, expected refcount 1 instead of -1073741824,
> of_node_get()/of_node_put() unbalanced - destroy cset entry:
> attach overlay node /i2c/pmic@62/powerkey
>
> The of_node of device assigned in mfd_match_of_node_to_dev() need be
> get, and it will be put in platform_device_release().
>
> Fixes: 002be8114007 ("mfd: core: Add missing of_node_put for loop iteration")
> Signed-off-by: Yang Yingliang <[email protected]>
Is this patch good or do you have any suggestion ?

Thanks,
Yang
> ---
> drivers/mfd/mfd-core.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/mfd/mfd-core.c b/drivers/mfd/mfd-core.c
> index 16d1861e9682..8e57f67719cf 100644
> --- a/drivers/mfd/mfd-core.c
> +++ b/drivers/mfd/mfd-core.c
> @@ -161,7 +161,7 @@ static int mfd_match_of_node_to_dev(struct platform_device *pdev,
> of_entry->np = np;
> list_add_tail(&of_entry->list, &mfd_of_node_list);
>
> - pdev->dev.of_node = np;
> + pdev->dev.of_node = of_node_get(np);
> pdev->dev.fwnode = &np->fwnode;
> #endif
> return 0;

2022-12-08 15:34:19

by Lee Jones

Subject: Re: [PATCH] mfd: core: fix UAF while using device of node

On Thu, 08 Dec 2022, Yang Yingliang wrote:

> Hi,
>
> On 2022/11/16 15:41, Yang Yingliang wrote:
> > I got the following UAF report:
> >
> > refcount_t: underflow; use-after-free.
> > WARNING: CPU: 1 PID: 270 at lib/refcount.c:29 refcount_warn_saturate+0x121/0x180
> > ...
> > OF: ERROR: memory leak, expected refcount 1 instead of -1073741824,
> > of_node_get()/of_node_put() unbalanced - destroy cset entry:
> > attach overlay node /i2c/pmic@62/powerkey
> >
> > The of_node of device assigned in mfd_match_of_node_to_dev() need be
> > get, and it will be put in platform_device_release().
> >
> > Fixes: 002be8114007 ("mfd: core: Add missing of_node_put for loop iteration")
> > Signed-off-by: Yang Yingliang <[email protected]>
> Is this patch good or do you have any suggestion ?

Looks okay, but I'm not applying it this late in the cycle.

Please wait until the next merge-window closes.

--
Lee Jones [李琼斯]