2023-03-07 16:02:04

by Zheng Wang

Subject: [PATCH] media: hantro: fix use after free bug in hantro_remove due to race condition

In hantro_probe, vpu->watchdog_work is bound with
hantro_watchdog. Then hantro_end_prepare_run may
be called to start the work.

If we close the file or remove the module which will
call hantro_release and hantro_remove to make cleanup,
there may be a unfinished work. The possible sequence
is as follows, which will cause a typical UAF bug.

The same thing will happen in hantro_release, and use
ctx after freeing it.

Fix it by canceling the work before cleanup in hantro_release.

CPU0 CPU1

|hantro_watchdog
hantro_remove |
v4l2_m2m_release |
kfree(m2m_dev); |
|
| v4l2_m2m_get_curr_priv
| m2m_dev->curr_ctx //use

Signed-off-by: Zheng Wang <[email protected]>
---
drivers/media/platform/verisilicon/hantro_drv.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/drivers/media/platform/verisilicon/hantro_drv.c b/drivers/media/platform/verisilicon/hantro_drv.c
index b0aeedae7b65..80bd856a4da9 100644
--- a/drivers/media/platform/verisilicon/hantro_drv.c
+++ b/drivers/media/platform/verisilicon/hantro_drv.c
@@ -1099,6 +1099,7 @@ static int hantro_remove(struct platform_device *pdev)

v4l2_info(&vpu->v4l2_dev, "Removing %s\n", pdev->name);

+ cancel_delayed_work(&vpu->watchdog_work);
media_device_unregister(&vpu->mdev);
hantro_remove_dec_func(vpu);
hantro_remove_enc_func(vpu);
--
2.25.1

2023-03-13 15:09:13

by Hans Verkuil

Subject: Re: [PATCH] media: hantro: fix use after free bug in hantro_remove due to race condition

On 07/03/2023 16:41, Zheng Wang wrote:
> In hantro_probe, vpu->watchdog_work is bound with
> hantro_watchdog. Then hantro_end_prepare_run may
> be called to start the work.
>
> If we close the file or remove the module which will
> call hantro_release and hantro_remove to make cleanup,
> there may be a unfinished work. The possible sequence
> is as follows, which will cause a typical UAF bug.
>
> The same thing will happen in hantro_release, and use
> ctx after freeing it.
>
> Fix it by canceling the work before cleanup in hantro_release.
>
> CPU0 CPU1
>
> |hantro_watchdog
> hantro_remove |
> v4l2_m2m_release |
> kfree(m2m_dev); |
> |
> | v4l2_m2m_get_curr_priv
> | m2m_dev->curr_ctx //use
>
> Signed-off-by: Zheng Wang <[email protected]>
> ---
> drivers/media/platform/verisilicon/hantro_drv.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/media/platform/verisilicon/hantro_drv.c b/drivers/media/platform/verisilicon/hantro_drv.c
> index b0aeedae7b65..80bd856a4da9 100644
> --- a/drivers/media/platform/verisilicon/hantro_drv.c
> +++ b/drivers/media/platform/verisilicon/hantro_drv.c
> @@ -1099,6 +1099,7 @@ static int hantro_remove(struct platform_device *pdev)
>
> v4l2_info(&vpu->v4l2_dev, "Removing %s\n", pdev->name);
>
> + cancel_delayed_work(&vpu->watchdog_work);

Use cancel_delayed_work_sync(): that ensures the code waits for a running
watchdog function to finish.

Ditto for the other patch.

Regards,

Hans

> media_device_unregister(&vpu->mdev);
> hantro_remove_dec_func(vpu);
> hantro_remove_enc_func(vpu);

2023-03-13 15:10:17

by Hans Verkuil

Subject: Re: [PATCH] media: hantro: fix use after free bug in hantro_remove due to race condition

On 13/03/2023 16:08, Hans Verkuil wrote:
> On 07/03/2023 16:41, Zheng Wang wrote:
>> In hantro_probe, vpu->watchdog_work is bound with
>> hantro_watchdog. Then hantro_end_prepare_run may
>> be called to start the work.
>>
>> If we close the file or remove the module which will
>> call hantro_release and hantro_remove to make cleanup,
>> there may be a unfinished work. The possible sequence
>> is as follows, which will cause a typical UAF bug.
>>
>> The same thing will happen in hantro_release, and use
>> ctx after freeing it.
>>
>> Fix it by canceling the work before cleanup in hantro_release.
>>
>> CPU0 CPU1
>>
>> |hantro_watchdog
>> hantro_remove |
>> v4l2_m2m_release |
>> kfree(m2m_dev); |
>> |
>> | v4l2_m2m_get_curr_priv
>> | m2m_dev->curr_ctx //use
>>
>> Signed-off-by: Zheng Wang <[email protected]>
>> ---
>> drivers/media/platform/verisilicon/hantro_drv.c | 1 +
>> 1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/media/platform/verisilicon/hantro_drv.c b/drivers/media/platform/verisilicon/hantro_drv.c
>> index b0aeedae7b65..80bd856a4da9 100644
>> --- a/drivers/media/platform/verisilicon/hantro_drv.c
>> +++ b/drivers/media/platform/verisilicon/hantro_drv.c
>> @@ -1099,6 +1099,7 @@ static int hantro_remove(struct platform_device *pdev)
>>
>> v4l2_info(&vpu->v4l2_dev, "Removing %s\n", pdev->name);
>>
>> + cancel_delayed_work(&vpu->watchdog_work);
>
> Use cancel_delayed_work_sync(): that ensures the code waits for a running
> watchdog function to finish.
>
> Ditto for the other patch.

And also the cedrus and rkvdec patches.

Regards,

Hans

>
> Regards,
>
> Hans
>
>> media_device_unregister(&vpu->mdev);
>> hantro_remove_dec_func(vpu);
>> hantro_remove_enc_func(vpu);
>

2023-03-13 15:22:09

by Zheng Hacker

Subject: Re: [PATCH] media: hantro: fix use after free bug in hantro_remove due to race condition

Hans Verkuil <[email protected]> 于2023年3月13日周一 23:08写道：
>
> On 07/03/2023 16:41, Zheng Wang wrote:
> > In hantro_probe, vpu->watchdog_work is bound with
> > hantro_watchdog. Then hantro_end_prepare_run may
> > be called to start the work.
> >
> > If we close the file or remove the module which will
> > call hantro_release and hantro_remove to make cleanup,
> > there may be a unfinished work. The possible sequence
> > is as follows, which will cause a typical UAF bug.
> >
> > The same thing will happen in hantro_release, and use
> > ctx after freeing it.
> >
> > Fix it by canceling the work before cleanup in hantro_release.
> >
> > CPU0 CPU1
> >
> > |hantro_watchdog
> > hantro_remove |
> > v4l2_m2m_release |
> > kfree(m2m_dev); |
> > |
> > | v4l2_m2m_get_curr_priv
> > | m2m_dev->curr_ctx //use
> >
> > Signed-off-by: Zheng Wang <[email protected]>
> > ---
> > drivers/media/platform/verisilicon/hantro_drv.c | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/media/platform/verisilicon/hantro_drv.c b/drivers/media/platform/verisilicon/hantro_drv.c
> > index b0aeedae7b65..80bd856a4da9 100644
> > --- a/drivers/media/platform/verisilicon/hantro_drv.c
> > +++ b/drivers/media/platform/verisilicon/hantro_drv.c
> > @@ -1099,6 +1099,7 @@ static int hantro_remove(struct platform_device *pdev)
> >
> > v4l2_info(&vpu->v4l2_dev, "Removing %s\n", pdev->name);
> >
> > + cancel_delayed_work(&vpu->watchdog_work);
>
> Use cancel_delayed_work_sync(): that ensures the code waits for a running
> watchdog function to finish.
>
> Ditto for the other patch.
>

Dear Hans,

Thanks for your advice. Will do in the next version of patch :)

Best regards,
Zheng

>
> > media_device_unregister(&vpu->mdev);
> > hantro_remove_dec_func(vpu);
> > hantro_remove_enc_func(vpu);
>

2023-03-13 15:23:36

by Zheng Hacker

Subject: Re: [PATCH] media: hantro: fix use after free bug in hantro_remove due to race condition

Hans Verkuil <[email protected]> 于2023年3月13日周一 23:09写道：
>
> On 13/03/2023 16:08, Hans Verkuil wrote:
> > On 07/03/2023 16:41, Zheng Wang wrote:
> >> In hantro_probe, vpu->watchdog_work is bound with
> >> hantro_watchdog. Then hantro_end_prepare_run may
> >> be called to start the work.
> >>
> >> If we close the file or remove the module which will
> >> call hantro_release and hantro_remove to make cleanup,
> >> there may be a unfinished work. The possible sequence
> >> is as follows, which will cause a typical UAF bug.
> >>
> >> The same thing will happen in hantro_release, and use
> >> ctx after freeing it.
> >>
> >> Fix it by canceling the work before cleanup in hantro_release.
> >>
> >> CPU0 CPU1
> >>
> >> |hantro_watchdog
> >> hantro_remove |
> >> v4l2_m2m_release |
> >> kfree(m2m_dev); |
> >> |
> >> | v4l2_m2m_get_curr_priv
> >> | m2m_dev->curr_ctx //use
> >>
> >> Signed-off-by: Zheng Wang <[email protected]>
> >> ---
> >> drivers/media/platform/verisilicon/hantro_drv.c | 1 +
> >> 1 file changed, 1 insertion(+)
> >>
> >> diff --git a/drivers/media/platform/verisilicon/hantro_drv.c b/drivers/media/platform/verisilicon/hantro_drv.c
> >> index b0aeedae7b65..80bd856a4da9 100644
> >> --- a/drivers/media/platform/verisilicon/hantro_drv.c
> >> +++ b/drivers/media/platform/verisilicon/hantro_drv.c
> >> @@ -1099,6 +1099,7 @@ static int hantro_remove(struct platform_device *pdev)
> >>
> >> v4l2_info(&vpu->v4l2_dev, "Removing %s\n", pdev->name);
> >>
> >> + cancel_delayed_work(&vpu->watchdog_work);
> >
> > Use cancel_delayed_work_sync(): that ensures the code waits for a running
> > watchdog function to finish.
> >
> > Ditto for the other patch.
>
> And also the cedrus and rkvdec patches.
>
> Regards,
>
Dear Hans,

Thanks for your review. I'll use cancel_delayed_work_sync in the next version.

Best regards,
Zheng
> >
> >> media_device_unregister(&vpu->mdev);
> >> hantro_remove_dec_func(vpu);
> >> hantro_remove_enc_func(vpu);
> >
>