Re: [PATCH 02/38] media: v4l2-ioctl: avoid memory leaks on some time32 compat functions

[Arnd Bergmann]

On Wed, Sep 2, 2020 at 6:10 PM Mauro Carvalho Chehab
<mchehab+huawei@xxxxxxxxxx> wrote:
>
> There are some reports about possible memory leaks:
>
>         drivers/media/v4l2-core//v4l2-ioctl.c:3203 video_put_user() warn: check that 'ev32' doesn't leak information (struct has a hole after 'type')
>         drivers/media/v4l2-core//v4l2-ioctl.c:3230 video_put_user() warn: check that 'vb32' doesn't leak information (struct has a hole after 'memory')
>
> While smatch seems to be reporting a false positive (line 3203),
> there's indeed a possible leak with reserved2 at vb32.
>
> We might have fixed just that one, but smatch checks won't
> be able to check leaks at ev32. So, re-work the code in a way
> that will ensure that the var contents will be zeroed before
> filling it.
>
> With that, we don't need anymore to touch reserved fields.
>
> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@xxxxxxxxxx>

Isn't this the same as commit 4ffb879ea648 ("media: media/v4l2-core:
Fix kernel-infoleak
in video_put_user()") that you already applied (aside from the issue
that Laurent
pointed out)?

       Arnd