Re: [PATCH v3 1/4] fbdev: Prevent possible use-after-free in fb_release()

[Javier Martinez Canillas]

Hello Thomas,

On 5/10/22 10:04, Thomas Zimmermann wrote:
> Hi
> 
> Am 10.05.22 um 00:42 schrieb Javier Martinez Canillas:
>> On 5/10/22 00:22, Andrzej Hajda wrote:
>>
>> [snip]
>>
>>>>    static void drm_fbdev_fb_destroy(struct fb_info *info)
>>>>    {
>>>> +       if (info->cmap.len)
>>>> +               fb_dealloc_cmap(&info->cmap);
>>>> +
>>>>           drm_fbdev_release(info->par);
>>>> +       framebuffer_release(info);
>>>
>>> I would put drm_fbdev_release at the beginning - it cancels workers
>>> which could expect cmap to be still valid.
>>>
>>
>> Indeed, you are correct again. [0] is the final version of the patch I've
>> but don't have an i915 test machine to give it a try. I'll test tomorrow
>> on my test systems to verify that it doesn't cause any regressions since
>> with other DRM drivers.
> 
> You have to go through all DRM drivers that call drm_fb_helper_fini() 
> and make sure that they free fb_info. For example armada appears to be 
> leaking now. [1]
>

But shouldn't fb_info be freed when unregister_framebuffer() is called
through drm_dev_unregister() ? AFAICT the call chain is the following:

drm_put_dev()
  drm_dev_unregister()
    drm_client_dev_unregister()
      drm_fbdev_client_unregister()
        drm_fb_helper_unregister_fbi()
          unregister_framebuffer()
            do_unregister_framebuffer()
              put_fb_info()
                drm_fbdev_fb_destroy()
                  framebuffer_release()

which is the reason why I believe that drm_fb_helper_fini() should be
an internal static function and only called from drm_fbdev_fb_destroy().

Drivers shouldn't really explicitly call this helper in my opinion.

-- 
Best regards,

Javier Martinez Canillas
Linux Engineering
Red Hat