Announcement

Collapse
No announcement yet.

AMD Adds Secure Video Playback To Their Open-Source Linux Driver

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #51
    Originally posted by atomsymbol View Post

    Will it be possible to take screenshots? If not, I am going to disable Secure Video Playback on my machine. In Windows, the screen area protected by DRM is blank in a screenshot.
    Well isn't that the whole point? Its to prevent people capturing (by screenshot or otherwise) a DRM video (note that I am not justifying this, just stating that this is one of the DRM goals).

    Comment


    • #52
      DRM. Defective by design.

      Comment


      • #53
        Originally posted by bemerk View Post
        On the one hand i often hear that every Windows system should be replaced with Linux but on the other hand the same hardliners are unwilling to do the necessary trade-offs. If there is no good streaming experience due to the lack of DRM methods on linux systems, it becomes hard to convince people to switch.
        Nobody forces you to use it, Linux was always about multiple options. KDE, Gnome, XFCE or whatever, nobody forces you to something there either.
        Some trade-offs are harder to make than others. For me, the question is not Windows, it's been entirely removed from any home computer at least 12 years ago, hence out of the equation.

        15 months ago I bought a 65" TV on which I have the netflix app that gives me 4K/HDR + Atmos via ARC on my AV receiver. It's not theatre quality but it's pretty damn good.
        Really fortunate that I did, because until then I was stuck with 720p and 2.0 audio on my HTPC, let's say 1080p when abovementioned add-on was actually working,
        The HTPC AV output is passing through the same capable AV receiver and displayed on the same TV. And yet there's such an entire world of difference.

        I don't turn that HTPC on as often these days (Prime video and Spotify do not require it either). So if your leisureflow involves watching shows/movies a lot with good video AND audio quality, then it might be a reason good enough not to use Linux. If getting this secure video playback on Linux allows to turn that reason around, I only see a win. And I don't care about any "yeah, but it's DRM".

        Comment


        • #54
          Originally posted by MadeUpName View Post
          The streaming services don't care and are never going to care about Linux DRMed up the whazoo or not. Netflix just increased my rates again even though they only give us a fraction of the content the US gets and they haven't been producing any new content this year so I dropped the hammer on them.
          Try using a VPN. I live in Germany and lived in Italy and ExpressVPN has allowed me to see whatever country I wish.
          GOD is REAL unless declared as an INTEGER.

          Comment


          • #55
            Atrocious

            Anyway... I wonder if this could actually be disabled during buildtime.

            Alternatively it would be interesting to see if it's possible to hackz it and use it to rip 4k content

            Comment


            • #56
              Originally posted by FireBurn View Post
              I'll hopefully get a Van Gough laptop with AV1 decode when they come out - any sign of discreet mobile RDNA2 yet?
              Nothing much since September. Expect a maximum of 4 cores, not 6-8. Van Gogh sometime in 2021, and Dragon Crest will be the successor, probably with Zen 3 cores.

              Comment


              • #57
                Originally posted by smitty3268 View Post
                If you think about it logically, if you are able to take a screenshot of the DRM'd video, that means there is nothing stopping you from taking screenshots of every single frame of that video. At which point, you have duplicated the entire video - making the DRM broken.
                The implication that if it is possible to take a screenshot then it is possible to do it every single frame is false. Preventing a screenshot every frame can be trivially solved by a adding a timeout of 1 second to the DRM implementation.

                I don't have a deep knowledge of Secure Video Playback, for example I don't understand how it can prevent redirection of the si_vid_create_tmz_buffer function to si_vid_create_buffer in a local kernel source code patch.

                Comment


                • #58
                  Originally posted by atomsymbol View Post

                  The implication that if it is possible to take a screenshot then it is possible to do it every single frame is false. Preventing a screenshot every frame can be trivially solved by a adding a timeout of 1 second to the DRM implementation.
                  That sounds incredibly easy to workaround in various ways, like slowing down the video or trimming the start.
                  I don't have a deep knowledge of Secure Video Playback, for example I don't understand how it can prevent redirection of the si_vid_create_tmz_buffer function to si_vid_create_buffer in a local kernel source code patch.
                  It can't. The buffer itself is accessible but it's encrypted so if you try to view it all you'll get is random noise rather than the picture you expect to see. Decryption is only available on the hardware (or more likely it's baked into the firmware).

                  At least, I assume that's how it works. Presumably the DRM works by using the hardware video decoding which generates the encrypted buffer before it's accessible on the kernel side.
                  Last edited by smitty3268; 06 November 2020, 03:24 PM.

                  Comment


                  • #59
                    Originally posted by smitty3268 View Post
                    That sounds incredibly easy to workaround in various ways, like slowing down the video or trimming the start.
                    Trimming the start? I don't understand.

                    Comment


                    • #60
                      Originally posted by atomsymbol View Post

                      Trimming the start? I don't understand.
                      Play the video and record a frame every second.

                      Start the video again from the start, but offset 1 frame forward, then record again a frame every second.

                      Repeat until you've gone through all the frames in the full second.

                      Or, just pause the video for 1 second, then unpause and move to the next frame, etc.

                      It might take 30 times longer to copy, but that's still trivial to script and automate.

                      The hardware can't know the details of what exactly the software stack above it is doing, so it has to work on a very low level.

                      Comment

                      Working...
                      X