|
|
|
|
|
|
by pornel
3639 days ago
|
|
|
> it was better to allow the w3c to specify a "black box" with well defined inputs and outputs. But they didn't! EME is a spec for only for inputs, and no outputs. EME entirely depends on CDMs, and their interface is deliberately left completely undefined (W3C uses that as an excuse to say they didn't—strictly speaking—define a DRM). Plug-ins at least had an open NPAPI interface that anybody could integrate with. CDMs don't have any public interface. The spec allows them to be anything, including kernel modules or hardware (and in practice they're… plug-ins). So now it's illegal under DMCA to write your own browser that plays EME-protected video with CDM compatible with Chrome's, Safari's or IE's. It's a loss of freedom, and no DRM has been removed or even relaxed in the process. |
|
|
On mobile platforms, they generally are system-integrated (and hardware-supported) components, often running at privilege levels exceeding the running Android/Linux kernel.
See the recent Qualcomm case where a DRM component (Widevine) running in TrustZone context[0] was used to attack Android's full disk encryption scheme.
[0] TrustZone is an ARM architecture feature for running code in a different execution context not accessible from the "normal" running kernel. Useful for running small amounts of code dedicated to protecting crypto keys, but horrible if you load gigantic DRM blobs into it that no one could reasonably audit due to sheer size even if their source code was available.