You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
…te (#265)
Add a magic header to measured information that should be loaded by the
host. Additionally, zero out the VTL2 boot data range, to validate that
on servicing operations, the host actually deposits new data.
This fixes the issue on ARM64 where this data was not actually written.
Thus, on a cold boot where memory is zeroed, the correct value of vtom =
0 was read, but during a servicing operation because this memory was not
part of the launch context, the value could be garbage and cause
failures later during initialization.
In the future, we should just have the bootloader parse the information
here, but getting the SLIT and PPTT info reconstructed is a bit
trickier. Tracked by #263
---------
Co-authored-by: Brian Perkins <[email protected]>
Unfortunately this gets a bit harder with the reserved region #304 . I don't know if we can easily remove some of these binary-only info, since packing them in device tree seems like the wrong thing to do
We shouldn't use /dev/mem to read data, and instead should use device tree and reconstruct what we need.
This is easy for the measured information, but harder for the SLIT/PPTT. However, we should be able to reconstruct this information from device tree.
The text was updated successfully, but these errors were encountered: