Tom Dohrmann

I'd argue that side effects from the reads are irrelevant here. What matters are the values that are read. Unless LLVM can somehow statically determine the value at the addresses...

We could wait for #404, but this also wouldn't be a deal breaker for me, especially because we don't seem to have come to a final design on the changes...

I'll get started on rebasing next onto master.

> I'll get started on rebasing next onto master. Hm, never mind, I forgot we don't do rebases. I'll get started on merging next into master instead.

> > I'll get started on rebasing next onto master. > > Hm, never mind, I forgot we don't do rebases. I'll get started on merging next into master instead....

@jarkkojs are you still working on this? If not, I'd like to pick this up. I already got a hacked-up prototype working, I would only have to clean it up.

[My work](https://github.com/Freax13/enarx/tree/feat/sev/upm) is compatible with the v7 patch set. It requires one additional patch: https://github.com/Freax13/linux/commit/b910c2271c4dac1353443fca98c7647414f8eebc. I'm not 100% if that patch is required though.

> > > [My work](https://github.com/Freax13/enarx/tree/feat/sev/upm) is compatible with the v7 patch set. It requires one additional patch: [Freax13/linux@b910c22](https://github.com/Freax13/linux/commit/b910c2271c4dac1353443fca98c7647414f8eebc). I'm not 100% if that patch is required though > > >...

This is still relevant,

correct