Re: [RFC PATCH v3 6/8] x86/pti: don't mark the user PGD with _PAGE_NX.

From: Willy Tarreau
Date: Thu Jan 11 2018 - 01:28:17 EST


On Wed, Jan 10, 2018 at 12:20:05PM -0800, Dave Hansen wrote:
> Granted, you have an RFC on this, but please, for the love of everything
> that is good the world, please stop sending this patch set until you
> have a halfway reasonable method of dealing with NX that doesn't involve
> #ifdefs. Please.

Dave, people have been disagreeing for a while on two critical points :
- per-task vs per-mm setting
- stop or pass execve()

The discussion resulted from the patchset publication with some good
and bad points detected in each case (including your comment about the
conditional jump in the entry code based on CR3).

The method of dealing with NX will result from design choices, so for
me it should be handled last (but I'm find with changing the commit
message to mark it "BROKEN" to avoid confusion).

Thanks,
Willy