Dave Hansen [off-list ref] writes:

On 7/23/20 10:08 AM, Andy Lutomirski wrote:

quoted

Suppose some kernel code (a syscall or kernel thread) changes PKRS
then takes a page fault. The page fault handler needs a fresh PKRS.
Then the page fault handler (say a VMA’s .fault handler) changes
PKRS.  The we get an interrupt. The interrupt *also* needs a fresh
PKRS and the page fault value needs to be saved somewhere.

So we have more than one saved value per thread, and thread_struct
isn’t going to solve this problem.

Taking a step back...  This is all true only if we decide that we want
protection keys to provide protection during exceptions and interrupts.
 Right now, the code supports nesting:

	kmap(foo);
		kmap(bar);
		kunmap(bar);
	kunmap(foo);

with a reference count.  So, the nested kmap() will see the count
elevated and do nothing.

Hopefully with a big fat warning if the nested map requires a different
key than the outer one.