+/*M:008 We hold reference to pages, which prevents them from being swapped.
+ * It'd be nice to have a callback in the "struct mm_struct" when Linux wants
+ * to swap out. If we had this, and a shrinker callback to trim PTE pages, we
+ * could probably consider launching Guests as non-root. :*/
+
+/*H:300
+ * The Page Table Code
+ *
+ * We use two-level page tables for the Guest. If you're not entirely
+ * comfortable with virtual addresses, physical addresses and page tables then
+ * I recommend you review arch/x86/lguest/boot.c's "Page Table Handling" (with
+ * diagrams!).
+ *
+ * The Guest keeps page tables, but we maintain the actual ones here: these are
+ * called "shadow" page tables. Which is a very Guest-centric name: these are
+ * the real page tables the CPU uses, although we keep them up to date to
+ * reflect the Guest's. (See what I mean about weird naming? Since when do
+ * shadows reflect anything?)
+ *
+ * Anyway, this is the most complicated part of the Host code. There are seven
+ * parts to this:
+ * (i) Looking up a page table entry when the Guest faults,
+ * (ii) Making sure the Guest stack is mapped,
+ * (iii) Setting up a page table entry when the Guest tells us one has changed,
+ * (iv) Switching page tables,
+ * (v) Flushing (throwing away) page tables,
+ * (vi) Mapping the Switcher when the Guest is about to run,
+ * (vii) Setting up the page tables initially.
+ :*/
+
+
+/* 1024 entries in a page table page maps 1024 pages: 4MB. The Switcher is
+ * conveniently placed at the top 4MB, so it uses a separate, complete PTE
+ * page. */
+#define SWITCHER_PGD_INDEX (PTRS_PER_PGD - 1)
+
+/* We actually need a separate PTE page for each CPU. Remember that after the
+ * Switcher code itself comes two pages for each CPU, and we don't want this
+ * CPU's guest to see the pages of any other CPU. */
+static DEFINE_PER_CPU(pte_t *, switcher_pte_pages);