xen: defer building p2m mfn structures until kernel is mapped

When building mfn parts of p2m structure, we rely on being able to
use mfn_to_virt, which in turn requires kernel to be mapped into
the linear area (which is distinct from the kernel image mapping
on 64-bit).  Defer calling xen_build_mfn_list_list() until after
xen_setup_kernel_pagetable();

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

2 files changed, 3 insertions, 3 deletions

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index ee304b52d8b7..d8873014b5ed 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1178,6 +1178,9 @@ asmlinkage void __init xen_start_kernel(void)
         xen_raw_console_write("mapping kernel into physical memory\n");
         pgd = xen_setup_kernel_pagetable(pgd, xen_start_info->nr_pages);
 
+        /* Allocate and initialize top and mid mfn levels for p2m structure */
+        xen_build_mfn_list_list();
+
         init_mm.pgd = pgd;
 
         /* keep using Xen gdt for now; no urgent need to change it */
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index b96513437236..9b43bb398d37 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -374,9 +374,6 @@ void __init xen_build_dynamic_phys_to_machine(void)
 
                 p2m_top[topidx][mididx] = &mfn_list[pfn];
         }
-
-        /* Allocate and initialize top and mid mfn levels */
-        xen_build_mfn_list_list();
 }
 
 unsigned long get_phys_to_machine(unsigned long pfn)