[ARM] Fix mm initialisation with write buffered write allocate caches

It seems that without the extra tlb flush, we may end up faulting
during the early kernel initialisation because the TLB can't see
the updated page tables.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

1 files changed, 8 insertions, 1 deletions

diff --git a/arch/arm/mm/init.c b/arch/arm/mm/init.c
index fd079ff1fc53..c168f322ef8c 100644
--- a/arch/arm/mm/init.c
+++ b/arch/arm/mm/init.c
@@ -486,10 +486,17 @@ static void __init devicemaps_init(struct machine_desc *mdesc)
 
         /*
          * Ask the machine support to map in the statically mapped devices.
-         * After this point, we can start to touch devices again.
          */
         if (mdesc->map_io)
                 mdesc->map_io();
+
+        /*
+         * Finally flush the tlb again - this ensures that we're in a
+         * consistent state wrt the writebuffer if the writebuffer needs
+         * draining.  After this point, we can start to touch devices
+         * again.
+         */
+        local_flush_tlb_all();
 }
 
 /*