[PATCH] ppc32: kernel mapping breakage

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

Hi !

Christoph Lameter's patch that change page allocators to use GFP_ZERO
broke ppc32 in a subtle way. Our allocator is designed to work before
mem_init_done, in which cases it uses a ppc specific early_get_page()
which doesn't return zeroed pages. However, he removed the call to
clear_page() unconditionally, thus causing the kernel initial page
tables to have random data in them.

They are initialized with set_pte, which means it's _mostly_ harmless,
except that set_pte on ppc32 preserves the _PAGE_HASHPTE bit, thus we
end up with random bits there, which can cause issues with further
manipulation of the kernel page tables and will slow down all hash
faults to them causing unnecessary searches.

Please apply in 2.6.11 if still possible ...

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

Index: linux-work/arch/ppc/mm/pgtable.c
===================================================================
--- linux-work.orig/arch/ppc/mm/pgtable.c	2005-01-24 17:09:23.000000000 +1100
+++ linux-work/arch/ppc/mm/pgtable.c	2005-02-23 14:31:41.000000000 +1100
@@ -107,8 +107,11 @@
 			ptepage->mapping = (void *) mm;
 			ptepage->index = address & PMD_MASK;
 		}
-	} else
+	} else {
 		pte = (pte_t *)early_get_page();
+		if (pte)
+			clear_page(pte);
+	}
 	return pte;
 }