Re: Badness at arch/powerpc/mm/pgtable.c:286

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

On Tue, 2009-12-08 at 19:14 +0200, kordex - wrote:
> Hello,
> 
> I turned some debuging options in order to trace rcutree problems but
> after turning debugging on it began just to loop on this:
> 
> http://xnet.fi/opt/pics/pgtable_badness.jpg
> 1.5M   pgtable_badness.jpg
> 
> I am sorry that I could not produce full dmesg.
> 
> Kernel config: http://xnet.fi/opt/apps/lkml-2.6.32-vanilla.debug.config.txt

Thanks for your report.

Does that patch fixes it ?

powerpc/mm: Fix a WARN_ON() with CONFIG_DEBUG_PAGEALLOC and CONFIG_DEBUG_VM

Set need to call __set_pte_at() and not set_pte_at() from __change_page_attr()
since the later will perform checks with CONFIG_DEBUG_VM that aren't suitable
to the way we override an existing PTE. (More specifically, it doesn't let
you write over a present PTE).

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

Index: linux-work/arch/powerpc/mm/pgtable_32.c
===================================================================
--- linux-work.orig/arch/powerpc/mm/pgtable_32.c	2009-12-09 18:06:14.000000000 +1100
+++ linux-work/arch/powerpc/mm/pgtable_32.c	2009-12-09 18:06:16.000000000 +1100
@@ -356,7 +356,7 @@ static int __change_page_attr(struct pag
 		return 0;
 	if (!get_pteptr(&init_mm, address, &kpte, &kpmd))
 		return -EINVAL;
-	set_pte_at(&init_mm, address, kpte, mk_pte(page, prot));
+	__set_pte_at(&init_mm, address, kpte, mk_pte(page, prot));
 	wmb();
 #ifdef CONFIG_PPC_STD_MMU
 	flush_hash_pages(0, address, pmd_val(*kpmd), 1);