[PATCH] Fix possible race with 4level-fixup.h

[PATCH] Fix possible race with 4level-fixup.h

[Benjamin Herrenschmidt]

Hi !

When using 4level-fixup.h, a PMD page may end up beeing freed before the
matching PGD entry is cleared due to the way the compatibility macros
work. This can cause nasty races on some architectures.

This patch fixes it by defining pud_clear() to be pgd_clear(). That
means we'll actually write 0 twice, a small price to pay here,
especially seeing how easy it is to convert to the new headers anyway
(hint hint, ppc & ppc64 patches as soon as 2.6.11 is out).

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

Index: linux-work/include/asm-generic/4level-fixup.h
===================================================================
--- linux-work.orig/include/asm-generic/4level-fixup.h	2005-01-24 17:09:49.000000000 +1100
+++ linux-work/include/asm-generic/4level-fixup.h	2005-02-17 18:10:38.000000000 +1100
@@ -24,7 +24,7 @@
 #define pud_bad(pud)			0
 #define pud_present(pud)		1
 #define pud_ERROR(pud)			do { } while (0)
-#define pud_clear(pud)			do { } while (0)
+#define pud_clear(pud)			pgd_clear((pgd_t *)(pud))
 
 #undef pud_free_tlb
 #define pud_free_tlb(tlb, x)            do { } while (0)

Re: [PATCH] Fix possible race with 4level-fixup.h

[Nick Piggin]

Benjamin Herrenschmidt wrote:
> Hi !
> 
> When using 4level-fixup.h, a PMD page may end up beeing freed before the
> matching PGD entry is cleared due to the way the compatibility macros
> work. This can cause nasty races on some architectures.
> 
> This patch fixes it by defining pud_clear() to be pgd_clear(). That
> means we'll actually write 0 twice, a small price to pay here,
> especially seeing how easy it is to convert to the new headers anyway
> (hint hint, ppc & ppc64 patches as soon as 2.6.11 is out).
> 
> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> 
> Index: linux-work/include/asm-generic/4level-fixup.h
> ===================================================================
> --- linux-work.orig/include/asm-generic/4level-fixup.h	2005-01-24 17:09:49.000000000 +1100
> +++ linux-work/include/asm-generic/4level-fixup.h	2005-02-17 18:10:38.000000000 +1100
> @@ -24,7 +24,7 @@
>  #define pud_bad(pud)			0
>  #define pud_present(pud)		1
>  #define pud_ERROR(pud)			do { } while (0)
> -#define pud_clear(pud)			do { } while (0)
> +#define pud_clear(pud)			pgd_clear((pgd_t *)(pud))
>  

Just a small nit - no cast needed here.

Re: [PATCH] Fix possible race with 4level-fixup.h

[Benjamin Herrenschmidt]

On Thu, 2005-02-17 at 19:33 +1100, Nick Piggin wrote:
> Benjamin Herrenschmidt wrote:
> > Hi !
> > 
> > When using 4level-fixup.h, a PMD page may end up beeing freed before the
> > matching PGD entry is cleared due to the way the compatibility macros
> > work. This can cause nasty races on some architectures.
> > 
> > This patch fixes it by defining pud_clear() to be pgd_clear(). That
> > means we'll actually write 0 twice, a small price to pay here,
> > especially seeing how easy it is to convert to the new headers anyway
> > (hint hint, ppc & ppc64 patches as soon as 2.6.11 is out).
> > 
> > Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> > 
> > Index: linux-work/include/asm-generic/4level-fixup.h
> > ===================================================================
> > --- linux-work.orig/include/asm-generic/4level-fixup.h	2005-01-24 17:09:49.000000000 +1100
> > +++ linux-work/include/asm-generic/4level-fixup.h	2005-02-17 18:10:38.000000000 +1100
> > @@ -24,7 +24,7 @@
> >  #define pud_bad(pud)			0
> >  #define pud_present(pud)		1
> >  #define pud_ERROR(pud)			do { } while (0)
> > -#define pud_clear(pud)			do { } while (0)
> > +#define pud_clear(pud)			pgd_clear((pgd_t *)(pud))
> >  
> 
> Just a small nit - no cast needed here.

Well, do you know ? pud is a pud_t* and the arch is free to implement
pgd_clear as an inline with strong typing no ? 

Ben.

Re: [PATCH] Fix possible race with 4level-fixup.h

[Nick Piggin]

Benjamin Herrenschmidt wrote:

>>>Index: linux-work/include/asm-generic/4level-fixup.h
>>>===================================================================
>>>--- linux-work.orig/include/asm-generic/4level-fixup.h	2005-01-24 17:09:49.000000000 +1100
>>>+++ linux-work/include/asm-generic/4level-fixup.h	2005-02-17 18:10:38.000000000 +1100
>>>@@ -24,7 +24,7 @@
>>> #define pud_bad(pud)			0
>>> #define pud_present(pud)		1
>>> #define pud_ERROR(pud)			do { } while (0)
>>>-#define pud_clear(pud)			do { } while (0)
>>>+#define pud_clear(pud)			pgd_clear((pgd_t *)(pud))
>>> 
>>
>>Just a small nit - no cast needed here.
> 
> 
> Well, do you know ? pud is a pud_t* and the arch is free to implement
> pgd_clear as an inline with strong typing no ? 
> 

Yeah but if you're using the 4level-fixup.h header, then you get
#define pud_t    pgd_t

Not that I really mind, but in this header we've just avoided
doing casts for that reason.

Nick