[PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set

[PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set

[Andi Kleen]

Based on patch from Frank van Maarseveen <frankvm@frankvm.com>, but
extended.

Signed-off-by: Andi Kleen <ak@suse.de>

---
 arch/i386/kernel/process.c   |    3 ++-
 arch/x86_64/kernel/process.c |    2 +-
 fs/binfmt_elf.c              |    3 ++-
 3 files changed, 5 insertions(+), 3 deletions(-)

Index: linux/arch/x86_64/kernel/process.c
===================================================================
--- linux.orig/arch/x86_64/kernel/process.c
+++ linux/arch/x86_64/kernel/process.c
@@ -845,7 +845,7 @@ int dump_task_regs(struct task_struct *t
 
 unsigned long arch_align_stack(unsigned long sp)
 {
-	if (randomize_va_space)
+	if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
 		sp -= get_random_int() % 8192;
 	return sp & ~0xf;
 }
Index: linux/arch/i386/kernel/process.c
===================================================================
--- linux.orig/arch/i386/kernel/process.c
+++ linux/arch/i386/kernel/process.c
@@ -37,6 +37,7 @@
 #include <linux/kallsyms.h>
 #include <linux/ptrace.h>
 #include <linux/random.h>
+#include <linux/personality.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -905,7 +906,7 @@ asmlinkage int sys_get_thread_area(struc
 
 unsigned long arch_align_stack(unsigned long sp)
 {
-	if (randomize_va_space)
+	if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
 		sp -= get_random_int() % 8192;
 	return sp & ~0xf;
 }
Index: linux/fs/binfmt_elf.c
===================================================================
--- linux.orig/fs/binfmt_elf.c
+++ linux/fs/binfmt_elf.c
@@ -515,7 +515,8 @@ static unsigned long randomize_stack_top
 {
 	unsigned int random_variable = 0;
 
-	if (current->flags & PF_RANDOMIZE) {
+	if ((current->flags & PF_RANDOMIZE) &&
+		!(current->personality & ADDR_NO_RANDOMIZE)) {
 		random_variable = get_random_int() & STACK_RND_MASK;
 		random_variable <<= PAGE_SHIFT;
 	}

Re: [PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set

[Ingo Molnar]

* Andi Kleen <ak@suse.de> wrote:

>  unsigned long arch_align_stack(unsigned long sp)
>  {
> -	if (randomize_va_space)
> +	if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
>  		sp -= get_random_int() % 8192;
>  	return sp & ~0xf;

i'm not opposing this patch at all, but didnt the performance problems 
go away when the 0xf was changed to 0x7f?

looks good otherwise.

Acked-by: Ingo Molnar <mingo@elte.hu>

	Ingo

Re: [PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set

[Ingo Molnar]

* Andi Kleen <ak@suse.de> wrote:

> On Sunday 16 July 2006 22:47, Ingo Molnar wrote:
> > * Andi Kleen <ak@suse.de> wrote:
> > >  unsigned long arch_align_stack(unsigned long sp)
> > >  {
> > > -	if (randomize_va_space)
> > > +	if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
> > >  		sp -= get_random_int() % 8192;
> > >  	return sp & ~0xf;
> >
> > i'm not opposing this patch at all, but didnt the performance problems
> > go away when the 0xf was changed to 0x7f?
> 
> Yes, but i sent the patch before that other patch was available.

excuses, excuses ;)

> I guess it's a separate issue anyways - this patch is just concerned 
> about disabling randomization consistently. Performance optimization 
> can be done in another one.

yeah. There's one security issue: the 'dont randomize' flag must be 
cleared when we cross a protection domain. When for example suid-ing in 
exec().

	Ingo

Re: [PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set

[Andi Kleen]

On Sunday 16 July 2006 22:47, Ingo Molnar wrote:
> * Andi Kleen <ak@suse.de> wrote:
> >  unsigned long arch_align_stack(unsigned long sp)
> >  {
> > -	if (randomize_va_space)
> > +	if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
> >  		sp -= get_random_int() % 8192;
> >  	return sp & ~0xf;
>
> i'm not opposing this patch at all, but didnt the performance problems
> go away when the 0xf was changed to 0x7f?

Yes, but i sent the patch before that other patch was available.

I guess it's a separate issue anyways - this patch is just concerned about 
disabling randomization consistently. Performance optimization can be done
in another one.

-Andi

Re: [PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set

[Andi Kleen]

> yeah. There's one security issue: the 'dont randomize' flag must be
> cleared when we cross a protection domain. When for example suid-ing in
> exec().

Good point.

-Andi

Re: [discuss] Re: [PATCH for 2.6.18-rc2] [2/8] i386/x86-64: Don't randomize stack top when no randomization personality is set II

[Andi Kleen]

> yeah. There's one security issue: the 'dont randomize' flag must be
> cleared when we cross a protection domain. When for example suid-ing in
> exec().

Just checked. It should be already done because ADDR_NO_RANDOMIZE
is in PER_CLEAR_ON_SETID which is cleared in exec.

-Andi