[PATCH v2] x86/tls: Fix possible spectre-v1 in do_get_thread_area()

[PATCH v2] x86/tls: Fix possible spectre-v1 in do_get_thread_area()

[Dianzhang Chen]

The index to access the threads tls array is controlled by userspace
via syscall: sys_ptrace(), hence leading to a potential exploitation
of the Spectre variant 1 vulnerability.
The idx can be controlled from:
        ptrace -> arch_ptrace -> do_get_thread_area.

Fix this by sanitizing idx before using it to index p->thread.tls_array.

Signed-off-by: Dianzhang Chen <dianzhangchen0@gmail.com>
---
 arch/x86/kernel/tls.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/tls.c b/arch/x86/kernel/tls.c
index a5b802a..424cff5 100644
--- a/arch/x86/kernel/tls.c
+++ b/arch/x86/kernel/tls.c
@@ -5,6 +5,7 @@
 #include <linux/user.h>
 #include <linux/regset.h>
 #include <linux/syscalls.h>
+#include <linux/nospec.h>
 
 #include <linux/uaccess.h>
 #include <asm/desc.h>
@@ -220,6 +221,7 @@ int do_get_thread_area(struct task_struct *p, int idx,
 		       struct user_desc __user *u_info)
 {
 	struct user_desc info;
+	int index;
 
 	if (idx == -1 && get_user(idx, &u_info->entry_number))
 		return -EFAULT;
@@ -227,8 +229,9 @@ int do_get_thread_area(struct task_struct *p, int idx,
 	if (idx < GDT_ENTRY_TLS_MIN || idx > GDT_ENTRY_TLS_MAX)
 		return -EINVAL;
 
-	fill_user_desc(&info, idx,
-		       &p->thread.tls_array[idx - GDT_ENTRY_TLS_MIN]);
+	index = idx - GDT_ENTRY_TLS_MIN;
+
+	fill_user_desc(&info, idx, &p->thread.tls_array[index]);
 
 	if (copy_to_user(u_info, &info, sizeof(info)))
 		return -EFAULT;
-- 
2.7.4

Re: [PATCH v2] x86/tls: Fix possible spectre-v1 in do_get_thread_area()

[Thomas Gleixner]

On Wed, 26 Jun 2019, Dianzhang Chen wrote:

> The index to access the threads tls array is controlled by userspace
> via syscall: sys_ptrace(), hence leading to a potential exploitation
> of the Spectre variant 1 vulnerability.
> The idx can be controlled from:
>         ptrace -> arch_ptrace -> do_get_thread_area.
> 
> Fix this by sanitizing idx before using it to index p->thread.tls_array.

Just that I can't find a place which sanitizes the value....

> +#include <linux/nospec.h>

and nothing which uses anything from this header file.

>  #include <linux/uaccess.h>
>  #include <asm/desc.h>
> @@ -220,6 +221,7 @@ int do_get_thread_area(struct task_struct *p, int idx,
>  		       struct user_desc __user *u_info)
>  {
>  	struct user_desc info;
> +	int index;
>  
>  	if (idx == -1 && get_user(idx, &u_info->entry_number))
>  		return -EFAULT;
> @@ -227,8 +229,9 @@ int do_get_thread_area(struct task_struct *p, int idx,
>  	if (idx < GDT_ENTRY_TLS_MIN || idx > GDT_ENTRY_TLS_MAX)
>  		return -EINVAL;
>  
> -	fill_user_desc(&info, idx,
> -		       &p->thread.tls_array[idx - GDT_ENTRY_TLS_MIN]);
> +	index = idx - GDT_ENTRY_TLS_MIN;
> +
> +	fill_user_desc(&info, idx, &p->thread.tls_array[index]);

So this is just a cosmetic change and the compiler will create probably
exactly the same binary.

Thanks,

	tglx

Re: [PATCH v2] x86/tls: Fix possible spectre-v1 in do_get_thread_area()

[Dianzhang Chen]

On Wed, Jun 26, 2019 at 12:38 AM Thomas Gleixner <tglx@linutronix.de> wrote:
>
> On Wed, 26 Jun 2019, Dianzhang Chen wrote:
>
> > The index to access the threads tls array is controlled by userspace
> > via syscall: sys_ptrace(), hence leading to a potential exploitation
> > of the Spectre variant 1 vulnerability.
> > The idx can be controlled from:
> >         ptrace -> arch_ptrace -> do_get_thread_area.
> >
> > Fix this by sanitizing idx before using it to index p->thread.tls_array.
>
> Just that I can't find a place which sanitizes the value....
>
> > +#include <linux/nospec.h>
>
> and nothing which uses anything from this header file.
>
> >  #include <linux/uaccess.h>
> >  #include <asm/desc.h>
> > @@ -220,6 +221,7 @@ int do_get_thread_area(struct task_struct *p, int idx,
> >                      struct user_desc __user *u_info)
> >  {
> >       struct user_desc info;
> > +     int index;
> >
> >       if (idx == -1 && get_user(idx, &u_info->entry_number))
> >               return -EFAULT;
> > @@ -227,8 +229,9 @@ int do_get_thread_area(struct task_struct *p, int idx,
> >       if (idx < GDT_ENTRY_TLS_MIN || idx > GDT_ENTRY_TLS_MAX)
> >               return -EINVAL;
> >
> > -     fill_user_desc(&info, idx,
> > -                    &p->thread.tls_array[idx - GDT_ENTRY_TLS_MIN]);
> > +     index = idx - GDT_ENTRY_TLS_MIN;
> > +
> > +     fill_user_desc(&info, idx, &p->thread.tls_array[index]);
>
> So this is just a cosmetic change and the compiler will create probably
> exactly the same binary.
>
> Thanks,
>
>         tglx
>

sorry for being careless, my bad.
And thanks for suggestion, i'll fix that in next version.