Skip to contentSkip to conversation

core, x86: make LIST_POISON less deadly

PATCH Superseded Avi Kivity 2 participants 2 messages Dec 20, 2009 – Dec 28, 2009 x86

lore
Share & tools
lore

Avi Kivity

author
#0 lore raw

PATCH 1/1 core, x86: make LIST_POISON less deadly

From: Avi Kivity <[email protected]>

The list macros use LIST_POISON1 and LIST_POISON2 as undereferencable pointers in order to trap erronous use of freed list_heads. Unfortunately userspace can arrange for those pointers to actually be dereferencable, potentially turning an oops to an expolit.

To avoid this allow architectures (currently x86_64 only) to override the default values for these pointers with truly-undereferencable values. This is easy on x86_64 as the virtual address space is large and contains areas that cannot be mapped.

Other 64-bit architectures will likely find similar unmapped ranges.

[ingo: switch to 0xdead000000000000 as the unmapped area] [ingo: add comments, cleanup]
[jaswinder: eliminate sparse warnings]

---

Linus, Ingo won't push this for some reason, but he has no objection to me submitting the patch and AFAICT no objections to the patch itself.

 arch/x86/Kconfig       |    5 +++++
 include/linux/poison.h |   16 ++++++++++++++--
 2 files changed, 19 insertions(+), 2 deletions(-)
arch/x86/Kconfig +5 -0
index 3b2a5ac..098f0d3 100644
@@ -1244,6 +1244,11 @@ config ARCH_MEMORY_PROBE
 	def_bool X86_64
 	depends on MEMORY_HOTPLUG
 
+config ILLEGAL_POINTER_VALUE
+       hex
+       default 0 if X86_32
+       default 0xdead000000000000 if X86_64
+
 source "mm/Kconfig"
 
 config HIGHPTE
include/linux/poison.h +14 -2
index 7fc194a..2110a81 100644
@@ -2,13 +2,25 @@
 #define _LINUX_POISON_H
 
 /********** include/linux/list.h **********/
+
+/*
+ * Architectures might want to move the poison pointer offset
+ * into some well-recognized area such as 0xdead000000000000,
+ * that is also not mappable by user-space exploits:
+ */
+#ifdef CONFIG_ILLEGAL_POINTER_VALUE
+# define POISON_POINTER_DELTA _AC(CONFIG_ILLEGAL_POINTER_VALUE, UL)
+#else
+# define POISON_POINTER_DELTA 0
+#endif
+
 /*
  * These are non-NULL pointers that will result in page faults
  * under normal circumstances, used to verify that nobody uses
  * non-initialized list entries.
  */
-#define LIST_POISON1  ((void *) 0x00100100)
-#define LIST_POISON2  ((void *) 0x00200200)
+#define LIST_POISON1  ((void *) 0x00100100 + POISON_POINTER_DELTA)
+#define LIST_POISON2  ((void *) 0x00200200 + POISON_POINTER_DELTA)
 
 /********** include/linux/timer.h **********/
 /*
Signature · 1 line
1.6.5.3
#1 lore raw

Hi Linus,

Quoted 15 lines · On Sun, 20 Dec 2009 15:22:30 +0200 Avi Kivity <[email protected]> wrote:

From: Avi Kivity <[email protected]>

The list macros use LIST_POISON1 and LIST_POISON2 as undereferencable pointers in order to trap erronous use of freed list_heads. Unfortunately userspace can arrange for those pointers to actually be dereferencable, potentially turning an oops to an expolit.

To avoid this allow architectures (currently x86_64 only) to override the default values for these pointers with truly-undereferencable values. This is easy on x86_64 as the virtual address space is large and contains areas that cannot be mapped.

Other 64-bit architectures will likely find similar unmapped ranges.

Quoted 8 lines

[ingo: switch to 0xdead000000000000 as the unmapped area] [ingo: add comments, cleanup]
[jaswinder: eliminate sparse warnings]

The above fixed version of this patch has been in linux-next (via the tip tree) for one year minus one day.

Signature · 3 lines
Cheers,
Stephen Rothwell                    [email protected]
http://www.canb.auug.org.au/~sfr/