Skip to contentSkip to conversation

Security-related list changes for 4.10

PATCH New Paul E. McKenney 1 participant 8 messages Nov 14, 2016

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover Security-related list changes for 4.10

Hello!

This series contains security-related list changes, some of which stems in part from PaX and Grsecurity work:

1. Split __list_add() debug checks into a separate functions,
courtesy of Kees Cook.

2. Consolidate DEBUG_LIST() for list_add_rcu(), courtesy of Kees Cook.

3. Split list_del() debug checking into separate function, courtesy
of Kees Cook.

4. Provide toggle for BUG() on data corruption, so that people who
care more about security than uptime can crash their machines in the face of list data corruption, courtesy of Kees Cook.

Note that CHECK_DATA_CORRUPTION() contains unconventional control flow, however, this is intended and is used correctly.

5. Add tests for list corruption, courtesy of Kees Cook.

6. Fix typo in "select" Kconfig statement in #4 above, courtesy
of Valentin Rothberg.

7. Avoid Kconfig warning from CONFIG_DEBUG_LIST from #4 above,
courtesy of Arnd Bergmann.

Thanx, Paul

------------------------------------------------------------------------

 drivers/misc/lkdtm.h      |    2 
 drivers/misc/lkdtm_bugs.c |   68 +++++++++++++++++
 drivers/misc/lkdtm_core.c |    2 
 include/linux/bug.h       |   17 ++++
 include/linux/list.h      |   37 ++++++---
 include/linux/rculist.h   |    8 --
 lib/Kconfig.debug         |   14 +++
 lib/list_debug.c          |  177 +++++++++++++++++-----------------------------
 8 files changed, 197 insertions(+), 128 deletions(-)
#1 lore raw

PATCH 1/7 list: Split list_add() debug checking into separate function

From: Kees Cook <[email protected]>

Right now, __list_add() code is repeated either in list.h or in list_debug.c, but the only differences between the two versions are the debug checks. This commit therefore extracts these debug checks into a separate __list_add_valid() function and consolidates __list_add(). Additionally this new __list_add_valid() function will stop list manipulations if a corruption is detected, instead of allowing for further corruption that may lead to even worse conditions.

This is slight refactoring of the same hardening done in PaX and Grsecurity.

---
 include/linux/list.h | 22 ++++++++++++++++------
 lib/list_debug.c     | 48 +++++++++++++++++++++++-------------------------
 2 files changed, 39 insertions(+), 31 deletions(-)
include/linux/list.h +16 -6
index 5809e9a2de5b..b6da9b1dce4d 100644
@@ -28,27 +28,37 @@ static inline void INIT_LIST_HEAD(struct list_head *list)
 	list->prev = list;
 }
 
+#ifdef CONFIG_DEBUG_LIST
+extern bool __list_add_valid(struct list_head *new,
+			      struct list_head *prev,
+			      struct list_head *next);
+#else
+static inline bool __list_add_valid(struct list_head *new,
+				struct list_head *prev,
+				struct list_head *next)
+{
+	return true;
+}
+#endif
+
 /*
  * Insert a new entry between two known consecutive entries.
  *
  * This is only for internal list manipulation where we know
  * the prev/next entries already!
  */
-#ifndef CONFIG_DEBUG_LIST
 static inline void __list_add(struct list_head *new,
 			      struct list_head *prev,
 			      struct list_head *next)
 {
+	if (!__list_add_valid(new, prev, next))
+		return;
+
 	next->prev = new;
 	new->next = next;
 	new->prev = prev;
 	WRITE_ONCE(prev->next, new);
 }
-#else
-extern void __list_add(struct list_head *new,
-			      struct list_head *prev,
-			      struct list_head *next);
-#endif
 
 /**
  * list_add - add a new entry
lib/list_debug.c +23 -25

64 lines not on this page · raw message

Signature · 1 line
2.5.2
#2 lore raw

PATCH 2/7 rculist: Consolidate DEBUG_LIST for list_add_rcu()

In reply to Paul E. McKenney

From: Kees Cook <[email protected]>

This commit consolidates the debug checking for list_add_rcu() into the new single __list_add_valid() debug function. Notably, this commit fixes the sanity check that was added in commit 17a801f4bfeb ("list_debug:
WARN for adding something already in the list"), which wasn't checking RCU-protected lists.

---
 include/linux/rculist.h |  8 +++-----
 lib/list_debug.c        | 19 -------------------
 2 files changed, 3 insertions(+), 24 deletions(-)
include/linux/rculist.h +3 -5
index 8beb98dcf14f..4f7a9561b8c4 100644
@@ -45,19 +45,17 @@ static inline void INIT_LIST_HEAD_RCU(struct list_head *list)
  * This is only for internal list manipulation where we know
  * the prev/next entries already!
  */
-#ifndef CONFIG_DEBUG_LIST
 static inline void __list_add_rcu(struct list_head *new,
 		struct list_head *prev, struct list_head *next)
 {
+	if (!__list_add_valid(new, prev, next))
+		return;
+
 	new->next = next;
 	new->prev = prev;
 	rcu_assign_pointer(list_next_rcu(prev), new);
 	next->prev = new;
 }
-#else
-void __list_add_rcu(struct list_head *new,
-		    struct list_head *prev, struct list_head *next);
-#endif
 
 /**
  * list_add_rcu - add a new entry to rcu-protected list
lib/list_debug.c +0 -19
index 149dd57b583b..d0b89b9d0736 100644
@@ -77,22 +77,3 @@ void list_del(struct list_head *entry)
 	entry->prev = LIST_POISON2;
 }
 EXPORT_SYMBOL(list_del);
-
-/*
- * RCU variants.
- */
-void __list_add_rcu(struct list_head *new,
-		    struct list_head *prev, struct list_head *next)
-{
-	WARN(next->prev != prev,
-		"list_add_rcu corruption. next->prev should be prev (%p), but was %p. (next=%p).\n",
-		prev, next->prev, next);
-	WARN(prev->next != next,
-		"list_add_rcu corruption. prev->next should be next (%p), but was %p. (prev=%p).\n",
-		next, prev->next, prev);
-	new->next = next;
-	new->prev = prev;
-	rcu_assign_pointer(list_next_rcu(prev), new);
-	next->prev = new;
-}
-EXPORT_SYMBOL(__list_add_rcu);
Signature · 1 line
2.5.2
#3 lore raw

PATCH 3/7 list: Split list_del() debug checking into separate function

In reply to Paul E. McKenney

From: Kees Cook <[email protected]>

Similar to the list_add() debug consolidation, this commit consolidates the debug checking performed during CONFIG_DEBUG_LIST into a new __list_del_entry_valid() function, and stops list updates when corruption is found.

Refactored from same hardening in PaX and Grsecurity.

---
 include/linux/list.h | 15 +++++++++------
 lib/list_debug.c     | 53 +++++++++++++++++++++++-----------------------------
 2 files changed, 32 insertions(+), 36 deletions(-)
include/linux/list.h +9 -6
index b6da9b1dce4d..d1039ecaf94f 100644
@@ -32,6 +32,7 @@ static inline void INIT_LIST_HEAD(struct list_head *list)
 extern bool __list_add_valid(struct list_head *new,
 			      struct list_head *prev,
 			      struct list_head *next);
+extern bool __list_del_entry_valid(struct list_head *entry);
 #else
 static inline bool __list_add_valid(struct list_head *new,
 				struct list_head *prev,
@@ -39,6 +40,10 @@ static inline bool __list_add_valid(struct list_head *new,
 {
 	return true;
 }
+static inline bool __list_del_entry_valid(struct list_head *entry)
+{
+	return true;
+}
 #endif
 
 /*
@@ -106,22 +111,20 @@ static inline void __list_del(struct list_head * prev, struct list_head * next)
  * Note: list_empty() on entry does not return true after this, the entry is
  * in an undefined state.
  */
-#ifndef CONFIG_DEBUG_LIST
 static inline void __list_del_entry(struct list_head *entry)
 {
+	if (!__list_del_entry_valid(entry))
+		return;
+
 	__list_del(entry->prev, entry->next);
 }
 
 static inline void list_del(struct list_head *entry)
 {
-	__list_del(entry->prev, entry->next);
+	__list_del_entry(entry);
 	entry->next = LIST_POISON1;
 	entry->prev = LIST_POISON2;
 }
-#else
-extern void __list_del_entry(struct list_head *entry);
-extern void list_del(struct list_head *entry);
-#endif
 
 /**
  * list_replace - replace old entry by new one
lib/list_debug.c +23 -30

64 lines not on this page · raw message

Signature · 1 line
2.5.2
#4 lore raw

PATCH 4/7 bug: Provide toggle for BUG on data corruption

In reply to Paul E. McKenney

From: Kees Cook <[email protected]>

The kernel checks for cases of data structure corruption under some CONFIGs (e.g. CONFIG_DEBUG_LIST). When corruption is detected, some systems may want to BUG() immediately instead of letting the system run with known corruption. Usually these kinds of manipulation primitives can be used by security flaws to gain arbitrary memory write control. This provides a new config CONFIG_BUG_ON_DATA_CORRUPTION and a corresponding macro CHECK_DATA_CORRUPTION for handling these situations. Notably, even if not BUGing, the kernel should not continue processing the corrupted structure.

This is inspired by similar hardening by Syed Rameez Mustafa in MSM kernels, and in PaX and Grsecurity, which is likely in response to earlier removal of the BUG calls in commit 924d9addb9b1 ("list debugging: use WARN() instead of BUG()").

---
 include/linux/bug.h | 17 ++++++++++++++++
 lib/Kconfig.debug   | 10 ++++++++++
 lib/list_debug.c    | 57 +++++++++++++++++++++--------------------------------
 3 files changed, 49 insertions(+), 35 deletions(-)
include/linux/bug.h +17 -0
index 292d6a10b0c2..baff2e8fc8a8 100644
@@ -121,4 +121,21 @@ static inline enum bug_trap_type report_bug(unsigned long bug_addr,
 }
 
 #endif	/* CONFIG_GENERIC_BUG */
+
+/*
+ * Since detected data corruption should stop operation on the affected
+ * structures, this returns false if the corruption condition is found.
+ */
+#define CHECK_DATA_CORRUPTION(condition, fmt, ...)			 \
+	do {								 \
+		if (unlikely(condition)) {				 \
+			if (IS_ENABLED(CONFIG_BUG_ON_DATA_CORRUPTION)) { \
+				pr_err(fmt, ##__VA_ARGS__);		 \
+				BUG();					 \
+			} else						 \
+				WARN(1, fmt, ##__VA_ARGS__);		 \
+			return false;					 \
+		}							 \
+	} while (0)
+
 #endif	/* _LINUX_BUG_H */
lib/Kconfig.debug +10 -0
index 33bc56cf60d7..07a6fac930c5 100644
@@ -1960,6 +1960,16 @@ config TEST_STATIC_KEYS
 
 	  If unsure, say N.
 
+config BUG_ON_DATA_CORRUPTION
+	bool "Trigger a BUG when data corruption is detected"
+	select CONFIG_DEBUG_LIST
+	help
+	  Select this option if the kernel should BUG when it encounters
+	  data corruption in kernel memory structures when they get checked
+	  for validity.
+
+	  If unsure, say N.
+
 source "samples/Kconfig"
 
 source "lib/Kconfig.kgdb"
lib/list_debug.c +22 -35

69 lines not on this page · raw message

Signature · 1 line
2.5.2
#5 lore raw

PATCH 5/7 lkdtm: Add tests for struct list corruption

In reply to Paul E. McKenney

From: Kees Cook <[email protected]>

When building under CONFIG_DEBUG_LIST, list addition and removal will be sanity-checked. This validates that the check is working as expected by setting up classic corruption attacks against list manipulations, available with the new lkdtm tests CORRUPT_LIST_ADD and CORRUPT_LIST_DEL.

---
 drivers/misc/lkdtm.h      |  2 ++
 drivers/misc/lkdtm_bugs.c | 68 +++++++++++++++++++++++++++++++++++++++++++++++
 drivers/misc/lkdtm_core.c |  2 ++
 3 files changed, 72 insertions(+)
drivers/misc/lkdtm.h +2 -0
index fdf954c2107f..cfa1039c62e7 100644
@@ -21,6 +21,8 @@ void lkdtm_SPINLOCKUP(void);
 void lkdtm_HUNG_TASK(void);
 void lkdtm_ATOMIC_UNDERFLOW(void);
 void lkdtm_ATOMIC_OVERFLOW(void);
+void lkdtm_CORRUPT_LIST_ADD(void);
+void lkdtm_CORRUPT_LIST_DEL(void);
 
 /* lkdtm_heap.c */
 void lkdtm_OVERWRITE_ALLOCATION(void);
drivers/misc/lkdtm_bugs.c +68 -0

79 lines not on this page · raw message

drivers/misc/lkdtm_core.c +2 -0
index f9154b8d67f6..7eeb71a75549 100644
@@ -197,6 +197,8 @@ struct crashtype crashtypes[] = {
 	CRASHTYPE(EXCEPTION),
 	CRASHTYPE(LOOP),
 	CRASHTYPE(OVERFLOW),
+	CRASHTYPE(CORRUPT_LIST_ADD),
+	CRASHTYPE(CORRUPT_LIST_DEL),
 	CRASHTYPE(CORRUPT_STACK),
 	CRASHTYPE(UNALIGNED_LOAD_STORE_WRITE),
 	CRASHTYPE(OVERWRITE_ALLOCATION),
Signature · 1 line
2.5.2
#6 lore raw

PATCH 6/7 lib/Kconfig.debug: Fix typo in select statement

In reply to Paul E. McKenney

From: Valentin Rothberg <[email protected]>

Commit 484f29c7430b3 ("bug: Provide toggle for BUG on data corruption") added a Kconfig select statement on CONFIG_DEBUG_LIST, but the CONFIG_ prefix is only used in Make and C(PP) syntax. Remove the CONFIG_ prefix to correctly select the Kconfig option DEBUG_LIST.

---
 lib/Kconfig.debug | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/Kconfig.debug +1 -1
index 07a6fac930c5..afa30fd52583 100644
@@ -1962,7 +1962,7 @@ config TEST_STATIC_KEYS
 
 config BUG_ON_DATA_CORRUPTION
 	bool "Trigger a BUG when data corruption is detected"
-	select CONFIG_DEBUG_LIST
+	select DEBUG_LIST
 	help
 	  Select this option if the kernel should BUG when it encounters
 	  data corruption in kernel memory structures when they get checked
Signature · 1 line
2.5.2
#7 lore raw

PATCH 7/7 bug: Avoid Kconfig warning for BUG_ON_DATA_CORRUPTION

In reply to Paul E. McKenney

From: Arnd Bergmann <[email protected]>

The CONFIG_DEBUG_LIST option is normally meant for kernel developers rather than production machines and is guarded by CONFIG_DEBUG_KERNEL.

In contrast, the newly added CONFIG_BUG_ON_DATA_CORRUPTION is meant for security hardening and may be used on systems that intentionally do not enable CONFIG_DEBUG_KERNEL. In this configuration, we get a warning from Kconfig about the mismatched dependencies:

warning: (BUG_ON_DATA_CORRUPTION) selects DEBUG_LIST which has unmet direct dependencies (DEBUG_KERNEL)

This annotates the DEBUG_LIST option to be selectable by BUG_ON_DATA_CORRUPTION when DEBUG_KERNEL is disabled.

Fixes 40cd725cfc7f ("bug: Provide toggle for BUG on data corruption")
---
 lib/Kconfig.debug | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/Kconfig.debug +1 -1
index afa30fd52583..0c6366b77062 100644
@@ -1214,7 +1214,7 @@ config DEBUG_BUGVERBOSE
 
 config DEBUG_LIST
 	bool "Debug linked list manipulation"
-	depends on DEBUG_KERNEL
+	depends on DEBUG_KERNEL || BUG_ON_DATA_CORRUPTION
 	help
 	  Enable this to turn on extended checks in the linked-list
 	  walking routines.
Signature · 1 line
2.5.2