Skip to contentSkip to conversation

lkdtm: add bad USER_DS test

PATCH Superseded Kees Cook 4 participants 6 messages Mar 23, 2017 – Mar 24, 2017

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Kees Cook

author
#0 lore raw

PATCH 1/1 lkdtm: add bad USER_DS test

This adds CORRUPT_USER_DS to check that the get_fs() test on syscall return still sees USER_DS during the new VERIFY_PRE_USERMODE_STATE checks.

---
 drivers/misc/lkdtm.h      |  1 +
 drivers/misc/lkdtm_bugs.c | 20 ++++++++++++++++++++
 drivers/misc/lkdtm_core.c |  1 +
 3 files changed, 22 insertions(+)
drivers/misc/lkdtm.h +1 -0
index 67d27be60405..3b4976396ec4 100644
@@ -27,6 +27,7 @@ void lkdtm_REFCOUNT_ZERO_SUB(void);
 void lkdtm_REFCOUNT_ZERO_ADD(void);
 void lkdtm_CORRUPT_LIST_ADD(void);
 void lkdtm_CORRUPT_LIST_DEL(void);
+void lkdtm_CORRUPT_USER_DS(void);
 
 /* lkdtm_heap.c */
 void lkdtm_OVERWRITE_ALLOCATION(void);
drivers/misc/lkdtm_bugs.c +20 -0
index e3f4cd8876b5..4906e53a6df3 100644
@@ -8,6 +8,7 @@
 #include <linux/list.h>
 #include <linux/refcount.h>
 #include <linux/sched.h>
+#include <linux/uaccess.h>
 
 struct lkdtm_list {
 	struct list_head node;
@@ -279,3 +280,22 @@ void lkdtm_CORRUPT_LIST_DEL(void)
 	else
 		pr_err("list_del() corruption not detected!\n");
 }
+
+void lkdtm_CORRUPT_USER_DS(void)
+{
+	/*
+	 * Test that USER_DS has been set correctly on exiting a syscall.
+	 * Since setting this higher than USER_DS (TASK_SIZE) would introduce
+	 * an exploitable condition, we lower it instead, since that should
+	 * not create as large a problem on an unprotected system.
+	 */
+	mm_segment_t lowfs;
+#ifdef MAKE_MM_SEG
+	lowfs = MAKE_MM_SEG(TASK_SIZE - PAGE_SIZE);
+#else
+	lowfs = TASK_SIZE - PAGE_SIZE;
+#endif
+
+	pr_info("setting bad task size limit\n");
+	set_fs(lowfs);
+}
drivers/misc/lkdtm_core.c +1 -0
index b9a4cd4a9b68..42d2b8e31e6b 100644
@@ -199,6 +199,7 @@ struct crashtype crashtypes[] = {
 	CRASHTYPE(OVERFLOW),
 	CRASHTYPE(CORRUPT_LIST_ADD),
 	CRASHTYPE(CORRUPT_LIST_DEL),
+	CRASHTYPE(CORRUPT_USER_DS),
 	CRASHTYPE(CORRUPT_STACK),
 	CRASHTYPE(UNALIGNED_LOAD_STORE_WRITE),
 	CRASHTYPE(OVERWRITE_ALLOCATION),
Signature · 6 lines
2.7.4


-- 
Kees Cook
Pixel Security
#1 lore raw
Quoted 4 lines · Kees Cook

This adds CORRUPT_USER_DS to check that the get_fs() test on syscall return still sees USER_DS during the new VERIFY_PRE_USERMODE_STATE checks.

...

Quoted 8 lines
drivers/misc/lkdtm_bugs.c +8 -0
+void lkdtm_CORRUPT_USER_DS(void)
+{
+	/*
+	 * Test that USER_DS has been set correctly on exiting a syscall.
+	 * Since setting this higher than USER_DS (TASK_SIZE) would introduce
+	 * an exploitable condition, we lower it instead, since that should
+	 * not create as large a problem on an unprotected system.
+	 */

drivers/misc/lkdtm_bugs.c void lkdtm_CORRUPT_LIST_DEL(void) · line 301

Quoted 10 lines
drivers/misc/lkdtm_bugs.c +10 -0
+	mm_segment_t lowfs;
+#ifdef MAKE_MM_SEG
+	lowfs = MAKE_MM_SEG(TASK_SIZE - PAGE_SIZE);
+#else
+	lowfs = TASK_SIZE - PAGE_SIZE;
+#endif
+
+	pr_info("setting bad task size limit\n");
+	set_fs(lowfs);
+}

This won't work on architectures where the set_fs() argument does not contain an address but an address space identifier. This is true e.g. for s390 and as far as I know also for sparc.
On s390 we have complete distinct address spaces for kernel and user space that each start at address zero.

#2 lore raw
Quoted 24 lines · Heiko Carstens
Quoted 4 lines · Kees Cook

This adds CORRUPT_USER_DS to check that the get_fs() test on syscall return still sees USER_DS during the new VERIFY_PRE_USERMODE_STATE checks.

...

Quoted 16 lines
drivers/misc/lkdtm_bugs.c +16 -0
+void lkdtm_CORRUPT_USER_DS(void)
+{
+     /*
+      * Test that USER_DS has been set correctly on exiting a syscall.
+      * Since setting this higher than USER_DS (TASK_SIZE) would introduce
+      * an exploitable condition, we lower it instead, since that should
+      * not create as large a problem on an unprotected system.
+      */
+     mm_segment_t lowfs;
+#ifdef MAKE_MM_SEG
+     lowfs = MAKE_MM_SEG(TASK_SIZE - PAGE_SIZE);
+#else
+     lowfs = TASK_SIZE - PAGE_SIZE;
+#endif
+
+     pr_info("setting bad task size limit\n");
Quoted 9 lines

drivers/misc/lkdtm_bugs.c void lkdtm_CORRUPT_LIST_DEL(void) · line 301

Quoted 2 lines
drivers/misc/lkdtm_bugs.c +2 -0
+     set_fs(lowfs);
+}

This won't work on architectures where the set_fs() argument does not contain an address but an address space identifier. This is true e.g. for s390 and as far as I know also for sparc.
On s390 we have complete distinct address spaces for kernel and user space that each start at address zero.

The patch that enforce USER_DS is disabled on s390 anyway. I guess, we can just do a set_fs(KERNEL_DS) for the others.

Signature · 1 line
Thomas
#3 lore raw
Quoted 36 lines · Thomas Garnier
Quoted 24 lines · Heiko Carstens
Quoted 4 lines · Kees Cook

This adds CORRUPT_USER_DS to check that the get_fs() test on syscall return still sees USER_DS during the new VERIFY_PRE_USERMODE_STATE checks.

...

Quoted 16 lines
drivers/misc/lkdtm_bugs.c +16 -0
+void lkdtm_CORRUPT_USER_DS(void)
+{
+     /*
+      * Test that USER_DS has been set correctly on exiting a syscall.
+      * Since setting this higher than USER_DS (TASK_SIZE) would introduce
+      * an exploitable condition, we lower it instead, since that should
+      * not create as large a problem on an unprotected system.
+      */
+     mm_segment_t lowfs;
+#ifdef MAKE_MM_SEG
+     lowfs = MAKE_MM_SEG(TASK_SIZE - PAGE_SIZE);
+#else
+     lowfs = TASK_SIZE - PAGE_SIZE;
+#endif
+
+     pr_info("setting bad task size limit\n");
Quoted 9 lines

drivers/misc/lkdtm_bugs.c void lkdtm_CORRUPT_LIST_DEL(void) · line 301

Quoted 2 lines
drivers/misc/lkdtm_bugs.c +2 -0
+     set_fs(lowfs);
+}

This won't work on architectures where the set_fs() argument does not contain an address but an address space identifier. This is true e.g. for s390 and as far as I know also for sparc.
On s390 we have complete distinct address spaces for kernel and user space that each start at address zero.

Quoted 2 lines

The patch that enforce USER_DS is disabled on s390 anyway. I guess, we can just do a set_fs(KERNEL_DS) for the others.

that would enable the test, but it would also mean that lkdtm can be used by a program to escalate its rights. I think that is the reason why Kees did this lowfs things.

#4 lore raw
Quoted 37 lines · Christian Borntraeger
Quoted 36 lines · Thomas Garnier
Quoted 33 lines · Heiko Carstens
Quoted 4 lines · Kees Cook

This adds CORRUPT_USER_DS to check that the get_fs() test on syscall return still sees USER_DS during the new VERIFY_PRE_USERMODE_STATE checks.

...

Quoted 8 lines
drivers/misc/lkdtm_bugs.c +8 -0
+void lkdtm_CORRUPT_USER_DS(void)
+{
+     /*
+      * Test that USER_DS has been set correctly on exiting a syscall.
+      * Since setting this higher than USER_DS (TASK_SIZE) would introduce
+      * an exploitable condition, we lower it instead, since that should
+      * not create as large a problem on an unprotected system.
+      */

drivers/misc/lkdtm_bugs.c void lkdtm_CORRUPT_LIST_DEL(void) · line 301

Quoted 10 lines
drivers/misc/lkdtm_bugs.c +10 -0
+     mm_segment_t lowfs;
+#ifdef MAKE_MM_SEG
+     lowfs = MAKE_MM_SEG(TASK_SIZE - PAGE_SIZE);
+#else
+     lowfs = TASK_SIZE - PAGE_SIZE;
+#endif
+
+     pr_info("setting bad task size limit\n");
+     set_fs(lowfs);
+}

This won't work on architectures where the set_fs() argument does not contain an address but an address space identifier. This is true e.g. for s390 and as far as I know also for sparc.
On s390 we have complete distinct address spaces for kernel and user space that each start at address zero.

Quoted 7 lines
Quoted 2 lines

The patch that enforce USER_DS is disabled on s390 anyway. I guess, we can just do a set_fs(KERNEL_DS) for the others.

that would enable the test, but it would also mean that lkdtm can be used by a program to escalate its rights. I think that is the reason why Kees did this lowfs things.

I see, I need to look at how lkdtm works exactly.

Signature · 1 line
Thomas
author
#5 lore raw
In reply to Christian Borntraeger
Quoted 37 lines · Christian Borntraeger
Quoted 36 lines · Thomas Garnier
Quoted 33 lines · Heiko Carstens
Quoted 4 lines · Kees Cook

This adds CORRUPT_USER_DS to check that the get_fs() test on syscall return still sees USER_DS during the new VERIFY_PRE_USERMODE_STATE checks.

...

Quoted 8 lines
drivers/misc/lkdtm_bugs.c +8 -0
+void lkdtm_CORRUPT_USER_DS(void)
+{
+     /*
+      * Test that USER_DS has been set correctly on exiting a syscall.
+      * Since setting this higher than USER_DS (TASK_SIZE) would introduce
+      * an exploitable condition, we lower it instead, since that should
+      * not create as large a problem on an unprotected system.
+      */

drivers/misc/lkdtm_bugs.c void lkdtm_CORRUPT_LIST_DEL(void) · line 301

Quoted 10 lines
drivers/misc/lkdtm_bugs.c +10 -0
+     mm_segment_t lowfs;
+#ifdef MAKE_MM_SEG
+     lowfs = MAKE_MM_SEG(TASK_SIZE - PAGE_SIZE);
+#else
+     lowfs = TASK_SIZE - PAGE_SIZE;
+#endif
+
+     pr_info("setting bad task size limit\n");
+     set_fs(lowfs);
+}

This won't work on architectures where the set_fs() argument does not contain an address but an address space identifier. This is true e.g. for s390 and as far as I know also for sparc.
On s390 we have complete distinct address spaces for kernel and user space that each start at address zero.

Quoted 6 lines
Quoted 2 lines

The patch that enforce USER_DS is disabled on s390 anyway. I guess, we can just do a set_fs(KERNEL_DS) for the others.

that would enable the test, but it would also mean that lkdtm can be used by a program to escalate its rights. I think that is the reason why Kees did this lowfs things.

Yeah, but it seems like getting this right for all architectures isn't sane. I'm going to change it to use KERNEL_DS but also post a SIGKILL to the process. That way it'll still trigger the syscall return checking, but will be unable to continue running with a potentially uncaught KERNEL_DS.

-Kees

Signature · 2 lines
Kees Cook
Pixel Security