Skip to contentSkip to conversation

Fix: kernel/ptrace.c: ptrace_peek_siginfo() missing __put_user() validation

PATCH In review Mathieu Desnoyers 3 participants 3 messages Jun 28, 2013 – Jun 29, 2013 ptrace support

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Fix: kernel/ptrace.c: ptrace_peek_siginfo() missing __put_user() validation

This __put_user() could be used by unprivileged processes to write into kernel memory. The issue here is that even if copy_siginfo_to_user() fails, the error code is not checked before __put_user() is executed. Luckily, ptrace_peek_siginfo() has been added within the 3.10-rc cycle, so it has not hit a stable release yet.

CC Andrey Vagin <[email protected]>
CC Roland McGrath <[email protected]>
CC Oleg Nesterov <[email protected]>
CC "Paul E. McKenney" <[email protected]>
CC David Howells <[email protected]>
CC Dave Jones <[email protected]>
CC Pavel Emelyanov <[email protected]>
CC Linus Torvalds <[email protected]>
Cc Pedro Alves <[email protected]>
CC Andrew Morton <[email protected]>
CC Linus Torvalds <[email protected]>
---
 kernel/ptrace.c |   20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)
kernel/ptrace.c +11 -9
Index: linux/kernel/ptrace.c
===================================================================
@@ -665,20 +665,22 @@ static int ptrace_peek_siginfo(struct ta
 		if (unlikely(is_compat_task())) {
 			compat_siginfo_t __user *uinfo = compat_ptr(data);
 
-			ret = copy_siginfo_to_user32(uinfo, &info);
-			ret |= __put_user(info.si_code, &uinfo->si_code);
+			if (copy_siginfo_to_user32(uinfo, &info) ||
+			    __put_user(info.si_code, &uinfo->si_code)) {
+				ret = -EFAULT;
+				break;
+			}
+
 		} else
 #endif
 		{
 			siginfo_t __user *uinfo = (siginfo_t __user *) data;
 
-			ret = copy_siginfo_to_user(uinfo, &info);
-			ret |= __put_user(info.si_code, &uinfo->si_code);
-		}
-
-		if (ret) {
-			ret = -EFAULT;
-			break;
+			if (copy_siginfo_to_user(uinfo, &info) ||
+			    __put_user(info.si_code, &uinfo->si_code)) {
+				ret = -EFAULT;
+				break;
+			}
 		}
 
 		data += sizeof(siginfo_t);
Signature · 3 lines
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com
maintainer Acked-by on v1 1/1
#1 lore raw
Quoted 6 lines · Mathieu Desnoyers

This __put_user() could be used by unprivileged processes to write into kernel memory. The issue here is that even if copy_siginfo_to_user() fails, the error code is not checked before __put_user() is executed. Luckily, ptrace_peek_siginfo() has been added within the 3.10-rc cycle, so it has not hit a stable release yet.

Agreed, this looks like 3.10 material.

Quoted 57 lines
CC Andrey Vagin <[email protected]>
CC Roland McGrath <[email protected]>
CC Oleg Nesterov <[email protected]>
CC "Paul E. McKenney" <[email protected]>
CC David Howells <[email protected]>
CC Dave Jones <[email protected]>
CC Pavel Emelyanov <[email protected]>
CC Linus Torvalds <[email protected]>
Cc Pedro Alves <[email protected]>
CC Andrew Morton <[email protected]>
CC Linus Torvalds <[email protected]>
---
 kernel/ptrace.c |   20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

Index: linux/kernel/ptrace.c
===================================================================

kernel/ptrace.c +11 -9
@@ -665,20 +665,22 @@ static int ptrace_peek_siginfo(struct ta
 		if (unlikely(is_compat_task())) {
 			compat_siginfo_t __user *uinfo = compat_ptr(data);
 
-			ret = copy_siginfo_to_user32(uinfo, &info);
-			ret |= __put_user(info.si_code, &uinfo->si_code);
+			if (copy_siginfo_to_user32(uinfo, &info) ||
+			    __put_user(info.si_code, &uinfo->si_code)) {
+				ret = -EFAULT;
+				break;
+			}
+
 		} else
 #endif
 		{
 			siginfo_t __user *uinfo = (siginfo_t __user *) data;
 
-			ret = copy_siginfo_to_user(uinfo, &info);
-			ret |= __put_user(info.si_code, &uinfo->si_code);
-		}
-
-		if (ret) {
-			ret = -EFAULT;
-			break;
+			if (copy_siginfo_to_user(uinfo, &info) ||
+			    __put_user(info.si_code, &uinfo->si_code)) {
+				ret = -EFAULT;
+				break;
+			}
 		}
 
 		data += sizeof(siginfo_t);
Signature · 4 lines
-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com
#2 lore raw
In reply to Mathieu Desnoyers
Quoted 5 lines · Mathieu Desnoyers

This __put_user() could be used by unprivileged processes to write into kernel memory. The issue here is that even if copy_siginfo_to_user() fails, the error code is not checked before __put_user() is executed. Luckily, ptrace_peek_siginfo() has been added within the 3.10-rc cycle, so it has not hit a stable release yet.

Why do those stupid

    __put_user(info.si_code, &uinfo->si_code))

things exist at all?

As far as I can tell, copy_siginfo_to_user[32]() already copies si_code. Is the field really so important that it has to be copied twice, just to make sure the write makes it?

Ok, ok, I see the explanation in the commit message that introduced this, but it sure as hell isn't obvious from actually looking at the code. I'm applying the patch, but I think this should have been a comment in the code as well.

Linus