Skip to contentSkip to conversation

x86/uaccess: Don't leak the AC flag into __put_user() value evaluation

PATCH In review Maxim Zhukov 3 participants 3 messages Mar 3, 2019 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/uaccess: Don't leak the AC flag into __put_user() value evaluation

From: Andy Lutomirski <[email protected]>

commit 2a418cf3f5f1caf911af288e978d61c9844b0695 upstream.

When calling __put_user(foo(), ptr), the __put_user() macro would call foo() in between __uaccess_begin() and __uaccess_end(). If that code were buggy, then those bugs would be run without SMAP protection.

Fortunately, there seem to be few instances of the problem in the kernel. Nevertheless, __put_user() should be fixed to avoid doing this. Therefore, evaluate __put_user()'s argument before setting AC.

This issue was noticed when an objtool hack by Peter Zijlstra complained about genregs_get() and I compared the assembly output to the C source.

[ bp: Massage commit message and fixed up whitespace. ]

Fixes 11f1a4b9755f ("x86: reorganize SMAP handling in user space accesses")
Cc Peter Zijlstra <[email protected]>
Cc Brian Gerst <[email protected]>
Cc Josh Poimboeuf <[email protected]>
Cc Denys Vlasenko <[email protected]>
---
 arch/x86/include/asm/uaccess.h | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/include/asm/uaccess.h +4 -2
index d788b0cdc0ad..4f4a9fa61b8a 100644
@@ -306,7 +306,7 @@ do {									\
 		__put_user_asm(x, ptr, retval, "l", "k", "ir", errret);	\
 		break;							\
 	case 8:								\
-		__put_user_asm_u64((__typeof__(*ptr))(x), ptr, retval,	\
+		__put_user_asm_u64(x, ptr, retval,	\
 				   errret);				\
 		break;							\
 	default:							\
@@ -410,7 +410,9 @@ do {									\
 #define __put_user_nocheck(x, ptr, size)			\
 ({								\
 	int __pu_err;						\
-	__put_user_size((x), (ptr), (size), __pu_err, -EFAULT);	\
+	__typeof__(*(ptr)) __pu_val;				\
+	__pu_val = x;						\
+	__put_user_size(__pu_val, (ptr), (size), __pu_err, -EFAULT);	\
 	__builtin_expect(__pu_err, 0);				\
 })
 
Signature · 1 line
2.21.0
#1 lore raw
Quoted 59 lines · Maxim Zhukov

From: Andy Lutomirski <[email protected]>

commit 2a418cf3f5f1caf911af288e978d61c9844b0695 upstream.

When calling __put_user(foo(), ptr), the __put_user() macro would call foo() in between __uaccess_begin() and __uaccess_end(). If that code were buggy, then those bugs would be run without SMAP protection.

Fortunately, there seem to be few instances of the problem in the kernel. Nevertheless, __put_user() should be fixed to avoid doing this. Therefore, evaluate __put_user()'s argument before setting AC.

This issue was noticed when an objtool hack by Peter Zijlstra complained about genregs_get() and I compared the assembly output to the C source.

[ bp: Massage commit message and fixed up whitespace. ]

Fixes 11f1a4b9755f ("x86: reorganize SMAP handling in user space accesses")
Cc Peter Zijlstra <[email protected]>
Cc Brian Gerst <[email protected]>
Cc Josh Poimboeuf <[email protected]>
Cc Denys Vlasenko <[email protected]>
---
 arch/x86/include/asm/uaccess.h | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/include/asm/uaccess.h +4 -2
index d788b0cdc0ad..4f4a9fa61b8a 100644
@@ -306,7 +306,7 @@ do {									\
 		__put_user_asm(x, ptr, retval, "l", "k", "ir", errret);	\
 		break;							\
 	case 8:								\
-		__put_user_asm_u64((__typeof__(*ptr))(x), ptr, retval,	\
+		__put_user_asm_u64(x, ptr, retval,	\
 				   errret);				\
 		break;							\
 	default:							\
@@ -410,7 +410,9 @@ do {									\
 #define __put_user_nocheck(x, ptr, size)			\
 ({								\
 	int __pu_err;						\
-	__put_user_size((x), (ptr), (size), __pu_err, -EFAULT);	\
+	__typeof__(*(ptr)) __pu_val;				\
+	__pu_val = x;						\
+	__put_user_size(__pu_val, (ptr), (size), __pu_err, -EFAULT);	\
 	__builtin_expect(__pu_err, 0);				\
 })
 
Signature · 1 line
2.21.0

Did you generate this against a current 4.4 tree? It fails to apply to 4.4.176 because this patch doesn't account for cc8c5450b6d7 ("x86:
reorganize SMAP handling in user space accesses"). Odds are you will only need one patch for 4.4 to 4.20.

Cheers,
Nathan

A message that is not in the archive (sent off-list or not yet received): CAFqsdWn2a4Q5kpmoei2WMix-cjVXuaciKQQYZJKmoNRhD9TSrA@mail.gmail.com

#3 lore raw
Quoted 3 lines · Maxim Zhukov

Whoops, this patch generated on old tree, sorry. I regenerate patch from new 4.4 tree. And test for all kernels (its work). Resend one patch for all stable branches?

If that is what works and is needed, yes.