Re: [PATCH] waitid(): Avoid unbalanced user_access_end() on access_ok() error

From: David Daney
Date: Fri Oct 20 2017 - 11:11:51 EST


On 10/20/2017 07:36 AM, Kees Cook wrote:
As pointed out by Linus and David, the earlier waitid() fix resulted in a
(currently harmless) unbalanced user_access_end() call. This fixes it to
just directly return EFAULT on access_ok() failure.

Fixes: 96ca579a1ecc ("waitid(): Add missing access_ok() checks")
Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
Cc: David Daney <ddaney@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>

Thanks Kees, this looks better to me ...

Acked-by: David Daney <david.daney@xxxxxxxxxx>

---
This seems best to get fixed up now (in the same release where it was
introduced) before it tries to bite us at some later time.
---
kernel/exit.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/exit.c b/kernel/exit.c
index cf28528842bc..f6cad39f35df 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -1611,7 +1611,7 @@ SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
return err;
if (!access_ok(VERIFY_WRITE, infop, sizeof(*infop)))
- goto Efault;
+ return -EFAULT;
user_access_begin();
unsafe_put_user(signo, &infop->si_signo, Efault);
@@ -1739,7 +1739,7 @@ COMPAT_SYSCALL_DEFINE5(waitid,
return err;
if (!access_ok(VERIFY_WRITE, infop, sizeof(*infop)))
- goto Efault;
+ return -EFAULT;
user_access_begin();
unsafe_put_user(signo, &infop->si_signo, Efault);