When mutex_lock_interruptible() fails, the error return code of
tomoyo_update_domain() is not properly assigned.
To fix this bug, error is assigned with the return value of
mutex_lock_interruptible(), and then error is checked.
Reported-by: TOTE Robot <[email protected]>
Signed-off-by: Jia-Ju Bai <[email protected]>
---
security/tomoyo/domain.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/security/tomoyo/domain.c b/security/tomoyo/domain.c
index 98d985895ec8..7b9f9664dbe3 100644
--- a/security/tomoyo/domain.c
+++ b/security/tomoyo/domain.c
@@ -118,7 +118,8 @@ int tomoyo_update_domain(struct tomoyo_acl_info *new_entry, const int size,
->perm == 1 << TOMOYO_TYPE_EXECUTE))
goto out;
}
- if (mutex_lock_interruptible(&tomoyo_policy_lock))
+ error = mutex_lock_interruptible(&tomoyo_policy_lock);
+ if (error)
goto out;
list_for_each_entry_rcu(entry, list, list,
srcu_read_lock_held(&tomoyo_ss)) {
--
2.17.1
On 2021/03/06 22:03, Jia-Ju Bai wrote:
> When mutex_lock_interruptible() fails, the error return code of
> tomoyo_update_domain() is not properly assigned.
> To fix this bug, error is assigned with the return value of
> mutex_lock_interruptible(), and then error is checked.
Thanks for a patch, but this patch is wrong.
Since the variable "error" is initialized as
int error = is_delete ? -ENOENT : -ENOMEM;
at the beginning of this function, unconditionally overwriting
this variable with the return code of mutex_lock_interruptible() breaks
if (error && !is_delete) {
}
block of this function.
And the caller does not check if the return code is -EINTR
instead of -ENOENT or -ENOMEM.