[PATCH v2] audit: do a quick exit when syscall number is less than 0

cgel.zte@gmail.com posted 1 patch 4 years, 2 months ago
kernel/auditsc.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
[PATCH v2] audit: do a quick exit when syscall number is less than 0
Posted by cgel.zte@gmail.com 4 years, 2 months ago
From: Yang Yang <yang.yang29@zte.com.cn>

Userspace may use syscall with syscall number less than 0 by calling
syscall(syscall_num,..). This kind of syscall could never be audited,
because auditctl requires rule with syscall number >=0. Therefore we
better do a quick handle no need to gohead with this situation.

Note that auditctl may set rules auditing invalid syscall with syscall
number bigger than NR_syscalls, to keep this mechanism working, we do
no more check(context->major bigger than NR_syscalls or not).

Signed-off-by: Yang Yang <yang.yang29@zte.com.cn>
Reported-by: Zeal Robot <zealci@zte.com.cn>
---
v2:
- cancel checking against NR_syscalls
---
 kernel/auditsc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/auditsc.c b/kernel/auditsc.c
index ea2ee1181921..79118c811853 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -2077,7 +2077,8 @@ void __audit_syscall_exit(int success, long return_code)
 	struct audit_context *context = audit_context();
 
 	if (!context || context->dummy ||
-	    context->context != AUDIT_CTX_SYSCALL)
+	    context->context != AUDIT_CTX_SYSCALL ||
+	    unlikely(context->major < 0))
 		goto out;
 
 	/* this may generate CONFIG_CHANGE records */
-- 
2.25.1
Re: [PATCH v2] audit: do a quick exit when syscall number is less than 0
Posted by Paul Moore 4 years, 2 months ago
On Sun, Apr 3, 2022 at 10:23 PM <cgel.zte@gmail.com> wrote:
>
> From: Yang Yang <yang.yang29@zte.com.cn>
>
> Userspace may use syscall with syscall number less than 0 by calling
> syscall(syscall_num,..). This kind of syscall could never be audited,
> because auditctl requires rule with syscall number >=0. Therefore we
> better do a quick handle no need to gohead with this situation.
>
> Note that auditctl may set rules auditing invalid syscall with syscall
> number bigger than NR_syscalls, to keep this mechanism working, we do
> no more check(context->major bigger than NR_syscalls or not).
>
> Signed-off-by: Yang Yang <yang.yang29@zte.com.cn>
> Reported-by: Zeal Robot <zealci@zte.com.cn>
> ---
> v2:
> - cancel checking against NR_syscalls
> ---
>  kernel/auditsc.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)

As discussed previously, this is not something I want to merge
upstream at this time.

-- 
paul-moore.com