Re: [PATCH v2 5/5] security: Add CONFIG_SECURITY_HOOK_LIKELY
From: KP Singh <kpsingh@kernel.org>
Date: 2023-09-18 13:27:48
Also in:
bpf
On Tue, Jun 20, 2023 at 10:59 PM Kees Cook [off-list ref] wrote:
On Fri, Jun 16, 2023 at 02:04:41AM +0200, KP Singh wrote:quoted
[...]@@ -110,6 +110,9 @@ static __initdata struct lsm_info *exclusive; #undef LSM_HOOK #undef DEFINE_LSM_STATIC_CALL +#define security_hook_active(n, h) \ + static_branch_maybe(CONFIG_SECURITY_HOOK_LIKELY, &SECURITY_HOOK_ACTIVE_KEY(h, n)) + /* * Initialise a table of static calls for each LSM hook. * DEFINE_STATIC_CALL_NULL invocation above generates a key (STATIC_CALL_KEY)@@ -816,7 +819,7 @@ static int lsm_superblock_alloc(struct super_block *sb) */ #define __CALL_STATIC_VOID(NUM, HOOK, ...) \ do { \ - if (static_branch_unlikely(&SECURITY_HOOK_ACTIVE_KEY(HOOK, NUM))) { \ + if (security_hook_active(NUM, HOOK)) { \ static_call(LSM_STATIC_CALL(HOOK, NUM))(__VA_ARGS__); \ } \ } while (0);@@ -828,7 +831,7 @@ do { \ #define __CALL_STATIC_INT(NUM, R, HOOK, LABEL, ...) \ do { \ - if (static_branch_unlikely(&SECURITY_HOOK_ACTIVE_KEY(HOOK, NUM))) { \ + if (security_hook_active(NUM, HOOK)) { \ R = static_call(LSM_STATIC_CALL(HOOK, NUM))(__VA_ARGS__); \ if (R != 0) \ goto LABEL; \I actually think I'd prefer there be no macro wrapping static_branch_maybe(), just for reading it more easily. i.e. people reading this code are going to expect the static_branch/static_call code patterns, and seeing "security_hook_active" only slows them down in understanding it. I don't think it's _that_ ugly to have it all typed out. e.g.:
Done and agreed, especially given that this is behind a macro anyways.
if (static_branch_maybe(CONFIG_SECURITY_HOOK_LIKELY, \
&SECURITY_HOOK_ACTIVE_KEY(HOOK, NUM)) { \
R = static_call(LSM_STATIC_CALL(HOOK, NUM))(__VA_ARGS__); \
if (R != 0) \
goto LABEL; \
--
Kees Cook