On Thu. 5 Dec. 2024 at 03:58, David Laight [off-list ref] wrote:

From: Vincent Mailhol

quoted

Sent: 02 December 2024 17:33

From: Vincent Mailhol <redacted>

is_const() is a one to one replacement of __is_constexpr(). Do the
replacement so that __is_constexpr() can be removed.

Signed-off-by: Vincent Mailhol <redacted>
---
 include/linux/fortify-string.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/include/linux/fortify-string.h b/include/linux/fortify-string.h
index 0d99bf11d260a3482bbe46e35c7553c0ccfb8b94..e3f2f772c5439ef71eb4a904b4ce27956bc69743 100644
--- a/include/linux/fortify-string.h
+++ b/include/linux/fortify-string.h

@@ -254,8 +254,8 @@ __FORTIFY_INLINE __kernel_size_t strnlen(const char * const POS p, __kernel_size
  * Returns number of characters in @p (NOT including the final NUL).
  *
  */
-#define strlen(p)                                                    \
-     __builtin_choose_expr(__is_constexpr(__builtin_strlen(p)),      \
+#define strlen(p)                                            \
+     __builtin_choose_expr(is_const(__builtin_strlen(p)),    \
              __builtin_strlen(p), __fortify_strlen(p))

I'm sure Linus suggested a way of doing that without replicating
the __builtin_strlen().

Indeed it may be valid to do:
        len = __builtin_strlen(p);
        __builtin_constant_p(len) ? len : __fortify_strlen(p);

Then, wouldn't it be better for strlen() to be an inline function
instead of a macro?

  __FORTIFY_INLINE __kernel_size_t strlen(const char *p)
  {
          __kernel_size_t ret = __builtin_strlen(p);

          if (__builtin_constant_p(ret))
                  return ret;
          return __fortify_strlen(p);
  }

I tested it and it worked on an allyesconfig. So if I receive no
objections, strlen() will become an inline function in v2.


Yours sincerely,
Vincent Mailhol