On Tue, Mar 02, 2021 at 09:02:54AM +1100, Daniel Axtens wrote:

Checkpatch does have one check that is relevant:

CHECK: Macro argument reuse 'p' - possible side-effects?
#36: FILE: arch/powerpc/include/asm/uaccess.h:482:
+#define unsafe_get_user(x, p, e) do {					\
+	if (unlikely(__get_user_nocheck((x), (p), sizeof(*(p)), false)))\
+		goto e;							\
+} while (0)

sizeof (of something other than a VLA) does not evaluate its operand.
The checkpatch warning is incorrect (well, it does say "possible" --
it just didn't find a possible problem here).

You can write
  bla = sizeof *p++;
and p is *not* incremented.


Segher