+#define external_array(type, name) \
+ ({ \
+ extern type name[]; \
+ type *name_ptr = name; \
+ asm ("" : "+r" (name_ptr)); \
+ name_ptr; \
+ })
I've had to pull similar tricks to persuade GCC to generate
the code I wanted (in my case, it was optimization: "evaluate it
in this order, damn it!"), and I prefer to use the asm itself
with overlapping operands to do the assignment.
#define external_array(type, name) \
({ \
extern type name[]; \
type *name_ptr;
asm ("" : "=g" (name_ptr) : "0" (name)); \
name_ptr; \
})
You could define a wrapper if you like, something like
/*
* Assign dst = src, but prevent the compiler from inferring anything
* about the assigned value, so it can't do any unwanted optimization.
*/
#define blind_assign(dst,src) asm("" : "=X" (dst) : "0" (src))
In case it helps, here's a list of architecture-independent operand
constraints (that I think is exhaustive, but I'm not 100% sure):
r - general-purpose register
f - floating-point register
m - memory
o - offsettable memory (excluding pre/post inc/decrement modes)
i - immediate
g - "general", any of the above
X - "wildcard". This matches anything at all, including special-purpose
registers that are not included in "r".
