GCC: "__unused__" vs just "unused" in variable attributes

Question

According to GCC's own documentation on variable attributes, the correct syntax for declaring an attribute unused is __attribute__((unused)).

However, in many examples and other code online, I frequently see __attribute__((__unused__)) instead, and they appear to both work.

Is there a reason for either specifying or omitting the __ in either case? Does it make any difference, and is there a preferred version? Are there any situations where using one and not the other might cause problems?

Presumably the same applies to other attribute parameters as well?

Answer 1

At the top of the very page you linked, it tells you:

You may also specify attributes with ‘__’ preceding and following each keyword. This allows you to use them in header files without being concerned about a possible macro of the same name. For example, you may use __aligned__ instead of aligned.

Identifiers containing double underscores (__) are reserved to the implementation. Hence no user program could legally define them as macros.