Re: [PATCH RFC] RISC-V: alter the special character used in FAKE_LABEL_NAME

[Nick Clifton <nickc@redhat.com>]

Hi Jan,

> What I don't think you addressed is my 2nd observation wrt quoted symbol
> names. There all characters can appear in symbol names, and hence
> deriving "local" (or not) based on symbol name looks wrong to me.

Ok, that is fair.  Let's see.  So the issue is, what should S_IS_LOCAL do
if a user provides a global, quoted symbol name that includes characters
that match DOLLAR_LABEL_CHAR, LOCAL_LABEL_CHAR or FAKE_LABEL_CHAR, right ?

I would imagine that this is an extremely unusual case, which is why it has
not come up before.

First off, I wonder why S_IS_LOCAL does not check to see if BSF_GLOBAL is
set in flags and return false if that is the case.  If it did, I think that
that would solve the problem.  I cannot think of any situation where you
could have a global symbol that needs to be treated as local for the purposes
of S_IS_LOCAL.

Alternatively, maybe it would be simpler to just document that quoted
symbol names that contain special characters might be treated as if they
were local symbols, and leave it up to the user to decide whether they
want to risk such behaviour.  Not very user friendly, but it would match
the current behaviour.  We could also add a warning to read_symbol_name()
and get_symbol_name() to alert users to the problem.

What do you think ?  Any preferences ?

Cheers
   Nick