Re: double underscores

[Seth Arnold <sarnold@wirex.com>]

On Tue, Jun 04, 2002 at 02:17:01PM -0500, Little, John wrote:
> why do some functions begin with double underscores?

Identifiers starting with an underscore are reserved for the C
implementation. Since the kernel doesn't use too much of the C
implementation (run time support.. libgcc, libcrt, perhaps), someone
decided that names beginning with an underscore was probably a safe
enough way to mark some functions as dangerous and for internal use
only.

You'll often see functions named like this:
 function ()
 __function ()

where the function() simply does some sanity checks on input before
calling __function (). It is done for speed reasons: if you've checked
that the inputs are sane and safe, you can call the __ version. If you
haven't, you should call the normal version, and let it do the checking.

As far as I know, this isn't a codified practice, so I wouldn't suggest
relying on it, unless you've looked at both versions, and are fairly
certain your checks are sufficient...

-- 
http://www.wirex.com/

PGP signature