This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [libiberty patch]: Change strncmp test in the way it uses mmap

At 2:46 PM -0500 12/31/01, DJ Delorie wrote:

> > What's wrong with the change I did?  Yes, both kinds of mmap's could
>> be done.  But what's wrong with doing the anonymous mmap if that is
>> available and the "/dev/zero" file mmap if not.  Why do both?
>
>Well, what's wrong with doing the /dev/zero file mmap is available,
>and anonymous if not?

I don't see why you seem so "protective" of doing that /dev/zero form when the other works equally as well (and in our case is necessary) and when it can be determined which to use based on the definitions of MAP_ANON and MAP_ANONYMOUS. 
I changed the original macro sequence,

  #ifndef MAP_ANON
  #ifdef MAP_ANONYMOUS
  #define MAP_ANON MAP_ANONYMOUS
  #else
  #define MAP_ANON MAP_FILE
  #endif
  #endif

to,

  #if defined(MAP_ANON) && !defined(MAP_ANONYMOUS)
  #define MAP_ANONYMOUS MAP_ANON
  #endif

And then MAP_ANONYMOUS controls which mmap to do.  I still see no reason to do both.

Ira
Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]