This is the mail archive of the gcc@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]

Re: RMS's aliasing suggestion appears unimplementable


    Catching simple reg copy propagations adds complexity, but can also be done
    as a variant of the existing code to propagate base values (ie, only propagate
    simple sets, but not any arithmetic).

Right.  The same code that keeps track of this for the sake
of making sure two values do not alias, will also do this subtask
for ascertaining that they do alias.


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