[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc-patches
Subject:    Re: [RFC] [PATCH] Implement -ffortify for C/C++
From:       Ulrich Drepper <drepper () redhat ! com>
Date:       2007-02-21 17:35:49
Message-ID: 45DC82F5.40203 () redhat ! com
[Download RAW message or body]


Dirk Mueller wrote:
>> - but once you have such support in the compiler there is no reason to
>>   add any bloat to gcc by making the built-ins more complicated,
> 
> They're not more complicated. 

How can you say that?  Of course it's more complicated in gcc.


> Unless I missed something, gcc still has to understand what an operator new 
> is, which is necessary for object sizes being properly propagated. This is an 
> essential part of the patch series. 

That's unrelated to introducing the changes to the built-ins.  Yes,
__builtin_object_size should track new calls just like malloc calls for
better coverage.  I would have imagined that the two types are handled
already but if new isn't, just add it.  No problem with this.

But changing the built-ins is wrong.

-- 
➧ Ulrich Drepper ➧ Red Hat, Inc. ➧ 444 Castro St ➧ Mountain View, CA ❖


["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]