This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: g++ memory bug.
- To: mrs at wrs dot com (Mike Stump)
- Subject: Re: g++ memory bug.
- From: Theodore Papadopoulo <Theodore dot Papadopoulo at sophia dot inria dot fr>
- Date: Fri, 08 Jan 1999 20:26:48 +0100
- cc: Theodore dot Papadopoulo at sophia dot inria dot fr, mark at markmitchell dot com, Jose dot Gomes at sophia dot inria dot fr, egcs-bugs at cygnus dot com, egcs-patches at cygnus dot com
mrs@wrs.com said:
> Nope, not removed yet (if you refer to the members). In general, only
> through link time feedback can we do this type of transformation, as
> another file or library may in fact do something with them. If you
> mean the one in main, then yes, it should wind up going.
I was referring to the stack (in main and the created functions).
This bring one corrolary: can a function be reconsidered for inlining
after having went throught the optimizer. If so, with your patch,
after a few cycles, the optimal code (with no function call at all
would be generated). The question is still, what is the cost and
is it worth doing so...
>> In this regard, egcs does much better since when it is able to do the
>> full inlining as it suppresses all the calls....
>
> It is nice to see that we do better than Kai...
Further checking shows that Kai does the inlining when the constructor of
the X class is the default constructor generated by the compiler and
does the behaviour I described when this constructor is implemented
by the user.
--------------------------------------------------------------------
Theodore Papadopoulo
Email: Theodore.Papadopoulo@sophia.inria.fr Tel: (33) 04 92 38 76 01
--------------------------------------------------------------------