[llvm-commits] [llvm-gcc-4.2] r53130 - /llvm-gcc-4.2/trunk/gcc/cgraphunit.c

[Dale Johannesen]

On Jul 4, 2008, at 9:38 AMPDT, Duncan Sands wrote:

> Sorry, problems with my email client :)  Here it is
> again, hopefully readable...
>
> Hi Dale,
>
>> Do not emit unreferenced static functions at -O0.
>> This is the gcc-4.0 behavior, changed deliberately
>> in gcc-4.2, see PR 24561.  After discussion Chris,
>> Evan and I agreed the 4.0 behavior is better so
>> we're going back to that.
>
> is there a similar issue with dead static globals?

The behavior is that dead static globals are removed in C++ at all  
optimization levels; this is the same across gcc-4.0, gcc-4.2, and  
llvm-gcc.
In C they are not removed at -O0; gcc-4.2 and llvm-gcc remove them at  
higher optimization levels, and gcc-4.0 does not.

My guess is that, since the processing involved for data objects is  
considerably less than that for functions, whatever you do here is not  
going to matter much to compile time.  That might not be true for big  
initialized static objects, especially given llvm IR's bulky notation  
for them, but how common can that be (unreferenced)?  I have no data  
to support my opinion, however.

> Also, how much of a win is this?

I don't know.  It showed up on a profile as a place where time was  
being spent, maybe the guys who made the profile have some idea.