[llvm] r193718 - Produce .weak_def_can_be_hidden for some linkonce_odr values

[John McCall]

On Feb 7, 2014, at 8:29 AM, Rafael Espíndola <rafael.espindola at gmail.com> wrote:
>> Hey, Rafael.  Sorry to dredge up old-ish history.  This is a great idea, but:
>> 
>>> +          if (!GlobalStatus::analyzeGlobal(GV, GS) && !GS.IsCompared)
>>> +            CanBeHidden = true;
>> 
>> This doesn’t seem to be a complete list of conditions where this is safe.  For a
>> mutable object, either a load or a store could require the object to be uniqued
>> across dynamic library boundaries to get the right semantics.
>> 
>> Actually, I’m not sure what uses a mutable linkonce_odr object could possibly
>> have that would allow it to be hidden.
> 
> Gosh. Thanks for catching this!
> 
>> Of course, for an immutable object (like a Function or a const GlobalVariable),
>> loads and stores don’t matter; and it’s always allowed if unnamed_addr is
>> present.
>> 
>> Simple clang test case:
>>  inline int foo() {
>>    static int CallCount = 0;
>>    return CallCount++;
>>  }
>> 
>>  int next() {
>>    return foo();
>>  }
>> 
>>        .section        __DATA,__datacoal_nt,coalesced
>>        .globl  __ZZ3foovE9CallCount    ## @_ZZ3foovE9CallCount
>>        .weak_def_can_be_hidden __ZZ3foovE9CallCount
>>        .align  2
>> __ZZ3foovE9CallCount:
>>        .long   0                       ## 0x0
> 
> Fixed in r200977. Sorry for the breakage.

No problem.  I actually only caught it by inspection — I was digging out a different problem exposed by this (but not your fault).

Thanks!

John.