[PATCH 1/6] dlls/msvcrt*: ensure variable sse2_cw is set for all code paths in _control87 (GCC11)

[Zebediah Figura]

On 9/30/21 12:14 PM, Eric Pouech wrote:
> 
>>>
>>> which still gives me the warnings, when compiling the 32bit part of a
>>> wow64 conf
>>>
>>> (but not on a pure 32bit conf)
>>>
>>> your solution doesn't generate warnings on neither of the two
>>>
>>> so will need further investigation on:
>>>
>>> - discrepency wrt wow
>>>
>>> - why the difference between the two patches
>>
>> I don't get warnings with that diff either. What version of GCC do you
>> have?
> 
> well, o well....
> 
> retesting today, and couldn't get any warnings with the patch I posted
> yesterday...<grrr>
> 
> 
> moreover, in this patch, if I change the if (_control... by if
> (!control... (inverting the test), I get no warnings either which makes
> absolutely no sense
> 
> 
> I'm starting to wonder if the  -Wmaybe-uninitialized warning could be a
> little fragile at this point
> 
> 
> there are a couple of bugs reports on gcc bugzilla reguarding this warning
> 
> potentially bugs report like may link to the same type of issues
> 
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102381
> 
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101768
> 
> 
> I may wait a bit to let things settle down before taking any action on
> this fix in msvcrt

I mean, -Wmaybe-uninitialized has always been at least somewhat 
heuristic, throughout the entire history of gcc, and there have been 
false positives and false negatives. But I've found that it tends to 
behave nicely for idiomatically structured code. I've found very few 
cases where "just initialize the variable to zero" seems like the best 
solution to me.