<div dir='auto'>11.2.1</div><div class="gmail_extra"><br><div class="gmail_quote">Le 29 sept. 2021 20:10, "Zebediah Figura (she/her)" <zfigura@codeweavers.com> a écrit :<br type="attribution" /><blockquote class="quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><p dir="ltr">On 9/29/21 12:59, Eric Pouech wrote:
<br>
> Le 29/09/2021 à 17:57, Zebediah Figura (she/her) a écrit :
<br>
>> On 9/29/21 02:43, Eric Pouech wrote:
<br>
>>> Le 28/09/2021 à 20:01, Zebediah Figura (she/her) a écrit :
<br>
>>>> On 9/28/21 11:49, Eric Pouech wrote:
<br>
>>>>> Signed-off-by: Eric Pouech <eric.pouech@gmail.com>
<br>
>>>>>
<br>
>>>>> ---
<br>
>>>>> dlls/msvcrt/math.c | 2 +-
<br>
>>>>> 1 file changed, 1 insertion(+), 1 deletion(-)
<br>
>>>>>
<br>
>>>>> diff --git a/dlls/msvcrt/math.c b/dlls/msvcrt/math.c
<br>
>>>>> index 7f59a4d20d4..ad632e70548 100644
<br>
>>>>> --- a/dlls/msvcrt/math.c
<br>
>>>>> +++ b/dlls/msvcrt/math.c
<br>
>>>>> @@ -5643,7 +5643,7 @@ unsigned int CDECL _control87(unsigned int
<br>
>>>>> newval, unsigned int mask)
<br>
>>>>> {
<br>
>>>>> unsigned int flags = 0;
<br>
>>>>> #ifdef __i386__
<br>
>>>>> - unsigned int sse2_cw;
<br>
>>>>> + unsigned int sse2_cw = 0;
<br>
>>>>> __control87_2( newval, mask, &flags, &sse2_cw );
<br>
>>>>>
<br>
>>>>>
<br>
>>>>
<br>
>>>> Wouldn't it be better to check for failure from __control87_2()?
<br>
>>>>
<br>
>>> unfortunately, gcc11 still complains when checking for failure of
<br>
>>> _control87_2()
<br>
>>>
<br>
>>> gcc doesn't seem to be smart enough to infer that ss2_cw is always
<br>
>>> when _control87_2() returns 1
<br>
>>
<br>
>> That doesn't match what I have here. With the attached patch gcc 11.1
<br>
>> doesn't complain.
<br>
>
<br>
> what I tried is:
<br>
>
<br>
> diff --git a/dlls/msvcrt/math.c b/dlls/msvcrt/math.c
<br>
> index 7f59a4d20d4..4560040eb9f 100644
<br>
> --- a/dlls/msvcrt/math.c
<br>
> +++ b/dlls/msvcrt/math.c
<br>
> @@ -5645,10 +5645,11 @@ unsigned int CDECL _control87(unsigned int
<br>
> newval, unsigned int mask)
<br>
> #ifdef __i386__
<br>
> unsigned int sse2_cw;
<br>
>
<br>
> - __control87_2( newval, mask, &flags, &sse2_cw );
<br>
> -
<br>
> - if ((flags ^ sse2_cw) & (_MCW_EM | _MCW_RC)) flags |= _EM_AMBIGUOUS;
<br>
> - flags |= sse2_cw;
<br>
> + if (__control87_2( newval, mask, &flags, &sse2_cw ))
<br>
> + {
<br>
> + if ((flags ^ sse2_cw) & (_MCW_EM | _MCW_RC)) flags |=
<br>
> _EM_AMBIGUOUS;
<br>
> + flags |= sse2_cw;
<br>
> + }
<br>
> #else
<br>
> flags = newval;
<br>
> _setfp(&flags, mask, NULL, 0);
<br>
>
<br>
>
<br>
> which still gives me the warnings, when compiling the 32bit part of a
<br>
> wow64 conf
<br>
>
<br>
> (but not on a pure 32bit conf)
<br>
>
<br>
> your solution doesn't generate warnings on neither of the two
<br>
>
<br>
> so will need further investigation on:
<br>
>
<br>
> - discrepency wrt wow
<br>
>
<br>
> - why the difference between the two patches
<br>
<br>
I don't get warnings with that diff either. What version of GCC do you have?
<br>
</p>
</blockquote></div><br></div>