[PATCH v4, middle end]: Fix PR 37908, thinko with atomic NAND operation

[Uros Bizjak]

Mark Mitchell wrote:

>> No, as explained in my previous message, it only matched wrong example.
>>     
>>> +@emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch}
>>>
>>> __sync_fetch_and_nand, right?
>>>   
>>>       
>> We have two types of sync builtins - __sync_op_and_fetch and
>> __sync_fetch_and_op. This one is correct.
>>
>> Attached patch revision implements suggested once-per-file warning about
>> changed NAND builtin semantics.
>>     
>
> The patch wasn't attached.
>
> However, given the explanations that have been provided, I think this is
> the right thing to do, so I retract my original concern about changing
> the meaning of the existing builtins.  If the new note provides a hint
> about how to disable the note, then the patch is OK.
>   
   
Uh, my apologies... The patch is now attached.

However, I'm not sure I understood the reason to disable the note. The 
note can be IMO considered as "less than a warning", just a heads-up 
that something has been changed that requires a bit of attention.


Thanks,
Uros.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: p.diff.txt
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20081116/bd93e18f/attachment.txt>