Memory Model

Memory Model

[Joe Gottman]

    According to the news on the front page,  the C++11 memory model has 
been added to gcc.  But C++11 status page 
(http://gcc.gnu.org/projects/cxx0x.html) says that it hasn't.  Is this 
an oversight?


Joe Gottman

Re: Memory Model

[Andrew MacLeod]

On 03/10/2012 09:17 AM, Joe Gottman wrote:
>    According to the news on the front page,  the C++11 memory model 
> has been added to gcc.  But C++11 status page 
> (http://gcc.gnu.org/projects/cxx0x.html) says that it hasn't.  Is this 
> an oversight?
>

Atomic support with memory model parameters have been added, but 
complete support is not quite there yet.  A full optimization audit has 
not been performed, so it is possible to introduce data races which were 
not there before.  Bitfields operations do not yet conform either.

That said, any one who encounters cases where those memory model 
non-compliances occur should open a bugzilla so we can make sure we have 
them covered in 4.8.


Andrew

Re: Memory Model

[Joe Gottman]

On 3/12/2012 8:44 AM, Andrew MacLeod wrote:
> On 03/10/2012 09:17 AM, Joe Gottman wrote:
>>    According to the news on the front page,  the C++11 memory model 
>> has been added to gcc.  But C++11 status page 
>> (http://gcc.gnu.org/projects/cxx0x.html) says that it hasn't.  Is 
>> this an oversight?
>>
>
> Atomic support with memory model parameters have been added, but 
> complete support is not quite there yet.  A full optimization audit 
> has not been performed, so it is possible to introduce data races 
> which were not there before.  Bitfields operations do not yet conform 
> either.
>
> That said, any one who encounters cases where those memory model 
> non-compliances occur should open a bugzilla so we can make sure we 
> have them covered in 4.8.
>
    OK.  Does it say anywhere on the website exactly what can and can't 
be done with respect to the memory model?

Joe Gottman

Re: Memory Model

[Lawrence Crowl]

On 3/12/12, Joe Gottman <josephgottman@comcast.net> wrote:
> On 3/12/2012 8:44 AM, Andrew MacLeod wrote:
> > On 03/10/2012 09:17 AM, Joe Gottman wrote:
> > > According to the news on the front page, the C++11
> > > memory model has been added to gcc.  But C++11 status page
> > > (http://gcc.gnu.org/projects/cxx0x.html) says that it hasn't.
> > > Is this an oversight?
> >
> > Atomic support with memory model parameters have been added, but
> > complete support is not quite there yet.  A full optimization
> > audit has not been performed, so it is possible to introduce
> > data races which were not there before.  Bitfields operations
> > do not yet conform either.
> >
> > That said, any one who encounters cases where those memory
> > model non-compliances occur should open a bugzilla so we can
> > make sure we have them covered in 4.8.
>
> OK.  Does it say anywhere on the website exactly what can and
> can't be done with respect to the memory model?

We cannot really know until after the optimization audit has been
completed.

It might be helpful to say the model is correct at some maximum
optimization level, probably -O0 or -O1.

-- 
Lawrence Crowl

Re: Memory Model

[Andrew MacLeod]

gcc.gnu.org bounced the previous email for some reason...

On 03/12/2012 06:51 PM, Joe Gottman wrote:
> On 3/12/2012 8:44 AM, Andrew MacLeod wrote:
>> On 03/10/2012 09:17 AM, Joe Gottman wrote:
>>>    According to the news on the front page,  the C++11 memory model 
>>> has been added to gcc.  But C++11 status page 
>>> (http://gcc.gnu.org/projects/cxx0x.html) says that it hasn't.  Is 
>>> this an oversight?
>>>
>>
>> Atomic support with memory model parameters have been added, but 
>> complete support is not quite there yet.  A full optimization audit 
>> has not been performed, so it is possible to introduce data races 
>> which were not there before.  Bitfields operations do not yet conform 
>> either.
>>
>> That said, any one who encounters cases where those memory model 
>> non-compliances occur should open a bugzilla so we can make sure we 
>> have them covered in 4.8.
>>
>    OK.  Does it say anywhere on the website exactly what can and can't 
> be done with respect to the memory model?
>

Well, perhaps not in so many words.   The information is all there, but 
not in a  nice summary format.  I have updated the atomics wiki (pointed 
to from the News page) to have a GCC 4.7 status which should spell out 
what you can expect.
http://gcc.gnu.org/wiki/Atomic/GCCMM


Does that help?

Andrew