why gets "invalidate_cache" a const void ptr?

[Gedare Bloom]

On Fri, Apr 9, 2010 at 1:53 PM, Joel Sherrill <joel.sherrill at oarcorp.com> wrote:
> On 04/09/2010 12:43 PM, Gedare Bloom wrote:
>>
>> On Fri, Apr 9, 2010 at 1:05 PM, Thomas Dörfler
>> <Thomas.Doerfler at embedded-brains.de>  wrote:
>>
>>>
>>> Hi,
>>>
>>> I just tripped over a strange thing: the "*invalidate_*_cache" functions
>>> in cpukit/rtems/rtems/include/rtems/cache.h are declared to get a "const
>>> void *" pointer passed in. IMHO the "const" is wrong, because the
>>> function may modify the storage where the pointer points to
>>>
>>>
>>
>> Is there a case where this happens?
>>
>
> I don't think the cache method is allowed to modify the contents.
> It is simply flushing it from the CPU cache.  The underlying memory
> is unchanged.
>>
>>
>>>
>>> Background:
>>> 1.) when a cache line is invalidated, the real memory behind it may
>>> become visible and have a different content than available in the cache
>>> previously.
>>>
>>>
>>
>> Yes, that is probably why the line was invalidated.
>>
>>
>
> I suppose this is possible if the cache is inconsistent with memory.
> But wouldn't that make volatile more appropriate?
>
> I am having trouble seeing where this case would happen.  Why would
> you have a system which cached some memory knowing that the
> memory and cache would be inconsistent?
>
The sparc (and sparc64) port deals with this, when it registers a trap
(interrupt) handler, it replaces the entry in the trap table going
through the data cache, and needs to make sure to invalidate the
instruction cache, because the instructions for the old handler might
exist.

-Gedare

>>> 2.) Having that function to get a "const void *" allos GCC to use more
>>> optimizations, which may result in wrong code.
>>>
>>>
>>
>> Are there examples of gcc optimizations that act on pointer to const?
>> Having const enforces code paths in the cache_invalidate family to not
>> attempt to write-back data to the address, I think this is expected
>> behavior when someone invalidates a cache line.
>> -Gedare
>>
>>
>>>
>>> what do you all think?
>>>
>>> I would vote to remove the "const" keyword in the declaration and the
>>> implementations of the functions.
>>> wkr,
>>> Thomas.
>>>
>>> --
>>>
>>> --------------------------------------------
>>> Embedded Brains GmbH
>>> Thomas Doerfler        Obere Lagerstrasse 30
>>> D-82178 Puchheim       Germany
>>> email: Thomas.Doerfler at embedded-brains.de
>>> Phone: +49-89-18908079-2
>>> Fax:   +49-89-18908079-9
>>> _______________________________________________
>>> rtems-users mailing list
>>> rtems-users at rtems.org
>>> http://www.rtems.org/mailman/listinfo/rtems-users
>>>
>>>
>>
>> _______________________________________________
>> rtems-users mailing list
>> rtems-users at rtems.org
>> http://www.rtems.org/mailman/listinfo/rtems-users
>>
>
>
> --
> Joel Sherrill, Ph.D.             Director of Research&  Development
> joel.sherrill at OARcorp.com        On-Line Applications Research
> Ask me about RTEMS: a free RTOS  Huntsville AL 35805
>   Support Available             (256) 722-9985
>
>
>