Gilles Chanteperdrix wrote:
> Tom Schouten wrote:
>> On Sat, Feb 07, 2009 at 10:39:19AM +0100, Gilles Chanteperdrix wrote:
>>> Tom Schouten wrote:
>>>> Hello,
>>>>
>>>> I tried with a different kernel .config (right.config) which made the 
>>>> userspace posix shm problem go away.  The significant difference I can see 
>>>> is this:
>>>>
>>>> ## cat right.config |grep SHM
>>>> CONFIG_SHMEM=y
>>>> # CONFIG_TINY_SHMEM is not set
>>>> # CONFIG_XENO_OPT_POSIX_SHM is not set
>>> Ok. In this case, you use Linux' shared memories, not xenomai's.
>>>
>> Aha.  
>>
>> Does this have any impact on real-time behaviour after the memory is 
>> mapped?
> 
> No, provided that the memory is locked with mlockall, there is no fault
> upon access to the mapped memory.
> 
> As explained in the Kconfig help, the only use of Xenomai posix skin
> shared memory is to share memory between kernel-space and user-space.
> So, if you only plan to write user-space applications, you do not need
> posix skin shared memory support.
> 
> Nevertheless, the bug you found needs fixing...

Ok. This should be fixed by now.

http://svn.gna.org/viewcvs/xenomai?rev=4617&view=rev

-- 
                                            Gilles.

_______________________________________________
Xenomai-help mailing list
[email protected]
https://mail.gna.org/listinfo/xenomai-help

Reply via email to