[Xenomai-help] rt_heap_bind in kernel module

[Xenomai-help] rt_heap_bind in kernel module

[Stefan Eletzhofer]

[-- Attachment #1: Type: text/plain, Size: 919 bytes --]

Hi,
reading the API docs (2.1.0) of the xenomai native API I noticed that the 
rt_heap_bind() call is a userspace-only call and i'm wondering why that is
the case? Or am I missing something here?

I'm in the process of porting applications which use the RTAI shm module
in both user space and kernel space; these applications basically allocate 
the shared memory chunks in user space and then have kernel modules dump
data into these areas.

To port things, I've coded a rt_heap_bind() for kernel space modelled after 
the code in skins/native/syscalls.c using the nucleus registry API, and that 
seems to work. So, is this missing call just laziness, then I'd post a patch.

Thanks,
Stefan E.
-- 
Stefan Eletzhofer
InQuant GmbH
Bahnhofstraße 11
D-88214 Ravensburg
http://www.inquant.de
http://www.eletztrick.de
+49 (0) 751 35 44 112
+49 (0) 171 23 24 529 (Mobil)
+49 (0) 751 35 44 115 (FAX)

[-- Attachment #2: Type: application/pgp-signature, Size: 191 bytes --]

Re: [Xenomai-help] rt_heap_bind in kernel module

[Philippe Gerum]

Stefan Eletzhofer wrote:
> Hi,
> reading the API docs (2.1.0) of the xenomai native API I noticed that the 
> rt_heap_bind() call is a userspace-only call and i'm wondering why that is
> the case? Or am I missing something here?
> 

Binding op are not supported in kernel space, basically because one sees 
a single address space there, and it's always possible to share the 
original object descriptor directly between modules as a plain memory 
area. The synchronization the binding ops provide between the creator 
and the users of a given resource has been ignored for kernel space.

> I'm in the process of porting applications which use the RTAI shm module
> in both user space and kernel space; these applications basically allocate 
> the shared memory chunks in user space and then have kernel modules dump
> data into these areas.
> 
> To port things, I've coded a rt_heap_bind() for kernel space modelled after 
> the code in skins/native/syscalls.c using the nucleus registry API, and that 
> seems to work. So, is this missing call just laziness, then I'd post a patch.
> 
> Thanks,
> Stefan E.
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Xenomai-help mailing list
> Xenomai-help@domain.hid
> https://mail.gna.org/listinfo/xenomai-help


-- 

Philippe.