[okl4-developer] Getting initial virtual and physical memory information within a cell

[Geoffrey Lee]

On Wed, Nov 19, 2008 at 12:13:30PM +0900, Alexandre Courbot wrote:
> Hello everybody,
> 
> These questions may sound like beginner's questions - well, that's
> because actually they are. Sorry if I am asking the obvious here.
> 
> I'm playing around with OKL4 3.0 and am trying to set up a cell that
> is able to create new protection domains and start threads within
> them. As far as I have seen, no example demonstrates such a feature.
> I'm trying to use the OKL4 library as much as possible, and as
> recommanded on the highest level available. The initial things my cell
> should therefore do are getting the initial virtual memory pool and
> physical segment pool as stated.

Alexandre - 

There are some examples in the source tarball, look in
okl4_3.0/examples/.


> 
> This is where I'm blocked, basically - I understand that I can get a
> list of available physical segments using okl4_env_segments and
> parsing the result, but this okl4_env_segment only provides very
> limited information about the physical memory. I cannot even know the
> start address or name under which it is defined by Elfweaver. In
> addition, I failed to find any interface to get virtual memory
> information. I feel like I'm missing a big point here, so any pointer
> would be greatly appreciated.

Normally, the physical memory information is not exported.  Mapping is 
done via giving a memory segment identifier and an offset into the segment.   
At no point does physical address need to be exported.  This is a 
little different where typically you would insert a physical address 
into the pagetable for mapping.  The exception is DMA, both use
cases can be found in the OKL4 3.0 release of OK Linux.


> 
> Also, how can I know which resource names are defined for a given
> cell? In the pingpong exemple, I have seen that the main kspace can be
> obtained through the MAIN_KSPACE name, but nothing appears about this
> entry name in the documentation.
> 

Yes, that seems to be missing from documentation.


> Thanks for any help,
> Alex.

	-gl

> 
> _______________________________________________
> Developer mailing list
> Developer at okl4.org
> https://lists.okl4.org/mailman/listinfo/developer
> 

--