On Mon, Dec 14, 2009 at 1:06 AM, Joel Fernandes <agnel.joel@xxxxxxxxx> wrote:
Hi Mulyadi,
Thanks for your response. But, DMA doesn't happen to the stack so its
ok if the (2-page)stack is not physically contiguous as long as it is
virtually contiguous?
I'm talking in terms of the processor architecture where all memory
accesses are infact virtual when paging
is turned on, so should accesses to the stack be?
I think the answer to my question might be kernel memory is directly
mapped to physical memory (LOW MEM), and so anything contiguous in the
kernel's virtual address space is also physically contiguous. Yes?
Thanks,
-Joel
Hi Joe/Mulyadi,
The reason I guess is:
The RAM is (low mem , up to 896 MB) is directly mapped to kernel virtual address space, starting from 0xC0000000. Any RAM (again low mem) physically contiguous is virtually contiguous too. So when accessing those pages, it does not require to go through the paging unit. To findout the physical adddress to operate , it is just a simple math of 'virtual address - 0xc00000000' will give physical address. Similarly we can findout VA, if we now PA. It will not go through paging unit. So while serving ISR, it does not have to go through paging unit etc while accessing the stack and that might be a reason that kernel stack should be two contiguous physical pages and in turn virtual pages.
Out of 1 GB kernel virtual address space, RAM is the biggest consumer of this space.
Thanks,
Gopal.
On Sun, Dec 13, 2009 at 11:52 PM, Mulyadi Santosa
<mulyadi.santosa@xxxxxxxxx> wrote:
> Sorry, pressing "send" button to send.
>
> Imagine if data copied span between those pages, then data on the
> second page will be left out because DMA operation thinks in physical
> address while the second page might not be physically adjacent.
>
> --
> regards,
>
> Mulyadi Santosa
> Freelance Linux trainer and consultant
>
> blog: the-hydra.blogspot.com
> training: mulyaditraining.blogspot.com
>
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
