[PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Huang, Ying]

This patchset adds the linked list of struct setup_data to boot
protocol. This is needed by EDD information and some SGI machine with
E820 entry number > 128.

Now, only the x86_64 is supported. The i386 can be supported after early
reservation mechanism for x86 is available.

This patchset is based x86.git latest branch and has been tested on
Intel x86_64 machine.

v2:

- Rebased on x86.git latest branch


Best Regards,
Huang Ying

Re: [PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Yinghai Lu]

On Thu, Mar 27, 2008 at 7:49 PM, Huang, Ying <ying.huang@intel.com> wrote:
> This patchset adds the linked list of struct setup_data to boot
>  protocol. This is needed by EDD information and some SGI machine with
>  E820 entry number > 128.

interesting, that 512nodes/4096 box physics memory is not continuous?

YH

Re: [PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Andi Kleen]

On Thu, Mar 27, 2008 at 10:33:27PM -0700, Yinghai Lu wrote:
> On Thu, Mar 27, 2008 at 7:49 PM, Huang, Ying <ying.huang@intel.com> wrote:
> > This patchset adds the linked list of struct setup_data to boot
> >  protocol. This is needed by EDD information and some SGI machine with
> >  E820 entry number > 128.
> 
> interesting, that 512nodes/4096 box physics memory is not continuous?

Likely not. That would require huge hardware overhead for all the
comparators to handle variable ranges.  Much easier to just use some higher bits 
in the address to index nodes.

-Andi

Re: [PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Paul Jackson]

YH wrote:
> interesting, that 512nodes/4096 box physics memory is not continuous?

If each node board is stuffed with the same amount of RAM,
and that amount of RAM is a power of two size, then you
are correct that it would be continuous.

But if the nodes are stuffed with unequal amounts of RAM,
or if its not a power of two, then, because we are doing
what Andi figures (breaking the addr into bit fields, for
efficient implementation) then we can end up with many holes.

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.940.382.4214

Re: [PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Andi Kleen]

On Fri, Mar 28, 2008 at 08:36:13AM -0500, Paul Jackson wrote:
> YH wrote:
> > interesting, that 512nodes/4096 box physics memory is not continuous?
> 
> If each node board is stuffed with the same amount of RAM,
> and that amount of RAM is a power of two size, then you

There are likely per node memory holes too, right?

So even in that scenario it would be not continuous.

-Andi

Re: [PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Paul Jackson]

Andi wrote:
> There are likely per node memory holes too, right?

Likely ... I don't have the details offhand.

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.940.382.4214

Re: [PATCH 0/4] x86_64 boot -v2: Add linked listof struct setup_data to boot protocol

[Ingo Molnar]

* Huang, Ying <ying.huang@intel.com> wrote:

> This patchset adds the linked list of struct setup_data to boot 
> protocol. This is needed by EDD information and some SGI machine with 
> E820 entry number > 128.
> 
> Now, only the x86_64 is supported. The i386 can be supported after 
> early reservation mechanism for x86 is available.
> 
> This patchset is based x86.git latest branch and has been tested on 
> Intel x86_64 machine.

thanks, applied. Note: as i mentioned, this code needs to be unified, 
and we can only merge this upstream into 2.6.26 once 32-bit support has 
been added too. The last thing we need is 64-bit only boot protocol 
extensions.

	Ingo