Re: [LinuxBIOS] serengeti support added

[Lu, Yinghai]

Great,

Next you need to work on 36 and 37...

YH

________________________________

From: [e-mail]
[mailto:[e-mail]] On Behalf Of ron minnich
Sent: Wednesday, November 23, 2005 6:47 PM
To: LinuxBIOS; yhlu; Stefan Reinauer; Li-Ta Lo
Subject: [LinuxBIOS] serengeti support added

Ying Hai, serengeti support is in. This is one part of issue 24.

ron

[LinuxBIOS] serengeti support added

[ron minnich]

On 11/23/05, Lu, Yinghai  wrote:
> Great,
>
> Next you need to work on 36 and 37…

we can not do that until we get the a_s1_core.diff to patch correctly,
right? I'm not totally done 24.

We're going to stop here and make sure the 2881 will work correctly with
current svn. We'll do these other patches next week.

ron

[LinuxBIOS] serengeti support added

[yhlu]

without 37 you may have problem for s2881 to work with kernel.....

Kernel need lifted apic are continous.

But LNXI apicid solution will default lift all apic as
A:
Node0/Core0: 0x10
Node0/Core1: 0x20
Node1/Core0: 0x11
Node1/Core1: 0x21

You need to make northbridge.c/amd_siblings.c to be consistent with init_cpus.c

current init_cpus.c is using
B:
ENABLE_APIC_EXT_ID =0
Node0/Core0: 0x00
Node0/Core1: 0x01
Node1/Core0: 0x02
Node1/Core1: 0x03
or
ENABLE_APIC_EXT_ID=1, APIC_ID_OFFSET=0x10, LIFT_BSP_APIC_ID=0
C:
Node0/Core0: 0x00
Node0/Core1: 0x11
Node1/Core0: 0x12
Node1/Core1: 0x13
or
D:
ENABLE_APIC_EXT_ID=1, APIC_ID_OFFSET=0x10, LIFT_BSP_APIC_ID=1
Node0/Core0: 0x10
Node0/Core1: 0x11
Node1/Core0: 0x12
Node1/Core1: 0x13

Kernel will work with B and D.

With AMD8111 We need C, and Kernel patch....

YH

On 11/23/05, ron minnich  wrote:
> On 11/23/05, Lu, Yinghai  wrote:
> > Great,
> >
> > Next you need to work on 36 and 37…
>
> we can not do that until we get the a_s1_core.diff to patch correctly,
> right? I'm not totally done 24.
>
> We're going to stop here and make sure the 2881 will work correctly with
> current svn. We'll do these other patches next week.
>
> ron

[LinuxBIOS] serengeti support added

[Li-Ta Lo]

On Wed, 2005-11-23 at 21:31 -0700, ron minnich wrote:
> On 11/23/05, Lu, Yinghai  wrote:
>         Great,
>     Next you need to work on 36 and 37…
>
>
>
> we can not do that until we get the a_s1_core.diff to patch correctly,
> right? I'm not totally done 24.

Do you mean the CAR enable patch depends on the ACPI patch? Why?
> We're going to stop here and make sure the 2881 will work correctly
> with current svn. We'll do these other patches next week.

If CAR enable patch does not depend on the ACPI patch, we should be
able to proceed. Revision 2098 has been tested on the real s2881 and
I don't think revision 2099 and 2100 will do anything the other part
of the tree.

[LinuxBIOS] serengeti support added

[yhlu]

Don't need acpi, but all related to northridge.c.

ACPI need to store node id and core id in apic id for SRAT and MADT in
bus_scan_bus....

with apic lifting, s2881 can work?

Can you check the apic id in LinuxBIOS and kernel?

YH

On 11/23/05, Li-Ta Lo  wrote:
> On Wed, 2005-11-23 at 21:31 -0700, ron minnich wrote:
> > On 11/23/05, Lu, Yinghai  wrote:
> >         Great,
> >     Next you need to work on 36 and 37…
> >
> >
> >
> > we can not do that until we get the a_s1_core.diff to patch correctly,
> > right? I'm not totally done 24.
>
> Do you mean the CAR enable patch depends on the ACPI patch? Why?
> > We're going to stop here and make sure the 2881 will work correctly
> > with current svn. We'll do these other patches next week.
>
> If CAR enable patch does not depend on the ACPI patch, we should be
> able to proceed. Revision 2098 has been tested on the real s2881 and
> I don't think revision 2099 and 2100 will do anything the other part
> of the tree.