[OpenBIOS] [commit] r1042 - trunk/openbios-devel/arch/ppc/qemu

[Alexander Graf]

On 17.06.2011, at 21:55, Andreas Färber wrote:

> Am 17.06.2011 um 02:55 schrieb repository service:
> 
>> Author: agraf
>> Date: Fri Jun 17 02:55:00 2011
>> New Revision: 1042
>> URL: http://tracker.coreboot.org/trac/openbios/changeset/1042
>> 
>> Log:
>> ppc: add POWER4 cpu description
>> 
>> So far, we only supported the 970fx CPU type inside guests. Unfortunately, that
>> CPU supports Altivec. Not all KVM capable host CPUs do have altivec though, so
>> they're left out in the rain when running openbios in the guest.
>> 
>> This patch adds POWER4 CPU description, allowing users to specify at least one
>> CPU type that does not require Altivec, but is still 64-bit capable.
>> 
>> Signed-off-by: Alexander Graf <agraf at suse.de>
>> 
>> Modified:
>>  trunk/openbios-devel/arch/ppc/qemu/init.c
>> 
>> Modified: trunk/openbios-devel/arch/ppc/qemu/init.c
>> ==============================================================================
>> --- trunk/openbios-devel/arch/ppc/qemu/init.c	Mon May 23 00:29:33 2011	(r1041)
>> +++ trunk/openbios-devel/arch/ppc/qemu/init.c	Fri Jun 17 02:55:00 2011	(r1042)
>> @@ -507,6 +507,18 @@
>>        .clock_frequency = 0x1dcd6500,
>>        .initfn = cpu_970_init,
>>    },
>> +    {
>> +        .iu_version = 0x00350000,
>> +        .name = "PowerPC,POWER4",
>> +        .icache_size = 0x10000,
>> +        .dcache_size = 0x8000,
>> +        .icache_sets = 0x100,
>> +        .dcache_sets = 0x40,
>> +        .icache_block_size = 0x80,
>> +        .dcache_block_size = 0x80,
>> +        .clock_frequency = 0x629b4940,
>> +        .initfn = cpu_970_init,
>> +    },
>> };
>> 
>> static const struct cpudef *
> 
> Looks okay.
> 
> Have you investigated passing this from QEMU via libfdt instead? POWER5, POWER7 and lots of others selectable in QEMU are missing, too.

Well, in this case fw_cfg with the fields for the CPU should be enough and wouldn't require us messing with two different dt frameworks. But I agree - it's probably a good idea to pass all of this from Qemu at the end of the day.


Alex