[LinuxBIOS] [PATCH v3] v2, v3: Bug in CAR setup with CacheSize!={4k, 8k, 16k}

[Carl-Daniel Hailfinger]

On 30.11.2007 18:36, Marc Jones wrote:
>
>
> Marc Jones wrote:
>>
>> yhlu wrote:
>>> my own tree have
>>>
>>> #if CacheSize < 0x8000
>>>         /* enable caching for 16K/8K/4K using fixed mtrr */
>>>         movl    $0x269, %ecx  /* fix4k_cc000*/
>>>     #if CacheSize == 0x4000
>>>         movl    $0x06060606, %edx /* WB IO type */
>>>     #endif
>>>     #if CacheSize == 0x2000
>>>         movl    $0x06060000, %edx /* WB IO type */
>>>     #endif
>>>     #if CacheSize == 0x1000
>>>         movl    $0x06000000, %edx /* WB IO type */
>>>     #endif
>>>         xorl    %eax, %eax
>>>         wrmsr
>>> #endif
>>>
>>> maybe i didn't post sth...
>>>
>> This is in the v2 code but not in v3.
>>
>>
>>> normally we can use 4k only at first.
>>>
>>> run_time decide is not good. for example REV E, you could use 64k. for
>>> for REV F you can only use 48K.
>>>
>> So this is still a problem since 48K isn't an option and it isn't a
>> power of 2 so I was incorrect on that point.
>>
>> It would be good to have an automatic runtime setting for each CPU
>> that could be adjusted by a mainboard override.
>>
>>
>>> Family 10h need more tricks.
>> These changes are coming soon.
>>
>> Marc
>>
>>
>
> Also, 48K is an option in v2. It seems that v3 should be brought up to
> date with v2 first.

48K is not an option for the v2 generic x86 code, only for the v2 K8
code (which is arguably less buggy, but still fails for some
non-power-of-2 settings).

Regards,
Carl-Daniel