[PATCH 3/3] include: sbi: Lift the maximum harts to 1024

[Heinrich Schuchardt]

On 10/9/25 15:29, Anup Patel wrote:
> On Thu, Oct 9, 2025 at 5:55 PM Wu Fei <atwufei at 163.com> wrote:
>>
>>
>> On 9/30/25 19:36, Anup Patel wrote:
>>> On Wed, Aug 6, 2025 at 6:39 AM <wu.fei9 at sanechips.com.cn> wrote:
>>>> It is expected to have riscv server boards with many cores very soon.
>>>>
>>>> Signed-off-by: Wu Fei <wu.fei9 at sanechips.com.cn>
>>>> ---
>>>>    include/sbi/sbi_hartmask.h | 2 +-
>>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/include/sbi/sbi_hartmask.h b/include/sbi/sbi_hartmask.h
>>>> index 200ab6e..2f94f39 100644
>>>> --- a/include/sbi/sbi_hartmask.h
>>>> +++ b/include/sbi/sbi_hartmask.h
>>>> @@ -20,7 +20,7 @@
>>>>     * also represents the maximum number of HART ids generic OpenSBI
>>>>     * can handle.
>>>>     */
>>>> -#define SBI_HARTMASK_MAX_BITS          128
>>>> +#define SBI_HARTMASK_MAX_BITS          1024
>>> Instead of making this 8x, it is better to move to a more realistic
>>> number like 256 or 512. NO ?
>>
>>
>> Do you think we need to make this configurable? It sounds good to make
>> it larger if it's configurable,
>>
>> 512 is not enough sooner or later on servers.
> 
> I would rather wait for a real platform with > 128 CPUs to show-up. Since
> it is a one-line change and we can do it any day.
> 
> Regards,
> Anup

The Linux kernel has raised the default to 256 CPUs with a maximum 
configuration value of 512. See
0fa749771993 ("riscv: Raise default NR_CPUS for 64BIT to 256")

Shouldn't OpenSBI match it and have a configuration variable that 
defaults to 256?

QEMU is still limited to 512: CPUs

Best regards

Heinrich