Re: [PATCH v5 04/15] x86/mtrr: support setting MTRR state for software defined MTRRs

[Juergen Gross <jgross@suse.com>]

[-- Attachment #1.1.1: Type: text/plain, Size: 1409 bytes --]

On 03.04.23 11:27, Huang, Kai wrote:
> 
>>>>    /**
>>>>     * mtrr_type_lookup - look up memory type in MTRR
>>>>     *
>>>> diff --git a/arch/x86/kernel/cpu/mtrr/mtrr.c b/arch/x86/kernel/cpu/mtrr/mtrr.c
>>>> index 1beb38f7a7a3..1c19d67ddab3 100644
>>>> --- a/arch/x86/kernel/cpu/mtrr/mtrr.c
>>>> +++ b/arch/x86/kernel/cpu/mtrr/mtrr.c
>>>> @@ -666,6 +666,15 @@ void __init mtrr_bp_init(void)
>>>>    	const char *why = "(not available)";
>>>>    	unsigned int phys_addr;
>>>>    
>>>> +	if (!generic_mtrrs && mtrr_state.enabled) {
>>>> +		/* Software overwrite of MTRR state, only for generic case. */
>>> 							      ^
>>> 							      !generic case?
>>
>> No. This test just verifies that the (visible) MTRR feature is switched off,
>> as there are no ways to modify any MTRR registers in the overwrite case.
>>
>> I can make this more obvious in a comment.
> 
> Should the comment say something like (because it applies to the code inside the
> check):
> 
> 
> 	If we have a static (synthetic) MTRR already established for special
> 	VMs, we still need to calculate the physical address bits using
> generic
> 	way, because the hardware to run those special VMs indeed has MTRR.
> 
> That explains why 'true' is passed to mtrr_calc_physbits().

I'd rather say that the interface of mtrr_overwrite_state() is based on the
interface of generic MTRRs.


Juergen

[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 3149 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]