Re: [Xen-devel] [PATCH v7 2/9] mm: Place unscrubbed pages at the end of pagelist

[Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>]

On 08/14/2017 06:37 AM, Jan Beulich wrote:
>>>> On 08.08.17 at 23:45, <boris.ostrovsky@xxxxxxxxxx> wrote:
>> .. so that it's easy to find pages that need to be scrubbed (those pages are
>> now marked with _PGC_need_scrub bit).
>>
>> We keep track of the first unscrubbed page in a page buddy using first_dirty
>> field. For now it can have two values, 0 (whole buddy needs scrubbing) or
>> INVALID_DIRTY_IDX (the buddy does not need to be scrubbed). Subsequent 
>> patches
>> will allow scrubbing to be interrupted, resulting in first_dirty taking any
>> value.
>>
>> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>
> Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
> with one remark:
>
>> --- a/xen/include/asm-x86/mm.h
>> +++ b/xen/include/asm-x86/mm.h
>> @@ -88,7 +88,15 @@ struct page_info
>>          /* Page is on a free list: ((count_info & PGC_count_mask) == 0). */
>>          struct {
>>              /* Do TLBs need flushing for safety before next page use? */
>> -            bool_t need_tlbflush;
>> +            bool need_tlbflush:1;
>> +
>> +            /*
>> +             * Index of the first *possibly* unscrubbed page in the buddy.
>> +             * One more bit than maximum possible order to accommodate
>> +             * INVALID_DIRTY_IDX.
>> +             */
>> +#define INVALID_DIRTY_IDX ((1UL << (MAX_ORDER + 1)) - 1)
>> +            unsigned long first_dirty:MAX_ORDER + 1;
>>          } free;
> I think generated code will be better with the two fields swapped:
> That way reading first_dirty won't involve a shift, and accessing a
> single bit doesn't require shifts at all on many architectures.

Ok, I will then keep need_tlbflush as the last field so the final struct
(as defined in patch 7) will look like

struct {
        unsigned long first_dirty:MAX_ORDER + 1;
        unsigned long scrub_state:2;
        bool need_tlbflush:1;
};


-boris


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel