Re: [Qemu-devel] [PATCH v3 09/10] migration: fix calculating xbzrle_counters.cache_miss_rate

[Xiao Guangrong <guangrong.xiao@gmail.com>]

On 08/08/2018 02:05 PM, Peter Xu wrote:
> On Tue, Aug 07, 2018 at 05:12:08PM +0800, guangrong.xiao@gmail.com wrote:
>> From: Xiao Guangrong <xiaoguangrong@tencent.com>
>>
>> As Peter pointed out:
>> | - xbzrle_counters.cache_miss is done in save_xbzrle_page(), so it's
>> |   per-guest-page granularity
>> |
>> | - RAMState.iterations is done for each ram_find_and_save_block(), so
>> |   it's per-host-page granularity
>> |
>> | An example is that when we migrate a 2M huge page in the guest, we
>> | will only increase the RAMState.iterations by 1 (since
>> | ram_find_and_save_block() will be called once), but we might increase
>> | xbzrle_counters.cache_miss for 2M/4K=512 times (we'll call
>> | save_xbzrle_page() that many times) if all the pages got cache miss.
>> | Then IMHO the cache miss rate will be 512/1=51200% (while it should
>> | actually be just 100% cache miss).
>>
>> And he also suggested as xbzrle_counters.cache_miss_rate is the only
>> user of rs->iterations we can adapt it to count guest page numbers
>>
>> After that, rename 'iterations' to 'handle_pages' to better reflect
>> its meaning
>>
>> Suggested-by: Peter Xu <peterx@redhat.com>
>> Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
>> ---
>>   migration/ram.c | 18 +++++++++---------
>>   1 file changed, 9 insertions(+), 9 deletions(-)
>>
>> diff --git a/migration/ram.c b/migration/ram.c
>> index 09be01dca2..bd7c18d1f9 100644
>> --- a/migration/ram.c
>> +++ b/migration/ram.c
>> @@ -300,10 +300,10 @@ struct RAMState {
>>       uint64_t num_dirty_pages_period;
>>       /* xbzrle misses since the beginning of the period */
>>       uint64_t xbzrle_cache_miss_prev;
>> -    /* number of iterations at the beginning of period */
>> -    uint64_t iterations_prev;
>> -    /* Iterations since start */
>> -    uint64_t iterations;
>> +    /* total handled pages at the beginning of period */
>> +    uint64_t handle_pages_prev;
>> +    /* total handled pages since start */
>> +    uint64_t handle_pages;
> 
> The name is not that straightforward to me.  I would think about
> "[guest|host]_page_count" or something better, or we just keep the old
> naming but with a better comment would be fine too.

The filed actually indicates total pages (target pages more precisely)
handled during live migration. 'iterations' confuses us completely.

It's target_page_count good to you?

> 
>>       /* number of dirty bits in the bitmap */
>>       uint64_t migration_dirty_pages;
>>       /* last dirty_sync_count we have seen */
>> @@ -1587,19 +1587,19 @@ uint64_t ram_pagesize_summary(void)
>>   
>>   static void migration_update_rates(RAMState *rs, int64_t end_time)
>>   {
>> -    uint64_t iter_count = rs->iterations - rs->iterations_prev;
>> +    uint64_t page_count = rs->handle_pages - rs->handle_pages_prev;
>>   
>>       /* calculate period counters */
>>       ram_counters.dirty_pages_rate = rs->num_dirty_pages_period * 1000
>>                   / (end_time - rs->time_last_bitmap_sync);
>>   
>> -    if (!iter_count) {
>> +    if (!page_count) {
>>           return;
>>       }
>>   
>>       if (migrate_use_xbzrle()) {
>>           xbzrle_counters.cache_miss_rate = (double)(xbzrle_counters.cache_miss -
>> -            rs->xbzrle_cache_miss_prev) / iter_count;
>> +            rs->xbzrle_cache_miss_prev) / page_count;
>>           rs->xbzrle_cache_miss_prev = xbzrle_counters.cache_miss;
>>       }
>>   }
>> @@ -1657,7 +1657,7 @@ static void migration_bitmap_sync(RAMState *rs)
>>   
>>           migration_update_rates(rs, end_time);
>>   
>> -        rs->iterations_prev = rs->iterations;
>> +        rs->handle_pages_prev = rs->handle_pages;
>>   
>>           /* reset period counters */
>>           rs->time_last_bitmap_sync = end_time;
>> @@ -3209,7 +3209,7 @@ static int ram_save_iterate(QEMUFile *f, void *opaque)
>>               break;
>>           }
>>   
>> -        rs->iterations++;
>> +        rs->handle_pages += pages;
> 
> So it's still counting host pages, is this your intention to only
> change the name in the patch?

Hmm... the value returned by ram_find_and_save_block() isn't the total
target pages posted out?

/**
  * ram_find_and_save_block: finds a dirty page and sends it to f
  *
  * Called within an RCU critical section.
  *
  * Returns the number of pages written where zero means no dirty pages,
  * or negative on error
...

  *
  * On systems where host-page-size > target-page-size it will send all the
  * pages in a host page that are dirty.
  */