Re: [PATCH] mm: reposition zbud page in lru list if not freed in zbud_free

[Wang Sheng-Hui <shhuiw@gmail.com>]

On 2014a1'09ae??02ae?JPY 22:39, Seth Jennings wrote:
> On Sun, Aug 31, 2014 at 09:29:29PM +0800, Wang Sheng-Hui wrote:
>> Reposition zbud page in the lru list of the pool if the zbud page
>> is not freed in zbud_free.
> 
> This doesn't mention what is wrong with the current code.  Afaict, the
> code is doing the right thing by not changing the position of the zpage
> in the LRU.  Why would we want to move a zpage to the front of the LRU
> just because one of the allocations it contains is freed?  The "age" of
> the other allocation is unchanged and its containing zpage should
> maintain its position in the LRU, right?
> 

Seth,

I have one question.

In zbud_reclaim_page, if one page contains handles not freed, then the page
is repositioned, in current code.

540                 } else if (zhdr->first_chunks == 0 ||
541                                 zhdr->last_chunks == 0) {
542                         /* add to unbuddied list */
543                         freechunks = num_free_chunks(zhdr);
544                         list_add(&zhdr->buddy, &pool->unbuddied[freechunks]);
545                 } else {
546                         /* add to buddied list */
547                         list_add(&zhdr->buddy, &pool->buddied);
548                 }
549
550                 /* add to beginning of LRU */
551                 list_add(&zhdr->lru, &pool->lru);

I want to why it is different for zbud_free?

Regards,
Sheng-Hui
> Thanks,
> Seth
> 
>>
>> Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>
>> ---
>>  mm/zbud.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/mm/zbud.c b/mm/zbud.c
>> index f26e7fc..b1d7777 100644
>> --- a/mm/zbud.c
>> +++ b/mm/zbud.c
>> @@ -432,15 +432,16 @@ void zbud_free(struct zbud_pool *pool, unsigned long handle)
>>  	/* Remove from existing buddy list */
>>  	list_del(&zhdr->buddy);
>>  
>> +	list_del(&zhdr->lru);
>>  	if (zhdr->first_chunks == 0 && zhdr->last_chunks == 0) {
>>  		/* zbud page is empty, free */
>> -		list_del(&zhdr->lru);
>>  		free_zbud_page(zhdr);
>>  		pool->pages_nr--;
>>  	} else {
>>  		/* Add to unbuddied list */
>>  		freechunks = num_free_chunks(zhdr);
>>  		list_add(&zhdr->buddy, &pool->unbuddied[freechunks]);
>> +		list_add(&zhdr->lru, &pool->lru);
>>  	}
>>  
>>  	spin_unlock(&pool->lock);
>> -- 
>> 1.8.3.2
>>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>