Re: [RFC v2] HWPOISON: soft offlining for non-lru movable page

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Minchan,
Thanks for reviewing.
On 2017/1/23 13:14, Minchan Kim wrote:
> Hello,
> 
> On Thu, Jan 19, 2017 at 10:59:03PM +0800, ysxie@xxxxxxxxxxx wrote:
>> From: Yisheng Xie <xieyisheng1@xxxxxxxxxx>
>>
>> @@ -1527,7 +1527,8 @@ static int get_any_page(struct page *page, unsigned long pfn, int flags)
>>  {
>>  	int ret = __get_any_page(page, pfn, flags);
>>  
>> -	if (ret == 1 && !PageHuge(page) && !PageLRU(page)) {
>> +	if (ret == 1 && !PageHuge(page) &&
>> +	    !PageLRU(page) && !__PageMovable(page)) {
> 
> __PageMovable without holding page_lock could be raced so need to check
> if it's okay to miss some of pages offlining by the race.
> When I read description of soft_offline_page, it seems to be okay.
> Just wanted double check. :)
Yes, I have thought about whether should add page_lock to avoid race. For it is ok to
miss some of pages caused by race, I do not add page_lock.

> 
>>  		/*
>>  		 * Try to free it.
>>  		 */
>> @@ -1609,7 +1610,7 @@ static int soft_offline_huge_page(struct page *page, int flags)
>>  
>>  static int __soft_offline_page(struct page *page, int flags)
>>  {
>> -	int ret;
>> +	int ret = -1;
>>  	unsigned long pfn = page_to_pfn(page);
>>  
>>  	/*
>> @@ -1619,7 +1620,8 @@ static int __soft_offline_page(struct page *page, int flags)
>>  	 * so there's no race between soft_offline_page() and memory_failure().
>>  	 */
>>  	lock_page(page);
>> -	wait_on_page_writeback(page);
>> +	if (PageLRU(page))
>> +		wait_on_page_writeback(page);
> 
> I doubt we need to add such limitation(i.e., Only LRU pages could be write-backed).
> Do you have some reason to add that code?

I add this check for not quite sure about whether non-lru page will as marked as
PageWriteBack(page). I will delete no need limitation in next version.

> 
>>  	if (PageHWPoison(page)) {
>>  		unlock_page(page);
>>  		put_hwpoison_page(page);
>> @@ -1630,7 +1632,8 @@ static int __soft_offline_page(struct page *page, int flags)
>>  	 * Try to invalidate first. This should work for
>>  	 * non dirty unmapped page cache pages.
>>  	 */
>> -	ret = invalidate_inode_page(page);
>> +	if (PageLRU(page))
>> +		ret = invalidate_inode_page(page);
> 
> Ditto.
> 
>>  	unlock_page(page);
>>  	/*
>>  	 * RED-PEN would be better to keep it isolated here, but we
>> @@ -1649,7 +1652,10 @@ static int __soft_offline_page(struct page *page, int flags)
>>  	 * Try to migrate to a new page instead. migrate.c
>>  	 * handles a large number of cases for us.
>>  	 */
>> -	ret = isolate_lru_page(page);
>> +	if (PageLRU(page))
>> +		ret = isolate_lru_page(page);
>> +	else
>> +		ret = !isolate_movable_page(page, ISOLATE_UNEVICTABLE);
>>  	/*
>>  	 * Drop page reference which is came from get_any_page()
>>  	 * successful isolate_lru_page() already took another one.
>> @@ -1657,18 +1663,15 @@ static int __soft_offline_page(struct page *page, int flags)
>>  	put_hwpoison_page(page);
>>  	if (!ret) {
>>  		LIST_HEAD(pagelist);
>> -		inc_node_page_state(page, NR_ISOLATED_ANON +
>> +		if (PageLRU(page))
> 
> isolate_lru_page removes PG_lru so this check will be false. Namely, happens
> isolated count mismatch happens.
> 
Really sorry about that. That's my mistake.
I will use !__PageMovable(page) instead in v3.

Thanks
Yisheng Xie.

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



[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]
  Powered by Linux