Re: [PATCH] mm/mlock: fix BUG_ON unlocked page for nolinear VMAs

[Wanpeng Li <liwanp@linux.vnet.ibm.com>]

[-- Attachment #1: Type: text/plain, Size: 888 bytes --]

Hi Sasha,
On Tue, Dec 17, 2013 at 10:32:53PM -0500, Sasha Levin wrote:
>On 12/17/2013 10:23 PM, Wanpeng Li wrote:
>>-			mlock_vma_page(page);   /* no-op if already mlocked */
>>-			if (page == check_page)
>>+			if (page != check_page && trylock_page(page)) {
>>+				mlock_vma_page(page);   /* no-op if already mlocked */
>>+				unlock_page(page);
>>+			} else if (page == check_page) {
>>+				mlock_vma_page(page);  /* no-op if already mlocked */
>>  				ret = SWAP_MLOCK;
>>+			}
>
>Previously, if page != check_page and the page was locked, we'd call mlock_vma_page()
>anyways. With this change, we don't. In fact, we'll just skip that entire block not doing
>anything.

Thanks for pointing out. ;-)

>
>If that's something that's never supposed to happen, can we add a
>
>	VM_BUG_ON(page != check_page && PageLocked(page))
>
>Just to cover this new code path?
>

How about this one? 


[-- Attachment #2: 0001-3.patch --]
[-- Type: text/x-diff, Size: 0 bytes --]