Re: [PATCH 5/7] mm,ksm: swapoff might need to copy

2013-02-22 Thread Hugh Dickins
On Thu, 21 Feb 2013, Johannes Weiner wrote: > On Thu, Feb 21, 2013 at 12:25:40AM -0800, Hugh Dickins wrote: > > Before establishing that KSM page migration was the cause of my > > WARN_ON_ONCE(page_mapped(page))s, I suspected that they came from the > > lack of a ksm_might_need_to_copy() in

Re: [PATCH 5/7] mm,ksm: swapoff might need to copy

2013-02-22 Thread Hugh Dickins
On Thu, 21 Feb 2013, Johannes Weiner wrote: On Thu, Feb 21, 2013 at 12:25:40AM -0800, Hugh Dickins wrote: Before establishing that KSM page migration was the cause of my WARN_ON_ONCE(page_mapped(page))s, I suspected that they came from the lack of a ksm_might_need_to_copy() in swapoff's

Re: [PATCH 5/7] mm,ksm: swapoff might need to copy

2013-02-21 Thread Johannes Weiner
On Thu, Feb 21, 2013 at 12:25:40AM -0800, Hugh Dickins wrote: > Before establishing that KSM page migration was the cause of my > WARN_ON_ONCE(page_mapped(page))s, I suspected that they came from the > lack of a ksm_might_need_to_copy() in swapoff's unuse_pte() - which > in many respects is

[PATCH 5/7] mm,ksm: swapoff might need to copy

2013-02-21 Thread Hugh Dickins
Before establishing that KSM page migration was the cause of my WARN_ON_ONCE(page_mapped(page))s, I suspected that they came from the lack of a ksm_might_need_to_copy() in swapoff's unuse_pte() - which in many respects is equivalent to faulting in a page. In fact I've never caught that as the

[PATCH 5/7] mm,ksm: swapoff might need to copy

2013-02-21 Thread Hugh Dickins
Before establishing that KSM page migration was the cause of my WARN_ON_ONCE(page_mapped(page))s, I suspected that they came from the lack of a ksm_might_need_to_copy() in swapoff's unuse_pte() - which in many respects is equivalent to faulting in a page. In fact I've never caught that as the

Re: [PATCH 5/7] mm,ksm: swapoff might need to copy

2013-02-21 Thread Johannes Weiner
On Thu, Feb 21, 2013 at 12:25:40AM -0800, Hugh Dickins wrote: Before establishing that KSM page migration was the cause of my WARN_ON_ONCE(page_mapped(page))s, I suspected that they came from the lack of a ksm_might_need_to_copy() in swapoff's unuse_pte() - which in many respects is equivalent