Re: [PATCHv5 5/7] mm: make compound_head() robust

2015-09-14 Thread Vlastimil Babka
On 09/11/2015 03:35 PM, Kirill A. Shutemov wrote: index 097c7a4bfbd9..330377f83ac7 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -1686,8 +1686,7 @@ static void __split_huge_page_refcount(struct page *page, (1L << PG_unevictable))); p

Re: [PATCHv5 5/7] mm: make compound_head() robust

2015-09-11 Thread Kirill A. Shutemov
On Thu, Sep 10, 2015 at 12:54:08PM +0200, Vlastimil Babka wrote: > On 09/03/2015 02:35 PM, Kirill A. Shutemov wrote: > > Hugh has pointed that compound_head() call can be unsafe in some > > context. There's one example: > > > > CPU0CPU1 > > > > isolate_migr

Re: [PATCHv5 5/7] mm: make compound_head() robust

2015-09-10 Thread Vlastimil Babka
On 09/03/2015 02:35 PM, Kirill A. Shutemov wrote: > Hugh has pointed that compound_head() call can be unsafe in some > context. There's one example: > > CPU0CPU1 > > isolate_migratepages_block() > page_count() > compound_head() > !!PageTail()

[PATCHv5 5/7] mm: make compound_head() robust

2015-09-03 Thread Kirill A. Shutemov
Hugh has pointed that compound_head() call can be unsafe in some context. There's one example: CPU0CPU1 isolate_migratepages_block() page_count() compound_head() !!PageTail() == true put_page()