From mboxrd@z Thu Jan 1 00:00:00 1970 From: Matthew Wilcox Subject: [PATCH 07/11] thp: Decrement refcount on huge zero page if it is split Date: Tue, 4 Aug 2015 15:58:01 -0400 Message-ID: <1438718285-21168-8-git-send-email-matthew.r.wilcox@intel.com> References: <1438718285-21168-1-git-send-email-matthew.r.wilcox@intel.com> Cc: "Kirill A. Shutemov" , willy@linux.intel.com To: Andrew Morton , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org Return-path: In-Reply-To: <1438718285-21168-1-git-send-email-matthew.r.wilcox@intel.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org From: "Kirill A. Shutemov" The DAX code neglected to put the refcount on the huge zero page. Also we must notify on splits. Signed-off-by: Kirill A. Shutemov Signed-off-by: Matthew Wilcox --- mm/huge_memory.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/mm/huge_memory.c b/mm/huge_memory.c index 5ffdcaa..326d17e 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -2951,7 +2951,9 @@ again: if (unlikely(!pmd_trans_huge(*pmd))) goto unlock; if (vma_is_dax(vma)) { - pmdp_huge_clear_flush(vma, haddr, pmd); + pmd_t _pmd = pmdp_huge_clear_flush_notify(vma, haddr, pmd); + if (is_huge_zero_pmd(_pmd)) + put_huge_zero_page(); } else if (is_huge_zero_pmd(*pmd)) { __split_huge_zero_page_pmd(vma, haddr, pmd); } else { -- 2.1.4