From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-qg0-f45.google.com (mail-qg0-f45.google.com [209.85.192.45]) by kanga.kvack.org (Postfix) with ESMTP id E431D6B0253 for ; Thu, 25 Feb 2016 13:45:06 -0500 (EST) Received: by mail-qg0-f45.google.com with SMTP id y9so47751544qgd.3 for ; Thu, 25 Feb 2016 10:45:06 -0800 (PST) Received: from mail-qg0-x230.google.com (mail-qg0-x230.google.com. [2607:f8b0:400d:c04::230]) by mx.google.com with ESMTPS id w9si9247474qhc.10.2016.02.25.10.45.06 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Thu, 25 Feb 2016 10:45:06 -0800 (PST) Received: by mail-qg0-x230.google.com with SMTP id y9so47751273qgd.3 for ; Thu, 25 Feb 2016 10:45:06 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <20160223192835.GJ9157@redhat.com> References: <20160223154950.GA22449@node.shutemov.name> <20160223180609.GC23289@redhat.com> <20160223183832.GB21820@node.shutemov.name> <20160223192835.GJ9157@redhat.com> Date: Thu, 25 Feb 2016 10:45:05 -0800 Message-ID: Subject: Re: THP race? From: Dan Williams Content-Type: text/plain; charset=UTF-8 Sender: owner-linux-mm@kvack.org List-ID: To: Andrea Arcangeli Cc: "Kirill A. Shutemov" , linux-mm On Tue, Feb 23, 2016 at 11:28 AM, Andrea Arcangeli wrote: > On Tue, Feb 23, 2016 at 09:38:32PM +0300, Kirill A. Shutemov wrote: >> pmd_trans_unstable(pmd), otherwise looks good: > > Yes sorry. > >> Acked-by: Kirill A. Shutemov > > Thanks for the quick ack, I just noticed or I would have added it to > the resubmit, but it can be still added to -mm. > >> BTW, I guess DAX would need to introduce the same infrastructure for >> pmd_devmap(). Dan? > > There is a i_mmap_lock_write in the truncate path that saves the day > for the pmd zapping in the truncate() case without mmap_sem (the only > case anon THP doesn't need to care about as truncate isn't possible in > the anon case), but not in the MADV_DONTNEED madvise case that runs > only with the mmap_sem for reading. > > The only objective of this "infrastructure" is to add no pmd_lock()ing > overhead to the page fault, if the mapping is already established but > not huge, and we've just to walk through the pmd to reach the > pte. All because MADV_DONTNEED is running with the mmap_sem for > reading unlike munmap and other slower syscalls that are forced to > mangle the vmas and have to take the mmap_sem for writing regardless. > > The question for DAX is if it should do a pmd_devmap check inside > pmd_none_or_trans_huge_or_clear_bad() after pmd_trans_huge() and get > away with a one liner, or add its own infrastructure with > pmd_devmap_unstable(). In the pmd_devmap case the problem isn't just > in __handle_mm_fault. If it could share the same infrastructure it'd > be ideal. > Yes, I see no reason why we can't/shoudn't move the pmd_devmap() check inside pmd_none_or_trans_huge_or_clear_bad(). -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org