From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-5.3 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_PASS,URIBL_BLOCKED,USER_AGENT_MUTT autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 3B4C8C6786E for ; Fri, 26 Oct 2018 11:05:51 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 0B0AB2064A for ; Fri, 26 Oct 2018 11:05:51 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 0B0AB2064A Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=suse.cz Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-btrfs-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727562AbeJZTmX (ORCPT ); Fri, 26 Oct 2018 15:42:23 -0400 Received: from mx2.suse.de ([195.135.220.15]:49402 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1726612AbeJZTmX (ORCPT ); Fri, 26 Oct 2018 15:42:23 -0400 X-Virus-Scanned: by amavisd-new at test-mx.suse.de Received: from relay1.suse.de (unknown [195.135.220.254]) by mx1.suse.de (Postfix) with ESMTP id DB5BBAF63; Fri, 26 Oct 2018 11:05:42 +0000 (UTC) Received: by quack2.suse.cz (Postfix, from userid 1000) id ABA851E06DB; Fri, 26 Oct 2018 11:47:33 +0200 (CEST) Date: Fri, 26 Oct 2018 11:47:33 +0200 From: Jan Kara To: Josef Bacik Cc: Jan Kara , kernel-team@fb.com, hannes@cmpxchg.org, linux-kernel@vger.kernel.org, tj@kernel.org, david@fromorbit.com, akpm@linux-foundation.org, linux-fsdevel@vger.kernel.org, linux-btrfs@vger.kernel.org, riel@fb.com, linux-mm@kvack.org Subject: Re: [PATCH 7/7] btrfs: drop mmap_sem in mkwrite for btrfs Message-ID: <20181026094733.GB25227@quack2.suse.cz> References: <20181018202318.9131-1-josef@toxicpanda.com> <20181018202318.9131-8-josef@toxicpanda.com> <20181025132230.GD7711@quack2.suse.cz> <20181025135849.bu3cmjnrvz5yysye@macbook-pro-91.dhcp.thefacebook.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20181025135849.bu3cmjnrvz5yysye@macbook-pro-91.dhcp.thefacebook.com> User-Agent: Mutt/1.10.1 (2018-07-13) Sender: linux-btrfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-btrfs@vger.kernel.org On Thu 25-10-18 09:58:51, Josef Bacik wrote: > On Thu, Oct 25, 2018 at 03:22:30PM +0200, Jan Kara wrote: > > On Thu 18-10-18 16:23:18, Josef Bacik wrote: > > > ->page_mkwrite is extremely expensive in btrfs. We have to reserve > > > space, which can take 6 lifetimes, and we could possibly have to wait on > > > writeback on the page, another several lifetimes. To avoid this simply > > > drop the mmap_sem if we didn't have the cached page and do all of our > > > work and return the appropriate retry error. If we have the cached page > > > we know we did all the right things to set this page up and we can just > > > carry on. > > > > > > Signed-off-by: Josef Bacik > > ... > > > @@ -8828,6 +8830,29 @@ vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf) > > > > > > reserved_space = PAGE_SIZE; > > > > > > + /* > > > + * We have our cached page from a previous mkwrite, check it to make > > > + * sure it's still dirty and our file size matches when we ran mkwrite > > > + * the last time. If everything is OK then return VM_FAULT_LOCKED, > > > + * otherwise do the mkwrite again. > > > + */ > > > + if (vmf->flags & FAULT_FLAG_USED_CACHED) { > > > + lock_page(page); > > > + if (vmf->cached_size == i_size_read(inode) && > > > + PageDirty(page)) > > > + return VM_FAULT_LOCKED; > > > + unlock_page(page); > > > + } > > > > I guess this is similar to Dave's comment: Why is i_size so special? What > > makes sure that file didn't get modified between time you've prepared > > cached_page and now such that you need to do the preparation again? > > And if indeed metadata prepared for a page cannot change, what's so special > > about it being that particular cached_page? > > > > Maybe to phrase my objections differently: Your preparations in > > btrfs_page_mkwrite() are obviously related to your filesystem metadata. So > > why cannot you infer from that metadata (extent tree, whatever - I'd use > > extent status tree in ext4) whether that particular file+offset is already > > prepared for writing and just bail out with success in that case? > > > > I was just being overly paranoid, I was afraid of the case where we would > truncate and then extend in between, but now that I actually think about it that > would end up with the page not being on the mapping anymore so we would catch > that case. I've dropped this part from my current version. I'm getting some > testing on these patches in production and I'll post them sometime next week > once I'm happy with them. Thanks, OK, but do you still need the vmf->cached_page stuff? Because I don't see why even that is necessary... Honza -- Jan Kara SUSE Labs, CR