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=-13.7 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI,NICE_REPLY_A, SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_SANE_1 autolearn=ham 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 87739C433E2 for ; Fri, 11 Sep 2020 11:14:31 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 2F17F21D92 for ; Fri, 11 Sep 2020 11:14:30 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1725770AbgIKLOa (ORCPT ); Fri, 11 Sep 2020 07:14:30 -0400 Received: from mx2.suse.de ([195.135.220.15]:46560 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1725710AbgIKLOT (ORCPT ); Fri, 11 Sep 2020 07:14:19 -0400 X-Virus-Scanned: by amavisd-new at test-mx.suse.de Received: from relay2.suse.de (unknown [195.135.221.27]) by mx2.suse.de (Postfix) with ESMTP id 56778ACDB; Fri, 11 Sep 2020 11:14:33 +0000 (UTC) Subject: Re: [PATCH 08/17] btrfs: refactor how we extract extent buffer from page for alloc_extent_buffer() To: Qu Wenruo , linux-btrfs@vger.kernel.org References: <20200908075230.86856-1-wqu@suse.com> <20200908075230.86856-9-wqu@suse.com> From: Nikolay Borisov Autocrypt: addr=nborisov@suse.com; prefer-encrypt=mutual; keydata= xsFNBFiKBz4BEADNHZmqwhuN6EAzXj9SpPpH/nSSP8YgfwoOqwrP+JR4pIqRK0AWWeWCSwmZ T7g+RbfPFlmQp+EwFWOtABXlKC54zgSf+uulGwx5JAUFVUIRBmnHOYi/lUiE0yhpnb1KCA7f u/W+DkwGerXqhhe9TvQoGwgCKNfzFPZoM+gZrm+kWv03QLUCr210n4cwaCPJ0Nr9Z3c582xc bCUVbsjt7BN0CFa2BByulrx5xD9sDAYIqfLCcZetAqsTRGxM7LD0kh5WlKzOeAXj5r8DOrU2 GdZS33uKZI/kZJZVytSmZpswDsKhnGzRN1BANGP8sC+WD4eRXajOmNh2HL4P+meO1TlM3GLl EQd2shHFY0qjEo7wxKZI1RyZZ5AgJnSmehrPCyuIyVY210CbMaIKHUIsTqRgY5GaNME24w7h TyyVCy2qAM8fLJ4Vw5bycM/u5xfWm7gyTb9V1TkZ3o1MTrEsrcqFiRrBY94Rs0oQkZvunqia c+NprYSaOG1Cta14o94eMH271Kka/reEwSZkC7T+o9hZ4zi2CcLcY0DXj0qdId7vUKSJjEep c++s8ncFekh1MPhkOgNj8pk17OAESanmDwksmzh1j12lgA5lTFPrJeRNu6/isC2zyZhTwMWs k3LkcTa8ZXxh0RfWAqgx/ogKPk4ZxOXQEZetkEyTFghbRH2BIwARAQABzSJOaWtvbGF5IEJv cmlzb3YgPG5ib3Jpc292QHN1c2UuZGU+wsF4BBMBAgAiBQJYijkSAhsDBgsJCAcDAgYVCAIJ CgsEFgIDAQIeAQIXgAAKCRBxvoJG5T8oV/B6D/9a8EcRPdHg8uLEPywuJR8URwXzkofT5bZE IfGF0Z+Lt2ADe+nLOXrwKsamhweUFAvwEUxxnndovRLPOpWerTOAl47lxad08080jXnGfYFS Dc+ew7C3SFI4tFFHln8Y22Q9075saZ2yQS1ywJy+TFPADIprAZXnPbbbNbGtJLoq0LTiESnD w/SUC6sfikYwGRS94Dc9qO4nWyEvBK3Ql8NkoY0Sjky3B0vL572Gq0ytILDDGYuZVo4alUs8 LeXS5ukoZIw1QYXVstDJQnYjFxYgoQ5uGVi4t7FsFM/6ykYDzbIPNOx49Rbh9W4uKsLVhTzG BDTzdvX4ARl9La2kCQIjjWRg+XGuBM5rxT/NaTS78PXjhqWNYlGc5OhO0l8e5DIS2tXwYMDY LuHYNkkpMFksBslldvNttSNei7xr5VwjVqW4vASk2Aak5AleXZS+xIq2FADPS/XSgIaepyTV tkfnyreep1pk09cjfXY4A7qpEFwazCRZg9LLvYVc2M2eFQHDMtXsH59nOMstXx2OtNMcx5p8 0a5FHXE/HoXz3p9bD0uIUq6p04VYOHsMasHqHPbsMAq9V2OCytJQPWwe46bBjYZCOwG0+x58 fBFreP/NiJNeTQPOa6FoxLOLXMuVtpbcXIqKQDoEte9aMpoj9L24f60G4q+pL/54ql2VRscK d87BTQRYigc+ARAAyJSq9EFk28++SLfg791xOh28tLI6Yr8wwEOvM3wKeTfTZd+caVb9gBBy wxYhIopKlK1zq2YP7ZjTP1aPJGoWvcQZ8fVFdK/1nW+Z8/NTjaOx1mfrrtTGtFxVBdSCgqBB jHTnlDYV1R5plJqK+ggEP1a0mr/rpQ9dFGvgf/5jkVpRnH6BY0aYFPprRL8ZCcdv2DeeicOO YMobD5g7g/poQzHLLeT0+y1qiLIFefNABLN06Lf0GBZC5l8hCM3Rpb4ObyQ4B9PmL/KTn2FV Xq/c0scGMdXD2QeWLePC+yLMhf1fZby1vVJ59pXGq+o7XXfYA7xX0JsTUNxVPx/MgK8aLjYW hX+TRA4bCr4uYt/S3ThDRywSX6Hr1lyp4FJBwgyb8iv42it8KvoeOsHqVbuCIGRCXqGGiaeX Wa0M/oxN1vJjMSIEVzBAPi16tztL/wQtFHJtZAdCnuzFAz8ue6GzvsyBj97pzkBVacwp3/Mw qbiu7sDz7yB0d7J2tFBJYNpVt/Lce6nQhrvon0VqiWeMHxgtQ4k92Eja9u80JDaKnHDdjdwq FUikZirB28UiLPQV6PvCckgIiukmz/5ctAfKpyYRGfez+JbAGl6iCvHYt/wAZ7Oqe/3Cirs5 KhaXBcMmJR1qo8QH8eYZ+qhFE3bSPH446+5oEw8A9v5oonKV7zMAEQEAAcLBXwQYAQIACQUC WIoHPgIbDAAKCRBxvoJG5T8oV1pyD/4zdXdOL0lhkSIjJWGqz7Idvo0wjVHSSQCbOwZDWNTN JBTP0BUxHpPu/Z8gRNNP9/k6i63T4eL1xjy4umTwJaej1X15H8Hsh+zakADyWHadbjcUXCkg OJK4NsfqhMuaIYIHbToi9K5pAKnV953xTrK6oYVyd/Rmkmb+wgsbYQJ0Ur1Ficwhp6qU1CaJ mJwFjaWaVgUERoxcejL4ruds66LM9Z1Qqgoer62ZneID6ovmzpCWbi2sfbz98+kW46aA/w8r 7sulgs1KXWhBSv5aWqKU8C4twKjlV2XsztUUsyrjHFj91j31pnHRklBgXHTD/pSRsN0UvM26 lPs0g3ryVlG5wiZ9+JbI3sKMfbdfdOeLxtL25ujs443rw1s/PVghphoeadVAKMPINeRCgoJH zZV/2Z/myWPRWWl/79amy/9MfxffZqO9rfugRBORY0ywPHLDdo9Kmzoxoxp9w3uTrTLZaT9M KIuxEcV8wcVjr+Wr9zRl06waOCkgrQbTPp631hToxo+4rA1jiQF2M80HAet65ytBVR2pFGZF zGYYLqiG+mpUZ+FPjxk9kpkRYz61mTLSY7tuFljExfJWMGfgSg1OxfLV631jV1TcdUnx+h3l Sqs2vMhAVt14zT8mpIuu2VNxcontxgVr1kzYA/tQg32fVRbGr449j1gw57BV9i0vww== Message-ID: <1a9a0712-bd2c-0fe1-8320-d0d239d87528@suse.com> Date: Fri, 11 Sep 2020 14:14:17 +0300 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.10.0 MIME-Version: 1.0 In-Reply-To: <20200908075230.86856-9-wqu@suse.com> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit Sender: linux-btrfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-btrfs@vger.kernel.org On 8.09.20 г. 10:52 ч., Qu Wenruo wrote: > This patch will extract the code to extract extent_buffer from > page::private into its own function, grab_extent_buffer_from_page(). > > Although it's just one line, for later sub-page size support it will > become way more larger. > > Also add some extra comments why we need to do such page::private > dancing. > > Signed-off-by: Qu Wenruo > --- > fs/btrfs/extent_io.c | 49 ++++++++++++++++++++++++++++++++++++-------- > 1 file changed, 40 insertions(+), 9 deletions(-) > > diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c > index 6fafbc1d047b..3c8fe40f67fa 100644 > --- a/fs/btrfs/extent_io.c > +++ b/fs/btrfs/extent_io.c > @@ -5214,6 +5214,44 @@ struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, > } > #endif > > +/* > + * A helper to grab the exist extent buffer from a page. > + * > + * There is a small race window where two callers of alloc_extent_buffer(): > + * Thread 1 | Thread 2 > + * -------------------------------------+--------------------------------------- > + * alloc_extent_buffer() | alloc_extent_buffer() > + * |- eb = __alloc_extent_buffer() | |- eb = __alloc_extent_buffer() > + * |- p = find_or_create_page() | |- p = find_or_create_page() > + * > + * In above case, two ebs get allocated for the same bytenr, and got the same > + * page. > + * We have to rely on the page->mapping->private_lock to make one of them to > + * give up and reuse the allocated eb: > + * > + * | |- grab_extent_buffer_from_page() > + * | |- get nothing > + * | |- attach_extent_buffer_page() > + * | | |- Now page->private is set > + * | |- spin_unlock(&mapping->private_lock); > + * |- spin_lock(private_lock); | |- Continue to insert radix tree. > + * |- grab_extent_buffer_from_page() | > + * |- got eb from thread 2 | > + * |- spin_unlock(private_lock); | > + * |- goto free_eb; | This comment is slightly misleading because it's not graB_extent_buffer_form_page's return value which decides who wins the race but rather the retval of PagePrivate which invoked _before_ grab_extent_buffer_from_page. > + * > + * The function here is to ensure we have proper locking and detect such race > + * so we won't allocating an eb twice. > + */ > +static struct extent_buffer *grab_extent_buffer_from_page(struct page *page) > +{ > + /* > + * For PAGE_SIZE == sectorsize case, a btree_inode page should have its > + * private pointer as extent buffer who owns this page. > + */ > + return (struct extent_buffer *)page->private; > +} > + > struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, > u64 start) > { > @@ -5258,15 +5296,8 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, > > spin_lock(&mapping->private_lock); > if (PagePrivate(p)) { > - /* > - * We could have already allocated an eb for this page > - * and attached one so lets see if we can get a ref on > - * the existing eb, and if we can we know it's good and > - * we can just return that one, else we know we can just > - * overwrite page->private. > - */ > - exists = (struct extent_buffer *)p->private; > - if (atomic_inc_not_zero(&exists->refs)) { > + exists = grab_extent_buffer_from_page(p); > + if (exists && atomic_inc_not_zero(&exists->refs)) { grab_extent_buffer_from_page is guaranteed to return an extent buffer due to the PagePrivate() check above. So simply doing atomic_in_not_zero is fine. > spin_unlock(&mapping->private_lock); > unlock_page(p); > put_page(p); >