From patchwork Mon Jun 19 06:47:41 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Christoph Hellwig X-Patchwork-Id: 13284025 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 923DEEB64DB for ; Mon, 19 Jun 2023 06:48:06 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230052AbjFSGsF (ORCPT ); Mon, 19 Jun 2023 02:48:05 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:49470 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229456AbjFSGsD (ORCPT ); Mon, 19 Jun 2023 02:48:03 -0400 Received: from bombadil.infradead.org (bombadil.infradead.org [IPv6:2607:7c80:54:3::133]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 55AF613D for ; Sun, 18 Jun 2023 23:48:02 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=infradead.org; s=bombadil.20210309; h=Content-Transfer-Encoding: MIME-Version:References:In-Reply-To:Message-Id:Date:Subject:Cc:To:From:Sender :Reply-To:Content-Type:Content-ID:Content-Description; bh=KsdBeZ20scZ+l4cg5+MIpGOwLSatgPbh99awfYXDByY=; b=Lgv7m5JIOisTxqJHs/ppumUtvL 63zdJK7DsYyyvSXGlEnAnaLrOwV4fKkL52nOSxg0d5nCR2G9yMPiGuPqZga56h8KeBJ0KZJk5Xwf/ KO0I16KeRP1jeqhF1VXWGeQbPCapq2mCMXxnxcVDpLAQ1i3i736h1Rzu7KUzKVNC4+1YcWdBBE4Rr wVM9SqvIuzoS2JhEI6F9DCk/PW7frvlUUzBMFiQgMJQqwiUkRNRo8Z2MB3EDdNZXrsXOGuk9zcebx xDq+w+1uruuPCu7GZW2XfBBCV+hZFJfwwPH0JbxWBqXDI53hw7V9dsmjc6Ktqmq4Ad3SpOU8D9bXy T24V5abg==; Received: from 2a02-8389-2341-5b80-39d3-4735-9a3c-88d8.cable.dynamic.v6.surfer.at ([2a02:8389:2341:5b80:39d3:4735:9a3c:88d8] helo=localhost) by bombadil.infradead.org with esmtpsa (Exim 4.96 #2 (Red Hat Linux)) id 1qB8gC-007cuH-2X; Mon, 19 Jun 2023 06:48:01 +0000 From: Christoph Hellwig To: Chris Mason , Josef Bacik , David Sterba Cc: linux-btrfs@vger.kernel.org Subject: [PATCH 1/2] btrfs: pass a flags argument to cow_file_range Date: Mon, 19 Jun 2023 08:47:41 +0200 Message-Id: <20230619064742.629345-2-hch@lst.de> X-Mailer: git-send-email 2.39.2 In-Reply-To: <20230619064742.629345-1-hch@lst.de> References: <20230619064742.629345-1-hch@lst.de> MIME-Version: 1.0 X-SRS-Rewrite: SMTP reverse-path rewritten from by bombadil.infradead.org. See http://www.infradead.org/rpr.html Precedence: bulk List-ID: X-Mailing-List: linux-btrfs@vger.kernel.org The int used as bool unlock is not a very good way to describe the behavior, and the next patch will have to add another beahvior modifier. Switch to pass a flag instead, with an inital CFR_KEEP_LOCKED flag that specifies the pages should always be kept locked. This is the inverse of the old unlock argument for the reason that it requires a flag for the exceptional behavior. Signed-off-by: Christoph Hellwig --- fs/btrfs/inode.c | 51 ++++++++++++++++++++++-------------------------- 1 file changed, 23 insertions(+), 28 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index dbbb67293e345c..92a78940991fcb 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -124,11 +124,13 @@ static struct kmem_cache *btrfs_inode_cachep; static int btrfs_setsize(struct inode *inode, struct iattr *attr); static int btrfs_truncate(struct btrfs_inode *inode, bool skip_writeback); + +#define CFR_KEEP_LOCKED (1 << 0) static noinline int cow_file_range(struct btrfs_inode *inode, struct page *locked_page, u64 start, u64 end, int *page_started, - unsigned long *nr_written, int unlock, - u64 *done_offset); + unsigned long *nr_written, u64 *done_offset, + u32 flags); static struct extent_map *create_io_em(struct btrfs_inode *inode, u64 start, u64 len, u64 orig_start, u64 block_start, u64 block_len, u64 orig_block_len, @@ -1148,7 +1150,7 @@ static int submit_uncompressed_range(struct btrfs_inode *inode, * can directly submit them without interruption. */ ret = cow_file_range(inode, locked_page, start, end, &page_started, - &nr_written, 0, NULL); + &nr_written, NULL, CFR_KEEP_LOCKED); /* Inline extent inserted, page gets unlocked and everything is done */ if (page_started) return 0; @@ -1362,25 +1364,18 @@ static u64 get_extent_allocation_hint(struct btrfs_inode *inode, u64 start, * locked_page is the page that writepage had locked already. We use * it to make sure we don't do extra locks or unlocks. * - * *page_started is set to one if we unlock locked_page and do everything - * required to start IO on it. It may be clean and already done with - * IO when we return. - * - * When unlock == 1, we unlock the pages in successfully allocated regions. - * When unlock == 0, we leave them locked for writing them out. + * When this function fails, it unlocks all pages except @locked_page. * - * However, we unlock all the pages except @locked_page in case of failure. + * When this function successfully creates an inline extent, it sets page_started + * to 1 and unlocks all pages including locked_page and starts I/O on them. + * (In reality inline extents are limited to a single page, so locked_page is + * the only page handled anyway). * - * In summary, page locking state will be as follow: + * When this function succeed and creates a normal extent, the page locking + * status depends on the passed in flags: * - * - page_started == 1 (return value) - * - All the pages are unlocked. IO is started. - * - Note that this can happen only on success - * - unlock == 1 - * - All the pages except @locked_page are unlocked in any case - * - unlock == 0 - * - On success, all the pages are locked for writing out them - * - On failure, all the pages except @locked_page are unlocked + * - If CFR_KEEP_LOCKED is set, all pages are kept locked. + * - Else all pages except for @locked_page are unlocked. * * When a failure happens in the second or later iteration of the * while-loop, the ordered extents created in previous iterations are kept @@ -1391,8 +1386,8 @@ static u64 get_extent_allocation_hint(struct btrfs_inode *inode, u64 start, static noinline int cow_file_range(struct btrfs_inode *inode, struct page *locked_page, u64 start, u64 end, int *page_started, - unsigned long *nr_written, int unlock, - u64 *done_offset) + unsigned long *nr_written, u64 *done_offset, + u32 flags) { struct btrfs_root *root = inode->root; struct btrfs_fs_info *fs_info = root->fs_info; @@ -1558,7 +1553,7 @@ static noinline int cow_file_range(struct btrfs_inode *inode, * Do set the Ordered (Private2) bit so we know this page was * properly setup for writepage. */ - page_ops = unlock ? PAGE_UNLOCK : 0; + page_ops = (flags & CFR_KEEP_LOCKED) ? 0 : PAGE_UNLOCK; page_ops |= PAGE_SET_ORDERED; extent_clear_unlock_delalloc(inode, start, start + ram_size - 1, @@ -1627,10 +1622,10 @@ static noinline int cow_file_range(struct btrfs_inode *inode, * EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV are handled by the cleanup * function. * - * However, in case of unlock == 0, we still need to unlock the pages - * (except @locked_page) to ensure all the pages are unlocked. + * However, in case of CFR_KEEP_LOCKED, we still need to unlock the + * pages (except @locked_page) to ensure all the pages are unlocked. */ - if (!unlock && orig_start < start) { + if ((flags & CFR_KEEP_LOCKED) && orig_start < start) { if (!locked_page) mapping_set_error(inode->vfs_inode.i_mapping, ret); extent_clear_unlock_delalloc(inode, orig_start, start - 1, @@ -1836,7 +1831,7 @@ static noinline int run_delalloc_zoned(struct btrfs_inode *inode, while (start <= end) { ret = cow_file_range(inode, locked_page, start, end, page_started, - nr_written, 0, &done_offset); + nr_written, &done_offset, CFR_KEEP_LOCKED); if (ret && ret != -EAGAIN) return ret; @@ -1956,7 +1951,7 @@ static int fallback_to_cow(struct btrfs_inode *inode, struct page *locked_page, } return cow_file_range(inode, locked_page, start, end, page_started, - nr_written, 1, NULL); + nr_written, NULL, 0); } struct can_nocow_file_extent_args { @@ -2433,7 +2428,7 @@ int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page page_started, nr_written, wbc); else ret = cow_file_range(inode, locked_page, start, end, - page_started, nr_written, 1, NULL); + page_started, nr_written, NULL, 0); out: ASSERT(ret <= 0); From patchwork Mon Jun 19 06:47:42 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Christoph Hellwig X-Patchwork-Id: 13284026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id DE414EB64DA for ; Mon, 19 Jun 2023 06:48:08 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230054AbjFSGsH (ORCPT ); Mon, 19 Jun 2023 02:48:07 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:49476 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229456AbjFSGsG (ORCPT ); Mon, 19 Jun 2023 02:48:06 -0400 Received: from bombadil.infradead.org (bombadil.infradead.org [IPv6:2607:7c80:54:3::133]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id AE4AE13D for ; Sun, 18 Jun 2023 23:48:05 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=infradead.org; s=bombadil.20210309; h=Content-Transfer-Encoding: MIME-Version:References:In-Reply-To:Message-Id:Date:Subject:Cc:To:From:Sender :Reply-To:Content-Type:Content-ID:Content-Description; bh=EmHjEGI5RwNCUkQMYYFchukDY4G0WoIisHyFoBWsgr4=; b=Mr5ucVUQFu/hmBy2we7C+Upsuf M5sZ/x5aEIyErhXSbvwAZBn9o6Y+OG86q2F5mBw4O5w7Op9VeOqPhxg9NHs5dlh+KkMsZaeUApnLz GQg2waSWGVOyMpL97YMzsS9CHgkMboLAQ1qKf/6DY4aM3xE9OJbBK09asnkCQLdZBCqmBT7eOkzNb 4o4axHacS2DcYnyezKThL3a0rO5kXRefYG3hqe2YEueF4Ceav/BjJCNOoyy+djHSidSqILQReZXef WMA8AycCmBhMYa5dqbguiJurpLEz850RwT2x71w4miRYnN5SRDkuV7UdRhOOrxvE2S0vKvGPZaw9e szn+g5ug==; Received: from 2a02-8389-2341-5b80-39d3-4735-9a3c-88d8.cable.dynamic.v6.surfer.at ([2a02:8389:2341:5b80:39d3:4735:9a3c:88d8] helo=localhost) by bombadil.infradead.org with esmtpsa (Exim 4.96 #2 (Red Hat Linux)) id 1qB8gG-007cuV-0Q; Mon, 19 Jun 2023 06:48:04 +0000 From: Christoph Hellwig To: Chris Mason , Josef Bacik , David Sterba Cc: linux-btrfs@vger.kernel.org Subject: [PATCH 2/2] btrfs: don't create inline extents in fallback_to_cow Date: Mon, 19 Jun 2023 08:47:42 +0200 Message-Id: <20230619064742.629345-3-hch@lst.de> X-Mailer: git-send-email 2.39.2 In-Reply-To: <20230619064742.629345-1-hch@lst.de> References: <20230619064742.629345-1-hch@lst.de> MIME-Version: 1.0 X-SRS-Rewrite: SMTP reverse-path rewritten from by bombadil.infradead.org. See http://www.infradead.org/rpr.html Precedence: bulk List-ID: X-Mailing-List: linux-btrfs@vger.kernel.org For nodatacow files, run_delalloc_nocow can still fall back to COW allocations when required and calls to fallback_to_cow helper for that. For such an allocation we can have multiple ordered_extents for existing extents that NOCOW overwrites and new allocations that fallback_to_cow creates. If one of the new extents is an inline extent, the writepages could would have to avoid normal page writeback for them as indicated by the page_started return argument, which run_delalloc_nocow can't return. Fix this by never creating inline extents from fallback_to_cow. Signed-off-by: Christoph Hellwig --- fs/btrfs/inode.c | 34 ++++++++++++++++++++-------------- 1 file changed, 20 insertions(+), 14 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 92a78940991fcb..cddf54bc330c44 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -126,6 +126,7 @@ static int btrfs_setsize(struct inode *inode, struct iattr *attr); static int btrfs_truncate(struct btrfs_inode *inode, bool skip_writeback); #define CFR_KEEP_LOCKED (1 << 0) +#define CFR_NOINLINE (1 << 1) static noinline int cow_file_range(struct btrfs_inode *inode, struct page *locked_page, u64 start, u64 end, int *page_started, @@ -1426,7 +1427,8 @@ static noinline int cow_file_range(struct btrfs_inode *inode, * This means we can trigger inline extent even if we didn't want to. * So here we skip inline extent creation completely. */ - if (start == 0 && fs_info->sectorsize == PAGE_SIZE) { + if (start == 0 && fs_info->sectorsize == PAGE_SIZE && + !(flags & CFR_NOINLINE)) { u64 actual_end = min_t(u64, i_size_read(&inode->vfs_inode), end + 1); @@ -1889,15 +1891,17 @@ static noinline int csum_exist_in_range(struct btrfs_fs_info *fs_info, } static int fallback_to_cow(struct btrfs_inode *inode, struct page *locked_page, - const u64 start, const u64 end, - int *page_started, unsigned long *nr_written) + const u64 start, const u64 end) { const bool is_space_ino = btrfs_is_free_space_inode(inode); const bool is_reloc_ino = btrfs_is_data_reloc_root(inode->root); const u64 range_bytes = end + 1 - start; struct extent_io_tree *io_tree = &inode->io_tree; + int page_started = 0; + unsigned long nr_written; u64 range_start = start; u64 count; + int ret; /* * If EXTENT_NORESERVE is set it means that when the buffered write was @@ -1950,8 +1954,15 @@ static int fallback_to_cow(struct btrfs_inode *inode, struct page *locked_page, NULL); } - return cow_file_range(inode, locked_page, start, end, page_started, - nr_written, NULL, 0); + /* + * Don't try to create inline extents, as a mix of inline extent that + * is written out and unlocked directly and a normal nocow extent + * doesn't work. + */ + ret = cow_file_range(inode, locked_page, start, end, &page_started, + &nr_written, NULL, CFR_NOINLINE); + ASSERT(!page_started); + return ret; } struct can_nocow_file_extent_args { @@ -2100,9 +2111,7 @@ static int can_nocow_file_extent(struct btrfs_path *path, */ static noinline int run_delalloc_nocow(struct btrfs_inode *inode, struct page *locked_page, - const u64 start, const u64 end, - int *page_started, - unsigned long *nr_written) + const u64 start, const u64 end) { struct btrfs_fs_info *fs_info = inode->root->fs_info; struct btrfs_root *root = inode->root; @@ -2270,8 +2279,7 @@ static noinline int run_delalloc_nocow(struct btrfs_inode *inode, */ if (cow_start != (u64)-1) { ret = fallback_to_cow(inode, locked_page, - cow_start, found_key.offset - 1, - page_started, nr_written); + cow_start, found_key.offset - 1); if (ret) goto error; cow_start = (u64)-1; @@ -2352,8 +2360,7 @@ static noinline int run_delalloc_nocow(struct btrfs_inode *inode, if (cow_start != (u64)-1) { cur_offset = end; - ret = fallback_to_cow(inode, locked_page, cow_start, end, - page_started, nr_written); + ret = fallback_to_cow(inode, locked_page, cow_start, end); if (ret) goto error; } @@ -2412,8 +2419,7 @@ int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page * preallocated inodes. */ ASSERT(!zoned || btrfs_is_data_reloc_root(inode->root)); - ret = run_delalloc_nocow(inode, locked_page, start, end, - page_started, nr_written); + ret = run_delalloc_nocow(inode, locked_page, start, end); goto out; }