From: Liu Bo <bo.li.liu@oracle.com>
To: linux-btrfs@vger.kernel.org
Subject: [PATCH v2 05/10] Btrfs: add extent map selftests
Date: Fri, 5 Jan 2018 12:51:12 -0700 [thread overview]
Message-ID: <20180105195117.5131-6-bo.li.liu@oracle.com> (raw)
In-Reply-To: <20180105195117.5131-1-bo.li.liu@oracle.com>
We've observed that btrfs_get_extent() and merge_extent_mapping() could
return -EEXIST in several cases, and they are caused by some racy
condition, e.g dio read vs dio write, which makes the problem very tricky
to reproduce.
This adds extent map selftests in order to simulate those racy situations.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
---
fs/btrfs/Makefile | 2 +-
fs/btrfs/tests/btrfs-tests.c | 3 +
fs/btrfs/tests/btrfs-tests.h | 1 +
fs/btrfs/tests/extent-map-tests.c | 202 ++++++++++++++++++++++++++++++++++++++
4 files changed, 207 insertions(+), 1 deletion(-)
create mode 100644 fs/btrfs/tests/extent-map-tests.c
diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
index 6fe881d..0c43736 100644
--- a/fs/btrfs/Makefile
+++ b/fs/btrfs/Makefile
@@ -19,4 +19,4 @@ btrfs-$(CONFIG_BTRFS_FS_REF_VERIFY) += ref-verify.o
btrfs-$(CONFIG_BTRFS_FS_RUN_SANITY_TESTS) += tests/free-space-tests.o \
tests/extent-buffer-tests.o tests/btrfs-tests.o \
tests/extent-io-tests.o tests/inode-tests.o tests/qgroup-tests.o \
- tests/free-space-tree-tests.o
+ tests/free-space-tree-tests.o tests/extent-map-tests.o
diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c
index d3f2537..9786d8c 100644
--- a/fs/btrfs/tests/btrfs-tests.c
+++ b/fs/btrfs/tests/btrfs-tests.c
@@ -277,6 +277,9 @@ int btrfs_run_sanity_tests(void)
goto out;
}
}
+ ret = btrfs_test_extent_map();
+ if (ret)
+ goto out;
out:
btrfs_destroy_test_fs();
return ret;
diff --git a/fs/btrfs/tests/btrfs-tests.h b/fs/btrfs/tests/btrfs-tests.h
index 266f1e3..bc0615b 100644
--- a/fs/btrfs/tests/btrfs-tests.h
+++ b/fs/btrfs/tests/btrfs-tests.h
@@ -33,6 +33,7 @@ int btrfs_test_extent_io(u32 sectorsize, u32 nodesize);
int btrfs_test_inodes(u32 sectorsize, u32 nodesize);
int btrfs_test_qgroups(u32 sectorsize, u32 nodesize);
int btrfs_test_free_space_tree(u32 sectorsize, u32 nodesize);
+int btrfs_test_extent_map(void);
struct inode *btrfs_new_test_inode(void);
struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize);
void btrfs_free_dummy_fs_info(struct btrfs_fs_info *fs_info);
diff --git a/fs/btrfs/tests/extent-map-tests.c b/fs/btrfs/tests/extent-map-tests.c
new file mode 100644
index 0000000..0407396
--- /dev/null
+++ b/fs/btrfs/tests/extent-map-tests.c
@@ -0,0 +1,202 @@
+/*
+ * Copyright (C) 2017 Oracle. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#include <linux/types.h>
+#include "btrfs-tests.h"
+#include "../ctree.h"
+
+static void free_extent_map_tree(struct extent_map_tree *em_tree)
+{
+ struct extent_map *em;
+ struct rb_node *node;
+
+ while (!RB_EMPTY_ROOT(&em_tree->map)) {
+ node = rb_first(&em_tree->map);
+ em = rb_entry(node, struct extent_map, rb_node);
+ remove_extent_mapping(em_tree, em);
+
+#ifdef CONFIG_BTRFS_DEBUG
+ if (refcount_read(&em->refs) != 1) {
+ test_msg(
+ "Oops we have a em leak: em (start 0x%llx len 0x%llx block_start 0x%llx block_len 0x%llx) refs %d\n",
+ em->start, em->len, em->block_start,
+ em->block_len, refcount_read(&em->refs));
+
+ refcount_set(&em->refs, 1);
+ }
+#endif
+ free_extent_map(em);
+ }
+}
+
+/*
+ * Test scenario:
+ *
+ * Suppose that no extent map has been loaded into memory yet, there is a file
+ * extent [0, 16K), followed by another file extent [16K, 20K), two dio reads
+ * are entering btrfs_get_extent() concurrently, t1 is reading [8K, 16K), t2 is
+ * reading [0, 8K)
+ *
+ * t1 t2
+ * btrfs_get_extent() btrfs_get_extent()
+ * -> lookup_extent_mapping() ->lookup_extent_mapping()
+ * -> add_extent_mapping(0, 16K)
+ * -> return em
+ * ->add_extent_mapping(0, 16K)
+ * -> #handle -EEXIST
+ */
+static void test_case_1(struct extent_map_tree *em_tree)
+{
+ struct extent_map *em;
+ u64 start = 0;
+ u64 len = SZ_8K;
+ int ret;
+
+ em = alloc_extent_map();
+ if (!em)
+ /* Skip the test on error. */
+ return;
+
+ /* Add [0, 16K) */
+ em->start = 0;
+ em->len = SZ_16K;
+ em->block_start = 0;
+ em->block_len = SZ_16K;
+ ret = add_extent_mapping(em_tree, em, 0);
+ ASSERT(ret == 0);
+ free_extent_map(em);
+
+ /* Add [16K, 20K) following [0, 16K) */
+ em = alloc_extent_map();
+ if (!em)
+ goto out;
+
+ em->start = SZ_16K;
+ em->len = SZ_4K;
+ em->block_start = SZ_32K; /* avoid merging */
+ em->block_len = SZ_4K;
+ ret = add_extent_mapping(em_tree, em, 0);
+ ASSERT(ret == 0);
+ free_extent_map(em);
+
+ em = alloc_extent_map();
+ if (!em)
+ goto out;
+
+ /* Add [0, 8K), should return [0, 16K) instead. */
+ em->start = start;
+ em->len = len;
+ em->block_start = start;
+ em->block_len = len;
+ ret = btrfs_add_extent_mapping(em_tree, &em, em->start, em->len);
+ if (ret)
+ test_msg("case1 [%llu %llu]: ret %d\n",
+ start, start + len, ret);
+ if (em &&
+ (em->start != 0 || extent_map_end(em) != SZ_16K ||
+ em->block_start != 0 || em->block_len != SZ_16K))
+ test_msg("case1 [%llu %llu]: ret %d return a wrong em (start %llu len %llu block_start %llu block_len %llu\n",
+ start, start + len, ret, em->start, em->len,
+ em->block_start, em->block_len);
+ free_extent_map(em);
+out:
+ /* free memory */
+ free_extent_map_tree(em_tree);
+}
+
+/*
+ * Test scenario:
+ *
+ * Reading the inline ending up with EEXIST, ie. read an inline
+ * extent and discard page cache and read it again.
+ */
+static void test_case_2(struct extent_map_tree *em_tree)
+{
+ struct extent_map *em;
+ int ret;
+
+ em = alloc_extent_map();
+ if (!em)
+ /* Skip the test on error. */
+ return;
+
+ /* Add [0, 1K) */
+ em->start = 0;
+ em->len = SZ_1K;
+ em->block_start = EXTENT_MAP_INLINE;
+ em->block_len = (u64)-1;
+ ret = add_extent_mapping(em_tree, em, 0);
+ ASSERT(ret == 0);
+ free_extent_map(em);
+
+ /* Add [4K, 4K) following [0, 1K) */
+ em = alloc_extent_map();
+ if (!em)
+ goto out;
+
+ em->start = SZ_4K;
+ em->len = SZ_4K;
+ em->block_start = SZ_4K;
+ em->block_len = SZ_4K;
+ ret = add_extent_mapping(em_tree, em, 0);
+ ASSERT(ret == 0);
+ free_extent_map(em);
+
+ em = alloc_extent_map();
+ if (!em)
+ goto out;
+
+ /* Add [0, 1K) */
+ em->start = 0;
+ em->len = SZ_1K;
+ em->block_start = EXTENT_MAP_INLINE;
+ em->block_len = (u64)-1;
+ ret = btrfs_add_extent_mapping(em_tree, &em, em->start, em->len);
+ if (ret)
+ test_msg("case2 [0 1K]: ret %d\n", ret);
+ if (em &&
+ (em->start != 0 || extent_map_end(em) != SZ_1K ||
+ em->block_start != EXTENT_MAP_INLINE || em->block_len != (u64)-1))
+ test_msg("case2 [0 1K]: ret %d return a wrong em (start %llu len %llu block_start %llu block_len %llu\n",
+ ret, em->start, em->len, em->block_start,
+ em->block_len);
+ free_extent_map(em);
+out:
+ /* free memory */
+ free_extent_map_tree(em_tree);
+}
+
+int btrfs_test_extent_map()
+{
+ struct extent_map_tree *em_tree;
+
+ test_msg("Running extent_map tests\n");
+
+ em_tree = kzalloc(sizeof(*em_tree), GFP_KERNEL);
+ if (!em_tree)
+ /* Skip the test on error. */
+ return 0;
+
+ extent_map_tree_init(em_tree);
+
+ test_case_1(em_tree);
+ test_case_2(em_tree);
+
+ kfree(em_tree);
+ return 0;
+}
--
2.9.4
next prev parent reply other threads:[~2018-01-05 20:54 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-01-05 19:51 [PATCH v2 00/10] bugfixes and regression tests of btrfs_get_extent Liu Bo
2018-01-05 19:51 ` [PATCH v2 01/10] Btrfs: fix incorrect block_len in merge_extent_mapping Liu Bo
2018-01-09 17:24 ` Josef Bacik
2018-01-05 19:51 ` [PATCH v2 02/10] Btrfs: fix unexpected EEXIST from btrfs_get_extent Liu Bo
2018-01-09 17:27 ` Josef Bacik
2018-01-05 19:51 ` [PATCH v2 03/10] Btrfs: add helper for em merge logic Liu Bo
2018-01-09 17:27 ` Josef Bacik
2018-01-05 19:51 ` [PATCH v2 04/10] Btrfs: move extent map specific code to extent_map.c Liu Bo
2018-01-09 17:29 ` Josef Bacik
2018-01-05 19:51 ` Liu Bo [this message]
2018-01-09 17:31 ` [PATCH v2 05/10] Btrfs: add extent map selftests Josef Bacik
2018-01-05 19:51 ` [PATCH v2 06/10] Btrfs: extent map selftest: buffered write vs dio read Liu Bo
2018-01-09 17:32 ` Josef Bacik
2018-01-05 19:51 ` [PATCH v2 07/10] Btrfs: extent map selftest: dio " Liu Bo
2018-01-09 17:32 ` Josef Bacik
2018-01-05 19:51 ` [PATCH v2 08/10] Btrfs: add WARN_ONCE to detect unexpected error from merge_extent_mapping Liu Bo
2018-01-09 17:33 ` Josef Bacik
2018-01-05 19:51 ` [PATCH v2 09/10] Btrfs: add tracepoint for em's EEXIST case Liu Bo
2018-01-09 17:35 ` Josef Bacik
2018-01-19 18:15 ` David Sterba
2018-01-19 18:22 ` Nikolay Borisov
2018-01-19 23:32 ` David Sterba
2018-01-05 19:51 ` [PATCH v2 10/10] Btrfs: noinline merge_extent_mapping Liu Bo
2018-01-09 17:35 ` Josef Bacik
2018-01-08 19:57 ` [PATCH v2 00/10] bugfixes and regression tests of btrfs_get_extent David Sterba
2018-01-18 16:51 ` David Sterba
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180105195117.5131-6-bo.li.liu@oracle.com \
--to=bo.li.liu@oracle.com \
--cc=linux-btrfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).