[PATCH v2 0/5] More unit test for mballoc

[PATCH v2 0/5] More unit test for mballoc

[Kemeng Shi]

This series covers more function to mark on-disk bitmap. Besides, some
code which is relevant to buddy cache is also tested.
Before more work is done, I want to be sure I'm not on a wrong
direction!

v1->v2:
-Fix unused variable warning which is reported at
https://lore.kernel.org/lkml/202311260042.kMxL6DnL-lkp@intel.com/T/

Kunit test result is as following:
# ./tools/testing/kunit/kunit.py run --kunitconfig=fs/ext4/.kunitconfig --raw_output
[18:39:42] Configuring KUnit Kernel ...
Regenerating .config ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
[18:39:45] Building KUnit Kernel ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
Building with:
$ make ARCH=um O=.kunit --jobs=88
[18:39:57] Starting KUnit Kernel (1/1)...
KTAP version 1
1..2
    KTAP version 1
    # Subtest: ext4_mballoc_test
    # module: ext4
    1..6
        KTAP version 1
        # Subtest: test_new_blocks_simple
        ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
    # test_new_blocks_simple: pass:3 fail:0 skip:0 total:3
    ok 1 test_new_blocks_simple
        KTAP version 1
        # Subtest: test_free_blocks_simple
        ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
    # test_free_blocks_simple: pass:3 fail:0 skip:0 total:3
    ok 2 test_free_blocks_simple
        KTAP version 1
        # Subtest: test_mb_generate_buddy
        ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
    # test_mb_generate_buddy: pass:3 fail:0 skip:0 total:3
    ok 3 test_mb_generate_buddy
        KTAP version 1
        # Subtest: test_mb_mark_used
        ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
 # SKIP blocksize exceeds pagesize
    # test_mb_mark_used: pass:2 fail:0 skip:1 total:3
    ok 4 test_mb_mark_used
        KTAP version 1
        # Subtest: test_mb_free_blocks
        ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
 # SKIP blocksize exceeds pagesize
    # test_mb_free_blocks: pass:2 fail:0 skip:1 total:3
    ok 5 test_mb_free_blocks
        KTAP version 1
        # Subtest: test_mark_diskspace_used
        ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
        ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
    # test_mark_diskspace_used: pass:3 fail:0 skip:0 total:3
    ok 6 test_mark_diskspace_used
# ext4_mballoc_test: pass:6 fail:0 skip:0 total:6
# Totals: pass:16 fail:0 skip:2 total:18
ok 1 ext4_mballoc_test
    KTAP version 1
    # Subtest: ext4_inode_test
    # module: ext4_inode_test
    1..1
        KTAP version 1
        # Subtest: inode_test_xtimestamp_decoding
        ok 1 1901-12-13 Lower bound of 32bit < 0 timestamp, no extra bits
        ok 2 1969-12-31 Upper bound of 32bit < 0 timestamp, no extra bits
        ok 3 1970-01-01 Lower bound of 32bit >=0 timestamp, no extra bits
        ok 4 2038-01-19 Upper bound of 32bit >=0 timestamp, no extra bits
        ok 5 2038-01-19 Lower bound of 32bit <0 timestamp, lo extra sec bit on
        ok 6 2106-02-07 Upper bound of 32bit <0 timestamp, lo extra sec bit on
        ok 7 2106-02-07 Lower bound of 32bit >=0 timestamp, lo extra sec bit on
        ok 8 2174-02-25 Upper bound of 32bit >=0 timestamp, lo extra sec bit on
        ok 9 2174-02-25 Lower bound of 32bit <0 timestamp, hi extra sec bit on
        ok 10 2242-03-16 Upper bound of 32bit <0 timestamp, hi extra sec bit on
        ok 11 2242-03-16 Lower bound of 32bit >=0 timestamp, hi extra sec bit on
        ok 12 2310-04-04 Upper bound of 32bit >=0 timestamp, hi extra sec bit on
        ok 13 2310-04-04 Upper bound of 32bit>=0 timestamp, hi extra sec bit 1. 1 ns
        ok 14 2378-04-22 Lower bound of 32bit>= timestamp. Extra sec bits 1. Max ns
        ok 15 2378-04-22 Lower bound of 32bit >=0 timestamp. All extra sec bits on
        ok 16 2446-05-10 Upper bound of 32bit >=0 timestamp. All extra sec bits on
    # inode_test_xtimestamp_decoding: pass:16 fail:0 skip:0 total:16
    ok 1 inode_test_xtimestamp_decoding
# Totals: pass:16 fail:0 skip:0 total:16
ok 2 ext4_inode_test
[18:39:57] Elapsed time: 14.400s total, 2.940s configuring, 11.343s building, 0.074s running

Kemeng Shi (5):
  ext4: Add unit test for test_free_blocks_simple
  ext4: Add unit test of ext4_mb_generate_buddy
  ext4: Add unit test for mb_mark_used
  ext4: Add unit test for mb_free_blocks
  ext4: Add unit test for ext4_mb_mark_diskspace_used

 fs/ext4/mballoc-test.c | 502 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 502 insertions(+)

-- 
2.30.0

[PATCH v2 1/5] ext4: Add unit test for test_free_blocks_simple

[Kemeng Shi]

Add unit test for test_free_blocks_simple.

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 fs/ext4/mballoc-test.c | 96 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)

diff --git a/fs/ext4/mballoc-test.c b/fs/ext4/mballoc-test.c
index f94901fd3..3aac42ea6 100644
--- a/fs/ext4/mballoc-test.c
+++ b/fs/ext4/mballoc-test.c
@@ -5,6 +5,7 @@
 
 #include <kunit/test.h>
 #include <kunit/static_stub.h>
+#include <linux/random.h>
 
 #include "ext4.h"
 
@@ -112,6 +113,13 @@ static void mbt_ctx_mark_used(struct super_block *sb, ext4_group_t group,
 	mb_set_bits(grp_ctx->bitmap_bh.b_data, start, len);
 }
 
+static void *mbt_ctx_bitmap(struct super_block *sb, ext4_group_t group)
+{
+	struct mbt_grp_ctx *grp_ctx = MBT_GRP_CTX(sb, group);
+
+	return grp_ctx->bitmap_bh.b_data;
+}
+
 /* called after mbt_init_sb_layout */
 static int mbt_ctx_init(struct super_block *sb)
 {
@@ -297,6 +305,93 @@ static void test_new_blocks_simple(struct kunit *test)
 		"unexpectedly get block when no block is available");
 }
 
+#define TEST_RANGE_COUNT 8
+
+struct test_range {
+	ext4_grpblk_t start;
+	ext4_grpblk_t len;
+};
+
+static void
+mbt_generate_test_ranges(struct super_block *sb, struct test_range *ranges,
+			 int count)
+{
+	ext4_grpblk_t start, len, max;
+	int i;
+
+	max = EXT4_CLUSTERS_PER_GROUP(sb) / count;
+	for (i = 0; i < count; i++) {
+		start = get_random_u32() % max;
+		len = get_random_u32() % max;
+		len = min(len, max - start);
+
+		ranges[i].start = start + i * max;
+		ranges[i].len = len;
+	}
+}
+
+static void
+validate_free_blocks_simple(struct kunit *test, struct super_block *sb,
+			    ext4_group_t goal_group, ext4_grpblk_t start,
+			    ext4_grpblk_t len)
+{
+	void *bitmap;
+	ext4_grpblk_t bit, max = EXT4_CLUSTERS_PER_GROUP(sb);
+	ext4_group_t i;
+
+	for (i = 0; i < ext4_get_groups_count(sb); i++) {
+		if (i == goal_group)
+			continue;
+
+		bitmap = mbt_ctx_bitmap(sb, i);
+		bit = mb_find_next_zero_bit(bitmap, max, 0);
+		KUNIT_ASSERT_EQ_MSG(test, bit, max,
+				    "free block on unexpected group %d", i);
+	}
+
+	bitmap = mbt_ctx_bitmap(sb, goal_group);
+	bit = mb_find_next_zero_bit(bitmap, max, 0);
+	KUNIT_ASSERT_EQ(test, bit, start);
+
+	bit = mb_find_next_bit(bitmap, max, bit + 1);
+	KUNIT_ASSERT_EQ(test, bit, start + len);
+}
+
+static void
+test_free_blocks_simple_range(struct kunit *test, ext4_group_t goal_group,
+			      ext4_grpblk_t start, ext4_grpblk_t len)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+	struct inode inode = { .i_sb = sb, };
+	ext4_fsblk_t block;
+
+	if (len == 0)
+		return;
+
+	block = ext4_group_first_block_no(sb, goal_group) +
+		EXT4_C2B(sbi, start);
+	ext4_free_blocks_simple(&inode, block, len);
+	validate_free_blocks_simple(test, sb, goal_group, start, len);
+	mbt_ctx_mark_used(sb, goal_group, 0, EXT4_CLUSTERS_PER_GROUP(sb));
+}
+
+static void test_free_blocks_simple(struct kunit *test)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	ext4_grpblk_t max = EXT4_CLUSTERS_PER_GROUP(sb);
+	ext4_group_t i;
+	struct test_range ranges[TEST_RANGE_COUNT];
+
+	for (i = 0; i < ext4_get_groups_count(sb); i++)
+		mbt_ctx_mark_used(sb, i, 0, max);
+
+	mbt_generate_test_ranges(sb, ranges, TEST_RANGE_COUNT);
+	for (i = 0; i < TEST_RANGE_COUNT; i++)
+		test_free_blocks_simple_range(test, TEST_GOAL_GROUP,
+			ranges[i].start, ranges[i].len);
+}
+
 static const struct mbt_ext4_block_layout mbt_test_layouts[] = {
 	{
 		.blocksize_bits = 10,
@@ -334,6 +429,7 @@ KUNIT_ARRAY_PARAM(mbt_layouts, mbt_test_layouts, mbt_show_layout);
 
 static struct kunit_case mbt_test_cases[] = {
 	KUNIT_CASE_PARAM(test_new_blocks_simple, mbt_layouts_gen_params),
+	KUNIT_CASE_PARAM(test_free_blocks_simple, mbt_layouts_gen_params),
 	{}
 };
 
-- 
2.30.0

[PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Kemeng Shi]

Add unit test of ext4_mb_generate_buddy

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 fs/ext4/mballoc-test.c | 207 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 207 insertions(+)

diff --git a/fs/ext4/mballoc-test.c b/fs/ext4/mballoc-test.c
index 3aac42ea6..6964974fa 100644
--- a/fs/ext4/mballoc-test.c
+++ b/fs/ext4/mballoc-test.c
@@ -28,6 +28,50 @@ struct mbt_ext4_super_block {
 #define MBT_CTX(_sb) (&(container_of((_sb), struct mbt_ext4_super_block, sb)->mbt_ctx))
 #define MBT_GRP_CTX(_sb, _group) (&MBT_CTX(_sb)->grp_ctx[_group])
 
+static const struct super_operations mbt_sops = {
+};
+
+static int mbt_mb_init(struct super_block *sb)
+{
+	int ret;
+
+	/* needed by ext4_mb_init->bdev_nonrot(sb->s_bdev) */
+	sb->s_bdev = kzalloc(sizeof(*sb->s_bdev), GFP_KERNEL);
+	if (sb->s_bdev == NULL)
+		return -ENOMEM;
+
+	sb->s_bdev->bd_queue = kzalloc(sizeof(struct request_queue), GFP_KERNEL);
+	if (sb->s_bdev->bd_queue == NULL) {
+		kfree(sb->s_bdev);
+		return -ENOMEM;
+	}
+
+	/*
+	 * needed by ext4_mb_init->ext4_mb_init_backend-> sbi->s_buddy_cache =
+	 * new_inode(sb);
+	 */
+	INIT_LIST_HEAD(&sb->s_inodes);
+	sb->s_op = &mbt_sops;
+
+	ret = ext4_mb_init(sb);
+	if (ret != 0)
+		goto err_out;
+
+	return 0;
+
+err_out:
+	kfree(sb->s_bdev->bd_queue);
+	kfree(sb->s_bdev);
+	return ret;
+}
+
+static void mbt_mb_release(struct super_block *sb)
+{
+	ext4_mb_release(sb);
+	kfree(sb->s_bdev->bd_queue);
+	kfree(sb->s_bdev);
+}
+
 static struct super_block *mbt_ext4_alloc_super_block(void)
 {
 	struct ext4_super_block *es = kzalloc(sizeof(*es), GFP_KERNEL);
@@ -37,8 +81,16 @@ static struct super_block *mbt_ext4_alloc_super_block(void)
 	if (fsb == NULL || sbi == NULL || es == NULL)
 		goto out;
 
+	sbi->s_blockgroup_lock =
+		kzalloc(sizeof(struct blockgroup_lock), GFP_KERNEL);
+	if (!sbi->s_blockgroup_lock)
+		goto out;
+
+	bgl_lock_init(sbi->s_blockgroup_lock);
+
 	sbi->s_es = es;
 	fsb->sb.s_fs_info = sbi;
+
 	return &fsb->sb;
 
 out:
@@ -54,6 +106,7 @@ static void mbt_ext4_free_super_block(struct super_block *sb)
 		container_of(sb, struct mbt_ext4_super_block, sb);
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 
+	kfree(sbi->s_blockgroup_lock);
 	kfree(sbi->s_es);
 	kfree(sbi);
 	kfree(fsb);
@@ -83,6 +136,9 @@ static void mbt_init_sb_layout(struct super_block *sb,
 	sbi->s_clusters_per_group = layout->blocks_per_group >>
 				    layout->cluster_bits;
 	sbi->s_desc_size = layout->desc_size;
+	sbi->s_desc_per_block_bits =
+		sb->s_blocksize_bits - (fls(layout->desc_size) - 1);
+	sbi->s_desc_per_block = 1 << sbi->s_desc_per_block_bits;
 
 	es->s_first_data_block = cpu_to_le32(0);
 	es->s_blocks_count_lo = cpu_to_le32(layout->blocks_per_group *
@@ -240,6 +296,14 @@ static int mbt_kunit_init(struct kunit *test)
 	kunit_activate_static_stub(test,
 				   ext4_mb_mark_context,
 				   ext4_mb_mark_context_stub);
+
+	/* stub function will be called in mt_mb_init->ext4_mb_init */
+	if (mbt_mb_init(sb) != 0) {
+		mbt_ctx_release(sb);
+		mbt_ext4_free_super_block(sb);
+		return -ENOMEM;
+	}
+
 	return 0;
 }
 
@@ -247,6 +311,7 @@ static void mbt_kunit_exit(struct kunit *test)
 {
 	struct super_block *sb = (struct super_block *)test->priv;
 
+	mbt_mb_release(sb);
 	mbt_ctx_release(sb);
 	mbt_ext4_free_super_block(sb);
 }
@@ -392,6 +457,147 @@ static void test_free_blocks_simple(struct kunit *test)
 			ranges[i].start, ranges[i].len);
 }
 
+static void mbt_generate_buddy(struct super_block *sb, void *buddy,
+			       void *bitmap, struct ext4_group_info *grp)
+{
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+	uint32_t order, off;
+	void *bb, *bb_h;
+	int max;
+
+	memset(buddy, 0xff, sb->s_blocksize);
+	memset(grp, 0, offsetof(struct ext4_group_info,
+				 bb_counters[MB_NUM_ORDERS(sb)]));
+
+	bb = bitmap;
+	max = EXT4_CLUSTERS_PER_GROUP(sb);
+	bb_h = buddy + sbi->s_mb_offsets[1];
+
+	off = mb_find_next_zero_bit(bb, max, 0);
+	grp->bb_first_free = off;
+	while (off < max) {
+		grp->bb_counters[0]++;
+		grp->bb_free++;
+
+		if (!(off & 1) && !mb_test_bit(off + 1, bb)) {
+			grp->bb_free++;
+			grp->bb_counters[0]--;
+			mb_clear_bit(off >> 1, bb_h);
+			grp->bb_counters[1]++;
+			grp->bb_largest_free_order = 1;
+			off++;
+		}
+
+		off = mb_find_next_zero_bit(bb, max, off + 1);
+	}
+
+	for (order = 1; order < MB_NUM_ORDERS(sb) - 1; order++) {
+		bb = buddy + sbi->s_mb_offsets[order];
+		bb_h = buddy + sbi->s_mb_offsets[order + 1];
+		max = max >> 1;
+		off = mb_find_next_zero_bit(bb, max, 0);
+
+		while (off < max) {
+			if (!(off & 1) && !mb_test_bit(off + 1, bb)) {
+				mb_set_bits(bb, off, 2);
+				grp->bb_counters[order] -= 2;
+				mb_clear_bit(off >> 1, bb_h);
+				grp->bb_counters[order + 1]++;
+				grp->bb_largest_free_order = order + 1;
+				off++;
+			}
+
+			off = mb_find_next_zero_bit(bb, max, off + 1);
+		}
+	}
+
+	max = EXT4_CLUSTERS_PER_GROUP(sb);
+	off = mb_find_next_zero_bit(bitmap, max, 0);
+	while (off < max) {
+		grp->bb_fragments++;
+
+		off = mb_find_next_bit(bitmap, max, off + 1);
+		if (off + 1 >= max)
+			break;
+
+		off = mb_find_next_zero_bit(bitmap, max, off + 1);
+	}
+}
+
+static void
+mbt_validate_group_info(struct kunit *test, struct ext4_group_info *grp1,
+			struct ext4_group_info *grp2)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	int i;
+
+	KUNIT_ASSERT_EQ(test, grp1->bb_first_free,
+			grp2->bb_first_free);
+	KUNIT_ASSERT_EQ(test, grp1->bb_fragments,
+			grp2->bb_fragments);
+	KUNIT_ASSERT_EQ(test, grp1->bb_free, grp2->bb_free);
+	KUNIT_ASSERT_EQ(test, grp1->bb_largest_free_order,
+			grp2->bb_largest_free_order);
+
+	for (i = 1; i < MB_NUM_ORDERS(sb); i++) {
+		KUNIT_ASSERT_EQ_MSG(test, grp1->bb_counters[i],
+				    grp2->bb_counters[i],
+				    "bb_counters[%d] diffs, expected %d, generated %d",
+				    i, grp1->bb_counters[i],
+				    grp2->bb_counters[i]);
+	}
+}
+
+static void
+do_test_generate_buddy(struct kunit *test, struct super_block *sb, void *bitmap,
+			   void *mbt_buddy, struct ext4_group_info *mbt_grp,
+			   void *ext4_buddy, struct ext4_group_info *ext4_grp)
+{
+	int i;
+
+	mbt_generate_buddy(sb, mbt_buddy, bitmap, mbt_grp);
+
+	for (i = 0; i < MB_NUM_ORDERS(sb); i++)
+		ext4_grp->bb_counters[i] = 0;
+	/* needed by validation in ext4_mb_generate_buddy */
+	ext4_grp->bb_free = mbt_grp->bb_free;
+	memset(ext4_buddy, 0xff, sb->s_blocksize);
+	ext4_mb_generate_buddy(sb, ext4_buddy, bitmap, TEST_GOAL_GROUP,
+			       ext4_grp);
+
+	KUNIT_ASSERT_EQ(test, memcmp(mbt_buddy, ext4_buddy, sb->s_blocksize),
+			0);
+	mbt_validate_group_info(test, mbt_grp, ext4_grp);
+}
+
+static void test_mb_generate_buddy(struct kunit *test)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	void *bitmap, *expected_bb, *generate_bb;
+	struct ext4_group_info *expected_grp, *generate_grp;
+	struct test_range ranges[TEST_RANGE_COUNT];
+	int i;
+
+	bitmap = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, bitmap);
+	expected_bb = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, expected_bb);
+	generate_bb = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, generate_bb);
+	expected_grp = kunit_kzalloc(test, offsetof(struct ext4_group_info,
+				bb_counters[MB_NUM_ORDERS(sb)]), GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, expected_grp);
+	generate_grp = ext4_get_group_info(sb, TEST_GOAL_GROUP);
+	KUNIT_ASSERT_NOT_NULL(test, generate_grp);
+
+	mbt_generate_test_ranges(sb, ranges, TEST_RANGE_COUNT);
+	for (i = 0; i < TEST_RANGE_COUNT; i++) {
+		mb_set_bits(bitmap, ranges[i].start, ranges[i].len);
+		do_test_generate_buddy(test, sb, bitmap, expected_bb,
+				       expected_grp, generate_bb, generate_grp);
+	}
+}
+
 static const struct mbt_ext4_block_layout mbt_test_layouts[] = {
 	{
 		.blocksize_bits = 10,
@@ -430,6 +636,7 @@ KUNIT_ARRAY_PARAM(mbt_layouts, mbt_test_layouts, mbt_show_layout);
 static struct kunit_case mbt_test_cases[] = {
 	KUNIT_CASE_PARAM(test_new_blocks_simple, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_free_blocks_simple, mbt_layouts_gen_params),
+	KUNIT_CASE_PARAM(test_mb_generate_buddy, mbt_layouts_gen_params),
 	{}
 };
 
-- 
2.30.0

Re: [PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Guenter Roeck]

Hi,

On Wed, Jan 03, 2024 at 06:48:57PM +0800, Kemeng Shi wrote:
> Add unit test of ext4_mb_generate_buddy
> 
> Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>

With this and other new ext4 tests test in the tree, I see a variety
of backtraces in the upstream kernel if debug options are enabled.
An example is

[    6.821447]         KTAP version 1
[    6.821769]         # Subtest: test_mb_generate_buddy
[    6.824787] =============================================================================
[    6.825568] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
...
[    6.894341] ok 7 ext4_inode_test
[    6.895411] =============================================================================
[    6.895777] BUG inode_cache (Tainted: G    B            N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232

Another example, from another test run, is

[    3.938551]         # Subtest: test_new_blocks_simple
[    3.947171]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
[    3.952988]         ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
[    3.958403]         ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
[    3.958890] =============================================================================
[    3.959159] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xffff8de881adbf68-0xffff8de881adbf6f @offset=16232

Another one:

[   18.730473]         # Subtest: test_new_blocks_simple
[   18.760547]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
[   18.778477] ==================================================================
[   18.778950] BUG: KFENCE: out-of-bounds write in ext4_mb_init+0x5d7/0xa60

This is just a sample, taken from a quick look at test results.

Are those backtraces expected ? If so, would it be possible to execute the
tests without generating such backtraces ? The backtraces, if intentional,
hide real problems in the noise.

Thanks,
Guenter

Re: [PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Guenter Roeck]

[ Adding more interested parties]

On Wed, Mar 20, 2024 at 09:23:24AM -0700, Guenter Roeck wrote:
> Hi,
> 
> On Wed, Jan 03, 2024 at 06:48:57PM +0800, Kemeng Shi wrote:
> > Add unit test of ext4_mb_generate_buddy
> > 
> > Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
> 
> With this and other new ext4 tests test in the tree, I see a variety
> of backtraces in the upstream kernel if debug options are enabled.
> An example is
> 
> [    6.821447]         KTAP version 1
> [    6.821769]         # Subtest: test_mb_generate_buddy
> [    6.824787] =============================================================================
> [    6.825568] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
> ...
> [    6.894341] ok 7 ext4_inode_test
> [    6.895411] =============================================================================
> [    6.895777] BUG inode_cache (Tainted: G    B            N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
> 
> Another example, from another test run, is
> 
> [    3.938551]         # Subtest: test_new_blocks_simple
> [    3.947171]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [    3.952988]         ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [    3.958403]         ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [    3.958890] =============================================================================
> [    3.959159] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xffff8de881adbf68-0xffff8de881adbf6f @offset=16232
> 
> Another one:
> 
> [   18.730473]         # Subtest: test_new_blocks_simple
> [   18.760547]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [   18.778477] ==================================================================
> [   18.778950] BUG: KFENCE: out-of-bounds write in ext4_mb_init+0x5d7/0xa60
> 
> This is just a sample, taken from a quick look at test results.
> 
> Are those backtraces expected ? If so, would it be possible to execute the
> tests without generating such backtraces ? The backtraces, if intentional,
> hide real problems in the noise.
> 
> Thanks,
> Guenter

Re: [PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Kemeng Shi]

on 3/21/2024 12:23 AM, Guenter Roeck wrote:
> Hi,
> 
> On Wed, Jan 03, 2024 at 06:48:57PM +0800, Kemeng Shi wrote:
>> Add unit test of ext4_mb_generate_buddy
>>
>> Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
> 
> With this and other new ext4 tests test in the tree, I see a variety
> of backtraces in the upstream kernel if debug options are enabled.
> An example is
> 
> [    6.821447]         KTAP version 1
> [    6.821769]         # Subtest: test_mb_generate_buddy
> [    6.824787] =============================================================================
> [    6.825568] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
> ...
> [    6.894341] ok 7 ext4_inode_test
> [    6.895411] =============================================================================
> [    6.895777] BUG inode_cache (Tainted: G    B            N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
> 
> Another example, from another test run, is
> 
> [    3.938551]         # Subtest: test_new_blocks_simple
> [    3.947171]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [    3.952988]         ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [    3.958403]         ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [    3.958890] =============================================================================
> [    3.959159] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xffff8de881adbf68-0xffff8de881adbf6f @offset=16232
> 
> Another one:
> 
> [   18.730473]         # Subtest: test_new_blocks_simple
> [   18.760547]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
> [   18.778477] ==================================================================
> [   18.778950] BUG: KFENCE: out-of-bounds write in ext4_mb_init+0x5d7/0xa60
> 
> This is just a sample, taken from a quick look at test results.
> 
> Are those backtraces expected ? If so, would it be possible to execute the
> tests without generating such backtraces ? The backtraces, if intentional,
> hide real problems in the noise.
Thanks for the report. The backtrace is not expected, I will look into this. Thansk!
> 
> Thanks,
> Guenter
> 

Re: [PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Kemeng Shi]

on 3/21/2024 3:16 PM, Kemeng Shi wrote:
> 
> 
> on 3/21/2024 12:23 AM, Guenter Roeck wrote:
>> Hi,
>>
>> On Wed, Jan 03, 2024 at 06:48:57PM +0800, Kemeng Shi wrote:
>>> Add unit test of ext4_mb_generate_buddy
>>>
>>> Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
>>
>> With this and other new ext4 tests test in the tree, I see a variety
>> of backtraces in the upstream kernel if debug options are enabled.
>> An example is
>>
>> [    6.821447]         KTAP version 1
>> [    6.821769]         # Subtest: test_mb_generate_buddy
>> [    6.824787] =============================================================================
>> [    6.825568] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
>> ...
>> [    6.894341] ok 7 ext4_inode_test
>> [    6.895411] =============================================================================
>> [    6.895777] BUG inode_cache (Tainted: G    B            N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
>>
>> Another example, from another test run, is
>>
>> [    3.938551]         # Subtest: test_new_blocks_simple
>> [    3.947171]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>> [    3.952988]         ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>> [    3.958403]         ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>> [    3.958890] =============================================================================
>> [    3.959159] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xffff8de881adbf68-0xffff8de881adbf6f @offset=16232
>>
>> Another one:
>>
>> [   18.730473]         # Subtest: test_new_blocks_simple
>> [   18.760547]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>> [   18.778477] ==================================================================
>> [   18.778950] BUG: KFENCE: out-of-bounds write in ext4_mb_init+0x5d7/0xa60
>>
>> This is just a sample, taken from a quick look at test results.
>>
>> Are those backtraces expected ? If so, would it be possible to execute the
>> tests without generating such backtraces ? The backtraces, if intentional,
>> hide real problems in the noise.
> Thanks for the report. The backtrace is not expected, I will look into this. Thansk!
>>
Hi Guenter, I could not reproduce this in my local vm. From the reported backtraces, it's
likely there is a out-of-bounds write to sbi->s_buddy_cache. I try to fix this in [1] and
it works fine in my local vm. I wish this work for you to elimate the reported nosie.
Look forward to your reply, Thanks.

Kemeng

[1] https://lore.kernel.org/linux-ext4/20240322165518.8147-1-shikemeng@huaweicloud.com/T/#u

>> Thanks,
>> Guenter
>>

Re: [PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Guenter Roeck]

On 3/22/24 02:27, Kemeng Shi wrote:
> 
> 
> on 3/21/2024 3:16 PM, Kemeng Shi wrote:
>>
>>
>> on 3/21/2024 12:23 AM, Guenter Roeck wrote:
>>> Hi,
>>>
>>> On Wed, Jan 03, 2024 at 06:48:57PM +0800, Kemeng Shi wrote:
>>>> Add unit test of ext4_mb_generate_buddy
>>>>
>>>> Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
>>>
>>> With this and other new ext4 tests test in the tree, I see a variety
>>> of backtraces in the upstream kernel if debug options are enabled.
>>> An example is
>>>
>>> [    6.821447]         KTAP version 1
>>> [    6.821769]         # Subtest: test_mb_generate_buddy
>>> [    6.824787] =============================================================================
>>> [    6.825568] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
>>> ...
>>> [    6.894341] ok 7 ext4_inode_test
>>> [    6.895411] =============================================================================
>>> [    6.895777] BUG inode_cache (Tainted: G    B            N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
>>>
>>> Another example, from another test run, is
>>>
>>> [    3.938551]         # Subtest: test_new_blocks_simple
>>> [    3.947171]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>> [    3.952988]         ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>> [    3.958403]         ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>> [    3.958890] =============================================================================
>>> [    3.959159] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xffff8de881adbf68-0xffff8de881adbf6f @offset=16232
>>>
>>> Another one:
>>>
>>> [   18.730473]         # Subtest: test_new_blocks_simple
>>> [   18.760547]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>> [   18.778477] ==================================================================
>>> [   18.778950] BUG: KFENCE: out-of-bounds write in ext4_mb_init+0x5d7/0xa60
>>>
>>> This is just a sample, taken from a quick look at test results.
>>>
>>> Are those backtraces expected ? If so, would it be possible to execute the
>>> tests without generating such backtraces ? The backtraces, if intentional,
>>> hide real problems in the noise.
>> Thanks for the report. The backtrace is not expected, I will look into this. Thansk!
>>>
> Hi Guenter, I could not reproduce this in my local vm. From the reported backtraces, it's
> likely there is a out-of-bounds write to sbi->s_buddy_cache. I try to fix this in [1] and
> it works fine in my local vm. I wish this work for you to elimate the reported nosie.
> Look forward to your reply, Thanks.
> 

You would need to have CONFIG_SLUB_DEBUG=y, CONFIG_SLUB_DEBUG_ON=y, and CONFIG_KFENCE=y
to see the problems.

Guenter

Re: [PATCH v2 2/5] ext4: Add unit test of ext4_mb_generate_buddy

[Kemeng Shi]

on 3/22/2024 10:49 PM, Guenter Roeck wrote:
> On 3/22/24 02:27, Kemeng Shi wrote:
>>
>>
>> on 3/21/2024 3:16 PM, Kemeng Shi wrote:
>>>
>>>
>>> on 3/21/2024 12:23 AM, Guenter Roeck wrote:
>>>> Hi,
>>>>
>>>> On Wed, Jan 03, 2024 at 06:48:57PM +0800, Kemeng Shi wrote:
>>>>> Add unit test of ext4_mb_generate_buddy
>>>>>
>>>>> Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
>>>>
>>>> With this and other new ext4 tests test in the tree, I see a variety
>>>> of backtraces in the upstream kernel if debug options are enabled.
>>>> An example is
>>>>
>>>> [    6.821447]         KTAP version 1
>>>> [    6.821769]         # Subtest: test_mb_generate_buddy
>>>> [    6.824787] =============================================================================
>>>> [    6.825568] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
>>>> ...
>>>> [    6.894341] ok 7 ext4_inode_test
>>>> [    6.895411] =============================================================================
>>>> [    6.895777] BUG inode_cache (Tainted: G    B            N): Padding overwritten. 0xfffff80006223f68-0xfffff80006223f6f @offset=16232
>>>>
>>>> Another example, from another test run, is
>>>>
>>>> [    3.938551]         # Subtest: test_new_blocks_simple
>>>> [    3.947171]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>>> [    3.952988]         ok 2 block_bits=12 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>>> [    3.958403]         ok 3 block_bits=16 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>>> [    3.958890] =============================================================================
>>>> [    3.959159] BUG inode_cache (Tainted: G                 N): Padding overwritten. 0xffff8de881adbf68-0xffff8de881adbf6f @offset=16232
>>>>
>>>> Another one:
>>>>
>>>> [   18.730473]         # Subtest: test_new_blocks_simple
>>>> [   18.760547]         ok 1 block_bits=10 cluster_bits=3 blocks_per_group=8192 group_count=4 desc_size=64
>>>> [   18.778477] ==================================================================
>>>> [   18.778950] BUG: KFENCE: out-of-bounds write in ext4_mb_init+0x5d7/0xa60
>>>>
>>>> This is just a sample, taken from a quick look at test results.
>>>>
>>>> Are those backtraces expected ? If so, would it be possible to execute the
>>>> tests without generating such backtraces ? The backtraces, if intentional,
>>>> hide real problems in the noise.
>>> Thanks for the report. The backtrace is not expected, I will look into this. Thansk!
>>>>
>> Hi Guenter, I could not reproduce this in my local vm. From the reported backtraces, it's
>> likely there is a out-of-bounds write to sbi->s_buddy_cache. I try to fix this in [1] and
>> it works fine in my local vm. I wish this work for you to elimate the reported nosie.
>> Look forward to your reply, Thanks.
>>
> 
> You would need to have CONFIG_SLUB_DEBUG=y, CONFIG_SLUB_DEBUG_ON=y, and CONFIG_KFENCE=y
> to see the problems.
Thanks for sharing this. I have already turn these configs on, but I use the ext4 tree
https://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4.git/log/?h=dev which may be
the reason why I can't reproduce the issue.
I see you have already tested the fix in upstream. Thanks a lot for that!

Kemeng
> 
> Guenter
> 

[PATCH v2 3/5] ext4: Add unit test for mb_mark_used

[Kemeng Shi]

Add unit test for mb_mark_used

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 fs/ext4/mballoc-test.c | 78 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 78 insertions(+)

diff --git a/fs/ext4/mballoc-test.c b/fs/ext4/mballoc-test.c
index 6964974fa..38e3644cb 100644
--- a/fs/ext4/mballoc-test.c
+++ b/fs/ext4/mballoc-test.c
@@ -148,9 +148,13 @@ static void mbt_init_sb_layout(struct super_block *sb,
 static int mbt_grp_ctx_init(struct super_block *sb,
 			    struct mbt_grp_ctx *grp_ctx)
 {
+	ext4_grpblk_t max = EXT4_CLUSTERS_PER_GROUP(sb);
+
 	grp_ctx->bitmap_bh.b_data = kzalloc(EXT4_BLOCK_SIZE(sb), GFP_KERNEL);
 	if (grp_ctx->bitmap_bh.b_data == NULL)
 		return -ENOMEM;
+	mb_set_bits(grp_ctx->bitmap_bh.b_data, max, sb->s_blocksize * 8 - max);
+	ext4_free_group_clusters_set(sb, &grp_ctx->desc, max);
 
 	return 0;
 }
@@ -197,6 +201,8 @@ static int mbt_ctx_init(struct super_block *sb)
 	 * block which will fail ext4_sb_block_valid check.
 	 */
 	mb_set_bits(ctx->grp_ctx[0].bitmap_bh.b_data, 0, 1);
+	ext4_free_group_clusters_set(sb, &ctx->grp_ctx[0].desc,
+				     EXT4_CLUSTERS_PER_GROUP(sb) - 1);
 
 	return 0;
 out:
@@ -231,6 +237,13 @@ static int ext4_wait_block_bitmap_stub(struct super_block *sb,
 				       ext4_group_t block_group,
 				       struct buffer_head *bh)
 {
+	/*
+	 * real ext4_wait_block_bitmap will set these flags and
+	 * functions like ext4_mb_init_cache will verify the flags.
+	 */
+	set_buffer_uptodate(bh);
+	set_bitmap_uptodate(bh);
+	set_buffer_verified(bh);
 	return 0;
 }
 
@@ -598,6 +611,70 @@ static void test_mb_generate_buddy(struct kunit *test)
 	}
 }
 
+static void
+test_mb_mark_used_range(struct kunit *test, struct ext4_buddy *e4b,
+			ext4_grpblk_t start, ext4_grpblk_t len, void *bitmap,
+			void *buddy, struct ext4_group_info *grp)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	struct ext4_free_extent ex;
+	int i;
+
+	/* mb_mark_used only accepts non-zero len */
+	if (len == 0)
+		return;
+
+	ex.fe_start = start;
+	ex.fe_len = len;
+	ex.fe_group = TEST_GOAL_GROUP;
+	mb_mark_used(e4b, &ex);
+
+	mb_set_bits(bitmap, start, len);
+	/* bypass bb_free validatoin in ext4_mb_generate_buddy */
+	grp->bb_free -= len;
+	memset(buddy, 0xff, sb->s_blocksize);
+	for (i = 0; i < MB_NUM_ORDERS(sb); i++)
+		grp->bb_counters[i] = 0;
+	ext4_mb_generate_buddy(sb, buddy, bitmap, 0, grp);
+
+	KUNIT_ASSERT_EQ(test, memcmp(buddy, e4b->bd_buddy, sb->s_blocksize),
+			0);
+	mbt_validate_group_info(test, grp, e4b->bd_info);
+}
+
+static void test_mb_mark_used(struct kunit *test)
+{
+	struct ext4_buddy e4b;
+	struct super_block *sb = (struct super_block *)test->priv;
+	void *bitmap, *buddy;
+	struct ext4_group_info *grp;
+	int ret;
+	struct test_range ranges[TEST_RANGE_COUNT];
+	int i;
+
+	/* buddy cache assumes that each page contains at least one block */
+	if (sb->s_blocksize > PAGE_SIZE)
+		kunit_skip(test, "blocksize exceeds pagesize");
+
+	bitmap = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, bitmap);
+	buddy = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, buddy);
+	grp = kunit_kzalloc(test, offsetof(struct ext4_group_info,
+				bb_counters[MB_NUM_ORDERS(sb)]), GFP_KERNEL);
+
+	ret = ext4_mb_load_buddy(sb, TEST_GOAL_GROUP, &e4b);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	grp->bb_free = EXT4_CLUSTERS_PER_GROUP(sb);
+	mbt_generate_test_ranges(sb, ranges, TEST_RANGE_COUNT);
+	for (i = 0; i < TEST_RANGE_COUNT; i++)
+		test_mb_mark_used_range(test, &e4b, ranges[i].start,
+					ranges[i].len, bitmap, buddy, grp);
+
+	ext4_mb_unload_buddy(&e4b);
+}
+
 static const struct mbt_ext4_block_layout mbt_test_layouts[] = {
 	{
 		.blocksize_bits = 10,
@@ -637,6 +714,7 @@ static struct kunit_case mbt_test_cases[] = {
 	KUNIT_CASE_PARAM(test_new_blocks_simple, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_free_blocks_simple, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_mb_generate_buddy, mbt_layouts_gen_params),
+	KUNIT_CASE_PARAM(test_mb_mark_used, mbt_layouts_gen_params),
 	{}
 };
 
-- 
2.30.0

[PATCH v2 4/5] ext4: Add unit test for mb_free_blocks

[Kemeng Shi]

Add unit test for mb_free_blocks.

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 fs/ext4/mballoc-test.c | 69 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)

diff --git a/fs/ext4/mballoc-test.c b/fs/ext4/mballoc-test.c
index 38e3644cb..b68f44740 100644
--- a/fs/ext4/mballoc-test.c
+++ b/fs/ext4/mballoc-test.c
@@ -675,6 +675,74 @@ static void test_mb_mark_used(struct kunit *test)
 	ext4_mb_unload_buddy(&e4b);
 }
 
+static void
+test_mb_free_blocks_range(struct kunit *test, struct ext4_buddy *e4b,
+			  ext4_grpblk_t start, ext4_grpblk_t len, void *bitmap,
+			  void *buddy, struct ext4_group_info *grp)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	int i;
+
+	/* mb_free_blocks will WARN if len is 0 */
+	if (len == 0)
+		return;
+
+	mb_free_blocks(NULL, e4b, start, len);
+
+	mb_clear_bits(bitmap, start, len);
+	/* bypass bb_free validatoin in ext4_mb_generate_buddy */
+	grp->bb_free += len;
+	memset(buddy, 0xff, sb->s_blocksize);
+	for (i = 0; i < MB_NUM_ORDERS(sb); i++)
+		grp->bb_counters[i] = 0;
+	ext4_mb_generate_buddy(sb, buddy, bitmap, 0, grp);
+
+	KUNIT_ASSERT_EQ(test, memcmp(buddy, e4b->bd_buddy, sb->s_blocksize),
+			0);
+	mbt_validate_group_info(test, grp, e4b->bd_info);
+
+}
+
+static void test_mb_free_blocks(struct kunit *test)
+{
+	struct ext4_buddy e4b;
+	struct super_block *sb = (struct super_block *)test->priv;
+	void *bitmap, *buddy;
+	struct ext4_group_info *grp;
+	struct ext4_free_extent ex;
+	int ret;
+	int i;
+	struct test_range ranges[TEST_RANGE_COUNT];
+
+	/* buddy cache assumes that each page contains at least one block */
+	if (sb->s_blocksize > PAGE_SIZE)
+		kunit_skip(test, "blocksize exceeds pagesize");
+
+	bitmap = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, bitmap);
+	buddy = kunit_kzalloc(test, sb->s_blocksize, GFP_KERNEL);
+	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, buddy);
+	grp = kunit_kzalloc(test, offsetof(struct ext4_group_info,
+				bb_counters[MB_NUM_ORDERS(sb)]), GFP_KERNEL);
+
+	ret = ext4_mb_load_buddy(sb, TEST_GOAL_GROUP, &e4b);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	ex.fe_start = 0;
+	ex.fe_len = EXT4_CLUSTERS_PER_GROUP(sb);
+	ex.fe_group = TEST_GOAL_GROUP;
+	mb_mark_used(&e4b, &ex);
+	grp->bb_free = 0;
+	memset(bitmap, 0xff, sb->s_blocksize);
+
+	mbt_generate_test_ranges(sb, ranges, TEST_RANGE_COUNT);
+	for (i = 0; i < TEST_RANGE_COUNT; i++)
+		test_mb_free_blocks_range(test, &e4b, ranges[i].start,
+					  ranges[i].len, bitmap, buddy, grp);
+
+	ext4_mb_unload_buddy(&e4b);
+}
+
 static const struct mbt_ext4_block_layout mbt_test_layouts[] = {
 	{
 		.blocksize_bits = 10,
@@ -715,6 +783,7 @@ static struct kunit_case mbt_test_cases[] = {
 	KUNIT_CASE_PARAM(test_free_blocks_simple, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_mb_generate_buddy, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_mb_mark_used, mbt_layouts_gen_params),
+	KUNIT_CASE_PARAM(test_mb_free_blocks, mbt_layouts_gen_params),
 	{}
 };
 
-- 
2.30.0

[PATCH v2 5/5] ext4: Add unit test for ext4_mb_mark_diskspace_used

[Kemeng Shi]

Add unit test for ext4_mb_mark_diskspace_used

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 fs/ext4/mballoc-test.c | 52 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 52 insertions(+)

diff --git a/fs/ext4/mballoc-test.c b/fs/ext4/mballoc-test.c
index b68f44740..12d0b22ca 100644
--- a/fs/ext4/mballoc-test.c
+++ b/fs/ext4/mballoc-test.c
@@ -470,6 +470,57 @@ static void test_free_blocks_simple(struct kunit *test)
 			ranges[i].start, ranges[i].len);
 }
 
+static void
+test_mark_diskspace_used_range(struct kunit *test,
+			       struct ext4_allocation_context *ac,
+			       ext4_grpblk_t start,
+			       ext4_grpblk_t len)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	int ret;
+	void *bitmap;
+	ext4_grpblk_t i, max;
+
+	/* ext4_mb_mark_diskspace_used will BUG if len is 0 */
+	if (len == 0)
+		return;
+
+	ac->ac_b_ex.fe_group = TEST_GOAL_GROUP;
+	ac->ac_b_ex.fe_start = start;
+	ac->ac_b_ex.fe_len = len;
+
+	bitmap = mbt_ctx_bitmap(sb, TEST_GOAL_GROUP);
+	memset(bitmap, 0, sb->s_blocksize);
+	ret = ext4_mb_mark_diskspace_used(ac, NULL, 0);
+	KUNIT_ASSERT_EQ(test, ret, 0);
+
+	max = EXT4_CLUSTERS_PER_GROUP(sb);
+	i = mb_find_next_bit(bitmap, max, 0);
+	KUNIT_ASSERT_EQ(test, i, start);
+	i = mb_find_next_zero_bit(bitmap, max, i + 1);
+	KUNIT_ASSERT_EQ(test, i, start + len);
+	i = mb_find_next_bit(bitmap, max, i + 1);
+	KUNIT_ASSERT_EQ(test, max, i);
+}
+
+static void test_mark_diskspace_used(struct kunit *test)
+{
+	struct super_block *sb = (struct super_block *)test->priv;
+	struct inode inode = { .i_sb = sb, };
+	struct ext4_allocation_context ac;
+	struct test_range ranges[TEST_RANGE_COUNT];
+	int i;
+
+	mbt_generate_test_ranges(sb, ranges, TEST_RANGE_COUNT);
+
+	ac.ac_status = AC_STATUS_FOUND;
+	ac.ac_sb = sb;
+	ac.ac_inode = &inode;
+	for (i = 0; i < TEST_RANGE_COUNT; i++)
+		test_mark_diskspace_used_range(test, &ac, ranges[i].start,
+					       ranges[i].len);
+}
+
 static void mbt_generate_buddy(struct super_block *sb, void *buddy,
 			       void *bitmap, struct ext4_group_info *grp)
 {
@@ -784,6 +835,7 @@ static struct kunit_case mbt_test_cases[] = {
 	KUNIT_CASE_PARAM(test_mb_generate_buddy, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_mb_mark_used, mbt_layouts_gen_params),
 	KUNIT_CASE_PARAM(test_mb_free_blocks, mbt_layouts_gen_params),
+	KUNIT_CASE_PARAM(test_mark_diskspace_used, mbt_layouts_gen_params),
 	{}
 };
 
-- 
2.30.0

Re: [PATCH v2 0/5] More unit test for mballoc

[Theodore Ts'o]

On Wed, 03 Jan 2024 18:48:55 +0800, Kemeng Shi wrote:
> This series covers more function to mark on-disk bitmap. Besides, some
> code which is relevant to buddy cache is also tested.
> Before more work is done, I want to be sure I'm not on a wrong
> direction!
> 
> v1->v2:
> -Fix unused variable warning which is reported at
> https://lore.kernel.org/lkml/202311260042.kMxL6DnL-lkp@intel.com/T/
> 
> [...]

Applied, thanks!

[1/5] ext4: Add unit test for test_free_blocks_simple
      commit: 6c5e0c9c21456fb561d0997fe2c4d3cf59745ba7
[2/5] ext4: Add unit test of ext4_mb_generate_buddy
      commit: 67d2a11b22b4d520072db62620851615df13183e
[3/5] ext4: Add unit test for mb_mark_used
      commit: ac96b56a2fbd9e05b28488bdc5d3bd8006b61d35
[4/5] ext4: Add unit test for mb_free_blocks
      commit: b7098e1fa7bcf350e089af9500c4f9992a1bd6cb
[5/5] ext4: Add unit test for ext4_mb_mark_diskspace_used
      commit: 2b81493f8eb6fc0c263dbca0064e10e4c00e0f91

Best regards,
-- 
Theodore Ts'o <tytso@mit.edu>

Re: [PATCH v2 0/5] More unit test for mballoc

[Guenter Roeck]

On Wed, Jan 03, 2024 at 06:48:55PM +0800, Kemeng Shi wrote:
> This series covers more function to mark on-disk bitmap. Besides, some
> code which is relevant to buddy cache is also tested.
> Before more work is done, I want to be sure I'm not on a wrong
> direction!
> 
> v1->v2:
> -Fix unused variable warning which is reported at
> https://lore.kernel.org/lkml/202311260042.kMxL6DnL-lkp@intel.com/T/
> 

With this patch series in linux-text, and with various debug options
enabled, unit tests for ext4 fail widely and result in crashes.

[    4.798582]         # Subtest: test_new_blocks_simple
[    4.803166] BUG: key 00000000000000a8 has not been registered!
[    4.803443] ------------[ cut here ]------------
ILLOPC: ffffffff9050cbf4: 0f 0b
[    4.803521] DEBUG_LOCKS_WARN_ON(1)
[    4.803720] WARNING: CPU: 0 PID: 154 at kernel/locking/lockdep.c:4895 lockdep_init_map_type+0x224/0x250
...
[    4.833277]         # Subtest: test_mb_mark_used
[    4.835875] ------------[ cut here ]------------
[    4.836055] kernel BUG at fs/ext4/mballoc.c:2053!

Guenter

Re: [PATCH v2 0/5] More unit test for mballoc

[Kemeng Shi]

on 2/26/2024 12:07 AM, Guenter Roeck wrote:
> On Wed, Jan 03, 2024 at 06:48:55PM +0800, Kemeng Shi wrote:
>> This series covers more function to mark on-disk bitmap. Besides, some
>> code which is relevant to buddy cache is also tested.
>> Before more work is done, I want to be sure I'm not on a wrong
>> direction!
>>
>> v1->v2:
>> -Fix unused variable warning which is reported at
>> https://lore.kernel.org/lkml/202311260042.kMxL6DnL-lkp@intel.com/T/
>>
> 
> With this patch series in linux-text, and with various debug options
> enabled, unit tests for ext4 fail widely and result in crashes.
> 
> [    4.798582]         # Subtest: test_new_blocks_simple
> [    4.803166] BUG: key 00000000000000a8 has not been registered!
> [    4.803443] ------------[ cut here ]------------
> ILLOPC: ffffffff9050cbf4: 0f 0b
> [    4.803521] DEBUG_LOCKS_WARN_ON(1)
> [    4.803720] WARNING: CPU: 0 PID: 154 at kernel/locking/lockdep.c:4895 lockdep_init_map_type+0x224/0x250
Not sure how this is triggerred.
> ...
> [    4.833277]         # Subtest: test_mb_mark_used
> [    4.835875] ------------[ cut here ]------------
> [    4.836055] kernel BUG at fs/ext4/mballoc.c:2053!
Internal functions mb_mark_used assumes group lock is held but unit
test code doesn't acquire it as there is no concurrent block
allocation/free in test code. Will add lock in unit test to fix this.
> 
> Guenter
> 
Hi Guenter, thanks for test and report this. I will appreciate it if
you could tell me how to reproduce this as I need to debug the first
issue and verify the fix. Thanks!