zdb: Report bad label checksum

In case if all label checksums will be invalid on any vdev, the pool
will become unimportable. From other side zdb with -l option will not
provide any useful information why it happened. Add notifications
about corrupted label checksums.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Fedor Uporov <fuporov.vstack@gmail.com>
Closes #2509
Closes #12685

5 files changed, 143 insertions, 6 deletions

diff --git a/cmd/zdb/zdb.c b/cmd/zdb/zdb.c
index 8bbb77479..dff61b7ba 100644
--- a/cmd/zdb/zdb.c
+++ b/cmd/zdb/zdb.c
@@ -4221,11 +4221,13 @@ print_label_numbers(char *prefix, cksum_record_t *rec)
 
 typedef struct zdb_label {
 	vdev_label_t label;
+	uint64_t label_offset;
 	nvlist_t *config_nv;
 	cksum_record_t *config;
 	cksum_record_t *uberblocks[MAX_UBERBLOCK_COUNT];
 	boolean_t header_printed;
 	boolean_t read_failed;
+	boolean_t cksum_valid;
 } zdb_label_t;
 
 static void
@@ -4239,7 +4241,8 @@ print_label_header(zdb_label_t *label, int l)
 		return;
 
 	(void) printf("------------------------------------\n");
-	(void) printf("LABEL %d\n", l);
+	(void) printf("LABEL %d %s\n", l,
+	    label->cksum_valid ? "" : "(Bad label cksum)");
 	(void) printf("------------------------------------\n");
 
 	label->header_printed = B_TRUE;
@@ -4751,6 +4754,42 @@ zdb_copy_object(objset_t *os, uint64_t srcobj, char *destfile)
 	return (err);
 }
 
+static boolean_t
+label_cksum_valid(vdev_label_t *label, uint64_t offset)
+{
+	zio_checksum_info_t *ci = &zio_checksum_table[ZIO_CHECKSUM_LABEL];
+	zio_cksum_t expected_cksum;
+	zio_cksum_t actual_cksum;
+	zio_cksum_t verifier;
+	zio_eck_t *eck;
+	int byteswap;
+
+	void *data = (char *)label + offsetof(vdev_label_t, vl_vdev_phys);
+	eck = (zio_eck_t *)((char *)(data) + VDEV_PHYS_SIZE) - 1;
+
+	offset += offsetof(vdev_label_t, vl_vdev_phys);
+	ZIO_SET_CHECKSUM(&verifier, offset, 0, 0, 0);
+
+	byteswap = (eck->zec_magic == BSWAP_64(ZEC_MAGIC));
+	if (byteswap)
+		byteswap_uint64_array(&verifier, sizeof (zio_cksum_t));
+
+	expected_cksum = eck->zec_cksum;
+	eck->zec_cksum = verifier;
+
+	abd_t *abd = abd_get_from_buf(data, VDEV_PHYS_SIZE);
+	ci->ci_func[byteswap](abd, VDEV_PHYS_SIZE, NULL, &actual_cksum);
+	abd_free(abd);
+
+	if (byteswap)
+		byteswap_uint64_array(&expected_cksum, sizeof (zio_cksum_t));
+
+	if (ZIO_CHECKSUM_EQUAL(actual_cksum, expected_cksum))
+		return (B_TRUE);
+
+	return (B_FALSE);
+}
+
 static int
 dump_label(const char *dev)
 {
@@ -4817,8 +4856,9 @@ dump_label(const char *dev)
 
 	/*
 	 * 1. Read the label from disk
-	 * 2. Unpack the configuration and insert in config tree.
-	 * 3. Traverse all uberblocks and insert in uberblock tree.
+	 * 2. Verify label cksum
+	 * 3. Unpack the configuration and insert in config tree.
+	 * 4. Traverse all uberblocks and insert in uberblock tree.
 	 */
 	for (int l = 0; l < VDEV_LABELS; l++) {
 		zdb_label_t *label = &labels[l];
@@ -4829,8 +4869,10 @@ dump_label(const char *dev)
 		zio_cksum_t cksum;
 		vdev_t vd;
 
+		label->label_offset = vdev_label_offset(psize, l, 0);
+
 		if (pread64(fd, &label->label, sizeof (label->label),
-		    vdev_label_offset(psize, l, 0)) != sizeof (label->label)) {
+		    label->label_offset) != sizeof (label->label)) {
 			if (!dump_opt['q'])
 				(void) printf("failed to read label %d\n", l);
 			label->read_failed = B_TRUE;
@@ -4839,6 +4881,8 @@ dump_label(const char *dev)
 		}
 
 		label->read_failed = B_FALSE;
+		label->cksum_valid = label_cksum_valid(&label->label,
+		    label->label_offset);
 
 		if (nvlist_unpack(buf, buflen, &config, 0) == 0) {
 			nvlist_t *vdev_tree = NULL;
diff --git a/tests/runfiles/common.run b/tests/runfiles/common.run
index 9f181b53e..220710548 100644
--- a/tests/runfiles/common.run
+++ b/tests/runfiles/common.run
@@ -122,8 +122,9 @@ tags = ['functional', 'clean_mirror']
 tests = ['zdb_002_pos', 'zdb_003_pos', 'zdb_004_pos', 'zdb_005_pos',
     'zdb_006_pos', 'zdb_args_neg', 'zdb_args_pos',
     'zdb_block_size_histogram', 'zdb_checksum', 'zdb_decompress',
-    'zdb_display_block', 'zdb_object_range_neg', 'zdb_object_range_pos',
-    'zdb_objset_id', 'zdb_decompress_zstd', 'zdb_recover', 'zdb_recover_2']
+    'zdb_display_block', 'zdb_label_checksum', 'zdb_object_range_neg',
+    'zdb_object_range_pos', 'zdb_objset_id', 'zdb_decompress_zstd',
+    'zdb_recover', 'zdb_recover_2']
 pre =
 post =
 tags = ['functional', 'cli_root', 'zdb']
diff --git a/tests/zfs-tests/include/blkdev.shlib b/tests/zfs-tests/include/blkdev.shlib
index bf7095290..18fc6b352 100644
--- a/tests/zfs-tests/include/blkdev.shlib
+++ b/tests/zfs-tests/include/blkdev.shlib
@@ -652,3 +652,16 @@ function corrupt_blocks_at_level # input_file corrupt_level
 	# This is necessary for pools made of loop devices.
 	sync
 }
+
+function corrupt_label_checksum # label_number vdev_path
+{
+	typeset label_size=$((256*1024))
+	typeset vdev_size=$(stat_size ${2})
+	typeset -a offsets=("$((128*1024 - 32))" \
+	    "$(($label_size + (128*1024 - 32)))" \
+	    "$(($vdev_size - $label_size - (128*1024 + 32)))" \
+	    "$(($vdev_size - (128*1024 + 32)))")
+
+	dd if=/dev/urandom of=${2} seek=${offsets[$1]} bs=1 count=32 \
+	    conv=notrunc
+}
diff --git a/tests/zfs-tests/tests/functional/cli_root/zdb/Makefile.am b/tests/zfs-tests/tests/functional/cli_root/zdb/Makefile.am
index d84a3dfc7..c1d4bf5a4 100644
--- a/tests/zfs-tests/tests/functional/cli_root/zdb/Makefile.am
+++ b/tests/zfs-tests/tests/functional/cli_root/zdb/Makefile.am
@@ -14,6 +14,7 @@ dist_pkgdata_SCRIPTS = \
 	zdb_object_range_neg.ksh \
 	zdb_object_range_pos.ksh \
 	zdb_display_block.ksh \
+	zdb_label_checksum.ksh \
 	zdb_objset_id.ksh \
 	zdb_recover.ksh \
 	zdb_recover_2.ksh
diff --git a/tests/zfs-tests/tests/functional/cli_root/zdb/zdb_label_checksum.ksh b/tests/zfs-tests/tests/functional/cli_root/zdb/zdb_label_checksum.ksh
new file mode 100755
index 000000000..6cc156041
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zdb/zdb_label_checksum.ksh
@@ -0,0 +1,78 @@
+#!/bin/ksh
+
+#
+# This file and its contents are supplied under the terms of the
+# Common Development and Distribution License ("CDDL"), version 1.0.
+# You may only use this file in accordance with the terms of version
+# 1.0 of the CDDL.
+#
+# A full copy of the text of the CDDL should have accompanied this
+# source.  A copy of the CDDL is also available via the Internet at
+# http://www.illumos.org/license/CDDL.
+#
+
+#
+# Copyright (c) 2021 by vStack. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/include/blkdev.shlib
+
+#
+# Description:
+# zdb -l will report corrupted labels checksums
+#
+# Strategy:
+# 1. Create pool with some number of vdevs and export it
+# 2. Corrupt label 0 and label 1, check that corrupted labels are reported
+# 3. Check that pool still be imported correctly
+# 4. Corrupt all labels, check that all corrupted labels are reported
+# 5. Check that pool cannot be imported
+#
+
+log_assert "Verify zdb -l will report corrupted labels checksums"
+log_onexit cleanup
+
+VIRTUAL_DISK=$TEST_BASE_DIR/disk
+
+function cleanup
+{
+	poolexists $TESTPOOL && log_must destroy_pool $TESTPOOL
+	[[ -f $VIRTUAL_DISK ]] && log_must rm $VIRTUAL_DISK
+}
+
+verify_runnable "global"
+
+log_must truncate -s $(($MINVDEVSIZE * 8)) $VIRTUAL_DISK
+
+log_must zpool create $TESTPOOL $VIRTUAL_DISK
+log_must zpool export $TESTPOOL
+
+corrupt_label_checksum 0 $VIRTUAL_DISK
+corrupt_label_checksum 1 $VIRTUAL_DISK
+
+msg_count=$(zdb -l $VIRTUAL_DISK | grep -c '(Bad label cksum)')
+[ $msg_count -ne 1 ] && \
+    log_fail "zdb -l produces an incorrect number of corrupted labels."
+
+msg_count=$(zdb -lll $VIRTUAL_DISK | grep -c '(Bad label cksum)')
+[ $msg_count -ne 2 ] && \
+    log_fail "zdb -l produces an incorrect number of corrupted labels."
+
+log_must zpool import $TESTPOOL -d $TEST_BASE_DIR
+log_must zpool export $TESTPOOL
+
+corrupt_label_checksum 0 $VIRTUAL_DISK
+corrupt_label_checksum 1 $VIRTUAL_DISK
+corrupt_label_checksum 2 $VIRTUAL_DISK
+corrupt_label_checksum 3 $VIRTUAL_DISK
+
+msg_count=$(zdb -lll $VIRTUAL_DISK | grep -c '(Bad label cksum)')
+[ $msg_count -ne 4 ] && \
+    log_fail "zdb -l produces an incorrect number of corrupted labels."
+
+log_mustnot zpool import $TESTPOOL -d $TEST_BASE_DIR
+
+cleanup
+
+log_pass "zdb -l bad cksum report is correct."