openzfs · behlendorf · Apr 26, 2022 · Aug 3, 2019
diff --git a/include/sys/dsl_dataset.h b/include/sys/dsl_dataset.h
@@ -487,6 +487,9 @@ boolean_t dsl_dataset_get_uint64_array_feature(dsl_dataset_t *ds,
 void dsl_dataset_activate_redaction(dsl_dataset_t *ds, uint64_t *redact_snaps,
     uint64_t num_redact_snaps, dmu_tx_t *tx);
 
+int dsl_dataset_oldest_snapshot(spa_t *spa, uint64_t head_ds, uint64_t min_txg,
+    uint64_t *oldest_dsobj);
+
 #ifdef ZFS_DEBUG
 #define	dprintf_ds(ds, fmt, ...) do { \
 	if (zfs_flags & ZFS_DEBUG_DPRINTF) { \

diff --git a/include/sys/spa.h b/include/sys/spa.h
@@ -1144,11 +1144,17 @@ extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
 extern void zfs_post_state_change(spa_t *spa, vdev_t *vd, uint64_t laststate);
 extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
 extern uint64_t spa_get_errlog_size(spa_t *spa);
-extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
+extern int spa_get_errlog(spa_t *spa, void *uaddr, uint64_t *count);
 extern void spa_errlog_rotate(spa_t *spa);
 extern void spa_errlog_drain(spa_t *spa);
 extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
 extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
+extern void spa_delete_dataset_errlog(spa_t *spa, uint64_t ds, dmu_tx_t *tx);
+extern void spa_swap_errlog(spa_t *spa, uint64_t new_head_ds,
+    uint64_t old_head_ds, dmu_tx_t *tx);
+extern void sync_error_list(spa_t *spa, avl_tree_t *t, uint64_t *obj,
+    dmu_tx_t *tx);
+extern void spa_upgrade_errlog(spa_t *spa, dmu_tx_t *tx);
 
 /* vdev cache */
 extern void vdev_cache_stat_init(void);

diff --git a/include/sys/zio.h b/include/sys/zio.h
@@ -283,6 +283,13 @@ extern const char *const zio_type_name[ZIO_TYPES];
  * Note: this structure is passed between userland and the kernel, and is
  * stored on disk (by virtue of being incorporated into other on-disk
  * structures, e.g. dsl_scan_phys_t).
+ *
+ * If the head_errlog feature is enabled a different on-disk format for error
+ * logs is used. This introduces the use of an error bookmark, a four-tuple
+ * <object, level, blkid, birth> that uniquely identifies any error block
+ * in the pool. The birth transaction group is used to track whether the block
+ * has been overwritten by newer data or added to a snapshot since its marking
+ * as an error.
  */
 struct zbookmark_phys {
 	uint64_t	zb_objset;
@@ -291,6 +298,13 @@ struct zbookmark_phys {
 	uint64_t	zb_blkid;
 };
 
+typedef struct zbookmark_err_phys {
+	uint64_t	zb_object;
+	int64_t		zb_level;
+	uint64_t	zb_blkid;
+	uint64_t	zb_birth;
+} zbookmark_err_phys_t;
+
 #define	SET_BOOKMARK(zb, objset, object, level, blkid)  \
 {                                                       \
 	(zb)->zb_objset = objset;                       \

diff --git a/include/zfeature_common.h b/include/zfeature_common.h
@@ -76,6 +76,7 @@ typedef enum spa_feature {
 	SPA_FEATURE_ZSTD_COMPRESS,
 	SPA_FEATURE_DRAID,
 	SPA_FEATURE_ZILSAXATTR,
+	SPA_FEATURE_HEAD_ERRLOG,
 	SPA_FEATURES
 } spa_feature_t;
 

diff --git a/lib/libnvpair/libnvpair.abi b/lib/libnvpair/libnvpair.abi
@@ -2794,7 +2794,7 @@
   </abi-instr>
   <abi-instr address-size='64' path='assert.c' language='LANG_C99'>
     <function-decl name='libspl_set_assert_ok' mangled-name='libspl_set_assert_ok' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libspl_set_assert_ok'>
-      <parameter type-id='f58c8277' name='val'/>
+      <parameter type-id='c19b74c3' name='val'/>
       <return type-id='48b5725f'/>
     </function-decl>
     <function-decl name='libspl_assertf' mangled-name='libspl_assertf' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libspl_assertf'>

diff --git a/lib/libuutil/libuutil.abi b/lib/libuutil/libuutil.abi
diff --git a/lib/libzfs/libzfs.abi b/lib/libzfs/libzfs.abi
@@ -596,7 +596,7 @@
     <elf-symbol name='fletcher_4_superscalar4_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='fletcher_4_superscalar_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='libzfs_config_ops' size='16' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
-    <elf-symbol name='spa_feature_table' size='1960' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
+    <elf-symbol name='spa_feature_table' size='2016' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfeature_checks_disable' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_deleg_perm_tab' size='512' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='zfs_history_event_names' size='328' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
@@ -1855,8 +1855,8 @@
     </function-decl>
   </abi-instr>
   <abi-instr address-size='64' path='../../module/zcommon/zfeature_common.c' language='LANG_C99'>
-    <array-type-def dimensions='1' type-id='83f29ca2' size-in-bits='15680' id='9d60dcc5'>
-      <subrange length='35' type-id='7359adad' id='6e6845b5'/>
+    <array-type-def dimensions='1' type-id='83f29ca2' size-in-bits='16128' id='9d5e9e2e'>
+      <subrange length='36' type-id='7359adad' id='ae666bde'/>
     </array-type-def>
     <enum-decl name='spa_feature' id='33ecb627'>
       <underlying-type type-id='9cac1fee'/>
@@ -1896,7 +1896,8 @@
       <enumerator name='SPA_FEATURE_ZSTD_COMPRESS' value='32'/>
       <enumerator name='SPA_FEATURE_DRAID' value='33'/>
       <enumerator name='SPA_FEATURE_ZILSAXATTR' value='34'/>
-      <enumerator name='SPA_FEATURES' value='35'/>
+      <enumerator name='SPA_FEATURE_HEAD_ERRLOG' value='35'/>
+      <enumerator name='SPA_FEATURES' value='36'/>
     </enum-decl>
     <typedef-decl name='spa_feature_t' type-id='33ecb627' id='d6618c78'/>
     <enum-decl name='zfeature_flags' id='6db816a4'>
@@ -1954,7 +1955,7 @@
     <qualified-type-def type-id='3eee3342' const='yes' id='0c1d5bbb'/>
     <pointer-type-def type-id='0c1d5bbb' size-in-bits='64' id='a3372543'/>
     <pointer-type-def type-id='d6618c78' size-in-bits='64' id='a8425263'/>
-    <var-decl name='spa_feature_table' type-id='9d60dcc5' mangled-name='spa_feature_table' visibility='default' elf-symbol-id='spa_feature_table'/>
+    <var-decl name='spa_feature_table' type-id='9d5e9e2e' mangled-name='spa_feature_table' visibility='default' elf-symbol-id='spa_feature_table'/>
     <var-decl name='zfeature_checks_disable' type-id='c19b74c3' mangled-name='zfeature_checks_disable' visibility='default' elf-symbol-id='zfeature_checks_disable'/>
     <function-decl name='zfeature_is_valid_guid' mangled-name='zfeature_is_valid_guid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfeature_is_valid_guid'>
       <parameter type-id='80f4b756' name='name'/>

diff --git a/lib/libzfsbootenv/libzfsbootenv.abi b/lib/libzfsbootenv/libzfsbootenv.abi
@@ -5,8 +5,6 @@
     <dependency name='libc.so.6'/>
   </elf-needed>
   <elf-function-symbols>
-    <elf-symbol name='_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
-    <elf-symbol name='_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='lzbe_add_pair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='lzbe_bootenv_print' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
     <elf-symbol name='lzbe_get_boot_device' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
@@ -20,6 +18,7 @@
     <type-decl name='char' size-in-bits='8' id='a84c031d'/>
     <type-decl name='int' size-in-bits='32' id='95e97e5e'/>
     <type-decl name='unnamed-enum-underlying-type-32' is-anonymous='yes' size-in-bits='32' alignment-in-bits='32' id='9cac1fee'/>
+    <type-decl name='void' id='48b5725f'/>
     <enum-decl name='lzbe_flags' id='2b77720b'>
       <underlying-type type-id='9cac1fee'/>
       <enumerator name='lzbe_add' value='0'/>
@@ -84,21 +83,16 @@
     <array-type-def dimensions='1' type-id='a84c031d' size-in-bits='160' id='664ac0b7'>
       <subrange length='20' type-id='7359adad' id='fdca39cf'/>
     </array-type-def>
+    <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
+    <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
+    <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
     <type-decl name='long int' size-in-bits='64' id='bd54fe1a'/>
     <type-decl name='signed char' size-in-bits='8' id='28577a57'/>
     <type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
+    <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
+    <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
+    <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
     <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
-    <class-decl name='_IO_marker' size-in-bits='192' is-struct='yes' visibility='default' id='010ae0b9'>
-      <data-member access='public' layout-offset-in-bits='0'>
-        <var-decl name='_next' type-id='e4c6fa61' visibility='default'/>
-      </data-member>
-      <data-member access='public' layout-offset-in-bits='64'>
-        <var-decl name='_sbuf' type-id='dca988a5' visibility='default'/>
-      </data-member>
-      <data-member access='public' layout-offset-in-bits='128'>
-        <var-decl name='_pos' type-id='95e97e5e' visibility='default'/>
-      </data-member>
-    </class-decl>
     <class-decl name='_IO_FILE' size-in-bits='1728' is-struct='yes' visibility='default' id='ec1ed955'>
       <data-member access='public' layout-offset-in-bits='0'>
         <var-decl name='_flags' type-id='95e97e5e' visibility='default'/>
@@ -167,16 +161,16 @@
         <var-decl name='_offset' type-id='724e4de6' visibility='default'/>
       </data-member>
       <data-member access='public' layout-offset-in-bits='1216'>
-        <var-decl name='__pad1' type-id='eaa32e2f' visibility='default'/>
+        <var-decl name='_codecvt' type-id='570f8c59' visibility='default'/>
       </data-member>
       <data-member access='public' layout-offset-in-bits='1280'>
-        <var-decl name='__pad2' type-id='eaa32e2f' visibility='default'/>
+        <var-decl name='_wide_data' type-id='c65a1f29' visibility='default'/>
       </data-member>
       <data-member access='public' layout-offset-in-bits='1344'>
-        <var-decl name='__pad3' type-id='eaa32e2f' visibility='default'/>
+        <var-decl name='_freeres_list' type-id='dca988a5' visibility='default'/>
       </data-member>
       <data-member access='public' layout-offset-in-bits='1408'>
-        <var-decl name='__pad4' type-id='eaa32e2f' visibility='default'/>
+        <var-decl name='_freeres_buf' type-id='eaa32e2f' visibility='default'/>
       </data-member>
       <data-member access='public' layout-offset-in-bits='1472'>
         <var-decl name='__pad5' type-id='b59d7dce' visibility='default'/>
@@ -188,19 +182,20 @@
         <var-decl name='_unused2' type-id='664ac0b7' visibility='default'/>
       </data-member>
     </class-decl>
-    <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
-    <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
-    <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
     <pointer-type-def type-id='aa12d1ba' size-in-bits='64' id='822cd80b'/>
     <pointer-type-def type-id='ec1ed955' size-in-bits='64' id='dca988a5'/>
+    <pointer-type-def type-id='a4036571' size-in-bits='64' id='570f8c59'/>
     <pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
     <pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
+    <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
+    <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
+    <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
+    <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
     <function-decl name='lzbe_bootenv_print' mangled-name='lzbe_bootenv_print' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_bootenv_print'>
       <parameter type-id='80f4b756' name='pool'/>
       <parameter type-id='80f4b756' name='nvlist'/>
       <parameter type-id='822cd80b' name='of'/>
       <return type-id='95e97e5e'/>
     </function-decl>
-    <type-decl name='void' id='48b5725f'/>
   </abi-instr>
 </abi-corpus>
diff --git a/man/man4/zfs.4 b/man/man4/zfs.4
@@ -454,6 +454,13 @@ If we have less than this amount of free space,
 most ZPL operations (e.g. write, create) will return
 .Sy ENOSPC .
 .
+.It Sy spa_upgrade_errlog_limit Ns = Ns Sy 0 Pq uint
+Limits the number of on-disk error log entries that will be converted to the
+new format when enabling the
+.Sy head_errlog
+feature.
+The default is to convert all log entries.
+.
 .It Sy vdev_removal_max_span Ns = Ns Sy 32768 Ns B Po 32kB Pc Pq int
 During top-level vdev removal, chunks of data are copied from the vdev
 which may include free space in order to trade bandwidth for IOPS.

diff --git a/man/man7/zpool-features.7 b/man/man7/zpool-features.7
@@ -507,6 +507,17 @@ once either of the limit properties has been set on a dataset
 and will never return to being
 .Sy enabled .
 .
+.feature com.delphix head_errlog no
+This feature enables the upgraded version of errlog, which required an on-disk
+error log format change.
+Now the error log of each head dataset is stored separately in the zap object
+and keyed by the head id.
+With this feature enabled, every dataset affected by an error block is listed
+in the output of
+.Nm zpool Cm status .
+.Pp
+\*[instant-never]
+.
 .feature com.delphix hole_birth no enabled_txg
 This feature has/had bugs, the result of which is that, if you do a
 .Nm zfs Cm send Fl i

diff --git a/module/zcommon/zfeature_common.c b/module/zcommon/zfeature_common.c
@@ -696,6 +696,7 @@ zpool_feature_init(void)
 	    ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL, sfeatures);
 
 	{
+
 	static const spa_feature_t zilsaxattr_deps[] = {
 		SPA_FEATURE_EXTENSIBLE_DATASET,
 		SPA_FEATURE_NONE
@@ -707,6 +708,12 @@ zpool_feature_init(void)
 	    ZFEATURE_TYPE_BOOLEAN, zilsaxattr_deps, sfeatures);
 	}
 
+	zfeature_register(SPA_FEATURE_HEAD_ERRLOG,
+	    "com.delphix:head_errlog", "head_errlog",
+	    "Support for per-dataset on-disk error logs.",
+	    ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, NULL,
+	    sfeatures);
+
 	zfs_mod_list_supported_free(sfeatures);
 }
 

diff --git a/module/zfs/dsl_dataset.c b/module/zfs/dsl_dataset.c
@@ -3708,6 +3708,15 @@ dsl_dataset_promote_sync(void *arg, dmu_tx_t *tx)
 
 	dsl_dir_rele(odd, FTAG);
 	promote_rele(ddpa, FTAG);
+
+	/*
+	 * Transfer common error blocks from old head to new head.
+	 */
+	if (spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_HEAD_ERRLOG)) {
+		uint64_t old_head = origin_head->ds_object;
+		uint64_t new_head = hds->ds_object;
+		spa_swap_errlog(dp->dp_spa, new_head, old_head, tx);
+	}
 }
 
 /*
@@ -4924,6 +4933,37 @@ dsl_dataset_activate_redaction(dsl_dataset_t *ds, uint64_t *redact_snaps,
 	ds->ds_feature[SPA_FEATURE_REDACTED_DATASETS] = ftuaa;
 }
 
+/*
+ * Find and return (in *oldest_dsobj) the oldest snapshot of the dsobj
+ * dataset whose birth time is >= min_txg.
+ */
+int
+dsl_dataset_oldest_snapshot(spa_t *spa, uint64_t head_ds, uint64_t min_txg,
+    uint64_t *oldest_dsobj)
+{
+	dsl_dataset_t *ds;
+	dsl_pool_t *dp = spa->spa_dsl_pool;
+
+	int error = dsl_dataset_hold_obj(dp, head_ds, FTAG, &ds);
+	if (error != 0)
+		return (error);
+
+	uint64_t prev_obj = dsl_dataset_phys(ds)->ds_prev_snap_obj;
+	uint64_t prev_obj_txg = dsl_dataset_phys(ds)->ds_prev_snap_txg;
+
+	while (prev_obj != 0 && min_txg < prev_obj_txg) {
+		dsl_dataset_rele(ds, FTAG);
+		if ((error = dsl_dataset_hold_obj(dp, prev_obj,
+		    FTAG, &ds)) != 0)
+			return (error);
+		prev_obj_txg = dsl_dataset_phys(ds)->ds_prev_snap_txg;
+		prev_obj = dsl_dataset_phys(ds)->ds_prev_snap_obj;
+	}
+	*oldest_dsobj = ds->ds_object;
+	dsl_dataset_rele(ds, FTAG);
+	return (0);
+}
+
 #if defined(_LP64)
 #define	RECORDSIZE_PERM ZMOD_RW
 #else

diff --git a/module/zfs/dsl_destroy.c b/module/zfs/dsl_destroy.c
@@ -1153,6 +1153,9 @@ dsl_destroy_head_sync_impl(dsl_dataset_t *ds, dmu_tx_t *tx)
 		dsl_destroy_snapshot_sync_impl(prev, B_FALSE, tx);
 		dsl_dataset_rele(prev, FTAG);
 	}
+	/* Delete errlog. */
+	if (spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_HEAD_ERRLOG))
+		spa_delete_dataset_errlog(dp->dp_spa, ds->ds_object, tx);
 }
 
 void