c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2024-12-27 03:19:35 +03:00
Code Issues Packages Projects Releases Wiki Activity

Illumos 6370 - ZFS send fails to transmit some holes

Browse Source 
6370 ZFS send fails to transmit some holes
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Stefan Ring <stefanrin@gmail.com>
Reviewed by: Steven Burgess <sburgess@datto.com>
Reviewed by: Arne Jansen <sensille@gmx.net>
Approved by: Robert Mustacchi <rm@joyent.com>

References:
  https://www.illumos.org/issues/6370
  https://github.com/illumos/illumos-gate/commit/286ef71

In certain circumstances, "zfs send -i" (incremental send) can produce
a stream which will result in incorrect sparse file contents on the
target.

The problem manifests as regions of the received file that should be
sparse (and read a zero-filled) actually contain data from a file that
was deleted (and which happened to share this file's object ID).

Note: this can happen only with filesystems (not zvols, because they do
not free (and thus can not reuse) object IDs).

Note: This can happen only if, since the incremental source (FromSnap),
a file was deleted and then another file was created, and the new file
is sparse (i.e. has areas that were never written to and should be
implicitly zero-filled).

We suspect that this was introduced by 4370 (applies only if hole_birth
feature is enabled), and made worse by 5243 (applies if hole_birth
feature is disabled, and we never send any holes).

The bug is caused by the hole birth feature. When an object is deleted
and replaced, all the holes in the object have birth time zero. However,
zfs send cannot tell that the holes are new since the file was replaced,
so it doesn't send them in an incremental. As a result, you can end up
with invalid data when you receive incremental send streams. As a
short-term fix, we can always send holes with birth time 0 (unless it's
a zvol or a dataset where we can guarantee that no objects have been
reused).

Ported-by: Steven Burgess <sburgess@datto.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4369
Closes #4050
This commit is contained in:
Paul Dagnelie 2016-02-25 20:45:19 -05:00 committed by Brian Behlendorf
parent a9977b37ca
commit c352ec27d5
2 changed files with 43 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
module/zfs/dmu_object.c
View File

@ -20,7 +20,7 @@
*/ */
/* /*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2013, 2014 by Delphix. All rights reserved. * Copyright (c) 2013, 2015 by Delphix. All rights reserved.
* Copyright 2014 HybridCluster. All rights reserved. * Copyright 2014 HybridCluster. All rights reserved.
*/ */
@ -50,6 +50,12 @@ dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize,
* reasonably sparse (at most 1/4 full). Look from the * reasonably sparse (at most 1/4 full). Look from the
* beginning once, but after that keep looking from here. * beginning once, but after that keep looking from here.
* If we can't find one, just keep going from here. * If we can't find one, just keep going from here.
*
* Note that dmu_traverse depends on the behavior that we use
* multiple blocks of the dnode object before going back to
* reuse objects. Any change to this algorithm should preserve
* that property or find another solution to the issues
* described in traverse_visitbp.
*/ */
if (P2PHASE(object, L2_dnode_count) == 0) { if (P2PHASE(object, L2_dnode_count) == 0) {
uint64_t offset = restarted ? object << DNODE_SHIFT : 0; uint64_t offset = restarted ? object << DNODE_SHIFT : 0;

49
module/zfs/dmu_traverse.c
View File

@ -20,7 +20,7 @@
*/ */
/* /*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2012, 2014 by Delphix. All rights reserved. * Copyright (c) 2012, 2016 by Delphix. All rights reserved.
*/ */
#include <sys/zfs_context.h> #include <sys/zfs_context.h>
@ -61,6 +61,7 @@ typedef struct traverse_data {
uint64_t td_hole_birth_enabled_txg; uint64_t td_hole_birth_enabled_txg;
blkptr_cb_t *td_func; blkptr_cb_t *td_func;
void *td_arg; void *td_arg;
boolean_t td_realloc_possible;
} traverse_data_t; } traverse_data_t;
static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp, static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
@ -228,18 +229,30 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp,
if (bp->blk_birth == 0) { if (bp->blk_birth == 0) {
/* /*
* Since this block has a birth time of 0 it must be a * Since this block has a birth time of 0 it must be one of
* hole created before the SPA_FEATURE_HOLE_BIRTH * two things: a hole created before the
* feature was enabled. If SPA_FEATURE_HOLE_BIRTH * SPA_FEATURE_HOLE_BIRTH feature was enabled, or a hole
* was enabled before the min_txg for this traveral we * which has always been a hole in an object.
* know the hole must have been created before the *
* min_txg for this traveral, so we can skip it. If * If a file is written sparsely, then the unwritten parts of
* SPA_FEATURE_HOLE_BIRTH was enabled after the min_txg * the file were "always holes" -- that is, they have been
* for this traveral we cannot tell if the hole was * holes since this object was allocated. However, we (and
* created before or after the min_txg for this * our callers) can not necessarily tell when an object was
* traversal, so we cannot skip it. * allocated. Therefore, if it's possible that this object
* was freed and then its object number reused, we need to
* visit all the holes with birth==0.
*
* If it isn't possible that the object number was reused,
* then if SPA_FEATURE_HOLE_BIRTH was enabled before we wrote
* all the blocks we will visit as part of this traversal,
* then this hole must have always existed, so we can skip
* it. We visit blocks born after (exclusive) td_min_txg.
*
* Note that the meta-dnode cannot be reallocated.
*/ */
if (td->td_hole_birth_enabled_txg < td->td_min_txg) if ((!td->td_realloc_possible ||
zb->zb_object == DMU_META_DNODE_OBJECT) &&
td->td_hole_birth_enabled_txg <= td->td_min_txg)
return (0); return (0);
} else if (bp->blk_birth <= td->td_min_txg) { } else if (bp->blk_birth <= td->td_min_txg) {
return (0); return (0);
@ -347,6 +360,15 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp,
prefetch_dnode_metadata(td, mdnp, zb->zb_objset, prefetch_dnode_metadata(td, mdnp, zb->zb_objset,
DMU_META_DNODE_OBJECT); DMU_META_DNODE_OBJECT);
/*
* See the block comment above for the goal of this variable.
* If the maxblkid of the meta-dnode is 0, then we know that
* we've never had more than DNODES_PER_BLOCK objects in the
* dataset, which means we can't have reused any object ids.
*/
if (osp->os_meta_dnode.dn_maxblkid == 0)
td->td_realloc_possible = B_FALSE;
if (arc_buf_size(buf) >= sizeof (objset_phys_t)) { if (arc_buf_size(buf) >= sizeof (objset_phys_t)) {
prefetch_dnode_metadata(td, gdnp, zb->zb_objset, prefetch_dnode_metadata(td, gdnp, zb->zb_objset,
DMU_GROUPUSED_OBJECT); DMU_GROUPUSED_OBJECT);
@ -554,12 +576,13 @@ traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp,
td->td_pfd = pd; td->td_pfd = pd;
td->td_flags = flags; td->td_flags = flags;
td->td_paused = B_FALSE; td->td_paused = B_FALSE;
td->td_realloc_possible = (txg_start == 0 ? B_FALSE : B_TRUE);
if (spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) { if (spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) {
VERIFY(spa_feature_enabled_txg(spa, VERIFY(spa_feature_enabled_txg(spa,
SPA_FEATURE_HOLE_BIRTH, &td->td_hole_birth_enabled_txg)); SPA_FEATURE_HOLE_BIRTH, &td->td_hole_birth_enabled_txg));
} else { } else {
td->td_hole_birth_enabled_txg = 0; td->td_hole_birth_enabled_txg = UINT64_MAX;
} }
pd->pd_flags = flags; pd->pd_flags = flags;