Commit f046f89a authored by Joe Thornber's avatar Joe Thornber Committed by Alasdair G Kergon

dm thin: fix discard corruption

Fix a bug in dm_btree_remove that could leave leaf values with incorrect
reference counts.  The effect of this was that removal of a shared block
could result in the space maps thinking the block was no longer used.
More concretely, if you have a thin device and a snapshot of it, sending
a discard to a shared region of the thin could corrupt the snapshot.

Thinp uses a 2-level nested btree to store it's mappings.  This first
level is indexed by thin device, and the second level by logical
block.

Often when we're removing an entry in this mapping tree we need to
rebalance nodes, which can involve shadowing them, possibly creating a
copy if the block is shared.  If we do create a copy then children of
that node need to have their reference counts incremented.  In this
way reference counts percolate down the tree as shared trees diverge.

The rebalance functions were incrementing the children at the
appropriate time, but they were always assuming the children were
internal nodes.  This meant the leaf values (in our case packed
block/flags entries) were not being incremented.

Cc: stable@vger.kernel.org
Signed-off-by: default avatarJoe Thornber <ejt@redhat.com>
Signed-off-by: default avatarAlasdair G Kergon <agk@redhat.com>
parent a937536b
...@@ -2544,7 +2544,7 @@ static struct target_type pool_target = { ...@@ -2544,7 +2544,7 @@ static struct target_type pool_target = {
.name = "thin-pool", .name = "thin-pool",
.features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE | .features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE |
DM_TARGET_IMMUTABLE, DM_TARGET_IMMUTABLE,
.version = {1, 6, 1}, .version = {1, 7, 0},
.module = THIS_MODULE, .module = THIS_MODULE,
.ctr = pool_ctr, .ctr = pool_ctr,
.dtr = pool_dtr, .dtr = pool_dtr,
...@@ -2831,7 +2831,7 @@ static int thin_iterate_devices(struct dm_target *ti, ...@@ -2831,7 +2831,7 @@ static int thin_iterate_devices(struct dm_target *ti,
static struct target_type thin_target = { static struct target_type thin_target = {
.name = "thin", .name = "thin",
.version = {1, 7, 1}, .version = {1, 8, 0},
.module = THIS_MODULE, .module = THIS_MODULE,
.ctr = thin_ctr, .ctr = thin_ctr,
.dtr = thin_dtr, .dtr = thin_dtr,
......
...@@ -139,15 +139,8 @@ struct child { ...@@ -139,15 +139,8 @@ struct child {
struct btree_node *n; struct btree_node *n;
}; };
static struct dm_btree_value_type le64_type = { static int init_child(struct dm_btree_info *info, struct dm_btree_value_type *vt,
.context = NULL, struct btree_node *parent,
.size = sizeof(__le64),
.inc = NULL,
.dec = NULL,
.equal = NULL
};
static int init_child(struct dm_btree_info *info, struct btree_node *parent,
unsigned index, struct child *result) unsigned index, struct child *result)
{ {
int r, inc; int r, inc;
...@@ -164,7 +157,7 @@ static int init_child(struct dm_btree_info *info, struct btree_node *parent, ...@@ -164,7 +157,7 @@ static int init_child(struct dm_btree_info *info, struct btree_node *parent,
result->n = dm_block_data(result->block); result->n = dm_block_data(result->block);
if (inc) if (inc)
inc_children(info->tm, result->n, &le64_type); inc_children(info->tm, result->n, vt);
*((__le64 *) value_ptr(parent, index)) = *((__le64 *) value_ptr(parent, index)) =
cpu_to_le64(dm_block_location(result->block)); cpu_to_le64(dm_block_location(result->block));
...@@ -236,7 +229,7 @@ static void __rebalance2(struct dm_btree_info *info, struct btree_node *parent, ...@@ -236,7 +229,7 @@ static void __rebalance2(struct dm_btree_info *info, struct btree_node *parent,
} }
static int rebalance2(struct shadow_spine *s, struct dm_btree_info *info, static int rebalance2(struct shadow_spine *s, struct dm_btree_info *info,
unsigned left_index) struct dm_btree_value_type *vt, unsigned left_index)
{ {
int r; int r;
struct btree_node *parent; struct btree_node *parent;
...@@ -244,11 +237,11 @@ static int rebalance2(struct shadow_spine *s, struct dm_btree_info *info, ...@@ -244,11 +237,11 @@ static int rebalance2(struct shadow_spine *s, struct dm_btree_info *info,
parent = dm_block_data(shadow_current(s)); parent = dm_block_data(shadow_current(s));
r = init_child(info, parent, left_index, &left); r = init_child(info, vt, parent, left_index, &left);
if (r) if (r)
return r; return r;
r = init_child(info, parent, left_index + 1, &right); r = init_child(info, vt, parent, left_index + 1, &right);
if (r) { if (r) {
exit_child(info, &left); exit_child(info, &left);
return r; return r;
...@@ -368,7 +361,7 @@ static void __rebalance3(struct dm_btree_info *info, struct btree_node *parent, ...@@ -368,7 +361,7 @@ static void __rebalance3(struct dm_btree_info *info, struct btree_node *parent,
} }
static int rebalance3(struct shadow_spine *s, struct dm_btree_info *info, static int rebalance3(struct shadow_spine *s, struct dm_btree_info *info,
unsigned left_index) struct dm_btree_value_type *vt, unsigned left_index)
{ {
int r; int r;
struct btree_node *parent = dm_block_data(shadow_current(s)); struct btree_node *parent = dm_block_data(shadow_current(s));
...@@ -377,17 +370,17 @@ static int rebalance3(struct shadow_spine *s, struct dm_btree_info *info, ...@@ -377,17 +370,17 @@ static int rebalance3(struct shadow_spine *s, struct dm_btree_info *info,
/* /*
* FIXME: fill out an array? * FIXME: fill out an array?
*/ */
r = init_child(info, parent, left_index, &left); r = init_child(info, vt, parent, left_index, &left);
if (r) if (r)
return r; return r;
r = init_child(info, parent, left_index + 1, &center); r = init_child(info, vt, parent, left_index + 1, &center);
if (r) { if (r) {
exit_child(info, &left); exit_child(info, &left);
return r; return r;
} }
r = init_child(info, parent, left_index + 2, &right); r = init_child(info, vt, parent, left_index + 2, &right);
if (r) { if (r) {
exit_child(info, &left); exit_child(info, &left);
exit_child(info, &center); exit_child(info, &center);
...@@ -434,7 +427,8 @@ static int get_nr_entries(struct dm_transaction_manager *tm, ...@@ -434,7 +427,8 @@ static int get_nr_entries(struct dm_transaction_manager *tm,
} }
static int rebalance_children(struct shadow_spine *s, static int rebalance_children(struct shadow_spine *s,
struct dm_btree_info *info, uint64_t key) struct dm_btree_info *info,
struct dm_btree_value_type *vt, uint64_t key)
{ {
int i, r, has_left_sibling, has_right_sibling; int i, r, has_left_sibling, has_right_sibling;
uint32_t child_entries; uint32_t child_entries;
...@@ -472,13 +466,13 @@ static int rebalance_children(struct shadow_spine *s, ...@@ -472,13 +466,13 @@ static int rebalance_children(struct shadow_spine *s,
has_right_sibling = i < (le32_to_cpu(n->header.nr_entries) - 1); has_right_sibling = i < (le32_to_cpu(n->header.nr_entries) - 1);
if (!has_left_sibling) if (!has_left_sibling)
r = rebalance2(s, info, i); r = rebalance2(s, info, vt, i);
else if (!has_right_sibling) else if (!has_right_sibling)
r = rebalance2(s, info, i - 1); r = rebalance2(s, info, vt, i - 1);
else else
r = rebalance3(s, info, i - 1); r = rebalance3(s, info, vt, i - 1);
return r; return r;
} }
...@@ -529,7 +523,7 @@ static int remove_raw(struct shadow_spine *s, struct dm_btree_info *info, ...@@ -529,7 +523,7 @@ static int remove_raw(struct shadow_spine *s, struct dm_btree_info *info,
if (le32_to_cpu(n->header.flags) & LEAF_NODE) if (le32_to_cpu(n->header.flags) & LEAF_NODE)
return do_leaf(n, key, index); return do_leaf(n, key, index);
r = rebalance_children(s, info, key); r = rebalance_children(s, info, vt, key);
if (r) if (r)
break; break;
...@@ -550,6 +544,14 @@ static int remove_raw(struct shadow_spine *s, struct dm_btree_info *info, ...@@ -550,6 +544,14 @@ static int remove_raw(struct shadow_spine *s, struct dm_btree_info *info,
return r; return r;
} }
static struct dm_btree_value_type le64_type = {
.context = NULL,
.size = sizeof(__le64),
.inc = NULL,
.dec = NULL,
.equal = NULL
};
int dm_btree_remove(struct dm_btree_info *info, dm_block_t root, int dm_btree_remove(struct dm_btree_info *info, dm_block_t root,
uint64_t *keys, dm_block_t *new_root) uint64_t *keys, dm_block_t *new_root)
{ {
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment