Skip to contentSkip to conversation

qcow2: Forbid discard in qcow2 v2 images with backing files

PATCH Superseded Alberto Garcia 3 participants 4 messages Mar 23, 2020 – Mar 24, 2020 Block layer

v1 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 qcow2: Forbid discard in qcow2 v2 images with backing files

A discard request deallocates the selected clusters so they read back as zeroes. This is done by clearing the cluster offset field and setting QCOW_OFLAG_ZERO in the L2 entry.

This flag is however only supported when qcow_version >= 3. In older images the cluster is simply deallocated, exposing any possible stale data from the backing file.

Since discard is an advisory operation it's safer to simply forbid it in this scenario.

Note that we are adding this check to qcow2_co_pdiscard() and not to qcow2_cluster_discard() or discard_in_l2_slice() because the last two are also used by qcow2_snapshot_create() to discard the clusters used by the VM state. In this case there's no risk of exposing stale data to the guest and we really want that the clusters are always discarded.

---
 block/qcow2.c              |  6 +++
 tests/qemu-iotests/060     |  5 ++-
 tests/qemu-iotests/060.out |  2 -
 tests/qemu-iotests/289     | 90 ++++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/289.out | 52 ++++++++++++++++++++++
 tests/qemu-iotests/group   |  1 +
 6 files changed, 152 insertions(+), 4 deletions(-)
 create mode 100755 tests/qemu-iotests/289
 create mode 100644 tests/qemu-iotests/289.out
block/qcow2.c +6 -0
index d44b45633d..7bb7e392e1 100644
@@ -3763,6 +3763,12 @@ static coroutine_fn int qcow2_co_pdiscard(BlockDriverState *bs,
     int ret;
     BDRVQcow2State *s = bs->opaque;
 
+    /* If the image does not support QCOW_OFLAG_ZERO then discarding
+     * clusters could expose stale data from the backing file. */
+    if (s->qcow_version < 3 && bs->backing) {
+        return -ENOTSUP;
+    }
+
     if (!QEMU_IS_ALIGNED(offset | bytes, s->cluster_size)) {
         assert(bytes < s->cluster_size);
         /* Ignore partial clusters, except for the special case of the
tests/qemu-iotests/060 +3 -2
index 043f12904a..4a4fdfb1e1 100755
@@ -167,9 +167,10 @@ _make_test_img -o 'compat=0.10' -b "$BACKING_IMG" 1G
 # Write two clusters, the second one enforces creation of an L2 table after
 # the first data cluster.
 $QEMU_IO -c 'write 0k 64k' -c 'write 512M 64k' "$TEST_IMG" | _filter_qemu_io
-# Discard the first cluster. This cluster will soon enough be reallocated and
+# Free the first cluster. This cluster will soon enough be reallocated and
 # used for COW.
-$QEMU_IO -c 'discard 0k 64k' "$TEST_IMG" | _filter_qemu_io
+poke_file "$TEST_IMG" '262144' "\x00\x00\x00\x00\x00\x00\x00\x00" # 0x40000 - L2 entry
+poke_file "$TEST_IMG" '131082' "\x00\x00" # 0x2000a - Refcount entry
 # Now, corrupt the image by marking the second L2 table cluster as free.
 poke_file "$TEST_IMG" '131084' "\x00\x00" # 0x2000c
 # Start a write operation requiring COW on the image stopping it right before
tests/qemu-iotests/060.out +0 -2
index d27692a33c..09caaea865 100644
@@ -105,8 +105,6 @@ wrote 65536/65536 bytes at offset 0
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
 wrote 65536/65536 bytes at offset 536870912
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
-discard 65536/65536 bytes at offset 0
-64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
 qcow2: Marking image as corrupt: Preventing invalid write on metadata (overlaps with active L2 table); further corruption events will be suppressed
 blkdebug: Suspended request '0'
 write failed: Input/output error
tests/qemu-iotests/289 new file +90 -0

90 lines not on this page · raw message

tests/qemu-iotests/289.out new file +52 -0

52 lines not on this page · raw message

tests/qemu-iotests/group +1 -0
index ec2b2302e5..891b3ce858 100644
@@ -295,3 +295,4 @@
 284 rw
 286 rw quick
 288 quick
+289 rw auto quick
Signature · 1 line
2.20.1
#1 lore raw
Quoted 11 lines · Alberto Garcia

A discard request deallocates the selected clusters so they read back as zeroes. This is done by clearing the cluster offset field and setting QCOW_OFLAG_ZERO in the L2 entry.

This flag is however only supported when qcow_version >= 3. In older images the cluster is simply deallocated, exposing any possible stale data from the backing file.

Since discard is an advisory operation it's safer to simply forbid it in this scenario.

Quoted 6 lines

Note that we are adding this check to qcow2_co_pdiscard() and not to qcow2_cluster_discard() or discard_in_l2_slice() because the last two are also used by qcow2_snapshot_create() to discard the clusters used by the VM state. In this case there's no risk of exposing stale data to the guest and we really want that the clusters are always discarded.

Sounds good to me.

tests/qemu-iotests/060 _make_test_img -o 'compat=0.10' -b "$BACKING_IMG" 1G

Quoted 34 lines
---
 block/qcow2.c              |  6 +++
 tests/qemu-iotests/060     |  5 ++-
 tests/qemu-iotests/060.out |  2 -
 tests/qemu-iotests/289     | 90 ++++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/289.out | 52 ++++++++++++++++++++++
 tests/qemu-iotests/group   |  1 +
 6 files changed, 152 insertions(+), 4 deletions(-)
 create mode 100755 tests/qemu-iotests/289
 create mode 100644 tests/qemu-iotests/289.out
block/qcow2.c +6 -0
index d44b45633d..7bb7e392e1 100644
@@ -3763,6 +3763,12 @@ static coroutine_fn int qcow2_co_pdiscard(BlockDriverState *bs,
     int ret;
     BDRVQcow2State *s = bs->opaque;
 
+    /* If the image does not support QCOW_OFLAG_ZERO then discarding
+     * clusters could expose stale data from the backing file. */
+    if (s->qcow_version < 3 && bs->backing) {
+        return -ENOTSUP;
+    }
+
     if (!QEMU_IS_ALIGNED(offset | bytes, s->cluster_size)) {
         assert(bytes < s->cluster_size);
         /* Ignore partial clusters, except for the special case of the
tests/qemu-iotests/060 +0 -0
index 043f12904a..4a4fdfb1e1 100755
@@ -167,9 +167,10 @@ _make_test_img -o 'compat=0.10' -b "$BACKING_IMG" 1G

More context: This image is created with -o 'compat=0.10', just because a discard on such an image would result in the cluster being freed. We can drop that compat=0.10 bit now.

Quoted 12 lines
tests/qemu-iotests/060 +3 -2
 # Write two clusters, the second one enforces creation of an L2 table after
 # the first data cluster.
 $QEMU_IO -c 'write 0k 64k' -c 'write 512M 64k' "$TEST_IMG" | _filter_qemu_io
-# Discard the first cluster. This cluster will soon enough be reallocated and
+# Free the first cluster. This cluster will soon enough be reallocated and
 # used for COW.
-$QEMU_IO -c 'discard 0k 64k' "$TEST_IMG" | _filter_qemu_io
+poke_file "$TEST_IMG" '262144' "\x00\x00\x00\x00\x00\x00\x00\x00" # 0x40000 - L2 entry
+poke_file "$TEST_IMG" '131082' "\x00\x00" # 0x2000a - Refcount entry
 # Now, corrupt the image by marking the second L2 table cluster as free.
 poke_file "$TEST_IMG" '131084' "\x00\x00" # 0x2000c
 # Start a write operation requiring COW on the image stopping it right before

[...]

Quoted 5 lines
tests/qemu-iotests/289 new file +0 -0
new file mode 100755
index 0000000000..13b4984721

[...]

tests/qemu-iotests/289 line 33

Quoted 4 lines
tests/qemu-iotests/289 new file +4 -0
+_cleanup()
+{
+    _cleanup_test_img
+    rm -f "$TEST_IMG.backing"

I’d call the image $TEST_IMG.base so _cleanup_test_img picks up on it. (rm-ing test images is also wrong, because with external data files, there will be more than one file. It doesn’t matter here anyway because this test doesn’t support external data files, but, well.)

tests/qemu-iotests/289 line 43

Quoted 10 lines
tests/qemu-iotests/289 new file +10 -0
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common.rc
+. ./common.filter
+
+_supported_fmt qcow2
+_supported_proto file
+_supported_os Linux

I’d mark the compat option unsupported because this test will ignore it.
Furthermore, the refcount_bits and data_file options are really
unsupported, because they won’t work with compat=0.10.

The test itself looks good.

Max

Reviewed-by on v1 1/1
#2 lore raw
In reply to Alberto Garcia
Quoted 27 lines · Alberto Garcia

A discard request deallocates the selected clusters so they read back as zeroes. This is done by clearing the cluster offset field and setting QCOW_OFLAG_ZERO in the L2 entry.

This flag is however only supported when qcow_version >= 3. In older images the cluster is simply deallocated, exposing any possible stale data from the backing file.

Since discard is an advisory operation it's safer to simply forbid it in this scenario.

Note that we are adding this check to qcow2_co_pdiscard() and not to qcow2_cluster_discard() or discard_in_l2_slice() because the last two are also used by qcow2_snapshot_create() to discard the clusters used by the VM state. In this case there's no risk of exposing stale data to the guest and we really want that the clusters are always discarded.

---
  block/qcow2.c              |  6 +++
  tests/qemu-iotests/060     |  5 ++-
  tests/qemu-iotests/060.out |  2 -
  tests/qemu-iotests/289     | 90 ++++++++++++++++++++++++++++++++++++++
  tests/qemu-iotests/289.out | 52 ++++++++++++++++++++++
  tests/qemu-iotests/group   |  1 +
  6 files changed, 152 insertions(+), 4 deletions(-)
Quoted 2 lines

create mode 100755 tests/qemu-iotests/289
create mode 100644 tests/qemu-iotests/289.out

The actual fix is much smaller than the iotest fallout ;)

tests/qemu-iotests/060 _make_test_img -o 'compat=0.10' -b "$BACKING_IMG" 1G · line 173

Quoted 11 lines

+++ b/tests/qemu-iotests/060

tests/qemu-iotests/060 +3 -2
@@ -167,9 +167,10 @@ _make_test_img -o 'compat=0.10' -b "$BACKING_IMG" 1G
  # Write two clusters, the second one enforces creation of an L2 table after
  # the first data cluster.
  $QEMU_IO -c 'write 0k 64k' -c 'write 512M 64k' "$TEST_IMG" | _filter_qemu_io
-# Discard the first cluster. This cluster will soon enough be reallocated and
+# Free the first cluster. This cluster will soon enough be reallocated and
  # used for COW.
-$QEMU_IO -c 'discard 0k 64k' "$TEST_IMG" | _filter_qemu_io
+poke_file "$TEST_IMG" '262144' "\x00\x00\x00\x00\x00\x00\x00\x00" # 0x40000 - L2 entry
+poke_file "$TEST_IMG" '131082' "\x00\x00" # 0x2000a - Refcount entry

Instead of writing '262144' ... # 0x40000, you could write $((0x40000)) in-place. Similarly for 131082 vs. 0x2000a.

Also, Max has pending patches for adding poke_file_be; if those land first, this becomes simpler as:

poke_file_be "$TEST_IMG" $((0x40000)) 8 0 # L2 entry poke_file_be "$TEST_IMG" $((0x2000a)) 2 0 # Refcount entry

tests/qemu-iotests/289 line 3

Quoted 5 lines
tests/qemu-iotests/289 new file +4 -0
+++ b/tests/qemu-iotests/289
@@ -0,0 +1,90 @@
+#!/usr/bin/env bash
+#
+# Test how 'qemu-io -c discard' behaves on v2 and v3 qcow2 images

At any rate, the new test looks reasonable to me. I see you have other review comments for improving it, with thos in, you can add

Signature · 3 lines
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org
#3 lore raw
Quoted 6 lines · Eric Blake

tests/qemu-iotests/060 _make_test_img -o 'compat=0.10' -b "$BACKING_IMG" 1G · line 173

Quoted 3 lines

-$QEMU_IO -c 'discard 0k 64k' "$TEST_IMG" | _filter_qemu_io

tests/qemu-iotests/060 +2 -0
+poke_file "$TEST_IMG" '262144' "\x00\x00\x00\x00\x00\x00\x00\x00" # 0x40000 - L2 entry
+poke_file "$TEST_IMG" '131082' "\x00\x00" # 0x2000a - Refcount entry

Instead of writing '262144' ... # 0x40000, you could write $((0x40000)) in-place. Similarly for 131082 vs. 0x2000a.

The exiting poke_file line in that test was using base 10 so I decided to use it too for consistency.

I actually realized that $rb_offset and $l2_offset are defined, so I could use those too.

Quoted 5 lines

Also, Max has pending patches for adding poke_file_be; if those land first, this becomes simpler as:

poke_file_be "$TEST_IMG" $((0x40000)) 8 0 # L2 entry poke_file_be "$TEST_IMG" $((0x2000a)) 2 0 # Refcount entry

I'm fine if those lines are changed when the patch is committed.

Berto