Fix qcow2+luks corruption introduced by commit 8ac0f15f335
v3 cover Fix qcow2+luks corruption introduced by commit 8ac0f15f335
Commit 8ac0f15f335 accidently broke the COW of non changed areas of newly allocated clusters, when the write spans multiple clusters, and needs COW both prior and after the write.
This results in 'after' COW area being encrypted with wrong sector address, which render it corrupted.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1745922
V2: grammar, spelling and code style fixes.
V3: more fixes after the review.
Best regards,
Maxim Levitsky
Maxim Levitsky (3):
block/qcow2: refactoring of threaded encryption code block/qcow2: fix the corruption when rebasing luks encrypted files
qemu-iotests: Add test for bz #1745922
block/qcow2-cluster.c | 29 +++++++----- block/qcow2-threads.c | 63 ++++++++++++++++++++------ tests/qemu-iotests/263 | 91 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/263.out | 40 +++++++++++++++++ tests/qemu-iotests/group | 1 + 5 files changed, 199 insertions(+), 25 deletions(-) create mode 100755 tests/qemu-iotests/263 create mode 100644 tests/qemu-iotests/263.out
Signature · 1 line
2.17.2
PATCH v3 1/3 block/qcow2: refactoring of threaded encryption code
This commit tries to clarify few function arguments, and add comments describing the encrypt/decrypt interface
--- block/qcow2-cluster.c | 8 +++--- block/qcow2-threads.c | 63 ++++++++++++++++++++++++++++++++++--------- 2 files changed, 54 insertions(+), 17 deletions(-)
block/qcow2-cluster.c +4 -4
@@ -463,8 +463,8 @@ static int coroutine_fn do_perform_cow_read(BlockDriverState *bs, } static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs, -uint64_t src_cluster_offset, -uint64_t cluster_offset, +uint64_t guest_cluster_offset, +uint64_t host_cluster_offset,unsigned offset_in_cluster,uint8_t *buffer,unsigned bytes)
@@ -474,8 +474,8 @@ static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs,assert ((offset_in_cluster & ~BDRV_SECTOR_MASK ) ==0 );assert ((bytes & ~BDRV_SECTOR_MASK ) ==0 );assert (s->crypto); -if (qcow2_co_encrypt (bs, cluster_offset, - src_cluster_offset + offset_in_cluster, +if (qcow2_co_encrypt (bs, host_cluster_offset, + guest_cluster_offset + offset_in_cluster, buffer, bytes) <0 ) {return false; }
block/qcow2-threads.c +50 -13
83 lines not on this page · raw message
Signature · 1 line
2.17.2
Re: [Qemu-devel] [PATCH v3 1/3] block/qcow2: refactoring of threaded encryption code
Quoted 67 lines · Maxim Levitsky
This commit tries to clarify few function arguments, and add comments describing the encrypt/decrypt interface
Signed-off-by Maxim Levitsky <[email protected]>--- block/qcow2-cluster.c | 8 +++--- block/qcow2-threads.c | 63 ++++++++++++++++++++++++++++++++++--------- 2 files changed, 54 insertions(+), 17 deletions(-)block/qcow2-cluster.c +4
-4@@ -463,8 +463,8 @@ static int coroutine_fn do_perform_cow_read(BlockDriverState *bs, } static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs, -uint64_t src_cluster_offset, -uint64_t cluster_offset, +uint64_t guest_cluster_offset, +uint64_t host_cluster_offset,unsigned offset_in_cluster,uint8_t *buffer,unsigned bytes)@@ -474,8 +474,8 @@ static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs,assert ((offset_in_cluster & ~BDRV_SECTOR_MASK ) ==0 );assert ((bytes & ~BDRV_SECTOR_MASK ) ==0 );assert (s->crypto); -if (qcow2_co_encrypt (bs, cluster_offset, - src_cluster_offset + offset_in_cluster, +if (qcow2_co_encrypt (bs, host_cluster_offset, + guest_cluster_offset + offset_in_cluster, buffer, bytes) <0 ) {return false; }block/qcow2-threads.c +9
-5@@ -234,15 +234,19 @@ static int qcow2_encdec_pool_func(void *opaque) } static int coroutine_fn -qcow2_co_encdec (BlockDriverState *bs,uint64_t file_cluster_offset, -uint64_t offset,void *buf,size_t len,Qcow2EncDecFunc func) +qcow2_co_encdec (BlockDriverState *bs,uint64_t host_cluster_offset, +uint64_t guest_offset,void *buf,size_t len, +Qcow2EncDecFunc func) {BDRVQcow2State *s = bs->opaque; + +uint64_t offset = s->crypt_physical_offset ? + host_cluster_offset +offset_into_cluster (s, guest_offset) : + guest_offset; +Qcow2EncDecData arg = { .block = s->crypto, - .offset = s->crypt_physical_offset ? - file_cluster_offset +offset_into_cluster (s, offset) : - offset, + .offset = offset, .buf = buf, .len = len, .func = func,@@ -251,18 +255,51 @@ qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset,return qcow2_co_process (bs, qcow2_encdec_pool_func, &arg); }
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 266
Quoted 10 lines
block/qcow2-threads.c +9
-0+ +/* +* qcow2_co_encrypt() +* +* Encrypts one or more contiguous aligned sectors +* +* @host_cluster_offset - underlying storage offset of the first cluster +* in which the encrypted data will be written +* Used as an initialization vector for encryption
s/as an/for generating the/
(AFAIU)
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 270
Quoted 4 lines
block/qcow2-threads.c +4
-0+* +* @guest_offset - guest (virtual) offset of the first sector of the +* data to be encrypted +* Used as an initialization vector for older, qcow2 native encryption
I wouldn’t be so specific here. I think it’d be better to just have a common sentence like “Depending on the encryption method, either of these offsets may be used for generating the initialization vector for encryption.”
Well, technically, host_cluster_offset will not be used itself. Before we can use it, of course we have to add the in-cluster offset to it (which qcow2_co_encdec() does).
This makes me wonder whether it wouldn’t make more sense to pass a host_offset instead of a host_cluster_offset (i.e. make the callers add the in-cluster offset to the host offset already)?
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 274
Quoted 4 lines
block/qcow2-threads.c +4
-0+* +* @buf - buffer with the data to encrypt, that after encryption +* will be written to the underlying storage device at +* @host_cluster_offset
I think just “buffer with the data to encrypt” is sufficient. (The rest is just the same as above.)
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 276
Quoted 2 lines
block/qcow2-threads.c +2
-0+* +* @len - length of the buffer (in sector size multiplies)
“In sector size multiples” to me means that it is a sector count (in that “one sector size multiple” is equal to 512 byes).
Maybe “must be a BDRV_SECTOR_SIZE multiple” instead?
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 279
Quoted 3 lines
block/qcow2-threads.c +3
-0+* +* Note that the group of the sectors, don't have to be aligned +* on cluster boundary and can also cross a cluster boundary.
Maybe “Note that while the whole range must be aligned on sectors, it does not have to be aligned on clusters and can also cross cluster boundaries”?
(“The group of sectors” sounds a bit weird to me. I don’t quite know, why. I think that for some reason it makes me think of a non-continuous set of sectors. Also, the caller doesn’t pass sector indices, but byte offsets, that just happen to have to be aligned to sectors. (I suppose because that’s the simplest way to make it a multiple of the encryption block size.))
Quoted 10 lines
block/qcow2-threads.c +4
-4+* +*/ int coroutine_fn -qcow2_co_encrypt (BlockDriverState *bs,uint64_t file_cluster_offset, -uint64_t offset,void *buf,size_t len) +qcow2_co_encrypt (BlockDriverState *bs,uint64_t host_cluster_offset, +uint64_t guest_offset,void *buf,size_t len) { -return qcow2_co_encdec (bs, file_cluster_offset, offset, buf, len, - qcrypto_block_encrypt);
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 295
Quoted 10 lines
block/qcow2-threads.c +8
-0+return qcow2_co_encdec (bs, host_cluster_offset, guest_offset, buf, len, + qcrypto_block_encrypt); } + +/* +* qcow2_co_decrypt() +* +* Decrypts one or more contiguous aligned sectors +* Similar to qcow2_co_encrypt
Hm. This would make me wonder in what way it is only similar to qcow2_co_encrypt(). Sure, it decrypts instead of encrypting, but maybe there’s more...?
So maybe “Its interface is the same as qcow2_co_encrypt()'s”?
Max
Quoted 15 lines
block/qcow2-threads.c +7
-4+* +*/ + int coroutine_fn -qcow2_co_decrypt (BlockDriverState *bs,uint64_t file_cluster_offset, -uint64_t offset,void *buf,size_t len) +qcow2_co_decrypt (BlockDriverState *bs,uint64_t host_cluster_offset, +uint64_t guest_offset,void *buf,size_t len) { -return qcow2_co_encdec (bs, file_cluster_offset, offset, buf, len, - qcrypto_block_decrypt); +return qcow2_co_encdec (bs, host_cluster_offset, guest_offset, buf, len, + qcrypto_block_decrypt); }
Re: [Qemu-devel] [PATCH v3 1/3] block/qcow2: refactoring of threaded encryption code
Quoted 81 lines · Max Reitz
Quoted 67 lines · Maxim Levitsky
This commit tries to clarify few function arguments, and add comments describing the encrypt/decrypt interface
Signed-off-by Maxim Levitsky <[email protected]>--- block/qcow2-cluster.c | 8 +++--- block/qcow2-threads.c | 63 ++++++++++++++++++++++++++++++++++--------- 2 files changed, 54 insertions(+), 17 deletions(-)block/qcow2-cluster.c +4
-4@@ -463,8 +463,8 @@ static int coroutine_fn do_perform_cow_read(BlockDriverState *bs, } static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs, -uint64_t src_cluster_offset, -uint64_t cluster_offset, +uint64_t guest_cluster_offset, +uint64_t host_cluster_offset,unsigned offset_in_cluster,uint8_t *buffer,unsigned bytes)@@ -474,8 +474,8 @@ static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs,assert ((offset_in_cluster & ~BDRV_SECTOR_MASK ) ==0 );assert ((bytes & ~BDRV_SECTOR_MASK ) ==0 );assert (s->crypto); -if (qcow2_co_encrypt (bs, cluster_offset, - src_cluster_offset + offset_in_cluster, +if (qcow2_co_encrypt (bs, host_cluster_offset, + guest_cluster_offset + offset_in_cluster, buffer, bytes) <0 ) {return false; }block/qcow2-threads.c +9
-5@@ -234,15 +234,19 @@ static int qcow2_encdec_pool_func(void *opaque) } static int coroutine_fn -qcow2_co_encdec (BlockDriverState *bs,uint64_t file_cluster_offset, -uint64_t offset,void *buf,size_t len,Qcow2EncDecFunc func) +qcow2_co_encdec (BlockDriverState *bs,uint64_t host_cluster_offset, +uint64_t guest_offset,void *buf,size_t len, +Qcow2EncDecFunc func) {BDRVQcow2State *s = bs->opaque; + +uint64_t offset = s->crypt_physical_offset ? + host_cluster_offset +offset_into_cluster (s, guest_offset) : + guest_offset; +Qcow2EncDecData arg = { .block = s->crypto, - .offset = s->crypt_physical_offset ? - file_cluster_offset +offset_into_cluster (s, offset) : - offset, + .offset = offset, .buf = buf, .len = len, .func = func,@@ -251,18 +255,51 @@ qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset,return qcow2_co_process (bs, qcow2_encdec_pool_func, &arg); }block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 266
Quoted 10 lines
block/qcow2-threads.c +9
-0+ +/* +* qcow2_co_encrypt() +* +* Encrypts one or more contiguous aligned sectors +* +* @host_cluster_offset - underlying storage offset of the first cluster +* in which the encrypted data will be written +* Used as an initialization vector for encryption s/as an/for generating the/
Quoted 1 line
(AFAIU)
Yes, this is better.
Quoted 10 lines
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 270
Quoted 5 lines
block/qcow2-threads.c +4
-0+* +* @guest_offset - guest (virtual) offset of the first sector of the +* data to be encrypted +* Used as an initialization vector for older, qcow2 native encryption I wouldn’t be so specific here. I think it’d be better to just have a common sentence like “Depending on the encryption method, either of these offsets may be used for generating the initialization vector for encryption.”
Nothing against this either.
Quoted 8 lines
Well, technically,
host_cluster_offsetwill not be used itself. Before we can use it, of course we have to add the in-cluster offset to it (whichqcow2_co_encdec()does).This makes me wonder whether it wouldn’t make more sense to pass a
host_offsetinstead of ahost_cluster_offset(i.e. make the callers add the in-cluster offset to the host offset already)?
This is what I wanted to do in first place, and it would be the cleanest solution, but that would need to update the 3rd caller of this function, the qcow2_co_pwritev_part, which does pass the cluster offset and guest full offset. You know what, I'll just do it. A bit more changes, but much cleaner code that eliminates the possibility of this bug of happening again.
Quoted 8 lines
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 274
Quoted 5 lines
block/qcow2-threads.c +4
-0+* +* @buf - buffer with the data to encrypt, that after encryption +* will be written to the underlying storage device at +* @host_cluster_offset I think just “buffer with the data to encrypt” is sufficient. (The rest is just the same as above.)
I wrote it to clarify this since Vladimir told me that its not clear enough. Note though that I made a mistake here since the data will be written not at host_cluster_offset but in host_cluster_offset + offset_into_cluster(s, guest_offset
Quoted 6 lines
+ * @len - length of the buffer (in sector size multiplies)
“In sector size multiples” to me means that it is a sector count (in that “one sector size multiple” is equal to 512 byes).
Maybe “must be a
BDRV_SECTOR_SIZEmultiple” instead?
All right.
Quoted 9 lines
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 279
Quoted 4 lines
block/qcow2-threads.c +3
-0+* +* Note that the group of the sectors, don't have to be aligned +* on cluster boundary and can also cross a cluster boundary. Maybe “Note that while the whole range must be aligned on sectors, it does not have to be aligned on clusters and can also cross cluster boundaries”?
Quoted 6 lines
(“The group of sectors” sounds a bit weird to me. I don’t quite know, why. I think that for some reason it makes me think of a non-continuous set of sectors. Also, the caller doesn’t pass sector indices, but byte offsets, that just happen to have to be aligned to sectors. (I suppose because that’s the simplest way to make it a multiple of the encryption block size.))
All right, this sounds better
Quoted 15 lines
Quoted 15 lines
block/qcow2-threads.c +6
-4+* +*/ int coroutine_fn -qcow2_co_encrypt (BlockDriverState *bs,uint64_t file_cluster_offset, -uint64_t offset,void *buf,size_t len) +qcow2_co_encrypt (BlockDriverState *bs,uint64_t host_cluster_offset, +uint64_t guest_offset,void *buf,size_t len) { -return qcow2_co_encdec (bs, file_cluster_offset, offset, buf, len, - qcrypto_block_encrypt); +return qcow2_co_encdec (bs, host_cluster_offset, guest_offset, buf, len, + qcrypto_block_encrypt); }
Quoted 10 lines
block/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 295
Quoted 6 lines
block/qcow2-threads.c +6
-0+ +/* +* qcow2_co_decrypt() +* +* Decrypts one or more contiguous aligned sectors +* Similar to qcow2_co_encrypt Hm. This would make me wonder in what way it is only similar to
qcow2_co_encrypt(). Sure, it decrypts instead of encrypting, but maybe there’s more...?
I don't think so, since we have symmetrical encryption here.
Quoted 2 lines
So maybe “Its interface is the same as qcow2_co_encrypt()'s”?
That would be lawyer territory, since interface is not technically the same, since it decrypts rather that encrypts the data in the buffer... I think that this wording should be good enough.
Quoted 20 lines
Max
Quoted 5 lines
block/qcow2-threads.c +3
-0+* +*/ +int coroutine_fnblock/qcow2-threads.c qcow2_co_encdec(BlockDriverState *bs, uint64_t file_cluster_offset, · line 305
Quoted 10 lines
block/qcow2-threads.c +4
-4-qcow2_co_decrypt(BlockDriverState *bs, uint64_t file_cluster_offset, - uint64_t offset, void *buf, size_t len) +qcow2_co_decrypt(BlockDriverState *bs, uint64_t host_cluster_offset, + uint64_t guest_offset, void *buf, size_t len) { -return qcow2_co_encdec (bs, file_cluster_offset, offset, buf, len, - qcrypto_block_decrypt); +return qcow2_co_encdec (bs, host_cluster_offset, guest_offset, buf, len, + qcrypto_block_decrypt); }
Best regards,
Maxim Levitsky
PATCH v3 2/3 block/qcow2: fix the corruption when rebasing luks encrypted files
In reply to Maxim LevitskyThis fixes subtle corruption introduced by luks threaded encryption in commit 8ac0f15f335
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1745922
The corruption happens when we do a write that
* writes to two or more unallocated clusters at once
* doesn't fully cover the first sector
* doesn't fully cover the last sector
In this case, when allocating the new clusters we COW both areas prior to the write and after the write, and we encrypt them.
The above mentioned commit accidentally made it so we encrypt the second COW area using the physical cluster offset of the first area.
Fix this by:
* Remove the offset_in_cluster parameter of do_perform_cow_encrypt, since it is misleading. That offset can be larger than cluster size currently.
Instead just add the start and the end COW area offsets to both host and guest offsets that do_perform_cow_encrypt receives.
* in do_perform_cow_encrypt, remove the cluster offset from the host_offset, and thus pass correctly to the qcow2_co_encrypt, the host cluster offset and full guest offset
In the bugreport that was triggered by rebasing a luks image to new, zero filled base, which lot of such writes, and causes some files with zero areas to contain garbage there instead.
But as described above it can happen elsewhere as well
--- block/qcow2-cluster.c | 29 +++++++++++++++++------------ 1 file changed, 17 insertions(+), 12 deletions(-)
block/qcow2-cluster.c +17 -12
@@ -463,20 +463,21 @@ static int coroutine_fn do_perform_cow_read(BlockDriverState *bs, } static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs, -uint64_t guest_cluster_offset, -uint64_t host_cluster_offset, -unsigned offset_in_cluster, +uint64_t guest_offset, +uint64_t host_offset,uint8_t *buffer,unsigned bytes) {if (bytes && bs->encrypted) {BDRVQcow2State *s = bs->opaque; -assert ((offset_in_cluster & ~BDRV_SECTOR_MASK ) ==0 ); -assert ((bytes & ~BDRV_SECTOR_MASK ) ==0 ); + +assert (QEMU_IS_ALIGNED (guest_offset,BDRV_SECTOR_SIZE )); +assert (QEMU_IS_ALIGNED (host_offset,BDRV_SECTOR_SIZE )); +assert (QEMU_IS_ALIGNED (bytes,BDRV_SECTOR_SIZE ));assert (s->crypto); -if (qcow2_co_encrypt (bs, host_cluster_offset, - guest_cluster_offset + offset_in_cluster, - buffer, bytes) <0 ) { + +if (qcow2_co_encrypt (bs,start_of_cluster (s, host_offset), + guest_offset, buffer, bytes) <0 ) {return false; } }
@@ -890,11 +891,15 @@ static int perform_cow(BlockDriverState *bs, QCowL2Meta *m)/* Encrypt the data if necessary before writing it */ if (bs->encrypted) { -if (!do_perform_cow_encrypt (bs, m->offset, m->alloc_offset, - start->offset, start_buffer, +if (!do_perform_cow_encrypt (bs, + m->offset + start->offset, + m->alloc_offset + start->offset, + start_buffer, start->nb_bytes) || - !do_perform_cow_encrypt (bs, m->offset, m->alloc_offset, - end->offset, end_buffer, end->nb_bytes)) { + !do_perform_cow_encrypt (bs, + m->offset + end->offset, + m->alloc_offset + end->offset, + end_buffer, end->nb_bytes)) { ret = -EIO ; goto fail; }
Signature · 1 line
2.17.2
Re: [Qemu-devel] [PATCH v3 2/3] block/qcow2: fix the corruption when rebasing luks encrypted files
Quoted 34 lines · Maxim Levitsky
This fixes subtle corruption introduced by luks threaded encryption in commit
8ac0f15f335Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1745922
The corruption happens when we do a write that
* writes to two or more unallocated clusters at once
* doesn't fully cover the first sector
* doesn't fully cover the last sectorIn this case, when allocating the new clusters we COW both areas prior to the write and after the write, and we encrypt them.
The above mentioned commit accidentally made it so we encrypt the second COW area using the physical cluster offset of the first area.
Fix this by:
* Remove theoffset_in_clusterparameter ofdo_perform_cow_encrypt, since it is misleading. That offset can be larger than cluster size currently.Instead just add the start and the end COW area offsets to both host and guest offsets that
do_perform_cow_encryptreceives.* in
do_perform_cow_encrypt, remove the cluster offset from thehost_offset, and thus pass correctly to theqcow2_co_encrypt, the host cluster offset and full guest offsetIn the bugreport that was triggered by rebasing a luks image to new, zero filled base, which lot of such writes, and causes some files with zero areas to contain garbage there instead.
But as described above it can happen elsewhere as well
Quoted 5 lines
Signed-off-by Maxim Levitsky <[email protected]>Reviewed-by Vladimir Sementsov-Ogievskiy <[email protected]>--- block/qcow2-cluster.c | 29 +++++++++++++++++------------ 1 file changed, 17 insertions(+), 12 deletions(-)
PATCH v3 3/3 qemu-iotests: Add test for bz #1745922
In reply to Maxim Levitsky--- tests/qemu-iotests/263 | 91 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/263.out | 40 +++++++++++++++++ tests/qemu-iotests/group | 1 + 3 files changed, 132 insertions(+) create mode 100755 tests/qemu-iotests/263 create mode 100644 tests/qemu-iotests/263.out
tests/qemu-iotests/263 new file +91 -0
@@ -0,0 +1,91 @@ +#!/usr/bin/env bash +# +# Test encrypted write that crosses cluster boundary of two unallocated clusters +# Based on 188 +# +# Copyright (C) 2019 Red Hat, Inc. +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +# creator [email protected] + +seq=`basename $0` +echo "QA output created by $seq" + +status=1 # failure is the default! + +_cleanup() +{ + _cleanup_test_img +} +trap "_cleanup; exit \$status" 0 1 2 3 15 + +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +_supported_fmt qcow2 +_supported_proto generic +_supported_os Linux + + +size=1M + +SECRET="secret,id=sec0,data=astrochicken" +QEMU_IO_OPTIONS=$QEMU_IO_OPTIONS_NO_FMT + + +_run_test() +{ + echo "== reading the whole image ==" + $QEMU_IO --object $SECRET -c "read -P 0 0 $size" --image-opts $1 | _filter_qemu_io | _filter_testdir + + echo + echo "== write two 512 byte sectors on a cluster boundary ==" + $QEMU_IO --object $SECRET -c "write -P 0xAA 0xFE00 0x400" --image-opts $1 | _filter_qemu_io | _filter_testdir + + echo + echo "== verify that the rest of the image is not changed ==" + $QEMU_IO --object $SECRET -c "read -P 0x00 0x00000 0xFE00" --image-opts $1 | _filter_qemu_io | _filter_testdir + $QEMU_IO --object $SECRET -c "read -P 0xAA 0x0FE00 0x400" --image-opts $1 | _filter_qemu_io | _filter_testdir + $QEMU_IO --object $SECRET -c "read -P 0x00 0x10200 0xEFE00" --image-opts $1 | _filter_qemu_io | _filter_testdir + +} + + +echo +echo "testing LUKS qcow2 encryption" +echo + +_make_test_img --object $SECRET -o "encrypt.format=luks,encrypt.key-secret=sec0,encrypt.iter-time=10,cluster_size=64K" $size +_run_test "driver=$IMGFMT,encrypt.key-secret=sec0,file.filename=$TEST_IMG" +_cleanup_test_img + +echo +echo "testing legacy AES qcow2 encryption" +echo + + +_make_test_img --object $SECRET -o "encrypt.format=aes,encrypt.key-secret=sec0,cluster_size=64K" $size +_run_test "driver=$IMGFMT,encrypt.key-secret=sec0,file.filename=$TEST_IMG" +_cleanup_test_img + + + +# success, all done +echo "*** done" +rm -f $seq.full +status=0
tests/qemu-iotests/263.out new file +40 -0
@@ -0,0 +1,40 @@ +QA output created by 263 + +testing LUKS qcow2 encryption + +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576 encrypt.format=luks encrypt.key-secret=sec0 encrypt.iter-time=10 +== reading the whole image == +read 1048576/1048576 bytes at offset 0 +1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) + +== write two 512 byte sectors on a cluster boundary == +wrote 1024/1024 bytes at offset 65024 +1 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) + +== verify that the rest of the image is not changed == +read 65024/65024 bytes at offset 0 +63.500 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 1024/1024 bytes at offset 65024 +1 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 982528/982528 bytes at offset 66048 +959.500 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) + +testing legacy AES qcow2 encryption + +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576 encrypt.format=aes encrypt.key-secret=sec0 +== reading the whole image == +read 1048576/1048576 bytes at offset 0 +1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) + +== write two 512 byte sectors on a cluster boundary == +wrote 1024/1024 bytes at offset 65024 +1 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) + +== verify that the rest of the image is not changed == +read 65024/65024 bytes at offset 0 +63.500 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 1024/1024 bytes at offset 65024 +1 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +read 982528/982528 bytes at offset 66048 +959.500 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +*** done
tests/qemu-iotests/group +1 -0
@@ -274,3 +274,4 @@ 257 rw 258 rw quick 262 rw quick migration +263 rw quick
Signature · 1 line
2.17.2
Re: [Qemu-devel] [PATCH v3 3/3] qemu-iotests: Add test for bz #1745922
Quoted 69 lines · Maxim Levitsky
Signed-off-by Maxim Levitsky <[email protected]>--- tests/qemu-iotests/263 | 91 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/263.out | 40 +++++++++++++++++ tests/qemu-iotests/group | 1 + 3 files changed, 132 insertions(+) create mode 100755 tests/qemu-iotests/263 create mode 100644 tests/qemu-iotests/263.outtests/qemu-iotests/263 new file +54
-0@@ -0,0 +1,91 @@ +#!/usr/bin/env bash +# +# Test encrypted write that crosses cluster boundary of two unallocated clusters +# Based on 188 +# +# Copyright (C) 2019 Red Hat, Inc. +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +# creator [email protected] + +seq=`basename $0` +echo "QA output created by $seq" + +status=1 # failure is the default! + +_cleanup() +{ + _cleanup_test_img +} +trap "_cleanup; exit \$status" 0 1 2 3 15 + +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +_supported_fmt qcow2 +_supported_proto generic +_supported_os Linux + + +size=1M + +SECRET="secret,id=sec0,data=astrochicken" +QEMU_IO_OPTIONS=$QEMU_IO_OPTIONS_NO_FMT + + +_run_test() +{ + echo "== reading the whole image ==" + $QEMU_IO --object $SECRET -c "read -P 0 0 $size" --image-opts $1 | _filter_qemu_io | _filter_testdir
tests/qemu-iotests/263 line 64
Quoted 10 lines
tests/qemu-iotests/263 new file +10
-0+ + echo + echo "== write two 512 byte sectors on a cluster boundary ==" + $QEMU_IO --object $SECRET -c "write -P 0xAA 0xFE00 0x400" --image-opts $1 | _filter_qemu_io | _filter_testdir + + echo + echo "== verify that the rest of the image is not changed ==" + $QEMU_IO --object $SECRET -c "read -P 0x00 0x00000 0xFE00" --image-opts $1 | _filter_qemu_io | _filter_testdir + $QEMU_IO --object $SECRET -c "read -P 0xAA 0x0FE00 0x400" --image-opts $1 | _filter_qemu_io | _filter_testdir + $QEMU_IO --object $SECRET -c "read -P 0x00 0x10200 0xEFE00" --image-opts $1 | _filter_qemu_io | _filter_testdir
Looks good to me overall, but in case you respin, I don’t think there’s reason not to quote the $1.
Max
Re: [Qemu-devel] [PATCH v3 3/3] qemu-iotests: Add test for bz #1745922
Quoted 84 lines · Max Reitz
Quoted 69 lines · Maxim Levitsky
Signed-off-by Maxim Levitsky <[email protected]>--- tests/qemu-iotests/263 | 91 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/263.out | 40 +++++++++++++++++ tests/qemu-iotests/group | 1 + 3 files changed, 132 insertions(+) create mode 100755 tests/qemu-iotests/263 create mode 100644 tests/qemu-iotests/263.outtests/qemu-iotests/263 new file +54
-0@@ -0,0 +1,91 @@ +#!/usr/bin/env bash +# +# Test encrypted write that crosses cluster boundary of two unallocated clusters +# Based on 188 +# +# Copyright (C) 2019 Red Hat, Inc. +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +# creator [email protected] + +seq=`basename $0` +echo "QA output created by $seq" + +status=1 # failure is the default! + +_cleanup() +{ + _cleanup_test_img +} +trap "_cleanup; exit \$status" 0 1 2 3 15 + +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +_supported_fmt qcow2 +_supported_proto generic +_supported_os Linux + + +size=1M + +SECRET="secret,id=sec0,data=astrochicken" +QEMU_IO_OPTIONS=$QEMU_IO_OPTIONS_NO_FMT + + +_run_test() +{ + echo "== reading the whole image ==" + $QEMU_IO --object $SECRET -c "read -P 0 0 $size" --image-opts $1 | _filter_qemu_io | _filter_testdirtests/qemu-iotests/263 line 64
Quoted 10 lines
tests/qemu-iotests/263 new file +10
-0+ + echo + echo "== write two 512 byte sectors on a cluster boundary ==" + $QEMU_IO --object $SECRET -c "write -P 0xAA 0xFE00 0x400" --image-opts $1 | _filter_qemu_io | _filter_testdir + + echo + echo "== verify that the rest of the image is not changed ==" + $QEMU_IO --object $SECRET -c "read -P 0x00 0x00000 0xFE00" --image-opts $1 | _filter_qemu_io | _filter_testdir + $QEMU_IO --object $SECRET -c "read -P 0xAA 0x0FE00 0x400" --image-opts $1 | _filter_qemu_io | _filter_testdir + $QEMU_IO --object $SECRET -c "read -P 0x00 0x10200 0xEFE00" --image-opts $1 | _filter_qemu_io | _filter_testdirLooks good to me overall, but in case you respin, I don’t think there’s reason not to quote the $1.
Quoted 2 lines
Max
Fixed, good idea.
Best regards,
Maxim Levitsky