Fix qcow2+luks corruption introduced by commit 8ac0f15f335
v7 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.
V4: addressed review comments from Max Reitz,
and futher refactored the qcow2_co_encrypt to just take full host and guest offset which simplifies everything.
V5: reworked the patches so one of them fixes the bug
only and other one is just refactoring
V6: removed do_perform_cow_encrypt
V7: removed do_perform_cow_encrypt take two, this
time I hopefully did that correctly :-)
Also updated commit names and messages a bit
Best regards,
Maxim Levitsky
Maxim Levitsky (3):
block/qcow2: Fix corruption introduced by commit 8ac0f15f335 block/qcow2: refactor encryption code
qemu-iotests: Add test for bz #1745922
block/qcow2-cluster.c | 40 ++++++----------- block/qcow2-threads.c | 63 ++++++++++++++++++++------ block/qcow2.c | 5 ++- block/qcow2.h | 8 ++-- tests/qemu-iotests/263 | 91 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/263.out | 40 +++++++++++++++++ tests/qemu-iotests/group | 1 + 7 files changed, 202 insertions(+), 46 deletions(-) create mode 100755 tests/qemu-iotests/263 create mode 100644 tests/qemu-iotests/263.out
Signature · 1 line
2.17.2
PATCH v7 1/3 block/qcow2: Fix corruption introduced by commit 8ac0f15f335
This 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
* uses luks encryption
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.
The problem is that offset_in_cluster in do_perform_cow_encrypt can be larger that the cluster size, thus cluster_offset will no longer point to the start of the cluster at which encrypted area starts.
Next patch in this series will refactor the code to avoid all these assumptions.
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 | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
block/qcow2-cluster.c +4 -3
@@ -474,9 +474,10 @@ 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, - buffer, bytes) <0 ) { +if (qcow2_co_encrypt (bs, +start_of_cluster (s, cluster_offset + offset_in_cluster), + src_cluster_offset + offset_in_cluster, + buffer, bytes) <0 ) {return false; } }
Signature · 1 line
2.17.2
Re: [Qemu-devel] [PATCH v7 1/3] block/qcow2: Fix corruption introduced by commit 8ac0f15f335
Quoted 31 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 sector
* uses luks encryptionIn 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.
The problem is that
offset_in_clusterindo_perform_cow_encryptcan be larger that the cluster size, thuscluster_offsetwill no longer point to the start of the cluster at which encrypted area starts.Next patch in this series will refactor the code to avoid all these assumptions.
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
Quoted 5 lines
Signed-off-by Maxim Levitsky <[email protected]>Reviewed-by Vladimir Sementsov-Ogievskiy <[email protected]>--- block/qcow2-cluster.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)
PATCH v7 2/3 block/qcow2: refactor encryption code
In reply to Maxim Levitsky* Change the qcow2_co_{encrypt|decrypt} to just receive full host and guest offsets and use this function directly instead of calling do_perform_cow_encrypt (which is removed by that patch).
* Adjust qcow2_co_encdec to take full host and guest offsets as well.
* Document the qcow2_co_{encrypt|decrypt} arguments to prevent the bug fixed in former commit from hopefully happening again.
--- block/qcow2-cluster.c | 41 ++++++++++------------------ block/qcow2-threads.c | 63 +++++++++++++++++++++++++++++++++---------- block/qcow2.c | 5 ++-- block/qcow2.h | 8 +++--- 4 files changed, 70 insertions(+), 47 deletions(-)
block/qcow2-cluster.c +14 -27
@@ -462,28 +462,6 @@ static int coroutine_fn do_perform_cow_read(BlockDriverState *bs,return 0 ; } -static bool coroutine_fn do_perform_cow_encrypt(BlockDriverState *bs, -uint64_t src_cluster_offset, -uint64_t cluster_offset, -unsigned offset_in_cluster, -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 (s->crypto); -if (qcow2_co_encrypt (bs, -start_of_cluster (s, cluster_offset + offset_in_cluster), - src_cluster_offset + offset_in_cluster, - buffer, bytes) <0 ) { -return false; - } - } -return true; -} - static int coroutine_fn do_perform_cow_write(BlockDriverState *bs, uint64_t cluster_offset, unsigned offset_in_cluster,
@@ -891,11 +869,20 @@ 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, - start->nb_bytes) || - !do_perform_cow_encrypt (bs, m->offset, m->alloc_offset, - end->offset, end_buffer, end->nb_bytes)) { + ret =qcow2_co_encrypt (bs, + m->alloc_offset + start->offset, + m->offset + start->offset, + start_buffer, start->nb_bytes); +if (ret <0 ) { + ret = -EIO ; + goto fail; + } + + ret =qcow2_co_encrypt (bs, + m->alloc_offset + end->offset, + m->offset + end->offset, + end_buffer, end->nb_bytes); +if (ret <0 ) { ret = -EIO ; goto fail; }
block/qcow2-threads.c +49 -14
84 lines not on this page · raw message
block/qcow2.c +3 -2
@@ -2069,7 +2069,8 @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,assert ((offset & (BDRV_SECTOR_SIZE -1 )) ==0 );assert ((cur_bytes & (BDRV_SECTOR_SIZE -1 )) ==0 ); -if (qcow2_co_decrypt (bs, cluster_offset, offset, +if (qcow2_co_decrypt (bs, cluster_offset + offset_in_cluster, + offset, cluster_data, cur_bytes) <0 ) { ret = -EIO ; goto fail;
@@ -2288,7 +2289,7 @@ static coroutine_fn int qcow2_co_pwritev_part(qemu_iovec_to_buf (qiov, qiov_offset + bytes_done, cluster_data, cur_bytes); -if (qcow2_co_encrypt (bs, cluster_offset, offset, +if (qcow2_co_encrypt (bs, cluster_offset + offset_in_cluster, offset, cluster_data, cur_bytes) <0 ) { ret = -EIO ; goto out_unlocked;
block/qcow2.h +4 -4
@@ -758,10 +758,10 @@ ssize_t coroutine_fn qcow2_co_decompress(BlockDriverState *bs, void *dest, size_t dest_size, const void *src, size_t src_size); 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_offset, +uint64_t guest_offset,void *buf,size_t len); 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_offset, +uint64_t guest_offset,void *buf,size_t len);#endif
Signature · 1 line
2.17.2
Re: [Qemu-devel] [PATCH v7 2/3] block/qcow2: refactor encryption code
Quoted 22 lines · Maxim Levitsky
* Change the qcow2_co_{encrypt|decrypt} to just receive full host and guest offsets and use this function directly instead of calling
do_perform_cow_encrypt(which is removed by that patch).* Adjust
qcow2_co_encdecto take full host and guest offsets as well.* Document the qcow2_co_{encrypt|decrypt} arguments to prevent the bug fixed in former commit from hopefully happening again.
Signed-off-by Maxim Levitsky <[email protected]>--- block/qcow2-cluster.c | 41 ++++++++++------------------ block/qcow2-threads.c | 63 +++++++++++++++++++++++++++++++++---------- block/qcow2.c | 5 ++-- block/qcow2.h | 8 +++--- 4 files changed, 70 insertions(+), 47 deletions(-)block/qcow2-cluster.c +0
-0
[..]
Quoted 9 lines
static int
coroutine_fndo_perform_cow_write(BlockDriverState *bs,
uint64_tcluster_offset, unsignedoffset_in_cluster,block/qcow2-cluster.c +0
-2@@ -891,11 +869,20 @@ 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,
block/qcow2-cluster.c static int perform_cow(BlockDriverState *bs, QCowL2Meta *m) · line 878
Quoted 10 lines
block/qcow2-cluster.c +7
-3- start->nb_bytes) || - !do_perform_cow_encrypt(bs, m->offset, m->alloc_offset, - end->offset, end_buffer, end->nb_bytes)) { + ret =qcow2_co_encrypt (bs, + m->alloc_offset + start->offset, + m->offset + start->offset, + start_buffer, start->nb_bytes); +if (ret <0 ) { + ret = -EIO ; + goto fail;
Just go to fail I think, don't reassign ret variable.
block/qcow2-cluster.c static int perform_cow(BlockDriverState *bs, QCowL2Meta *m) · line 886
Quoted 8 lines
block/qcow2-cluster.c +7
-0+ } + + ret =qcow2_co_encrypt (bs, + m->alloc_offset + end->offset, + m->offset + end->offset, + end_buffer, end->nb_bytes); +if (ret <0 ) { ret = -EIO ;
and here.
with these two places fixed:
(I think, these simple changes may be applied while queuing, so you don't need to resend, if there no other reasons)
Signature · 2 lines
Best regards, Vladimir
Re: [Qemu-devel] [PATCH v7 2/3] block/qcow2: refactor encryption code
In reply to Maxim LevitskyQuoted 10 lines · Maxim Levitsky
* Change the qcow2_co_{encrypt|decrypt} to just receive full host and guest offsets and use this function directly instead of calling
do_perform_cow_encrypt(which is removed by that patch).* Adjust
qcow2_co_encdecto take full host and guest offsets as well.* Document the qcow2_co_{encrypt|decrypt} arguments to prevent the bug fixed in former commit from hopefully happening again.
Quoted 7 lines
Signed-off-by Maxim Levitsky <[email protected]>--- block/qcow2-cluster.c | 41 ++++++++++------------------ block/qcow2-threads.c | 63 +++++++++++++++++++++++++++++++++---------- block/qcow2.c | 5 ++-- block/qcow2.h | 8 +++--- 4 files changed, 70 insertions(+), 47 deletions(-)
Looks good to me, but there are conflicts with two series I’ve already taken to my branch: Namely “qcow2: async handling of fragmented io” by Vladimir and “Alignment checks cleanup” by Nir.
Unfortunately, those conflicts (while not difficult to resolve) are not quite trivial, so I’d rather not resolve them myself.
I’ll send a pull request today, but until they are in master, you could rebase on my branch (either of
- https://git.xanclic.moe/XanClic/qemu.git block
- https://github.com/XanClic/qemu.git block
, whichever you prefer that works).
Max
PATCH v7 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
@@ -274,5 +274,6 @@ 257 rw 258 rw quick 262 rw quick migration +263 rw quick 265 rw auto quick 266 rw quick
Signature · 1 line
2.17.2
Re: [Qemu-devel] [PATCH v7 3/3] qemu-iotests: Add test for bz #1745922
Quoted 9 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.out
Quoted 16 lines · Maxim Levitsky
Commit
8ac0f15f335accidently 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
CC qemu-stable <[email protected]>V2: grammar, spelling and code style fixes.
V3: more fixes after the review.
V4: addressed review comments from Max Reitz,
and futher refactored theqcow2_co_encryptto just take full host and guest offset which simplifies everything.
Quoted 8 lines
V5: reworked the patches so one of them fixes the bug
only and other one is just refactoringV6: removed
do_perform_cow_encryptV7: removed
do_perform_cow_encrypttake two, this
time I hopefully did that correctly :-)
Also updated commit names and messages a bit
Luckily for you (maybe), Vladimir’s series doesn‘t quite pass the iotests for me, so unfortunately (I find it unfortunate) I had to remove it from my branch. Thus, the conflicts are much more tame and I felt comfortable taking the series to my branch (with the remaining trivial conflicts resolved, and with Vladimir’s suggestion applied):
https://git.xanclic.moe/XanClic/qemu/commits/branch/block
Max
Quoted 26 lines · Max Reitz
Quoted 16 lines · Maxim Levitsky
Commit
8ac0f15f335accidently 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
CC qemu-stable <[email protected]>V2: grammar, spelling and code style fixes.
V3: more fixes after the review.
V4: addressed review comments from Max Reitz,
and futher refactored theqcow2_co_encryptto just take full host and guest offset which simplifies everything.Quoted 8 lines
V5: reworked the patches so one of them fixes the bug
only and other one is just refactoringV6: removed
do_perform_cow_encryptV7: removed
do_perform_cow_encrypttake two, this
time I hopefully did that correctly :-)
Also updated commit names and messages a bit
Quoted 7 lines
Luckily for you (maybe), Vladimir’s series doesn‘t quite pass the iotests for me, so unfortunately (I find it unfortunate) I had to remove it from my branch. Thus, the conflicts are much more tame and I felt comfortable taking the series to my branch (with the remaining trivial conflicts resolved, and with Vladimir’s suggestion applied):
First of all, Thanks!
I don't know if this is luckily for me since I already rebased my series on top of
https://git.xanclic.moe/XanClic/qemu.git,
and run all qcow2 iotests, and only tests
162 169 194 196 234 262 failed, and I know that 162 always fails
due to that kernel change I talked about here few days ago,
and rest for the AF_UNIX path len, which I need to do something
about in the long term. I sometimes do a separate build in
directory which path doesn't trigger this, and sometimes,
when I know that I haven't done significant changes to the patches,
I just let these tests fail. In long term, maybe even in a few days I'll allocate some time to rethink the build environment here to
fix that permanently.
Now I am rerunning the iotests just for fun, in short enough directory to see if I can reproduce the failure that you had. After looking in your report, that iotest 026 fails, it does pass here, but then I am only running these iotests on my laptop so I probably don't trigger the race you were able to.
So thanks again!
Best regards,
Maxim Levitsky
Quoted 51 lines · Maxim Levitsky
Quoted 26 lines · Max Reitz
Quoted 16 lines · Maxim Levitsky
Commit
8ac0f15f335accidently 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
CC qemu-stable <[email protected]>V2: grammar, spelling and code style fixes.
V3: more fixes after the review.
V4: addressed review comments from Max Reitz,
and futher refactored theqcow2_co_encryptto just take full host and guest offset which simplifies everything.Quoted 8 lines
V5: reworked the patches so one of them fixes the bug
only and other one is just refactoringV6: removed
do_perform_cow_encryptV7: removed
do_perform_cow_encrypttake two, this
time I hopefully did that correctly :-)
Also updated commit names and messages a bitQuoted 7 lines
Luckily for you (maybe), Vladimir’s series doesn‘t quite pass the iotests for me, so unfortunately (I find it unfortunate) I had to remove it from my branch. Thus, the conflicts are much more tame and I felt comfortable taking the series to my branch (with the remaining trivial conflicts resolved, and with Vladimir’s suggestion applied):
First of all, Thanks!
I don't know if this is luckily for me since I already rebased my series on top of
https://git.xanclic.moe/XanClic/qemu.git,
and run all qcow2 iotests, and only tests
162 169 194 196 234 262 failed, and I know that 162 always fails
due to that kernel change I talked about here few days ago,
and rest for theAF_UNIXpath len, which I need to do something
about in the long term. I sometimes do a separate build in
directory which path doesn't trigger this, and sometimes,
when I know that I haven't done significant changes to the patches,
I just let these tests fail. In long term, maybe even in a few days I'll allocate some time to rethink the build environment here to
fix that permanently.
Quoted 5 lines
Now I am rerunning the iotests just for fun, in short enough directory to see if I can reproduce the failure that you had. After looking in your report, that iotest 026 fails, it does pass here, but then I am only running these iotests on my laptop so I probably don't trigger the race you were able to.
Passed all the tests but 162, not that it matters much to be honest. Best regards,
Maxim Levitsky