* [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions"
@ 2022-04-27 11:40 Hanna Reitz
2022-04-27 11:40 ` [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function Hanna Reitz
` (4 more replies)
0 siblings, 5 replies; 11+ messages in thread
From: Hanna Reitz @ 2022-04-27 11:40 UTC (permalink / raw)
To: qemu-block
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, Hanna Reitz, qemu-devel,
Paolo Bonzini
Hi,
This assertion was disabled in commit
b1c073490553f80594b903ceedfc7c1aef6b1b19. We wanted to re-enable it
once the 7.1 tree has opened (which is now), but to do so, we should
also fix the bug reported in
https://gitlab.com/qemu-project/qemu/-/issues/945 .
Patches 1 and 2 fix that problem (I hope), patch 3 re-enables the
assertion, and patch 4 adds a regression test for issue 945.
Hanna Reitz (4):
block: Classify bdrv_get_flags() as I/O function
qcow2: Do not reopen data_file in invalidate_cache
Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions"
iotests: Add regression test for issue 945
include/block/block-global-state.h | 1 -
include/block/block-io.h | 1 +
include/qemu/main-loop.h | 3 +-
block.c | 2 +-
block/qcow2.c | 104 +++++++++++-------
.../tests/export-incoming-iothread | 81 ++++++++++++++
.../tests/export-incoming-iothread.out | 5 +
7 files changed, 151 insertions(+), 46 deletions(-)
create mode 100755 tests/qemu-iotests/tests/export-incoming-iothread
create mode 100644 tests/qemu-iotests/tests/export-incoming-iothread.out
--
2.35.1
^ permalink raw reply [flat|nested] 11+ messages in thread
* [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function
2022-04-27 11:40 [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
@ 2022-04-27 11:40 ` Hanna Reitz
2022-04-27 13:14 ` Eric Blake
2022-04-27 11:40 ` [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache Hanna Reitz
` (3 subsequent siblings)
4 siblings, 1 reply; 11+ messages in thread
From: Hanna Reitz @ 2022-04-27 11:40 UTC (permalink / raw)
To: qemu-block
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, Hanna Reitz, qemu-devel,
Paolo Bonzini
This function is safe to call in an I/O context, and qcow2_do_open()
does so (invoked in an I/O context by qcow2_co_invalidate_cache()).
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
include/block/block-global-state.h | 1 -
include/block/block-io.h | 1 +
block.c | 2 +-
3 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/include/block/block-global-state.h b/include/block/block-global-state.h
index 25bb69bbef..21265e3966 100644
--- a/include/block/block-global-state.h
+++ b/include/block/block-global-state.h
@@ -172,7 +172,6 @@ void bdrv_next_cleanup(BdrvNextIterator *it);
BlockDriverState *bdrv_next_monitor_owned(BlockDriverState *bs);
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
void *opaque, bool read_only);
-int bdrv_get_flags(BlockDriverState *bs);
char *bdrv_get_full_backing_filename(BlockDriverState *bs, Error **errp);
char *bdrv_dirname(BlockDriverState *bs, Error **errp);
diff --git a/include/block/block-io.h b/include/block/block-io.h
index 5e3f346806..62c84f0519 100644
--- a/include/block/block-io.h
+++ b/include/block/block-io.h
@@ -103,6 +103,7 @@ int bdrv_apply_auto_read_only(BlockDriverState *bs, const char *errmsg,
bool bdrv_is_read_only(BlockDriverState *bs);
bool bdrv_is_writable(BlockDriverState *bs);
bool bdrv_is_sg(BlockDriverState *bs);
+int bdrv_get_flags(BlockDriverState *bs);
bool bdrv_is_inserted(BlockDriverState *bs);
void bdrv_lock_medium(BlockDriverState *bs, bool locked);
void bdrv_eject(BlockDriverState *bs, bool eject_flag);
diff --git a/block.c b/block.c
index 8cd16e757e..2c00dddd80 100644
--- a/block.c
+++ b/block.c
@@ -6298,7 +6298,7 @@ const char *bdrv_get_device_or_node_name(const BlockDriverState *bs)
int bdrv_get_flags(BlockDriverState *bs)
{
- GLOBAL_STATE_CODE();
+ IO_CODE();
return bs->open_flags;
}
--
2.35.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache
2022-04-27 11:40 [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
2022-04-27 11:40 ` [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function Hanna Reitz
@ 2022-04-27 11:40 ` Hanna Reitz
2022-04-27 13:20 ` Eric Blake
2022-04-27 16:05 ` Kevin Wolf
2022-04-27 11:40 ` [PATCH 3/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
` (2 subsequent siblings)
4 siblings, 2 replies; 11+ messages in thread
From: Hanna Reitz @ 2022-04-27 11:40 UTC (permalink / raw)
To: qemu-block
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, Hanna Reitz, qemu-devel,
Paolo Bonzini
qcow2_co_invalidate_cache() closes and opens the qcow2 file, by calling
qcow2_close() and qcow2_do_open(). These two functions must thus be
usable from both a global-state and an I/O context.
As they are, they are not safe to call in an I/O context, because they
use bdrv_unref_child() and bdrv_open_child() to close/open the data_file
child, respectively, both of which are global-state functions. When
used from qcow2_co_invalidate_cache(), we do not need to close/open the
data_file child, though (we do not do this for bs->file or bs->backing
either), and so we should skip it in the qcow2_co_invalidate_cache()
path.
To do so, add a parameter to qcow2_do_open() and qcow2_close() to make
them skip handling s->data_file, and have qcow2_co_invalidate_cache()
exempt it from the memset() on the BDRVQcow2State.
(Note that the QED driver similarly closes/opens the QED image by
invoking bdrv_qed_close()+bdrv_qed_do_open(), but both functions seem
safe to use in an I/O context.)
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/945
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
block/qcow2.c | 104 ++++++++++++++++++++++++++++++--------------------
1 file changed, 62 insertions(+), 42 deletions(-)
diff --git a/block/qcow2.c b/block/qcow2.c
index b5c47931ef..4f5e6440fb 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1296,7 +1296,8 @@ static int validate_compression_type(BDRVQcow2State *s, Error **errp)
/* Called with s->lock held. */
static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options,
- int flags, Error **errp)
+ int flags, bool open_data_file,
+ Error **errp)
{
ERRP_GUARD();
BDRVQcow2State *s = bs->opaque;
@@ -1614,50 +1615,52 @@ static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options,
goto fail;
}
- /* Open external data file */
- s->data_file = bdrv_open_child(NULL, options, "data-file", bs,
- &child_of_bds, BDRV_CHILD_DATA,
- true, errp);
- if (*errp) {
- ret = -EINVAL;
- goto fail;
- }
+ if (open_data_file) {
+ /* Open external data file */
+ s->data_file = bdrv_open_child(NULL, options, "data-file", bs,
+ &child_of_bds, BDRV_CHILD_DATA,
+ true, errp);
+ if (*errp) {
+ ret = -EINVAL;
+ goto fail;
+ }
- if (s->incompatible_features & QCOW2_INCOMPAT_DATA_FILE) {
- if (!s->data_file && s->image_data_file) {
- s->data_file = bdrv_open_child(s->image_data_file, options,
- "data-file", bs, &child_of_bds,
- BDRV_CHILD_DATA, false, errp);
+ if (s->incompatible_features & QCOW2_INCOMPAT_DATA_FILE) {
+ if (!s->data_file && s->image_data_file) {
+ s->data_file = bdrv_open_child(s->image_data_file, options,
+ "data-file", bs, &child_of_bds,
+ BDRV_CHILD_DATA, false, errp);
+ if (!s->data_file) {
+ ret = -EINVAL;
+ goto fail;
+ }
+ }
if (!s->data_file) {
+ error_setg(errp, "'data-file' is required for this image");
ret = -EINVAL;
goto fail;
}
- }
- if (!s->data_file) {
- error_setg(errp, "'data-file' is required for this image");
- ret = -EINVAL;
- goto fail;
- }
- /* No data here */
- bs->file->role &= ~BDRV_CHILD_DATA;
+ /* No data here */
+ bs->file->role &= ~BDRV_CHILD_DATA;
- /* Must succeed because we have given up permissions if anything */
- bdrv_child_refresh_perms(bs, bs->file, &error_abort);
- } else {
- if (s->data_file) {
- error_setg(errp, "'data-file' can only be set for images with an "
- "external data file");
- ret = -EINVAL;
- goto fail;
- }
+ /* Must succeed because we have given up permissions if anything */
+ bdrv_child_refresh_perms(bs, bs->file, &error_abort);
+ } else {
+ if (s->data_file) {
+ error_setg(errp, "'data-file' can only be set for images with "
+ "an external data file");
+ ret = -EINVAL;
+ goto fail;
+ }
- s->data_file = bs->file;
+ s->data_file = bs->file;
- if (data_file_is_raw(bs)) {
- error_setg(errp, "data-file-raw requires a data file");
- ret = -EINVAL;
- goto fail;
+ if (data_file_is_raw(bs)) {
+ error_setg(errp, "data-file-raw requires a data file");
+ ret = -EINVAL;
+ goto fail;
+ }
}
}
@@ -1839,7 +1842,7 @@ static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options,
fail:
g_free(s->image_data_file);
- if (has_data_file(bs)) {
+ if (open_data_file && has_data_file(bs)) {
bdrv_unref_child(bs, s->data_file);
s->data_file = NULL;
}
@@ -1876,7 +1879,8 @@ static void coroutine_fn qcow2_open_entry(void *opaque)
BDRVQcow2State *s = qoc->bs->opaque;
qemu_co_mutex_lock(&s->lock);
- qoc->ret = qcow2_do_open(qoc->bs, qoc->options, qoc->flags, qoc->errp);
+ qoc->ret = qcow2_do_open(qoc->bs, qoc->options, qoc->flags, true,
+ qoc->errp);
qemu_co_mutex_unlock(&s->lock);
}
@@ -2714,7 +2718,7 @@ static int qcow2_inactivate(BlockDriverState *bs)
return result;
}
-static void qcow2_close(BlockDriverState *bs)
+static void qcow2_do_close(BlockDriverState *bs, bool close_data_file)
{
BDRVQcow2State *s = bs->opaque;
qemu_vfree(s->l1_table);
@@ -2740,7 +2744,7 @@ static void qcow2_close(BlockDriverState *bs)
g_free(s->image_backing_file);
g_free(s->image_backing_format);
- if (has_data_file(bs)) {
+ if (close_data_file && has_data_file(bs)) {
bdrv_unref_child(bs, s->data_file);
s->data_file = NULL;
}
@@ -2749,11 +2753,17 @@ static void qcow2_close(BlockDriverState *bs)
qcow2_free_snapshots(bs);
}
+static void qcow2_close(BlockDriverState *bs)
+{
+ qcow2_do_close(bs, true);
+}
+
static void coroutine_fn qcow2_co_invalidate_cache(BlockDriverState *bs,
Error **errp)
{
ERRP_GUARD();
BDRVQcow2State *s = bs->opaque;
+ BdrvChild *data_file;
int flags = s->flags;
QCryptoBlock *crypto = NULL;
QDict *options;
@@ -2767,14 +2777,24 @@ static void coroutine_fn qcow2_co_invalidate_cache(BlockDriverState *bs,
crypto = s->crypto;
s->crypto = NULL;
- qcow2_close(bs);
+ /*
+ * Do not reopen s->data_file (i.e., have qcow2_do_close() not close it,
+ * and then prevent qcow2_do_open() from opening it), because this function
+ * runs in the I/O path and as such we must not invoke global-state
+ * functions like bdrv_unref_child() and bdrv_open_child().
+ */
+ qcow2_do_close(bs, false);
+
+ data_file = s->data_file;
memset(s, 0, sizeof(BDRVQcow2State));
+ s->data_file = data_file;
+
options = qdict_clone_shallow(bs->options);
flags &= ~BDRV_O_INACTIVE;
qemu_co_mutex_lock(&s->lock);
- ret = qcow2_do_open(bs, options, flags, errp);
+ ret = qcow2_do_open(bs, options, flags, false, errp);
qemu_co_mutex_unlock(&s->lock);
qobject_unref(options);
if (ret < 0) {
--
2.35.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH 3/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions"
2022-04-27 11:40 [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
2022-04-27 11:40 ` [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function Hanna Reitz
2022-04-27 11:40 ` [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache Hanna Reitz
@ 2022-04-27 11:40 ` Hanna Reitz
2022-04-27 13:22 ` Eric Blake
2022-04-27 11:40 ` [PATCH 4/4] iotests: Add regression test for issue 945 Hanna Reitz
2022-04-27 16:13 ` [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Kevin Wolf
4 siblings, 1 reply; 11+ messages in thread
From: Hanna Reitz @ 2022-04-27 11:40 UTC (permalink / raw)
To: qemu-block
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, Hanna Reitz, qemu-devel,
Paolo Bonzini
This reverts commit b1c073490553f80594b903ceedfc7c1aef6b1b19. (We
wanted to do so once the 7.1 tree opens, which has happened. The issue
reported in https://gitlab.com/qemu-project/qemu/-/issues/945 should be
fixed by the preceding patches.)
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
include/qemu/main-loop.h | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/include/qemu/main-loop.h b/include/qemu/main-loop.h
index d3750c8e76..89bd9edefb 100644
--- a/include/qemu/main-loop.h
+++ b/include/qemu/main-loop.h
@@ -284,8 +284,7 @@ bool qemu_in_main_thread(void);
#else
#define GLOBAL_STATE_CODE() \
do { \
- /* FIXME: Re-enable after 7.0 release */ \
- /* assert(qemu_in_main_thread()); */ \
+ assert(qemu_in_main_thread()); \
} while (0)
#endif /* CONFIG_COCOA */
--
2.35.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH 4/4] iotests: Add regression test for issue 945
2022-04-27 11:40 [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
` (2 preceding siblings ...)
2022-04-27 11:40 ` [PATCH 3/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
@ 2022-04-27 11:40 ` Hanna Reitz
2022-04-27 13:52 ` Eric Blake
2022-04-27 16:13 ` [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Kevin Wolf
4 siblings, 1 reply; 11+ messages in thread
From: Hanna Reitz @ 2022-04-27 11:40 UTC (permalink / raw)
To: qemu-block
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, Hanna Reitz, qemu-devel,
Paolo Bonzini
Create a VM with a BDS in an iothread, add -incoming defer to the
command line, and then export this BDS via NBD. Doing so should not
fail an assertion.
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
---
.../tests/export-incoming-iothread | 81 +++++++++++++++++++
.../tests/export-incoming-iothread.out | 5 ++
2 files changed, 86 insertions(+)
create mode 100755 tests/qemu-iotests/tests/export-incoming-iothread
create mode 100644 tests/qemu-iotests/tests/export-incoming-iothread.out
diff --git a/tests/qemu-iotests/tests/export-incoming-iothread b/tests/qemu-iotests/tests/export-incoming-iothread
new file mode 100755
index 0000000000..7679e49103
--- /dev/null
+++ b/tests/qemu-iotests/tests/export-incoming-iothread
@@ -0,0 +1,81 @@
+#!/usr/bin/env python3
+# group: rw quick migration
+#
+# Regression test for issue 945:
+# https://gitlab.com/qemu-project/qemu/-/issues/945
+# Test adding an export on top of an iothread-ed block device while in
+# -incoming defer.
+#
+# Copyright (C) 2022 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/>.
+#
+
+import os
+import iotests
+from iotests import qemu_img_create
+
+
+image_size = 1 * 1024 * 1024
+test_img = os.path.join(iotests.test_dir, 'test.img')
+node_name = 'node0'
+iothread_id = 'iothr0'
+
+nbd_sock = os.path.join(iotests.sock_dir, 'nbd.sock')
+
+
+class TestExportIncomingIothread(iotests.QMPTestCase):
+ def setUp(self) -> None:
+ qemu_img_create('-f', iotests.imgfmt, test_img, str(image_size))
+
+ self.vm = iotests.VM()
+ self.vm.add_object(f'iothread,id={iothread_id}')
+ self.vm.add_blockdev((
+ f'driver={iotests.imgfmt}',
+ f'node-name={node_name}',
+ 'file.driver=file',
+ f'file.filename={test_img}'
+ ))
+ self.vm.add_incoming('defer')
+ self.vm.launch()
+
+ def tearDown(self):
+ self.vm.shutdown()
+ os.remove(test_img)
+
+ def test_export_add(self):
+ result = self.vm.qmp('nbd-server-start', {
+ 'addr': {
+ 'type': 'unix',
+ 'data': {
+ 'path': nbd_sock
+ }
+ }
+ })
+ self.assert_qmp(result, 'return', {})
+
+ # Regression test for issue 945: This should not fail an assertion
+ result = self.vm.qmp('block-export-add', {
+ 'type': 'nbd',
+ 'id': 'exp0',
+ 'node-name': node_name,
+ 'iothread': iothread_id
+ })
+ self.assert_qmp(result, 'return', {})
+
+
+if __name__ == '__main__':
+ iotests.main(supported_fmts=['generic'],
+ unsupported_fmts=['luks'], # Would need a secret
+ supported_protocols=['file'])
diff --git a/tests/qemu-iotests/tests/export-incoming-iothread.out b/tests/qemu-iotests/tests/export-incoming-iothread.out
new file mode 100644
index 0000000000..ae1213e6f8
--- /dev/null
+++ b/tests/qemu-iotests/tests/export-incoming-iothread.out
@@ -0,0 +1,5 @@
+.
+----------------------------------------------------------------------
+Ran 1 tests
+
+OK
--
2.35.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function
2022-04-27 11:40 ` [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function Hanna Reitz
@ 2022-04-27 13:14 ` Eric Blake
0 siblings, 0 replies; 11+ messages in thread
From: Eric Blake @ 2022-04-27 13:14 UTC (permalink / raw)
To: Hanna Reitz
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, qemu-devel, qemu-block,
Paolo Bonzini
On Wed, Apr 27, 2022 at 01:40:54PM +0200, Hanna Reitz wrote:
> This function is safe to call in an I/O context, and qcow2_do_open()
> does so (invoked in an I/O context by qcow2_co_invalidate_cache()).
>
> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
> ---
> include/block/block-global-state.h | 1 -
> include/block/block-io.h | 1 +
> block.c | 2 +-
> 3 files changed, 2 insertions(+), 2 deletions(-)
Reviewed-by: Eric Blake <eblake@redhat.com>
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache
2022-04-27 11:40 ` [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache Hanna Reitz
@ 2022-04-27 13:20 ` Eric Blake
2022-04-27 16:05 ` Kevin Wolf
1 sibling, 0 replies; 11+ messages in thread
From: Eric Blake @ 2022-04-27 13:20 UTC (permalink / raw)
To: Hanna Reitz
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, qemu-devel, qemu-block,
Paolo Bonzini
On Wed, Apr 27, 2022 at 01:40:55PM +0200, Hanna Reitz wrote:
> qcow2_co_invalidate_cache() closes and opens the qcow2 file, by calling
> qcow2_close() and qcow2_do_open(). These two functions must thus be
> usable from both a global-state and an I/O context.
>
> As they are, they are not safe to call in an I/O context, because they
> use bdrv_unref_child() and bdrv_open_child() to close/open the data_file
> child, respectively, both of which are global-state functions. When
> used from qcow2_co_invalidate_cache(), we do not need to close/open the
> data_file child, though (we do not do this for bs->file or bs->backing
> either), and so we should skip it in the qcow2_co_invalidate_cache()
> path.
>
> To do so, add a parameter to qcow2_do_open() and qcow2_close() to make
> them skip handling s->data_file, and have qcow2_co_invalidate_cache()
> exempt it from the memset() on the BDRVQcow2State.
>
> (Note that the QED driver similarly closes/opens the QED image by
> invoking bdrv_qed_close()+bdrv_qed_do_open(), but both functions seem
> safe to use in an I/O context.)
>
> Fixes: https://gitlab.com/qemu-project/qemu/-/issues/945
> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
> ---
> block/qcow2.c | 104 ++++++++++++++++++++++++++++++--------------------
> 1 file changed, 62 insertions(+), 42 deletions(-)
Reviewed-by: Eric Blake <eblake@redhat.com>
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 3/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions"
2022-04-27 11:40 ` [PATCH 3/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
@ 2022-04-27 13:22 ` Eric Blake
0 siblings, 0 replies; 11+ messages in thread
From: Eric Blake @ 2022-04-27 13:22 UTC (permalink / raw)
To: Hanna Reitz
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, qemu-devel, qemu-block,
Paolo Bonzini
On Wed, Apr 27, 2022 at 01:40:56PM +0200, Hanna Reitz wrote:
> This reverts commit b1c073490553f80594b903ceedfc7c1aef6b1b19. (We
> wanted to do so once the 7.1 tree opens, which has happened. The issue
> reported in https://gitlab.com/qemu-project/qemu/-/issues/945 should be
> fixed by the preceding patches.)
>
> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
> ---
> include/qemu/main-loop.h | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
Reviewed-by: Eric Blake <eblake@redhat.com>
[Here's hoping we don't have to re-disable it because it finds more
bugs - but maximizing the soak time during development with it enabled
is good]
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 4/4] iotests: Add regression test for issue 945
2022-04-27 11:40 ` [PATCH 4/4] iotests: Add regression test for issue 945 Hanna Reitz
@ 2022-04-27 13:52 ` Eric Blake
0 siblings, 0 replies; 11+ messages in thread
From: Eric Blake @ 2022-04-27 13:52 UTC (permalink / raw)
To: Hanna Reitz
Cc: Kevin Wolf, Emanuele Giuseppe Esposito, qemu-devel, qemu-block,
Paolo Bonzini
On Wed, Apr 27, 2022 at 01:40:57PM +0200, Hanna Reitz wrote:
> Create a VM with a BDS in an iothread, add -incoming defer to the
> command line, and then export this BDS via NBD. Doing so should not
> fail an assertion.
>
> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
> ---
> .../tests/export-incoming-iothread | 81 +++++++++++++++++++
> .../tests/export-incoming-iothread.out | 5 ++
> 2 files changed, 86 insertions(+)
> create mode 100755 tests/qemu-iotests/tests/export-incoming-iothread
> create mode 100644 tests/qemu-iotests/tests/export-incoming-iothread.out
The test looks sane, and appears to match the formula posted in issue
945. When applying just patches 3-4 (skipping 1-2), './check
export-incoming-iothread' passed, but './check -qcow2
export-incoming-iothread' failed; so the image format is important.
And the failure was rather verbose, which is a GOOD thing - our
efforts to make the python framework point out abnormal exits is
working!
+WARNING:qemu.machine.machine:qemu received signal 6; command: "/home/eblake/qemu/build/tests/qemu-iotests/...
...
+Traceback (most recent call last):
+ File "/home/eblake/qemu/tests/qemu-iotests/tests/export-incoming-iothread", line 69, in test_export_add
+ result = self.vm.qmp('block-export-add', {
...
+ File "/home/eblake/qemu/python/qemu/qmp/qmp_client.py", line 463, in _reply
+ raise result
+qemu.qmp.qmp_client.ExecInterruptedError: Disconnected
+
+======================================================================
+ERROR: test_export_add (__main__.TestExportIncomingIothread)
+----------------------------------------------------------------------
+Traceback (most recent call last):
+ File "/home/eblake/qemu/python/qemu/machine/machine.py", line 533, in _soft_shutdown
+ self.qmp('quit')
...
+qemu.qmp.protocol.StateError: QMPClient is disconnecting. Call disconnect() to return to IDLE state.
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+ File "/home/eblake/qemu/python/qemu/machine/machine.py", line 554, in _do_shutdown
+ self._soft_shutdown(timeout)
...
+ File "/home/eblake/qemu/python/qemu/qmp/protocol.py", line 968, in _readline
+ raise EOFError
+EOFError
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+ File "/home/eblake/qemu/tests/qemu-iotests/tests/export-incoming-iothread", line 54, in tearDown
+ self.vm.shutdown()
+ File "/home/eblake/qemu/python/qemu/machine/machine.py", line 583, in shutdown
+ self._do_shutdown(timeout)
+ File "/home/eblake/qemu/python/qemu/machine/machine.py", line 557, in _do_shutdown
+ raise AbnormalShutdown("Could not perform graceful shutdown") \
+qemu.machine.machine.AbnormalShutdown: Could not perform graceful shutdown
+
----------------------------------------------------------------------
Ran 1 tests
-OK
+FAILED (errors=2)
Then applying patches 1-2 and repeating the test passed, so your test
is a good proof that we identified and fixed the problem at hand.
Reviewed-by: Eric Blake <eblake@redhat.com>
Tested-by: Eric Blake <eblake@redhat.com>
Given that it is visible with NBD, I'm happy to queue this series
through my tree if no one else grabs it first.
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache
2022-04-27 11:40 ` [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache Hanna Reitz
2022-04-27 13:20 ` Eric Blake
@ 2022-04-27 16:05 ` Kevin Wolf
1 sibling, 0 replies; 11+ messages in thread
From: Kevin Wolf @ 2022-04-27 16:05 UTC (permalink / raw)
To: Hanna Reitz
Cc: Emanuele Giuseppe Esposito, Paolo Bonzini, qemu-devel, qemu-block
Am 27.04.2022 um 13:40 hat Hanna Reitz geschrieben:
> qcow2_co_invalidate_cache() closes and opens the qcow2 file, by calling
> qcow2_close() and qcow2_do_open(). These two functions must thus be
> usable from both a global-state and an I/O context.
>
> As they are, they are not safe to call in an I/O context, because they
> use bdrv_unref_child() and bdrv_open_child() to close/open the data_file
> child, respectively, both of which are global-state functions. When
> used from qcow2_co_invalidate_cache(), we do not need to close/open the
> data_file child, though (we do not do this for bs->file or bs->backing
> either), and so we should skip it in the qcow2_co_invalidate_cache()
> path.
>
> To do so, add a parameter to qcow2_do_open() and qcow2_close() to make
> them skip handling s->data_file, and have qcow2_co_invalidate_cache()
> exempt it from the memset() on the BDRVQcow2State.
>
> (Note that the QED driver similarly closes/opens the QED image by
> invoking bdrv_qed_close()+bdrv_qed_do_open(), but both functions seem
> safe to use in an I/O context.)
>
> Fixes: https://gitlab.com/qemu-project/qemu/-/issues/945
> Signed-off-by: Hanna Reitz <hreitz@redhat.com>
This feels a bit like a hack, and we'll have to change it again if we
ever want to allow changing the data_file with reopen. But it should do
the job for now.
Kevin
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions"
2022-04-27 11:40 [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
` (3 preceding siblings ...)
2022-04-27 11:40 ` [PATCH 4/4] iotests: Add regression test for issue 945 Hanna Reitz
@ 2022-04-27 16:13 ` Kevin Wolf
4 siblings, 0 replies; 11+ messages in thread
From: Kevin Wolf @ 2022-04-27 16:13 UTC (permalink / raw)
To: Hanna Reitz
Cc: Emanuele Giuseppe Esposito, Paolo Bonzini, qemu-devel, qemu-block
Am 27.04.2022 um 13:40 hat Hanna Reitz geschrieben:
> Hi,
>
> This assertion was disabled in commit
> b1c073490553f80594b903ceedfc7c1aef6b1b19. We wanted to re-enable it
> once the 7.1 tree has opened (which is now), but to do so, we should
> also fix the bug reported in
> https://gitlab.com/qemu-project/qemu/-/issues/945 .
>
> Patches 1 and 2 fix that problem (I hope), patch 3 re-enables the
> assertion, and patch 4 adds a regression test for issue 945.
Thanks, applied to the block branch.
Kevin
^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2022-04-27 16:15 UTC | newest]
Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-27 11:40 [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
2022-04-27 11:40 ` [PATCH 1/4] block: Classify bdrv_get_flags() as I/O function Hanna Reitz
2022-04-27 13:14 ` Eric Blake
2022-04-27 11:40 ` [PATCH 2/4] qcow2: Do not reopen data_file in invalidate_cache Hanna Reitz
2022-04-27 13:20 ` Eric Blake
2022-04-27 16:05 ` Kevin Wolf
2022-04-27 11:40 ` [PATCH 3/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Hanna Reitz
2022-04-27 13:22 ` Eric Blake
2022-04-27 11:40 ` [PATCH 4/4] iotests: Add regression test for issue 945 Hanna Reitz
2022-04-27 13:52 ` Eric Blake
2022-04-27 16:13 ` [PATCH 0/4] Revert "main-loop: Disable GLOBAL_STATE_CODE() assertions" Kevin Wolf
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).