From: Gollu Appalanaidu
Add the Compare command.
This implementation uses a bounce buffer to read in the data from
storage and then compare with the host supplied buffer.
Signed-off-by: Gollu Appalanaidu
[k.jensen: rebased]
Signed-off-by: Klaus Jensen
---
hw/block/nvme.c | 100
From: Klaus Jensen
Add support for TP 4065a ("Simple Copy Command"), v2020.05.04
("Ratified").
The implementation uses a bounce buffer to first read in the source
logical blocks, then issue a write of that bounce buffer. The default
maximum number of source logical blocks is 128, translating to
From: Klaus Jensen
Add new data structures and types for the Simple Copy command.
Signed-off-by: Klaus Jensen
Cc: Stefan Hajnoczi
Cc: Fam Zheng
---
include/block/nvme.h | 45 ++--
1 file changed, 43 insertions(+), 2 deletions(-)
diff --git
From: Klaus Jensen
Add support for TP 4065 ("Simple Copy Command").
Klaus Jensen (2):
nvme: updated shared header for copy command
hw/block/nvme: add simple copy command
hw/block/nvme-ns.h| 4 +
hw/block/nvme.h | 1 +
include/block/nvme.h | 45 -
hw/block/nvme-ns.c
On Nov 12 20:59, Klaus Jensen wrote:
> From: Klaus Jensen
>
> This adds support for the Deallocated or Unwritten Logical Block error
> recovery feature as well as the Dataset Management command.
>
> v8:
> - Move req->opaque clearing to nvme_req_clear.
> - Add two preparation/cleanup
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 42 --
1 file changed, 40 insertions(+), 2 deletions(-)
diff --git a/block.c b/block.c
index cf7b859a81..6ea926e2c1 100644
--- a/block.c
+++ b/block.c
@@ -2974,12 +2974,19 @@ BdrvChild
We don't need this workaround anymore: bdrv_append is already smart
enough and we can use new bdrv_drop_filter().
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block/backup-top.c | 38 +-
tests/qemu-iotests/283.out | 2 +-
2 files changed, 2
Using bdrv_replace_node() for removing filter is not good enough: it
keeps child reference of the filter, which may conflict with original
top node during permission update.
Instead let's create new interface, which will do all graph
modifications first and then update permissions.
Let's modify
Refactor calling driver callbacks to a separate transaction action to
be used later.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 71 -
1 file changed, 55 insertions(+), 16 deletions(-)
diff --git a/block.c b/block.c
index
Return int status to avoid extra error propagation schemes.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
include/block/block.h | 4 ++--
block.c | 15 ---
block/commit.c | 6 ++
block/mirror.c | 6 ++
blockdev.c
We are going to drop recursive bdrv_child_* functions, so stop use them
in bdrv_child_try_set_perm() as a first step.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 14 --
1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/block.c b/block.c
index
Add test to show that simple DFS recursion order is not correct for
permission update. Correct order is topological-sort order, which will
be introduced later.
Consider the block driver which has two filter children: one active
with exclusive write access and one inactive with no specific
We have too much comments for this feature. It seems better just don't
do it. Most of real users (tests don't count) have to create additional
reference.
Drop also comment in external_snapshot_prepare:
- bdrv_append doesn't "remove" old bs in common sense, it sounds
strange
- the fact that
inore_children thing doesn't help to track all propagated permissions
of children we want to ignore. The simplest way to correctly update
permissions is update graph first and then do permission update. In
this case we just referesh permissions for the whole subgraph (in
topological-sort defined
Functions with errp argument are not recommened to be void-functions.
Improve bdrv_replace_node().
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
include/block/block.h | 4 ++--
block.c | 14 --
2 files changed, 10 insertions(+), 8 deletions(-)
diff --git
To be used in the following commit.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 54 ++
1 file changed, 54 insertions(+)
diff --git a/block.c b/block.c
index d799afeedd..1b10b6fb5e 100644
--- a/block.c
+++ b/block.c
@@ -82,6
Split part of bdrv_replace_node_common() to be used separately.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 47 ++-
1 file changed, 30 insertions(+), 17 deletions(-)
diff --git a/block.c b/block.c
index 5f6ad1d016..1327254b8e 100644
---
The code partly duplicates bdrv_root_attach_child() and
bdrv_attach_child(). Still refactoring these two functions by renaming
them to *_common with new noperm argument is more complicating. When
all operations moved to new graph update paradigm (update permissions
only on updated graph) all
First, permission update loop tries to do iterations transactionally,
but the whole update is not transactional: nobody roll-back successful
loop iterations when some iteration fails.
Second, in the iteration we have nested permission update:
c->klass->update_filename may point to
bdrv_append is not very good for inserting filters: it does extra
permission update as part of bdrv_set_backing_hd(). During this update
filter may conflict with other parents of top_bs.
Instead, let's first do all graph modifications and after it update
permissions.
Note: bdrv_append() is still
Rewrite bdrv_check_perm(), bdrv_abort_perm_update() and bdrv_set_perm()
to update nodes in topological sort order instead of simple DFS. With
topologically sorted nodes, we update a node only when all its parents
already updated. With DFS it's not so.
Consider the following example:
A -+
Each of them has only one caller. Open-coding simplifies further
pemission-update system changes.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 59 +
1 file changed, 17 insertions(+), 42 deletions(-)
diff --git a/block.c
Add new interface, allowing use of existing node list. It will be used
to fix bdrv_replace_node() in the following commit.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 112 ++--
1 file changed, 76 insertions(+), 36 deletions(-)
Split out non-recursive parts, and refactor as block graph transaction
action.
Signed-off-by: Vladimir Sementsov-Ogievskiy
---
block.c | 79 ++---
1 file changed, 59 insertions(+), 20 deletions(-)
diff --git a/block.c b/block.c
index
Add additional check that node parents do not interfere with each
other. This should not hurt existing callers and allows in further
patch use bdrv_refresh_perms() to update a subtree of changed
BdrvChild (check that change is correct).
New check will substitute bdrv_check_update_perm() in
Add simple transaction API to use in further update of block graph
operations.
Supposed usage is:
- "prepare" is main function of the action and it should make the main
effect of the action to be visible for the following actions, keeping
possibility of roll-back, saving necessary things in
Add the test that shows that concept of ignore_children is incomplete.
Actually, when we want to update something, ignoring permission of some
existing BdrvChild, we should ignore also the propagated effect of this
child to the other children. But that's not done. Better approach
(update
On permission update commit we must set same permissions as on _check_.
Let's add assertions. Next step may be to drop permission parameters
from _set_.
Note that prior to previous commit, fixing bdrv_drop_intermediate(),
new assertion in bdrv_child_set_perm() crashes on iotests 30 and 40.
Hi all!
Here is a proposal of updating graph changing procedures.
The thing brought me here is a question about "activating" filters after
insertion, which is done in mirror_top and backup_top. The problem is
that we can't simply avoid permission conflict when inserting the
filter: during
On Mon, 2020-11-23 at 19:20 +0100, Alberto Garcia wrote:
> On Mon 23 Nov 2020 05:09:41 PM CET, Kevin Wolf wrote:
> > > Commit 205fa50750 ("qcow2: Add subcluster support to zero_in_l2_slice()")
> > > introduced a subtle change to code in zero_in_l2_slice:
> > >
> > > It swapped the order of
> > >
On Mon 23 Nov 2020 05:09:41 PM CET, Kevin Wolf wrote:
>> Commit 205fa50750 ("qcow2: Add subcluster support to zero_in_l2_slice()")
>> introduced a subtle change to code in zero_in_l2_slice:
>>
>> It swapped the order of
>>
>> 1. qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
>> 2.
On Mon, 2020-11-23 at 18:38 +0100, Kevin Wolf wrote:
> Am 23.11.2020 um 16:49 hat Maxim Levitsky geschrieben:
> > Commit 205fa50750 ("qcow2: Add subcluster support to zero_in_l2_slice()")
> > introduced a subtle change to code in zero_in_l2_slice:
> >
> > It swapped the order of
> >
> > 1.
Am 23.11.2020 um 16:49 hat Maxim Levitsky geschrieben:
> Commit 205fa50750 ("qcow2: Add subcluster support to zero_in_l2_slice()")
> introduced a subtle change to code in zero_in_l2_slice:
>
> It swapped the order of
>
> 1. qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
> 2.
On 23.11.2020 18:44, Andrey Shinkevich wrote:
The subject was discussed here:
https://lists.gnu.org/archive/html/qemu-devel/2017-05/msg00206.html
This series is a solution for the issue with QMP monitor buffered input.
A JSON little parser is introduced to separate QMP commands read from the
On 23.11.2020 18:44, Andrey Shinkevich wrote:
The subject was discussed here:
https://lists.gnu.org/archive/html/qemu-devel/2017-05/msg00206.html
This series is a solution for the issue with QMP monitor buffered input.
A JSON little parser is introduced to separate QMP commands read from the
Am 23.11.2020 um 16:49 hat Maxim Levitsky geschrieben:
> Commit 205fa50750 ("qcow2: Add subcluster support to zero_in_l2_slice()")
> introduced a subtle change to code in zero_in_l2_slice:
>
> It swapped the order of
>
> 1. qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
> 2.
On this weekend, I had discovered that one of my VMs started to act weird.
Due to this, I found out that it and most of the other VMs I have,
have grown an qcow2 corruption.
So after some bisecting, digging through dumps, and debugging,
I think I found the root cause and a fix.
In addition to
Commit 205fa50750 ("qcow2: Add subcluster support to zero_in_l2_slice()")
introduced a subtle change to code in zero_in_l2_slice:
It swapped the order of
1. qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
2. set_l2_entry(s, l2_slice, l2_index + i, QCOW_OFLAG_ZERO);
3.
This patch paves the way for the one that follows. The following patch
makes the QMP monitor to read up to 4K from stdin at once. That results
in running the bash 'sleep' command before the _qemu_proc_exec() starts
in subshell. Another 'sleep' command with an unobtrusive 'query-status'
plays as a
QMP and HMP monitors read one byte at a time from the socket or stdin,
which is very inefficient. With 100+ VMs on the host, this results in
multiple extra system calls and CPU overuse.
This patch increases the amount of read data up to 4096 bytes that fits
the buffer size on the channel level.
A
The subject was discussed here:
https://lists.gnu.org/archive/html/qemu-devel/2017-05/msg00206.html
This series is a solution for the issue with QMP monitor buffered input.
A JSON little parser is introduced to separate QMP commands read from the
input buffer so that incoming requests do not
On Mon, 23 Nov 2020 at 12:55, Philippe Mathieu-Daudé wrote:
>
> On 11/4/20 9:59 PM, Peter Maydell wrote:
> > On Wed, 4 Nov 2020 at 15:18, Stefan Hajnoczi wrote:
> >>
> >> The following changes since commit
> >> 8507c9d5c9a62de2a0e281b640f995e26eac46af:
> >>
> >> Merge remote-tracking branch
23.11.2020 14:10, Kevin Wolf wrote:
Am 23.11.2020 um 11:29 hat Vladimir Sementsov-Ogievskiy geschrieben:
23.11.2020 13:10, Kevin Wolf wrote:
Am 20.11.2020 um 19:19 hat Vladimir Sementsov-Ogievskiy geschrieben:
20.11.2020 20:22, Kevin Wolf wrote:
Am 20.11.2020 um 17:43 hat Vladimir
On 11/4/20 9:59 PM, Peter Maydell wrote:
> On Wed, 4 Nov 2020 at 15:18, Stefan Hajnoczi wrote:
>>
>> The following changes since commit 8507c9d5c9a62de2a0e281b640f995e26eac46af:
>>
>> Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into
>> staging (2020-11-03 15:59:44 +)
>>
[ Cc: qemu-block ]
Am 23.11.2020 um 10:36 hat Quentin Grolleau geschrieben:
> Hello,
>
> In our company, we are hosting a large number of Vm, hosted behind Openstack
> (so libvirt/qemu).
> A large majority of our Vms are runnign with local data only, stored on NVME,
> and most of them are RAW
On Wed, Nov 18, 2020 at 09:16:40AM +, Stefan Hajnoczi wrote:
Markus Armbruster pointed out that g_return_val_if() is meant for programming
errors. It must not be used for input validation since it can be compiled out.
Use explicit if statements instead.
This patch series converts vhost-user
Am 23.11.2020 um 11:29 hat Vladimir Sementsov-Ogievskiy geschrieben:
> 23.11.2020 13:10, Kevin Wolf wrote:
> > Am 20.11.2020 um 19:19 hat Vladimir Sementsov-Ogievskiy geschrieben:
> > > 20.11.2020 20:22, Kevin Wolf wrote:
> > > > Am 20.11.2020 um 17:43 hat Vladimir Sementsov-Ogievskiy geschrieben:
23.11.2020 13:10, Kevin Wolf wrote:
Am 20.11.2020 um 19:19 hat Vladimir Sementsov-Ogievskiy geschrieben:
20.11.2020 20:22, Kevin Wolf wrote:
Am 20.11.2020 um 17:43 hat Vladimir Sementsov-Ogievskiy geschrieben:
20.11.2020 19:36, Kevin Wolf wrote:
Am 20.11.2020 um 17:16 hat Vladimir
Am 20.11.2020 um 19:19 hat Vladimir Sementsov-Ogievskiy geschrieben:
> 20.11.2020 20:22, Kevin Wolf wrote:
> > Am 20.11.2020 um 17:43 hat Vladimir Sementsov-Ogievskiy geschrieben:
> > > 20.11.2020 19:36, Kevin Wolf wrote:
> > > > Am 20.11.2020 um 17:16 hat Vladimir Sementsov-Ogievskiy geschrieben:
49 matches
Mail list logo
