On Fri, Apr 12, 2024 at 08:52:37PM +0200, Michal Privoznik wrote:
> clang on Fedora started to complain about some calls to g_new0()
> we're making in vbox_snapshot_conf.c. Specifically, we're passing
> zero as number of elements to allocate. And while usually SA
> tools are not clever, in this spe
clang on Fedora started to complain about some calls to g_new0()
we're making in vbox_snapshot_conf.c. Specifically, we're passing
zero as number of elements to allocate. And while usually SA
tools are not clever, in this specific case clang is right.
There are three cases where such call is made,
On Fri, Apr 12, 2024 at 17:18:41 +0200, Michal Prívozník wrote:
> On 4/12/24 16:33, Peter Krempa wrote:
> > On Fri, Apr 12, 2024 at 12:19:05 +0200, Michal Privoznik wrote:
> >> In QEMU and LXC drivers in a few places just
> >
> > s/just/only/ so that's more clear that something is missing.
> >
>
On 4/12/24 16:33, Peter Krempa wrote:
> On Fri, Apr 12, 2024 at 12:19:05 +0200, Michal Privoznik wrote:
>> In QEMU and LXC drivers in a few places just
>
> s/just/only/ so that's more clear that something is missing.
>
Fixed locally. Does it warrant your Reviewed-by? ;-)
Michal
On Fri, Apr 12, 2024 at 12:19:05 +0200, Michal Privoznik wrote:
> In QEMU and LXC drivers in a few places just
s/just/only/ so that's more clear that something is missing.
> virNetDevBandwidthClear() is called. This means that if an
> interface is of openvswitch vport profile, its QoS is not
> re
Yu,
On Thu, Apr 11, 2024 at 06:36:54PM +0200, Yu Zhang wrote:
> > 1) Either a CI test covering at least the major RDMA paths, or at least
> > periodically tests for each QEMU release will be needed.
> We use a batch of regression test cases for the stack, which covers the
> test for QEMU. I di
On Fri, Apr 12, 2024 at 03:36 PM +0200, Marc Hartmayer
wrote:
> From: Boris Fiuczynski
>
> Two situations will trigger an udev add event:
> 1) the mdev is created when started (transient) or
> 2) the mdev was defined and is started
> In case 1 there is no node object existing and no config dat
There is no reason to serialize the `mdevctl list` calls.
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 7 +--
1 file changed, 1 insertion(+), 6 deletions(-)
diff --git a/src/node_device/node_device_udev.c
b/src/node_device/node_device_udev.c
index 9282afdd3241..77
Inline `udevRemoveOneDevice` as it's used only once.
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 17 +
1 file changed, 5 insertions(+), 12 deletions(-)
diff --git a/src/node_device/node_device_udev.c
b/src/node_device/node_device_udev.c
index a121ad99
When an udev event occurs the mdev active config data requires an update via
mdevctl as the udev does not contain all config data. This update needs to occur
immediate and to be finished before the libvirt nodedev event is issued to keep
the API usage reliable.
The only case where a direct `nodeDe
Use a worker pool for processing the udev events and the initialization instead
of a separate initThread and a mdevctl-thread. This has the large advantage that
we can leverage the job API and now this thread pool is responsible to do all
the "costly-work" and the libvirt nodedev event creation.
T
There is only one case where force is true, therefore let's inline that case.
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 25 +++--
1 file changed, 11 insertions(+), 14 deletions(-)
diff --git a/src/node_device/node_device_udev.c
b/src/node_device
Use the proper driver functions for the node state shutdown preparation and
wait. In the next patch, these functions will be extended.
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 54 +++---
1 file changed, 42 insertions(+), 12 deletions(-)
diff
Since @driver->privateData is modified take the lock.
Reviewed-by: Boris Fiuczynski
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 15 +++
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/src/node_device/node_device_udev.c
b/src/node_device/nod
From: Boris Fiuczynski
When an udev add event occurs the mdev active config data requires an
update via mdevctl as the udev does not contain all config data.
This update needs to occur immediate and to be finished before the
libvirt CREATE event is issued to keep the API usage reliable.
After th
The new names make it easier to understand the purpose of the data.
Reviewed-by: Boris Fiuczynski
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 48 +++---
1 file changed, 24 insertions(+), 24 deletions(-)
diff --git a/src/node_device/node_device
Commit a99d876a0f58 ("node_device: Use automatic mutex management") replaced the
locking mechanism and accidentally removed the comment with the reason why the
lock is taken. Restore this comment and add a new comment about the lock.
Reviewed-by: Boris Fiuczynski
Signed-off-by: Marc Hartmayer
--
It is done a little differently everywhere in libvirt, but most common is to
test for != -1.
Reviewed-by: Boris Fiuczynski
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/src/node_device/node_device_
Remove the timeout when the udevEventData is disposed, analogous to priv->watch.
Reviewed-by: Boris Fiuczynski
Signed-off-by: Marc Hartmayer
---
src/node_device/node_device_udev.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/node_device/node_device_udev.c
b/src/node_device/node_d
From: Boris Fiuczynski
When a mdev device is destroyed or stopped the udev remove event
handling needs to reset the active config data of the node object
representing a persisted mdev.
Reviewed-by: Marc Hartmayer
Signed-off-by: Boris Fiuczynski
Signed-off-by: Marc Hartmayer
---
src/node_devi
@def is owned by @obj after adding it the node device object list. As soon as
the @obj lock has been released, another thread could free @obj and therefore
@def. If now someone accesses @def this would lead to a heap-use-after-free and
therefore most likely to a segmentation fault, therefore set @d
From: Boris Fiuczynski
Two situations will trigger an udev add event:
1) the mdev is created when started (transient) or
2) the mdev was defined and is started
In case 1 there is no node object existing and no config data is copied.
In case 2 copying the active config data of an existing node o
When an udev event occurs for a mediated device (mdev) the mdev config data
requires an update via mdevctl as the udev event does not contain all config
data. This update needs to occur immediate and to be finished before the libvirt
nodedev event is issued to keep the API usage reliable.
This pat
On Fri, Apr 12, 2024 at 12:19:04 +0200, Michal Privoznik wrote:
> The reason virDomainClearNetBandwidth() exists in src/conf/ is
> that at the time its introduction we did not have a better place.
> But now we do. Firstly, virDomainClearNetBandwidth() is
> hypervisor agnostic code, but really has n
On Fri, Apr 12, 2024 at 12:19:03 +0200, Michal Privoznik wrote:
> The @brname argument of virNetDevOpenvswitchRemovePort() is and
> was unused ever since its introduction in v0.9.11-rc1~257. Just
> remove it.
>
> Signed-off-by: Michal Privoznik
> ---
Reviewed-by: Peter Krempa
__
On Fri, Apr 12, 2024 at 12:19:02 +0200, Michal Privoznik wrote:
> Both LXC and QEMU drivers have the same code to remove vport when
> removing a domain's interface. Instead of repeating the same
> pattern in both drivers, move the code into hypervisor agnostic
> location (src/hypervisor/) and switc
On Fri, Apr 12, 2024 at 12:19:01 +0200, Michal Privoznik wrote:
> The comment to virNetDevOpenvswitchInterfaceGetMaster() contains
> wrong function name. Fix this.
>
> Signed-off-by: Michal Privoznik
> ---
Reviewed-by: Peter Krempa
___
Devel mailing l
On Fri, Apr 12, 2024 at 12:31:26 +0200, Michal Privoznik wrote:
> The vshFindTypedParamByName() function no longer exists (as of
> v1.0.2-rc1~82), but its header file declaration was still kept
> around. Drop it.
>
> Signed-off-by: Michal Privoznik
> ---
Trivial;
Reviewed-by: Peter Krempa
The vshFindTypedParamByName() function no longer exists (as of
v1.0.2-rc1~82), but its header file declaration was still kept
around. Drop it.
Signed-off-by: Michal Privoznik
---
tools/vsh.h | 3 ---
1 file changed, 3 deletions(-)
diff --git a/tools/vsh.h b/tools/vsh.h
index f06d65407d..eeba1d4
In QEMU and LXC drivers in a few places just
virNetDevBandwidthClear() is called. This means that if an
interface is of openvswitch vport profile, its QoS is not
removed. And to make matters worse - OVS is designed to remember
state even when corresponding interface is gone. This leads to
stale QoS
The reason virDomainClearNetBandwidth() exists in src/conf/ is
that at the time its introduction we did not have a better place.
But now we do. Firstly, virDomainClearNetBandwidth() is
hypervisor agnostic code, but really has nothing to do with
domain configuration (it doesn't parse/format XML). Se
Both LXC and QEMU drivers have the same code to remove vport when
removing a domain's interface. Instead of repeating the same
pattern in both drivers, move the code into hypervisor agnostic
location (src/hypervisor/) and switch to calling this new
function.
Signed-off-by: Michal Privoznik
---
s
The @brname argument of virNetDevOpenvswitchRemovePort() is and
was unused ever since its introduction in v0.9.11-rc1~257. Just
remove it.
Signed-off-by: Michal Privoznik
---
src/hypervisor/domain_interface.c | 4 +---
src/util/virnetdevopenvswitch.c | 2 +-
src/util/virnetdevopenvswitch.h |
The comment to virNetDevOpenvswitchInterfaceGetMaster() contains
wrong function name. Fix this.
Signed-off-by: Michal Privoznik
---
src/util/virnetdevopenvswitch.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/util/virnetdevopenvswitch.c b/src/util/virnetdevopenvswitch.
*** BLURB HERE ***
Michal Prívozník (5):
virnetdevopenvswitch: Fix comment to
virNetDevOpenvswitchInterfaceGetMaster()
hypervisor: Introduce and use virDomainInterfaceVportRemove()
virnetdevopenvswitch: Drop @brname arg from
virNetDevOpenvswitchRemovePort()
conf: Move virDomainClea
On Thu, Apr 11, 2024 at 05:50 PM +0200, Boris Fiuczynski
wrote:
> Reviewed-by: Boris Fiuczynski
>
> On 4/3/24 16:03, Marc Hartmayer wrote:
>> Since @driver->privateData is modified take the lock.
>>
>> Question: In theory we could take the udevEventData->mdevctlLock?
>
> Isn't that protecting t
36 matches
Mail list logo
