[libvirt PATCH v3] hypervisor: Move interface mgmt methods to hypervisor

[Praveen K Paladugu]

Move guest interface management methods from qemu to hypervisor. These
methods will be shared by networking support in ch driver.

Signed-off-by: Praveen K Paladugu 
---
 po/POTFILES   |   1 +
 src/hypervisor/domain_interface.c | 368 ++
 src/hypervisor/domain_interface.h |  41 
 src/hypervisor/meson.build|   1 +
 src/libvirt_private.syms  |   8 +
 src/qemu/qemu_command.c   |   6 +-
 src/qemu/qemu_hotplug.c   |   5 +-
 src/qemu/qemu_interface.c | 339 +--
 src/qemu/qemu_interface.h |  11 -
 src/qemu/qemu_process.c   |   5 +-
 10 files changed, 434 insertions(+), 351 deletions(-)
 create mode 100644 src/hypervisor/domain_interface.c
 create mode 100644 src/hypervisor/domain_interface.h

diff --git a/po/POTFILES b/po/POTFILES
index 3a51aea5cb..023c041f61 100644
--- a/po/POTFILES
+++ b/po/POTFILES
@@ -92,6 +92,7 @@ src/hyperv/hyperv_util.c
 src/hyperv/hyperv_wmi.c
 src/hypervisor/domain_cgroup.c
 src/hypervisor/domain_driver.c
+src/hypervisor/domain_interface.c
 src/hypervisor/virhostdev.c
 src/interface/interface_backend_netcf.c
 src/interface/interface_backend_udev.c
diff --git a/src/hypervisor/domain_interface.c 
b/src/hypervisor/domain_interface.c
new file mode 100644
index 00..f726cb2bea
--- /dev/null
+++ b/src/hypervisor/domain_interface.c
@@ -0,0 +1,368 @@
+/*
+ * Copyright (C) 2015-2016 Red Hat, Inc.
+ * Copyright IBM Corp. 2014
+ *
+ * domain_interface.c: methods to manage guest/domain interfaces
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library 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
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library.  If not, see
+ * .
+ */
+
+#include 
+
+#include "virconftypes.h"
+#include "virmacaddr.h"
+#include "virnetdevtap.h"
+#include "domain_audit.h"
+#include "domain_conf.h"
+#include "domain_interface.h"
+#include "domain_nwfilter.h"
+#include "viralloc.h"
+#include "virebtables.h"
+#include "virfile.h"
+#include "virlog.h"
+#include "virnetdevbridge.h"
+#include "network_conf.h"
+
+#define VIR_FROM_THIS VIR_FROM_INTERFACE
+
+VIR_LOG_INIT("interface.interface_connect");
+
+bool
+virDomainInterfaceIsVnetCompatModel(const virDomainNetDef *net)
+{
+return (virDomainNetIsVirtioModel(net) ||
+net->model == VIR_DOMAIN_NET_MODEL_E1000E ||
+net->model == VIR_DOMAIN_NET_MODEL_IGB ||
+net->model == VIR_DOMAIN_NET_MODEL_VMXNET3);
+}
+
+/* virDomainInterfaceEthernetConnect:
+ * @def: the definition of the VM
+ * @driver: qemu driver data
+ * @net: pointer to the VM's interface description
+ * @tapfd: array of file descriptor return value for the new device
+ * @tapfdsize: number of file descriptors in @tapfd
+ *
+ * Called *only* called if actualType is VIR_DOMAIN_NET_TYPE_ETHERNET
+ * (i.e. if the connection is made with a tap device)
+ */
+int
+virDomainInterfaceEthernetConnect(virDomainDef *def,
+virDomainNetDef *net,
+ebtablesContext *ebtables,
+bool macFilter,
+bool privileged,
+int *tapfd,
+size_t tapfdSize)
+{
+virMacAddr tapmac;
+int ret = -1;
+unsigned int tap_create_flags = VIR_NETDEV_TAP_CREATE_IFUP;
+bool template_ifname = false;
+const char *tunpath = "/dev/net/tun";
+const char *auditdev = tunpath;
+
+if (net->backend.tap) {
+tunpath = net->backend.tap;
+if (!privileged) {
+virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
+   _("cannot use custom tap device in session mode"));
+goto cleanup;
+}
+}
+
+if (virDomainInterfaceIsVnetCompatModel(net))
+tap_create_flags |= VIR_NETDEV_TAP_CREATE_VNET_HDR;
+
+if (net->managed_tap == VIR_TRISTATE_BOOL_NO) {
+if (!net->ifname) {
+virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+   _("target dev must be supplied when managed='no'"));
+goto cleanup;
+}
+if (virNetDevExists(net->ifname) != 1) {
+virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+   _("target managed='no' but specified dev doesn't 
exist"));
+goto cleanup;
+}
+
+tap_create_flags |= VIR_NETDEV_TAP_CREATE_ALLOW_EXISTING;

Re: [PATCH libvirt v1 0/3] Ensure full early console access with libvirt

[Marc Hartmayer]

On Wed, Oct 11, 2023 at 10:05 AM +0200, "Marc Hartmayer" 
 wrote:
> On Thu, Sep 28, 2023 at 05:37 PM +0200, Marc Hartmayer 
>  wrote:
>> Currently, early console output may be lost, e.g. if starting a guest with
>> `virsh start --console` guest, which can make debugging of early failures 
>> very
>> difficult
>> (like zipl messages or disabled wait conditions happening early). This is
>> because QEMU may emit serial console output before the libvirt console client
>> starts to consume data from the pts. This can be prevented by starting the 
>> guest
>> in paused state, connect to the console and then resume the guest.
>>

>> Note: There is still a problem in QEMU itself, see QEMU patch series `[PATCH]
>> chardev/char-pty: Avoid losing bytes when the other side just (re-)connected`
>> [1]

This patch is now accepted upstream.

>>
>> Changelog:
>> RFCv1->v1:
>> + rebased on current master
>> + worked in comments from Daniel
>>
>> [1] https://lists.gnu.org/archive/html/qemu-devel/2023-08/msg02725.html
>>
>> Marc Hartmayer (3):
>>   virsh: add `console --resume` support
>>   Improve `virsh start --console` behavior
>>   Improve `virsh create --console` behavior
>>
>>  tools/virsh-console.c |  8 
>>  tools/virsh-console.h |  1 +
>>  tools/virsh-domain.c  | 94 ---
>>  3 files changed, 80 insertions(+), 23 deletions(-)
>>
>>
>> base-commit: dd403f8873cf8de7675b89ed757a4228af7bc05e
>> -- 
>> 2.34.1
>>
>
> Polite ping and adding Daniel to CC - sry I missed that in the
> beginning.
>
> -- 
> Kind regards / Beste Grüße
>Marc Hartmayer
>
> IBM Deutschland Research & Development GmbH
> Vorsitzender des Aufsichtsrats: Gregor Pillen
> Geschäftsführung: David Faller
> Sitz der Gesellschaft: Böblingen
> Registergericht: Amtsgericht Stuttgart, HRB 243294

Very friendly ping.

-- 
Kind regards / Beste Grüße
   Marc Hartmayer

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Gregor Pillen
Geschäftsführung: David Faller
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294

Re: [libvirt PATCH v2] hypervisor: Move interface mgmt methods to hypervisor

[Praveen Paladugu]

On Wed, Oct 18, 2023 at 03:36:47PM -0400, Laine Stump wrote:
> On 10/16/23 3:34 PM, Praveen K Paladugu wrote:
> >Move guest interface management methods from qemu to hypervisor. These
> >methods will be shared by networking support in ch driver.
> >
> >Signed-off-by: Praveen K Paladugu 
> >---
> >  po/POTFILES   |   1 +
> >  src/hypervisor/domain_interface.c | 280 ++
> >  src/hypervisor/domain_interface.h |  39 +
> >  src/hypervisor/meson.build|   1 +
> >  src/libvirt_private.syms  |   6 +
> >  src/qemu/qemu_command.c   |   7 +-
> >  src/qemu/qemu_hotplug.c   |   3 +-
> >  src/qemu/qemu_interface.c | 246 +-
> >  src/qemu/qemu_interface.h |   6 -
> >  src/qemu/qemu_process.c   |   3 +-
> >  10 files changed, 343 insertions(+), 249 deletions(-)
> >  create mode 100644 src/hypervisor/domain_interface.c
> >  create mode 100644 src/hypervisor/domain_interface.h
> >
> >diff --git a/po/POTFILES b/po/POTFILES
> >index 3a51aea5cb..023c041f61 100644
> >--- a/po/POTFILES
> >+++ b/po/POTFILES
> >@@ -92,6 +92,7 @@ src/hyperv/hyperv_util.c
> >  src/hyperv/hyperv_wmi.c
> >  src/hypervisor/domain_cgroup.c
> >  src/hypervisor/domain_driver.c
> >+src/hypervisor/domain_interface.c
> >  src/hypervisor/virhostdev.c
> >  src/interface/interface_backend_netcf.c
> >  src/interface/interface_backend_udev.c
> >diff --git a/src/hypervisor/domain_interface.c 
> >b/src/hypervisor/domain_interface.c
> >new file mode 100644
> >index 00..592c4253df
> >--- /dev/null
> >+++ b/src/hypervisor/domain_interface.c
> >@@ -0,0 +1,280 @@
> >+/*
> >+ * Copyright (C) 2015-2016 Red Hat, Inc.
> >+ * Copyright IBM Corp. 2014
> >+ *
> >+ * domain_interface.c: methods to manage guest/domain interfaces
> >+ *
> >+ * This library is free software; you can redistribute it and/or
> >+ * modify it under the terms of the GNU Lesser General Public
> >+ * License as published by the Free Software Foundation; either
> >+ * version 2.1 of the License, or (at your option) any later version.
> >+ *
> >+ * This library 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
> >+ * Lesser General Public License for more details.
> >+ *
> >+ * You should have received a copy of the GNU Lesser General Public
> >+ * License along with this library.  If not, see
> >+ * .
> >+ */
> >+
> >+#include 
> >+
> >+#include "virconftypes.h"
> >+#include "virmacaddr.h"
> >+#include "virnetdevtap.h"
> >+#include "domain_audit.h"
> >+#include "domain_conf.h"
> >+#include "domain_interface.h"
> >+#include "domain_nwfilter.h"
> >+#include "viralloc.h"
> >+#include "virebtables.h"
> >+#include "virfile.h"
> >+#include "virlog.h"
> >+#include "virnetdevbridge.h"
> >+#include "network_conf.h"
> >+
> >+#define VIR_FROM_THIS VIR_FROM_INTERFACE
> >+
> >+VIR_LOG_INIT("interface.interface_connect");
> >+
> >+bool
> >+virDomainInterfaceIsVnetCompatModel(const virDomainNetDef *net)
> >+{
> >+return (virDomainNetIsVirtioModel(net) ||
> >+net->model == VIR_DOMAIN_NET_MODEL_E1000E ||
> >+net->model == VIR_DOMAIN_NET_MODEL_IGB ||
> >+net->model == VIR_DOMAIN_NET_MODEL_VMXNET3);
> >+}
> >+
> >+/* virDomainInterfaceEthernetConnect:
> >+ * @def: the definition of the VM
> >+ * @driver: qemu driver data
> >+ * @net: pointer to the VM's interface description
> >+ * @tapfd: array of file descriptor return value for the new device
> >+ * @tapfdsize: number of file descriptors in @tapfd
> >+ *
> >+ * Called *only* called if actualType is VIR_DOMAIN_NET_TYPE_ETHERNET
> >+ * (i.e. if the connection is made with a tap device)
> >+ */
> >+int
> >+virDomainInterfaceEthernetConnect(virDomainDef *def,
> >+virDomainNetDef *net,
> >+ebtablesContext *ebtables,
> >+bool macFilter,
> >+bool privileged,
> >+int *tapfd,
> >+size_t tapfdSize)
> 
> This is all fine as far as it goes, but I was expecting that all of
> the functions at the same level (qemuInterface*Connect()) would be
> genericized and moved to the new file. I guess you're only planning
> to use the plain "ethernet" type of interface in the ch driver, but
> for QEMU to have some of the *Connect functions in one place and
> some in another makes the code more confusing; this is especially so
> since virDomainInterfaceStartDevice() has cases for these other
> interface types.

I started with enabling the "ethernet" mode, so I moved all the related
methods. I am planning to also enable "Bridge" and "NAT" modes too in follow
up patches. I did this to test the methods properly before I push.

> 
> I'm guessing

[PATCH] Send event on persistent config modification

[Fima Shevrin]

Currently, libvirt doesn't send events when devices are attached,
detached or updated. Thus, any services that listen to events are
unaware of the change to persistent config.

Signed-off-by: Fima Shevrin 
---
 src/qemu/qemu_driver.c | 25 -
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c
index 43d96739d5..86da8da777 100644
--- a/src/qemu/qemu_driver.c
+++ b/src/qemu/qemu_driver.c
@@ -7209,6 +7209,7 @@ qemuDomainAttachDeviceLiveAndConfig(virDomainObj *vm,
 unsigned int flags)
 {
 qemuDomainObjPrivate *priv = vm->privateData;
+virObjectEvent *event = NULL;
 g_autoptr(virDomainDef) vmdef = NULL;
 g_autoptr(virQEMUDriverConfig) cfg = NULL;
 g_autoptr(virDomainDeviceDef) devConf = NULL;
@@ -7292,6 +7293,12 @@ qemuDomainAttachDeviceLiveAndConfig(virDomainObj *vm,
 return -1;
 
 virDomainObjAssignDef(vm, &vmdef, false, NULL);
+
+/* Event sending if persistent config has changed */
+event = virDomainEventLifecycleNewFromObj(vm,
+  VIR_DOMAIN_EVENT_DEFINED,
+  
VIR_DOMAIN_EVENT_DEFINED_UPDATED);
+virObjectEventStateQueue(driver->domainEventState, event);
 }
 
 return 0;
@@ -7348,6 +7355,7 @@ qemuDomainUpdateDeviceFlags(virDomainPtr dom,
 virQEMUDriver *driver = dom->conn->privateData;
 virDomainObj *vm = NULL;
 qemuDomainObjPrivate *priv;
+virObjectEvent *event = NULL;
 g_autoptr(virDomainDef) vmdef = NULL;
 g_autoptr(virDomainDeviceDef) dev_config = NULL;
 g_autoptr(virDomainDeviceDef) dev_live = NULL;
@@ -7419,8 +7427,16 @@ qemuDomainUpdateDeviceFlags(virDomainPtr dom,
 /* Finally, if no error until here, we can save config. */
 if (flags & VIR_DOMAIN_AFFECT_CONFIG) {
 ret = virDomainDefSave(vmdef, driver->xmlopt, cfg->configDir);
-if (!ret)
+if (!ret) {
 virDomainObjAssignDef(vm, &vmdef, false, NULL);
+
+/* Event sending if persistent config has changed */
+event = virDomainEventLifecycleNewFromObj(vm,
+ VIR_DOMAIN_EVENT_DEFINED,
+ 
VIR_DOMAIN_EVENT_DEFINED_UPDATED);
+
+virObjectEventStateQueue(driver->domainEventState, event);
+}
 }
 
  endjob:
@@ -7438,6 +7454,7 @@ qemuDomainDetachDeviceLiveAndConfig(virQEMUDriver *driver,
 unsigned int flags)
 {
 qemuDomainObjPrivate *priv = vm->privateData;
+virObjectEvent *event = NULL;
 g_autoptr(virQEMUDriverConfig) cfg = NULL;
 g_autoptr(virDomainDeviceDef) dev_config = NULL;
 g_autoptr(virDomainDeviceDef) dev_live = NULL;
@@ -7495,6 +7512,12 @@ qemuDomainDetachDeviceLiveAndConfig(virQEMUDriver 
*driver,
 return -1;
 
 virDomainObjAssignDef(vm, &vmdef, false, NULL);
+
+/* Event sending if persistent config has changed */
+event = virDomainEventLifecycleNewFromObj(vm,
+  VIR_DOMAIN_EVENT_DEFINED,
+  
VIR_DOMAIN_EVENT_DEFINED_UPDATED);
+virObjectEventStateQueue(driver->domainEventState, event);
 }
 
 return 0;
-- 
2.39.3

Plans for 9.9.0 release (freeze on Thursday 26 Oct)

[Jiri Denemark]

We are getting close to 9.9.0 release of libvirt. To aim for the
release on Wednesday 01 Nov I suggest entering the freeze on Thursday
26 Oct and tagging RC2 on Monday 30 Oct.

I hope this works for everyone.

Jirka