Add pci passthrough implementation to libxl driver, using hostdev common
library.
---
src/libxl/libxl_conf.c | 44
src/libxl/libxl_driver.c | 21 -
2 files changed, 64 insertions(+), 1 deletions(-)
diff --git
Write separate module for hostdev passthrough so that it could be used by all
hypervisor drivers and maintain a global hostdev state.
---
po/POTFILES.in |1 +
src/Makefile.am |1 +
src/libvirt.c|5 +
src/libvirt_private.syms | 15 +
This patch sets include two patches:
1/2 is the implementation of the hostdev passthrough common library.
To meet two purposes:
a. move qemu hostdev APIs to common library so that it could be used by all
hypervisor drivers.
b. maintain a global hostdev in-use state.
2/2 is the implementation of
On 05/13/2013 01:23 PM, Michal Privoznik wrote:
With multiqueue network feature, we are advised to pass multiple
vhost-net FDs as well. The ratio should be 1:1. Therefore we must
alter the qemuOpenVhostNet function to allow that.
---
src/qemu/qemu_command.c | 40
Write separate module for hostdev passthrough so that it could be used by all
hypervisor drivers and maintain a global hostdev state.
Signed-off-by: Chunyan Liu cy...@suse.com
---
po/POTFILES.in |1 +
src/Makefile.am |1 +
src/libvirt.c|5 +
This patch sets include two patches:
1/2 is the implementation of the hostdev passthrough common library.
To meet two purposes:
a. move qemu hostdev APIs to common library so that it could be used by all
hypervisor drivers.
b. maintain a global hostdev in-use state.
2/2 is the implementation of
Add pci passthrough implementation to libxl driver, using hostdev common
library.
Signed-off-by: Chunyan Liu cy...@suse.com
---
src/libxl/libxl_conf.c | 44
src/libxl/libxl_driver.c | 21 -
2 files changed, 64 insertions(+),
DRIVER_SOURCE_FILES mixes files with absolute path (inherited from
REMOTE_DRIVER_GENERATED) with file paths that are relative to srcdir but
check-driverimpls.pl needs full paths.
---
src/Makefile.am | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/Makefile.am
On 05/16/2013 02:23 AM, Eric Blake wrote:
Several files called out COPYING or COPYING.LIB instead of using
the normal boilerplate. It's especially important that we don't
call out COPYING from an LGPL file, since COPYING is traditionally
used for the GPL. A few files were lacking copyright
On 05/16/2013 02:23 AM, Eric Blake wrote:
https://www.gnu.org/licenses/gpl-howto.html states:
You should also include a copy of the license itself somewhere in the
distribution of your program. All programs, whether they are released
under the GPL or LGPL, should include the text version of
On 05/16/2013 02:23 AM, Eric Blake wrote:
Now that COPYING no longer contains the text of the LGPL,
modify the LGPLv2-only files from vbox to call out the
correct file.
* src/vbox/vbox_CAPI_v2_2.h: Refer to correct file.
* src/vbox/vbox_CAPI_v3_0.h: Likewise.
* src/vbox/vbox_CAPI_v3_1.h:
On 05/16/2013 02:23 AM, Eric Blake wrote:
To ensure we don't regress and cause the need for further
cleanups, add a 'make syntax-check' rule that ensures new
files have proper copyright contents.
* cfg.mk (sc_copyright_file): New rule.
Signed-off-by: Eric Blake ebl...@redhat.com
---
On 05/15/2013 12:33 PM, Daniel P. Berrange wrote:
From: Daniel P. Berrange berra...@redhat.com
Change the build process driver initialization so that the
VirtualBox driver is built into libvirtd, instead of libvirt.so
This change avoids the VirtualBox GPLv2-only license causing
On 15.05.2013 23:23, Matthias Bolte wrote:
2013/5/3 Michal Privoznik mpriv...@redhat.com:
---
src/vmware/vmware_conf.c | 28 ++--
src/vmware/vmware_driver.c | 11 +--
2 files changed, 15 insertions(+), 24 deletions(-)
diff --git a/src/vmware/vmware_conf.c
On Wed, May 15, 2013 at 01:09:44PM -0400, Laine Stump wrote:
On 05/15/2013 07:23 AM, Osier Yang wrote:
On 15/05/13 19:09, Daniel P. Berrange wrote:
Instead of
adding many parameters to the API, we could provide a struct
Agreed, many params is always not good. .
containing all the
On Wed, May 15, 2013 at 02:36:32PM -0400, dwa...@redhat.com wrote:
From: Dan Walsh dwa...@redhat.com
mcstransd is a translation tool that can translate MCS Labels into human
understandable code. I have patched it to watch for translation files in the
/run/setrans directory. This allows us
On 16/05/13 14:07, Chunyan Liu wrote:
This patch sets include two patches:
1/2 is the implementation of the hostdev passthrough common library.
To meet two purposes:
a. move qemu hostdev APIs to common library so that it could be used by all
hypervisor drivers.
b. maintain a global hostdev
On 16/05/13 14:07, Chunyan Liu wrote:
Write separate module for hostdev passthrough so that it could be used by all
hypervisor drivers and maintain a global hostdev state.
Signed-off-by: Chunyan Liucy...@suse.com
---
po/POTFILES.in |1 +
src/Makefile.am |1
On Thu, May 16, 2013 at 12:33:56AM +0800, Osier Yang wrote:
On 15/05/13 23:32, Daniel P. Berrange wrote:
On Wed, May 15, 2013 at 11:25:19PM +0800, Osier Yang wrote:
Since 0d70656afded, it starts to access the sysfs files to build
the qemu command line (by virSCSIDeviceGetSgName, which is to
On Thu, May 16, 2013 at 12:37:23AM +0800, Osier Yang wrote:
On 15/05/13 23:36, Daniel P. Berrange wrote:
On Wed, May 15, 2013 at 11:25:20PM +0800, Osier Yang wrote:
fchostdata is also about sysfs files, it can go home now.
---
Not sure I really agree with this change. The 'fchostdata' files
stdlib.h header file needed for getenv was only transitively included
through libdevmapper.h.
---
Pushed as a trivial build-breaker.
src/util/virutil.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/util/virutil.c b/src/util/virutil.c
index ab3c82f..7f4ecd6 100644
---
Since QEMU 1.1 qcow2 supports a features bitfield, allowing
to implement additional features without breaking compatibility
with older QEMU versions.
The version header on-disk is bumped to 3, however the format name
is still 'qcow2', not 'qcow3' (which is what I used in v2).
v2:
Split out option string generation to make adding new options easier
and simplify the code.
---
src/storage/storage_backend.c | 111 ++
1 file changed, 59 insertions(+), 52 deletions(-)
diff --git a/src/storage/storage_backend.c
---
docs/schemas/domainsnapshot.rng | 7 +++
src/conf/snapshot_conf.c | 74
src/conf/snapshot_conf.h | 2 +
tests/domainsnapshotxml2xmlin/disk_snapshot.xml | 5 ++
Add features and compat elements to volume target XML.
compat is a string which for qcow2 represents the QEMU version
it should be compatible with. Valid values are 0.10 and 1.1.
1.1 is implicit if the features element is present, otherwise
qemu-img default is used. 0.10 can be specified to
When cmd is NULL, virCommandToString reports an OOM error,
overwriting the error reported by virStorageBackendCreateQemuImgCmd.
---
tests/storagevolxml2argvtest.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tests/storagevolxml2argvtest.c
For live external snapshots, the image creation is done by QEMU
and we can't pass the needed options.
---
docs/formatsnapshot.html.in | 5
src/qemu/qemu_driver.c | 72 +++--
2 files changed, 62 insertions(+), 15 deletions(-)
diff --git
Just to reduce the indentation levels. Remove the unneeded
NULL check for disk-file, as virBufferEscapeString doesn't
print anything with NULL arguments.
---
src/conf/snapshot_conf.c | 53 ++--
1 file changed, 29 insertions(+), 24 deletions(-)
diff
Detect qcow2 images with version 3 in the image header as
VIR_STORAGE_FILE_QCOW2.
These images have a feature bitfield, with just one feature supported
so far: lazy_refcounts.
The header length changed too, moving the location of the backing
format name.
---
src/util/virstoragefile.c | 164
Add -o compat= and -o lazy_refcounts options for qemu-img.
---
src/storage/storage_backend.c | 36 +++--
tests/storagevolxml2argvdata/qcow2-1.1.argv | 1 +
tests/storagevolxml2argvdata/qcow2-lazy.argv| 1 +
virStorageDefParsePerms:
* Use uid_t/gid_t to do casting
virStoragePoolDefParseSource:
* Improve the error message for source name parsing
* Remove the useless casting (const char *)
virStoragePoolDefParseAuthChap:
* Fix the wrong error message
virStoragePoolDefParseAuthCephx:
* Don't
*** BLURB HERE ***
Osier Yang (5):
storage_conf: Fix the coding stype in storage_conf.c
storage_conf: Fix indentions in storage_conf.c
storage_conf: Various fixes or improvements on pool def parsing
storage_conf: Left fixes or improvements for storage_conf.c
storage_conf: Improve the
Changes:
* Remove the useless space in for statement (e.g.
for (i = 0 ; i something ; i++)
* Change the function's style to:
void
foo(bar)
{
printf(foo is not bar\n);
}
* Don't lose {} for if...else branches if one of the branch
has more than one line
virStorageVolDefParseXML:
* Create virStorageVolDefPtr def, and use ret to track the
return value; frees the strings at cleanup label instead
of freeing them in the middle.
virStorageVolDefFormat:
* Use macro NULLSTR
---
src/conf/storage_conf.c | 93
Commit 7f15ebc7a2b599ab10dbc15bca6f823591213e67 introduced a bug
happening when guests without a graphics element are migrated.
The initialization of listenAddress happens unconditionally
from the cookie even if the cookie-graphics pointer was NULL.
Moved the initialization to where it is safe.
Uses the 4 spaces for indention.
---
src/conf/storage_conf.c | 176
1 file changed, 88 insertions(+), 88 deletions(-)
diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
index 643c3cc..8fa805b 100644
--- a/src/conf/storage_conf.c
+++
---
docs/formatdomain.html.in | 26 +-
docs/schemas/domaincommon.rng | 5 +
src/conf/domain_conf.c | 20 +
src/conf/domain_conf.h | 1 +
Changes:
* Remove useless comments
* Remove useless blank lines
* If the struct member is a enum type, comment it like
/* enum fooBar */
* Break the long lines
* Prefer the common function style for the inline function
---
src/conf/storage_conf.h | 120
Currently, only one tapfd and one vhostfd could be passed.
However, multiqueue network requires several FDs to be passed to
qemu so we must adapt out monitor handling functions to cope with
that.
---
src/qemu/qemu_hotplug.c | 7 +--
src/qemu/qemu_monitor.c | 39
For future work it's better, if tapfd is passed as pointer.
Moreover, we need to be able to return multiple values now.
---
src/qemu/qemu_command.c | 89 ++---
src/qemu/qemu_command.h | 4 ++-
src/qemu/qemu_hotplug.c | 4 +--
3 files changed, 53
Currently, the function knows how to open a TAP device for a single
time. However, in case of multiqueue network we need to open it multiple
times. Moreover, when doing TUNSETIFF ioctl, the IFF_MULTI_QUEUE flag
shall be requested. This commit changes a behaviour slightly as well.
Till now it was
Fixed version with Laine's comments worked in [1].
Patches 2, 3 and 5 has been ACKed already. BTW, patch 4 is really a different
to patch 3. The only thing they share is a commit message body.
Kernel and subsequently QEMU learned multiple transmit queues a while ago. The
feature has a nice
---
src/qemu/qemu_command.c | 26 ++
src/qemu/qemu_hotplug.c | 27 ---
2 files changed, 34 insertions(+), 19 deletions(-)
diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
index 5d64705..7661b13 100644
--- a/src/qemu/qemu_command.c
+++
Currently, we have one huge function to construct qemu command line.
This is very ineffective esp. if there's a fault somewhere.
---
src/qemu/qemu_command.c | 224 +---
1 file changed, 117 insertions(+), 107 deletions(-)
diff --git
---
src/qemu/qemu_hotplug.c | 96 +++--
1 file changed, 61 insertions(+), 35 deletions(-)
diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
index 953339b..695edc7 100644
--- a/src/qemu/qemu_hotplug.c
+++ b/src/qemu/qemu_hotplug.c
@@
The qemuBuildHostNetStr() function which is responsible for
generating command line for a network interface needs to be aware
of multiqueue network interface as we are required to use:
- fd=%d in case of one FD
- fds=%d:%d:%d:...:%d in case of multiple FDs
These two approaches can't be
On 16/05/13 20:40, Osier Yang wrote:
Changes:
* Remove useless comments
* Remove useless blank lines
* If the struct member is a enum type, comment it like
/* enum fooBar */
* Break the long lines
* Prefer the common function style for the inline function
---
Currently, only one tapfd and one vhostfd could be passed.
However, multiqueue network requires several FDs to be passed to
qemu so we must adapt out monitor handling functions to cope with
that.
---
src/qemu/qemu_hotplug.c | 7 +--
src/qemu/qemu_monitor.c | 39
---
src/qemu/qemu_command.c | 82 ++---
1 file changed, 58 insertions(+), 24 deletions(-)
diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
index f2c6d47..5d64705 100644
--- a/src/qemu/qemu_command.c
+++ b/src/qemu/qemu_command.c
@@
This attribute is going to represent number of queues for
multique vhost network interface. This commit implements XML
extension part of the feature and add one test as well. For now,
we can only do xml2xml test as qemu command line generation code
is not adapted yet.
---
With multiqueue network feature, we are advised to pass multiple
vhost-net FDs as well. The ratio should be 1:1. Therefore we must
alter the qemuOpenVhostNet function to allow that.
---
src/qemu/qemu_command.c | 60 -
src/qemu/qemu_command.h | 3
Commit 632f78c introduced a regression which causes schedinfo being
unable to set some parameters. When migrating to priv-cgroup there
was missing variable left out and due to passed NULL to underlying
function, the setting failed.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=963592
---
QEMU_CAPS_VNC_SHARE_POLICY (qemu = 1.1)
---
src/qemu/qemu_capabilities.c | 6 +-
src/qemu/qemu_capabilities.h | 1 +
tests/qemuhelptest.c | 9 ++---
3 files changed, 12 insertions(+), 4 deletions(-)
diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c
index
v2-v3
rebase work.
v1-v2:
changed attribute name from 'policy' to 'sharePolicy'
renamed caps flag name: QEMU_CAPS_VNC_SHARE_POLICY
fixed issues pointed out in v1 review
have to keep hard-coded version probe after checking qemu -help and qmp
command.
As there were some vnc patches
example: qemu ${otherargs} \
-vnc 127.0.0.1:0,share=allow-exclusive
---
src/qemu/qemu_command.c | 36
tests/qemuargv2xmltest.c | 1 +
tests/qemuxml2argvtest.c | 1 +
3 files changed, 38 insertions(+)
diff --git a/src/qemu/qemu_command.c
-vnc :5900,share=allow-exclusive
allows clients to ask for exclusive access which is
implemented by dropping other connections Connecting
multiple clients in parallel requires all clients asking
for a shared session (vncviewer: -shared switch)
-vnc :5900,share=force-shared
disables exclusive
On 05/16/2013 12:20 AM, Jiri Denemark wrote:
DRIVER_SOURCE_FILES mixes files with absolute path (inherited from
REMOTE_DRIVER_GENERATED) with file paths that are relative to srcdir but
check-driverimpls.pl needs full paths.
---
src/Makefile.am | 3 ++-
1 file changed, 2 insertions(+), 1
On 16/05/13 21:02, Martin Kletzander wrote:
Commit 632f78c introduced a regression which causes schedinfo being
unable to set some parameters. When migrating to priv-cgroup there
was missing variable left out and due to passed NULL to underlying
function, the setting failed.
Resolves:
On Thu, May 16, 2013 at 06:58:56 -0600, Eric Blake wrote:
On 05/16/2013 12:20 AM, Jiri Denemark wrote:
DRIVER_SOURCE_FILES mixes files with absolute path (inherited from
REMOTE_DRIVER_GENERATED) with file paths that are relative to srcdir but
check-driverimpls.pl needs full paths.
---
On Thu, May 16, 2013 at 14:41:29 +0200, Viktor Mihajlovski wrote:
Commit 7f15ebc7a2b599ab10dbc15bca6f823591213e67 introduced a bug
happening when guests without a graphics element are migrated.
The initialization of listenAddress happens unconditionally
from the cookie even if the
In virLXCProcessReboot, if the 'conn' value is NULL, the eventual
call to virLXCNetworkLookupByName() via virLXCProcessStart() and
virLXCProcessSetupInterfaces() will cause a core since it references
conn-networkDriver directly.
---
src/lxc/lxc_process.c | 4 ++--
1 file changed, 2 insertions(+),
The shunloadStart function didn't check the status of virInitialize which
was flagged by Coverity. Adjust the function and shunloadtest in order
to handle the situation.
---
tests/shunloadhelper.c | 12
tests/shunloadtest.c | 26 +-
2 files changed, 25
The return from virInitialize() needs to be checked.
---
tests/xencapstest.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tests/xencapstest.c b/tests/xencapstest.c
index e220234..8bdd6cd 100644
--- a/tests/xencapstest.c
+++ b/tests/xencapstest.c
@@ -161,7 +161,8 @@
Upgraded to Coverity 6.5.3 which uncovered a few code paths with issues.
John Ferlan (3):
lxc_process: Resolve Coverity NULL_RETURNS error
xencapstest: Resolve Coverity CHECKED_RETURN error
shunloadtest: Resolve Coverity CHECKED_RETURN error
src/lxc/lxc_process.c | 4 ++--
CVE-2013-1962
remoteDispatchStoragePoolListAllVolumes wasn't freeing the pool.
The pool also held a reference to the connection, preventing it from
getting freed and closing the netcf interface driver, which held two
sockets open.
---
daemon/remote.c | 2 ++
1 file changed, 2 insertions(+)
diff
On 05/16/2013 08:03 AM, Ján Tomko wrote:
CVE-2013-1962
remoteDispatchStoragePoolListAllVolumes wasn't freeing the pool.
The pool also held a reference to the connection, preventing it from
getting freed and closing the netcf interface driver, which held two
sockets open.
---
On 05/16/2013 04:06 PM, Eric Blake wrote:
On 05/16/2013 08:03 AM, Ján Tomko wrote:
CVE-2013-1962
remoteDispatchStoragePoolListAllVolumes wasn't freeing the pool.
The pool also held a reference to the connection, preventing it from
getting freed and closing the netcf interface driver, which
On Thu, May 16, 2013 at 10:01:51AM -0400, John Ferlan wrote:
In virLXCProcessReboot, if the 'conn' value is NULL, the eventual
call to virLXCNetworkLookupByName() via virLXCProcessStart() and
virLXCProcessSetupInterfaces() will cause a core since it references
conn-networkDriver directly.
---
On 05/16/2013 08:49 AM, Michal Privoznik wrote:
Fixed version with Laine's comments worked in [1].
Patches 2, 3 and 5 has been ACKed already. BTW, patch 4 is really a different
to patch 3. The only thing they share is a commit message body.
Ah, I see now. I started reading through patch 4
On 05/16/2013 08:01 AM, John Ferlan wrote:
In virLXCProcessReboot, if the 'conn' value is NULL, the eventual
call to virLXCNetworkLookupByName() via virLXCProcessStart() and
virLXCProcessSetupInterfaces() will cause a core since it references
conn-networkDriver directly.
Did you investigate
On 05/16/2013 08:01 AM, John Ferlan wrote:
The shunloadStart function didn't check the status of virInitialize which
was flagged by Coverity. Adjust the function and shunloadtest in order
to handle the situation.
---
tests/shunloadhelper.c | 12
tests/shunloadtest.c | 26
On 05/16/2013 08:01 AM, John Ferlan wrote:
The return from virInitialize() needs to be checked.
---
tests/xencapstest.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
ACK.
diff --git a/tests/xencapstest.c b/tests/xencapstest.c
index e220234..8bdd6cd 100644
---
---
Pushed under the trivial rule.
src/datatypes.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/datatypes.c b/src/datatypes.c
index 884eb3e..940d968 100644
--- a/src/datatypes.c
+++ b/src/datatypes.c
@@ -421,7 +421,7 @@ virInterfaceDispose(void *obj)
* and register
Make target path mandatory for pool types that require target,
since we refuse to parse a target without a path.
https://bugzilla.redhat.com/show_bug.cgi?id=893273
---
docs/schemas/storagepool.rng | 8 +++-
tests/Makefile.am
We don't parse it anyway.
https://bugzilla.redhat.com/show_bug.cgi?id=893273
---
docs/schemas/storagevol.rng| 4 +++-
tests/Makefile.am | 3 ++-
tests/storagevolschemadata/qcow2-no-source.xml | 29 ++
Fix https://bugzilla.redhat.com/show_bug.cgi?id=893273
Ján Tomko (2):
schema: require target path in storage pool xml
schema: make source optional in volume XML
docs/schemas/storagepool.rng | 8 +++---
docs/schemas/storagevol.rng| 4 ++-
On 05/14/2013 02:56 PM, Osier Yang wrote:
On 14/05/13 17:29, Ján Tomko wrote:
On 05/14/2013 11:03 AM, Osier Yang wrote:
The reason for it's not exposed for such long time is that there is
no xml2xml test for it. This fixes the bug and adds test.
It was not exposed, because
On 05/16/2013 10:19 AM, Eric Blake wrote:
On 05/16/2013 08:01 AM, John Ferlan wrote:
In virLXCProcessReboot, if the 'conn' value is NULL, the eventual
call to virLXCNetworkLookupByName() via virLXCProcessStart() and
virLXCProcessSetupInterfaces() will cause a core since it references
Since 0d70656afded, it starts to access the sysfs files to build
the qemu command line (by virSCSIDeviceGetSgName, which is to find
out the scsi generic device name by adpater:bus:target:unit), there
is no way to work around, qemu wants to see the scsi generic device
like /dev/sg6 anyway.
And
On 16/05/13 22:44, Ján Tomko wrote:
On 05/14/2013 02:56 PM, Osier Yang wrote:
On 14/05/13 17:29, Ján Tomko wrote:
On 05/14/2013 11:03 AM, Osier Yang wrote:
The reason for it's not exposed for such long time is that there is
no xml2xml test for it. This fixes the bug and adds test.
It was not
On 07/05/13 21:45, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
Unlike disk device, the scsi-generic always writethrough the data,
so no need to introduce a cache tag, and set cache=off.
---
docs/formatdomain.html.in | 6
Running make check in a VPATH configured build directory fails
in fchosttest as the test data files are searched for relative to
the current working directory.
Signed-off-by: Viktor Mihajlovski mihaj...@linux.vnet.ibm.com
---
This may become obsolete with Osier's pending patches, but
it helps
On 07/05/13 22:51, Osier Yang wrote:
On 07/05/13 22:00, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
E.g.
% sg_map
/dev/sg0 /dev/sda
/dev/sg1 /dev/sr0
What the helper gets for /dev/sg0 is /dev/sda, it will be used by
later patch.
---
src/libvirt_private.syms | 1 +
On 07/05/13 21:52, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
Shared disk is not only the thing we should care about after scsi
hostdev is introduced. A same scsi device can be used as disk for
one domain, and as scsi hostdev for another domain at the same time.
That's why
libxl expects the event handler to free the event passed to it. From
libxl_event.h:
event becomes owned by the application and must be freed, either
by event_occurs or later
---
src/libxl/libxl_driver.c |8 +++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git
Xen 4.3 fixes a mistake in the libxl event handler signature where the
event owned by the application was defined as const. Detect this and
define the libvirt libxl event handler signature appropriately.
---
src/libxl/libxl_driver.c |7 ++-
1 files changed, 6 insertions(+), 1
On Wed, May 15, 2013 at 10:35:48AM -0400, dwa...@redhat.com wrote:
From: Dan Walsh dwa...@redhat.com
We do not want to allow contained applications to be able to read fusefs_t.
So we want /proc/meminfo label to match the system default proc_t.
Fix checking of error codes
---
On Thu, May 16, 2013 at 05:04:06PM +0100, Daniel P. Berrange wrote:
On Wed, May 15, 2013 at 10:35:48AM -0400, dwa...@redhat.com wrote:
From: Dan Walsh dwa...@redhat.com
We do not want to allow contained applications to be able to read fusefs_t.
So we want /proc/meminfo label to match the
Linux netfilter at some point inverted the meaning of the '--ctdir reply'
and newer netfilter implementations now expect '--ctdir original'
instead and vice-versa.
We probe for this netfilter change via an IMCP message over loopback and 3
filtering rules applied to INPUT. If the sent byte
Hello Daniel,
I've been working on improving scalability in OpenStack on libvirt+kvm
for the last couple of months. I'm particularly interested in reducing
the time it takes to create VMs when many VMs are requested in
parallel.
One apparent bottleneck during virtual machine creation is libvirt.
On Thu, May 16, 2013 at 12:09:39PM -0400, Peter Feiner wrote:
Hello Daniel,
I've been working on improving scalability in OpenStack on libvirt+kvm
for the last couple of months. I'm particularly interested in reducing
the time it takes to create VMs when many VMs are requested in
parallel.
On 08/05/13 03:01, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
This adds the shared device entry when starting domain (more
exactly, when preparing host devices), and remove the entry
when destroying domain (when reattaching host devices).
---
src/qemu/qemu_hostdev.c | 25
On 08/05/13 02:57, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
This changes the helpers qemu{Add,Remove}SharedDisk into
qemu{Add,Remove}SharedDevice, as most of the code in the helpers
can be reused for scsi host device.
To track the shared scsi host device, first it finds out
On 08/05/13 03:17, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
sgio is only valid for scsi host device.
---
docs/formatdomain.html.in | 7 -
docs/schemas/domaincommon.rng | 8 +
src/conf/domain_conf.c
On 08/05/13 07:42, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
unpriv_sgio setting is tight with the shared device helpers, let's
put them together in qemu_conf.c
---
src/qemu/qemu_conf.c| 36
src/qemu/qemu_conf.h| 2 ++
On 08/05/13 07:44, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
Just like what previous patches do, it refactors qemuSetUnprivSGIO
to take the virDomainDeviceDefPtr as argument instead.
---
src/qemu/qemu_conf.c| 67 -
On 08/05/13 07:54, John Ferlan wrote:
On 05/03/2013 02:07 PM, Osier Yang wrote:
---
src/qemu/qemu_hostdev.c | 3 +++
1 file changed, 3 insertions(+)
ACK
Pushed.
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
How many CPU cores are you testing on ? That's a good improvement,
but I'd expect the improvement to be greater as # of core is larger.
I'm testing on 12 Cores x 2 HT per code. As I'm working on teasing out
software bottlenecks, I'm intentionally running fewer tasks (20 parallel
creations)
Xen upstream commit 5fb11a07 fixed the signature of the event handler,
causing build failures of the libxl driver on xen-unstable / Xen 4.3.
While at it, plug leaking of libxl_event objects in the event handler.
Jim Fehlig (2):
libxl: fix build with Xen4.3
libxl: fix leaking libxl events
On Thu, May 16, 2013 at 01:00:15PM -0400, Peter Feiner wrote:
How many CPU cores are you testing on ? That's a good improvement,
but I'd expect the improvement to be greater as # of core is larger.
I'm testing on 12 Cores x 2 HT per code. As I'm working on teasing out
software
1 - 100 of 131 matches
Mail list logo