于 2011年04月29日 23:11, Eric Blake 写道:
On 04/29/2011 01:21 AM, Osier Yang wrote:
Otherwise if there are inactive lxc domains, lxcStartup will
try to reconnect to sockets of these domains, which results in
errors in libvirtd log.
---
src/lxc/lxc_driver.c |3 +++
1 files changed, 3
On Mon, 2011-05-02 at 12:07 -0400, Laine Stump wrote:
Just so I understand correctly - the 802.1Qbg standard requires using
a
VLAN, but the vepa implementation doesn't have that restriction, so
it
is still okay for someone using a vepa switch to not use a VLAN.
Correct?
yes
--
Best
On Fri, Apr 22, 2011 at 02:24:54PM +0200, Michal Privoznik wrote:
The patch which moved libpciaccess initialization to one place caused
regression - we were not able to run on system with no PCI bus, like
s390(x).
---
src/node_device/node_device_udev.c |9 +
1 files changed, 9
This adds a streaming-video=filter|all|off attribute. It is used to change
the behavior of video stream detection in spice, the default is filter (the
default for libvirt is not to specify it - the actual default is defined in
libspice-server.so).
Usage:
graphics type='spice' autoport='yes'
On Fri, Apr 29, 2011 at 03:21:03PM +0800, Osier Yang wrote:
Otherwise if there are inactive lxc domains, lxcStartup will
try to reconnect to sockets of these domains, which results in
errors in libvirtd log.
---
src/lxc/lxc_driver.c |3 +++
1 files changed, 3 insertions(+), 0
On 05/03/2011 03:05 PM, Alon Levy wrote:
This adds a streaming-video=filter|all|off attribute. It is used to change
the behavior of video stream detection in spice, the default is filter (the
default for libvirt is not to specify it - the actual default is defined in
libspice-server.so).
Usage:
On Mon, May 02, 2011 at 08:09:07AM -0700, Badari Pulavarty wrote:
Hi,
How can I make use of multifunction functionality to support
more than 32 virtio devices in QEMU through libvirt ?
It isn't supported because it doesn't work with hotplug.
Are there corresponding changes in libvirt ?
To
When setting up a FIFO for QEMU, it allows either a pair
of fifos used unidirectionally, or a single fifo used
bidirectionally. Look for the bidirectional fifo first
when labelling since that is more useful
* src/security/security_dac.c,
src/security/security_selinux.c: Fix fifo handling
---
On Mon, May 02, 2011 at 04:29:49PM -0500, Adam Litke wrote:
typedef enum {
/* If set, virDomainBlockAllocate() will return immediately
* allowing polling for operation completion status
*/
VIR_DOMAIN_DISK_STREAM_NONBLOCK,
}
On 05/02/2011 09:36 PM, Daniel Veillard wrote:
On Mon, May 02, 2011 at 04:56:53PM -0600, Eric Blake wrote:
* tests/.valgrind.supp: Consolidate bash suppressions. Ignore
more libnl issues.
---
Matthias tested on a debian-based system; I found more issues when
testing on Fedora 14. With
On 05/03/2011 07:09 AM, Daniel P. Berrange wrote:
On Fri, Apr 29, 2011 at 03:21:03PM +0800, Osier Yang wrote:
Otherwise if there are inactive lxc domains, lxcStartup will
try to reconnect to sockets of these domains, which results in
errors in libvirtd log.
---
src/lxc/lxc_driver.c |3
On 05/03/2011 08:41 AM, Daniel P. Berrange wrote:
When setting up a FIFO for QEMU, it allows either a pair
of fifos used unidirectionally, or a single fifo used
bidirectionally. Look for the bidirectional fifo first
when labelling since that is more useful
* src/security/security_dac.c,
Clang detected a null-pointer dereference regression, introduced
in commit 4e8969eb. Without this patch, a device with
unbind_from_stub set to false would eventually try to call
virFileExists on uncomputed drvdir.
* src/util/pci.c (pciUnbindDeviceFromStub): Ensure drvdir is set
before use.
---
On Tue, May 03, 2011 at 09:19:59AM -0600, Eric Blake wrote:
On 05/03/2011 07:09 AM, Daniel P. Berrange wrote:
On Fri, Apr 29, 2011 at 03:21:03PM +0800, Osier Yang wrote:
Otherwise if there are inactive lxc domains, lxcStartup will
try to reconnect to sockets of these domains, which results
This code has had problems historically. As originally
written, in commit 6bcf2501 (Jun 08), it could call unlink
on a random string, nuking an unrelated file.
Then commit 182a80b9 (Sep 09), the code was rewritten to
allocate tmp, with both a use-after-free bug and a chance to
call unlink(NULL).
On 05/03/2011 09:37 AM, Daniel P. Berrange wrote:
Osier, can you get this done before 0.9.1, or are we better off
reverting commit 0e7f7f85 for the release and resolving this post-release?
I vote for reverting it, since the original problem was merely a
harmless message in the logs which
Unlikely to hit in real life, but clang noticed it.
* tests/commandtest.c (test4): Avoid unlink(NULL) on OOM.
---
Much like a similar patch I sent for qemu_driver, but this time
the unlink(NULL) is much harder to trigger.
tests/commandtest.c |3 ++-
1 files changed, 2 insertions(+), 1
On Tue, May 03, 2011 at 10:38:00AM -0600, Eric Blake wrote:
Unlikely to hit in real life, but clang noticed it.
* tests/commandtest.c (test4): Avoid unlink(NULL) on OOM.
---
Much like a similar patch I sent for qemu_driver, but this time
the unlink(NULL) is much harder to trigger.
Clang detected that vol-download will call unlink(NULL) if there
is a parse error during option parsing. Also, mingw doesn't like
unlinking an open file.
* tools/virsh.c (cmdVolDownload): Only unlink file if created.
---
tools/virsh.c |9 +
1 files changed, 5 insertions(+), 4
On Tue, May 03, 2011 at 09:46:15AM -0600, Eric Blake wrote:
This code has had problems historically. As originally
written, in commit 6bcf2501 (Jun 08), it could call unlink
on a random string, nuking an unrelated file.
Then commit 182a80b9 (Sep 09), the code was rewritten to
allocate tmp,
On Tue, May 03, 2011 at 09:34:11AM -0600, Eric Blake wrote:
Clang detected a null-pointer dereference regression, introduced
in commit 4e8969eb. Without this patch, a device with
unbind_from_stub set to false would eventually try to call
virFileExists on uncomputed drvdir.
* src/util/pci.c
On Tue, May 03, 2011 at 10:45:36AM -0600, Eric Blake wrote:
Clang detected that vol-download will call unlink(NULL) if there
is a parse error during option parsing. Also, mingw doesn't like
unlinking an open file.
* tools/virsh.c (cmdVolDownload): Only unlink file if created.
---
On 05/03/2011 10:42 AM, Daniel P. Berrange wrote:
On Tue, May 03, 2011 at 10:38:00AM -0600, Eric Blake wrote:
Unlikely to hit in real life, but clang noticed it.
* tests/commandtest.c (test4): Avoid unlink(NULL) on OOM.
---
cleanup:
virCommandFree(cmd);
-unlink(pidfile);
+
On 05/03/2011 10:49 AM, Daniel P. Berrange wrote:
On Tue, May 03, 2011 at 10:45:36AM -0600, Eric Blake wrote:
Clang detected that vol-download will call unlink(NULL) if there
is a parse error during option parsing. Also, mingw doesn't like
unlinking an open file.
* tools/virsh.c
On 05/03/2011 10:48 AM, Daniel P. Berrange wrote:
On Tue, May 03, 2011 at 09:34:11AM -0600, Eric Blake wrote:
Clang detected a null-pointer dereference regression, introduced
in commit 4e8969eb. Without this patch, a device with
unbind_from_stub set to false would eventually try to call
On 05/03/2011 10:47 AM, Daniel P. Berrange wrote:
On Tue, May 03, 2011 at 09:46:15AM -0600, Eric Blake wrote:
This code has had problems historically. As originally
written, in commit 6bcf2501 (Jun 08), it could call unlink
on a random string, nuking an unrelated file.
Then commit 182a80b9
Version 2.0.0 or yajl changed API. It is fairly trivial for us to
cope with both APIs in libvirt, so adapt.
* configure.ac: Probe for yajl2 API
* src/util/json.c: Conditional support for yajl2 API
---
configure.ac|8
src/util/json.c | 44
On 05/03/2011 11:12 AM, Daniel P. Berrange wrote:
Version 2.0.0 or yajl changed API. It is fairly trivial for us to
cope with both APIs in libvirt, so adapt.
* configure.ac: Probe for yajl2 API
* src/util/json.c: Conditional support for yajl2 API
+++ b/src/util/json.c
@@ -709,7 +709,12 @@
Clang found three instances of uninitialized use of nparams in
the cleanup path. Unfortunately, one is a false positive: clang
couldn't see that ret-params.params_val is guaranteed to be
NULL unless allocated within a function, and that nparams is
guaranteed to be assigned prior to the
On 05/03/2011 11:28 AM, Eric Blake wrote:
Clang found three instances of uninitialized use of nparams in
the cleanup path. Unfortunately, one is a false positive: clang
couldn't see that ret-params.params_val is guaranteed to be
NULL unless allocated within a function, and that nparams is
Detected by clang. NULL deref added in commit 343a27a (Mar 11),
but leak of voldef present since commit 2cd9b2d (Apr 09).
* src/storage/storage_driver.c (storageVolumeCreateXML): Don't
leak voldef or dereference null volobj.
---
src/storage/storage_driver.c |4 +++-
1 files changed, 3
Detected by clang.
* src/esx/esx_driver.c (esxDomainGetInfo): Fail early on error.
---
src/esx/esx_driver.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/src/esx/esx_driver.c b/src/esx/esx_driver.c
index 1f8f90b..e929208 100644
--- a/src/esx/esx_driver.c
+++
Clang 2.8 wasn't quite able to follow that persistentDef was
assigned earlier if (flags VIR_DOMAIN_MEM_CONFIG) is true.
Silence this false positive, to make clang analysis easier to use.
* src/qemu/qemu_driver.c (qemudDomainSetMemoryFlags): Add an
annotation to silence clang's claim of a NULL
Clang warned about a dead assignment. In the process, I noticed
that we are only using the function for a bool value. I audited
all other callers in qemu_{migration,cgroup,driver,hotplug), and
all were making the call in a bool context.
* src/qemu/qemu_cgroup.c (qemuSetupCgroup): Delete dead
Clang complained about this, and it was easy enough to fix.
* src/util/util.c (virFileOpenAs): Drop dead assignment.
---
src/util/util.c |3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/src/util/util.c b/src/util/util.c
index 37472bb..9041ab6 100644
---
On 05/03/2011 04:16 PM, Scott Dowdle wrote:
Greetings,
I wasn't sure who to report this to... and I did a bugzilla search and didn't
find anything... and I'm not sure where to report this to... since it isn't
Fedora 14 exactly since I'm using the virt-preview repo.
I *think* the best place
Clang noticed a dead assignment, which turned out to be the use
of the wrong variable. rc starts life as -1, and is only ever
assigned to 0 just before a successful cleanup.
* src/lxc/lxc_driver.c (lxcSetupInterfaces): Don't call
virReportSystemError(-1).
---
src/lxc/lxc_driver.c |3 +--
1
Detected by clang.
* src/qemu/qemu_migration.c (qemuMigrationToFile): Nothing later
uses is_reg.
---
src/qemu/qemu_migration.c |1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/src/qemu/qemu_migration.c b/src/qemu/qemu_migration.c
index 7f4b111..6c5bf66 100644
---
Greetings,
- Original Message from Laine Stump -
I'm not familiar with that part of the code, but it looks like the cert
check happens if listen_tls = 1, which is its default value. You can
disable it by uncommenting listen_tls=0 in
/etc/libvirt/libvirtd.conf.
I'm not sure what
Clang detected a dead store to rc. It turns out that in fixing this,
I also found a FILE* leak.
* src/util/cgroup.c (virCgroupKillInternal): Abort rather than
resuming loop on fscanf failure, and cleanup file on error.
---
As a bonus, this also fixes a decl-after-statement for fp.
I've encountered an interesting scenario:
1. define a guest via virsh define xml
2 start this guest via virsh
3. one of the disk elements is a multipath device that is currently
misconfigured such that any io to the device hangs the calling process
4. libvirt times out when attemping to
clang didn't like the last increment to nargs. But why even
track nargs ourselves, when virCommand does it for us?
* src/storage/storage_backend_iscsi.c
(virStorageBackendISCSIConnection): Switch to virCommand to avoid
a dead-store warning on nargs.
---
src/storage/storage_backend_iscsi.c |
Detected by clang.
* src/esx/esx_util.c (esxUtil_ParseDatastorePath): No need to
increment.
---
src/esx/esx_util.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/esx/esx_util.c b/src/esx/esx_util.c
index 9ef947c..640e984 100644
--- a/src/esx/esx_util.c
+++
On 05/03/2011 09:33 AM, Daniel P. Berrange wrote:
On Mon, May 02, 2011 at 08:09:07AM -0700, Badari Pulavarty wrote:
Hi,
How can I make use of multifunction functionality to support
more than 32 virtio devices in QEMU through libvirt ?
It isn't supported because it doesn't work with hotplug.
I'm getting this when trying to build libvirt on rawhide (using the
package gnutls-devel-2.12.3-1.fc16.x86_64):
remote/remote_driver.c: In function 'negotiate_gnutls_on_connection':
remote/remote_driver.c:1361:9: error:
'gnutls_certificate_type_set_priority' is deprecated (declared at
* Ryan Harper ry...@us.ibm.com [2011-05-03 16:57]:
I've encountered an interesting scenario:
1. define a guest via virsh define xml
2 start this guest via virsh
3. one of the disk elements is a multipath device that is currently
misconfigured such that any io to the device hangs the
于 2011年05月04日 00:11, Eric Blake 写道:
On 05/03/2011 09:37 AM, Daniel P. Berrange wrote:
Osier, can you get this done before 0.9.1, or are we better off
reverting commit 0e7f7f85 for the release and resolving this post-release?
I vote for reverting it, since the original problem was merely a
On Tue, May 03, 2011 at 05:33:11PM -0600, Eric Blake wrote:
I'm getting this when trying to build libvirt on rawhide (using the
package gnutls-devel-2.12.3-1.fc16.x86_64):
remote/remote_driver.c: In function 'negotiate_gnutls_on_connection':
remote/remote_driver.c:1361:9: error:
On 05/03/2011 06:14 PM, Eric Blake wrote:
clang didn't like the last increment to nargs. But why even
track nargs ourselves, when virCommand does it for us?
* src/storage/storage_backend_iscsi.c
(virStorageBackendISCSIConnection): Switch to virCommand to avoid
a dead-store warning on nargs.
On 05/03/2011 06:17 PM, Eric Blake wrote:
Detected by clang.
* src/esx/esx_util.c (esxUtil_ParseDatastorePath): No need to
increment.
---
src/esx/esx_util.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/esx/esx_util.c b/src/esx/esx_util.c
index
On 05/03/2011 05:04 PM, Eric Blake wrote:
Clang noticed a dead assignment, which turned out to be the use
of the wrong variable. rc starts life as -1, and is only ever
assigned to 0 just before a successful cleanup.
* src/lxc/lxc_driver.c (lxcSetupInterfaces): Don't call
On 05/03/2011 04:29 PM, Eric Blake wrote:
Clang complained about this, and it was easy enough to fix.
* src/util/util.c (virFileOpenAs): Drop dead assignment.
---
src/util/util.c |3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/src/util/util.c b/src/util/util.c
On 05/03/2011 04:22 PM, Eric Blake wrote:
Clang warned about a dead assignment. In the process, I noticed
that we are only using the function for a bool value. I audited
all other callers in qemu_{migration,cgroup,driver,hotplug), and
all were making the call in a bool context.
*
On 05/03/2011 03:10 PM, Eric Blake wrote:
Detected by clang.
* src/esx/esx_driver.c (esxDomainGetInfo): Fail early on error.
---
src/esx/esx_driver.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/src/esx/esx_driver.c b/src/esx/esx_driver.c
index 1f8f90b..e929208
On 05/03/2011 01:58 PM, Eric Blake wrote:
Detected by clang. NULL deref added in commit 343a27a (Mar 11),
but leak of voldef present since commit 2cd9b2d (Apr 09).
* src/storage/storage_driver.c (storageVolumeCreateXML): Don't
leak voldef or dereference null volobj.
---
55 matches
Mail list logo