gt; > */
> > +QEMU_CAPS_KVM, /* Whether KVM is enabled by default */
> > +QEMU_CAPS_DRIVE_FORMAT, /* Is -drive format= avail */
>
> ...
>
> That is, group them by 5 and separate the groups with empty lines so
> that
> the list looks like the corresponding definition in
> qemu_c
This gets rid of the partially enforced alignment and makes it less
likely for a bogus value to be introduced in the enumeration.
Use #define for QEMU_CAPS_NET_NAME and QEMU_CAPS_HOST_NET_ADD, both
of which are aliases for QEMU_CAPS_0_10.
---
First attempt was
[libvirt] [PATCH] qemu: Align
The primary maintainers and people with commit access rights:
Alex Jia <a...@redhat.com>
+Andrea Bolognani <abolo...@redhat.com>
Cédric Bosdonnat <cbosdon...@suse.com>
Christophe Fergeau <cferg...@redhat.com>
Claudio Bley <claudio.b...@gmail.com>
--
2.4.3
--
libvir-list
?
Definitely.
I've just sent a patch that tries to fix the same issue
with a different approach, please check it out.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
Since a9fe620372144db, we are generating virkeymaps.h at build
time; however, we are not including $(builddir)/util in the
header search path, so when doing a VPATH build the compiler
is unable to locate the file.
make[2]: Entering directory
On Fri, 2015-10-16 at 09:47 +0200, Martin Kletzander wrote:
> On Thu, Oct 15, 2015 at 04:12:07PM +0200, Andrea Bolognani wrote:
> > As agreed, I've followed up with the cleanups by splitting
> > the huge news.html.in file into separate, smaller files,
> > one per
No functional changes.
---
src/qemu/qemu_hostdev.c | 4 ++--
src/qemu/qemu_hostdev.h | 2 +-
src/qemu/qemu_hotplug.c | 4 ++--
3 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/src/qemu/qemu_hostdev.c b/src/qemu/qemu_hostdev.c
index a4409d6..ec708c8 100644
---
I was working on the QEMU hostdev code and the inconsistent
naming used kept making things just that little harder for
me to follow, so I unified it.
Cheers.
Andrea Bolognani (4):
qemu: hostdev: Unify naming for qemuPrepareHost*Devices()
qemu: hostdev: Unify naming
This calls the PCI-, USB- and SCSI-specific functions just like
qemuPrepareHostDevices() and qemuDomainReAttachHostDevices()
already do.
Update qemuProcessReconnect() to use the new function.
---
src/qemu/qemu_hostdev.c | 18 ++
src/qemu/qemu_hostdev.h | 2 ++
No functional changes.
---
src/qemu/qemu_hostdev.c | 40
src/qemu/qemu_hostdev.h | 22 +++---
src/qemu/qemu_hotplug.c | 9 -
3 files changed, 35 insertions(+), 36 deletions(-)
diff --git a/src/qemu/qemu_hostdev.c
No functional changes.
---
src/qemu/qemu_hostdev.c | 12 ++--
src/qemu/qemu_hostdev.h | 12 ++--
src/qemu/qemu_process.c | 6 +++---
3 files changed, 15 insertions(+), 15 deletions(-)
diff --git a/src/qemu/qemu_hostdev.c b/src/qemu/qemu_hostdev.c
index ec708c8..d395271 100644
le missing them.
> >
> > Thanks for spotting that, I will include your fix.
> >
>
> Then ACK with that fix then.
Pushed, thanks :)
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
ing of negative values,
but make the error message when count == 0 more
specific as Peter suggested.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
e symbolic names leading to it. I think
> the simplest solution of this mess is just
>
> #define QEMU_CAPS_NET_NAME QEMU_CAPS_0_10
> #define QEMU_CAPS_HOST_NET_ADD QEMU_CAPS_0_10
>
> and only keep QEMU_CAPS_0_10 in the enum.
I did just that and sent a v2 to the list
e minutes too late, that the #define statements were
incorrectly indented and were breaking syntax-check.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
Two #define lines introduced by b527aa0 did not respect the
indentation rules, thus breaking syntax-check.
---
Pushed as trivial.
src/qemu/qemu_capabilities.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/qemu/qemu_capabilities.h b/src/qemu/qemu_capabilities.h
index
ood point. I've changed the script to just copy the
data we're interested in instead of copying everything
and cleaning it up afterwards, and would you believe it?
It's not only easier to understand and more robust, but
also quite a bit shorter :)
I'll send a v2 right away.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
should have received a copy of the GNU Lesser General Public
+# License along with this library. If not, see
+# <http://www.gnu.org/licenses/>.
+#
+# Author: Andrea Bolognani <abolo...@redhat.com>
+
+SYSFS_PATH="/sys/devices/system"
+NODE_TOPLEVEL_COPY="online possible&quo
. If not, see
+# <http://www.gnu.org/licenses/>.
+#
+# Author: Andrea Bolognani <abolo...@redhat.com>
+
+die() {
+local message=$1
+
+echo "$message" >&2
+
+exit 1
+}
+
+list_cpus() {
+local datadir=$1
+
+local index
+
+ls -d "$datadi
A bunch of files that we don't currently parse, and are very
unlikely to ever start parsing, made their way into the nodeinfo
test data. Get rid of them.
---
tests/nodeinfodata/linux-f21-mustang/cpu/modalias| 1 -
tests/nodeinfodata/linux-raspberrypi/cpu/cpuidle/current_driver
Changes from v1:
* use a better approach to copy host data (thanks Martin)
* acknowledge the fact that the scripts require bash
* other small cleanups and improvements
Andrea Bolognani (2):
tests: Add script to display nodeinfo test data
tests: Add script to copy nodeinfo test data
On Thu, 2015-10-08 at 18:12 +0100, Daniel P. Berrange wrote:
> On Thu, Oct 08, 2015 at 06:55:23PM +0200, Andrea Bolognani wrote:
> > Since a9fe620372144db, we are generating virkeymaps.h at build
> > time; however, we are not including $(builddir)/util in the
> > header searc
ah, using $(builddir) is not good
enough as older version of autoconf don't define it.
I've just sent the patch after testing it on CentOS 5,
please check it out.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
way ;)
Pushed, thanks!
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
Commit 4e8032272f1704f7 used $(builddir) in the header search
path to fix a build issue; however, $(builddir) is not defined
by old autoconf versions such as the one available in CentOS 5,
resulting in the following error:
cc1: error: /util: No such file or directory
make[3]: ***
On Fri, 2015-10-09 at 09:55 +0200, Martin Kletzander wrote:
> With this we could stop generating whole bunch of other stuff in
> $(srcdir) See output of `git grep -F '> $(srcdir)'` for details =)
That would be very nice indeed! I'll look into it.
Cheers.
--
Andrea Bolognani
Software
with the former option (because it requires less
changes to the code) and add a paragraph about it to the documentation,
but I'd love to hear your opinion on the subject.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https
On Thu, 2015-07-09 at 11:46 -0400, John Ferlan wrote:
On 07/07/2015 03:25 AM, Andrea Bolognani wrote:
Changes from v3 to v4:
* removed a printf() statement;
* fixed typo in a commit message.
Shivaprasad G Bhat (2):
Fix nodeinfo output on PPC64 KVM hosts
Add testcase
names, and eg.
the Nautilus file manager happily allows me to use foo;reboot
as name when creating or renaming files...
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
,
+virNodeCapsFillCPUInfo(const char *cpupath,
+ int cpu_id ATTRIBUTE_UNUSED,
virCapsHostNUMACellCPUPtr cpu
ATTRIBUTE_UNUSED)
cpupath should be marked ATTRIBUTE_UNUSED as well.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list
e bits of memory that need to be locked.
So there's no comprehensive documentation that I know of,
except for the code itself :)
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
atus quo.
Would you prefer it if I pulled this patch from the series
for now and posted it again once it supports restoring the
limit back to the default once the last VFIO device has been
removed from the guest? I'd be okay with that.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization
On Wed, 2015-11-18 at 16:11 +0100, Peter Krempa wrote:
> On Wed, Nov 18, 2015 at 15:13:19 +0100, Andrea Bolognani wrote:
> > Unlike other architectures, ppc64 domains need to lock memory
> > even when VFIO is not used.
> >
> > Change qemuDomainRequiresM
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1254044
---
src/qemu/qemu_domain.c | 3 ++-
.../qemuxml2argvdata/qemuxml2argv-pseries-net-default.args | 6 ++
.../qemuxml2argvdata/qemuxml2argv-pseries-net-default.xml | 14 ++
As the name suggests, this capability can be used to check whether
a QEMU binary supports the virtio-net-* devices.
---
src/qemu/qemu_capabilities.c | 7 ++-
src/qemu/qemu_capabilities.h | 1 +
tests/qemucapabilitiesdata/caps_1.2.2-1.caps | 1 +
This applies to all architectures except for ARM, which already
has its own logic to pick the best default.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1254044
---
Changes from v1:
* make sure virtio-net is available using capabilities
instead of blindly using it (thanks Martin)
This new version of the patch addresses Martin's comments
by using capabilities to detect whether virtio-net can be
used, and if that is the case, does so on all architectures.
Plus, the commit messages are slightly less terse this time
around :)
Cheers.
Andrea Bolognani (2):
qemu: Introduce
s robust on non-x86
> > arches.
>
> Hum... I didn't even know that file would change with arch'es.
Take a look at linuxNodeInfoCPUPopulate() in src/nodeinfo.c
for inspiration. Sharing the parsing code would also be nice.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Tea
er maintained in qemu than the rtl8139, and 3)
> performs
> better than rtl8139 (although obviously not as good as virtio).
Good point. We could switch the default to virtio just for
ppc64 then, leaving the capability check in place of course,
and change the global default from rtl8139
the case eg. on RHEL ppc64.
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
d the RHEL QEMU PPC64 changes.
Yeah, Martin suggested doing something similar to the
first option as well.
Let's just probe for a bunch of network devices and use
the first one that's available, okay?
1. rtl8139
2. e1000
3. virtio-net
Any other we should try? Any drawbacks to this appr
we need a 'qemuDomainMachineIsS390CCW' helper?
I'd also use STRPREFIX() instead of STREQLEN().
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
On Wed, 2015-09-09 at 21:43 +0200, Martin Kletzander wrote:
> > Andrea Bolognani (4):
> > qemu: Introduce QEMU_CAPS_DEVICE_RTL8139
> > qemu: Introduce QEMU_CAPS_DEVICE_E1000
> > qemu: Introduce QEMU_CAPS_DEVICE_VIRTIO_NET
> > qemu: Try several network devi
This capability can be used to detect whether or not the QEMU
binary supports the rtl8139 network device.
---
src/qemu/qemu_capabilities.c | 5 -
src/qemu/qemu_capabilities.h | 1 +
tests/qemucapabilitiesdata/caps_1.2.2-1.caps | 1 +
Up until now, the default has been rtl8139, but no check was in
place to make sure that device was actually available.
Now we try rtl8139, e1000 and virtio-net in turn, checking for
availability before using any of them: this means we have a much
better chance for the guest to be able to boot.
This capability can be used to detect whether or not the QEMU
binary supports the e1000 network device.
---
src/qemu/qemu_capabilities.c | 2 ++
src/qemu/qemu_capabilities.h | 1 +
tests/qemucapabilitiesdata/caps_1.2.2-1.caps | 1 +
This capability can be used to detect whether or not the QEMU
binary supports the virtio-net-* network device.
---
src/qemu/qemu_capabilities.c | 5 +
src/qemu/qemu_capabilities.h | 1 +
tests/qemucapabilitiesdata/caps_1.2.2-1.caps | 1 +
in the test suite
Changes in v2[1]:
* detect virtio-net availability
* try to user it if available
Cheers.
[1] https://www.redhat.com/archives/libvir-list/2015-September/msg00058.html
Andrea Bolognani (4):
qemu: Introduce QEMU_CAPS_DEVICE_RTL8139
qemu: Introduce QEMU_CAPS_DEVICE_E1000
When looking for a QEMU binary suitable for running ppc64le guests
we have to take into account the fact that we use the QEMU target
as key for the hash, so direct comparison is not good enough.
Factor out the logic from virQEMUCapsFindBinaryForArch() to a new
virQEMUCapsFindTarget() function and
a.arch == data.arch from before.
Good idea! After all, the definition of insanity is looking up
using the same key twice and expecting a result the second time
around :)
I've just posted a new version[1] that implements your suggestion.
Cheers.
[1] https://www.redhat.com/archives/libvir-list
On Wed, 2015-09-16 at 09:22 +0100, Daniel P. Berrange wrote:
> On Wed, Sep 16, 2015 at 09:13:09AM +0200, Andrea Bolognani wrote:
> > When looking for a QEMU binary suitable for running ppc64le guests
> > we have to take into account the fact that we use the QEMU target
> > as
When looking for a QEMU binary suitable for running ppc64le guests
we have to take into account the fact that we use the QEMU target
as key for the hash, so direct comparison is not good enough:
normalize everything that can run on the ppc64 target to
VIR_ARCH_PPC64 instead.
This approach was
When looking for a QEMU binary suitable for running ppc64le guests
we have to take into account the fact that we use the QEMU target
as key for the hash, so direct comparison is not good enough.
Factor out the logic from virQEMUCapsFindBinaryForArch() to a new
virQEMUCapsFindTarget() function and
separate
function which is then reused when looking up capabilities.
Of course it's a slightly larger patch this time around :)
Cheers.
[1] https://www.redhat.com/archives/libvir-list/2015-September/msg00475
.html
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list
This allows the Wikipedia link to be recognized correctly by eg.
gnome-terminal's Open Link and Copy Link Address features.
---
Pushed as trivial
src/util/virarch.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/util/virarch.h b/src/util/virarch.h
index 3206ce2..af5ff83
qemu-kvm can be used to run ppc64 guests on ppc64le hosts and vice
versa, since the hardware is actually the same and the endianness
is chosen by the guest kernel.
Up until now, however, libvirt didn't allow the use of qemu-kvm
to run guests if their endianness didn't match the host's.
Resolves:
solated environment that libvirt, amongsth
> +other backends, can be used.
> +
> +
> +
>
>
s/amongsth/amongst/
Or even s/amongsth/among/ [1]
Plus the whole last sentence looks like some words went missing...
Maybe take another look at it.
Cheers.
[1] http://grammarist.co
ment." ?
Much better :)
ACK
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
UCapsFindTarget() can be considered a first step in that
direction. A lot more work is still needed, though.
> In other words, ACK :-)
Thanks :)
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
aightforward wrapper,
eg. the return statement is the only one in the body.
That's of course just my personal taste :)
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
On Thu, 2015-09-24 at 19:03 +0200, Pavel Hrdina wrote:
> On Thu, Jun 11, 2015 at 03:15:19PM +0200, Andrea Bolognani wrote:
> > ---
> > COPYING.LESSER | 18 +-
> > 1 file changed, 9 insertions(+), 9 deletions(-)
> >
>
> ACK :)
Oh, I had co
---
This pushes a lot of lines well beyond 80 columns, but then again some
of the entries were already as long as 92 columns...
src/qemu/qemu_capabilities.h | 412 +--
1 file changed, 205 insertions(+), 207 deletions(-)
diff --git
On Mon, 2015-10-05 at 15:24 +0200, Jiri Denemark wrote:
> On Mon, Oct 05, 2015 at 14:13:05 +0200, Andrea Bolognani wrote:
> > ---
> > This pushes a lot of lines well beyond 80 columns, but then again
> > some
> > of the entries were already as long as 92 co
The value is not inspected inside the function, so it makes more
sense for the caller to change the device's setting explicitly.
---
src/util/virpci.c | 10 ++
src/util/virpci.h | 2 +-
tests/virpcitest.c | 2 +-
3 files changed, 8 insertions(+), 6 deletions(-)
diff --git
When reattaching VFIO devices to the host, we have to be careful not to
end up in a situation where ownership of the IOMMU group is shared
between the hosts and a domain, as doing so will result in host crashes.
If a PCI device is not safe for reattach, as detected by
This replaces the virPCIKnownStubs string array that was used
internally for stub driver validation.
Advantages:
* possible values are well-defined
* typos in driver names will be detected at compile time
* avoids having several copies of the same string around
* no error checking
When we determine that a VFIO device can be safely reattached to the
host, we have to also reattach all devices we might have delayed
earlier in order to clean up properly.
The new function virHostdevExpandPCIDeviceListIOMMUGroups() helps by
adding inactive devices that are part of the same IOMMU
We mark all VFIO devices as inactive after we've detached them from the
domain, so that step doesn't need to be performed afterwards.
---
src/util/virhostdev.c | 8 +++-
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/src/util/virhostdev.c b/src/util/virhostdev.c
index
This function, previously called virHostdevIsPCINodeDeviceUsed(), was
written to work both when used on its own and when used with
virPCIDeviceAddressIOMMUGroupIterate(); as a consequence, it was
slightly more convoluted than it needed to be.
Rework it to perform a single task (checking that a
The newly-introduced virHostdevIsPCIDeviceSafeForDetach() function
checks against all problematic situations that can arise when detaching
a PCI device from the host for VFIO device assignment.
Some of the checks are simply reintroduced after being removed when
reworking
This is a straightforward wrapper around
virPCIDeviceAddressIOMMUGroupIterate() that will make some code less
awkward to write later on.
---
src/libvirt_private.syms | 1 +
src/util/virpci.c| 26 ++
src/util/virpci.h| 8 ++--
3 files changed, 33
Add a couple of useful debug messages, update comments to reflect recent
changes, use the proper case for VFIO in variable names and fix
parameter alignment in a function definition.
---
src/util/virhostdev.c | 27 ---
1 file changed, 16 insertions(+), 11 deletions(-)
, thus preventing the crash and fixing the bug
10: Spit and polish
Cheers.
[1] Luckily, it doesn't break the existing tests either
[2] If you call 'virsh nodedev-reattach' on a device that's
assigned to a guest, libvirt won't stop you and you will
end up crashing your system
Andrea
On Thu, 2015-12-03 at 14:06 -0500, Laine Stump wrote:
> On 12/02/2015 12:39 PM, Andrea Bolognani wrote:
> > This is a straightforward wrapper around
> > virPCIDeviceAddressIOMMUGroupIterate() that will make some code less
> > awkward to write later on.
> > ---
> >
On Fri, 2015-12-04 at 16:31 +0100, Michal Privoznik wrote:
> ACK, although I'd rather wait until after release. If you (or somebody
> else) feels otherwise, do push it any time you want.
No need to rush, I'll push after release.
Thanks for the review!
--
Andrea Bolognani
Software En
The old name is no longer accurate, since now we're using its value as
the root of the fake filesystem.
No functional changes.
---
tests/vircgroupmock.c | 10 +-
tests/vircgrouptest.c | 16
tests/virhostdevtest.c | 16
tests/virpcimock.c | 6
We might need to mock files living outside SYSFS_PREFIX later on,
so it's better to treat the temporary directory we are passed via
the environment as the root of the fake filesystem and create
SYSFS_PREFIX inside it.
The environment variable name will be changed to reflect the new use
we're
We might need to mock files living outside PCI_SYSFS_PREFIX later on,
so it's better to treat the temporary directory we are passed via
the environment as the root of the fake filesystem and create
PCI_SYSFS_PREFIX inside it.
The environment variable name will be changed to reflect the new use
://www.redhat.com/archives/libvir-list/2015-November/msg00532.html
is an example of why we would need to do that
Andrea Bolognani (7):
tests: scsihost: Don't set LIBVIRT_FAKE_SYSFS_DIR
tests: pcimock: Remove check for fakesysfsdir
tests: pcimock: Use the temporary directory as fake root
Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and
fakesysfscgroupdir. This makes it explicit what part of the fake sysfs
filesystem they're referring to, and also leaves open the possibility of
handling files in two unrelated parts of the fake sysfs filesystem.
No functional
This updates the test program to make it consistent with recent changes
to the mock libraries, and also opens up the possibility of mocking more
than just /sys in the future.
---
tests/scsihosttest.c | 17 -
1 file changed, 12 insertions(+), 5 deletions(-)
diff --git
The test program is not preloading any of the mock libraries that read
that environment variable, so setting it is pointless.
---
tests/scsihosttest.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/tests/scsihosttest.c b/tests/scsihosttest.c
index eecb1c3..eb5f522 100644
---
init_env() will return right away if fakesysfsdir is already
initialized, so this check is redundant.
---
tests/virpcimock.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/tests/virpcimock.c b/tests/virpcimock.c
index 0b49290..6d00a4f 100644
--- a/tests/virpcimock.c
+++
This function detects whether a domain needs RLIMIT_MEMLOCK
to be set, and if so, uses an appropriate value.
---
src/qemu/qemu_domain.c | 30 ++
src/qemu/qemu_domain.h | 1 +
2 files changed, 31 insertions(+)
diff --git a/src/qemu/qemu_domain.c
We increase the limit before plugging in a PCI hostdev or a memory
module because some memory might need to be locked due to eg. VFIO.
Of course we should do the opposite after unplugging a device: this
was already the case for memory modules, but not for PCI hostdevs.
---
* small changes here and there as pointed out during review
Cheers.
[1] https://www.redhat.com/archives/libvir-list/2015-November/msg01021.html
Andrea Bolognani (7):
process: Allow virProcessPrLimit() to get current limit
process: Add virProcessGetMaxMemLock()
qemu: Add
Replace all uses of the qemuDomainRequiresMlock/virProcessSetMaxMemLock
combination with the equivalent qemuDomainAdjustMaxMemLock() call.
---
src/qemu/qemu_hotplug.c | 40 +---
1 file changed, 13 insertions(+), 27 deletions(-)
diff --git
passthrough - are legacy KVM
assignment and Xen assignment really not affected at all? Is the IOMMU
not involved in the process?
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
inRemovePCIHostDevice code.
I have now posted v2:
https://www.redhat.com/archives/libvir-list/2015-December/msg00454.html
Cheers.
--
Andrea Bolognani
Software Engineer - Virtualization Team
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list
On Fri, 2015-12-04 at 09:46 +0100, Andrea Bolognani wrote:
> On Thu, 2015-12-03 at 14:06 -0500, Laine Stump wrote:
> > Instead of creating this new function, how about if we change
> > virPCIDevice to contain a virPCIDeviceAddress rather than individual
> > domain, bus, slot,
, but extending it so that 'host' is accepted as well should be
easy to do in a backwards-compatible way.
The behaviour of the value '2' has been defined as "don't specify any
GIC version" though, and I'm afraid we'll be unable to change that
because it would affect existing guests. Marti
On Tue, 2015-12-15 at 12:27 -0500, Laine Stump wrote:
> On 12/15/2015 05:48 AM, Andrea Bolognani wrote:
> > On Tue, 2015-12-15 at 10:05 +0100, Andrea Bolognani wrote:
> > > @@ -1661,25 +1661,14 @@ virPCIDeviceFree(virPCIDevicePtr dev)
> > > * @dev:
Instead of replicating the information (domain, bus, slot, function)
inside the virPCIDevice structure, use the already-existing
virPCIDeviceAddress structure.
For users of the module, this means that the object returned by
virPCIDeviceGetAddress() can no longer be NULL and must no longer
be
On Wed, 2015-12-16 at 14:32 -0500, John Ferlan wrote:
>
> On 12/11/2015 09:45 AM, Andrea Bolognani wrote:
> > During my testing, I've realized that even with the series applied
> > there's still one case in which we're unable to restore the previous
> > memory lock
On Fri, 2015-12-04 at 17:07 +0100, Andrea Bolognani wrote:
> On Fri, 2015-12-04 at 16:31 +0100, Michal Privoznik wrote:
> > ACK, although I'd rather wait until after release. If you (or somebody
> > else) feels otherwise, do push it any time you want.
>
> No need to rush, I'
On Wed, 2015-12-02 at 18:17 +0100, Andrea Bolognani wrote:
> This series is my attempt at fixing
>
> https://bugzilla.redhat.com/show_bug.cgi?id=1272300
>
[...]
>
> The problem being solved is that, when using VFIO, IOMMU group
> ownership can't be sh
MemLock is already used in other modules and, while still an
abbreviation, is not ambiguous.
---
src/qemu/qemu_command.c | 4 ++--
src/qemu/qemu_domain.c | 10 +-
src/qemu/qemu_domain.h | 4 ++--
3 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/src/qemu/qemu_command.c
The prlimit() function allows both getting and setting limits for
a process; expose the same functionality in our wrapper.
Add the const modifier for new_limit, in accordance with the
prototype for prlimit().
---
src/util/virprocess.c | 16 ++--
1 file changed, 10 insertions(+), 6
When the function changes the memory lock limit for the first time,
it will retrieve the current value and store it inside the
virDomainObj for the domain.
When the function is called again, if memory locking is no longer
needed, it will be able to restore the memory locking limit to its
original
301 - 400 of 8356 matches
Mail list logo