On 5/23/22 07:39, Martin Kletzander wrote:
On Sat, May 21, 2022 at 01:59:05PM -0300, Daniel Henrique Barboza wrote:
qemuxml2xmltests that have "pseries" in the name now use the
DO_TEST_CAPS_LATEST_ARCH() macro.
Signed-off-by: Daniel Henrique Barboza
Reviewed-by: Martin Kletzander
Pushed fixing the typo in patch 2. Thanks!
Daniel
On 5/20/22 17:47, Daniel Henrique Barboza wrote:
Hi,
This series attempts to fix a problem I found when converting the
"pseries-cpu-compat-power9" to use DO_TEST_CAPS_ARCH_LATEST* macros.
More information about the problem per se can be found
_equal is not used anywhere; the rest of the code implements the
syntax-check target, which takes care of figuring out the list of
checks that have been defined and running them, printing the name
of each check along with its execution time.
This was useful when we were using autotools, but these
All checks are added to the syntax-check suite, and this name is
displayed prominently in the output of 'meson test', so there
really is no need to include the sc_ prefix too.
Signed-off-by: Andrea Bolognani
---
build-aux/meson.build | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
Basically all files in the repository are already passing the
check, except for syntax-check.mk itself. Fix that, and stop
limiting the files on which the test is performed.
These changes have been generated by running
$ sed -Ei 's/[ '$'\t'']+\\$/ \\/g' $(git grep -El '[ '$'\t'']+\\$')
Due to the way make works, we are not forced to follow a strict
order in defining rules and variables. In fact _sc_search_regexp,
which is used by all checks, is only defined halfway through the
file.
Shuffle things around so that the things that we need to look at
the most frequently are closer
The makefile is an implementation detail, so point users towards
the proper way of running syntax-check if they happen to call it
directly.
Signed-off-by: Andrea Bolognani
---
build-aux/syntax-check.mk | 5 +
1 file changed, 5 insertions(+)
diff --git a/build-aux/syntax-check.mk
We can assume that VC_LIST_ALWAYS_EXCLUDE_REGEX will not be
defined in a way that would catch backup files.
Signed-off-by: Andrea Bolognani
---
build-aux/syntax-check.mk | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/build-aux/syntax-check.mk b/build-aux/syntax-check.mk
Signed-off-by: Andrea Bolognani
---
build-aux/Makefile.in | 1 +
build-aux/meson.build | 3 +++
build-aux/syntax-check.mk | 5 -
3 files changed, 4 insertions(+), 5 deletions(-)
diff --git a/build-aux/Makefile.in b/build-aux/Makefile.in
index 9ccbec7b1b..7ee4680847 100644
---
The pattern in build-aux/syntax-check.mk is written specifically
so that it won't match itself, which makes having an exception
for the file unnecessary.
Signed-off-by: Andrea Bolognani
---
build-aux/syntax-check.mk | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git
It's only used in diagnostics, and even there it's not
particularly useful and can make it more difficult to spot the
actual error message.
Signed-off-by: Andrea Bolognani
---
build-aux/syntax-check.mk | 58 ++-
1 file changed, 27 insertions(+), 31
The file src/util/vireventglib.c doesn't contain a main() function
and so it's not even considered by the check.
Signed-off-by: Andrea Bolognani
---
build-aux/syntax-check.mk | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/build-aux/syntax-check.mk
This is one of the standard checks that we have inherited from
gnulib, but it's not applicable to libvirt because we don't want
plain bindtextdomain() to be used: virGettextInitialize() is our
own private API that should be used instead.
The sc_gettext_init check ensures that our private API is
Most of the pattern is no longer relevant, because the files it
was intended to match have been dropped from the repository.
Specifically:
filescommitdate
-- --
*.gif 6cb131e5cbd0 2022-01-19
In its current form, the check will not only catch the intended
#include
but also stuff like
#include
#include "qemu_interop_config.h"
#include
The last one is problematic, because it's used in config.h itself.
Making the pattern more strict allows us to drop the exception.
$ meson test --suite blurb
Andrea Bolognani (13):
syntax-check: Drop sc_bindtextdomain check
syntax-check: Remove sc_gettext_init exception
syntax-check: Remove sc_copyright_usage exception
syntax-check: Improve sc_prohibit_config_h_in_headers check
syntax-check: Use VC_LIST_EXCEPT in
On Mon, May 09, 2022 at 05:02:13PM +0200, Michal Privoznik wrote:
> Since its 5.14 release the Linux kernel allows userspace to
> define trusted groups of processes/threads that can run on
> sibling Hyper Threads (HT) at the same time. This is to mitigate
> side channel attacks like L1TF or MDS.
On Mon, May 09, 2022 at 05:02:12PM +0200, Michal Privoznik wrote:
> In near future it will be necessary to know the PID of virtiofsd
> started for QEMU. Move the code into a separate function
> (qemuVirtioFSGetPid()) and export it in the header file.
>
> Signed-off-by: Michal Privoznik
> ---
>
On Mon, May 09, 2022 at 05:02:11PM +0200, Michal Privoznik wrote:
> In near future it will be necessary to know the PID of swtpm
> process for QEMU. Export the function that does just that
> (qemuTPMEmulatorGetPid()).
>
> Signed-off-by: Michal Privoznik
> ---
> src/qemu/qemu_tpm.c | 2 +-
>
On Mon, May 09, 2022 at 05:02:10PM +0200, Michal Privoznik wrote:
> In near future it will be necessary to know the PID of
> vhost-user-gpu process for QEMU. Export the function that does
> just that (qemuVhostUserGPUGetPid()).
>
> Signed-off-by: Michal Privoznik
> ---
>
On Mon, May 09, 2022 at 05:02:09PM +0200, Michal Privoznik wrote:
> In near future it will be necessary to know the PID of DBus
> daemon started for QEMU. Move the code into a separate function
> (qemuDBusGetPID()) and export it in the header file.
>
> Signed-off-by: Michal Privoznik
> ---
>
On Mon, May 09, 2022 at 05:02:08PM +0200, Michal Privoznik wrote:
> In qemu_extdevice.c lives code that handles helper daemons that
> are required for some types of devices (e.g. virtiofsd,
> vhost-user-gpu, swtpm, etc.). These devices have their own
> handling code in separate files, with only a
On Mon, May 09, 2022 at 05:02:14PM +0200, Michal Privoznik wrote:
> There are two modes of core scheduling that are handy wrt
> virCommand:
>
> 1) create new trusted group when executing a virCommand
>
> 2) place freshly executed virCommand into the trusted group of
>another process.
>
>
On Mon, May 09, 2022 at 05:02:17PM +0200, Michal Privoznik wrote:
> Since the level of trust that QEMU has is the same level of trust
> that helper processes have there's no harm in placing all of them
> into the same group.
This assumption feels like it might be a bit of a stretch. I
recall
On 5/23/22 11:22 AM, Michal Privoznik wrote:
We have virDomainUpdateDeviceFlags() API that allows changing of
some attributes of a device whilst domain is still running (e.g.
setting different QoS, link state change on vNICs). But only very
limited set of attributes can be changed and we have to
On Mon, May 09, 2022 at 05:02:07PM +0200, Michal Privoznik wrote:
> The Linux kernel offers a way to mitigate side channel attacks on Hyper
> Threads (e.g. MDS and L1TF). Long story short, userspace can define
> groups of processes (aka trusted groups) and only processes within one
> group can run
As "none" is a legal value represented in the sysfs attribute dev_busid
this patch prevents libvirt from incorrectly reporting an internal error.
Signed-off-by: Boris Fiuczynski
Suggested-by: Michal Privoznik
---
src/node_device/node_device_udev.c | 2 +-
1 file changed, 1 insertion(+), 1
On 5/23/22 4:40 PM, Michal Prívozník wrote:
I'm fixing small memleaks I've raised in 16/17 and merging.
Reviewed-by: Michal Privoznik
Michal
Thanks Michal
--
Mit freundlichen Grüßen/Kind regards
Boris Fiuczynski
IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats:
On 5/23/22 4:40 PM, Michal Prívozník wrote:
On 5/13/22 12:31, Boris Fiuczynski wrote:
Add the new introduced sysfs attribute dev_busid which provides the address
of the device in the subchannel independent from the bound device driver.
It is added if available in the sysfs as optional
We have virDomainUpdateDeviceFlags() API that allows changing of
some attributes of a device whilst domain is still running (e.g.
setting different QoS, link state change on vNICs). But only very
limited set of attributes can be changed and we have to check
whether user isn't trying to sneak in a
On 5/23/22 16:46, Jiri Denemark wrote:
> My recent commit v8.3.0-201-gc500955e95 tried to fix a regression which
> would cause the function to return success even if virCloseCallbacksSet
> failed. But due to a strange code flow in the function introduced an
> opposite regression. The function
My recent commit v8.3.0-201-gc500955e95 tried to fix a regression which
would cause the function to return success even if virCloseCallbacksSet
failed. But due to a strange code flow in the function introduced an
opposite regression. The function would return NULL on success when
called without
On 5/13/22 12:31, Boris Fiuczynski wrote:
> Add the new introduced sysfs attribute dev_busid which provides the address
> of the device in the subchannel independent from the bound device driver.
> It is added if available in the sysfs as optional channel_dev_addr element
> into
> the css device
On 5/13/22 12:30, Boris Fiuczynski wrote:
> While this series started with the intend to add the optional device
> address of a subchannel device to the nodedev css device the outcome now
> also includes a small fix in the error reporting of css cap XML parsing
> as well as a refactoring of
We already allow users to provide TFTP root path in network XML
and not specify any DHCP. This makes sense, because dnsmasq is
not only DHCP server but also TFTP server and users might have
a DHCP server configured on their own, outside of libvirt's
control and want just the TFTP part.
By moving
Generating configuration file for dnsmasq is done in
networkDnsmasqConfContents() which is this big, self-contained
function. Separate at least DHCP part into its own function for
better readability.
Signed-off-by: Michal Privoznik
---
src/network/bridge_driver.c | 227
In networkDnsmasqConfContents() there's a for() loop which
initializes some variables in its initialization block. This
makes both the loop() statement and variable declaration block
look needlessly ugly. Speaking of variable declaration, also move
some variables which are used only within blocks
This is rebased version of:
https://listman.redhat.com/archives/libvir-list/2021-December/226045.html
Michal Prívozník (3):
network: Initialize variables in networkDnsmasqConfContents()
network: Separate DHCP config generator into a function
network: Generate TFTP config regardless of DHCP
The virDomainChrDefParseXML() function uses old style of parsing
XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/ch/ch_monitor.c | 2 +-
src/conf/domain_conf.c | 82
The virDomainDiskDefParseSourceXML() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 14 +-
1 file changed, 5 insertions(+), 9
The virDomainHubDefParseXML() function uses old style of parsing
XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 17 -
src/conf/domain_conf.h | 14
The virDomainInputDefParseXML() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_audit.c | 2 +-
src/conf/domain_conf.c | 86
The virDomainTimerDefParseXML() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 105 +--
The virDomainPanicDefParseXML() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 11 ---
src/conf/domain_conf.h | 2 +-
The virDomainStorageNetworkParseHost() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c| 19 +++
The virDomainTPMDefParseXML() function uses old style of parsing
XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 43
The virDomainHostdevSubsysSCSIDefParseXML() function uses old
style of parsing XML (virXMLPropString + str2enum conversion).
Use virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 19 ++-
src/conf/domain_conf.h
The virDomainDiskDefMirrorParse() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 26 ++
src/conf/domain_conf.h |
The virDomainDiskSourceNVMeParse() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 15 +++
1 file changed, 3 insertions(+), 12
The virDomainHostdevSubsysSCSIVHostDefParseXML() function uses
old style of parsing XML (virXMLPropString + str2enum
conversion). Use virXMLPropEnumDefault() which encapsulates those
steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 19 +--
The virDomainDiskSourcePoolDefParse() function uses old style of
parsing XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/domain_conf.c | 12
src/conf/storage_source_conf.h
There are three places (two in domain_conf.c and one in
qemu_migration.c) where a virStorageSource->type is typecasted to
virStorageType (for the purpose of catching missing enum member
in a switch() statement at compile time). This is needless,
because as of v8.2.0-rc1~120 the struct member is of
The virStorageSourceGetActualType() function returns either
virStorageSource->type (which is of type virStorageType), or
virStorageSourcePoolDef->type, which really stores a value of the
same enum. Thus, the latter struct can be changed so that the
virStorageSourceGetActualType() function can
The virDomainBackupDefParse() function uses old style of parsing
XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/backup_conf.c | 16 ++--
src/conf/backup_conf.h | 2 +-
2 files
The virDomainDeviceAddressParseXML() function uses old style of parsing
XML (virXMLPropString + str2enum conversion). Use
virXMLPropEnumDefault() which encapsulates those steps.
Signed-off-by: Michal Privoznik
---
src/conf/device_conf.c | 12 -
src/conf/device_conf.h | 4 +--
This is rebased version of:
https://listman.redhat.com/archives/libvir-list/2022-April/229941.html
Michal Prívozník (17):
Drop needless typecast to virStorageType enum
virStorageSourceGetActualType: Change type of retval
virDomainBackupDefParse: Switch to virXMLPropEnumDefault()
On 5/23/22 13:56, Peter Krempa wrote:
> Peter Krempa (2):
> docs: domain: Remove extraneous quotes
> schemas: Allow 'manual' snapshot mode in domain definition
>
> docs/formatdomain.rst | 20 ++--
> src/conf/schemas/domaincommon.rng | 1 +
> 2 files changed, 11
Commit a1465e661e7 added the 'manual' disk snapshot mode documentation
but didn't allow it in the schema as default snapshot mode for a disk.
Add the needed value into the schema.
Signed-off-by: Peter Krempa
---
src/conf/schemas/domaincommon.rng | 1 +
1 file changed, 1 insertion(+)
diff
Certain documentation bits tried to put a reference of a value into
quotes, but that's not needed for both the pure view of the rST source
and the rendered output.
Signed-off-by: Peter Krempa
---
docs/formatdomain.rst | 20 ++--
1 file changed, 10 insertions(+), 10 deletions(-)
Peter Krempa (2):
docs: domain: Remove extraneous quotes
schemas: Allow 'manual' snapshot mode in domain definition
docs/formatdomain.rst | 20 ++--
src/conf/schemas/domaincommon.rng | 1 +
2 files changed, 11 insertions(+), 10 deletions(-)
--
2.35.3
On 5/18/22 11:52, Andrea Bolognani wrote:
> Changes from [v1]
>
> * instead of checking that the potfile doesn't contain unwanted
> comments at syntax-check time, prevent them from being added by
> passing all xgettext options explicitly ourselves.
>
> [v1]
On Thu, May 19, 2022 at 04:49:32PM +0200, Michal Privoznik wrote:
When a SIGHUP is received a thread is spawned that runs
virStateReload(). However, if SIGINT is received while the former
thread is still running then we may get into problematic
situation: the cleanup code in main() sees drivers
On Mon, May 23, 2022 at 12:55:17 +0200, Jiri Denemark wrote:
> Commit v8.3.0-152-g49ef0f95c6 removed explicit VIR_FREE from
> qemuMigrationBegin, effectively reverting v1.2.14-57-g77ddd0bba2
>
> The xml variable was used to hold the return value and thus had to be
> unset when an error happened
Commit v8.3.0-152-g49ef0f95c6 removed explicit VIR_FREE from
qemuMigrationBegin, effectively reverting v1.2.14-57-g77ddd0bba2
The xml variable was used to hold the return value and thus had to be
unset when an error happened after xml was already non-NULL. Such code
may be quite confusing though
On Sat, May 21, 2022 at 01:59:05PM -0300, Daniel Henrique Barboza wrote:
qemuxml2xmltests that have "pseries" in the name now use the
DO_TEST_CAPS_LATEST_ARCH() macro.
Signed-off-by: Daniel Henrique Barboza
Reviewed-by: Martin Kletzander
signature.asc
Description: PGP signature
On Fri, May 20, 2022 at 05:47:04PM -0300, Daniel Henrique Barboza wrote:
Use the newly added ARG_CAPS_HOST_CPU_MODEL to set which host CPU we
expect the test to use - the test should fail when using a POWER8 host
cpu but complete when using a POWER9 host cpu.
Two new macros were added because
On Fri, May 20, 2022 at 05:47:03PM -0300, Daniel Henrique Barboza wrote:
When loading a latest caps for an arch for the first time the following
occurs in testQemuInfoInitArgs():
- the caps file is located. It's not in the cache since it's the first time
it's being read;
- the cachecaps are
On Fri, May 20, 2022 at 05:47:02PM -0300, Daniel Henrique Barboza wrote:
Fix identation of virQEMUCapsUpdateHostCPUModel() params.
Signed-off-by: Daniel Henrique Barboza
trivial,
Reviewed-by: Martin Kletzander
---
src/qemu/qemu_capspriv.h | 4 ++--
1 file changed, 2 insertions(+), 2
On 5/18/22 14:48, Michal Prívozník wrote:
> On 5/9/22 17:02, Michal Privoznik wrote:
>>
>
> Polite ping.
Less polite ping.
Michal
On Mon, May 23, 2022 at 10:33:39AM +0200, Andrea Bolognani wrote:
We already allow this for OVMF.
Closes: https://gitlab.com/libvirt/libvirt/-/issues/312
Signed-off-by: Andrea Bolognani
Reviewed-by: Martin Kletzander
---
src/security/apparmor/libvirt-qemu | 2 +-
1 file changed, 1
We already allow this for OVMF.
Closes: https://gitlab.com/libvirt/libvirt/-/issues/312
Signed-off-by: Andrea Bolognani
---
src/security/apparmor/libvirt-qemu | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/security/apparmor/libvirt-qemu
71 matches
Mail list logo