Loading
Commits on Source 59
-
Daniel P. Berrangé authoredSigned-off-by:Daniel P. Berrangé <berrange@redhat.com>
-
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Jobs starting with a '.' are templates so never run unless referenced by another job. Fixes: 6872a7cc Signed-off-by:
-
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Michal Privoznik authored
Distro changes include: * Fedora 39/40 -> 40/41 Signed-off-by:Michal Privoznik <mprivozn@redhat.com>
-
Michal Privoznik authored
Golang 1.24 is the latest version. We already test with version 1.11 (the oldest version required), then 1.16 - those can stay. But 1.20 should be switched to the most recent. Signed-off-by: -
Daniel P. Berrangé authored
Currently all return values less than zero get turned into an error return. This is incorrect because -2 indicates "blocking" and -3 indicates "in a hole" and in neither of these cases will libvirt report any error. These special return codes must be propagated as is. Closes: #19 Signed-off-by:
-
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This would have caught the typo fixed in a recent commit. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This would have prevented the missing 'serial' field fixed in a recent commit. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This would have avoided the many missing fields fixed in a recent commit. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
The VM stats passthrough is documented have fields ending in one of three different suffixes. Rather than blindly expose everything, filter based on the suffixes. This will prevent exposing any future fields that might be better done as named struct fields. Signed-off-by: -
Daniel P. Berrangé authored
As well as the 'vm.*' custom hypervisor stats already exposed, there are further custom stats per VCPU. Using the custom stats filter that validates the suffix ensures we don't expose any of the normal vCPU stats this way. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This would have avoided the missing field fixed in a recent commit. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This would have prevented us from missing these fields before. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Michal Privoznik authored
Couple of things happening here: 1) new VIR_DOMAIN_EVENT_ID_NIC_MAC_CHANGE event 2) new error codes 3) new APIs 4) some macros were defined twice, the redefinition is removed. Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
Actual generation is done in the next commit. Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
There's one API that's not exposed: virAdmInitialize() because it's kind of special. It's covered by the next commit. Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
The symbol is just a typedef for a callback. Put it onto an ignore list, right next to virConnectCloseFunc which faces the same fate. Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Michal Privoznik authored
Signed-off-by: -
Daniel P. Berrangé authored
If a enum typedef is backported to an old release, the public header file definition will clash with our typedef, because the compiler will not consider an "int" to be equivalent to an "enum". This is seen currently with CentOS Stream 9: In file included from ./libvirt_generated.h:41, from ./libvirt_admin_generated.h:39, from ./admin_helper.h:29, from ./admin.go:36: ./libvirt_generated_typedefs.h:280:13: error: conflicting types for 'virConnectGetDomainCapabilitiesFlags'; have 'int' 280 | typedef int virConnectGetDomainCapabilitiesFlags; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ In file included from /usr/include/libvirt/libvirt.h:36, from /usr/include/libvirt/virterror.h:27, from ./libvirt_admin_generated.h:36, from ./admin_helper.h:29, from ./admin.go:36: /usr/include/libvirt/libvirt-domain.h:1504:3: note: previous declaration of 'virConnectGetDomainCapabilitiesFlags' with type 'virConnectGetDomainCapabilitiesFlags' 1504 | } virConnectGetDomainCapabilitiesFlags; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Since the Go bindings don't require the enum typedef, only the constants, simplest solution is to simply not generate the typedefs. The only exception is the 'virErrorLevel' enum which was mistakenly included in the public declaration of the 'virError' struct. Signed-off-by: -
Daniel P. Berrangé authored
Debian has purged the repos for Buster, which is needed by the min golang 1.11 version. Alpine doesn't purge their old repos, so switch over to that instead. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: