[PATCH] test_driver: Implement virConnectGetDomainCapabilities()
by Michal Privoznik
Our test driver lacks implementation for
virConnectGetDomainCapabilities(). Provide one, though a trivial
one. Mostly so that something else than VIR_ERR_NO_SUPPORT error
is returned.
Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
---
src/test/test_driver.c | 36 ++++++++++++++++++++++++++++++++++++
1 file changed, 36 insertions(+)
diff --git a/src/test/test_driver.c b/src/test/test_driver.c
index c1d4d71b78..a8705a8da2 100644
--- a/src/test/test_driver.c
+++ b/src/test/test_driver.c
@@ -10000,6 +10000,41 @@ testNetworkSetMetadata(virNetworkPtr net,
return ret;
}
+static char *
+testConnectGetDomainCapabilities(virConnectPtr conn G_GNUC_UNUSED,
+ const char *emulatorbin,
+ const char *arch_str,
+ const char *machine,
+ const char *virttype_str,
+ unsigned int flags)
+{
+ g_autoptr(virDomainCaps) domCaps = NULL;
+ virArch arch = VIR_ARCH_NONE;
+ int virttype = VIR_DOMAIN_VIRT_NONE;
+
+ virCheckFlags(0, NULL);
+
+ if (arch_str &&
+ (arch = virArchFromString(arch_str)) == VIR_ARCH_NONE) {
+ virReportError(VIR_ERR_INVALID_ARG,
+ _("unknown architecture: %1$s"), arch_str);
+ return NULL;
+ }
+
+ if (virttype_str &&
+ (virttype = virDomainVirtTypeFromString(virttype_str)) < 0) {
+ virReportError(VIR_ERR_INVALID_ARG,
+ _("unknown virttype: %1$s"), virttype_str);
+ return NULL;
+ }
+
+ if (!(domCaps = virDomainCapsNew(emulatorbin, machine, arch, virttype)))
+ return NULL;
+
+ return virDomainCapsFormat(domCaps);
+}
+
+
/*
* Test driver
*/
@@ -10167,6 +10202,7 @@ static virHypervisorDriver testHypervisorDriver = {
.domainCheckpointGetParent = testDomainCheckpointGetParent, /* 5.6.0 */
.domainCheckpointDelete = testDomainCheckpointDelete, /* 5.6.0 */
.domainGetMessages = testDomainGetMessages, /* 7.6.0 */
+ .connectGetDomainCapabilities = testConnectGetDomainCapabilities, /* 9.8.0 */
};
static virNetworkDriver testNetworkDriver = {
--
2.41.0
1 year, 2 months
[libvirt PATCH] qemu_nbdkit: fix possible null dereference
by Pavel Hrdina
Function virGetConnectSecret() can return NULL so we need to check it
since in virSecretGetSecretString() it gets dereferenced.
Reported-by: coverity
Signed-off-by: Pavel Hrdina <phrdina(a)redhat.com>
---
src/qemu/qemu_nbdkit.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/qemu/qemu_nbdkit.c b/src/qemu/qemu_nbdkit.c
index 66b09cd240..17819ca992 100644
--- a/src/qemu/qemu_nbdkit.c
+++ b/src/qemu/qemu_nbdkit.c
@@ -971,6 +971,9 @@ qemuNbdkitProcessBuildCommandAuth(virStorageAuthDef *authdef,
}
conn = virGetConnectSecret();
+ if (!conn)
+ return -1;
+
if (virSecretGetSecretString(conn,
&authdef->seclookupdef,
secrettype,
--
2.41.0
1 year, 2 months
[PATCH 0/4] virerror: Make virReportEnumRangeError() check for type mismatch
by Michal Privoznik
There are few cases where virReportEnumRangeError() is passed to a
mismatching enum. With patch 4/4 we can check for this at compile time
(assuming -Wenum-compare is available).
For instance with clang I get:
FAILED: src/conf/libvirt_conf.a.p/domain_validate.c.o
clang ...
../src/conf/domain_validate.c:184:9: error: comparison of different enumeration types ('virDomainVideoBackendType' and 'virDomainInputType') [-Werror,-Wenum-compare]
virReportEnumRangeError(virDomainInputType, video->backend);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/util/virerror.h:175:47: note: expanded from macro 'virReportEnumRangeError'
(__typeof__(value))1 == (typname)1 && sizeof((typname)1) != 0 ? #typname : #typname)
~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~
Whereas with gcc I get more obscure error:
FAILED: src/conf/libvirt_conf.a.p/domain_validate.c.o
cc ...
In file included from ../src/util/virthread.h:25,
from ../src/util/virobject.h:25,
from ../src/conf/capabilities.h:28,
from ../src/conf/domain_conf.h:31,
from ../src/conf/domain_validate.h:25,
from ../src/conf/domain_validate.c:23:
../src/conf/domain_validate.c: In function ‘virDomainVideoDefValidate’:
../src/util/virerror.h:175:47: error: comparison between ‘enum <anonymous>’ and ‘enum <anonymous>’ [-Werror=enum-compare]
175 | (__typeof__(value))1 == (typname)1 && sizeof((typname)1) != 0 ? #typname : #typname)
| ^~
../src/conf/domain_validate.c:184:9: note: in expansion of macro ‘virReportEnumRangeError’
184 | virReportEnumRangeError(virDomainInputType, video->backend);
| ^~~~~~~~~~~~~~~~~~~~~~~
But I guess this is also okay-ish - it reports there is a problem with
code.
NB this only works with variables that are of a certain enum types. It
does not work with those generic 'int var /* someEnum */' declarations,
understandably. But as we use virXMLPropEnum() more, we can declare
variables of their true type (enum) and get more and more calls to
virReportEnumRangeError() checked.
Michal Prívozník (4):
virnetdevvportprofile: Turn virNetDevVPortProfileLinkOp enum into a
proper typedef
virNetDevVPortProfileOp8021Qbh: Use proper type in
virReportEnumRangeError()
virDomainVideoDefValidate: Use proper type in
virReportEnumRangeError()
virerror: Make virReportEnumRangeError() check for type mismatch
src/conf/domain_validate.c | 2 +-
src/util/virerror.h | 14 ++++++++------
src/util/virnetdevvportprofile.c | 10 +++++-----
3 files changed, 14 insertions(+), 12 deletions(-)
--
2.41.0
1 year, 2 months
[PATCH] libxl: Fix Domain-0 ballooning logic
by Jim Fehlig
When Domain-0 autoballooning is enabled, it's possible that memory may
need to be ballooned down in Domain-0 to accommodate the needs of another
virtual machine. libxlDomainFreeMemory handles this task, but due to a
logic bug is underflowing the variable containing Domain-0 new
target memory. The resulting huge numbers are filtered by
libxlSetMemoryTargetWrapper and memory is not changed.
Under the covers, libxlDomainFreeMemory uses Xen's libxl_set_memory_target
API, which includes a 'relative' parameter for specifying how to set the
target. If true, the target is an increment/decrement value over the
current memory, otherwise target is taken as an absolute value.
libxlDomainFreeMemory sets 'relative' to true, but never allows for
negative values by declaring the target memory variable as an unsigned.
Fix by declaring the variable as signed, which also requried adjusting
libxlSetMemoryTargetWrapper.
Signed-off-by: Jim Fehlig <jfehlig(a)suse.com>
---
src/libxl/libxl_api_wrapper.h | 16 ++++++----------
src/libxl/libxl_domain.c | 2 +-
2 files changed, 7 insertions(+), 11 deletions(-)
diff --git a/src/libxl/libxl_api_wrapper.h b/src/libxl/libxl_api_wrapper.h
index a9627f7983..c9582294cc 100644
--- a/src/libxl/libxl_api_wrapper.h
+++ b/src/libxl/libxl_api_wrapper.h
@@ -193,24 +193,20 @@ libxlSendTriggerWrapper(libxl_ctx *ctx,
static inline int
libxlSetMemoryTargetWrapper(libxl_ctx *ctx,
uint32_t domid,
- uint64_t target_memkb,
+ int64_t target_memkb,
int relative,
int enforce)
{
int ret = -1;
- /* Technically this guard could be LIBXL_HAVE_MEMKB_64BITS */
-#if LIBXL_API_VERSION < 0x040800
- if (target_memkb < UINT_MAX) {
- uint32_t val32 = target_memkb;
+#ifdef LIBXL_HAVE_MEMKB_64BITS
+ ret = libxl_set_memory_target(ctx, domid, target_memkb, relative, enforce);
+#else
+ if (target_memkb < INT_MAX) {
+ int32_t val32 = target_memkb;
ret = libxl_set_memory_target(ctx, domid, val32, relative, enforce);
}
-#else
- if (target_memkb < LLONG_MAX) {
- int64_t val64 = target_memkb;
- ret = libxl_set_memory_target(ctx, domid, val64, relative, enforce);
- }
#endif
return ret;
diff --git a/src/libxl/libxl_domain.c b/src/libxl/libxl_domain.c
index 6c167df63e..0c4beffd6a 100644
--- a/src/libxl/libxl_domain.c
+++ b/src/libxl/libxl_domain.c
@@ -926,7 +926,7 @@ libxlDomainFreeMem(libxl_ctx *ctx, libxl_domain_config *d_config)
{
uint64_t needed_mem;
uint64_t free_mem;
- uint64_t target_mem;
+ int64_t target_mem;
int tries = 3;
int wait_secs = 10;
--
2.42.0
1 year, 2 months
Share qemuInterfacexxxxConnect methods with ch
by Praveen Paladugu
Folks,
I am working on upstreaming network support for ch driver. Like qemu driver,ch
driver invokes steps in qemuInterfaceEthernetConnect,
qemuInterfaceBridgeConnect methods to connect tap devices to appropriate
host backends.
Current implementation clones aboves methods to ch_interface files and
uses them. I'd like to drop driver specific args from above methods and
move them to a common place so that qemu and ch drivers can share above
methods.
int
virInterfaceEthernetConnect(virDomainDef *def,
virDomainNetDef *net,
ebtablesContext *ebtables,
bool macFilter,
bool privileged,
int *tapfd,
size_t tapfdSize)
I started with `qemuInterfaceEthernetConnect` and modified the signature as
shown above. I initially tried putting `virInterfaceEthernetConnect` in
'src/util' but that caused a lot of cross inclusions. Above definition pulls in
domain_conf.h, domain_nwfilter.h, domain_audit.h and more headers, into utils
which causes cross inclusion.
I later created interface/interface_connect.{c,h} files and moved the methods to
these files. This would require driver code to include
interface/interface_connect.h, which seemed better than above cross inclusion
scenario.
Do you see any issues with moving above methods to interface/interface_connect.x
files? Any other ideas on how to reorg above methods so that they can be shared
by qemu and ch drivers?
Praveen
1 year, 2 months
[PATCH 00/16] nwfilter: Add support for user defined metadata
by K Shiva Kiran
This patchset adds support for the following user defined metadata
fields for network filters.
- title: A short description of the filter.
- description: Any documentation that the user wants to store.
- metadata: Other metadata in XML form.
K Shiva Kiran (16):
xml: Add <title>, <description> and <metadata> to nwfilter xml schema
conf: Add parser logic for nwfilter metadata fields
nwfilter: Add enum to operate on user defined metadata
nwfilter: Add error code and message for missing metadata
nwfilter: Introduce public API to modify user metadata
nwfilter: Introduce public API to retrieve user-defined metadata
nwfilter: Implement RPC
virsh: Add helper method to retrieve xml from NWFilter def
virsh: Add new command `nwfilter-desc`
virsh: Add new command `nwfilter-metadata`
virsh: Add option --title for nwfilter-list
docs: Document nwfilter metadata related commands
virnwfilterobj: Add virNWFilterObjGetMetadata()
virnwfilterobj: Add virNWFilterObjSetMetadata()
nwfilter_driver: Add Driver implementation for metadata
NEWS: Introduce user-defined metadata fields for NWFilter object
NEWS.rst | 18 ++
docs/formatnwfilter.rst | 31 +++
docs/manpages/virsh.rst | 98 +++++++-
include/libvirt/libvirt-nwfilter.h | 27 ++
include/libvirt/virterror.h | 1 +
src/conf/nwfilter_conf.c | 30 +++
src/conf/nwfilter_conf.h | 5 +
src/conf/schemas/nwfilter.rng | 9 +
src/conf/virnwfilterobj.c | 148 +++++++++++
src/conf/virnwfilterobj.h | 13 +
src/driver-nwfilter.h | 15 ++
src/libvirt-nwfilter.c | 154 ++++++++++++
src/libvirt_private.syms | 2 +
src/libvirt_public.syms | 6 +
src/nwfilter/nwfilter_driver.c | 61 +++++
src/remote/remote_driver.c | 2 +
src/remote/remote_protocol.x | 34 ++-
src/remote_protocol-structs | 19 ++
src/util/virerror.c | 3 +
tools/virsh-nwfilter.c | 387 ++++++++++++++++++++++++++++-
tools/virsh-util.c | 25 ++
tools/virsh-util.h | 9 +
22 files changed, 1089 insertions(+), 8 deletions(-)
--
2.42.0
1 year, 2 months
[libvirt PATCH v3 00/12] Extract the integration job commands to a shell script
by Erik Skultety
v1 here: https://listman.redhat.com/archives/libvir-list/2023-September/242121.html
Since v1:
- introduced a new run_cmd_quiet helper function instead of using eval to deal
with Shell redirections
- added more info from the POSIX standard on test's -a,-o options
- added commit references to the vendor locking of Avocado to v98.0 in the past
Erik Skultety (12):
syntax-check: Drop the shell's 'check for minus' rule
ci: integration: Extract the integration CI main recipe to jobs.sh
ci: integration: Adjust the check for CentOS Stream version
ci: integration: Drop the 'install-deps' hidden job and reference
ci: jobs.sh: Drop comment about the need for Avocado 98.0
ci: jobs.sh: integration: Use --quiet with virsh
ci: jobs.sh: run_integration: Add/Rewrite/Reformat commentaries
ci: jobs.sh: run_integration: Make POSIX-compliant
ci: jobs.sh: Introduce a quiet version of run_cmd
ci: jobs.sh: integration: Execute commands via 'run_cmd[_quiet]'
helpers
ci: jobs.sh: run_integration: Print DAEMONS variable for debugging
ci: jobs.sh: Define and create SCRATCH_DIR for local executions
build-aux/syntax-check.mk | 9 -------
ci/integration-template.yml | 42 +++--------------------------
ci/jobs.sh | 54 +++++++++++++++++++++++++++++++++++++
3 files changed, 57 insertions(+), 48 deletions(-)
--
2.41.0
1 year, 2 months
[PATCH] fix uint64 underflow
by Dmitry Frolov
According to previous statement,
'free_mem' is less than 'needed_mem'.
So, the subtraction 'free_mem - needed_mem' is negative,
and will raise uint64 underflow.
Signed-off-by: Dmitry Frolov <frolov(a)swemel.ru>
---
src/libxl/libxl_domain.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/libxl/libxl_domain.c b/src/libxl/libxl_domain.c
index 6c167df63e..36be042971 100644
--- a/src/libxl/libxl_domain.c
+++ b/src/libxl/libxl_domain.c
@@ -940,7 +940,7 @@ libxlDomainFreeMem(libxl_ctx *ctx, libxl_domain_config *d_config)
if (free_mem >= needed_mem)
return 0;
- target_mem = free_mem - needed_mem;
+ target_mem = needed_mem - free_mem;
if (libxlSetMemoryTargetWrapper(ctx, 0, target_mem,
/* relative */ 1, 0) < 0)
goto error;
--
2.34.1
1 year, 2 months
[PATCH] src: Avoid needless checks before calling g_strdup()
by Michal Privoznik
There are few places where the following pattern occurs:
if (var)
other = g_strdup(var);
where @other wasn't initialized before g_strdup(). Checking for
var != NULL is useless in this case, as that's exactly what
g_strdup() does (in which case it returns NULL).
Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
---
src/conf/domain_event.c | 3 +--
src/libxl/libxl_conf.c | 6 ++----
src/lxc/lxc_native.c | 3 +--
src/qemu/qemu_monitor_json.c | 3 +--
src/util/virconf.c | 3 +--
5 files changed, 6 insertions(+), 12 deletions(-)
diff --git a/src/conf/domain_event.c b/src/conf/domain_event.c
index 75603a933a..09f3368064 100644
--- a/src/conf/domain_event.c
+++ b/src/conf/domain_event.c
@@ -1561,8 +1561,7 @@ virDomainEventMetadataChangeNew(int id,
return NULL;
ev->type = type;
- if (nsuri)
- ev->nsuri = g_strdup(nsuri);
+ ev->nsuri = g_strdup(nsuri);
return (virObjectEvent *)ev;
}
diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
index 3ffb46fddd..4582126d19 100644
--- a/src/libxl/libxl_conf.c
+++ b/src/libxl/libxl_conf.c
@@ -583,8 +583,7 @@ libxlMakeDomBuildInfo(virDomainDef *def,
#endif
/* copy SLIC table path to acpi_firmware */
- if (def->os.slic_table)
- b_info->u.hvm.acpi_firmware = g_strdup(def->os.slic_table);
+ b_info->u.hvm.acpi_firmware = g_strdup(def->os.slic_table);
if (def->nsounds > 0) {
/*
@@ -1198,8 +1197,7 @@ libxlMakeDisk(virDomainDiskDef *l_disk, libxl_device_disk *x_disk)
return -1;
}
- if (l_disk->domain_name)
- x_disk->backend_domname = g_strdup(l_disk->domain_name);
+ x_disk->backend_domname = g_strdup(l_disk->domain_name);
return 0;
}
diff --git a/src/lxc/lxc_native.c b/src/lxc/lxc_native.c
index 5a63efc35f..c0011e0600 100644
--- a/src/lxc/lxc_native.c
+++ b/src/lxc/lxc_native.c
@@ -67,8 +67,7 @@ lxcCreateFSDef(int type,
def->type = type;
def->accessmode = VIR_DOMAIN_FS_ACCESSMODE_PASSTHROUGH;
- if (src)
- def->src->path = g_strdup(src);
+ def->src->path = g_strdup(src);
def->dst = g_strdup(dst);
def->readonly = readonly;
def->usage = usage;
diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c
index 137cb4e293..8152eea9a0 100644
--- a/src/qemu/qemu_monitor_json.c
+++ b/src/qemu/qemu_monitor_json.c
@@ -2960,8 +2960,7 @@ qemuMonitorJSONGetMigrationStatsReply(virJSONValue *reply,
case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
if (error) {
tmp = virJSONValueObjectGetString(ret, "error-desc");
- if (tmp)
- *error = g_strdup(tmp);
+ *error = g_strdup(tmp);
}
break;
diff --git a/src/util/virconf.c b/src/util/virconf.c
index 934632a35f..8fdf40e9d0 100644
--- a/src/util/virconf.c
+++ b/src/util/virconf.c
@@ -939,8 +939,7 @@ int virConfGetValueStringList(virConf *conf,
case VIR_CONF_STRING:
if (compatString) {
*values = g_new0(char *, cval->str ? 2 : 1);
- if (cval->str)
- (*values)[0] = g_strdup(cval->str);
+ (*values)[0] = g_strdup(cval->str);
break;
}
G_GNUC_FALLTHROUGH;
--
2.41.0
1 year, 2 months
[libvirt PATCH 00/12] Extract the integration job commands to a shell script
by Erik Skultety
Now that we have our base GitLab jobs extracted to ci/jobs.sh file, let's stay
consistent and do the same for the core integration tests job template.
Technically a v2 of:
https://listman.redhat.com/archives/libvir-list/2023-January/237201.html
Despite the above, quite a few things have changed, so I'm proposing this as a
new thing.
Here are test pipelines (both integration as well as normal one) after the
change:
https://gitlab.com/eskultety/libvirt/-/pipelines/1008044353
https://gitlab.com/eskultety/libvirt/-/pipelines/1008098378
Erik Skultety (12):
syntax-check: Drop the shell's 'check for minus' rule
ci: integration: Extract the integration CI main recipe to jobs.sh
ci: integration: Adjust the check for CentOS Stream version
ci: integration: Drop the 'install-deps' hidden job and reference
ci: jobs.sh: Drop comment about the need for Avocado 98.0
ci: jobs.sh: integration: Use --quiet with virsh
ci: jobs.sh: run_integration: Add/Rewrite/Reformat commentaries
ci: jobs: run_integration: Make POSIX-compliant
ci: jobs.sh: run_cmd: Use eval to run commands
ci: jobs: integration: Execute raw commands via 'run_cmd' helper
ci: jobs: run_integration: Print DAEMONS variable for debugging
ci: jobs.sh: Define and create SCRATCH_DIR for local executions
build-aux/syntax-check.mk | 9 -------
ci/integration-template.yml | 42 +++--------------------------
ci/jobs.sh | 53 +++++++++++++++++++++++++++++++++++--
3 files changed, 54 insertions(+), 50 deletions(-)
--
2.41.0
1 year, 2 months
