[libvirt] [PATCH v5 00/36] BaselineHypervisorCPU using QEMU QMP exchanges
by Chris Venteicher
Some architectures (S390) depend on QEMU to compute baseline CPU model and
expand a models feature set.
Interacting with QEMU requires starting the QEMU process and completing one or
more query-cpu-model-baseline QMP exchanges with QEMU in addition to a
query-cpu-model-expansion QMP exchange to expand all features in the model.
See "s390x CPU models: exposing features" patch set on Qemu-devel for discussion
of QEMU aspects.
This is part of resolution of: https://bugzilla.redhat.com/show_bug.cgi?id=1511999
-----
This patch set fixes all process code issues identified here:
https://www.redhat.com/archives/libvir-list/2018-November/msg00349.html
in patches 1-22 of the series.
The remaining patches implement the BaselineHypervisorCPU changes using
the non-domain qemu process code.
The process changes (patches 1-22)...
- Make the process code generic (not capabilities specific) for use by
BaselineHypervisorCPU
- Many of the process patches are simple code moves with implementation
changes in other distinct patches
- A thread safe library function creates a unique directory under libDir for each QEMU
process (for QMP messaging) to decouple processes in terms of sockets and
file system footprint.
The BaselineHypervisorCPU changes (patches 22-36)...
- Fix all issues raised in patch sets 1-4.
Thanks,
Chris
Chris Venteicher (36):
qemu_process: Move process code from qemu_capabilities to qemu_process
qemu_process: Use qemuProcessQmp prefix
qemu_process: Limit qemuProcessQmpNew to const input strings
qemu_process: Refer to proc not cmd in process code
qemu_process: Use consistent name for stop process function
qemu_capabilities: Stop QEMU process before freeing
qemu_process: Use qemuProcessQmp struct for a single process
qemu_process: All ProcessQMP errors are fatal
qemu_process: Persist stderr in qemuProcessQmp struct
qemu_process: Introduce qemuProcessQmpStart
qemu_process: Collect monitor code in single function
qemu_process: Store libDir in qemuProcessQmp struct
qemu_process: Setup paths within qemuProcessQmpInit
qemu_process: Stop retaining Monitor config in qemuProcessQmp
qemu_process: Don't open monitor if process failed
qemu_process: Cleanup qemuProcessQmp alloc function
qemu_process: Cleanup qemuProcessQmpStop function
qemu_process: Catch process free before process stop
qemu_monitor: Make monitor callbacks optional
qemu_process: Enter QMP command mode when starting QEMU Process
qemu_process: Use unique directories for QMP processes
qemu_process: Stop locking QMP process monitor immediately
qemu_monitor: Introduce qemuMonitorCPUModelInfoNew
qemu_monitor: Introduce qemuMonitorCPUModelInfo / JSON conversion
qemu_capabilities: Introduce virQEMuCapsMigratablePropsDiff
qemu_monitor: qemuMonitorGetCPUModelExpansion inputs and outputs
CPUModelInfo
qemu_capabilities: Introduce CPUModelInfo to virCPUDef function
qemu_capabilities: Introduce virCPUDef to CPUModelInfo function
qemu_monitor: Support query-cpu-model-baseline QMP command
qemu_driver: Consolidate code to baseline using libvirt
qemu_driver: Decouple code for baseline using libvirt
qemu_driver: Identify using libvirt as a distinct way to compute
baseline
qemu_driver: Support baseline calculation using QEMU
qemu_driver: Support feature expansion via QEMU when baselining cpu
qemu_driver: Remove unsupported props in expanded hypervisor baseline
output
qemu_monitor: Default props to migratable when expanding cpu model
src/qemu/qemu_capabilities.c | 617 ++++++++----------
src/qemu/qemu_capabilities.h | 4 +
src/qemu/qemu_driver.c | 216 +++++-
src/qemu/qemu_monitor.c | 184 +++++-
src/qemu/qemu_monitor.h | 29 +-
src/qemu/qemu_monitor_json.c | 226 +++++--
src/qemu/qemu_monitor_json.h | 12 +-
src/qemu/qemu_process.c | 356 ++++++++++
src/qemu/qemu_process.h | 32 +
tests/cputest.c | 11 +-
.../caps_2.10.0.s390x.xml | 60 +-
.../caps_2.11.0.s390x.xml | 58 +-
.../caps_2.12.0.s390x.xml | 56 +-
.../qemucapabilitiesdata/caps_2.8.0.s390x.xml | 32 +-
.../qemucapabilitiesdata/caps_2.9.0.s390x.xml | 34 +-
.../qemucapabilitiesdata/caps_3.0.0.s390x.xml | 64 +-
tests/qemucapabilitiestest.c | 7 +
17 files changed, 1396 insertions(+), 602 deletions(-)
--
2.17.1
6 years, 3 months
[libvirt] [PATCH 0/4] Fix check for accept_ra when starting an IPv6 network
by Laine Stump
These patches fix the crash described in:
https://bugzilla.redhat.com/1583131
They also fix a deficiency revealed by that crash - we weren't
checking the accept_ra value of interfaces in the "nexthop" elements
of multipath RA routes.
Laine Stump (4):
util: remove const specifier from nlmsghdr arg to
virNetlinkDumpCallback()
util: add a function to insert new interfaces to IPv6CheckForwarding
list
util: use nlmsg_find_attr() instead of an open-coded loop
util: check accept_ra for all nexthop interfaces of multipath routes
src/util/virnetdevip.c | 115 ++++++++++++++++++++++++++++-------------
src/util/virnetlink.h | 2 +-
2 files changed, 79 insertions(+), 38 deletions(-)
--
2.20.1
6 years, 3 months
[libvirt] [PATCH 0/2] Fix a couple of SCSI hostdev issues
by John Ferlan
See patches for details
John Ferlan (2):
qemu: Filter non SCSI hostdevs in qemuHostdevPrepareSCSIDevices
qemu: Remove virHostdevIsSCSIDevice from qemuIsSharedHostdev
src/qemu/qemu_conf.c | 3 +--
src/qemu/qemu_hostdev.c | 3 +++
2 files changed, 4 insertions(+), 2 deletions(-)
--
2.20.1
6 years, 3 months
[libvirt] [PATCH] src: Don't use double-colon rules
by Andrea Bolognani
According to the GNU Make manual, "double-colon rules are
somewhat obscure and not often very useful". Looking at
the few instances we have in libvirt, that certainly seems
to be the case, so just drop them.
Signed-off-by: Andrea Bolognani <abologna(a)redhat.com>
---
src/Makefile.am | 6 +++---
src/remote/Makefile.inc.am | 4 ++--
2 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/src/Makefile.am b/src/Makefile.am
index cd386297ed..8c8dfe3dcf 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -816,14 +816,14 @@ if WITH_LIBVIRTD
if LIBVIRT_INIT_SCRIPT_RED_HAT
initdir = $(sysconfdir)/rc.d/init.d
-install-init:: $(SYSVINIT_FILES) install-sysconfig
+install-init: $(SYSVINIT_FILES) install-sysconfig
$(MKDIR_P) $(DESTDIR)$(initdir)
for f in $(SYSVINIT_FILES:%.init=%) ; \
do \
$(INSTALL_SCRIPT) $$f.init $(DESTDIR)$(initdir)/$$f; \
done
-uninstall-init:: uninstall-sysconfig
+uninstall-init: uninstall-sysconfig
rm -f $(SYSVINIT_FILES:%.init=$(DESTDIR)$(initdir)/%)
rmdir $(DESTDIR)$(initdir) || :
@@ -1010,7 +1010,7 @@ install-data-local: $(INSTALL_DATA_LOCAL) \
$(MKDIR_P) "$(DESTDIR)$(localstatedir)/lib/libvirt/filesystems"
$(MKDIR_P) "$(DESTDIR)$(localstatedir)/lib/libvirt/boot"
-uninstall-local:: $(UNINSTALL_LOCAL) \
+uninstall-local: $(UNINSTALL_LOCAL) \
$(INSTALL_DATA_DIRS:%=uninstall-data-%)
rmdir "$(DESTDIR)$(localstatedir)/cache/libvirt" ||:
rmdir "$(DESTDIR)$(localstatedir)/lib/libvirt/images" ||:
diff --git a/src/remote/Makefile.inc.am b/src/remote/Makefile.inc.am
index 0988435eed..3d0ff29548 100644
--- a/src/remote/Makefile.inc.am
+++ b/src/remote/Makefile.inc.am
@@ -213,7 +213,7 @@ polkitdir = $(datadir)/polkit-1
polkitactionsdir = $(polkitdir)/actions
polkitrulesdir = $(polkitdir)/rules.d
-install-polkit::
+install-polkit:
$(MKDIR_P) $(DESTDIR)$(polkitactionsdir)
$(INSTALL_DATA) $(srcdir)/remote/libvirtd.policy \
$(DESTDIR)$(polkitactionsdir)/org.libvirt.unix.policy
@@ -221,7 +221,7 @@ install-polkit::
$(INSTALL_DATA) $(srcdir)/remote/libvirtd.rules \
$(DESTDIR)$(polkitrulesdir)/50-libvirt.rules
-uninstall-polkit::
+uninstall-polkit:
rm -f $(DESTDIR)$(polkitactionsdir)/org.libvirt.unix.policy
rmdir $(DESTDIR)$(polkitactionsdir) || :
rm -f $(DESTDIR)$(polkitrulesdir)/50-libvirt.rules
--
2.20.1
6 years, 3 months
[libvirt] [PATCH] util: Remove unnecessary ATTRIBUTE_NONNULL for virCommandAddArg[Pair]
by John Ferlan
Commit 912c6b22 modified the virCommandAddArg and virCommandAddArgPair
to perform NULL argument checking; however, no corresponding change
to the prototypes was made to remove the ATTRIBUTE_NONNULL, so the
Coverity build failed. Adjust the prototypes accordingly.
Signed-off-by: John Ferlan <jferlan(a)redhat.com>
---
It's a build breaker for Coverity or anything that enables STATIC_ANALYSIS
so I'll push under that rule.
src/util/vircommand.h | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/src/util/vircommand.h b/src/util/vircommand.h
index dbf5041890..100f7a06e0 100644
--- a/src/util/vircommand.h
+++ b/src/util/vircommand.h
@@ -123,7 +123,7 @@ void virCommandAddEnvPassAllowSUID(virCommandPtr cmd,
void virCommandAddEnvPassCommon(virCommandPtr cmd);
void virCommandAddArg(virCommandPtr cmd,
- const char *val) ATTRIBUTE_NONNULL(2);
+ const char *val);
void virCommandAddArgBuffer(virCommandPtr cmd,
virBufferPtr buf);
@@ -134,8 +134,7 @@ void virCommandAddArgFormat(virCommandPtr cmd,
void virCommandAddArgPair(virCommandPtr cmd,
const char *name,
- const char *val)
- ATTRIBUTE_NONNULL(2) ATTRIBUTE_NONNULL(3);
+ const char *val);
void virCommandAddArgSet(virCommandPtr cmd,
const char *const*vals) ATTRIBUTE_NONNULL(2);
--
2.17.2
6 years, 4 months
[libvirt] [PATCH] qemu: require reply from guest agent in qemuAgentGetInterfaces
by Ján Tomko
Since its introduction in commit 0977b8aa071 (released in v1.2.14)
qemuAgentGetInterfaces calls qemuAgentCommand with needReply=false,
which allows qemuAgentCommand to return 0 even when it did not get
any reply from the agent.
Set needReply to true, since we dereference it right after.
This can be hit if libvirt is waiting for an event from the agent
(e.g. shutdown) and the agent cannot reply in time (e.g. due to
the guest being shut down), as reported in:
https://bugzilla.redhat.com/show_bug.cgi?id=1663051
Signed-off-by: Ján Tomko <jtomko(a)redhat.com>
---
src/qemu/qemu_agent.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/qemu/qemu_agent.c b/src/qemu/qemu_agent.c
index 0f4dae37e3..80b789b5e0 100644
--- a/src/qemu/qemu_agent.c
+++ b/src/qemu/qemu_agent.c
@@ -2043,7 +2043,7 @@ qemuAgentGetInterfaces(qemuAgentPtr mon,
if (!(cmd = qemuAgentMakeCommand("guest-network-get-interfaces", NULL)))
goto cleanup;
- if (qemuAgentCommand(mon, cmd, &reply, false,
+ if (qemuAgentCommand(mon, cmd, &reply, true,
VIR_DOMAIN_QEMU_AGENT_COMMAND_BLOCK) < 0)
goto cleanup;
--
2.20.1
6 years, 4 months
[libvirt] [PATCH] cpu_map: Add support for arch-capabilities feature
by Jiri Denemark
The feature was added to QEMU in 3.1.0 and it is currently blocking
migration, which is expected to change in the future. Luckily 3.1.0 is
new enough to give us migratability hints on each feature via
query-cpu-model-expension, which means we don't need to use the
"migratable" attribute on the CPU map XML.
The kernel calls this feature arch_capabilities and RHEL/CentOS 7.* use
arch-facilities. Apparently some CPU test files were gathered with the
RHEL version of QEMU. Let's update the test files to avoid possible
confusion about the correct naming.
Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com>
---
src/cpu_map/x86_features.xml | 3 +++
tests/cputestdata/x86_64-cpuid-EPYC-7601-32-Core-ibpb.json | 2 +-
tests/cputestdata/x86_64-cpuid-Xeon-E5-2609-v3.json | 2 +-
tests/cputestdata/x86_64-cpuid-Xeon-E5-2623-v4.json | 2 +-
tests/cputestdata/x86_64-cpuid-Xeon-Gold-5115.json | 2 +-
5 files changed, 7 insertions(+), 4 deletions(-)
diff --git a/src/cpu_map/x86_features.xml b/src/cpu_map/x86_features.xml
index 59236456d6..02431bea29 100644
--- a/src/cpu_map/x86_features.xml
+++ b/src/cpu_map/x86_features.xml
@@ -326,6 +326,9 @@
<feature name='stibp'>
<cpuid eax_in='0x07' ecx_in='0x00' edx='0x08000000'/>
</feature>
+ <feature name='arch-capabilities'> <!-- arch_capabilities, arch-facilities -->
+ <cpuid eax_in='0x07' ecx_in='0x00' edx='0x20000000'/>
+ </feature>
<feature name='ssbd'>
<cpuid eax_in='0x07' ecx_in='0x00' edx='0x80000000'/>
</feature>
diff --git a/tests/cputestdata/x86_64-cpuid-EPYC-7601-32-Core-ibpb.json b/tests/cputestdata/x86_64-cpuid-EPYC-7601-32-Core-ibpb.json
index 94a60fcc8f..1f53bb8bf3 100644
--- a/tests/cputestdata/x86_64-cpuid-EPYC-7601-32-Core-ibpb.json
+++ b/tests/cputestdata/x86_64-cpuid-EPYC-7601-32-Core-ibpb.json
@@ -232,7 +232,7 @@
"avx512vbmi": false,
"kvm-asyncpf": true,
"spec-ctrl": false,
- "arch-facilities": false,
+ "arch-capabilities": false,
"model": 1,
"node-id": -1
}
diff --git a/tests/cputestdata/x86_64-cpuid-Xeon-E5-2609-v3.json b/tests/cputestdata/x86_64-cpuid-Xeon-E5-2609-v3.json
index 10c5434263..6bdaf6e83a 100644
--- a/tests/cputestdata/x86_64-cpuid-Xeon-E5-2609-v3.json
+++ b/tests/cputestdata/x86_64-cpuid-Xeon-E5-2609-v3.json
@@ -232,7 +232,7 @@
"avx512vbmi": false,
"kvm-asyncpf": true,
"spec-ctrl": true,
- "arch-facilities": false,
+ "arch-capabilities": false,
"model": 63,
"node-id": -1
}
diff --git a/tests/cputestdata/x86_64-cpuid-Xeon-E5-2623-v4.json b/tests/cputestdata/x86_64-cpuid-Xeon-E5-2623-v4.json
index 0506dec0a7..2c6be20768 100644
--- a/tests/cputestdata/x86_64-cpuid-Xeon-E5-2623-v4.json
+++ b/tests/cputestdata/x86_64-cpuid-Xeon-E5-2623-v4.json
@@ -232,7 +232,7 @@
"avx512vbmi": false,
"kvm-asyncpf": true,
"spec-ctrl": true,
- "arch-facilities": false,
+ "arch-capabilities": false,
"model": 79,
"node-id": -1
}
diff --git a/tests/cputestdata/x86_64-cpuid-Xeon-Gold-5115.json b/tests/cputestdata/x86_64-cpuid-Xeon-Gold-5115.json
index 79f3580219..79b47a56fb 100644
--- a/tests/cputestdata/x86_64-cpuid-Xeon-Gold-5115.json
+++ b/tests/cputestdata/x86_64-cpuid-Xeon-Gold-5115.json
@@ -232,7 +232,7 @@
"avx512vbmi": false,
"kvm-asyncpf": true,
"spec-ctrl": true,
- "arch-facilities": false,
+ "arch-capabilities": false,
"model": 85,
"node-id": -1
}
--
2.20.1
6 years, 4 months
[libvirt] [PATCH] qemu: Don't enable seclabel remembering for session mode
by Michal Privoznik
The session daemon is unable to set XATTRs in 'trusted'
namespace because it doesn't run as privileged process.
Therefore, when creating the default qemu config enable
rememberOwner only when running as privileged process.
Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
---
src/qemu/qemu_conf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c
index 3718ca6c22..20952e9607 100644
--- a/src/qemu/qemu_conf.c
+++ b/src/qemu/qemu_conf.c
@@ -145,7 +145,7 @@ virQEMUDriverConfigPtr virQEMUDriverConfigNew(bool privileged)
cfg->group = (gid_t)-1;
}
cfg->dynamicOwnership = privileged;
- cfg->rememberOwner = true;
+ cfg->rememberOwner = privileged;
cfg->cgroupControllers = -1; /* -1 == auto-detect */
--
2.19.2
6 years, 4 months
[libvirt] [PATCH v2 0/3] Better syntax-check on BSD
by Eric Blake
Since v1: fix the gnulib bug I accidentally introduced, then
copy the gist of Roman's gnulib changes to also apply to our
cfg.mk syntax checks.
Eric Blake (3):
maint: update gnulib for syntax-check on BSD
maint: prefer $(GREP) in cfg.mk
maint: split long lines for BSD syntax-check
.gnulib | 2 +-
cfg.mk | 95 +++++++++++++++++++++++++++++----------------------------
2 files changed, 49 insertions(+), 48 deletions(-)
--
2.20.1
6 years, 4 months
