December 2012 - Devel - libvirt List Archives

[libvirt] [PATCH] Fix arch datatype in vahControl in virt-aa-helper.c
by Daniel P. Berrange 19 Dec '12

19 Dec '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> When changing to virArch, the virt-aa-helper.c file was not completely changed. The vahControl struct was left with a char *arch field, instead of virArch arch field. Pushed as a build breaker fix --- src/security/virt-aa-helper.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/security/virt-aa-helper.c b/src/security/virt-aa-helper.c index c07e69d..16ce7f3 100644 --- a/src/security/virt-aa-helper.c +++ b/src/security/virt-aa-helper.c @@ -72,7 +72,7 @@ typedef struct { virDomainDefPtr def; /* VM definition */ virCapsPtr caps; /* VM capabilities */ char *hvm; /* type of hypervisor (eg hvm, xen) */ - char *arch; /* machine architecture */ + virArch arch; /* machine architecture */ char *newfile; /* newly added file */ bool append; /* append to .files instead of rewrite */ } vahControl; @@ -87,7 +87,6 @@ vahDeinit(vahControl * ctl) virCapabilitiesFree(ctl->caps); VIR_FREE(ctl->files); VIR_FREE(ctl->hvm); - VIR_FREE(ctl->arch); VIR_FREE(ctl->newfile); return 0; -- 1.7.11.7

1 0

[libvirt] [PATCH] Change string form of VIR_ARCH_ITANIUM back to ia64
by Daniel P. Berrange 19 Dec '12

19 Dec '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> Historically there was an inconsistency in handling of the itanium arch. The xen driver & CPU model code treated it as 'ia64' but the QEMU capabilities code used 'itanium'. On the grounds that no one has ever seriously used itanium with QEMU, while RHEL shipped itanium with Xen, we should favour 'ia64' as the canonical format Pushed as a test breaker fix from virArch fallout --- src/qemu/qemu_capabilities.c | 4 ---- src/util/virarch.c | 4 +--- 2 files changed, 1 insertion(+), 7 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index ae75bc6..50d999a 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -251,8 +251,6 @@ VIR_ONCE_GLOBAL_INIT(qemuCaps) static virArch qemuCapsArchFromString(const char *arch) { - if (STREQ(arch, "ia64")) - return VIR_ARCH_ITANIUM; if (STREQ(arch, "i386")) return VIR_ARCH_I686; if (STREQ(arch, "arm")) @@ -266,8 +264,6 @@ static const char *qemuCapsArchToString(virArch arch) { if (arch == VIR_ARCH_I686) return "i386"; - else if (arch == VIR_ARCH_ITANIUM) - return "ia64"; else if (arch == VIR_ARCH_ARMV7L) return "arm"; diff --git a/src/util/virarch.c b/src/util/virarch.c index 5458b74..1376887 100644 --- a/src/util/virarch.c +++ b/src/util/virarch.c @@ -39,7 +39,7 @@ static const struct virArchData { { "cris", 32, VIR_ARCH_LITTLE_ENDIAN }, { "i686", 32, VIR_ARCH_LITTLE_ENDIAN }, - { "itanium", 64, VIR_ARCH_LITTLE_ENDIAN }, + { "ia64", 64, VIR_ARCH_LITTLE_ENDIAN }, { "lm32", 32, VIR_ARCH_BIG_ENDIAN }, { "m68k", 32, VIR_ARCH_BIG_ENDIAN }, { "microblaze", 32, VIR_ARCH_BIG_ENDIAN }, @@ -159,8 +159,6 @@ virArch virArchFromHost(void) ut.machine[3] == '6' && ut.machine[4] == '\0') { arch = VIR_ARCH_I686; - } else if (STREQ(ut.machine, "ia64")) { - arch = VIR_ARCH_ITANIUM; } else if (STREQ(ut.machine, "amd64")) { arch = VIR_ARCH_X86_64; } else { -- 1.7.11.7

1 0

[libvirt] (no subject)
by Daniel P. Berrange 19 Dec '12

19 Dec '12

o

1 1

[libvirt] [PATCH] Fix CPU baseline to not hardcode arch
by Daniel P. Berrange 19 Dec '12

19 Dec '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> Prior to the virArch changes, the CPU baseline method would free the arch string in the returned CPU. Fix the regression by setting arch to VIR_ARCH_NONE at the end Pushed as a test breaker fix from virArch fallout Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/cpu/cpu_x86.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/cpu/cpu_x86.c b/src/cpu/cpu_x86.c index 2001b99..cb21910 100644 --- a/src/cpu/cpu_x86.c +++ b/src/cpu/cpu_x86.c @@ -1715,6 +1715,8 @@ x86Baseline(virCPUDefPtr *cpus, if (!outputVendor) VIR_FREE(cpu->vendor); + cpu->arch = VIR_ARCH_NONE; + cleanup: x86ModelFree(base_model); x86MapFree(map); -- 1.7.11.7

1 0

[libvirt] [PATCH] Fix SELinux security label test
by Daniel P. Berrange 19 Dec '12

19 Dec '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> If securityselinuxtest was run on a system with newer SELinux policy it would fail, due to using svirt_tcg_t instead of svirt_t. Fixing the domain type to be KVM avoids this issue. Pushed as a build fix --- tests/securityselinuxtest.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/securityselinuxtest.c b/tests/securityselinuxtest.c index 8bcf3a1..0fa70ba 100644 --- a/tests/securityselinuxtest.c +++ b/tests/securityselinuxtest.c @@ -77,6 +77,7 @@ testBuildDomainDef(bool dynamic, if (VIR_ALLOC(secdef) < 0) goto no_memory; + def->virtType = VIR_DOMAIN_VIRT_KVM; def->seclabels[0] = secdef; def->seclabels[0]->type = dynamic ? VIR_DOMAIN_SECLABEL_DYNAMIC : VIR_DOMAIN_SECLABEL_STATIC; -- 1.7.11.7

1 0

[libvirt] [PATCH 0/8 v7] Unprivileged SG_IO support
by Osier Yang 18 Dec '12

18 Dec '12

As a result of RFC [1], this implements the unprivleged SG_IO support. v6 - v7: * No restoring of unpriv_sgio per Daniel's thought. * Use "major:minor" as the hash key per Jirka's suggestion. Osier Yang (8): util: Prepare helpers for unpriv_sgio setting qemu: Add a hash table for the shared disks docs: Add docs and rng schema for new XML cdbfilter conf: Parse and format the new XML tag cdbfilter qemu: set unpriv_sgio when domain starting qemu: Check if the shared disk's cdbfilter conflicts with others qemu: Set unpriv_sgio when attaching disk qemu: Error out if the shared disk conf conflicts with others when attaching docs/formatdomain.html.in | 13 ++- docs/schemas/domaincommon.rng | 52 ++++-- src/conf/domain_conf.c | 58 +++++-- src/conf/domain_conf.h | 10 + src/libvirt_private.syms | 4 + src/qemu/qemu_conf.c | 92 ++++++++++ src/qemu/qemu_conf.h | 12 ++ src/qemu/qemu_driver.c | 38 ++++ src/qemu/qemu_process.c | 75 ++++++++ src/qemu/qemu_process.h | 2 + src/util/util.c | 180 ++++++++++++++++++++ src/util/util.h | 8 + ...ml2argv-disk-scsi-lun-passthrough-cdbfilter.xml | 32 ++++ tests/qemuxml2xmltest.c | 1 + 14 files changed, 547 insertions(+), 30 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-scsi-lun-passthrough-cdbfilter.xml [1] https://www.redhat.com/archives/libvir-list/2012-November/msg00988.html [2] https://www.redhat.com/archives/libvir-list/2012-December/msg00325.html Regards, Osier

3 35

[libvirt] VirtualBox 4.2 Drivers Complete?
by Arslan Waqar 18 Dec '12

18 Dec '12

I was using a VM on Vbox (Ubuntu 12.04) and from the virsh I tried: virsh # dommemstat 2 and I got the following errors: error: Failed to get memory statistics for domain 2 error: this function is not supported by the connection driver: virDomainMemoryStats Are the Drivers for Vbox 4.2 incomplete? -- Regards, Arslan Waqar.

2 1

[libvirt] [PATCH 0/6] Use enum for tracking valid architectures
by Daniel P. Berrange 18 Dec '12

18 Dec '12

Different operating systems have different names for architectures. eg x86_64 x64, amd64. This can cause inconsistency in libvirt code dealing with architectures. To deal with this define an enum for valid architectures and a canonical string mapping. Update all code internally to use the enum instead of 'char *' for arch data.

5 21

[libvirt] [PATCH] spec: Do not install *.py[co] in python examples
by Jiri Denemark 18 Dec '12

18 Dec '12

Unfortunately, rpm is stupid enough to bytycompile python scripts even though they are located in /usr/share/doc/libvirt-python-*/examples and it does so after %install phase is finished. Thus there's no way we could remove those files from BUILDROOT. As a workaround, we may safely remove the examples subdirectory completely without losing anything. The python scripts that were installed there are also copied directly into /usr/share/doc/libvirt-python-*/ by %doc python/tests/*.py rule. And yes, the files are actually tests, not examples. --- libvirt.spec.in | 2 ++ 1 file changed, 2 insertions(+) diff --git a/libvirt.spec.in b/libvirt.spec.in index 7457a48..13cc8a2 100644 --- a/libvirt.spec.in +++ b/libvirt.spec.in @@ -1392,6 +1392,8 @@ rm -f $RPM_BUILD_ROOT%{_datadir}/augeas/lenses/tests/test_libvirtd_lxc.aug %if ! %{with_python} rm -rf $RPM_BUILD_ROOT%{_datadir}/doc/libvirt-python-%{version} +%else +rm -rf $RPM_BUILD_ROOT%{_datadir}/doc/libvirt-python-%{version}/examples %endif %if ! %{with_qemu} -- 1.8.0.2

2 2

[libvirt] [PATCH 00/28] Rename files in src/util to have a 'vir' prefix
by Daniel P. Berrange 18 Dec '12

18 Dec '12

Currently the files in src/util are 50/50 split between those having a 'vir' prefix and those which don't. Finish the job by remaining all remaining files to have a 'vir' prefix too, guaranteeing that they can't clash with system headers[1]. The only case which was not a plain rename, was merging processinfo.{c,h} into the virprocess.{c,h} since the functionality logically belongs together. At the end src/Makefile.am is updated to sort util files alphabetically. Left as further work - make sure all functions in these files also have a 'vir' prefix. eg dnsmasq, ebtables, iptables, usb and pci files are violators in this regard. Also left, much of virutil.c should move into virfile.c Daniel [1] Not a theoretical problem - we already hit a clsah on 'usb.h' before forcing us to use 'hostusb.h' and 'memory.h' clashes with a system header but we never noticed so far.

3 38