August 2018 - Devel - libvirt List Archives

[libvirt] [PATCH for v4.6.0 0/2] Fix some random problems
by Michal Privoznik 02 Aug '18

02 Aug '18

Bad pun. But while debugging the issue Bjoern raised [1] I've noticed these problems. https://www.redhat.com/archives/libvir-list/2018-July/msg02101.html Michal Prívozník (2): util: Don't overflow in virRandomBits viriscsi: Request more random bits for interface name src/util/viriscsi.c | 2 +- src/util/virrandom.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) -- 2.16.4

5 15

[libvirt] [jenkins-ci PATCH 0/8] lcitool: Cleanups and improvements
by Andrea Bolognani 01 Aug '18

01 Aug '18

This series makes the guests/ directory less crowded by moving a bunch of files to subdirectories, enhances the lcitool script so that it works when called from any directory, and introduces the ability to skip installation of build dependencies for projects you don't care about. Some of these changes pave the way for introducing [secret feature redacted] later on. Andrea Bolognani (8): lcitool: Make the script location-independent lcitool: Use JSON to pass extra-vars lcitool: Pass base and playbook_base to Ansible guests: Move update playbook and related resources guests: Drop install_config from inventory guests: Move install configs lcitool: Allow installing a subset of build dependencies guests: Update documentation guests/README.markdown | 38 ++++++-- guests/{ => configs}/kickstart.cfg | 0 guests/{ => configs}/preseed.cfg | 0 guests/host_vars/libvirt-centos-7/install.yml | 1 - guests/host_vars/libvirt-debian-8/install.yml | 1 - guests/host_vars/libvirt-debian-9/install.yml | 1 - .../host_vars/libvirt-debian-sid/install.yml | 1 - .../host_vars/libvirt-fedora-27/install.yml | 1 - .../host_vars/libvirt-fedora-28/install.yml | 1 - .../libvirt-fedora-rawhide/install.yml | 1 - .../host_vars/libvirt-ubuntu-16/install.yml | 1 - .../host_vars/libvirt-ubuntu-18/install.yml | 1 - guests/lcitool | 96 ++++++++++++++----- .../{site.yml => playbooks/update/main.yml} | 30 +++--- guests/{ => playbooks/update}/tasks/base.yml | 6 +- .../update}/tasks/bootloader.yml | 0 .../update}/tasks/bootstrap.yml | 0 .../{ => playbooks/update}/tasks/jenkins.yml | 4 +- .../{ => playbooks/update}/tasks/kludges.yml | 0 .../{ => playbooks/update}/tasks/packages.yml | 2 +- guests/{ => playbooks/update}/tasks/paths.yml | 0 .../{ => playbooks/update}/tasks/services.yml | 0 guests/{ => playbooks/update}/tasks/users.yml | 4 +- .../update}/templates/bash_profile.j2 | 0 .../update}/templates/bashrc.j2 | 0 .../update}/templates/ccache.conf.j2 | 0 .../fedora-rawhide-kernel-nodebug.repo.j2 | 0 .../update}/templates/jenkins.service.j2 | 0 .../templates/jessie-backports.preferences.j2 | 0 .../templates/jessie-backports.sources.j2 | 0 30 files changed, 128 insertions(+), 61 deletions(-) rename guests/{ => configs}/kickstart.cfg (100%) rename guests/{ => configs}/preseed.cfg (100%) rename guests/{site.yml => playbooks/update/main.yml} (52%) rename guests/{ => playbooks/update}/tasks/base.yml (92%) rename guests/{ => playbooks/update}/tasks/bootloader.yml (100%) rename guests/{ => playbooks/update}/tasks/bootstrap.yml (100%) rename guests/{ => playbooks/update}/tasks/jenkins.yml (92%) rename guests/{ => playbooks/update}/tasks/kludges.yml (100%) rename guests/{ => playbooks/update}/tasks/packages.yml (96%) rename guests/{ => playbooks/update}/tasks/paths.yml (100%) rename guests/{ => playbooks/update}/tasks/services.yml (100%) rename guests/{ => playbooks/update}/tasks/users.yml (95%) rename guests/{ => playbooks/update}/templates/bash_profile.j2 (100%) rename guests/{ => playbooks/update}/templates/bashrc.j2 (100%) rename guests/{ => playbooks/update}/templates/ccache.conf.j2 (100%) rename guests/{ => playbooks/update}/templates/fedora-rawhide-kernel-nodebug.repo.j2 (100%) rename guests/{ => playbooks/update}/templates/jenkins.service.j2 (100%) rename guests/{ => playbooks/update}/templates/jessie-backports.preferences.j2 (100%) rename guests/{ => playbooks/update}/templates/jessie-backports.sources.j2 (100%) -- 2.17.1

2 17

[libvirt] [PATCH] tests: viriscsitest: make it work on big endian archs
by Pino Toscano 01 Aug '18

01 Aug '18

viriscsitest tries to ensure the interface IQN used is a specific one, checking later on that it is the same all during the whole test. Since the IQN generation involves random bytes, viriscsitest got a fake virRandomBytes from the virrandommock helper library, setting static values. virRandomBits(), called by virStorageBackendCreateIfaceIQN(), always requests 8 random bytes, chopping off the ones not requested by the caller -- this meant that on big endian machines it would chop bytes from the wrong size of the data buffer, and thus not returning the expected numbers. As a fix, do not rely on the mock virRandomBytes, but provide an own version of it: this version will fill the values in the expected order, depending on the endianness of the system. This way, the result of virStorageBackendCreateIfaceIQN() will be what the test actually expects. Signed-off-by: Pino Toscano <ptoscano(a)redhat.com> --- tests/viriscsitest.c | 30 ++++++++++++++++++++++++++++-- 1 file changed, 28 insertions(+), 2 deletions(-) diff --git a/tests/viriscsitest.c b/tests/viriscsitest.c index 4fd7ff6e19..d43c0115d3 100644 --- a/tests/viriscsitest.c +++ b/tests/viriscsitest.c @@ -21,6 +21,7 @@ #include <config.h> #include "testutils.h" +#include "virrandom.h" #ifdef WIN32 int @@ -36,6 +37,32 @@ main(void) # define VIR_FROM_THIS VIR_FROM_NONE +bool isLittleEndian(void); +bool isLittleEndian(void) +{ + static int cache = -1; + if (cache < 0) { + int val = 1; + char *ptr = (char *)&val; + cache = *ptr ? 1 : 0; + } + return cache > 0; +} + + +int +virRandomBytes(unsigned char *buf, + size_t buflen) +{ + size_t i; + const bool isLE = isLittleEndian(); + + for (i = 0; i < buflen; i++) + buf[i] = isLE ? i : buflen - i - 1; + + return 0; +} + static const char *iscsiadmSessionOutput = "tcp: [1] 10.20.30.40:3260,1 iqn.2004-06.example:example1:iscsi.test\n" "tcp: [2] 10.20.30.41:3260,1 iqn.2005-05.example:example1:iscsi.hello\n" @@ -368,6 +395,5 @@ mymain(void) return EXIT_SUCCESS; } -VIR_TEST_MAIN_PRELOAD(mymain, - abs_builddir "/.libs/virrandommock.so") +VIR_TEST_MAIN(mymain) #endif /* WIN32 */ -- 2.17.1

3 3

Re: [libvirt] [PATCH v2] conf: virDomainDefValidateInternal prohibit some characters in shmem name
by skobyda＠redhat.com 01 Aug '18

01 Aug '18

On Fri, 2018-07-27 at 13:19 -0400, John Ferlan wrote: > You shouldn't drop libvir-list when you reply... Sorry, I dropped it by mistake. > > On 07/27/2018 06:35 AM, skobyda(a)redhat.com wrote: > > On Mon, 2018-07-23 at 17:32 -0400, John Ferlan wrote: > > > > > > On 07/12/2018 09:10 AM, Simon Kobyda wrote: > > > > XML shmem name will not include character '/', and will not be > > > > equal to strings > > > > "." or "..", as shmem name is used in a path. > > > > > > Validate that the provided XML shmem name is not directory > > > specific > > > "." > > > or ".." names as well as ensuring that there is no path separator > > > '/' > > > in > > > the name. > > > > https://bugzilla.redhat.com/show_bug.cgi?id=1192400 > > > > --- > > > > > > > > Changes in V2 > > > > - Added error reports > > > > - Error situation will happen only if shmem name is > > > > equal to > > > > "." or "..", however their occurence in a name > > > > compromised of > > > > more > > > > characters is allowed. > > > > > > > > src/conf/domain_conf.c | 22 ++++++++++++++++++++++ > > > > 1 file changed, 22 insertions(+) > > > > > > > > > > I believe this actually belongs in > > > virDomainDeviceDefValidateInternal > > > for case VIR_DOMAIN_DEVICE_SHMEM. > > > Also, should the docs/schemas/domaincommon.rng be modified? > > > Currently > > > it > > > has: > > > > > > <define name="shmem"> > > > <element name="shmem"> > > > <attribute name="name"> > > > <data type="string"> > > > <param name="pattern">[^/]*</param> > > > </data> > > > > > > > Is it a good idea? To create such regular expression, I believe it > > would make it unreadable for another person, ergo useless. I don't > > want > > it to end up as IPv6. I've benn told that actual parser can be, and > > sometimes is stricter than scheme. > > > > Well this is what I was hoping others would be able to chime in on. > Hence why you ask on list. I don't disagree that being stricter in > Parse > is fine. Doing regexes is like reading a foreign language to me at > times. Some just don't make sense. Yeah, that's why I don't want to try to insert complicated regex in docs/schemas/domaincommon.rng. > > I think that regex is indicating - any character except '/', but I'm > not > sure. If it is, it's ironic that we have to check for '/' > specifically > since it shouldn't be passing the xml validation test - OK at least > if > one was editing via virsh. > > John > You are right about what that regex does. But from what I heard, that functionality is rather new to libvirt, so users have option to bypass that verification. So I would like to also leave it in code in case users try to bypass that regex verification. Simon. > > > Consider how other names are limited in their scope. The > > > basictypes.rng > > > has a number of examples. > > > > > > Naturally, the problem with changing it is that someone somewhere > > > will > > > complain, but libvirt used to accept this other format. Right now > > > I > > > would think the scope a bit too broad. > > > > But then we cannot fix most of the bugs, since there could always > > be > > some user which relies on bug behavior. > > > > > > If we are to limit the name we should also document in > > > docs/formatdomain.html.in that the shmem name is "limited" in > > > name to > > > avoid the '/' character, ".", and "..". > > > > > > BTW: My regex isn't that good, but it would seem '/' is an > > > invalid > > > character by XML standards even though the code never checked for > > > it. > > > Using virt-xml-validate <file> <schema> would "validate" whether > > > someone > > > provides valid XML. > > > > > > > > > John > > > > > > > diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c > > > > index 7ab2953d83..6b34c17de4 100644 > > > > --- a/src/conf/domain_conf.c > > > > +++ b/src/conf/domain_conf.c > > > > @@ -6107,6 +6107,8 @@ virDomainDefLifecycleActionValidate(const > > > > virDomainDef *def) > > > > static int > > > > virDomainDefValidateInternal(const virDomainDef *def) > > > > { > > > > + size_t i; > > > > + > > > > if (virDomainDefCheckDuplicateDiskInfo(def) < 0) > > > > return -1; > > > > > > > > @@ -6136,6 +6138,26 @@ virDomainDefValidateInternal(const > > > > virDomainDef *def) > > > > return -1; > > > > } > > > > > > > > + for (i = 0; i < def->nshmems; i++) { > > > > + if (strchr(def->shmems[i]->name, '/')) { > > > > + virReportError(VIR_ERR_XML_ERROR, "%s", > > > > + _("shmem name cannot include '/' > > > > character")); > > > > + return -1; > > > > + } > > > > + > > > > + if (STREQ(def->shmems[i]->name, ".")) { > > > > + virReportError(VIR_ERR_XML_ERROR, "%s", > > > > + _("shmem name cannot be equal to > > > > '.'")); > > > > + return -1; > > > > + } > > > > + > > > > + if (STREQ(def->shmems[i]->name, "..")) { > > > > + virReportError(VIR_ERR_XML_ERROR, "%s", > > > > + _("shmem name cannot be equal to > > > > '..'")); > > > > + return -1; > > > > + } > > > > + } > > > > + > > > > if (virDomainDefLifecycleActionValidate(def) < 0) > > > > return -1; > > > > > > > >

2 1

[libvirt] [PATCH for 4.6.0] tests: qemucapsprobe: Fix output after swithching to jansson
by Peter Krempa 01 Aug '18

01 Aug '18

Jansson does not put a newline at the end of formatted JSON strings. This breaks the qemucapsprobe utility as we need to keep the spacing so that tests work. Add an explicit newline. Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- tests/qemucapsprobemock.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/qemucapsprobemock.c b/tests/qemucapsprobemock.c index 5936975505..049b16273a 100644 --- a/tests/qemucapsprobemock.c +++ b/tests/qemucapsprobemock.c @@ -76,6 +76,7 @@ qemuMonitorSend(qemuMonitorPtr mon, printLineSkipEmpty("\n", stdout); printLineSkipEmpty(reformatted, stdout); + printLineSkipEmpty("\n", stdout); VIR_FREE(reformatted); return realQemuMonitorSend(mon, msg); @@ -116,6 +117,7 @@ qemuMonitorJSONIOProcessLine(qemuMonitorPtr mon, printLineSkipEmpty("\n", stdout); printLineSkipEmpty(json, stdout); + printLineSkipEmpty("\n", stdout); } cleanup: -- 2.16.2

4 5

[libvirt] [dockerfiles PATCH] Update OS when building images
by Andrea Bolognani 01 Aug '18

01 Aug '18

The corresponding libvirt-jenkins-ci commit is dcbad35f45fe. Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- Pushed under the Dockerfile refresh rule. buildenv-centos-7.Dockerfile | 3 ++- buildenv-debian-8.Dockerfile | 1 + buildenv-debian-9.Dockerfile | 1 + buildenv-debian-sid.Dockerfile | 1 + buildenv-fedora-27.Dockerfile | 3 ++- buildenv-fedora-28.Dockerfile | 3 ++- buildenv-fedora-rawhide.Dockerfile | 3 ++- buildenv-ubuntu-16.Dockerfile | 1 + buildenv-ubuntu-18.Dockerfile | 1 + 9 files changed, 13 insertions(+), 4 deletions(-) diff --git a/buildenv-centos-7.Dockerfile b/buildenv-centos-7.Dockerfile index 2c4c26d..896bf68 100644 --- a/buildenv-centos-7.Dockerfile +++ b/buildenv-centos-7.Dockerfile @@ -65,6 +65,7 @@ ENV PACKAGES audit-libs-devel \ sudo \ systemtap-sdt-devel \ vim -RUN yum install -y ${PACKAGES} && \ +RUN yum update -y && \ + yum install -y ${PACKAGES} && \ yum autoremove -y && \ yum clean all -y diff --git a/buildenv-debian-8.Dockerfile b/buildenv-debian-8.Dockerfile index 8a8c3b0..8877528 100644 --- a/buildenv-debian-8.Dockerfile +++ b/buildenv-debian-8.Dockerfile @@ -71,6 +71,7 @@ ENV PACKAGES augeas-tools \ xsltproc \ zfs-fuse RUN apt-get update && \ + apt-get dist-upgrade -y && \ apt-get install -y ${PACKAGES} && \ apt-get autoremove -y && \ apt-get autoclean -y diff --git a/buildenv-debian-9.Dockerfile b/buildenv-debian-9.Dockerfile index c3af421..0592c97 100644 --- a/buildenv-debian-9.Dockerfile +++ b/buildenv-debian-9.Dockerfile @@ -73,6 +73,7 @@ ENV PACKAGES augeas-tools \ xsltproc \ zfs-fuse RUN apt-get update && \ + apt-get dist-upgrade -y && \ apt-get install -y ${PACKAGES} && \ apt-get autoremove -y && \ apt-get autoclean -y diff --git a/buildenv-debian-sid.Dockerfile b/buildenv-debian-sid.Dockerfile index a922457..2d0310d 100644 --- a/buildenv-debian-sid.Dockerfile +++ b/buildenv-debian-sid.Dockerfile @@ -73,6 +73,7 @@ ENV PACKAGES augeas-tools \ xsltproc \ zfs-fuse RUN apt-get update && \ + apt-get dist-upgrade -y && \ apt-get install -y ${PACKAGES} && \ apt-get autoremove -y && \ apt-get autoclean -y diff --git a/buildenv-fedora-27.Dockerfile b/buildenv-fedora-27.Dockerfile index 2f35a2e..09ec721 100644 --- a/buildenv-fedora-27.Dockerfile +++ b/buildenv-fedora-27.Dockerfile @@ -73,6 +73,7 @@ ENV PACKAGES audit-libs-devel \ wireshark-devel \ xen-devel \ zfs-fuse -RUN yum install -y ${PACKAGES} && \ +RUN yum update -y && \ + yum install -y ${PACKAGES} && \ yum autoremove -y && \ yum clean all -y diff --git a/buildenv-fedora-28.Dockerfile b/buildenv-fedora-28.Dockerfile index f4f7198..60afba5 100644 --- a/buildenv-fedora-28.Dockerfile +++ b/buildenv-fedora-28.Dockerfile @@ -73,6 +73,7 @@ ENV PACKAGES audit-libs-devel \ wireshark-devel \ xen-devel \ zfs-fuse -RUN yum install -y ${PACKAGES} && \ +RUN yum update -y && \ + yum install -y ${PACKAGES} && \ yum autoremove -y && \ yum clean all -y diff --git a/buildenv-fedora-rawhide.Dockerfile b/buildenv-fedora-rawhide.Dockerfile index 189ea59..33b47e1 100644 --- a/buildenv-fedora-rawhide.Dockerfile +++ b/buildenv-fedora-rawhide.Dockerfile @@ -97,6 +97,7 @@ ENV PACKAGES audit-libs-devel \ wireshark-devel \ xen-devel \ zfs-fuse -RUN yum install -y ${PACKAGES} && \ +RUN yum update -y && \ + yum install -y ${PACKAGES} && \ yum autoremove -y && \ yum clean all -y diff --git a/buildenv-ubuntu-16.Dockerfile b/buildenv-ubuntu-16.Dockerfile index bc387a1..c904c6e 100644 --- a/buildenv-ubuntu-16.Dockerfile +++ b/buildenv-ubuntu-16.Dockerfile @@ -74,6 +74,7 @@ ENV PACKAGES augeas-tools \ xsltproc \ zfs-fuse RUN apt-get update && \ + apt-get dist-upgrade -y && \ apt-get install -y ${PACKAGES} && \ apt-get autoremove -y && \ apt-get autoclean -y diff --git a/buildenv-ubuntu-18.Dockerfile b/buildenv-ubuntu-18.Dockerfile index 2a15f14..bbd32c1 100644 --- a/buildenv-ubuntu-18.Dockerfile +++ b/buildenv-ubuntu-18.Dockerfile @@ -74,6 +74,7 @@ ENV PACKAGES augeas-tools \ xsltproc \ zfs-fuse RUN apt-get update && \ + apt-get dist-upgrade -y && \ apt-get install -y ${PACKAGES} && \ apt-get autoremove -y && \ apt-get autoclean -y -- 2.17.1

1 0

[libvirt] [jenkins-ci PATCH] lcitool: Update OS when building Docker images
by Andrea Bolognani 01 Aug '18

01 Aug '18

The last build of the Fedora Rawhide Docker image failed with Transaction check error: file /usr/lib64/libgdbm_compat.so.4.0.0 from install of gdbm-libs-1:1.16-1.fc29.x86_64 conflicts with file from package gdbm-1:1.14.1-3.fc28.x86_64 caused by the gdbm package having recently been split. Especially with fast-moving distribution such as Fedora Rawhide and Debian Sid, we can't rely on the base OS not changing over time, so merely installing the packages we need on top of that is not enough. Additionally, updating the OS brings the Docker images closer to the guests running on CentOS CI, which are updated daily. Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- guests/lcitool | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/guests/lcitool b/guests/lcitool index 22b08dd..33c1afe 100755 --- a/guests/lcitool +++ b/guests/lcitool @@ -478,13 +478,15 @@ class Application: if package_format == "deb": sys.stdout.write(textwrap.dedent(""" RUN apt-get update && \\ + apt-get dist-upgrade -y && \\ apt-get install -y ${PACKAGES} && \\ apt-get autoremove -y && \\ apt-get autoclean -y """)) elif package_format == "rpm": sys.stdout.write(textwrap.dedent(""" - RUN yum install -y ${PACKAGES} && \\ + RUN yum update -y && \\ + yum install -y ${PACKAGES} && \\ yum autoremove -y && \\ yum clean all -y """)) -- 2.17.1

3 3

[libvirt] [dbus PATCH] Install daemon under @sbindir@
by Andrea Bolognani 01 Aug '18

01 Aug '18

The libvirt-dbus daemon is not supposed to be invoked explicitly by the user, but rather to be spawned on-demand by the D-Bus daemon: as such, @sbindir@ is a more suitable location in which to install it. Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- data/Makefile.am | 4 ++-- data/session/org.libvirt.service.in | 2 +- data/system/org.libvirt.service.in | 2 +- libvirt-dbus.spec.in | 2 +- src/Makefile.am | 2 +- 5 files changed, 6 insertions(+), 6 deletions(-) diff --git a/data/Makefile.am b/data/Makefile.am index 7b523da..660a100 100644 --- a/data/Makefile.am +++ b/data/Makefile.am @@ -62,12 +62,12 @@ CLEANFILES = \ session/org.libvirt.service: session/org.libvirt.service.in $(AM_V_GEN)$(MKDIR_P) session && \ - sed -e 's|[@]bindir[@]|$(bindir)|g' < $< > $@-t && \ + sed -e 's|[@]sbindir[@]|$(sbindir)|g' < $< > $@-t && \ mv $@-t $@ system/org.libvirt.service: system/org.libvirt.service.in $(AM_V_GEN)$(MKDIR_P) system && \ - sed -e 's|[@]bindir[@]|$(bindir)|g' \ + sed -e 's|[@]sbindir[@]|$(sbindir)|g' \ -e 's|[@]SYSTEM_USER[@]|$(SYSTEM_USER)|' \ < $< > $@-t && mv $@-t $@ diff --git a/data/session/org.libvirt.service.in b/data/session/org.libvirt.service.in index a8cb6a9..8a56b95 100644 --- a/data/session/org.libvirt.service.in +++ b/data/session/org.libvirt.service.in @@ -1,3 +1,3 @@ [D-BUS Service] Name=org.libvirt -Exec=@bindir@/libvirt-dbus --session +Exec=@sbindir@/libvirt-dbus --session diff --git a/data/system/org.libvirt.service.in b/data/system/org.libvirt.service.in index 0d3abdd..781a503 100644 --- a/data/system/org.libvirt.service.in +++ b/data/system/org.libvirt.service.in @@ -1,4 +1,4 @@ [D-BUS Service] Name=org.libvirt -Exec=@bindir@/libvirt-dbus --system +Exec=@sbindir@/libvirt-dbus --system User=@SYSTEM_USER@ diff --git a/libvirt-dbus.spec.in b/libvirt-dbus.spec.in index f99e041..626e2da 100644 --- a/libvirt-dbus.spec.in +++ b/libvirt-dbus.spec.in @@ -51,7 +51,7 @@ exit 0 %files %doc README.md HACKING.md AUTHORS NEWS %license COPYING -%{_bindir}/libvirt-dbus +%{_sbindir}/libvirt-dbus %{_datadir}/dbus-1/services/org.libvirt.service %{_datadir}/dbus-1/system-services/org.libvirt.service %{_datadir}/dbus-1/system.d/org.libvirt.conf diff --git a/src/Makefile.am b/src/Makefile.am index b5bf129..bc13feb 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -28,7 +28,7 @@ libutil_la_SOURCES = \ util.h \ $(NULL) -bin_PROGRAMS = \ +sbin_PROGRAMS = \ libvirt-dbus \ $(NULL) -- 2.17.1

3 3

[libvirt] [PATCH] qemu: qemuDomainChangeNet: don't overwrite device info when pci addr is missing
by Katerina Koukiou 01 Aug '18

01 Aug '18

When trying to update an interface's rom settings with an device XML that is missing the PCI addr element, all new rom settings where not applied. https://bugzilla.redhat.com/show_bug.cgi?id=1599513 Signed-off-by: Katerina Koukiou <kkoukiou(a)redhat.com> --- Not sure why we chose to overwrite the whole info before though, I hope that this doesn't cause side problems. src/qemu/qemu_hotplug.c | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 1488f0a7c2..f45192b1d3 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -3445,17 +3445,14 @@ qemuDomainChangeNet(virQEMUDriverPtr driver, goto cleanup; } - /* info: if newdev->info is empty, fill it in from olddev, - * otherwise verify that it matches - nothing is allowed to - * change. (There is no helper function to do this, so - * individually check the few feidls of virDomainDeviceInfo that - * are relevant in this case). + /* info: if newdev->info.addr.pci is empty, fill it in from olddev, + * otherwise verify that it matches. */ if (!virDomainDeviceAddressIsValid(&newdev->info, - VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) && - virDomainDeviceInfoCopy(&newdev->info, &olddev->info) < 0) { - goto cleanup; + VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) { + newdev->info.addr.pci = olddev->info.addr.pci; } + if (!virPCIDeviceAddressEqual(&olddev->info.addr.pci, &newdev->info.addr.pci)) { virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s", -- 2.17.1

2 4

[libvirt] [PATCH RFC 0/2] ESX: Fixing SetAutoStart
by Marcos Paulo de Souza 01 Aug '18

01 Aug '18

Hi guys, while doing some random tests using the ESX driver, I faced a crash, and these two patches are trying to solve. I don't know if there is a better way to solve it, since ESX driver *generated* code to alloc and free some *generated* structs, but, please let me know if you have a better idea. Thanks, Marcos Paulo de Souza (2): esx: Do not crash SetAutoStart by double free esx: Fix SetAutoStart invalid pointer free src/esx/esx_driver.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) -- 2.17.1

2 3