- Devel - libvirt List Archives

[libvirt] Libvirt and qemu multiqueue net
by Michal Privoznik 13 Mar '13

13 Mar '13

It's been a while since qemu added support for multiqueue virtio-net. For more info follow [1]. From usage POV, libvirt is required to pass multiple FDs: qemu-kvm -netdev tap,id=hn0,fds=x:y:...:z,vhost=on,vhostfds=a:b:...:c -device virtio-net-pci,netdev=hn0,vectors=N ... where count(x:...:z) = M. We are strongly advised to use 1:1 ratio => M=N. The FDs for 'fds' argument can be obtained via: fd1 = open("/dev/net/tun") ioctl(fd1, TUNSETIFF, "tap") fd2 = open("/dev/net/tun") ioctl(fd2, TUNATTACHQUEUE, "tap") And what I am wondering about is, how to expose this to XML. Or do we even want to expose this instead of some smart automatic setting? There are not any other configurable knobs right now, but they may come in the future. Michal 1: http://www.linux-kvm.org/page/Multiqueue

3 2

[libvirt] we can't build a new libvirt package (error 1)
by Erdembileg. G 13 Mar '13

13 Mar '13

Hi, I installed Xen 4.2.1-4.el6 on CentOS 6.4. kernel version is 3.8.2-3.el6xen.x86_64. Libvirt version is libvirt-0.10.2-18.el6 Before configure and installion are OK. But I can't build a new libvirt package. PASS: test_conf.sh PASS: cpuset PASS: define-dev-segfault PASS: int-overflow PASS: libvirtd-fail PASS: libvirtd-pool PASS: read-bufsiz PASS: read-non-seekable PASS: start PASS: vcpupin TEST: virsh-all ........................................ 40 ........................................ 80 ........................................ 120 ........................................ 160 ................................. 193 OK PASS: virsh-all PASS: virsh-optparse PASS: virsh-schedinfo PASS: virsh-synopsis PASS: virsh-undefine ======================================= 1 of 71 tests failed (1 test was not run) Please report to libvir-list(a)redhat.com ======================================= make[1]: *** [check-TESTS] Error 1 make[1]: Leaving directory `/root/rpmbuild/BUILD/libvirt-0.10.2/tests' make: *** [check-am] Error 2 error: Bad exit status from /var/tmp/rpm-tmp.rsnyPA (%check) RPM build errors: Bad exit status from /var/tmp/rpm-tmp.rsnyPA (%check) [root@xebin SPECS]# -- Best Regards, Erdembileg.G

2 1

[libvirt] [PATCH] audit: Audit operations done by using VirtIO RNG
by Peter Krempa 12 Mar '13

12 Mar '13

This patch adds auditing of resources used by the 'random' backend of virtio RNG. --- If there's desire to audit also use of the "egd" backend that uses a generic character device, a way how to audit this device will need to be introduced. We don't audit useage of chardevs right now. src/conf/domain_audit.c | 76 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 76 insertions(+) diff --git a/src/conf/domain_audit.c b/src/conf/domain_audit.c index 8cd522a..c80bdb4 100644 --- a/src/conf/domain_audit.c +++ b/src/conf/domain_audit.c @@ -100,6 +100,79 @@ cleanup: } +static void +virDomainAuditRNG(virDomainObjPtr vm, + virDomainRNGDefPtr newDef, virDomainRNGDefPtr oldDef, + const char *reason, bool success) +{ + char uuidstr[VIR_UUID_STRING_BUFLEN]; + char *vmname; + char *oldsrc = NULL; + char *newsrc = NULL; + const char *virt; + + virUUIDFormat(vm->def->uuid, uuidstr); + if (!(vmname = virAuditEncode("vm", vm->def->name))) + goto no_memory; + + if (!(virt = virDomainVirtTypeToString(vm->def->virtType))) { + VIR_WARN("Unexpected virt type %d while encoding audit message", vm->def->virtType); + virt = "?"; + } + + if (newDef) { + switch (newDef->backend) { + case VIR_DOMAIN_RNG_BACKEND_RANDOM: + if (!(newsrc = virAuditEncode("new-rng", VIR_AUDIT_STR(newDef->source.file)))) + goto no_memory; + break; + + case VIR_DOMAIN_RNG_BACKEND_EGD: + case VIR_DOMAIN_RNG_BACKEND_LAST: + if (!(newsrc = virAuditEncode("new-rng", "?"))) + goto no_memory; + break; + } + } else { + if (!(newsrc = virAuditEncode("new-rng", "?"))) + goto no_memory; + } + + if (oldDef) { + switch (oldDef->backend) { + case VIR_DOMAIN_RNG_BACKEND_RANDOM: + if (!(oldsrc = virAuditEncode("old-rng", VIR_AUDIT_STR(oldDef->source.file)))) + goto no_memory; + break; + + case VIR_DOMAIN_RNG_BACKEND_EGD: + case VIR_DOMAIN_RNG_BACKEND_LAST: + if (!(oldsrc = virAuditEncode("old-rng", "?"))) + goto no_memory; + break; + } + } else { + if (!(oldsrc = virAuditEncode("old-rng", "?"))) + goto no_memory; + } + + VIR_AUDIT(VIR_AUDIT_RECORD_RESOURCE, success, + "virt=%s resrc=rng reason=%s %s uuid=%s %s %s", + virt, reason, vmname, uuidstr, + oldsrc, newsrc); + +cleanup: + VIR_FREE(vmname); + VIR_FREE(oldsrc); + VIR_FREE(newsrc); + return; + +no_memory: + VIR_WARN("OOM while encoding audit message"); + goto cleanup; +} + + void virDomainAuditFS(virDomainObjPtr vm, virDomainFSDefPtr oldDef, virDomainFSDefPtr newDef, @@ -641,6 +714,9 @@ virDomainAuditStart(virDomainObjPtr vm, const char *reason, bool success) virDomainAuditRedirdev(vm, redirdev, "start", true); } + if (vm->def->rng) + virDomainAuditRNG(vm, vm->def->rng, NULL, "start", true); + virDomainAuditMemory(vm, 0, vm->def->mem.cur_balloon, "start", true); virDomainAuditVcpu(vm, 0, vm->def->vcpus, "start", true); -- 1.8.1.5

2 3

[libvirt] [PATCH v3 0/3] Keep original file label
by Michal Privoznik 12 Mar '13

12 Mar '13

Yet another rework of $subj. I am still not solving atomicity problem for now. See diff to the patches if you want to know what's changed. Michal Privoznik (3): virFile: Add APIs for extended attributes handling virfile: Introduce internal API for managing ACL security_dac: Favour ACLs over chown() configure.ac | 2 + libvirt.spec.in | 1 + m4/virt-acl.m4 | 9 ++ src/Makefile.am | 4 +- src/libvirt_private.syms | 6 + src/security/security_dac.c | 209 ++++++++++++++++++++++++++----- src/util/virfile.c | 290 ++++++++++++++++++++++++++++++++++++++++++++ src/util/virfile.h | 23 ++++ 8 files changed, 515 insertions(+), 29 deletions(-) create mode 100644 m4/virt-acl.m4 -- 1.8.1.5

2 7

[libvirt] [PATCH] Add missing break in LXC loop device setup
by Daniel P. Berrange 12 Mar '13

12 Mar '13

From: "Daniel P. Berrange" <berrange(a)redhat.com> When setting up disks with loop devices for LXC, one of the switch cases was missing a 'break' causing it to fallthrough to an error condition. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/lxc/lxc_controller.c | 1 + 1 file changed, 1 insertion(+) diff --git a/src/lxc/lxc_controller.c b/src/lxc/lxc_controller.c index ce46070..128271f 100644 --- a/src/lxc/lxc_controller.c +++ b/src/lxc/lxc_controller.c @@ -448,6 +448,7 @@ static int virLXCControllerSetupLoopDevices(virLXCControllerPtr ctrl) fd = virLXCControllerSetupLoopDeviceDisk(disk); if (fd < 0) goto cleanup; + break; default: virReportError(VIR_ERR_CONFIG_UNSUPPORTED, -- 1.8.1.4

2 1

[libvirt] using virSetUIDGID() with unprivileged qemu defeats setuid helper
by Csaba Henk 12 Mar '13

12 Mar '13

Hi, I recently experienced that my qemu guest (which I'm using with unprivileged user) fails to start with: error: internal error process exited while connecting to monitor: chardev: opening backend "pty" failed This happens upon trying to facilitate the <serial type='pty'> <target port='0'/> </serial> <console type='pty'> <target type='serial' port='0'/> </console> stanzas, for which qemu wants to grab a pty through openpty(3). openpty needs to have the assigned pty to be chown'd to the qemu user, which is attempted via running the setuid helper program pt_chown. However, chown(2) fails with EPERM. The culprit seems to be the commits v1.0.3-rc1~113: util: virSetUIDGIDWithCaps - change uid while keeping caps v1.0.3-rc1~112: util: maintain caps when running command with uid != 0 which change how capabilities are manipulated before program execution. Just immediately before the execve(2) call, the qemu process used to have the following capabilities: CapInh: 0000000000000000 CapPrm: 0000000000000000 CapEff: 0000000000000000 CapBnd: ffffffffffffffff since said commits, it looks like: CapInh: 0000000000000000 CapPrm: 0000000000000000 CapEff: 0000000000000000 CapBnd: ffffffe000000000 as far as my capability-noob eyes can see, the bounding set lacks CAP_CHOWN and thus pt_chown won't attain CAP_CHOWN despite running on uid 0, and the EPERM is triggered. How could we fix it? Qemu invocation should be customized or virExec() adjusted? Or is there some configuration workaround? (For the record, I've seen it on Arch Linux; tried their binary package and also my own builds, which included a current git checkout.) Thanks Csaba

3 2

[libvirt] [PATCH] Add support for disks backed by plain files in LXC
by Daniel P. Berrange 12 Mar '13

12 Mar '13

From: "Daniel P. Berrange" <berrange(a)redhat.com> By using a loopback device, disks backed by plain files can be made available to LXC containers. We make no attempt to auto-detect format if <driver type="raw"/> is not set, instead we unconditionally treat that as meaning raw. This is to avoid the security issues inherant with format auto-detection Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/lxc/lxc_controller.c | 67 ++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 65 insertions(+), 2 deletions(-) diff --git a/src/lxc/lxc_controller.c b/src/lxc/lxc_controller.c index a7e715e..176e1be 100644 --- a/src/lxc/lxc_controller.c +++ b/src/lxc/lxc_controller.c @@ -356,7 +356,7 @@ static int virLXCControllerValidateConsoles(virLXCControllerPtr ctrl) } -static int virLXCControllerSetupLoopDevice(virDomainFSDefPtr fs) +static int virLXCControllerSetupLoopDeviceFS(virDomainFSDefPtr fs) { int lofd; char *loname = NULL; @@ -377,6 +377,27 @@ static int virLXCControllerSetupLoopDevice(virDomainFSDefPtr fs) } +static int virLXCControllerSetupLoopDeviceDisk(virDomainDiskDefPtr disk) +{ + int lofd; + char *loname = NULL; + + if ((lofd = virFileLoopDeviceAssociate(disk->src, &loname)) < 0) + return -1; + + /* + * We now change it into a block device type, so that + * the rest of container setup 'just works' + */ + disk->type = VIR_DOMAIN_DISK_TYPE_BLOCK; + VIR_FREE(disk->src); + disk->src = loname; + loname = NULL; + + return lofd; +} + + static int virLXCControllerSetupLoopDevices(virLXCControllerPtr ctrl) { size_t i; @@ -389,7 +410,7 @@ static int virLXCControllerSetupLoopDevices(virLXCControllerPtr ctrl) if (fs->type != VIR_DOMAIN_FS_TYPE_FILE) continue; - fd = virLXCControllerSetupLoopDevice(fs); + fd = virLXCControllerSetupLoopDeviceFS(fs); if (fd < 0) goto cleanup; @@ -402,6 +423,48 @@ static int virLXCControllerSetupLoopDevices(virLXCControllerPtr ctrl) ctrl->loopDevFds[ctrl->nloopDevs - 1] = fd; } + for (i = 0 ; i < ctrl->def->ndisks ; i++) { + virDomainDiskDefPtr disk = ctrl->def->disks[i]; + int fd; + + if (disk->type != VIR_DOMAIN_DISK_TYPE_FILE) + continue; + + switch (disk->format) { + /* We treat 'none' as meaning 'raw' since we + * don't want to go into the auto-probing + * business for security reasons + */ + case VIR_STORAGE_FILE_RAW: + case VIR_STORAGE_FILE_NONE: + if (disk->driverName && + STRNEQ(disk->driverName, "loop")) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("disk driver %s is not supported"), + disk->driverName); + goto cleanup; + } + + fd = virLXCControllerSetupLoopDeviceDisk(disk); + if (fd < 0) + goto cleanup; + + default: + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, + _("disk format %s is not supported"), + virStorageFileFormatTypeToString(disk->format)); + goto cleanup; + } + + VIR_DEBUG("Saving loop fd %d", fd); + if (VIR_EXPAND_N(ctrl->loopDevFds, ctrl->nloopDevs, 1) < 0) { + VIR_FORCE_CLOSE(fd); + virReportOOMError(); + goto cleanup; + } + ctrl->loopDevFds[ctrl->nloopDevs - 1] = fd; + } + VIR_DEBUG("Setup all loop devices"); ret = 0; -- 1.7.11.7

3 3

[libvirt] [PATCH] util: escapes special characters in VIR_LOG_REGEX
by Guannan Ren 12 Mar '13

12 Mar '13

In debug mode, the bug failed to start vm error: Failed to start domain rhel5u9 error: internal error Out of space while reading console log output: ... --- src/util/virlog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/util/virlog.c b/src/util/virlog.c index 130bdff..957d993 100644 --- a/src/util/virlog.c +++ b/src/util/virlog.c @@ -80,7 +80,7 @@ static regex_t *virLogRegex = NULL; #define VIR_LOG_DATE_REGEX "[0-9][0-9][0-9][0-9]-[0-9][0-9]-[0-9][0-9]" -#define VIR_LOG_TIME_REGEX "[0-9][0-9]:[0-9][0-9]:[0-9][0-9].[0-9][0-9][0-9]+[0-9][0-9][0-9][0-9]" +#define VIR_LOG_TIME_REGEX "[0-9][0-9]:[0-9][0-9]:[0-9][0-9]\\.[0-9][0-9][0-9]\\+[0-9][0-9][0-9][0-9]" #define VIR_LOG_PID_REGEX "[0-9]+" #define VIR_LOG_LEVEL_REGEX "debug|info|warning|error" -- 1.7.11.2

3 4

[libvirt] [libvirt-sandbox][PATCH] docs: fix some typos in TODO
by Alex Jia 12 Mar '13

12 Mar '13

Signed-off-by: Alex Jia <ajia(a)redhat.com> --- TODO | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/TODO b/TODO index 1ac37a0..fc63361 100644 --- a/TODO +++ b/TODO @@ -12,10 +12,10 @@ If we run a syslog within the container will it get messages from the outside? EXECUTE: virt-sandbox-service execute --command "BLAH" does not work. We need to have the ability to execute any random command within the container, and get stdin, stdout, stderror outside the container. (Partially Completed) -Still needs kernel to implement missing container namespace files under /proc/PID/ns, Also need a mechansm to get the PID of systemd from libvirt. +Still needs kernel to implement missing container namespace files under /proc/PID/ns, Also need a mechanism to get the PID of systemd from libvirt. HOSTNAME: - Currently if I execute hostname within the conatiner it sees the name of the host not the name based on the container name or the IP Address associated with dhclient. (Completed) + Currently if I execute hostname within the container it sees the name of the host not the name based on the container name or the IP Address associated with dhclient. (Completed) virt-sandbox-service connect NAME hangs when you attempt to end the connection. ^d should bring you back to the host terminal. -- 1.7.1

2 1

[libvirt] [libvirt-sandbox][PATCH] docs: fix some typos in architecture.txt
by Alex Jia 12 Mar '13

12 Mar '13

Signed-off-by: Alex Jia <ajia(a)redhat.com> --- docs/architecture.txt | 16 ++++++++-------- 1 files changed, 8 insertions(+), 8 deletions(-) diff --git a/docs/architecture.txt b/docs/architecture.txt index 16bffad..c227cbd 100644 --- a/docs/architecture.txt +++ b/docs/architecture.txt @@ -17,7 +17,7 @@ Thus the foundation of the sandbox is the host filesystem passthrough capabiliti of the virtualization technology. There needs to be the ability to pass the entire root filesystem of the host through to the virtual domain in readonly mode. There are then zero or more additional host locations passed through in read-write mode, -to be mounted at the specific locations in the virtual donain to which the application +to be mounted at the specific locations in the virtual domain to which the application will require write access. The host and guest paths for the additional locations need not, and typically will not, be the same. @@ -136,7 +136,7 @@ second problem is that the primary console is typically operating in a non-raw mode initially, which means that any data sent from the host to the guest is automatically echoed back out. This is not desirable, since the application to be run needs to be in charge of -this. Thus it is often neccessary to put the guest console into raw +this. Thus it is often necessary to put the guest console into raw mode. Unfortunately with a virtual machine based sandbox, there is no way to tell the kernel to put its console in raw mode from the moment it boots. Thus it is critical to prevent the host from sending @@ -157,11 +157,11 @@ sequences to be sent. Any literal '\' in the stream is itself escaped as '\\'. Initially the host process starts off in receive mode only. ie it will -not attmpt to send any data to the virtual guest. +not attempt to send any data to the virtual guest. If the sandbox successfully starts the application, the magic byte sequence "xoqpuɐs" will be transmitted from the guest to the host. -This byte sequence is guarenteed to be the first data sent from the +This byte sequence is guaranteed to be the first data sent from the guest to the host in normal circumstances. Thus if the host process receives any other byte sequences it knows that sandbox startup has failed. In this case, further data received from the guest should @@ -181,14 +181,14 @@ Kernels and initrds For application sandboxes based on virtual machines, a kernel and initrd is required to boot the guest. The goal once again is to run -the same kernel in the guest, as curently runs on the host OS. The +the same kernel in the guest, as currently runs on the host OS. The initrd though, will typically need to be different, since at the time of writing all distro initrd's lack the ability to boot from a 9p based host filesystem. In addition startup performance of the virtual machine startup is absolutely critical. The hardware configured for the virtual machine -is well known ahead of time, thus a highly targetted initrd can be +is well known ahead of time, thus a highly targeted initrd can be built and all hardware probing can be avoided. In fact all that is required is an initrd containing a 9p module and the virtio-net modules (and their dependencies). @@ -227,14 +227,14 @@ inefficiency in libvirt probing QEMU command line arguments. The kernel command line is tuned in an attempt to minimize the time it spends initializing hardware - - loglevel=0 - to supress all extraneous kernel output on the primary + - loglevel=0 - to suppress all extraneous kernel output on the primary console which would get mixed up with application data - quiet - as above - edd=off - stop probing for EDD support which does not exist for QEMU - noreplace-smp - don't attempt to switch SMP alternatives, which wastes many cycles - pci=noearly - minimize time spent initializing the PCI bus - - cgroup_disable=memory - don't waste time on unsed subsystem + - cgroup_disable=memory - don't waste time on unused subsystem Still todo -- 1.7.1

2 1