- Devel - libvirt List Archives

[libvirt] [PATCH 1/3] qemu: hotplug: Only label hostdev after checking device conflicts
by Cole Robinson 10 Dec '13

10 Dec '13

Similar to what Jiri did for cgroup setup/teardown in 05e149f94, push it all into the device handler functions so we can do the necessary prep work before claiming the device. This also fixes hotplugging USB devices by product/vendor (virt-manager's default behavior): https://bugzilla.redhat.com/show_bug.cgi?id=1016511 --- src/qemu/qemu_hotplug.c | 42 ++++++++++++++++++++++++++++++++++-------- 1 file changed, 34 insertions(+), 8 deletions(-) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index f4fc723..bff9e23 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -1153,6 +1153,7 @@ qemuDomainAttachHostPciDevice(virQEMUDriverPtr driver, char *configfd_name = NULL; bool releaseaddr = false; bool teardowncgroup = false; + bool teardownlabel = false; int backend = hostdev->source.subsys.u.pci.backend; if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs + 1) < 0) @@ -1193,6 +1194,11 @@ qemuDomainAttachHostPciDevice(virQEMUDriverPtr driver, goto error; teardowncgroup = true; + if (virSecurityManagerSetHostdevLabel(driver->securityManager, + vm->def, hostdev, NULL) < 0) + goto error; + teardownlabel = true; + if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) { if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0) goto error; @@ -1250,6 +1256,10 @@ qemuDomainAttachHostPciDevice(virQEMUDriverPtr driver, error: if (teardowncgroup && qemuTeardownHostdevCgroup(vm, hostdev) < 0) VIR_WARN("Unable to remove host device cgroup ACL on hotplug fail"); + if (teardownlabel && + virSecurityManagerRestoreHostdevLabel(driver->securityManager, + vm->def, hostdev, NULL) < 0) + VIR_WARN("Unable to restore host device labelling on hotplug fail"); if (releaseaddr) qemuDomainReleaseDeviceAddress(vm, hostdev->info, NULL); @@ -1445,6 +1455,7 @@ qemuDomainAttachHostUsbDevice(virQEMUDriverPtr driver, char *devstr = NULL; bool added = false; bool teardowncgroup = false; + bool teardownlabel = false; int ret = -1; if (qemuFindHostdevUSBDevice(hostdev, true, &usb) < 0) @@ -1466,6 +1477,11 @@ qemuDomainAttachHostUsbDevice(virQEMUDriverPtr driver, goto cleanup; teardowncgroup = true; + if (virSecurityManagerSetHostdevLabel(driver->securityManager, + vm->def, hostdev, NULL) < 0) + goto cleanup; + teardownlabel = true; + if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) { if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0) goto cleanup; @@ -1492,10 +1508,14 @@ qemuDomainAttachHostUsbDevice(virQEMUDriverPtr driver, ret = 0; cleanup: - if (ret < 0 && - teardowncgroup && - qemuTeardownHostdevCgroup(vm, hostdev) < 0) - VIR_WARN("Unable to remove host device cgroup ACL on hotplug fail"); + if (ret < 0) { + if (teardowncgroup && qemuTeardownHostdevCgroup(vm, hostdev) < 0) + VIR_WARN("Unable to remove host device cgroup ACL on hotplug fail"); + if (teardownlabel && + virSecurityManagerRestoreHostdevLabel(driver->securityManager, + vm->def, hostdev, NULL) < 0) + VIR_WARN("Unable to restore host device labelling on hotplug fail"); + } if (added) virUSBDeviceListSteal(driver->activeUsbHostdevs, usb); virUSBDeviceFree(usb); @@ -1515,6 +1535,7 @@ qemuDomainAttachHostScsiDevice(virQEMUDriverPtr driver, char *devstr = NULL; char *drvstr = NULL; bool teardowncgroup = false; + bool teardownlabel = false; if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_DRIVE) || !virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE) || @@ -1543,6 +1564,11 @@ qemuDomainAttachHostScsiDevice(virQEMUDriverPtr driver, goto cleanup; teardowncgroup = true; + if (virSecurityManagerSetHostdevLabel(driver->securityManager, + vm->def, hostdev, NULL) < 0) + goto cleanup; + teardownlabel = true; + if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0) goto cleanup; @@ -1584,6 +1610,10 @@ cleanup: qemuDomainReAttachHostScsiDevices(driver, vm->def->name, &hostdev, 1); if (teardowncgroup && qemuTeardownHostdevCgroup(vm, hostdev) < 0) VIR_WARN("Unable to remove host device cgroup ACL on hotplug fail"); + if (teardownlabel && + virSecurityManagerRestoreHostdevLabel(driver->securityManager, + vm->def, hostdev, NULL) < 0) + VIR_WARN("Unable to restore host device labelling on hotplug fail"); } VIR_FREE(drvstr); VIR_FREE(devstr); @@ -1601,10 +1631,6 @@ int qemuDomainAttachHostDevice(virQEMUDriverPtr driver, return -1; } - if (virSecurityManagerSetHostdevLabel(driver->securityManager, - vm->def, hostdev, NULL) < 0) - return -1; - switch (hostdev->source.subsys.type) { case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI: if (qemuDomainAttachHostPciDevice(driver, vm, -- 1.8.4.2

2 6

[libvirt] [PATCH] Read PCI class from sysfs class file instead of config space.
by Thadeu Lima de Souza Cascardo 10 Dec '13

10 Dec '13

When determining if a device is behind a PCI bridge, the PCI device class is checked by reading the config space. However, there are some devices which have the wrong class on the config space, but the class is initialized by Linux correctly as a PCI BRIDGE. This class can be read by the sysfs file '/sys/bus/pci/devices/xxxx:xx:xx.x/class'. One example of such bridge is IBM PCI Bridge 1014:03b9, which is identified as a Host Bridge when reading the config space. Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo(a)linux.vnet.ibm.com> --- src/util/virpci.c | 38 +++++++++++++++++++++++++++++++++++--- 1 files changed, 35 insertions(+), 3 deletions(-) diff --git a/src/util/virpci.c b/src/util/virpci.c index 8ec642f..8353411 100644 --- a/src/util/virpci.c +++ b/src/util/virpci.c @@ -344,6 +344,34 @@ virPCIDeviceRead32(virPCIDevicePtr dev, int cfgfd, unsigned int pos) } static int +virPCIDeviceReadClass(virPCIDevicePtr dev, uint16_t *device_class) +{ + char *path = NULL; + char *id_str; + int ret = 0; + unsigned int value; + + if (virPCIFile(&path, dev->name, "class") < 0) + return -1; + + /* class string is '0xNNNNNN\n' ... i.e. 9 bytes */ + if (virFileReadAll(path, 9, &id_str) < 0) { + VIR_FREE(path); + return -1; + } + + VIR_FREE(path); + + id_str[8] = '\0'; + ret = virStrToLong_ui(id_str, NULL, 16, &value); + if (ret == 0) + *device_class = (value >> 8) & 0xFFFF; + + VIR_FREE(id_str); + return ret; +} + +static int virPCIDeviceWrite(virPCIDevicePtr dev, int cfgfd, unsigned int pos, @@ -645,8 +673,8 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) return 0; /* Is it a bridge? */ - device_class = virPCIDeviceRead16(check, fd, PCI_CLASS_DEVICE); - if (device_class != PCI_CLASS_BRIDGE_PCI) + ret = virPCIDeviceReadClass(check, &device_class); + if (ret == -1 || device_class != PCI_CLASS_BRIDGE_PCI) goto cleanup; /* Is it a plane? */ @@ -2110,6 +2138,7 @@ virPCIDeviceDownstreamLacksACS(virPCIDevicePtr dev) unsigned int pos; int fd; int ret = 0; + uint16_t device_class; if ((fd = virPCIDeviceConfigOpen(dev, true)) < 0) return -1; @@ -2119,8 +2148,11 @@ virPCIDeviceDownstreamLacksACS(virPCIDevicePtr dev) goto cleanup; } + if (virPCIDeviceReadClass(dev, &device_class)) + goto cleanup; + pos = dev->pcie_cap_pos; - if (!pos || virPCIDeviceRead16(dev, fd, PCI_CLASS_DEVICE) != PCI_CLASS_BRIDGE_PCI) + if (!pos || device_class != PCI_CLASS_BRIDGE_PCI) goto cleanup; flags = virPCIDeviceRead16(dev, fd, pos + PCI_EXP_FLAGS); -- 1.7.1

1 0

[libvirt] [libvirt-python 1/2] override: Fix exception handling syntax
by Doug Goldstein 09 Dec '13

09 Dec '13

Python 3 no longer accepts 'except Exception, e:' as valid while Python 2.4 does not accept the new syntax 'except Exception as e:' so this uses a fall back method that is compatible with both. --- libvirt-override-virStream.py | 3 ++- libvirt-override.py | 8 ++++++-- 2 files changed, 8 insertions(+), 3 deletions(-) diff --git a/libvirt-override-virStream.py b/libvirt-override-virStream.py index 53000da..189d062 100644 --- a/libvirt-override-virStream.py +++ b/libvirt-override-virStream.py @@ -50,7 +50,8 @@ ret = handler(self, got, opaque) if type(ret) is int and ret < 0: raise RuntimeError("recvAll handler returned %d" % ret) - except Exception, e: + except Exception: + e = sys.exc_info()[1] try: self.abort() except: diff --git a/libvirt-override.py b/libvirt-override.py index 87996f8..63f8ecb 100644 --- a/libvirt-override.py +++ b/libvirt-override.py @@ -3,12 +3,16 @@ # # On cygwin, the DLL is called cygvirtmod.dll +import sys + try: import libvirtmod -except ImportError, lib_e: +except ImportError: + lib_e = sys.exc_info()[1] try: import cygvirtmod as libvirtmod - except ImportError, cyg_e: + except ImportError: + cyg_e = sys.exc_info()[1] if str(cyg_e).count("No module named"): raise lib_e -- 1.8.3.2

2 3

[libvirt] [libvirt-python 0/3] Python 3 support mixed with Dan's patches
by Doug Goldstein 09 Dec '13

09 Dec '13

Turns out Dan and I both started working on Python 3 support and we talked about some changes to his patchset and these are the result of that conversation. This really belongs in the middle of his patchset but he's pushed the first part already so this is on top of master. Doug Goldstein (3): setup: Drop unused exception variable generator: Support exceptions in Python 2 and 3 Update exception catching in generated code generator.py | 18 ++++++++++++------ setup.py | 2 +- 2 files changed, 13 insertions(+), 7 deletions(-) -- 1.8.3.2

2 6

[libvirt] [PATCH] Workaround startup race condition
by Adam Walters 09 Dec '13

09 Dec '13

This patch works around a race condition on libvirt start. It feels a little hackish to me, but I have a hard time letting go of problems, so I had to at least try and implement something before going to sleep. I'd love some comments on the workaround, and welcome any improvement suggestions, as well. Please note that I implemented this in qemu_conf.c because it is the only place in the codebase I found VIR_DOMAIN_DISK_TYPE_VOLUME used, which tells me that only qemu has storage pool support at this time (not to mention that the qemu code is what I've become familiar with. It also doesn't hurt that I only have qemu configured and ready to go for testing, too). Adam Walters (1): qemu: conf: Work around race condition on libvirt start src/qemu/qemu_conf.c | 11 +++++++++++ 1 file changed, 11 insertions(+) -- 1.8.4.2

3 5

[libvirt] save-image-edit fails to change the memory configuration of the image
by Shivaprasad bhat 09 Dec '13

09 Dec '13

Hi, I see the below error when i change the memory. virsh save-image-edit /MANAS/rhel7VM1-running-save2 --running error: unsupported configuration: Target domain max memory 4194304 does not match source 2097152 Failed. Try again? [y,n,f,?]: n error: unsupported configuration: Target domain max memory 4194304 does not match source 2097152 Any idea why is this disallowed ? Thanks, Shiva

2 2

[libvirt] Libvirt support for KVM/ARM on armv7l hardware
by 林佳緯 09 Dec '13

09 Dec '13

I have found Libvirt supporting for armv7b in http://libvirt.org/news.html. But armv7l is not found there. Does Libvirt support run KVM/ARM for hardware virtualization on armv7l platform? Thanks!

1 0

[libvirt] Libvirt support for KVM/ARM
by 林佳緯 09 Dec '13

09 Dec '13

Does libvirt API support KVM/ARM now? Thanks for any answer. Gavin

1 0

[libvirt] [Ruby] Next release?
by Hiroshi Miura 08 Dec '13

08 Dec '13

Hi Chris, I'm waiting a next release of ruby-libvitrt in 4 month. Recent development git snapshot works well with vagnrant-kvm that I worked. thank you for hard work. Hiroshi

2 1

[libvirt] [PATCHv5 0/4] Implement RBD storage pool support
by Adam Walters 07 Dec '13

07 Dec '13

Here are updated patches that implement RBD storage pool support in libvirt. By and large, my changes are confined to qemu_conf.c, but I did need a couple of changes in domain_conf.c in order to get things working. I've tested these patches on my equipment, and they worked there, but they could probably use testing by others if anyone else has Ceph configured. The last patch in this series isn't needed in order to have RBD volumes work for domains, but it is needed if you don't want to have to kill the domain to get it back under the control of libvirt. Adam Walters (4): qemu: conf: Add qemuAddRBDPoolSourceHost helper function qemu: conf: Implement RBD storage pool support domain: conf: Bypass virDomainDiskDefForeachPath for network volumes domain: conf: Resolve bug related to network volume types src/conf/domain_conf.c | 6 +++-- src/qemu/qemu_conf.c | 62 +++++++++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 65 insertions(+), 3 deletions(-) -- 1.8.4.2

1 6