July 2013 - Devel - libvirt List Archives

[libvirt] [PATCH] Add a colon after 'internal error'
by Ján Tomko 24 Jul '13

24 Jul '13

As we do for other errors with an extra string. --- src/util/virerror.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/util/virerror.c b/src/util/virerror.c index b92c6d1..ca25678 100644 --- a/src/util/virerror.c +++ b/src/util/virerror.c @@ -730,7 +730,7 @@ virErrorMsg(virErrorNumber error, const char *info) return NULL; case VIR_ERR_INTERNAL_ERROR: if (info != NULL) - errmsg = _("internal error %s"); + errmsg = _("internal error: %s"); else errmsg = _("internal error"); break; -- 1.8.1.5

2 2

[libvirt] [PATCH] Fix virCgroupAvailable() w/o HAVE_GETMNTENT_R defined
by Roman Bogorodskiy 24 Jul '13

24 Jul '13

virCgroupAvailable() implementation calls getmntent_r without checking if HAVE_GETMNTENT_R is defined, so it fails to build on platforms without getmntent_r support. Make virCgroupAvailable() just return false without HAVE_GETMNTENT_R. --- src/util/vircgroup.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/util/vircgroup.c b/src/util/vircgroup.c index 5251611..d328212 100644 --- a/src/util/vircgroup.c +++ b/src/util/vircgroup.c @@ -69,9 +69,10 @@ typedef enum { bool virCgroupAvailable(void) { + bool ret = false; +#ifdef HAVE_GETMNTENT_R FILE *mounts = NULL; struct mntent entry; - bool ret = false; char buf[CGROUP_MAX_VAL]; if (!virFileExists("/proc/cgroups")) @@ -88,6 +89,7 @@ bool virCgroupAvailable(void) } VIR_FORCE_FCLOSE(mounts); +#endif return ret; } -- 1.7.11.5

2 1

[libvirt] [PATCH 0/3] qemu: Fix how files are being opened
by Martin Kletzander 24 Jul '13

24 Jul '13

There were some places in the code, where files were being opened with uid:gid of the daemon instead of the qemu process related to the file. First patch exposes the parseIds() function in order for it to be used somewhere else in the code than in the DAC security driver. The next patch fixes how the files are opened and the last one fixes occurences of open() that should use different uid:gid for opening files. There maybe should be a check for whether the file being opened is an image and whether the label used to open the file should be imagelabel or not. But, the QEMU process opening the file is running as the label (not imagelabel) and accessing the files as such. Martin Kletzander (3): Expose ownership ID parsing Make qemuOpenFile aware of per-VM DAC seclabel. Use qemuOpenFile in qemu_driver.c src/libvirt_private.syms | 1 + src/qemu/qemu_driver.c | 87 +++++++++++++++++++++++++++++++-------------- src/security/security_dac.c | 51 ++------------------------ src/util/virutil.c | 56 +++++++++++++++++++++++++++++ src/util/virutil.h | 2 ++ 5 files changed, 122 insertions(+), 75 deletions(-) -- 1.8.2.1

2 8

[libvirt] [PATCH V4] add console support in libxl
by Bamvor Jian Zhang 24 Jul '13

24 Jul '13

this patch introduce the console api in libxl driver for both pv and hvm guest. and import and update the libxlMakeChrdevStr function which was deleted in commit dfa1e1dd. Signed-off-by: Bamvor Jian Zhang <bjzhang(a)suse.com> --- Changes since V3: implicity forbit dev_name pass to libxl driver. Changes since V2: 1), forbid parallel configure because libxl do not support it 2), only support one serial on libxl driver. 3), also remove console code in libxl driver, AFAICS serial is enough for connecting to libxl console. changes since V1: 1), add virDomainOpenConsoleEnsureACL 3), remove virReportOOMErrorFull when virAsprintf fail. 4), change size_t for non-nagetive number in libxlDomainOpenConsole size_t i; size_t num = 0; 5), fix for make check (1), replace virAsprintf with VIR_STRDUP in two places (2), delete space. src/libxl/libxl_conf.c | 100 +++++++++++++++++++++++++++++++++++++++++++++++ src/libxl/libxl_conf.h | 3 ++ src/libxl/libxl_driver.c | 94 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 197 insertions(+) diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c index 4a0fba9..539d537 100644 --- a/src/libxl/libxl_conf.c +++ b/src/libxl/libxl_conf.c @@ -331,6 +331,90 @@ error: } static int +libxlMakeChrdevStr(virDomainChrDefPtr def, char **buf) +{ + const char *type = virDomainChrTypeToString(def->source.type); + int ret; + + if (!type) { + virReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("unexpected chr device type")); + return -1; + } + + switch (def->source.type) { + case VIR_DOMAIN_CHR_TYPE_NULL: + case VIR_DOMAIN_CHR_TYPE_STDIO: + case VIR_DOMAIN_CHR_TYPE_VC: + case VIR_DOMAIN_CHR_TYPE_PTY: + if (VIR_STRDUP(*buf, type) < 0) + return -1; + break; + + case VIR_DOMAIN_CHR_TYPE_FILE: + case VIR_DOMAIN_CHR_TYPE_PIPE: + if (virAsprintf(buf, "%s:%s", type, + def->source.data.file.path) < 0) + return -1; + break; + + case VIR_DOMAIN_CHR_TYPE_DEV: + if (VIR_STRDUP(*buf, def->source.data.file.path) < 0) + return -1; + break; + + case VIR_DOMAIN_CHR_TYPE_UDP: { + const char *connectHost = def->source.data.udp.connectHost; + const char *bindHost = def->source.data.udp.bindHost; + const char *bindService = def->source.data.udp.bindService; + + if (connectHost == NULL) + connectHost = ""; + if (bindHost == NULL) + bindHost = ""; + if (bindService == NULL) + bindService = "0"; + + ret = virAsprintf(buf, "udp:%s:%s@%s:%s", + connectHost, + def->source.data.udp.connectService, + bindHost, + bindService); + if (ret < 0) + return -1; + break; + + } + case VIR_DOMAIN_CHR_TYPE_TCP: + if (def->source.data.tcp.protocol == VIR_DOMAIN_CHR_TCP_PROTOCOL_TELNET) + ret = virAsprintf(buf, "telnet:%s:%s%s", + def->source.data.tcp.host, + def->source.data.tcp.service, + def->source.data.tcp.listen ? ",server,nowait" : ""); + else + ret = virAsprintf(buf, "tcp:%s:%s%s", + def->source.data.tcp.host, + def->source.data.tcp.service, + def->source.data.tcp.listen ? ",server,nowait" : ""); + + if (ret < 0) + return -1; + break; + + case VIR_DOMAIN_CHR_TYPE_UNIX: + ret = virAsprintf(buf, "unix:%s%s", + def->source.data.nix.path, + def->source.data.nix.listen ? ",server,nowait" : ""); + if (ret < 0) + return -1; + break; + + } + + return 0; +} + +static int libxlMakeDomBuildInfo(virDomainDefPtr def, libxl_domain_config *d_config) { libxl_domain_build_info *b_info = &d_config->b_info; @@ -403,6 +487,22 @@ libxlMakeDomBuildInfo(virDomainDefPtr def, libxl_domain_config *d_config) if (VIR_STRDUP(b_info->u.hvm.boot, bootorder) < 0) goto error; + if (def->nserials) { + if (def->nserials > 1) { + virReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("Only one serial is supported by libxl")); + goto error; + } + if (libxlMakeChrdevStr(def->serials[0], &b_info->u.hvm.serial) < 0) + goto error; + } + + if (def->nparallels) { + virReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("Parallel is not supported")); + goto error; + } + /* * The following comment and calculation were taken directly from * libxenlight's internal function libxl_get_required_shadow_memory(): diff --git a/src/libxl/libxl_conf.h b/src/libxl/libxl_conf.h index 2b4a281..861d689 100644 --- a/src/libxl/libxl_conf.h +++ b/src/libxl/libxl_conf.h @@ -34,6 +34,7 @@ # include "configmake.h" # include "virportallocator.h" # include "virobject.h" +# include "virchrdev.h" # define LIBXL_VNC_PORT_MIN 5900 @@ -94,6 +95,8 @@ struct _libxlDomainObjPrivate { /* per domain libxl ctx */ libxl_ctx *ctx; + /* console */ + virChrdevsPtr devs; libxl_evgen_domain_death *deathW; /* list of libxl timeout registrations */ diff --git a/src/libxl/libxl_driver.c b/src/libxl/libxl_driver.c index 358d387..9299484 100644 --- a/src/libxl/libxl_driver.c +++ b/src/libxl/libxl_driver.c @@ -417,6 +417,9 @@ libxlDomainObjPrivateAlloc(void) libxl_osevent_register_hooks(priv->ctx, &libxl_event_callbacks, priv); + if (!(priv->devs = virChrdevAlloc())) + return NULL; + return priv; } @@ -429,6 +432,7 @@ libxlDomainObjPrivateDispose(void *obj) libxl_evdisable_domain_death(priv->ctx, priv->deathW); libxl_ctx_free(priv->ctx); + virChrdevFree(priv->devs); } static void @@ -4493,6 +4497,95 @@ cleanup: return ret; } + +static int +libxlDomainOpenConsole(virDomainPtr dom, + const char *dev_name, + virStreamPtr st, + unsigned int flags) +{ + libxlDriverPrivatePtr driver = dom->conn->privateData; + virDomainObjPtr vm = NULL; + int ret = -1; + virDomainChrDefPtr chr = NULL; + libxlDomainObjPrivatePtr priv; + char *console = NULL; + + virCheckFlags(VIR_DOMAIN_CONSOLE_SAFE | + VIR_DOMAIN_CONSOLE_FORCE, -1); + + if (dev_name) { + /* XXX support device aliases in future */ + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("Named device aliases are not supported")); + goto cleanup; + } + + libxlDriverLock(driver); + vm = virDomainObjListFindByUUID(driver->domains, dom->uuid); + libxlDriverUnlock(driver); + if (!vm) { + char uuidstr[VIR_UUID_STRING_BUFLEN]; + virUUIDFormat(dom->uuid, uuidstr); + virReportError(VIR_ERR_NO_DOMAIN, + _("No domain with matching uuid '%s'"), uuidstr); + goto cleanup; + } + + if (virDomainOpenConsoleEnsureACL(dom->conn, vm->def) < 0) + goto cleanup; + + if (!virDomainObjIsActive(vm)) { + virReportError(VIR_ERR_OPERATION_INVALID, + "%s", _("domain is not running")); + goto cleanup; + } + + priv = vm->privateData; + + if (vm->def->nserials) + chr = vm->def->serials[0]; + + if (!chr) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find character device %s"), + NULLSTR(dev_name)); + goto cleanup; + } + + if (chr->source.type != VIR_DOMAIN_CHR_TYPE_PTY) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("character device %s is not using a PTY"), + NULLSTR(dev_name)); + goto cleanup; + } + + ret = libxl_primary_console_get_tty(priv->ctx, vm->def->id, &console); + if (ret) + goto cleanup; + + if (VIR_STRDUP(chr->source.data.file.path, console) < 0) + goto cleanup; + + /* handle mutually exclusive access to console devices */ + ret = virChrdevOpen(priv->devs, + &chr->source, + st, + (flags & VIR_DOMAIN_CONSOLE_FORCE) != 0); + + if (ret == 1) { + virReportError(VIR_ERR_OPERATION_FAILED, "%s", + _("Active console session exists for this domain")); + ret = -1; + } + +cleanup: + VIR_FREE(console); + if (vm) + virObjectUnlock(vm); + return ret; +} + static int libxlDomainSetSchedulerParameters(virDomainPtr dom, virTypedParameterPtr params, int nparams) @@ -4875,6 +4968,7 @@ static virDriver libxlDriver = { .domainManagedSave = libxlDomainManagedSave, /* 0.9.2 */ .domainHasManagedSaveImage = libxlDomainHasManagedSaveImage, /* 0.9.2 */ .domainManagedSaveRemove = libxlDomainManagedSaveRemove, /* 0.9.2 */ + .domainOpenConsole = libxlDomainOpenConsole, /* 1.1.1 */ .domainIsActive = libxlDomainIsActive, /* 0.9.0 */ .domainIsPersistent = libxlDomainIsPersistent, /* 0.9.0 */ .domainIsUpdated = libxlDomainIsUpdated, /* 0.9.0 */ -- 1.8.1.4

1 0

[libvirt] [PATCH v3 0/7] Support to use iscsi storage in domain conf
by John Ferlan 24 Jul '13

24 Jul '13

An update to yesterday's posting: https://www.redhat.com/archives/libvir-list/2013-July/msg01213.html Changes in v3 over v2 * Update patch 1 per code review * Use 'direct' instead of 'uri' and supporting documentaiton * In patch 4, use qemuAddISCSIPoolSourceHost() instead of qemuAddDiskISCSIUri. Also added a check in the function that (pooldef->source.nhost != 1) to make sure we at least have a host defined in the pool and of course that there's only one. John Ferlan (5): storage_iscsi: Reflect the default target port conf: Introduce new XML tag "mode" for disk source conf: Introduce virDomainDiskSourceIsBlockType qemu: Translate the iscsi pool/volume disk source tests: Add various network and volume definitions Osier Yang (2): conf: Ignore the volume type disk if its mode is "direct" qemu: Translate the volume type disk source before cgroup setting docs/formatdomain.html.in | 11 +- docs/schemas/domaincommon.rng | 8 ++ src/conf/domain_conf.c | 67 ++++++++++- src/conf/domain_conf.h | 26 +++++ src/libvirt_private.syms | 1 + src/qemu/qemu_command.c | 20 +++- src/qemu/qemu_conf.c | 124 +++++++++++++++++---- src/qemu/qemu_process.c | 13 ++- src/storage/storage_backend_iscsi.c | 19 ++-- .../qemuxml2argv-disk-source-pool-mode.xml | 48 ++++++++ tests/qemuxml2xmltest.c | 1 + tests/securityselinuxlabeldata/netdisks.txt | 5 + tests/securityselinuxlabeldata/netdisks.xml | 58 ++++++++++ tests/securityselinuxlabeldata/voldisks.txt | 5 + tests/securityselinuxlabeldata/voldisks.xml | 45 ++++++++ tests/securityselinuxlabeltest.c | 2 + 16 files changed, 414 insertions(+), 39 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-source-pool-mode.xml create mode 100644 tests/securityselinuxlabeldata/netdisks.txt create mode 100644 tests/securityselinuxlabeldata/netdisks.xml create mode 100644 tests/securityselinuxlabeldata/voldisks.txt create mode 100644 tests/securityselinuxlabeldata/voldisks.xml -- 1.8.1.4

4 41

[libvirt] [PATCH] dbus: work with older dbus
by Eric Blake 24 Jul '13

24 Jul '13

dbus 1.2.24 (on RHEL 6) lacks DBUS_TYPE_UNIX_FD; but as we aren't trying to pass one of those anyways, we can just drop support for it in our wrapper. Solves this build error: CC libvirt_util_la-virdbus.lo util/virdbus.c:242: error: 'DBUS_TYPE_UNIX_FD' undeclared here (not in a function) * src/util/virdbus.c (virDBusBasicTypes): Drop support for unix fds. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing under the build-breaker rule. src/util/virdbus.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/src/util/virdbus.c b/src/util/virdbus.c index 9b0977a..4ce6c46 100644 --- a/src/util/virdbus.c +++ b/src/util/virdbus.c @@ -1,7 +1,7 @@ /* * virdbus.c: helper for using DBus * - * Copyright (C) 2012 Red Hat, Inc. + * Copyright (C) 2012-2013 Red Hat, Inc. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public @@ -239,7 +239,6 @@ static const char virDBusBasicTypes[] = { DBUS_TYPE_STRING, DBUS_TYPE_OBJECT_PATH, DBUS_TYPE_SIGNATURE, - DBUS_TYPE_UNIX_FD }; static bool virDBusIsBasicType(char c) { @@ -1024,6 +1023,9 @@ int virDBusMessageDecode(DBusMessage* msg, * '{' - start of a dictionary entry (pair of types) * '}' - start of a dictionary entry (pair of types) * + * At this time, there is no support for Unix fd's ('h'), which only + * newer DBus supports. + * * Passing values in variadic args for basic types is * simple, the value is just passed directly using the * corresponding C type listed against the type code -- 1.8.3.1

2 1

[libvirt] [PATCH 0/5] Support for pcie-root / Q35 machine type
by Laine Stump 24 Jul '13

24 Jul '13

This is *almost* usable. I still need to add support for a dmi-to-pci-bridge before the q35 machine type can be used. 1/5 and 3/5 are just cleaning up some things that bothered me while writing patch 4/5. 2/5 should help out other machine types (e.g. arm) just a bit, by eliminating the extra PIIX3 devices unless the machine actually has a PIIX3 chip. 4/5 has the bulk of the changes - it restructures the PCI address validation/allocation so that it doesn't assume that all PCI addresses are the same, or that all PCI slots are the same. It sets up an enum where different types of PCI addresses can be listed, and the validation code checks that the slot about to be used for a device is actually compatible with that device. 5/5 adds the pcie-root controller which is implicit in the q35 machinetype. Of course since we can only connect pcie devices to this controller, and there are no pcie devices supported, it is not really worth much. However, the next patch (in progress) will add a dmi-to-pci-bridge controller, which *can* plug into the pcie-root and provide plain PCI slots, making a working a35 possible. Note that the auto-allocation doesn't attempt to auto-allocate a slot for any type of device other than plain PCI. This means that even when the dmi-to-pci-bridge controller is added, its address on pcie-root (and a pci-bridge device that must be connected to the dmi-to-pci-bridge controller) will need to be manually specified. Auto-placement will be an exercise for later. Things missing: 1) I need a test case for auto-adding multiple bridges. I will add that this afternoon. 2) I need a test case for a q35 domain. That wil also be added this afternoon. Laine Stump (5): qemu: turn qemuDomainPCIAddressBus into a struct qemu: only check for PIIX3-specific device addrs on pc-* machinetypes qemu: make QEMU_PCI_ADDRESS_(SLOT|FUNCTION)_LAST less misleading qemu: set/validate slot/connection type when assigning slots for PCI devices conf: add pcie-root controller docs/formatdomain.html.in | 17 +- src/conf/domain_conf.c | 4 +- src/conf/domain_conf.h | 5 +- src/libvirt_private.syms | 1 + src/qemu/qemu_command.c | 559 ++++++++++++++++++++++++++++++---------------- src/qemu/qemu_command.h | 28 ++- src/qemu/qemu_domain.c | 23 +- 7 files changed, 434 insertions(+), 203 deletions(-) -- 1.7.11.7

3 16

[libvirt] [PATCH 0/2] Integration with systemd-machined
by Daniel P. Berrange 24 Jul '13

24 Jul '13

From: "Daniel P. Berrange" <berrange(a)redhat.com> This includes a v2 of a previously submitted patch https://www.redhat.com/archives/libvir-list/2013-July/msg01084.html This posting includes a second patch to actually wrap the systemd call Changes since that first posting - Use 'int' instead of 'size_t' for the ARRAY var-arg, since a plain number "3" does not promote to 'size_t' when passed through variadic args, only to 'int'. Using 'int' thus avoids need to cast every time. - Handle DBus error replies - Make sure building --without-dbus works Daniel P. Berrange (2): Introduce virDBusCallMethod & virDBusMessageRead methods Add API for calling systemd-machined's DBus API .gitignore | 2 + include/libvirt/virterror.h | 2 + src/Makefile.am | 1 + src/libvirt_private.syms | 8 + src/util/virdbus.c | 966 +++++++++++++++++++++++++++++++++++++++++++- src/util/virdbus.h | 11 + src/util/virdbuspriv.h | 45 +++ src/util/virerror.c | 7 + src/util/virerror.h | 11 + src/util/virsystemd.c | 139 +++++++ src/util/virsystemd.h | 36 ++ tests/Makefile.am | 29 ++ tests/testutils.h | 2 + tests/virdbustest.c | 389 ++++++++++++++++++ tests/virsystemdmock.c | 77 ++++ tests/virsystemdtest.c | 131 ++++++ 16 files changed, 1855 insertions(+), 1 deletion(-) create mode 100644 src/util/virdbuspriv.h create mode 100644 src/util/virsystemd.c create mode 100644 src/util/virsystemd.h create mode 100644 tests/virdbustest.c create mode 100644 tests/virsystemdmock.c create mode 100644 tests/virsystemdtest.c -- 1.8.1.4

2 8

[libvirt] [PATCH] Don't overwrite errors in qemuTranslateDiskSourcePool
by Ján Tomko 24 Jul '13

24 Jul '13

Both virStoragePoolFree and virStorageVolFree reset the last error, which might lead to the cryptic message: An error occurred, but the cause is unknown When the volume wasn't found, virStorageVolFree was called with NULL, leading to an error: invalid storage volume pointer in virStorageVolFree This patch changes it to: Storage volume not found: no storage vol with matching name 'tomato' --- src/qemu/qemu_conf.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index 3e7b78a..80b8156 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -1248,6 +1248,7 @@ qemuTranslateDiskSourcePool(virConnectPtr conn, char *poolxml = NULL; virStorageVolInfo info; int ret = -1; + virErrorPtr savedError; if (def->type != VIR_DOMAIN_DISK_TYPE_VOLUME) return 0; @@ -1324,8 +1325,14 @@ qemuTranslateDiskSourcePool(virConnectPtr conn, def->srcpool->voltype = info.type; ret = 0; cleanup: - virStoragePoolFree(pool); - virStorageVolFree(vol); + savedError = virSaveLastError(); + if (pool) + virStoragePoolFree(pool); + if (vol) + virStorageVolFree(vol); + virSetError(savedError); + virFreeError(savedError); + VIR_FREE(poolxml); virStoragePoolDefFree(pooldef); return ret; -- 1.8.1.5

2 1

Re: [libvirt] [Qemu-devel] [Question] why x2apic's set by default without host support(on Nehalem CPU).
by Eduardo Habkost 23 Jul '13

23 Jul '13

On Mon, Jul 22, 2013 at 07:24:24PM +0800, Peter Huang(Peng) wrote: > Hi, Everyone > > I have been encountered a problem recently. > My box uses Nehalem E5520 CPU, and this model doesn't support x2apic feature. > I created a VM on KVM, and uses cpu mode=pass-through, I can get the right cpu > model from the VM, but I can also see that the CPU in the vm supports x2apic. > > This seems werid. > From QEMU's CPU model definition, Nehalem also doesn't support x2apic. > > Can anyone explain it? > libvirt's "host-passthrough" uses "-cpu host', and it "-cpu host" enables every feature that can be enabled on the host. As KVM x2apic emulation doesn't require host CPU support, it gets enabled. If the behavior of "-cpu host" doesn't match libvirt expectations, we need to clarify what are the requirements, and maybe have a "try to be close to host CPU mode" as opposed to the current "enable everything that can be enabled" mode. -- Eduardo

3 4