January 2011 - Devel - Libvirt List Archives

[libvirt] [PATCH] storage: Allow to delete device mapper disk partition

by Osier Yang

From: root <root(a)amd-1352-8-2.englab.nay.redhat.com> The name convention of device mapper disk is different, and 'parted' can't be used to delete a device mapper disk partition. e.g. [root@amd-1352-8-2 ~]# virsh vol-list --pool osier Name Path ----------------------------------------- 3600a0b80005ad1d7000093604cae912fp1 /dev/mapper/3600a0b80005ad1d7000093604cae912fp1 [root@amd-1352-8-2 ~]# parted /dev/mapper/3600a0b80005adb0b0000ab2d4cae9254 rm p1 Error: Expecting a partition number. This patch introduces 'dmsetup' to fix it. Changes: - New function 'virIsDeviceMapperDevice' in src/utils/utils.c - Delete 'is_dm_device' in src/storage/parthelper.c, use 'virIsDeviceMapperDevice' instead. - Requires device-mapper for 'with-storage-disk' in libvirt.spec.in - Check 'dmsetup' in configure.ac for 'with-storage-disk' - Changes on src/Makefile.am to link against libdevmapper - New entry for 'virIsDeviceMapperDevice' in src/libvirt_private.syms e.g. [root@amd-1352- ~]# virsh vol-list --pool osier Name Path ----------------------------------------- 3600a0b80005ad1d7000093604cae912fp1 /dev/mapper/3600a0b80005ad1d7000093604cae912fp1 [root@amd-1352- ~]# virsh vol-delete /dev/mapper/3600a0b80005ad1d7000093604cae912fp1 Vol /dev/mapper/3600a0b80005ad1d7000093604cae912fp1 deleted [root@amd-1352- ~]# virsh vol-list --pool osier Name Path ----------------------------------------- --- configure.ac | 21 ++++++++++++++ libvirt.spec.in | 1 + src/Makefile.am | 7 ++-- src/libvirt_private.syms | 1 + src/storage/parthelper.c | 14 +-------- src/storage/storage_backend_disk.c | 54 ++++++++++++++++++++++------------- src/util/util.c | 14 +++++++++ src/util/util.h | 1 + 8 files changed, 76 insertions(+), 37 deletions(-) diff --git a/configure.ac b/configure.ac index f310a5e..b101faa 100644 --- a/configure.ac +++ b/configure.ac @@ -1705,6 +1705,8 @@ LIBPARTED_CFLAGS= LIBPARTED_LIBS= if test "$with_storage_disk" = "yes" || test "$with_storage_disk" = "check"; then AC_PATH_PROG([PARTED], [parted], [], [$PATH:/sbin:/usr/sbin]) + AC_PATH_PROG([DMSETUP], [dmsetup], [], [$PATH:/sbin:/usr/sbin]) + if test -z "$PARTED" ; then with_storage_disk=no PARTED_FOUND=no @@ -1712,9 +1714,17 @@ if test "$with_storage_disk" = "yes" || test "$with_storage_disk" = "check"; the PARTED_FOUND=yes fi + if test -z "$DMSETUP" ; then + with_storage_disk=no + DMSETUP_FOUND=no + else + DMSETUP_FOUND=yes + fi + if test "$with_storage_disk" != "no" && test "x$PKG_CONFIG" != "x" ; then PKG_CHECK_MODULES(LIBPARTED, libparted >= $PARTED_REQUIRED, [], [PARTED_FOUND=no]) fi + if test "$PARTED_FOUND" = "no"; then # RHEL-5 vintage parted is missing pkg-config files save_LIBS="$LIBS" @@ -1738,9 +1748,20 @@ if test "$with_storage_disk" = "yes" || test "$with_storage_disk" = "check"; the with_storage_disk=yes fi + if test "$DMSETUP_FOUND" = "no"; then + if test "$with_storage_disk" = "yes" ; then + AC_MSG_ERROR([We need dmsetup for disk storage driver]) + else + with_storage_disk=no + fi + else + with_storage_disk=yes + fi + if test "$with_storage_disk" = "yes"; then AC_DEFINE_UNQUOTED([WITH_STORAGE_DISK], 1, [whether Disk backend for storage driver is enabled]) AC_DEFINE_UNQUOTED([PARTED],["$PARTED"], [Location or name of the parted program]) + AC_DEFINE_UNQUOTED([DMSETUP],["$DMSETUP"], [Location or name of the dmsetup program]) fi fi AM_CONDITIONAL([WITH_STORAGE_DISK], [test "$with_storage_disk" = "yes"]) diff --git a/libvirt.spec.in b/libvirt.spec.in index 0a2d10e..27e6a99 100644 --- a/libvirt.spec.in +++ b/libvirt.spec.in @@ -279,6 +279,7 @@ Requires: iscsi-initiator-utils %if %{with_storage_disk} # For disk driver Requires: parted +Requires: device-mapper %endif %if %{with_storage_mpath} # For multipath support diff --git a/src/Makefile.am b/src/Makefile.am index f8b8434..3680074 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -437,9 +437,9 @@ libvirt_la_BUILT_LIBADD = libvirt_util.la libvirt_util_la_SOURCES = \ $(UTIL_SOURCES) libvirt_util_la_CFLAGS = $(CAPNG_CFLAGS) $(YAJL_CFLAGS) $(LIBNL_CFLAGS) \ - $(AM_CFLAGS) $(AUDIT_CFLAGS) + $(AM_CFLAGS) $(AUDIT_CFLAGS) $(DEVMAPPER_LIBS) libvirt_util_la_LIBADD = $(CAPNG_LIBS) $(YAJL_LIBS) $(LIBNL_LIBS) \ - $(LIB_PTHREAD) $(AUDIT_LIBS) + $(LIB_PTHREAD) $(AUDIT_LIBS) $(DEVMAPPER_LIBS) noinst_LTLIBRARIES += libvirt_conf.la @@ -1154,7 +1154,6 @@ libvirt_parthelper_SOURCES = $(STORAGE_HELPER_DISK_SOURCES) libvirt_parthelper_LDFLAGS = $(WARN_LDFLAGS) $(AM_LDFLAGS) libvirt_parthelper_LDADD = \ $(LIBPARTED_LIBS) \ - $(DEVMAPPER_LIBS) \ libvirt_util.la \ ../gnulib/lib/libgnu.la @@ -1179,7 +1178,7 @@ libvirt_lxc_SOURCES = \ libvirt_lxc_LDFLAGS = $(WARN_CFLAGS) $(AM_LDFLAGS) libvirt_lxc_LDADD = $(CAPNG_LIBS) $(YAJL_LIBS) \ $(LIBXML_LIBS) $(NUMACTL_LIBS) $(LIB_PTHREAD) \ - $(LIBNL_LIBS) $(AUDIT_LIBS) ../gnulib/lib/libgnu.la + $(LIBNL_LIBS) $(AUDIT_LIBS) $(DEVMAPPER_LIBS) ../gnulib/lib/libgnu.la libvirt_lxc_CFLAGS = \ $(LIBPARTED_CFLAGS) \ $(NUMACTL_CFLAGS) \ diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index 3a5aec9..9d974e8 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -900,6 +900,7 @@ virStrcpy; virStrncpy; virTimestamp; virVasprintf; +virIsDeviceMapperDevice; # uuid.h diff --git a/src/storage/parthelper.c b/src/storage/parthelper.c index 6ef413d..1f662fb 100644 --- a/src/storage/parthelper.c +++ b/src/storage/parthelper.c @@ -59,18 +59,6 @@ enum diskCommand { DISK_GEOMETRY }; -static int -is_dm_device(const char *devname) -{ - struct stat buf; - - if (devname && !stat(devname, &buf) && dm_is_dm_major(major(buf.st_rdev))) { - return 1; - } - - return 0; -} - int main(int argc, char **argv) { PedDevice *dev; @@ -96,7 +84,7 @@ int main(int argc, char **argv) } path = argv[1]; - if (is_dm_device(path)) { + if (virIsDeviceMapperDevice(path)) { partsep = "p"; canonical_path = strdup(path); if (canonical_path == NULL) { diff --git a/src/storage/storage_backend_disk.c b/src/storage/storage_backend_disk.c index c7ade6b..45e43c4 100644 --- a/src/storage/storage_backend_disk.c +++ b/src/storage/storage_backend_disk.c @@ -660,34 +660,48 @@ virStorageBackendDiskDeleteVol(virConnectPtr conn ATTRIBUTE_UNUSED, srcname = basename(pool->def->source.devices[0].path); DEBUG("devname=%s, srcname=%s", devname, srcname); - if (!STRPREFIX(devname, srcname)) { + if (!virIsDeviceMapperDevice(devpath) && !STRPREFIX(devname, srcname)) { virStorageReportError(VIR_ERR_INTERNAL_ERROR, _("Volume path '%s' did not start with parent " "pool source device name."), devname); goto cleanup; } - part_num = devname + strlen(srcname); + if (!virIsDeviceMapperDevice(devpath)) { + part_num = devname + strlen(srcname); - if (*part_num == 0) { - virStorageReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot parse partition number from target " - "'%s'"), devname); - goto cleanup; - } - - /* eg parted /dev/sda rm 2 */ - const char *prog[] = { - PARTED, - pool->def->source.devices[0].path, - "rm", - "--script", - part_num, - NULL, - }; + if (*part_num == 0) { + virStorageReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot parse partition number from target " + "'%s'"), devname); + goto cleanup; + } - if (virRun(prog, NULL) < 0) - goto cleanup; + /* eg parted /dev/sda rm 2 */ + const char *prog[] = { + PARTED, + pool->def->source.devices[0].path, + "rm", + "--script", + part_num, + NULL, + }; + + if (virRun(prog, NULL) < 0) + goto cleanup; + } else { + const char *prog[] = { + DMSETUP, + "remove", + "-f", + devpath, + NULL, + NULL, + }; + + if (virRun(prog, NULL) < 0) + goto cleanup; + } rc = 0; cleanup: diff --git a/src/util/util.c b/src/util/util.c index ee04ca9..6e2f966 100644 --- a/src/util/util.c +++ b/src/util/util.c @@ -45,6 +45,7 @@ #include <string.h> #include <signal.h> #include <termios.h> +#include <libdevmapper.h> #include "c-ctype.h" #ifdef HAVE_PATHS_H @@ -3100,3 +3101,16 @@ virTimestamp(void) return timestamp; } + +int +virIsDeviceMapperDevice(const char *devname) +{ + struct stat buf; + + if (devname && !stat(devname, &buf) && dm_is_dm_major(major(buf.st_rdev))) { + return 1; + } + + return 0; +} + diff --git a/src/util/util.h b/src/util/util.h index 8373038..dd683ea 100644 --- a/src/util/util.h +++ b/src/util/util.h @@ -296,4 +296,5 @@ int virBuildPathInternal(char **path, ...) ATTRIBUTE_SENTINEL; char *virTimestamp(void); +int virIsDeviceMapperDevice(const char *devname); #endif /* __VIR_UTIL_H__ */ -- 1.7.1

13 years, 3 months

3
4
0 / 0

[libvirt] [PATCH] virsh: Error prompt if one passes negative value for scheduler setting

by Osier Yang

As cgroup doesn't allow one writes negative into files like cpu.shares, (e.g. echo -1> /cgroup/cpu/libvirt/qemu/rhel6/cpu.shares), user will be confused if libvirt accepts negative value and converts it into unsigned int (or long int, etc) silently. * tools/virsh.c --- tools/virsh.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/tools/virsh.c b/tools/virsh.c index cd54174..a0f2527 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -1652,7 +1652,8 @@ cmdSchedInfoUpdate(vshControl *ctl, const vshCmd *cmd, } break; case VIR_DOMAIN_SCHED_FIELD_UINT: - if (virStrToLong_ui(val, NULL, 10, &param->value.ui) < 0) { + if (STRPREFIX(val, "-") || + virStrToLong_ui(val, NULL, 10, &param->value.ui) < 0) { vshError(ctl, "%s", _("Invalid value for parameter, expecting an unsigned int")); return -1; @@ -1666,7 +1667,8 @@ cmdSchedInfoUpdate(vshControl *ctl, const vshCmd *cmd, } break; case VIR_DOMAIN_SCHED_FIELD_ULLONG: - if (virStrToLong_ull(val, NULL, 10, &param->value.ul) < 0) { + if (STRPREFIX(val, "-") || + virStrToLong_ull(val, NULL, 10, &param->value.ul) < 0) { vshError(ctl, "%s", _("Invalid value for parameter, expecting an unsigned long long")); return -1; -- 1.7.3.2

13 years, 3 months

3
9
0 / 0

[libvirt] [PATCH] qemu: Build command line for incoming tunneled migration

by Osier Yang

Command line building for imcoming tunneled migration is missed, as a result, all the tunneled migration will fail with "unknown migration protocol". * src/qemu/qemu_command.c --- src/qemu/qemu_command.c | 8 ++++++++ 1 files changed, 8 insertions(+), 0 deletions(-) diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index a0f86a3..8f681fd 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -4067,6 +4067,14 @@ qemuBuildCommandLine(virConnectPtr conn, } virCommandAddArg(cmd, migrateFrom); virCommandPreserveFD(cmd, migrateFd); + } else if (STRPREFIX(migrateFrom, "unix")) { + if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX)) { + qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, + "%s", _("UNIX migration is not supported " + "with this QEMU binary")); + goto error; + } + virCommandAddArg(cmd, migrateFrom); } else { qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("unknown migration protocol")); -- 1.7.3.2

13 years, 3 months

2
4
0 / 0

[libvirt] [PATCHv3 0/5] smartcard: round 3

by Eric Blake

This series has hopefully taken into account all the feedback from v2 (https://www.redhat.com/archives/libvir-list/2011-January/msg00608.html). Major changes: - enhance the XML to support optional ccid <controller> (missing controllers are added according to <address> elements) and optional <address> per smartcard (missing address assume the next available port on controller 0) - enhance the XML to support an optional <source dev='/path'/> for host mode. For now, this path is only used in SELinux labeling; I suspect that this needs more work, since the point is that a single device in the host should be shared among the NSS implementation of multiple guests (so labeling the host device to belong to a single guest is wrong); but fixing it correctly requires a better understanding of what NSS actually needs to access, as well as possibly modifying qemu's smartcard implementation to take the host device either as a pathname or even as an already-opened fd. - enhance the XML to support an optional <database> element for host-certificates mode. - enhance the qemu command line to fully populate all parameters, rather than the bare minimum defaults, and reflect that in the tests. It requires this pre-requisite patch for qemu -chardev aliases: https://www.redhat.com/archives/libvir-list/2011-January/msg01032.html Eric Blake (5): smartcard: add XML support for <smartcard> device smartcard: add domain conf support smartcard: check for qemu capability smartcard: enable SELinux support smartcard: turn on qemu support cfg.mk | 1 + docs/formatdomain.html.in | 95 +++++- docs/schemas/domain.rng | 73 ++++ src/conf/domain_conf.c | 396 +++++++++++++++++++- src/conf/domain_conf.h | 53 +++- src/libvirt_private.syms | 4 + src/qemu/qemu_capabilities.c | 2 + src/qemu/qemu_capabilities.h | 1 + src/qemu/qemu_command.c | 90 +++++- src/security/security_selinux.c | 94 +++++ .../qemuxml2argv-smartcard-controller.args | 1 + .../qemuxml2argv-smartcard-controller.xml | 20 + .../qemuxml2argv-smartcard-host-certificates.args | 1 + .../qemuxml2argv-smartcard-host-certificates.xml | 20 + .../qemuxml2argv-smartcard-host.args | 1 + .../qemuxml2argv-smartcard-host.xml | 16 + .../qemuxml2argv-smartcard-passthrough-tcp.args | 1 + .../qemuxml2argv-smartcard-passthrough-tcp.xml | 19 + tests/qemuxml2argvtest.c | 13 + 19 files changed, 887 insertions(+), 14 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-controller.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-controller.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-host-certificates.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-host-certificates.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-host.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-host.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-passthrough-tcp.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-smartcard-passthrough-tcp.xml -- 1.7.3.5

13 years, 3 months

3
22
0 / 0

[libvirt] [PATCH 2/4 v2] cgroup: Implement blkio.weight tuning API.

by Gui Jianfeng

Implement blkio.weight tuning API. Signed-off-by: Gui Jianfeng <guijianfeng(a)cn.fujitsu.com> --- src/libvirt_private.syms | 2 ++ src/util/cgroup.c | 39 +++++++++++++++++++++++++++++++++++++++ src/util/cgroup.h | 3 +++ 3 files changed, 44 insertions(+), 0 deletions(-) diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index 2ce4bed..97b9851 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -77,6 +77,8 @@ virCgroupMounted; virCgroupRemove; virCgroupSetCpuShares; virCgroupSetFreezerState; +virCgroupSetWeight; +virCgroupGetWeight; virCgroupSetMemory; virCgroupSetMemoryHardLimit; virCgroupSetMemorySoftLimit; diff --git a/src/util/cgroup.c b/src/util/cgroup.c index 309f4e9..4b7e629 100644 --- a/src/util/cgroup.c +++ b/src/util/cgroup.c @@ -851,6 +851,45 @@ int virCgroupForDomain(virCgroupPtr driver ATTRIBUTE_UNUSED, #endif /** + * virCgroupSetWeight: + * + * @group: The cgroup to change io weight for + * @weight: The Weight for this cgroup + * + * Returns: 0 on success + */ +int virCgroupSetWeight(virCgroupPtr group, unsigned long weight) +{ + if (weight > 1000 || weight < 100) + return -EINVAL; + + return virCgroupSetValueU64(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "blkio.weight", + weight); +} + +/** + * virCgroupGetWeight: + * + * @group: The cgroup to get weight for + * @Weight: Pointer to returned weight + * + * Returns: 0 on success + */ +int virCgroupGetWeight(virCgroupPtr group, unsigned long *weight) +{ + long long unsigned int __weight; + int ret; + ret = virCgroupGetValueU64(group, + VIR_CGROUP_CONTROLLER_BLKIO, + "blkio.weight", &__weight); + if (ret == 0) + *weight = (unsigned long) __weight; + return ret; +} + +/** * virCgroupSetMemory: * * @group: The cgroup to change memory for diff --git a/src/util/cgroup.h b/src/util/cgroup.h index 67b1299..99c1cfe 100644 --- a/src/util/cgroup.h +++ b/src/util/cgroup.h @@ -41,6 +41,9 @@ int virCgroupForDomain(virCgroupPtr driver, int virCgroupAddTask(virCgroupPtr group, pid_t pid); +int virCgroupSetWeight(virCgroupPtr group, unsigned long weight); +int virCgroupGetWeight(virCgroupPtr group, unsigned long *weight); + int virCgroupSetMemory(virCgroupPtr group, unsigned long long kb); int virCgroupGetMemoryUsage(virCgroupPtr group, unsigned long *kb); -- 1.7.1

13 years, 3 months

2
1
0 / 0

[libvirt] [PATCHv4] bridge_driver: handle DNS over IPv6

by Paweł Krześniak

* dnsmasq listens on all defined IPv[46] addresses for network * Add ip6tables rules to allow DNS traffic to host --- src/network/bridge_driver.c | 51 ++++++++++++++++++++++++++++++++++-------- 1 files changed, 41 insertions(+), 10 deletions(-) diff --git a/src/network/bridge_driver.c b/src/network/bridge_driver.c index c098ab5..24be0b7 100644 --- a/src/network/bridge_driver.c +++ b/src/network/bridge_driver.c @@ -432,6 +432,8 @@ networkBuildDnsmasqArgv(virNetworkObjPtr network, int r, ret = -1; int nbleases = 0; char *bridgeaddr; + int ii; + virNetworkIpDefPtr tmpipdef; if (!(bridgeaddr = virSocketFormatAddr(&ipdef->address))) goto cleanup; @@ -468,20 +470,28 @@ networkBuildDnsmasqArgv(virNetworkObjPtr network, /* *no* conf file */ virCommandAddArgList(cmd, "--conf-file=", "", NULL); - /* - * XXX does not actually work, due to some kind of - * race condition setting up ipv6 addresses on the - * interface. A sleep(10) makes it work, but that's - * clearly not practical - * - * virCommandAddArg(cmd, "--interface"); - * virCommandAddArg(cmd, ipdef->bridge); - */ virCommandAddArgList(cmd, - "--listen-address", bridgeaddr, "--except-interface", "lo", NULL); + /* + * --interface does not actually work with dnsmasq < 2.47, + * due to DAD for ipv6 addresses on the interface. + * + * virCommandAddArgList(cmd, "--interface", ipdef->bridge, NULL); + * + * So listen on all defined IPv[46] addresses + */ + for (ii = 0; + (tmpipdef = virNetworkDefGetIpByIndex(network->def, AF_UNSPEC, ii)); + ii++) { + char *ipaddr = virSocketFormatAddr(&tmpipdef->address); + if (!ipaddr) + goto cleanup; + virCommandAddArgList(cmd, "--listen-address", ipaddr, NULL); + VIR_FREE(ipaddr); + } + for (r = 0 ; r < ipdef->nranges ; r++) { char *saddr = virSocketFormatAddr(&ipdef->ranges[r].start); if (!saddr) @@ -1027,9 +1037,30 @@ networkAddGeneralIp6tablesRules(struct network_driver *driver, goto err3; } + /* allow DNS over IPv6 */ + if (iptablesAddTcpInput(driver->iptables, AF_INET6, + network->def->bridge, 53) < 0) { + networkReportError(VIR_ERR_SYSTEM_ERROR, + _("failed to add ip6tables rule to allow DNS requests from '%s'"), + network->def->bridge); + goto err4; + } + + if (iptablesAddUdpInput(driver->iptables, AF_INET6, + network->def->bridge, 53) < 0) { + networkReportError(VIR_ERR_SYSTEM_ERROR, + _("failed to add ip6tables rule to allow DNS requests from '%s'"), + network->def->bridge); + goto err5; + } + return 0; /* unwind in reverse order from the point of failure */ +err5: + iptablesRemoveTcpInput(driver->iptables, AF_INET6, network->def->bridge, 53); +err4: + iptablesRemoveForwardAllowCross(driver->iptables, AF_INET6, network->def->bridge); err3: iptablesRemoveForwardRejectIn(driver->iptables, AF_INET6, network->def->bridge); err2: -- 1.7.3.5

13 years, 3 months

2
1
0 / 0

[libvirt] [PATCH] Move all the QEMU migration code to a new file

by Daniel P. Berrange

The introduction of the v3 migration protocol, along with support for migration cookies, will significantly expand the size of the migration code. Move it all to a separate file to make it more manageable The functions are not moved 100%. The API entry points remain in the main QEMU driver, but once the public virDomainPtr is resolved to the internal virDomainObjPtr, all following code is moved. This will allow the new v3 API entry points to call into the same shared internal migration functions * src/qemu/qemu_domain.c, src/qemu/qemu_domain.h: Add qemuDomainFormatXML helper method * src/qemu/qemu_driver.c: Remove all migration code * src/qemu/qemu_migration.c, src/qemu/qemu_migration.h: Add all migration code. --- src/Makefile.am | 3 +- src/qemu/qemu_domain.c | 39 ++ src/qemu/qemu_domain.h | 4 + src/qemu/qemu_driver.c | 1296 ++------------------------------------------- src/qemu/qemu_migration.c | 1295 ++++++++++++++++++++++++++++++++++++++++++++ src/qemu/qemu_migration.h | 63 +++ 6 files changed, 1444 insertions(+), 1256 deletions(-) create mode 100644 src/qemu/qemu_migration.c create mode 100644 src/qemu/qemu_migration.h diff --git a/src/Makefile.am b/src/Makefile.am index 9ef1d36..e4440dc 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -281,7 +281,8 @@ QEMU_DRIVER_SOURCES = \ qemu/qemu_hostdev.c qemu/qemu_hostdev.h \ qemu/qemu_hotplug.c qemu/qemu_hotplug.h \ qemu/qemu_conf.c qemu/qemu_conf.h \ - qemu/qemu_process.c qemu/qemu_process.h \ + qemu/qemu_process.c qemu/qemu_process.h \ + qemu/qemu_migration.c qemu/qemu_migration.h \ qemu/qemu_monitor.c qemu/qemu_monitor.h \ qemu/qemu_monitor_text.c \ qemu/qemu_monitor_text.h \ diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index e3163ab..8a2b9cc 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -30,6 +30,7 @@ #include "virterror_internal.h" #include "c-ctype.h" #include "event.h" +#include "cpu/cpu.h" #include <sys/time.h> @@ -653,3 +654,41 @@ void qemuDomainObjExitRemoteWithDriver(struct qemud_driver *driver, virDomainObjLock(obj); virDomainObjUnref(obj); } + + +char *qemuDomainFormatXML(struct qemud_driver *driver, + virDomainObjPtr vm, + int flags) +{ + char *ret = NULL; + virCPUDefPtr cpu = NULL; + virDomainDefPtr def; + virCPUDefPtr def_cpu; + + if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) + def = vm->newDef; + else + def = vm->def; + def_cpu = def->cpu; + + /* Update guest CPU requirements according to host CPU */ + if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) && def_cpu && def_cpu->model) { + if (!driver->caps || !driver->caps->host.cpu) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("cannot get host CPU capabilities")); + goto cleanup; + } + + if (!(cpu = virCPUDefCopy(def_cpu)) + || cpuUpdate(cpu, driver->caps->host.cpu)) + goto cleanup; + def->cpu = cpu; + } + + ret = virDomainDefFormat(def, flags); + +cleanup: + def->cpu = def_cpu; + virCPUDefFree(cpu); + return ret; +} diff --git a/src/qemu/qemu_domain.h b/src/qemu/qemu_domain.h index 4333a78..ebb2050 100644 --- a/src/qemu/qemu_domain.h +++ b/src/qemu/qemu_domain.h @@ -107,4 +107,8 @@ void qemuDomainObjEnterRemoteWithDriver(struct qemud_driver *driver, void qemuDomainObjExitRemoteWithDriver(struct qemud_driver *driver, virDomainObjPtr obj); +char *qemuDomainFormatXML(struct qemud_driver *driver, + virDomainObjPtr vm, + int flags); + #endif /* __QEMU_DOMAIN_H__ */ diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index dcc1050..aaf2ea1 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -57,6 +57,7 @@ #include "qemu_bridge_filter.h" #include "qemu_audit.h" #include "qemu_process.h" +#include "qemu_migration.h" #include "virterror_internal.h" #include "logging.h" @@ -1676,176 +1677,6 @@ cleanup: } -/** qemuDomainMigrateOffline: - * Pause domain for non-live migration. - */ -static int -qemuDomainMigrateOffline(struct qemud_driver *driver, - virDomainObjPtr vm) -{ - int ret; - - ret = qemuProcessStopCPUs(driver, vm); - if (ret == 0) { - virDomainEventPtr event; - - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_SUSPENDED, - VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED); - if (event) - qemuDomainEventQueue(driver, event); - } - - return ret; -} - - -static int -qemuDomainWaitForMigrationComplete(struct qemud_driver *driver, virDomainObjPtr vm) -{ - int ret = -1; - int status; - unsigned long long memProcessed; - unsigned long long memRemaining; - unsigned long long memTotal; - qemuDomainObjPrivatePtr priv = vm->privateData; - - priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; - - while (priv->jobInfo.type == VIR_DOMAIN_JOB_UNBOUNDED) { - /* Poll every 50ms for progress & to allow cancellation */ - struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull }; - struct timeval now; - int rc; - const char *job; - - switch (priv->jobActive) { - case QEMU_JOB_MIGRATION_OUT: - job = _("migration job"); - break; - case QEMU_JOB_SAVE: - job = _("domain save job"); - break; - case QEMU_JOB_DUMP: - job = _("domain core dump job"); - break; - default: - job = _("job"); - } - - - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), - job, _("guest unexpectedly quit")); - goto cleanup; - } - - if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) { - priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL; - VIR_DEBUG0("Cancelling job at client request"); - qemuDomainObjEnterMonitorWithDriver(driver, vm); - rc = qemuMonitorMigrateCancel(priv->mon); - qemuDomainObjExitMonitorWithDriver(driver, vm); - if (rc < 0) { - VIR_WARN0("Unable to cancel job"); - } - } else if (priv->jobSignals & QEMU_JOB_SIGNAL_SUSPEND) { - priv->jobSignals ^= QEMU_JOB_SIGNAL_SUSPEND; - VIR_DEBUG0("Pausing domain for non-live migration"); - if (qemuDomainMigrateOffline(driver, vm) < 0) - VIR_WARN0("Unable to pause domain"); - } else if (priv->jobSignals & QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME) { - unsigned long long ms = priv->jobSignalsData.migrateDowntime; - - priv->jobSignals ^= QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME; - priv->jobSignalsData.migrateDowntime = 0; - VIR_DEBUG("Setting migration downtime to %llums", ms); - qemuDomainObjEnterMonitorWithDriver(driver, vm); - rc = qemuMonitorSetMigrationDowntime(priv->mon, ms); - qemuDomainObjExitMonitorWithDriver(driver, vm); - if (rc < 0) - VIR_WARN0("Unable to set migration downtime"); - } - - /* Repeat check because the job signals might have caused - * guest to die - */ - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), - job, _("guest unexpectedly quit")); - goto cleanup; - } - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - rc = qemuMonitorGetMigrationStatus(priv->mon, - &status, - &memProcessed, - &memRemaining, - &memTotal); - qemuDomainObjExitMonitorWithDriver(driver, vm); - - if (rc < 0) { - priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED; - goto cleanup; - } - - if (gettimeofday(&now, NULL) < 0) { - priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED; - virReportSystemError(errno, "%s", - _("cannot get time of day")); - goto cleanup; - } - priv->jobInfo.timeElapsed = timeval_to_ms(now) - priv->jobStart; - - switch (status) { - case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE: - priv->jobInfo.type = VIR_DOMAIN_JOB_NONE; - qemuReportError(VIR_ERR_OPERATION_FAILED, - _("%s: %s"), job, _("is not active")); - break; - - case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE: - priv->jobInfo.dataTotal = memTotal; - priv->jobInfo.dataRemaining = memRemaining; - priv->jobInfo.dataProcessed = memProcessed; - - priv->jobInfo.memTotal = memTotal; - priv->jobInfo.memRemaining = memRemaining; - priv->jobInfo.memProcessed = memProcessed; - break; - - case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED: - priv->jobInfo.type = VIR_DOMAIN_JOB_COMPLETED; - ret = 0; - break; - - case QEMU_MONITOR_MIGRATION_STATUS_ERROR: - priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED; - qemuReportError(VIR_ERR_OPERATION_FAILED, - _("%s: %s"), job, _("unexpectedly failed")); - break; - - case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED: - priv->jobInfo.type = VIR_DOMAIN_JOB_CANCELLED; - qemuReportError(VIR_ERR_OPERATION_FAILED, - _("%s: %s"), job, _("canceled by client")); - break; - } - - virDomainObjUnlock(vm); - qemuDriverUnlock(driver); - - nanosleep(&ts, NULL); - - qemuDriverLock(driver); - virDomainObjLock(vm); - } - -cleanup: - return ret; -} - - #define QEMUD_SAVE_MAGIC "LibvirtQemudSave" #define QEMUD_SAVE_VERSION 2 @@ -2146,7 +1977,7 @@ static int qemudDomainSaveFlag(struct qemud_driver *driver, virDomainPtr dom, if (rc < 0) goto endjob; - rc = qemuDomainWaitForMigrationComplete(driver, vm); + rc = qemuMigrationWaitForCompletion(driver, vm); if (rc < 0) goto endjob; @@ -2454,7 +2285,7 @@ static int doCoreDump(struct qemud_driver *driver, if (ret < 0) goto cleanup; - ret = qemuDomainWaitForMigrationComplete(driver, vm); + ret = qemuMigrationWaitForCompletion(driver, vm); if (ret < 0) goto cleanup; @@ -3590,44 +3421,6 @@ cleanup: } -static char *qemudVMDumpXML(struct qemud_driver *driver, - virDomainObjPtr vm, - int flags) -{ - char *ret = NULL; - virCPUDefPtr cpu = NULL; - virDomainDefPtr def; - virCPUDefPtr def_cpu; - - if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) - def = vm->newDef; - else - def = vm->def; - def_cpu = def->cpu; - - /* Update guest CPU requirements according to host CPU */ - if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) && def_cpu && def_cpu->model) { - if (!driver->caps || !driver->caps->host.cpu) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("cannot get host CPU capabilities")); - goto cleanup; - } - - if (!(cpu = virCPUDefCopy(def_cpu)) - || cpuUpdate(cpu, driver->caps->host.cpu)) - goto cleanup; - def->cpu = cpu; - } - - ret = virDomainDefFormat(def, flags); - -cleanup: - def->cpu = def_cpu; - virCPUDefFree(cpu); - return ret; -} - - static char *qemudDomainDumpXML(virDomainPtr dom, int flags) { struct qemud_driver *driver = dom->conn->privateData; @@ -3673,7 +3466,7 @@ static char *qemudDomainDumpXML(virDomainPtr dom, } } - ret = qemudVMDumpXML(driver, vm, flags); + ret = qemuDomainFormatXML(driver, vm, flags); cleanup: if (vm) @@ -5537,18 +5330,6 @@ qemuDomainEventDeregisterAny(virConnectPtr conn, /* Migration support. */ -static bool ATTRIBUTE_NONNULL(1) -qemuDomainIsMigratable(virDomainDefPtr def) -{ - if (def->nhostdevs > 0) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("Domain with assigned host devices cannot be migrated")); - return false; - } - - return true; -} - /* Prepare is the first step, and it runs on the destination host. * * This version starts an empty VM listening on a localhost TCP port, and @@ -5563,24 +5344,8 @@ qemudDomainMigratePrepareTunnel(virConnectPtr dconn, const char *dom_xml) { struct qemud_driver *driver = dconn->privateData; - virDomainDefPtr def = NULL; - virDomainObjPtr vm = NULL; - char *migrateFrom; - virDomainEventPtr event = NULL; int ret = -1; - int internalret; - char *unixfile = NULL; - unsigned long long qemuCmdFlags; - qemuDomainObjPrivatePtr priv = NULL; - struct timeval now; - - if (gettimeofday(&now, NULL) < 0) { - virReportSystemError(errno, "%s", - _("cannot get time of day")); - return -1; - } - qemuDriverLock(driver); if (!dom_xml) { qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("no domain XML passed")); @@ -5597,139 +5362,12 @@ qemudDomainMigratePrepareTunnel(virConnectPtr dconn, goto cleanup; } - /* Parse the domain XML. */ - if (!(def = virDomainDefParseString(driver->caps, dom_xml, - VIR_DOMAIN_XML_INACTIVE))) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("failed to parse XML")); - goto cleanup; - } - - if (!qemuDomainIsMigratable(def)) - goto cleanup; - - /* Target domain name, maybe renamed. */ - if (dname) { - VIR_FREE(def->name); - def->name = strdup(dname); - if (def->name == NULL) - goto cleanup; - } - - if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0) - goto cleanup; - - if (!(vm = virDomainAssignDef(driver->caps, - &driver->domains, - def, true))) { - /* virDomainAssignDef already set the error */ - goto cleanup; - } - def = NULL; - priv = vm->privateData; - - if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) - goto cleanup; - priv->jobActive = QEMU_JOB_MIGRATION_OUT; - - /* Domain starts inactive, even if the domain XML had an id field. */ - vm->def->id = -1; - - if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.dest.%s", - driver->libDir, vm->def->name) < 0) { - virReportOOMError(); - goto endjob; - } - unlink(unixfile); - - /* check that this qemu version supports the interactive exec */ - if (qemuCapsExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Cannot determine QEMU argv syntax %s"), - vm->def->emulator); - goto endjob; - } - if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) - internalret = virAsprintf(&migrateFrom, "unix:%s", unixfile); - else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) - internalret = virAsprintf(&migrateFrom, "exec:nc -U -l %s", unixfile); - else { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("Destination qemu is too old to support tunnelled migration")); - goto endjob; - } - if (internalret < 0) { - virReportOOMError(); - goto endjob; - } - /* Start the QEMU daemon, with the same command-line arguments plus - * -incoming unix:/path/to/file or exec:nc -U /path/to/file - */ - internalret = qemuProcessStart(dconn, driver, vm, migrateFrom, true, - -1, NULL, VIR_VM_OP_MIGRATE_IN_START); - VIR_FREE(migrateFrom); - if (internalret < 0) { - qemuDomainStartAudit(vm, "migrated", false); - /* Note that we don't set an error here because qemuProcessStart - * should have already done that. - */ - if (!vm->persistent) { - virDomainRemoveInactive(&driver->domains, vm); - vm = NULL; - } - goto endjob; - } - - if (virFDStreamConnectUNIX(st, - unixfile, - false) < 0) { - qemuDomainStartAudit(vm, "migrated", false); - qemuProcessStop(driver, vm, 0); - if (!vm->persistent) { - if (qemuDomainObjEndJob(vm) > 0) - virDomainRemoveInactive(&driver->domains, vm); - vm = NULL; - } - virReportSystemError(errno, - _("cannot open unix socket '%s' for tunnelled migration"), - unixfile); - goto endjob; - } - - qemuDomainStartAudit(vm, "migrated", true); - - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_STARTED, - VIR_DOMAIN_EVENT_STARTED_MIGRATED); - ret = 0; - -endjob: - if (vm && - qemuDomainObjEndJob(vm) == 0) - vm = NULL; - - /* We set a fake job active which is held across - * API calls until the finish() call. This prevents - * any other APIs being invoked while incoming - * migration is taking place - */ - if (vm && - virDomainObjIsActive(vm)) { - priv->jobActive = QEMU_JOB_MIGRATION_IN; - priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; - priv->jobStart = timeval_to_ms(now); - } + qemuDriverLock(driver); + ret = qemuMigrationPrepareTunnel(driver, dconn, st, + dname, dom_xml); + qemuDriverUnlock(driver); cleanup: - virDomainDefFree(def); - if (unixfile) - unlink(unixfile); - VIR_FREE(unixfile); - if (vm) - virDomainObjUnlock(vm); - if (event) - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); return ret; } @@ -5748,25 +5386,8 @@ qemudDomainMigratePrepare2 (virConnectPtr dconn, unsigned long resource ATTRIBUTE_UNUSED, const char *dom_xml) { - static int port = 0; struct qemud_driver *driver = dconn->privateData; - virDomainDefPtr def = NULL; - virDomainObjPtr vm = NULL; - int this_port; - char *hostname = NULL; - char migrateFrom [64]; - const char *p; - virDomainEventPtr event = NULL; int ret = -1; - int internalret; - qemuDomainObjPrivatePtr priv = NULL; - struct timeval now; - - if (gettimeofday(&now, NULL) < 0) { - virReportSystemError(errno, "%s", - _("cannot get time of day")); - return -1; - } virCheckFlags(VIR_MIGRATE_LIVE | VIR_MIGRATE_PEER2PEER | @@ -5795,666 +5416,38 @@ qemudDomainMigratePrepare2 (virConnectPtr dconn, goto cleanup; } - /* The URI passed in may be NULL or a string "tcp://somehostname:port". - * - * If the URI passed in is NULL then we allocate a port number - * from our pool of port numbers and return a URI of - * "tcp://ourhostname:port". - * - * If the URI passed in is not NULL then we try to parse out the - * port number and use that (note that the hostname is assumed - * to be a correct hostname which refers to the target machine). - */ - if (uri_in == NULL) { - this_port = QEMUD_MIGRATION_FIRST_PORT + port++; - if (port == QEMUD_MIGRATION_NUM_PORTS) port = 0; - - /* Get hostname */ - if ((hostname = virGetHostname(NULL)) == NULL) - goto cleanup; - - if (STRPREFIX(hostname, "localhost")) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("hostname on destination resolved to localhost, but migration requires an FQDN")); - goto cleanup; - } - - /* XXX this really should have been a properly well-formed - * URI, but we can't add in tcp:// now without breaking - * compatability with old targets. We at least make the - * new targets accept both syntaxes though. - */ - /* Caller frees */ - internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port); - if (internalret < 0) { - virReportOOMError(); - goto cleanup; - } - } else { - /* Check the URI starts with "tcp:". We will escape the - * URI when passing it to the qemu monitor, so bad - * characters in hostname part don't matter. - */ - if (!STRPREFIX (uri_in, "tcp:")) { - qemuReportError (VIR_ERR_INVALID_ARG, - "%s", _("only tcp URIs are supported for KVM/QEMU migrations")); - goto cleanup; - } + ret = qemuMigrationPrepareDirect(driver, dconn, + uri_in, uri_out, + dname, dom_xml); - /* Get the port number. */ - p = strrchr (uri_in, ':'); - if (p == strchr(uri_in, ':')) { - /* Generate a port */ - this_port = QEMUD_MIGRATION_FIRST_PORT + port++; - if (port == QEMUD_MIGRATION_NUM_PORTS) - port = 0; +cleanup: + qemuDriverUnlock(driver); + return ret; +} - /* Caller frees */ - if (virAsprintf(uri_out, "%s:%d", uri_in, this_port) < 0) { - virReportOOMError(); - goto cleanup; - } - } else { - p++; /* definitely has a ':' in it, see above */ - this_port = virParseNumber (&p); - if (this_port == -1 || p-uri_in != strlen (uri_in)) { - qemuReportError(VIR_ERR_INVALID_ARG, - "%s", _("URI ended with incorrect ':port'")); - goto cleanup; - } - } - } +/* Perform is the second step, and it runs on the source host. */ +static int +qemudDomainMigratePerform (virDomainPtr dom, + const char *cookie ATTRIBUTE_UNUSED, + int cookielen ATTRIBUTE_UNUSED, + const char *uri, + unsigned long flags, + const char *dname, + unsigned long resource) +{ + struct qemud_driver *driver = dom->conn->privateData; + virDomainObjPtr vm; + int ret = -1; - if (*uri_out) - VIR_DEBUG("Generated uri_out=%s", *uri_out); - - /* Parse the domain XML. */ - if (!(def = virDomainDefParseString(driver->caps, dom_xml, - VIR_DOMAIN_XML_INACTIVE))) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("failed to parse XML")); - goto cleanup; - } - - if (!qemuDomainIsMigratable(def)) - goto cleanup; - - /* Target domain name, maybe renamed. */ - if (dname) { - VIR_FREE(def->name); - def->name = strdup(dname); - if (def->name == NULL) - goto cleanup; - } - - if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0) - goto cleanup; - - if (!(vm = virDomainAssignDef(driver->caps, - &driver->domains, - def, true))) { - /* virDomainAssignDef already set the error */ - goto cleanup; - } - def = NULL; - priv = vm->privateData; - - if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) - goto cleanup; - priv->jobActive = QEMU_JOB_MIGRATION_OUT; - - /* Domain starts inactive, even if the domain XML had an id field. */ - vm->def->id = -1; - - /* Start the QEMU daemon, with the same command-line arguments plus - * -incoming tcp:0.0.0.0:port - */ - snprintf (migrateFrom, sizeof (migrateFrom), "tcp:0.0.0.0:%d", this_port); - if (qemuProcessStart(dconn, driver, vm, migrateFrom, true, - -1, NULL, VIR_VM_OP_MIGRATE_IN_START) < 0) { - qemuDomainStartAudit(vm, "migrated", false); - /* Note that we don't set an error here because qemuProcessStart - * should have already done that. - */ - if (!vm->persistent) { - if (qemuDomainObjEndJob(vm) > 0) - virDomainRemoveInactive(&driver->domains, vm); - vm = NULL; - } - goto endjob; - } - - qemuDomainStartAudit(vm, "migrated", true); - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_STARTED, - VIR_DOMAIN_EVENT_STARTED_MIGRATED); - ret = 0; - -endjob: - if (vm && - qemuDomainObjEndJob(vm) == 0) - vm = NULL; - - /* We set a fake job active which is held across - * API calls until the finish() call. This prevents - * any other APIs being invoked while incoming - * migration is taking place - */ - if (vm && - virDomainObjIsActive(vm)) { - priv->jobActive = QEMU_JOB_MIGRATION_IN; - priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; - priv->jobStart = timeval_to_ms(now); - } - -cleanup: - VIR_FREE(hostname); - virDomainDefFree(def); - if (ret != 0) - VIR_FREE(*uri_out); - if (vm) - virDomainObjUnlock(vm); - if (event) - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); - return ret; - -} - - -/* Perform migration using QEMU's native TCP migrate support, - * not encrypted obviously - */ -static int doNativeMigrate(struct qemud_driver *driver, - virDomainObjPtr vm, - const char *uri, - unsigned int flags, - const char *dname ATTRIBUTE_UNUSED, - unsigned long resource) -{ - int ret = -1; - xmlURIPtr uribits = NULL; - qemuDomainObjPrivatePtr priv = vm->privateData; - unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND; - - /* Issue the migrate command. */ - if (STRPREFIX(uri, "tcp:") && !STRPREFIX(uri, "tcp://")) { - /* HACK: source host generates bogus URIs, so fix them up */ - char *tmpuri; - if (virAsprintf(&tmpuri, "tcp://%s", uri + strlen("tcp:")) < 0) { - virReportOOMError(); - goto cleanup; - } - uribits = xmlParseURI(tmpuri); - VIR_FREE(tmpuri); - } else { - uribits = xmlParseURI(uri); - } - if (!uribits) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot parse URI %s"), uri); - goto cleanup; - } - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - if (resource > 0 && - qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) { - qemuDomainObjExitMonitorWithDriver(driver, vm); - goto cleanup; - } - - if (flags & VIR_MIGRATE_NON_SHARED_DISK) - background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK; - - if (flags & VIR_MIGRATE_NON_SHARED_INC) - background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC; - - if (qemuMonitorMigrateToHost(priv->mon, background_flags, uribits->server, - uribits->port) < 0) { - qemuDomainObjExitMonitorWithDriver(driver, vm); - goto cleanup; - } - qemuDomainObjExitMonitorWithDriver(driver, vm); - - if (qemuDomainWaitForMigrationComplete(driver, vm) < 0) - goto cleanup; - - ret = 0; - -cleanup: - xmlFreeURI(uribits); - return ret; -} - - -#define TUNNEL_SEND_BUF_SIZE 65536 - -static int doTunnelSendAll(virStreamPtr st, - int sock) -{ - char *buffer; - int nbytes = TUNNEL_SEND_BUF_SIZE; - - if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) { - virReportOOMError(); - virStreamAbort(st); - return -1; - } - - /* XXX should honour the 'resource' parameter here */ - for (;;) { - nbytes = saferead(sock, buffer, nbytes); - if (nbytes < 0) { - virReportSystemError(errno, "%s", - _("tunnelled migration failed to read from qemu")); - virStreamAbort(st); - VIR_FREE(buffer); - return -1; - } - else if (nbytes == 0) - /* EOF; get out of here */ - break; - - if (virStreamSend(st, buffer, nbytes) < 0) { - qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", - _("Failed to write migration data to remote libvirtd")); - VIR_FREE(buffer); - return -1; - } - } - - VIR_FREE(buffer); - - if (virStreamFinish(st) < 0) - /* virStreamFinish set the error for us */ - return -1; - - return 0; -} - -static int doTunnelMigrate(virDomainPtr dom, - struct qemud_driver *driver, - virConnectPtr dconn, - virDomainObjPtr vm, - const char *dom_xml, - const char *uri, - unsigned long flags, - const char *dname, - unsigned long resource) -{ - qemuDomainObjPrivatePtr priv = vm->privateData; - int client_sock = -1; - int qemu_sock = -1; - struct sockaddr_un sa_qemu, sa_client; - socklen_t addrlen; - virDomainPtr ddomain = NULL; - int retval = -1; - virStreamPtr st = NULL; - char *unixfile = NULL; - int internalret; - unsigned long long qemuCmdFlags; - int status; - unsigned long long transferred, remaining, total; - unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND; - - /* - * The order of operations is important here to avoid touching - * the source VM until we are very sure we can successfully - * start the migration operation. - * - * 1. setup local support infrastructure (eg sockets) - * 2. setup destination fully - * 3. start migration on source - */ - - - /* Stage 1. setup local support infrastructure */ - - if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s", - driver->libDir, vm->def->name) < 0) { - virReportOOMError(); - goto cleanup; - } - - qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0); - if (qemu_sock < 0) { - virReportSystemError(errno, "%s", - _("cannot open tunnelled migration socket")); - goto cleanup; - } - memset(&sa_qemu, 0, sizeof(sa_qemu)); - sa_qemu.sun_family = AF_UNIX; - if (virStrcpy(sa_qemu.sun_path, unixfile, - sizeof(sa_qemu.sun_path)) == NULL) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Unix socket '%s' too big for destination"), - unixfile); - goto cleanup; - } - unlink(unixfile); - if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) { - virReportSystemError(errno, - _("Cannot bind to unix socket '%s' for tunnelled migration"), - unixfile); - goto cleanup; - } - if (listen(qemu_sock, 1) < 0) { - virReportSystemError(errno, - _("Cannot listen on unix socket '%s' for tunnelled migration"), - unixfile); - goto cleanup; - } - - if (chown(unixfile, qemu_driver->user, qemu_driver->group) < 0) { - virReportSystemError(errno, - _("Cannot change unix socket '%s' owner"), - unixfile); - goto cleanup; - } - - /* check that this qemu version supports the unix migration */ - if (qemuCapsExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Cannot extract Qemu version from '%s'"), - vm->def->emulator); - goto cleanup; - } - - if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) && - !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("Source qemu is too old to support tunnelled migration")); - goto cleanup; - } - - - /* Stage 2. setup destination fully - * - * Once stage 2 has completed successfully, we *must* call finish - * to cleanup the target whether we succeed or fail - */ - st = virStreamNew(dconn, 0); - if (st == NULL) - /* virStreamNew only fails on OOM, and it reports the error itself */ - goto cleanup; - - qemuDomainObjEnterRemoteWithDriver(driver, vm); - internalret = dconn->driver->domainMigratePrepareTunnel(dconn, st, - flags, dname, - resource, dom_xml); - qemuDomainObjExitRemoteWithDriver(driver, vm); - - if (internalret < 0) - /* domainMigratePrepareTunnel sets the error for us */ - goto cleanup; - - /* the domain may have shutdown or crashed while we had the locks dropped - * in qemuDomainObjEnterRemoteWithDriver, so check again - */ - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("guest unexpectedly quit")); - goto cleanup; - } - - /* 3. start migration on source */ - qemuDomainObjEnterMonitorWithDriver(driver, vm); - if (flags & VIR_MIGRATE_NON_SHARED_DISK) - background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK; - if (flags & VIR_MIGRATE_NON_SHARED_INC) - background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC; - if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX){ - internalret = qemuMonitorMigrateToUnix(priv->mon, background_flags, - unixfile); - } - else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) { - const char *args[] = { "nc", "-U", unixfile, NULL }; - internalret = qemuMonitorMigrateToCommand(priv->mon, QEMU_MONITOR_MIGRATE_BACKGROUND, args); - } else { - internalret = -1; - } - qemuDomainObjExitMonitorWithDriver(driver, vm); - if (internalret < 0) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("tunnelled migration monitor command failed")); - goto finish; - } - - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("guest unexpectedly quit")); - goto cleanup; - } - - /* From this point onwards we *must* call cancel to abort the - * migration on source if anything goes wrong */ - - /* it is also possible that the migrate didn't fail initially, but - * rather failed later on. Check the output of "info migrate" - */ - qemuDomainObjEnterMonitorWithDriver(driver, vm); - if (qemuMonitorGetMigrationStatus(priv->mon, - &status, - &transferred, - &remaining, - &total) < 0) { - qemuDomainObjExitMonitorWithDriver(driver, vm); - goto cancel; - } - qemuDomainObjExitMonitorWithDriver(driver, vm); - - if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s",_("migrate failed")); - goto cancel; - } - - addrlen = sizeof(sa_client); - while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) { - if (errno == EAGAIN || errno == EINTR) - continue; - virReportSystemError(errno, "%s", - _("tunnelled migration failed to accept from qemu")); - goto cancel; - } - - retval = doTunnelSendAll(st, client_sock); - -cancel: - if (retval != 0 && virDomainObjIsActive(vm)) { - qemuDomainObjEnterMonitorWithDriver(driver, vm); - qemuMonitorMigrateCancel(priv->mon); - qemuDomainObjExitMonitorWithDriver(driver, vm); - } - -finish: - dname = dname ? dname : dom->name; - qemuDomainObjEnterRemoteWithDriver(driver, vm); - ddomain = dconn->driver->domainMigrateFinish2 - (dconn, dname, NULL, 0, uri, flags, retval); - qemuDomainObjExitRemoteWithDriver(driver, vm); - -cleanup: - VIR_FORCE_CLOSE(client_sock); - VIR_FORCE_CLOSE(qemu_sock); - - if (ddomain) - virUnrefDomain(ddomain); - - if (unixfile) { - unlink(unixfile); - VIR_FREE(unixfile); - } - - if (st) - /* don't call virStreamFree(), because that resets any pending errors */ - virUnrefStream(st); - return retval; -} - - -/* This is essentially a simplified re-impl of - * virDomainMigrateVersion2 from libvirt.c, but running in source - * libvirtd context, instead of client app context */ -static int doNonTunnelMigrate(virDomainPtr dom, - struct qemud_driver *driver, - virConnectPtr dconn, - virDomainObjPtr vm, - const char *dom_xml, - const char *uri ATTRIBUTE_UNUSED, - unsigned long flags, - const char *dname, - unsigned long resource) -{ - virDomainPtr ddomain = NULL; - int retval = -1; - char *uri_out = NULL; - int rc; - - qemuDomainObjEnterRemoteWithDriver(driver, vm); - /* NB we don't pass 'uri' into this, since that's the libvirtd - * URI in this context - so we let dest pick it */ - rc = dconn->driver->domainMigratePrepare2(dconn, - NULL, /* cookie */ - 0, /* cookielen */ - NULL, /* uri */ - &uri_out, - flags, dname, - resource, dom_xml); - qemuDomainObjExitRemoteWithDriver(driver, vm); - if (rc < 0) - /* domainMigratePrepare2 sets the error for us */ - goto cleanup; - - /* the domain may have shutdown or crashed while we had the locks dropped - * in qemuDomainObjEnterRemoteWithDriver, so check again - */ - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("guest unexpectedly quit")); - goto cleanup; - } - - if (uri_out == NULL) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("domainMigratePrepare2 did not set uri")); - goto cleanup; - } - - if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0) - goto finish; - - retval = 0; - -finish: - dname = dname ? dname : dom->name; - qemuDomainObjEnterRemoteWithDriver(driver, vm); - ddomain = dconn->driver->domainMigrateFinish2 - (dconn, dname, NULL, 0, uri_out, flags, retval); - qemuDomainObjExitRemoteWithDriver(driver, vm); - - if (ddomain) - virUnrefDomain(ddomain); - -cleanup: - return retval; -} - - -static int doPeer2PeerMigrate(virDomainPtr dom, - struct qemud_driver *driver, - virDomainObjPtr vm, - const char *uri, - unsigned long flags, - const char *dname, - unsigned long resource) -{ - int ret = -1; - virConnectPtr dconn = NULL; - char *dom_xml; - bool p2p; - - /* the order of operations is important here; we make sure the - * destination side is completely setup before we touch the source - */ - - qemuDomainObjEnterRemoteWithDriver(driver, vm); - dconn = virConnectOpen(uri); - qemuDomainObjExitRemoteWithDriver(driver, vm); - if (dconn == NULL) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - _("Failed to connect to remote libvirt URI %s"), uri); - return -1; - } - - qemuDomainObjEnterRemoteWithDriver(driver, vm); - p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn, - VIR_DRV_FEATURE_MIGRATION_P2P); - qemuDomainObjExitRemoteWithDriver(driver, vm); - if (!p2p) { - qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", - _("Destination libvirt does not support peer-to-peer migration protocol")); - goto cleanup; - } - - /* domain may have been stopped while we were talking to remote daemon */ - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("guest unexpectedly quit")); - goto cleanup; - } - - dom_xml = qemudVMDumpXML(driver, vm, - VIR_DOMAIN_XML_SECURE | - VIR_DOMAIN_XML_UPDATE_CPU); - if (!dom_xml) { - qemuReportError(VIR_ERR_OPERATION_FAILED, - "%s", _("failed to get domain xml")); - goto cleanup; - } - - if (flags & VIR_MIGRATE_TUNNELLED) - ret = doTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource); - else - ret = doNonTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource); - -cleanup: - VIR_FREE(dom_xml); - /* don't call virConnectClose(), because that resets any pending errors */ - qemuDomainObjEnterRemoteWithDriver(driver, vm); - virUnrefConnect(dconn); - qemuDomainObjExitRemoteWithDriver(driver, vm); - - return ret; -} - - -/* Perform is the second step, and it runs on the source host. */ -static int -qemudDomainMigratePerform (virDomainPtr dom, - const char *cookie ATTRIBUTE_UNUSED, - int cookielen ATTRIBUTE_UNUSED, - const char *uri, - unsigned long flags, - const char *dname, - unsigned long resource) -{ - struct qemud_driver *driver = dom->conn->privateData; - virDomainObjPtr vm; - virDomainEventPtr event = NULL; - int ret = -1; - int resume = 0; - qemuDomainObjPrivatePtr priv; - - virCheckFlags(VIR_MIGRATE_LIVE | - VIR_MIGRATE_PEER2PEER | - VIR_MIGRATE_TUNNELLED | - VIR_MIGRATE_PERSIST_DEST | - VIR_MIGRATE_UNDEFINE_SOURCE | - VIR_MIGRATE_PAUSED | - VIR_MIGRATE_NON_SHARED_DISK | - VIR_MIGRATE_NON_SHARED_INC, -1); + virCheckFlags(VIR_MIGRATE_LIVE | + VIR_MIGRATE_PEER2PEER | + VIR_MIGRATE_TUNNELLED | + VIR_MIGRATE_PERSIST_DEST | + VIR_MIGRATE_UNDEFINE_SOURCE | + VIR_MIGRATE_PAUSED | + VIR_MIGRATE_NON_SHARED_DISK | + VIR_MIGRATE_NON_SHARED_INC, -1); qemuDriverLock(driver); vm = virDomainFindByUUID(&driver->domains, dom->uuid); @@ -6465,120 +5458,16 @@ qemudDomainMigratePerform (virDomainPtr dom, _("no domain with matching uuid '%s'"), uuidstr); goto cleanup; } - priv = vm->privateData; - - if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) - goto cleanup; - priv->jobActive = QEMU_JOB_MIGRATION_OUT; - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("domain is not running")); - goto endjob; - } - - memset(&priv->jobInfo, 0, sizeof(priv->jobInfo)); - priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; - - resume = vm->state == VIR_DOMAIN_RUNNING; - if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) { - if (qemuDomainMigrateOffline(driver, vm) < 0) - goto endjob; - } - - if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) { - if (doPeer2PeerMigrate(dom, driver, vm, uri, flags, dname, resource) < 0) - /* doPeer2PeerMigrate already set the error, so just get out */ - goto endjob; - } else { - if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0) - goto endjob; - } - - /* Clean up the source domain. */ - qemuProcessStop(driver, vm, 1); - qemuDomainStopAudit(vm, "migrated"); - resume = 0; - - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_STOPPED, - VIR_DOMAIN_EVENT_STOPPED_MIGRATED); - if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) { - virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm); - if (qemuDomainObjEndJob(vm) > 0) - virDomainRemoveInactive(&driver->domains, vm); - vm = NULL; - } - ret = 0; - -endjob: - if (resume && vm->state == VIR_DOMAIN_PAUSED) { - /* we got here through some sort of failure; start the domain again */ - if (qemuProcessStartCPUs(driver, vm, dom->conn) < 0) { - /* Hm, we already know we are in error here. We don't want to - * overwrite the previous error, though, so we just throw something - * to the logs and hope for the best - */ - VIR_ERROR(_("Failed to resume guest %s after failure"), - vm->def->name); - } - - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_RESUMED, - VIR_DOMAIN_EVENT_RESUMED_MIGRATED); - } - if (vm && - qemuDomainObjEndJob(vm) == 0) - vm = NULL; + ret = qemuMigrationPerform(driver, dom->conn, vm, + uri, flags, + dname, resource); cleanup: - if (vm) - virDomainObjUnlock(vm); - if (event) - qemuDomainEventQueue(driver, event); qemuDriverUnlock(driver); return ret; } -#if WITH_MACVTAP -static void -qemudVPAssociatePortProfiles(virDomainDefPtr def) { - int i; - int last_good_net = -1; - virDomainNetDefPtr net; - - for (i = 0; i < def->nnets; i++) { - net = def->nets[i]; - if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { - if (vpAssociatePortProfileId(net->ifname, - net->mac, - net->data.direct.linkdev, - &net->data.direct.virtPortProfile, - def->uuid, - VIR_VM_OP_MIGRATE_IN_FINISH) != 0) - goto err_exit; - } - last_good_net = i; - } - - return; - -err_exit: - for (i = 0; i < last_good_net; i++) { - net = def->nets[i]; - if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { - vpDisassociatePortProfileId(net->ifname, - net->mac, - net->data.direct.linkdev, - &net->data.direct.virtPortProfile, - VIR_VM_OP_MIGRATE_IN_FINISH); - } - } -} -#else /* !WITH_MACVTAP */ -static void -qemudVPAssociatePortProfiles(virDomainDefPtr def ATTRIBUTE_UNUSED) { } -#endif /* WITH_MACVTAP */ /* Finish is the third and final step, and it runs on the destination host. */ static virDomainPtr @@ -6593,10 +5482,7 @@ qemudDomainMigrateFinish2 (virConnectPtr dconn, struct qemud_driver *driver = dconn->privateData; virDomainObjPtr vm; virDomainPtr dom = NULL; - virDomainEventPtr event = NULL; virErrorPtr orig_err; - int newVM = 1; - qemuDomainObjPrivatePtr priv = NULL; virCheckFlags(VIR_MIGRATE_LIVE | VIR_MIGRATE_PEER2PEER | @@ -6618,118 +5504,18 @@ qemudDomainMigrateFinish2 (virConnectPtr dconn, goto cleanup; } - priv = vm->privateData; - if (priv->jobActive != QEMU_JOB_MIGRATION_IN) { - qemuReportError(VIR_ERR_NO_DOMAIN, - _("domain '%s' is not processing incoming migration"), dname); - goto cleanup; - } - priv->jobActive = QEMU_JOB_NONE; - memset(&priv->jobInfo, 0, sizeof(priv->jobInfo)); - - if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) - goto cleanup; - - /* Did the migration go as planned? If yes, return the domain - * object, but if no, clean up the empty qemu process. - */ - if (retcode == 0) { - if (!virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("guest unexpectedly quit")); - goto cleanup; - } - - qemudVPAssociatePortProfiles(vm->def); - - if (flags & VIR_MIGRATE_PERSIST_DEST) { - if (vm->persistent) - newVM = 0; - vm->persistent = 1; - - if (virDomainSaveConfig(driver->configDir, vm->def) < 0) { - /* Hmpf. Migration was successful, but making it persistent - * was not. If we report successful, then when this domain - * shuts down, management tools are in for a surprise. On the - * other hand, if we report failure, then the management tools - * might try to restart the domain on the source side, even - * though the domain is actually running on the destination. - * Return a NULL dom pointer, and hope that this is a rare - * situation and management tools are smart. - */ - vm = NULL; - goto endjob; - } - - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_DEFINED, - newVM ? - VIR_DOMAIN_EVENT_DEFINED_ADDED : - VIR_DOMAIN_EVENT_DEFINED_UPDATED); - if (event) - qemuDomainEventQueue(driver, event); - event = NULL; - - } - dom = virGetDomain (dconn, vm->def->name, vm->def->uuid); - - if (!(flags & VIR_MIGRATE_PAUSED)) { - /* run 'cont' on the destination, which allows migration on qemu - * >= 0.10.6 to work properly. This isn't strictly necessary on - * older qemu's, but it also doesn't hurt anything there - */ - if (qemuProcessStartCPUs(driver, vm, dconn) < 0) { - if (virGetLastError() == NULL) - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("resume operation failed")); - goto endjob; - } - } - - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_RESUMED, - VIR_DOMAIN_EVENT_RESUMED_MIGRATED); - if (vm->state == VIR_DOMAIN_PAUSED) { - qemuDomainEventQueue(driver, event); - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_SUSPENDED, - VIR_DOMAIN_EVENT_SUSPENDED_PAUSED); - } - if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) { - VIR_WARN("Failed to save status on vm %s", vm->def->name); - goto endjob; - } - } else { - qemuProcessStop(driver, vm, 1); - qemuDomainStopAudit(vm, "failed"); - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_STOPPED, - VIR_DOMAIN_EVENT_STOPPED_FAILED); - if (!vm->persistent) { - if (qemuDomainObjEndJob(vm) > 0) - virDomainRemoveInactive(&driver->domains, vm); - vm = NULL; - } - } - -endjob: - if (vm && - qemuDomainObjEndJob(vm) == 0) - vm = NULL; + dom = qemuMigrationFinish(driver, dconn, vm, flags, retcode); cleanup: if (orig_err) { virSetError(orig_err); virFreeError(orig_err); } - if (vm) - virDomainObjUnlock(vm); - if (event) - qemuDomainEventQueue(driver, event); qemuDriverUnlock(driver); return dom; } + static int qemudNodeDeviceGetPciInfo (virNodeDevicePtr dev, unsigned *domain, diff --git a/src/qemu/qemu_migration.c b/src/qemu/qemu_migration.c new file mode 100644 index 0000000..8d23cc5 --- /dev/null +++ b/src/qemu/qemu_migration.c @@ -0,0 +1,1295 @@ +/* + * qemu_migration.c: QEMU migration handling + * + * Copyright (C) 2006-2011 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 + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include <config.h> + +#include <sys/time.h> + +#include "qemu_migration.h" +#include "qemu_monitor.h" +#include "qemu_domain.h" +#include "qemu_process.h" +#include "qemu_capabilities.h" +#include "qemu_audit.h" + +#include "logging.h" +#include "virterror_internal.h" +#include "memory.h" +#include "util.h" +#include "files.h" +#include "datatypes.h" +#include "fdstream.h" + +#define VIR_FROM_THIS VIR_FROM_QEMU + +#define timeval_to_ms(tv) (((tv).tv_sec * 1000ull) + ((tv).tv_usec / 1000)) + + +bool +qemuMigrationIsAllowed(virDomainDefPtr def) +{ + if (def->nhostdevs > 0) { + qemuReportError(VIR_ERR_OPERATION_INVALID, + "%s", _("Domain with assigned host devices cannot be migrated")); + return false; + } + + return true; +} + +/** qemuMigrationSetOffline + * Pause domain for non-live migration. + */ +int +qemuMigrationSetOffline(struct qemud_driver *driver, + virDomainObjPtr vm) +{ + int ret; + + ret = qemuProcessStopCPUs(driver, vm); + if (ret == 0) { + virDomainEventPtr event; + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_SUSPENDED, + VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED); + if (event) + qemuDomainEventQueue(driver, event); + } + + return ret; +} + + +int +qemuMigrationWaitForCompletion(struct qemud_driver *driver, virDomainObjPtr vm) +{ + int ret = -1; + int status; + unsigned long long memProcessed; + unsigned long long memRemaining; + unsigned long long memTotal; + qemuDomainObjPrivatePtr priv = vm->privateData; + + priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; + + while (priv->jobInfo.type == VIR_DOMAIN_JOB_UNBOUNDED) { + /* Poll every 50ms for progress & to allow cancellation */ + struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull }; + struct timeval now; + int rc; + const char *job; + + switch (priv->jobActive) { + case QEMU_JOB_MIGRATION_OUT: + job = _("migration job"); + break; + case QEMU_JOB_SAVE: + job = _("domain save job"); + break; + case QEMU_JOB_DUMP: + job = _("domain core dump job"); + break; + default: + job = _("job"); + } + + + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), + job, _("guest unexpectedly quit")); + goto cleanup; + } + + if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) { + priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL; + VIR_DEBUG0("Cancelling job at client request"); + qemuDomainObjEnterMonitorWithDriver(driver, vm); + rc = qemuMonitorMigrateCancel(priv->mon); + qemuDomainObjExitMonitorWithDriver(driver, vm); + if (rc < 0) { + VIR_WARN0("Unable to cancel job"); + } + } else if (priv->jobSignals & QEMU_JOB_SIGNAL_SUSPEND) { + priv->jobSignals ^= QEMU_JOB_SIGNAL_SUSPEND; + VIR_DEBUG0("Pausing domain for non-live migration"); + if (qemuMigrationSetOffline(driver, vm) < 0) + VIR_WARN0("Unable to pause domain"); + } else if (priv->jobSignals & QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME) { + unsigned long long ms = priv->jobSignalsData.migrateDowntime; + + priv->jobSignals ^= QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME; + priv->jobSignalsData.migrateDowntime = 0; + VIR_DEBUG("Setting migration downtime to %llums", ms); + qemuDomainObjEnterMonitorWithDriver(driver, vm); + rc = qemuMonitorSetMigrationDowntime(priv->mon, ms); + qemuDomainObjExitMonitorWithDriver(driver, vm); + if (rc < 0) + VIR_WARN0("Unable to set migration downtime"); + } + + /* Repeat check because the job signals might have caused + * guest to die + */ + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), + job, _("guest unexpectedly quit")); + goto cleanup; + } + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + rc = qemuMonitorGetMigrationStatus(priv->mon, + &status, + &memProcessed, + &memRemaining, + &memTotal); + qemuDomainObjExitMonitorWithDriver(driver, vm); + + if (rc < 0) { + priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED; + goto cleanup; + } + + if (gettimeofday(&now, NULL) < 0) { + priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED; + virReportSystemError(errno, "%s", + _("cannot get time of day")); + goto cleanup; + } + priv->jobInfo.timeElapsed = timeval_to_ms(now) - priv->jobStart; + + switch (status) { + case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE: + priv->jobInfo.type = VIR_DOMAIN_JOB_NONE; + qemuReportError(VIR_ERR_OPERATION_FAILED, + _("%s: %s"), job, _("is not active")); + break; + + case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE: + priv->jobInfo.dataTotal = memTotal; + priv->jobInfo.dataRemaining = memRemaining; + priv->jobInfo.dataProcessed = memProcessed; + + priv->jobInfo.memTotal = memTotal; + priv->jobInfo.memRemaining = memRemaining; + priv->jobInfo.memProcessed = memProcessed; + break; + + case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED: + priv->jobInfo.type = VIR_DOMAIN_JOB_COMPLETED; + ret = 0; + break; + + case QEMU_MONITOR_MIGRATION_STATUS_ERROR: + priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED; + qemuReportError(VIR_ERR_OPERATION_FAILED, + _("%s: %s"), job, _("unexpectedly failed")); + break; + + case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED: + priv->jobInfo.type = VIR_DOMAIN_JOB_CANCELLED; + qemuReportError(VIR_ERR_OPERATION_FAILED, + _("%s: %s"), job, _("canceled by client")); + break; + } + + virDomainObjUnlock(vm); + qemuDriverUnlock(driver); + + nanosleep(&ts, NULL); + + qemuDriverLock(driver); + virDomainObjLock(vm); + } + +cleanup: + return ret; +} + + +/* Prepare is the first step, and it runs on the destination host. + * + * This version starts an empty VM listening on a localhost TCP port, and + * sets up the corresponding virStream to handle the incoming data. + */ +int +qemuMigrationPrepareTunnel(struct qemud_driver *driver, + virConnectPtr dconn, + virStreamPtr st, + const char *dname, + const char *dom_xml) +{ + virDomainDefPtr def = NULL; + virDomainObjPtr vm = NULL; + char *migrateFrom; + virDomainEventPtr event = NULL; + int ret = -1; + int internalret; + char *unixfile = NULL; + unsigned long long qemuCmdFlags; + qemuDomainObjPrivatePtr priv = NULL; + struct timeval now; + + if (gettimeofday(&now, NULL) < 0) { + virReportSystemError(errno, "%s", + _("cannot get time of day")); + return -1; + } + + /* Parse the domain XML. */ + if (!(def = virDomainDefParseString(driver->caps, dom_xml, + VIR_DOMAIN_XML_INACTIVE))) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("failed to parse XML")); + goto cleanup; + } + + if (!qemuMigrationIsAllowed(def)) + goto cleanup; + + /* Target domain name, maybe renamed. */ + if (dname) { + VIR_FREE(def->name); + def->name = strdup(dname); + if (def->name == NULL) + goto cleanup; + } + + if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0) + goto cleanup; + + if (!(vm = virDomainAssignDef(driver->caps, + &driver->domains, + def, true))) { + /* virDomainAssignDef already set the error */ + goto cleanup; + } + def = NULL; + priv = vm->privateData; + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) + goto cleanup; + priv->jobActive = QEMU_JOB_MIGRATION_OUT; + + /* Domain starts inactive, even if the domain XML had an id field. */ + vm->def->id = -1; + + if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.dest.%s", + driver->libDir, vm->def->name) < 0) { + virReportOOMError(); + goto endjob; + } + unlink(unixfile); + + /* check that this qemu version supports the interactive exec */ + if (qemuCapsExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Cannot determine QEMU argv syntax %s"), + vm->def->emulator); + goto endjob; + } + if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) + internalret = virAsprintf(&migrateFrom, "unix:%s", unixfile); + else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) + internalret = virAsprintf(&migrateFrom, "exec:nc -U -l %s", unixfile); + else { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("Destination qemu is too old to support tunnelled migration")); + goto endjob; + } + if (internalret < 0) { + virReportOOMError(); + goto endjob; + } + /* Start the QEMU daemon, with the same command-line arguments plus + * -incoming unix:/path/to/file or exec:nc -U /path/to/file + */ + internalret = qemuProcessStart(dconn, driver, vm, migrateFrom, true, + -1, NULL, VIR_VM_OP_MIGRATE_IN_START); + VIR_FREE(migrateFrom); + if (internalret < 0) { + qemuDomainStartAudit(vm, "migrated", false); + /* Note that we don't set an error here because qemuProcessStart + * should have already done that. + */ + if (!vm->persistent) { + virDomainRemoveInactive(&driver->domains, vm); + vm = NULL; + } + goto endjob; + } + + if (virFDStreamConnectUNIX(st, + unixfile, + false) < 0) { + qemuDomainStartAudit(vm, "migrated", false); + qemuProcessStop(driver, vm, 0); + if (!vm->persistent) { + if (qemuDomainObjEndJob(vm) > 0) + virDomainRemoveInactive(&driver->domains, vm); + vm = NULL; + } + virReportSystemError(errno, + _("cannot open unix socket '%s' for tunnelled migration"), + unixfile); + goto endjob; + } + + qemuDomainStartAudit(vm, "migrated", true); + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_STARTED, + VIR_DOMAIN_EVENT_STARTED_MIGRATED); + ret = 0; + +endjob: + if (vm && + qemuDomainObjEndJob(vm) == 0) + vm = NULL; + + /* We set a fake job active which is held across + * API calls until the finish() call. This prevents + * any other APIs being invoked while incoming + * migration is taking place + */ + if (vm && + virDomainObjIsActive(vm)) { + priv->jobActive = QEMU_JOB_MIGRATION_IN; + priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; + priv->jobStart = timeval_to_ms(now); + } + +cleanup: + virDomainDefFree(def); + if (unixfile) + unlink(unixfile); + VIR_FREE(unixfile); + if (vm) + virDomainObjUnlock(vm); + if (event) + qemuDomainEventQueue(driver, event); + qemuDriverUnlock(driver); + return ret; +} + + +int +qemuMigrationPrepareDirect(struct qemud_driver *driver, + virConnectPtr dconn, + const char *uri_in, + char **uri_out, + const char *dname, + const char *dom_xml) +{ + static int port = 0; + virDomainDefPtr def = NULL; + virDomainObjPtr vm = NULL; + int this_port; + char *hostname = NULL; + char migrateFrom [64]; + const char *p; + virDomainEventPtr event = NULL; + int ret = -1; + int internalret; + qemuDomainObjPrivatePtr priv = NULL; + struct timeval now; + + if (gettimeofday(&now, NULL) < 0) { + virReportSystemError(errno, "%s", + _("cannot get time of day")); + return -1; + } + + /* The URI passed in may be NULL or a string "tcp://somehostname:port". + * + * If the URI passed in is NULL then we allocate a port number + * from our pool of port numbers and return a URI of + * "tcp://ourhostname:port". + * + * If the URI passed in is not NULL then we try to parse out the + * port number and use that (note that the hostname is assumed + * to be a correct hostname which refers to the target machine). + */ + if (uri_in == NULL) { + this_port = QEMUD_MIGRATION_FIRST_PORT + port++; + if (port == QEMUD_MIGRATION_NUM_PORTS) port = 0; + + /* Get hostname */ + if ((hostname = virGetHostname(NULL)) == NULL) + goto cleanup; + + if (STRPREFIX(hostname, "localhost")) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("hostname on destination resolved to localhost, but migration requires an FQDN")); + goto cleanup; + } + + /* XXX this really should have been a properly well-formed + * URI, but we can't add in tcp:// now without breaking + * compatability with old targets. We at least make the + * new targets accept both syntaxes though. + */ + /* Caller frees */ + internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port); + if (internalret < 0) { + virReportOOMError(); + goto cleanup; + } + } else { + /* Check the URI starts with "tcp:". We will escape the + * URI when passing it to the qemu monitor, so bad + * characters in hostname part don't matter. + */ + if (!STRPREFIX (uri_in, "tcp:")) { + qemuReportError (VIR_ERR_INVALID_ARG, + "%s", _("only tcp URIs are supported for KVM/QEMU migrations")); + goto cleanup; + } + + /* Get the port number. */ + p = strrchr (uri_in, ':'); + if (p == strchr(uri_in, ':')) { + /* Generate a port */ + this_port = QEMUD_MIGRATION_FIRST_PORT + port++; + if (port == QEMUD_MIGRATION_NUM_PORTS) + port = 0; + + /* Caller frees */ + if (virAsprintf(uri_out, "%s:%d", uri_in, this_port) < 0) { + virReportOOMError(); + goto cleanup; + } + + } else { + p++; /* definitely has a ':' in it, see above */ + this_port = virParseNumber (&p); + if (this_port == -1 || p-uri_in != strlen (uri_in)) { + qemuReportError(VIR_ERR_INVALID_ARG, + "%s", _("URI ended with incorrect ':port'")); + goto cleanup; + } + } + } + + if (*uri_out) + VIR_DEBUG("Generated uri_out=%s", *uri_out); + + /* Parse the domain XML. */ + if (!(def = virDomainDefParseString(driver->caps, dom_xml, + VIR_DOMAIN_XML_INACTIVE))) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("failed to parse XML")); + goto cleanup; + } + + if (!qemuMigrationIsAllowed(def)) + goto cleanup; + + /* Target domain name, maybe renamed. */ + if (dname) { + VIR_FREE(def->name); + def->name = strdup(dname); + if (def->name == NULL) + goto cleanup; + } + + if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0) + goto cleanup; + + if (!(vm = virDomainAssignDef(driver->caps, + &driver->domains, + def, true))) { + /* virDomainAssignDef already set the error */ + goto cleanup; + } + def = NULL; + priv = vm->privateData; + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) + goto cleanup; + priv->jobActive = QEMU_JOB_MIGRATION_OUT; + + /* Domain starts inactive, even if the domain XML had an id field. */ + vm->def->id = -1; + + /* Start the QEMU daemon, with the same command-line arguments plus + * -incoming tcp:0.0.0.0:port + */ + snprintf (migrateFrom, sizeof (migrateFrom), "tcp:0.0.0.0:%d", this_port); + if (qemuProcessStart(dconn, driver, vm, migrateFrom, true, + -1, NULL, VIR_VM_OP_MIGRATE_IN_START) < 0) { + qemuDomainStartAudit(vm, "migrated", false); + /* Note that we don't set an error here because qemuProcessStart + * should have already done that. + */ + if (!vm->persistent) { + if (qemuDomainObjEndJob(vm) > 0) + virDomainRemoveInactive(&driver->domains, vm); + vm = NULL; + } + goto endjob; + } + + qemuDomainStartAudit(vm, "migrated", true); + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_STARTED, + VIR_DOMAIN_EVENT_STARTED_MIGRATED); + ret = 0; + +endjob: + if (vm && + qemuDomainObjEndJob(vm) == 0) + vm = NULL; + + /* We set a fake job active which is held across + * API calls until the finish() call. This prevents + * any other APIs being invoked while incoming + * migration is taking place + */ + if (vm && + virDomainObjIsActive(vm)) { + priv->jobActive = QEMU_JOB_MIGRATION_IN; + priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; + priv->jobStart = timeval_to_ms(now); + } + +cleanup: + VIR_FREE(hostname); + virDomainDefFree(def); + if (ret != 0) + VIR_FREE(*uri_out); + if (vm) + virDomainObjUnlock(vm); + if (event) + qemuDomainEventQueue(driver, event); + return ret; +} + + +/* Perform migration using QEMU's native TCP migrate support, + * not encrypted obviously + */ +static int doNativeMigrate(struct qemud_driver *driver, + virDomainObjPtr vm, + const char *uri, + unsigned int flags, + const char *dname ATTRIBUTE_UNUSED, + unsigned long resource) +{ + int ret = -1; + xmlURIPtr uribits = NULL; + qemuDomainObjPrivatePtr priv = vm->privateData; + unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND; + + /* Issue the migrate command. */ + if (STRPREFIX(uri, "tcp:") && !STRPREFIX(uri, "tcp://")) { + /* HACK: source host generates bogus URIs, so fix them up */ + char *tmpuri; + if (virAsprintf(&tmpuri, "tcp://%s", uri + strlen("tcp:")) < 0) { + virReportOOMError(); + goto cleanup; + } + uribits = xmlParseURI(tmpuri); + VIR_FREE(tmpuri); + } else { + uribits = xmlParseURI(uri); + } + if (!uribits) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot parse URI %s"), uri); + goto cleanup; + } + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + if (resource > 0 && + qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) { + qemuDomainObjExitMonitorWithDriver(driver, vm); + goto cleanup; + } + + if (flags & VIR_MIGRATE_NON_SHARED_DISK) + background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK; + + if (flags & VIR_MIGRATE_NON_SHARED_INC) + background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC; + + if (qemuMonitorMigrateToHost(priv->mon, background_flags, uribits->server, + uribits->port) < 0) { + qemuDomainObjExitMonitorWithDriver(driver, vm); + goto cleanup; + } + qemuDomainObjExitMonitorWithDriver(driver, vm); + + if (qemuMigrationWaitForCompletion(driver, vm) < 0) + goto cleanup; + + ret = 0; + +cleanup: + xmlFreeURI(uribits); + return ret; +} + + +#define TUNNEL_SEND_BUF_SIZE 65536 + +static int doTunnelSendAll(virStreamPtr st, + int sock) +{ + char *buffer; + int nbytes = TUNNEL_SEND_BUF_SIZE; + + if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) { + virReportOOMError(); + virStreamAbort(st); + return -1; + } + + /* XXX should honour the 'resource' parameter here */ + for (;;) { + nbytes = saferead(sock, buffer, nbytes); + if (nbytes < 0) { + virReportSystemError(errno, "%s", + _("tunnelled migration failed to read from qemu")); + virStreamAbort(st); + VIR_FREE(buffer); + return -1; + } + else if (nbytes == 0) + /* EOF; get out of here */ + break; + + if (virStreamSend(st, buffer, nbytes) < 0) { + qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", + _("Failed to write migration data to remote libvirtd")); + VIR_FREE(buffer); + return -1; + } + } + + VIR_FREE(buffer); + + if (virStreamFinish(st) < 0) + /* virStreamFinish set the error for us */ + return -1; + + return 0; +} + +static int doTunnelMigrate(struct qemud_driver *driver, + virConnectPtr dconn, + virDomainObjPtr vm, + const char *dom_xml, + const char *uri, + unsigned long flags, + const char *dname, + unsigned long resource) +{ + qemuDomainObjPrivatePtr priv = vm->privateData; + int client_sock = -1; + int qemu_sock = -1; + struct sockaddr_un sa_qemu, sa_client; + socklen_t addrlen; + virDomainPtr ddomain = NULL; + int retval = -1; + virStreamPtr st = NULL; + char *unixfile = NULL; + int internalret; + unsigned long long qemuCmdFlags; + int status; + unsigned long long transferred, remaining, total; + unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND; + + /* + * The order of operations is important here to avoid touching + * the source VM until we are very sure we can successfully + * start the migration operation. + * + * 1. setup local support infrastructure (eg sockets) + * 2. setup destination fully + * 3. start migration on source + */ + + + /* Stage 1. setup local support infrastructure */ + + if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s", + driver->libDir, vm->def->name) < 0) { + virReportOOMError(); + goto cleanup; + } + + qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0); + if (qemu_sock < 0) { + virReportSystemError(errno, "%s", + _("cannot open tunnelled migration socket")); + goto cleanup; + } + memset(&sa_qemu, 0, sizeof(sa_qemu)); + sa_qemu.sun_family = AF_UNIX; + if (virStrcpy(sa_qemu.sun_path, unixfile, + sizeof(sa_qemu.sun_path)) == NULL) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Unix socket '%s' too big for destination"), + unixfile); + goto cleanup; + } + unlink(unixfile); + if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) { + virReportSystemError(errno, + _("Cannot bind to unix socket '%s' for tunnelled migration"), + unixfile); + goto cleanup; + } + if (listen(qemu_sock, 1) < 0) { + virReportSystemError(errno, + _("Cannot listen on unix socket '%s' for tunnelled migration"), + unixfile); + goto cleanup; + } + + if (chown(unixfile, driver->user, driver->group) < 0) { + virReportSystemError(errno, + _("Cannot change unix socket '%s' owner"), + unixfile); + goto cleanup; + } + + /* check that this qemu version supports the unix migration */ + if (qemuCapsExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Cannot extract Qemu version from '%s'"), + vm->def->emulator); + goto cleanup; + } + + if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) && + !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("Source qemu is too old to support tunnelled migration")); + goto cleanup; + } + + + /* Stage 2. setup destination fully + * + * Once stage 2 has completed successfully, we *must* call finish + * to cleanup the target whether we succeed or fail + */ + st = virStreamNew(dconn, 0); + if (st == NULL) + /* virStreamNew only fails on OOM, and it reports the error itself */ + goto cleanup; + + qemuDomainObjEnterRemoteWithDriver(driver, vm); + internalret = dconn->driver->domainMigratePrepareTunnel(dconn, st, + flags, dname, + resource, dom_xml); + qemuDomainObjExitRemoteWithDriver(driver, vm); + + if (internalret < 0) + /* domainMigratePrepareTunnel sets the error for us */ + goto cleanup; + + /* the domain may have shutdown or crashed while we had the locks dropped + * in qemuDomainObjEnterRemoteWithDriver, so check again + */ + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("guest unexpectedly quit")); + goto cleanup; + } + + /* 3. start migration on source */ + qemuDomainObjEnterMonitorWithDriver(driver, vm); + if (flags & VIR_MIGRATE_NON_SHARED_DISK) + background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK; + if (flags & VIR_MIGRATE_NON_SHARED_INC) + background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC; + if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX){ + internalret = qemuMonitorMigrateToUnix(priv->mon, background_flags, + unixfile); + } + else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) { + const char *args[] = { "nc", "-U", unixfile, NULL }; + internalret = qemuMonitorMigrateToCommand(priv->mon, QEMU_MONITOR_MIGRATE_BACKGROUND, args); + } else { + internalret = -1; + } + qemuDomainObjExitMonitorWithDriver(driver, vm); + if (internalret < 0) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("tunnelled migration monitor command failed")); + goto finish; + } + + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("guest unexpectedly quit")); + goto cleanup; + } + + /* From this point onwards we *must* call cancel to abort the + * migration on source if anything goes wrong */ + + /* it is also possible that the migrate didn't fail initially, but + * rather failed later on. Check the output of "info migrate" + */ + qemuDomainObjEnterMonitorWithDriver(driver, vm); + if (qemuMonitorGetMigrationStatus(priv->mon, + &status, + &transferred, + &remaining, + &total) < 0) { + qemuDomainObjExitMonitorWithDriver(driver, vm); + goto cancel; + } + qemuDomainObjExitMonitorWithDriver(driver, vm); + + if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s",_("migrate failed")); + goto cancel; + } + + addrlen = sizeof(sa_client); + while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) { + if (errno == EAGAIN || errno == EINTR) + continue; + virReportSystemError(errno, "%s", + _("tunnelled migration failed to accept from qemu")); + goto cancel; + } + + retval = doTunnelSendAll(st, client_sock); + +cancel: + if (retval != 0 && virDomainObjIsActive(vm)) { + qemuDomainObjEnterMonitorWithDriver(driver, vm); + qemuMonitorMigrateCancel(priv->mon); + qemuDomainObjExitMonitorWithDriver(driver, vm); + } + +finish: + dname = dname ? dname : vm->def->name; + qemuDomainObjEnterRemoteWithDriver(driver, vm); + ddomain = dconn->driver->domainMigrateFinish2 + (dconn, dname, NULL, 0, uri, flags, retval); + qemuDomainObjExitRemoteWithDriver(driver, vm); + +cleanup: + VIR_FORCE_CLOSE(client_sock); + VIR_FORCE_CLOSE(qemu_sock); + + if (ddomain) + virUnrefDomain(ddomain); + + if (unixfile) { + unlink(unixfile); + VIR_FREE(unixfile); + } + + if (st) + /* don't call virStreamFree(), because that resets any pending errors */ + virUnrefStream(st); + return retval; +} + + +/* This is essentially a simplified re-impl of + * virDomainMigrateVersion2 from libvirt.c, but running in source + * libvirtd context, instead of client app context */ +static int doNonTunnelMigrate(struct qemud_driver *driver, + virConnectPtr dconn, + virDomainObjPtr vm, + const char *dom_xml, + const char *uri ATTRIBUTE_UNUSED, + unsigned long flags, + const char *dname, + unsigned long resource) +{ + virDomainPtr ddomain = NULL; + int retval = -1; + char *uri_out = NULL; + int rc; + + qemuDomainObjEnterRemoteWithDriver(driver, vm); + /* NB we don't pass 'uri' into this, since that's the libvirtd + * URI in this context - so we let dest pick it */ + rc = dconn->driver->domainMigratePrepare2(dconn, + NULL, /* cookie */ + 0, /* cookielen */ + NULL, /* uri */ + &uri_out, + flags, dname, + resource, dom_xml); + qemuDomainObjExitRemoteWithDriver(driver, vm); + if (rc < 0) + /* domainMigratePrepare2 sets the error for us */ + goto cleanup; + + /* the domain may have shutdown or crashed while we had the locks dropped + * in qemuDomainObjEnterRemoteWithDriver, so check again + */ + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("guest unexpectedly quit")); + goto cleanup; + } + + if (uri_out == NULL) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("domainMigratePrepare2 did not set uri")); + goto cleanup; + } + + if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0) + goto finish; + + retval = 0; + +finish: + dname = dname ? dname : vm->def->name; + qemuDomainObjEnterRemoteWithDriver(driver, vm); + ddomain = dconn->driver->domainMigrateFinish2 + (dconn, dname, NULL, 0, uri_out, flags, retval); + qemuDomainObjExitRemoteWithDriver(driver, vm); + + if (ddomain) + virUnrefDomain(ddomain); + +cleanup: + return retval; +} + + +static int doPeer2PeerMigrate(struct qemud_driver *driver, + virDomainObjPtr vm, + const char *uri, + unsigned long flags, + const char *dname, + unsigned long resource) +{ + int ret = -1; + virConnectPtr dconn = NULL; + char *dom_xml; + bool p2p; + + /* the order of operations is important here; we make sure the + * destination side is completely setup before we touch the source + */ + + qemuDomainObjEnterRemoteWithDriver(driver, vm); + dconn = virConnectOpen(uri); + qemuDomainObjExitRemoteWithDriver(driver, vm); + if (dconn == NULL) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + _("Failed to connect to remote libvirt URI %s"), uri); + return -1; + } + + qemuDomainObjEnterRemoteWithDriver(driver, vm); + p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn, + VIR_DRV_FEATURE_MIGRATION_P2P); + qemuDomainObjExitRemoteWithDriver(driver, vm); + if (!p2p) { + qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", + _("Destination libvirt does not support peer-to-peer migration protocol")); + goto cleanup; + } + + /* domain may have been stopped while we were talking to remote daemon */ + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("guest unexpectedly quit")); + goto cleanup; + } + + dom_xml = qemuDomainFormatXML(driver, vm, + VIR_DOMAIN_XML_SECURE | + VIR_DOMAIN_XML_UPDATE_CPU); + if (!dom_xml) { + qemuReportError(VIR_ERR_OPERATION_FAILED, + "%s", _("failed to get domain xml")); + goto cleanup; + } + + if (flags & VIR_MIGRATE_TUNNELLED) + ret = doTunnelMigrate(driver, dconn, vm, dom_xml, uri, flags, dname, resource); + else + ret = doNonTunnelMigrate(driver, dconn, vm, dom_xml, uri, flags, dname, resource); + +cleanup: + VIR_FREE(dom_xml); + /* don't call virConnectClose(), because that resets any pending errors */ + qemuDomainObjEnterRemoteWithDriver(driver, vm); + virUnrefConnect(dconn); + qemuDomainObjExitRemoteWithDriver(driver, vm); + + return ret; +} + + +int qemuMigrationPerform(struct qemud_driver *driver, + virConnectPtr conn, + virDomainObjPtr vm, + const char *uri, + unsigned long flags, + const char *dname, + unsigned long resource) +{ + virDomainEventPtr event = NULL; + int ret = -1; + int resume = 0; + qemuDomainObjPrivatePtr priv = vm->privateData; + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) + goto cleanup; + priv->jobActive = QEMU_JOB_MIGRATION_OUT; + + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_OPERATION_INVALID, + "%s", _("domain is not running")); + goto endjob; + } + + memset(&priv->jobInfo, 0, sizeof(priv->jobInfo)); + priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED; + + resume = vm->state == VIR_DOMAIN_RUNNING; + if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) { + if (qemuMigrationSetOffline(driver, vm) < 0) + goto endjob; + } + + if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) { + if (doPeer2PeerMigrate(driver, vm, uri, flags, dname, resource) < 0) + /* doPeer2PeerMigrate already set the error, so just get out */ + goto endjob; + } else { + if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0) + goto endjob; + } + + /* Clean up the source domain. */ + qemuProcessStop(driver, vm, 1); + qemuDomainStopAudit(vm, "migrated"); + resume = 0; + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_STOPPED, + VIR_DOMAIN_EVENT_STOPPED_MIGRATED); + if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) { + virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm); + if (qemuDomainObjEndJob(vm) > 0) + virDomainRemoveInactive(&driver->domains, vm); + vm = NULL; + } + ret = 0; + +endjob: + if (resume && vm->state == VIR_DOMAIN_PAUSED) { + /* we got here through some sort of failure; start the domain again */ + if (qemuProcessStartCPUs(driver, vm, conn) < 0) { + /* Hm, we already know we are in error here. We don't want to + * overwrite the previous error, though, so we just throw something + * to the logs and hope for the best + */ + VIR_ERROR(_("Failed to resume guest %s after failure"), + vm->def->name); + } + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_RESUMED, + VIR_DOMAIN_EVENT_RESUMED_MIGRATED); + } + if (vm && + qemuDomainObjEndJob(vm) == 0) + vm = NULL; + +cleanup: + if (vm) + virDomainObjUnlock(vm); + if (event) + qemuDomainEventQueue(driver, event); + return ret; +} + + +#if WITH_MACVTAP +static void +qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def) { + int i; + int last_good_net = -1; + virDomainNetDefPtr net; + + for (i = 0; i < def->nnets; i++) { + net = def->nets[i]; + if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { + if (vpAssociatePortProfileId(net->ifname, + net->mac, + net->data.direct.linkdev, + &net->data.direct.virtPortProfile, + def->uuid, + VIR_VM_OP_MIGRATE_IN_FINISH) != 0) + goto err_exit; + } + last_good_net = i; + } + + return; + +err_exit: + for (i = 0; i < last_good_net; i++) { + net = def->nets[i]; + if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { + vpDisassociatePortProfileId(net->ifname, + net->mac, + net->data.direct.linkdev, + &net->data.direct.virtPortProfile, + VIR_VM_OP_MIGRATE_IN_FINISH); + } + } +} +#else /* !WITH_MACVTAP */ +static void +qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def ATTRIBUTE_UNUSED) { } +#endif /* WITH_MACVTAP */ + + +virDomainPtr +qemuMigrationFinish(struct qemud_driver *driver, + virConnectPtr dconn, + virDomainObjPtr vm, + unsigned long flags, + int retcode) +{ + virDomainPtr dom = NULL; + virDomainEventPtr event = NULL; + int newVM = 1; + qemuDomainObjPrivatePtr priv = NULL; + + priv = vm->privateData; + if (priv->jobActive != QEMU_JOB_MIGRATION_IN) { + qemuReportError(VIR_ERR_NO_DOMAIN, + _("domain '%s' is not processing incoming migration"), vm->def->name); + goto cleanup; + } + priv->jobActive = QEMU_JOB_NONE; + memset(&priv->jobInfo, 0, sizeof(priv->jobInfo)); + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) + goto cleanup; + + /* Did the migration go as planned? If yes, return the domain + * object, but if no, clean up the empty qemu process. + */ + if (retcode == 0) { + if (!virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("guest unexpectedly quit")); + goto cleanup; + } + + qemuMigrationVPAssociatePortProfiles(vm->def); + + if (flags & VIR_MIGRATE_PERSIST_DEST) { + if (vm->persistent) + newVM = 0; + vm->persistent = 1; + + if (virDomainSaveConfig(driver->configDir, vm->def) < 0) { + /* Hmpf. Migration was successful, but making it persistent + * was not. If we report successful, then when this domain + * shuts down, management tools are in for a surprise. On the + * other hand, if we report failure, then the management tools + * might try to restart the domain on the source side, even + * though the domain is actually running on the destination. + * Return a NULL dom pointer, and hope that this is a rare + * situation and management tools are smart. + */ + vm = NULL; + goto endjob; + } + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_DEFINED, + newVM ? + VIR_DOMAIN_EVENT_DEFINED_ADDED : + VIR_DOMAIN_EVENT_DEFINED_UPDATED); + if (event) + qemuDomainEventQueue(driver, event); + event = NULL; + + } + dom = virGetDomain (dconn, vm->def->name, vm->def->uuid); + + if (!(flags & VIR_MIGRATE_PAUSED)) { + /* run 'cont' on the destination, which allows migration on qemu + * >= 0.10.6 to work properly. This isn't strictly necessary on + * older qemu's, but it also doesn't hurt anything there + */ + if (qemuProcessStartCPUs(driver, vm, dconn) < 0) { + if (virGetLastError() == NULL) + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("resume operation failed")); + goto endjob; + } + } + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_RESUMED, + VIR_DOMAIN_EVENT_RESUMED_MIGRATED); + if (vm->state == VIR_DOMAIN_PAUSED) { + qemuDomainEventQueue(driver, event); + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_SUSPENDED, + VIR_DOMAIN_EVENT_SUSPENDED_PAUSED); + } + if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) { + VIR_WARN("Failed to save status on vm %s", vm->def->name); + goto endjob; + } + } else { + qemuProcessStop(driver, vm, 1); + qemuDomainStopAudit(vm, "failed"); + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_STOPPED, + VIR_DOMAIN_EVENT_STOPPED_FAILED); + if (!vm->persistent) { + if (qemuDomainObjEndJob(vm) > 0) + virDomainRemoveInactive(&driver->domains, vm); + vm = NULL; + } + } + +endjob: + if (vm && + qemuDomainObjEndJob(vm) == 0) + vm = NULL; + +cleanup: + if (vm) + virDomainObjUnlock(vm); + if (event) + qemuDomainEventQueue(driver, event); + return dom; +} diff --git a/src/qemu/qemu_migration.h b/src/qemu/qemu_migration.h new file mode 100644 index 0000000..54660a4 --- /dev/null +++ b/src/qemu/qemu_migration.h @@ -0,0 +1,63 @@ +/* + * qemu_migration.h: QEMU migration handling + * + * Copyright (C) 2006-2011 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 + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#ifndef __QEMU_MIGRATION_H__ +# define __QEMU_MIGRATION_H__ + +#include "qemu_conf.h" + + +bool qemuMigrationIsAllowed(virDomainDefPtr def) + ATTRIBUTE_NONNULL(1); +int qemuMigrationSetOffline(struct qemud_driver *driver, + virDomainObjPtr vm); + +int qemuMigrationWaitForCompletion(struct qemud_driver *driver, virDomainObjPtr vm); + +int qemuMigrationPrepareTunnel(struct qemud_driver *driver, + virConnectPtr dconn, + virStreamPtr st, + const char *dname, + const char *dom_xml); + +int qemuMigrationPrepareDirect(struct qemud_driver *driver, + virConnectPtr dconn, + const char *uri_in, + char **uri_out, + const char *dname, + const char *dom_xml); + +int qemuMigrationPerform(struct qemud_driver *driver, + virConnectPtr conn, + virDomainObjPtr vm, + const char *uri, + unsigned long flags, + const char *dname, + unsigned long resource); + +virDomainPtr qemuMigrationFinish(struct qemud_driver *driver, + virConnectPtr dconn, + virDomainObjPtr vm, + unsigned long flags, + int retcode); + + +#endif /* __QEMU_MIGRATION_H__ */ -- 1.7.3.4

13 years, 3 months

2
3
0 / 0

Re: [libvirt] Raising events for testing

by arnaud.champion＠devatom.fr

Yep that's what I've just seen. I've decided to try via mingwin which seems to implement a poll function...Callbacks implementation are really hard for a noob :) Regards, Arnaud ---------------------------------------- De: "Eric Blake" <eblake(a)redhat.com> Envoyé: lundi 31 janvier 2011 20:12 A: arnaud.champion(a)si-vs1281.com Objet: Re: [libvirt] Raising events for testing On 01/31/2011 11:52 AM, arnaud.champion(a)devatom.fr wrote: >> There is - it involves using gnulib. But that is only for C and C++ - I >> have no idea how this translates over to C# > So if I understand well, are you saying that if I use gnulib windows version via C# (pinvoke, this kind of magic) I will have access to poll function ? > Intersting, I'll take a look. I hope gnulib if published via a dll under windows, if so, I'm pretty sure I can use it with C# > regards Sorry; gnulib is a source only library that currently targets only C and C++; it does not target C#, and does not provide any dll, so I have no idea how that affects your porting efforts. -- Eric Blake eblake(a)redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org

13 years, 3 months

1
0
0 / 0

Re: [libvirt] Raising events for testing

by arnaud.champion＠devatom.fr

>There is - it involves using gnulib. But that is only for C and C++ - I >have no idea how this translates over to C# So if I understand well, are you saying that if I use gnulib windows version via C# (pinvoke, this kind of magic) I will have access to poll function ? Intersting, I'll take a look. I hope gnulib if published via a dll under windows, if so, I'm pretty sure I can use it with C# regards Arnaud ---------------------------------------- De: "Eric Blake" <eblake(a)redhat.com> Envoyé: lundi 31 janvier 2011 19:37 A: arnaud.champion(a)devatom.fr Objet: Re: [libvirt] Raising events for testing On 01/30/2011 05:21 AM, arnaud.champion(a)devatom.fr wrote: > Hi, > > One thing I need to understand, in the callback python or c samples, the > fileDescriptor is on a socket or on a pipe ? Libvirt uses gnulib to guarantee that poll() works on file descriptors; all socket operations are magically mapped by gnulib to behave like file descriptors, so that you don't have to worry about the (less-useful) restrictions of windows poll() only working on sockets. >> Okay I have understand the use and mean of poll (pollfd... >> bla...bla...bla...) >> I have also discover that there is no Win32 equivalent which work with >> a file descriptor (unix way...). There is - it involves using gnulib. But that is only for C and C++ - I have no idea how this translates over to C#. -- Eric Blake eblake(a)redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org

13 years, 3 months

2
1
0 / 0

[libvirt] [PATCH] Split all QEMU process mangement code into separate file

by Daniel P. Berrange

Move the qemudStartVMDaemon and qemudShutdownVMDaemon methods into a separate file, renaming them to qemuProcessStart, qemuProcessStop. All helper methods called by these are also moved & renamed to match * src/Makefile.am: Add qemu_process.c/.h * src/qemu/qemu_command.c: Add emuDomainAssignPCIAddresses * src/qemu/qemu_command.h: Add VNC port min/max * src/qemu/qemu_domain.c, src/qemu/qemu_domain.h: Add domain event queue helpers * src/qemu/qemu_driver.c, src/qemu/qemu_driver.h: Remove all QEMU process startup/shutdown functions * src/qemu/qemu_process.c, src/qemu/qemu_process.h: Add all QEMU process startup/shutdown functions --- src/Makefile.am | 1 + src/qemu/qemu_command.c | 29 + src/qemu/qemu_command.h | 5 + src/qemu/qemu_domain.c | 56 + src/qemu/qemu_domain.h | 11 + src/qemu/qemu_driver.c | 3519 ++++++++--------------------------------------- src/qemu/qemu_driver.h | 31 +- src/qemu/qemu_process.c | 2412 ++++++++++++++++++++++++++++++++ src/qemu/qemu_process.h | 52 + 9 files changed, 3110 insertions(+), 3006 deletions(-) create mode 100644 src/qemu/qemu_process.c create mode 100644 src/qemu/qemu_process.h diff --git a/src/Makefile.am b/src/Makefile.am index f8b8434..9ef1d36 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -281,6 +281,7 @@ QEMU_DRIVER_SOURCES = \ qemu/qemu_hostdev.c qemu/qemu_hostdev.h \ qemu/qemu_hotplug.c qemu/qemu_hotplug.h \ qemu/qemu_conf.c qemu/qemu_conf.h \ + qemu/qemu_process.c qemu/qemu_process.h \ qemu/qemu_monitor.c qemu/qemu_monitor.h \ qemu/qemu_monitor_text.c \ qemu/qemu_monitor_text.h \ diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 1ce8941..4f4658b 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -711,6 +711,35 @@ static int qemuCollectPCIAddress(virDomainDefPtr def ATTRIBUTE_UNUSED, } +int +qemuDomainAssignPCIAddresses(virDomainDefPtr def) +{ + int ret = -1; + unsigned long long qemuCmdFlags = 0; + qemuDomainPCIAddressSetPtr addrs = NULL; + + if (qemuCapsExtractVersionInfo(def->emulator, + NULL, + &qemuCmdFlags) < 0) + goto cleanup; + + if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) { + if (!(addrs = qemuDomainPCIAddressSetCreate(def))) + goto cleanup; + + if (qemuAssignDevicePCISlots(def, addrs) < 0) + goto cleanup; + } + + ret = 0; + +cleanup: + qemuDomainPCIAddressSetFree(addrs); + + return ret; +} + + qemuDomainPCIAddressSetPtr qemuDomainPCIAddressSetCreate(virDomainDefPtr def) { qemuDomainPCIAddressSetPtr addrs; diff --git a/src/qemu/qemu_command.h b/src/qemu/qemu_command.h index e410259..b562919 100644 --- a/src/qemu/qemu_command.h +++ b/src/qemu/qemu_command.h @@ -37,6 +37,10 @@ # define QEMU_VIRTIO_SERIAL_PREFIX "virtio-serial" # define QEMU_FSDEV_HOST_PREFIX "fsdev-" +#define QEMU_VNC_PORT_MIN 5900 +#define QEMU_VNC_PORT_MAX 65535 + + virCommandPtr qemuBuildCommandLine(virConnectPtr conn, struct qemud_driver *driver, virDomainDefPtr def, @@ -130,6 +134,7 @@ virDomainDefPtr qemuParseCommandLine(virCapsPtr caps, virDomainDefPtr qemuParseCommandLineString(virCapsPtr caps, const char *args); +int qemuDomainAssignPCIAddresses(virDomainDefPtr def); qemuDomainPCIAddressSetPtr qemuDomainPCIAddressSetCreate(virDomainDefPtr def); int qemuDomainPCIAddressReserveSlot(qemuDomainPCIAddressSetPtr addrs, int slot); diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index fa7c8bd..e3163ab 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -29,6 +29,7 @@ #include "logging.h" #include "virterror_internal.h" #include "c-ctype.h" +#include "event.h" #include <sys/time.h> @@ -41,6 +42,61 @@ #define timeval_to_ms(tv) (((tv).tv_sec * 1000ull) + ((tv).tv_usec / 1000)) +static void qemuDomainEventDispatchFunc(virConnectPtr conn, + virDomainEventPtr event, + virConnectDomainEventGenericCallback cb, + void *cbopaque, + void *opaque) +{ + struct qemud_driver *driver = opaque; + + /* Drop the lock whle dispatching, for sake of re-entrancy */ + qemuDriverUnlock(driver); + virDomainEventDispatchDefaultFunc(conn, event, cb, cbopaque, NULL); + qemuDriverLock(driver); +} + +void qemuDomainEventFlush(int timer ATTRIBUTE_UNUSED, void *opaque) +{ + struct qemud_driver *driver = opaque; + virDomainEventQueue tempQueue; + + qemuDriverLock(driver); + + driver->domainEventDispatching = 1; + + /* Copy the queue, so we're reentrant safe */ + tempQueue.count = driver->domainEventQueue->count; + tempQueue.events = driver->domainEventQueue->events; + driver->domainEventQueue->count = 0; + driver->domainEventQueue->events = NULL; + + virEventUpdateTimeout(driver->domainEventTimer, -1); + virDomainEventQueueDispatch(&tempQueue, + driver->domainEventCallbacks, + qemuDomainEventDispatchFunc, + driver); + + /* Purge any deleted callbacks */ + virDomainEventCallbackListPurgeMarked(driver->domainEventCallbacks); + + driver->domainEventDispatching = 0; + qemuDriverUnlock(driver); +} + + +/* driver must be locked before calling */ +void qemuDomainEventQueue(struct qemud_driver *driver, + virDomainEventPtr event) +{ + if (virDomainEventQueuePush(driver->domainEventQueue, + event) < 0) + virDomainEventFree(event); + if (driver->domainEventQueue->count == 1) + virEventUpdateTimeout(driver->domainEventTimer, 0); +} + + static void *qemuDomainObjPrivateAlloc(void) { qemuDomainObjPrivatePtr priv; diff --git a/src/qemu/qemu_domain.h b/src/qemu/qemu_domain.h index f14fb79..4333a78 100644 --- a/src/qemu/qemu_domain.h +++ b/src/qemu/qemu_domain.h @@ -77,6 +77,17 @@ struct _qemuDomainObjPrivate { int persistentAddrs; }; +struct qemuDomainWatchdogEvent +{ + virDomainObjPtr vm; + int action; +}; + +void qemuDomainEventFlush(int timer ATTRIBUTE_UNUSED, void *opaque); + +/* driver must be locked before calling */ +void qemuDomainEventQueue(struct qemud_driver *driver, + virDomainEventPtr event); void qemuDomainSetPrivateDataHooks(virCapsPtr caps); void qemuDomainSetNamespaceHooks(virCapsPtr caps); diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 84d339b..dcc1050 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -46,9 +46,6 @@ #include <sys/un.h> -#include "virterror_internal.h" -#include "logging.h" -#include "datatypes.h" #include "qemu_driver.h" #include "qemu_conf.h" #include "qemu_capabilities.h" @@ -59,6 +56,11 @@ #include "qemu_monitor.h" #include "qemu_bridge_filter.h" #include "qemu_audit.h" +#include "qemu_process.h" + +#include "virterror_internal.h" +#include "logging.h" +#include "datatypes.h" #include "c-ctype.h" #include "event.h" #include "buf.h" @@ -88,169 +90,48 @@ #define VIR_FROM_THIS VIR_FROM_QEMU -#define QEMU_VNC_PORT_MIN 5900 -#define QEMU_VNC_PORT_MAX 65535 - #define QEMU_NB_MEM_PARAM 3 +# if HAVE_LINUX_KVM_H +# include <linux/kvm.h> +# endif -#define timeval_to_ms(tv) (((tv).tv_sec * 1000ull) + ((tv).tv_usec / 1000)) +/* device for kvm ioctls */ +# define KVM_DEVICE "/dev/kvm" -struct watchdogEvent -{ - virDomainObjPtr vm; - int action; -}; +/* add definitions missing in older linux/kvm.h */ +# ifndef KVMIO +# define KVMIO 0xAE +# endif +# ifndef KVM_CHECK_EXTENSION +# define KVM_CHECK_EXTENSION _IO(KVMIO, 0x03) +# endif +# ifndef KVM_CAP_NR_VCPUS +# define KVM_CAP_NR_VCPUS 9 /* returns max vcpus per vm */ +# endif + + +#define timeval_to_ms(tv) (((tv).tv_sec * 1000ull) + ((tv).tv_usec / 1000)) static void processWatchdogEvent(void *data, void *opaque); static int qemudShutdown(void); -static void qemuDomainEventFlush(int timer, void *opaque); -static void qemuDomainEventQueue(struct qemud_driver *driver, - virDomainEventPtr event); - static int qemudDomainObjStart(virConnectPtr conn, struct qemud_driver *driver, virDomainObjPtr vm, bool start_paused); -static int qemudStartVMDaemon(virConnectPtr conn, - struct qemud_driver *driver, - virDomainObjPtr vm, - const char *migrateFrom, - bool start_paused, - int stdin_fd, - const char *stdin_path, - enum virVMOperationType vmop); - -static void qemudShutdownVMDaemon(struct qemud_driver *driver, - virDomainObjPtr vm, - int migrated); - static int qemudDomainGetMaxVcpus(virDomainPtr dom); -static int qemuDetectVcpuPIDs(struct qemud_driver *driver, - virDomainObjPtr vm); - -static int qemudVMFiltersInstantiate(virConnectPtr conn, - virDomainDefPtr def); - -static struct qemud_driver *qemu_driver = NULL; - - -static int doStartCPUs(struct qemud_driver *driver, virDomainObjPtr vm, virConnectPtr conn) -{ - int ret; - qemuDomainObjPrivatePtr priv = vm->privateData; - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - ret = qemuMonitorStartCPUs(priv->mon, conn); - if (ret == 0) { - vm->state = VIR_DOMAIN_RUNNING; - } - qemuDomainObjExitMonitorWithDriver(driver, vm); - - return ret; -} - -static int doStopCPUs(struct qemud_driver *driver, virDomainObjPtr vm) -{ - int ret; - int oldState = vm->state; - qemuDomainObjPrivatePtr priv = vm->privateData; - - vm->state = VIR_DOMAIN_PAUSED; - qemuDomainObjEnterMonitorWithDriver(driver, vm); - ret = qemuMonitorStopCPUs(priv->mon); - qemuDomainObjExitMonitorWithDriver(driver, vm); - if (ret < 0) { - vm->state = oldState; - } - return ret; -} - - -static int -qemudLogFD(struct qemud_driver *driver, const char* name, bool append) -{ - char *logfile; - mode_t logmode; - int fd = -1; - - if (virAsprintf(&logfile, "%s/%s.log", driver->logDir, name) < 0) { - virReportOOMError(); - return -1; - } - - logmode = O_CREAT | O_WRONLY; - /* Only logrotate files in /var/log, so only append if running privileged */ - if (driver->privileged || append) - logmode |= O_APPEND; - else - logmode |= O_TRUNC; - - if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) { - virReportSystemError(errno, - _("failed to create logfile %s"), - logfile); - VIR_FREE(logfile); - return -1; - } - VIR_FREE(logfile); - if (virSetCloseExec(fd) < 0) { - virReportSystemError(errno, "%s", - _("Unable to set VM logfile close-on-exec flag")); - VIR_FORCE_CLOSE(fd); - return -1; - } - return fd; -} - - -static int -qemudLogReadFD(const char* logDir, const char* name, off_t pos) -{ - char *logfile; - mode_t logmode = O_RDONLY; - int fd = -1; - - if (virAsprintf(&logfile, "%s/%s.log", logDir, name) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("failed to build logfile name %s/%s.log"), - logDir, name); - return -1; - } - - if ((fd = open(logfile, logmode)) < 0) { - virReportSystemError(errno, - _("failed to create logfile %s"), - logfile); - VIR_FREE(logfile); - return -1; - } - if (virSetCloseExec(fd) < 0) { - virReportSystemError(errno, "%s", - _("Unable to set VM logfile close-on-exec flag")); - VIR_FORCE_CLOSE(fd); - VIR_FREE(logfile); - return -1; - } - if (pos < 0 || lseek(fd, pos, SEEK_SET) < 0) { - virReportSystemError(pos < 0 ? 0 : errno, - _("Unable to seek to %lld in %s"), - (long long) pos, logfile); - VIR_FORCE_CLOSE(fd); - } - VIR_FREE(logfile); - return fd; -} +struct qemud_driver *qemu_driver = NULL; struct qemuAutostartData { struct qemud_driver *driver; virConnectPtr conn; }; + static void qemuAutostartDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque) { @@ -283,8 +164,10 @@ qemuAutostartDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaq virDomainObjUnlock(vm); } + static void -qemudAutostartConfigs(struct qemud_driver *driver) { +qemuAutostartDomains(struct qemud_driver *driver) +{ /* XXX: Figure out a better way todo this. The domain * startup code needs a connection handle in order * to lookup the bridge associated with a virtual @@ -304,2847 +187,605 @@ qemudAutostartConfigs(struct qemud_driver *driver) { virConnectClose(conn); } - -/** - * qemudRemoveDomainStatus - * - * remove all state files of a domain from statedir - * - * Returns 0 on success - */ static int -qemudRemoveDomainStatus(struct qemud_driver *driver, - virDomainObjPtr vm) +qemuSecurityInit(struct qemud_driver *driver) { - char ebuf[1024]; - char *file = NULL; - - if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) { - virReportOOMError(); - return(-1); - } - - if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR) - VIR_WARN("Failed to remove domain XML for %s: %s", - vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf))); - VIR_FREE(file); + virSecurityManagerPtr mgr = virSecurityManagerNew(driver->securityDriverName, + driver->allowDiskFormatProbing); + if (!mgr) + goto error; - if (virFileDeletePid(driver->stateDir, vm->def->name) != 0) - VIR_WARN("Failed to remove PID file for %s: %s", - vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf))); + if (driver->privileged) { + virSecurityManagerPtr dac = virSecurityManagerNewDAC(driver->user, + driver->group, + driver->allowDiskFormatProbing, + driver->dynamicOwnership); + if (!dac) + goto error; + if (!(driver->securityManager = virSecurityManagerNewStack(mgr, + dac))) + goto error; + } else { + driver->securityManager = mgr; + } return 0; -} - -/* - * This is a callback registered with a qemuMonitorPtr instance, - * and to be invoked when the monitor console hits an end of file - * condition, or error, thus indicating VM shutdown should be - * performed - */ -static void -qemuHandleMonitorEOF(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm, - int hasError) { - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr event = NULL; - qemuDomainObjPrivatePtr priv; +error: + VIR_ERROR0(_("Failed to initialize security drivers")); + virSecurityManagerFree(mgr); + return -1; +} - VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name); - virDomainObjLock(vm); +static virCapsPtr +qemuCreateCapabilities(virCapsPtr oldcaps, + struct qemud_driver *driver) +{ + virCapsPtr caps; - if (!virDomainObjIsActive(vm)) { - VIR_DEBUG("Domain %p is not active, ignoring EOF", vm); - virDomainObjUnlock(vm); - return; + /* Basic host arch / guest machine capabilities */ + if (!(caps = qemuCapsInit(oldcaps))) { + virReportOOMError(); + return NULL; } - priv = vm->privateData; - if (!hasError && priv->monJSON && !priv->gotShutdown) { - VIR_DEBUG("Monitor connection to '%s' closed without SHUTDOWN event; " - "assuming the domain crashed", vm->def->name); - hasError = 1; + if (driver->allowDiskFormatProbing) { + caps->defaultDiskDriverName = NULL; + caps->defaultDiskDriverType = NULL; + } else { + caps->defaultDiskDriverName = "qemu"; + caps->defaultDiskDriverType = "raw"; } - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_STOPPED, - hasError ? - VIR_DOMAIN_EVENT_STOPPED_FAILED : - VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN); - - qemudShutdownVMDaemon(driver, vm, 0); - qemuDomainStopAudit(vm, hasError ? "failed" : "shutdown"); - - if (!vm->persistent) - virDomainRemoveInactive(&driver->domains, vm); - else - virDomainObjUnlock(vm); + qemuDomainSetPrivateDataHooks(caps); + qemuDomainSetNamespaceHooks(caps); - if (event) { - qemuDriverLock(driver); - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); + if (virGetHostUUID(caps->host.host_uuid)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("cannot get the host uuid")); + goto err_exit; } -} - - -static virDomainDiskDefPtr -findDomainDiskByPath(virDomainObjPtr vm, - const char *path) -{ - int i; - for (i = 0; i < vm->def->ndisks; i++) { - virDomainDiskDefPtr disk; + /* Security driver data */ + const char *doi, *model; - disk = vm->def->disks[i]; - if (disk->src != NULL && STREQ(disk->src, path)) - return disk; + doi = virSecurityManagerGetDOI(driver->securityManager); + model = virSecurityManagerGetModel(driver->securityManager); + if (STRNEQ(model, "none")) { + if (!(caps->host.secModel.model = strdup(model))) + goto no_memory; + if (!(caps->host.secModel.doi = strdup(doi))) + goto no_memory; } - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("no disk found with path %s"), - path); - return NULL; -} - -static virDomainDiskDefPtr -findDomainDiskByAlias(virDomainObjPtr vm, - const char *alias) -{ - int i; - - if (STRPREFIX(alias, QEMU_DRIVE_HOST_PREFIX)) - alias += strlen(QEMU_DRIVE_HOST_PREFIX); - - for (i = 0; i < vm->def->ndisks; i++) { - virDomainDiskDefPtr disk; + VIR_DEBUG("Initialized caps for security driver \"%s\" with " + "DOI \"%s\"", model, doi); - disk = vm->def->disks[i]; - if (disk->info.alias != NULL && STREQ(disk->info.alias, alias)) - return disk; - } + return caps; - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("no disk found with alias %s"), - alias); +no_memory: + virReportOOMError(); +err_exit: + virCapabilitiesFree(caps); return NULL; } -static int -getVolumeQcowPassphrase(virConnectPtr conn, - virDomainDiskDefPtr disk, - char **secretRet, - size_t *secretLen) +static void qemuDomainSnapshotLoad(void *payload, + const char *name ATTRIBUTE_UNUSED, + void *data) { - virSecretPtr secret; - char *passphrase; - unsigned char *data; - size_t size; - int ret = -1; - virStorageEncryptionPtr enc; - - if (!disk->encryption) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("disk %s does not have any encryption information"), - disk->src); - return -1; - } - enc = disk->encryption; - - if (!conn) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("cannot find secrets without a connection")); - goto cleanup; - } - - if (conn->secretDriver == NULL || - conn->secretDriver->lookupByUUID == NULL || - conn->secretDriver->getValue == NULL) { - qemuReportError(VIR_ERR_NO_SUPPORT, "%s", - _("secret storage not supported")); - goto cleanup; - } + virDomainObjPtr vm = (virDomainObjPtr)payload; + char *baseDir = (char *)data; + char *snapDir = NULL; + DIR *dir = NULL; + struct dirent *entry; + char *xmlStr; + int ret; + char *fullpath; + virDomainSnapshotDefPtr def = NULL; + char ebuf[1024]; - if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW || - enc->nsecrets != 1 || - enc->secrets[0]->type != - VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) { - qemuReportError(VIR_ERR_XML_ERROR, - _("invalid <encryption> for volume %s"), disk->src); + virDomainObjLock(vm); + if (virAsprintf(&snapDir, "%s/%s", baseDir, vm->def->name) < 0) { + VIR_ERROR(_("Failed to allocate memory for snapshot directory for domain %s"), + vm->def->name); goto cleanup; } - secret = conn->secretDriver->lookupByUUID(conn, - enc->secrets[0]->uuid); - if (secret == NULL) - goto cleanup; - data = conn->secretDriver->getValue(secret, &size, - VIR_SECRET_GET_VALUE_INTERNAL_CALL); - virUnrefSecret(secret); - if (data == NULL) - goto cleanup; + VIR_INFO("Scanning for snapshots for domain %s in %s", vm->def->name, + snapDir); - if (memchr(data, '\0', size) != NULL) { - memset(data, 0, size); - VIR_FREE(data); - qemuReportError(VIR_ERR_XML_ERROR, - _("format='qcow' passphrase for %s must not contain a " - "'\\0'"), disk->src); + if (!(dir = opendir(snapDir))) { + if (errno != ENOENT) + VIR_ERROR(_("Failed to open snapshot directory %s for domain %s: %s"), + snapDir, vm->def->name, + virStrerror(errno, ebuf, sizeof(ebuf))); goto cleanup; } - if (VIR_ALLOC_N(passphrase, size + 1) < 0) { - memset(data, 0, size); - VIR_FREE(data); - virReportOOMError(); - goto cleanup; - } - memcpy(passphrase, data, size); - passphrase[size] = '\0'; + while ((entry = readdir(dir))) { + if (entry->d_name[0] == '.') + continue; - memset(data, 0, size); - VIR_FREE(data); + /* NB: ignoring errors, so one malformed config doesn't + kill the whole process */ + VIR_INFO("Loading snapshot file '%s'", entry->d_name); - *secretRet = passphrase; - *secretLen = size; + if (virAsprintf(&fullpath, "%s/%s", snapDir, entry->d_name) < 0) { + VIR_ERROR0(_("Failed to allocate memory for path")); + continue; + } - ret = 0; + ret = virFileReadAll(fullpath, 1024*1024*1, &xmlStr); + if (ret < 0) { + /* Nothing we can do here, skip this one */ + VIR_ERROR(_("Failed to read snapshot file %s: %s"), fullpath, + virStrerror(errno, ebuf, sizeof(ebuf))); + VIR_FREE(fullpath); + continue; + } -cleanup: - return ret; -} + def = virDomainSnapshotDefParseString(xmlStr, 0); + if (def == NULL) { + /* Nothing we can do here, skip this one */ + VIR_ERROR(_("Failed to parse snapshot XML from file '%s'"), fullpath); + VIR_FREE(fullpath); + VIR_FREE(xmlStr); + continue; + } -static int -findVolumeQcowPassphrase(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virConnectPtr conn, - virDomainObjPtr vm, - const char *path, - char **secretRet, - size_t *secretLen) -{ - virDomainDiskDefPtr disk; - int ret = -1; + virDomainSnapshotAssignDef(&vm->snapshots, def); - virDomainObjLock(vm); - disk = findDomainDiskByPath(vm, path); + VIR_FREE(fullpath); + VIR_FREE(xmlStr); + } - if (!disk) - goto cleanup; + /* FIXME: qemu keeps internal track of snapshots. We can get access + * to this info via the "info snapshots" monitor command for running + * domains, or via "qemu-img snapshot -l" for shutoff domains. It would + * be nice to update our internal state based on that, but there is a + * a problem. qemu doesn't track all of the same metadata that we do. + * In particular we wouldn't be able to fill in the <parent>, which is + * pretty important in our metadata. + */ - ret = getVolumeQcowPassphrase(conn, disk, secretRet, secretLen); + virResetLastError(); cleanup: + if (dir) + closedir(dir); + VIR_FREE(snapDir); virDomainObjUnlock(vm); - return ret; } - +/** + * qemudStartup: + * + * Initialization function for the QEmu daemon + */ static int -qemuHandleDomainReset(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm) -{ - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr event; +qemudStartup(int privileged) { + char *base = NULL; + char *driverConf = NULL; + int rc; + virConnectPtr conn = NULL; - virDomainObjLock(vm); - event = virDomainEventRebootNewFromObj(vm); - virDomainObjUnlock(vm); + if (VIR_ALLOC(qemu_driver) < 0) + return -1; - if (event) { - qemuDriverLock(driver); - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); + if (virMutexInit(&qemu_driver->lock) < 0) { + VIR_ERROR0(_("cannot initialize mutex")); + VIR_FREE(qemu_driver); + return -1; } + qemuDriverLock(qemu_driver); + qemu_driver->privileged = privileged; - return 0; -} - - -static int -qemuHandleDomainShutdown(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm) -{ - virDomainObjLock(vm); - ((qemuDomainObjPrivatePtr) vm->privateData)->gotShutdown = true; - virDomainObjUnlock(vm); - - return 0; -} - + /* Don't have a dom0 so start from 1 */ + qemu_driver->nextvmid = 1; -static int -qemuHandleDomainStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm) -{ - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr event = NULL; + if (virDomainObjListInit(&qemu_driver->domains) < 0) + goto out_of_memory; - virDomainObjLock(vm); - if (vm->state == VIR_DOMAIN_RUNNING) { - VIR_DEBUG("Transitioned guest %s to paused state due to unknown event", vm->def->name); + /* Init callback list */ + if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0) + goto out_of_memory; + if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew())) + goto out_of_memory; - vm->state = VIR_DOMAIN_PAUSED; - event = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_SUSPENDED, - VIR_DOMAIN_EVENT_SUSPENDED_PAUSED); + if ((qemu_driver->domainEventTimer = + virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0) + goto error; - if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) - VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name); - } - virDomainObjUnlock(vm); + /* Allocate bitmap for vnc port reservation */ + if ((qemu_driver->reservedVNCPorts = + virBitmapAlloc(QEMU_VNC_PORT_MAX - QEMU_VNC_PORT_MIN)) == NULL) + goto out_of_memory; - if (event) { - qemuDriverLock(driver); - if (event) - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); - } + /* read the host sysinfo */ + if (privileged) + qemu_driver->hostsysinfo = virSysinfoRead(); - return 0; -} + if (privileged) { + if (virAsprintf(&qemu_driver->logDir, + "%s/log/libvirt/qemu", LOCALSTATEDIR) == -1) + goto out_of_memory; + if ((base = strdup (SYSCONFDIR "/libvirt")) == NULL) + goto out_of_memory; -static int -qemuHandleDomainRTCChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm, - long long offset) -{ - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr event; + if (virAsprintf(&qemu_driver->stateDir, + "%s/run/libvirt/qemu", LOCALSTATEDIR) == -1) + goto out_of_memory; - virDomainObjLock(vm); - event = virDomainEventRTCChangeNewFromObj(vm, offset); + if (virAsprintf(&qemu_driver->libDir, + "%s/lib/libvirt/qemu", LOCALSTATEDIR) == -1) + goto out_of_memory; - if (vm->def->clock.offset == VIR_DOMAIN_CLOCK_OFFSET_VARIABLE) - vm->def->clock.data.adjustment = offset; + if (virAsprintf(&qemu_driver->cacheDir, + "%s/cache/libvirt/qemu", LOCALSTATEDIR) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->saveDir, + "%s/lib/libvirt/qemu/save", LOCALSTATEDIR) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->snapshotDir, + "%s/lib/libvirt/qemu/snapshot", LOCALSTATEDIR) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->autoDumpPath, + "%s/lib/libvirt/qemu/dump", LOCALSTATEDIR) == -1) + goto out_of_memory; + } else { + uid_t uid = geteuid(); + char *userdir = virGetUserDirectory(uid); + if (!userdir) + goto error; - if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) - VIR_WARN0("unable to save domain status with RTC change"); + if (virAsprintf(&qemu_driver->logDir, + "%s/.libvirt/qemu/log", userdir) == -1) { + VIR_FREE(userdir); + goto out_of_memory; + } - virDomainObjUnlock(vm); + if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) { + VIR_FREE(userdir); + goto out_of_memory; + } + VIR_FREE(userdir); - if (event) { - qemuDriverLock(driver); - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); + if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->libDir, "%s/qemu/lib", base) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->cacheDir, "%s/qemu/cache", base) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1) + goto out_of_memory; + if (virAsprintf(&qemu_driver->autoDumpPath, "%s/qemu/dump", base) == -1) + goto out_of_memory; } - return 0; -} - - -static int -qemuHandleDomainWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm, - int action) -{ - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr watchdogEvent = NULL; - virDomainEventPtr lifecycleEvent = NULL; - - virDomainObjLock(vm); - watchdogEvent = virDomainEventWatchdogNewFromObj(vm, action); - - if (action == VIR_DOMAIN_EVENT_WATCHDOG_PAUSE && - vm->state == VIR_DOMAIN_RUNNING) { - VIR_DEBUG("Transitioned guest %s to paused state due to watchdog", vm->def->name); - - vm->state = VIR_DOMAIN_PAUSED; - lifecycleEvent = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_SUSPENDED, - VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG); - - if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) - VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name); + if (virFileMakePath(qemu_driver->stateDir) != 0) { + char ebuf[1024]; + VIR_ERROR(_("Failed to create state dir '%s': %s"), + qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf)); + goto error; } - - if (vm->def->watchdog->action == VIR_DOMAIN_WATCHDOG_ACTION_DUMP) { - struct watchdogEvent *wdEvent; - if (VIR_ALLOC(wdEvent) == 0) { - wdEvent->action = VIR_DOMAIN_WATCHDOG_ACTION_DUMP; - wdEvent->vm = vm; - ignore_value(virThreadPoolSendJob(driver->workerPool, wdEvent)); - } else - virReportOOMError(); + if (virFileMakePath(qemu_driver->libDir) != 0) { + char ebuf[1024]; + VIR_ERROR(_("Failed to create lib dir '%s': %s"), + qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf)); + goto error; } - - virDomainObjUnlock(vm); - - if (watchdogEvent || lifecycleEvent) { - qemuDriverLock(driver); - if (watchdogEvent) - qemuDomainEventQueue(driver, watchdogEvent); - if (lifecycleEvent) - qemuDomainEventQueue(driver, lifecycleEvent); - qemuDriverUnlock(driver); + if (virFileMakePath(qemu_driver->cacheDir) != 0) { + char ebuf[1024]; + VIR_ERROR(_("Failed to create cache dir '%s': %s"), + qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf)); + goto error; } - - return 0; -} - - -static int -qemuHandleDomainIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm, - const char *diskAlias, - int action, - const char *reason) -{ - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr ioErrorEvent = NULL; - virDomainEventPtr ioErrorEvent2 = NULL; - virDomainEventPtr lifecycleEvent = NULL; - const char *srcPath; - const char *devAlias; - virDomainDiskDefPtr disk; - - virDomainObjLock(vm); - disk = findDomainDiskByAlias(vm, diskAlias); - - if (disk) { - srcPath = disk->src; - devAlias = disk->info.alias; - } else { - srcPath = ""; - devAlias = ""; + if (virFileMakePath(qemu_driver->saveDir) != 0) { + char ebuf[1024]; + VIR_ERROR(_("Failed to create save dir '%s': %s"), + qemu_driver->saveDir, virStrerror(errno, ebuf, sizeof ebuf)); + goto error; + } + if (virFileMakePath(qemu_driver->snapshotDir) != 0) { + char ebuf[1024]; + VIR_ERROR(_("Failed to create save dir '%s': %s"), + qemu_driver->snapshotDir, virStrerror(errno, ebuf, sizeof ebuf)); + goto error; + } + if (virFileMakePath(qemu_driver->autoDumpPath) != 0) { + char ebuf[1024]; + VIR_ERROR(_("Failed to create dump dir '%s': %s"), + qemu_driver->autoDumpPath, virStrerror(errno, ebuf, sizeof ebuf)); + goto error; } - ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action); - ioErrorEvent2 = virDomainEventIOErrorReasonNewFromObj(vm, srcPath, devAlias, action, reason); - - if (action == VIR_DOMAIN_EVENT_IO_ERROR_PAUSE && - vm->state == VIR_DOMAIN_RUNNING) { - VIR_DEBUG("Transitioned guest %s to paused state due to IO error", vm->def->name); + /* Configuration paths are either ~/.libvirt/qemu/... (session) or + * /etc/libvirt/qemu/... (system). + */ + if (virAsprintf(&driverConf, "%s/qemu.conf", base) < 0 || + virAsprintf(&qemu_driver->configDir, "%s/qemu", base) < 0 || + virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) < 0) + goto out_of_memory; - vm->state = VIR_DOMAIN_PAUSED; - lifecycleEvent = virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_SUSPENDED, - VIR_DOMAIN_EVENT_SUSPENDED_IOERROR); + VIR_FREE(base); - if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) - VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name); + rc = virCgroupForDriver("qemu", &qemu_driver->cgroup, privileged, 1); + if (rc < 0) { + char buf[1024]; + VIR_WARN("Unable to create cgroup for driver: %s", + virStrerror(-rc, buf, sizeof(buf))); } - virDomainObjUnlock(vm); - if (ioErrorEvent || ioErrorEvent2 || lifecycleEvent) { - qemuDriverLock(driver); - if (ioErrorEvent) - qemuDomainEventQueue(driver, ioErrorEvent); - if (ioErrorEvent2) - qemuDomainEventQueue(driver, ioErrorEvent2); - if (lifecycleEvent) - qemuDomainEventQueue(driver, lifecycleEvent); - qemuDriverUnlock(driver); + if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) { + goto error; } + VIR_FREE(driverConf); - return 0; -} + if (qemuSecurityInit(qemu_driver) < 0) + goto error; - -static int -qemuHandleDomainGraphics(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm, - int phase, - int localFamily, - const char *localNode, - const char *localService, - int remoteFamily, - const char *remoteNode, - const char *remoteService, - const char *authScheme, - const char *x509dname, - const char *saslUsername) -{ - struct qemud_driver *driver = qemu_driver; - virDomainEventPtr event; - virDomainEventGraphicsAddressPtr localAddr = NULL; - virDomainEventGraphicsAddressPtr remoteAddr = NULL; - virDomainEventGraphicsSubjectPtr subject = NULL; - int i; - - virDomainObjLock(vm); - - if (VIR_ALLOC(localAddr) < 0) - goto no_memory; - localAddr->family = localFamily; - if (!(localAddr->service = strdup(localService)) || - !(localAddr->node = strdup(localNode))) - goto no_memory; - - if (VIR_ALLOC(remoteAddr) < 0) - goto no_memory; - remoteAddr->family = remoteFamily; - if (!(remoteAddr->service = strdup(remoteService)) || - !(remoteAddr->node = strdup(remoteNode))) - goto no_memory; - - if (VIR_ALLOC(subject) < 0) - goto no_memory; - if (x509dname) { - if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0) - goto no_memory; - if (!(subject->identities[subject->nidentity].type = strdup("x509dname")) || - !(subject->identities[subject->nidentity].name = strdup(x509dname))) - goto no_memory; - subject->nidentity++; - } - if (saslUsername) { - if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0) - goto no_memory; - if (!(subject->identities[subject->nidentity].type = strdup("saslUsername")) || - !(subject->identities[subject->nidentity].name = strdup(saslUsername))) - goto no_memory; - subject->nidentity++; - } - - event = virDomainEventGraphicsNewFromObj(vm, phase, localAddr, remoteAddr, authScheme, subject); - virDomainObjUnlock(vm); - - if (event) { - qemuDriverLock(driver); - qemuDomainEventQueue(driver, event); - qemuDriverUnlock(driver); - } - - return 0; - -no_memory: - virReportOOMError(); - if (localAddr) { - VIR_FREE(localAddr->service); - VIR_FREE(localAddr->node); - VIR_FREE(localAddr); - } - if (remoteAddr) { - VIR_FREE(remoteAddr->service); - VIR_FREE(remoteAddr->node); - VIR_FREE(remoteAddr); - } - if (subject) { - for (i = 0 ; i < subject->nidentity ; i++) { - VIR_FREE(subject->identities[i].type); - VIR_FREE(subject->identities[i].name); - } - VIR_FREE(subject->identities); - VIR_FREE(subject); - } - - return -1; -} - - -static void qemuHandleMonitorDestroy(qemuMonitorPtr mon, - virDomainObjPtr vm) -{ - qemuDomainObjPrivatePtr priv = vm->privateData; - if (priv->mon == mon) - priv->mon = NULL; - virDomainObjUnref(vm); -} - -static qemuMonitorCallbacks monitorCallbacks = { - .destroy = qemuHandleMonitorDestroy, - .eofNotify = qemuHandleMonitorEOF, - .diskSecretLookup = findVolumeQcowPassphrase, - .domainShutdown = qemuHandleDomainShutdown, - .domainStop = qemuHandleDomainStop, - .domainReset = qemuHandleDomainReset, - .domainRTCChange = qemuHandleDomainRTCChange, - .domainWatchdog = qemuHandleDomainWatchdog, - .domainIOError = qemuHandleDomainIOError, - .domainGraphics = qemuHandleDomainGraphics, -}; - -static int -qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm) -{ - qemuDomainObjPrivatePtr priv = vm->privateData; - int ret = -1; - - if (virSecurityManagerSetSocketLabel(driver->securityManager, vm) < 0) { - VIR_ERROR(_("Failed to set security context for monitor for %s"), - vm->def->name); - goto error; - } - - /* Hold an extra reference because we can't allow 'vm' to be - * deleted while the monitor is active */ - virDomainObjRef(vm); - - priv->mon = qemuMonitorOpen(vm, - priv->monConfig, - priv->monJSON, - &monitorCallbacks); - - if (priv->mon == NULL) - virDomainObjUnref(vm); - - if (virSecurityManagerClearSocketLabel(driver->securityManager, vm) < 0) { - VIR_ERROR(_("Failed to clear security context for monitor for %s"), - vm->def->name); - goto error; - } - - if (priv->mon == NULL) { - VIR_INFO("Failed to connect monitor for %s", vm->def->name); - goto error; - } - - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - ret = qemuMonitorSetCapabilities(priv->mon); - qemuDomainObjExitMonitorWithDriver(driver, vm); - -error: - - return ret; -} - -struct virReconnectDomainData { - virConnectPtr conn; - struct qemud_driver *driver; -}; -/* - * Open an existing VM's monitor, re-detect VCPU threads - * and re-reserve the security labels in use - */ -static void -qemuReconnectDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque) -{ - virDomainObjPtr obj = payload; - struct virReconnectDomainData *data = opaque; - struct qemud_driver *driver = data->driver; - qemuDomainObjPrivatePtr priv; - unsigned long long qemuCmdFlags; - virConnectPtr conn = data->conn; - - virDomainObjLock(obj); - - VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name); - - priv = obj->privateData; - - /* Hold an extra reference because we can't allow 'vm' to be - * deleted if qemuConnectMonitor() failed */ - virDomainObjRef(obj); - - /* XXX check PID liveliness & EXE path */ - if (qemuConnectMonitor(driver, obj) < 0) - goto error; - - if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) { - goto error; - } - - /* XXX we should be persisting the original flags in the XML - * not re-detecting them, since the binary may have changed - * since launch time */ - if (qemuCapsExtractVersionInfo(obj->def->emulator, - NULL, - &qemuCmdFlags) >= 0 && - (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) { - priv->persistentAddrs = 1; - - if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)) || - qemuAssignDevicePCISlots(obj->def, priv->pciaddrs) < 0) - goto error; - } - - if (virSecurityManagerReserveLabel(driver->securityManager, obj) < 0) - goto error; - - if (qemudVMFiltersInstantiate(conn, obj->def)) - goto error; - - if (obj->def->id >= driver->nextvmid) - driver->nextvmid = obj->def->id + 1; - - if (virDomainObjUnref(obj) > 0) - virDomainObjUnlock(obj); - return; - -error: - if (!virDomainObjIsActive(obj)) { - if (virDomainObjUnref(obj) > 0) - virDomainObjUnlock(obj); - return; - } - - if (virDomainObjUnref(obj) > 0) { - /* We can't get the monitor back, so must kill the VM - * to remove danger of it ending up running twice if - * user tries to start it again later */ - qemudShutdownVMDaemon(driver, obj, 0); - if (!obj->persistent) - virDomainRemoveInactive(&driver->domains, obj); - else - virDomainObjUnlock(obj); - } -} - -/** - * qemudReconnectDomains - * - * Try to re-open the resources for live VMs that we care - * about. - */ -static void -qemuReconnectDomains(virConnectPtr conn, struct qemud_driver *driver) -{ - struct virReconnectDomainData data = {conn, driver}; - virHashForEach(driver->domains.objs, qemuReconnectDomain, &data); -} - - -static int -qemuSecurityInit(struct qemud_driver *driver) -{ - virSecurityManagerPtr mgr = virSecurityManagerNew(driver->securityDriverName, - driver->allowDiskFormatProbing); - if (!mgr) - goto error; - - if (driver->privileged) { - virSecurityManagerPtr dac = virSecurityManagerNewDAC(driver->user, - driver->group, - driver->allowDiskFormatProbing, - driver->dynamicOwnership); - if (!dac) - goto error; - - if (!(driver->securityManager = virSecurityManagerNewStack(mgr, - dac))) - goto error; - } else { - driver->securityManager = mgr; - } - - return 0; - -error: - VIR_ERROR0(_("Failed to initialize security drivers")); - virSecurityManagerFree(mgr); - return -1; -} - - -static virCapsPtr -qemuCreateCapabilities(virCapsPtr oldcaps, - struct qemud_driver *driver) -{ - virCapsPtr caps; - - /* Basic host arch / guest machine capabilities */ - if (!(caps = qemuCapsInit(oldcaps))) { - virReportOOMError(); - return NULL; - } - - if (driver->allowDiskFormatProbing) { - caps->defaultDiskDriverName = NULL; - caps->defaultDiskDriverType = NULL; - } else { - caps->defaultDiskDriverName = "qemu"; - caps->defaultDiskDriverType = "raw"; - } - - qemuDomainSetPrivateDataHooks(caps); - qemuDomainSetNamespaceHooks(caps); - - if (virGetHostUUID(caps->host.host_uuid)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("cannot get the host uuid")); - goto err_exit; - } - - /* Security driver data */ - const char *doi, *model; - - doi = virSecurityManagerGetDOI(driver->securityManager); - model = virSecurityManagerGetModel(driver->securityManager); - if (STRNEQ(model, "none")) { - if (!(caps->host.secModel.model = strdup(model))) - goto no_memory; - if (!(caps->host.secModel.doi = strdup(doi))) - goto no_memory; - } - - VIR_DEBUG("Initialized caps for security driver \"%s\" with " - "DOI \"%s\"", model, doi); - - return caps; - -no_memory: - virReportOOMError(); -err_exit: - virCapabilitiesFree(caps); - return NULL; -} - -static void qemuDomainSnapshotLoad(void *payload, - const char *name ATTRIBUTE_UNUSED, - void *data) -{ - virDomainObjPtr vm = (virDomainObjPtr)payload; - char *baseDir = (char *)data; - char *snapDir = NULL; - DIR *dir = NULL; - struct dirent *entry; - char *xmlStr; - int ret; - char *fullpath; - virDomainSnapshotDefPtr def = NULL; - char ebuf[1024]; - - virDomainObjLock(vm); - if (virAsprintf(&snapDir, "%s/%s", baseDir, vm->def->name) < 0) { - VIR_ERROR(_("Failed to allocate memory for snapshot directory for domain %s"), - vm->def->name); - goto cleanup; - } - - VIR_INFO("Scanning for snapshots for domain %s in %s", vm->def->name, - snapDir); - - if (!(dir = opendir(snapDir))) { - if (errno != ENOENT) - VIR_ERROR(_("Failed to open snapshot directory %s for domain %s: %s"), - snapDir, vm->def->name, - virStrerror(errno, ebuf, sizeof(ebuf))); - goto cleanup; - } - - while ((entry = readdir(dir))) { - if (entry->d_name[0] == '.') - continue; - - /* NB: ignoring errors, so one malformed config doesn't - kill the whole process */ - VIR_INFO("Loading snapshot file '%s'", entry->d_name); - - if (virAsprintf(&fullpath, "%s/%s", snapDir, entry->d_name) < 0) { - VIR_ERROR0(_("Failed to allocate memory for path")); - continue; - } - - ret = virFileReadAll(fullpath, 1024*1024*1, &xmlStr); - if (ret < 0) { - /* Nothing we can do here, skip this one */ - VIR_ERROR(_("Failed to read snapshot file %s: %s"), fullpath, - virStrerror(errno, ebuf, sizeof(ebuf))); - VIR_FREE(fullpath); - continue; - } - - def = virDomainSnapshotDefParseString(xmlStr, 0); - if (def == NULL) { - /* Nothing we can do here, skip this one */ - VIR_ERROR(_("Failed to parse snapshot XML from file '%s'"), fullpath); - VIR_FREE(fullpath); - VIR_FREE(xmlStr); - continue; - } - - virDomainSnapshotAssignDef(&vm->snapshots, def); - - VIR_FREE(fullpath); - VIR_FREE(xmlStr); - } - - /* FIXME: qemu keeps internal track of snapshots. We can get access - * to this info via the "info snapshots" monitor command for running - * domains, or via "qemu-img snapshot -l" for shutoff domains. It would - * be nice to update our internal state based on that, but there is a - * a problem. qemu doesn't track all of the same metadata that we do. - * In particular we wouldn't be able to fill in the <parent>, which is - * pretty important in our metadata. - */ - - virResetLastError(); - -cleanup: - if (dir) - closedir(dir); - VIR_FREE(snapDir); - virDomainObjUnlock(vm); -} - -/** - * qemudStartup: - * - * Initialization function for the QEmu daemon - */ -static int -qemudStartup(int privileged) { - char *base = NULL; - char *driverConf = NULL; - int rc; - virConnectPtr conn = NULL; - - if (VIR_ALLOC(qemu_driver) < 0) - return -1; - - if (virMutexInit(&qemu_driver->lock) < 0) { - VIR_ERROR0(_("cannot initialize mutex")); - VIR_FREE(qemu_driver); - return -1; - } - qemuDriverLock(qemu_driver); - qemu_driver->privileged = privileged; - - /* Don't have a dom0 so start from 1 */ - qemu_driver->nextvmid = 1; - - if (virDomainObjListInit(&qemu_driver->domains) < 0) - goto out_of_memory; - - /* Init callback list */ - if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0) - goto out_of_memory; - if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew())) - goto out_of_memory; - - if ((qemu_driver->domainEventTimer = - virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0) - goto error; - - /* Allocate bitmap for vnc port reservation */ - if ((qemu_driver->reservedVNCPorts = - virBitmapAlloc(QEMU_VNC_PORT_MAX - QEMU_VNC_PORT_MIN)) == NULL) - goto out_of_memory; - - /* read the host sysinfo */ - if (privileged) - qemu_driver->hostsysinfo = virSysinfoRead(); - - if (privileged) { - if (virAsprintf(&qemu_driver->logDir, - "%s/log/libvirt/qemu", LOCALSTATEDIR) == -1) - goto out_of_memory; - - if ((base = strdup (SYSCONFDIR "/libvirt")) == NULL) - goto out_of_memory; - - if (virAsprintf(&qemu_driver->stateDir, - "%s/run/libvirt/qemu", LOCALSTATEDIR) == -1) - goto out_of_memory; - - if (virAsprintf(&qemu_driver->libDir, - "%s/lib/libvirt/qemu", LOCALSTATEDIR) == -1) - goto out_of_memory; - - if (virAsprintf(&qemu_driver->cacheDir, - "%s/cache/libvirt/qemu", LOCALSTATEDIR) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->saveDir, - "%s/lib/libvirt/qemu/save", LOCALSTATEDIR) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->snapshotDir, - "%s/lib/libvirt/qemu/snapshot", LOCALSTATEDIR) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->autoDumpPath, - "%s/lib/libvirt/qemu/dump", LOCALSTATEDIR) == -1) - goto out_of_memory; - } else { - uid_t uid = geteuid(); - char *userdir = virGetUserDirectory(uid); - if (!userdir) - goto error; - - if (virAsprintf(&qemu_driver->logDir, - "%s/.libvirt/qemu/log", userdir) == -1) { - VIR_FREE(userdir); - goto out_of_memory; - } - - if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) { - VIR_FREE(userdir); - goto out_of_memory; - } - VIR_FREE(userdir); - - if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->libDir, "%s/qemu/lib", base) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->cacheDir, "%s/qemu/cache", base) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1) - goto out_of_memory; - if (virAsprintf(&qemu_driver->autoDumpPath, "%s/qemu/dump", base) == -1) - goto out_of_memory; - } - - if (virFileMakePath(qemu_driver->stateDir) != 0) { - char ebuf[1024]; - VIR_ERROR(_("Failed to create state dir '%s': %s"), - qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf)); - goto error; - } - if (virFileMakePath(qemu_driver->libDir) != 0) { - char ebuf[1024]; - VIR_ERROR(_("Failed to create lib dir '%s': %s"), - qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf)); - goto error; - } - if (virFileMakePath(qemu_driver->cacheDir) != 0) { - char ebuf[1024]; - VIR_ERROR(_("Failed to create cache dir '%s': %s"), - qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf)); - goto error; - } - if (virFileMakePath(qemu_driver->saveDir) != 0) { - char ebuf[1024]; - VIR_ERROR(_("Failed to create save dir '%s': %s"), - qemu_driver->saveDir, virStrerror(errno, ebuf, sizeof ebuf)); - goto error; - } - if (virFileMakePath(qemu_driver->snapshotDir) != 0) { - char ebuf[1024]; - VIR_ERROR(_("Failed to create save dir '%s': %s"), - qemu_driver->snapshotDir, virStrerror(errno, ebuf, sizeof ebuf)); - goto error; - } - if (virFileMakePath(qemu_driver->autoDumpPath) != 0) { - char ebuf[1024]; - VIR_ERROR(_("Failed to create dump dir '%s': %s"), - qemu_driver->autoDumpPath, virStrerror(errno, ebuf, sizeof ebuf)); - goto error; - } - - /* Configuration paths are either ~/.libvirt/qemu/... (session) or - * /etc/libvirt/qemu/... (system). - */ - if (virAsprintf(&driverConf, "%s/qemu.conf", base) < 0 || - virAsprintf(&qemu_driver->configDir, "%s/qemu", base) < 0 || - virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) < 0) - goto out_of_memory; - - VIR_FREE(base); - - rc = virCgroupForDriver("qemu", &qemu_driver->cgroup, privileged, 1); - if (rc < 0) { - char buf[1024]; - VIR_WARN("Unable to create cgroup for driver: %s", - virStrerror(-rc, buf, sizeof(buf))); - } - - if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) { - goto error; - } - VIR_FREE(driverConf); - - if (qemuSecurityInit(qemu_driver) < 0) - goto error; - - if ((qemu_driver->caps = qemuCreateCapabilities(NULL, - qemu_driver)) == NULL) + if ((qemu_driver->caps = qemuCreateCapabilities(NULL, + qemu_driver)) == NULL) goto error; if ((qemu_driver->activePciHostdevs = pciDeviceListNew()) == NULL) - goto error; - - if (privileged) { - if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) { - virReportSystemError(errno, - _("unable to set ownership of '%s' to user %d:%d"), - qemu_driver->libDir, qemu_driver->user, qemu_driver->group); - goto error; - } - if (chown(qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group) < 0) { - virReportSystemError(errno, - _("unable to set ownership of '%s' to %d:%d"), - qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group); - goto error; - } - if (chown(qemu_driver->saveDir, qemu_driver->user, qemu_driver->group) < 0) { - virReportSystemError(errno, - _("unable to set ownership of '%s' to %d:%d"), - qemu_driver->saveDir, qemu_driver->user, qemu_driver->group); - goto error; - } - if (chown(qemu_driver->snapshotDir, qemu_driver->user, qemu_driver->group) < 0) { - virReportSystemError(errno, - _("unable to set ownership of '%s' to %d:%d"), - qemu_driver->snapshotDir, qemu_driver->user, qemu_driver->group); - goto error; - } - } - - /* If hugetlbfs is present, then we need to create a sub-directory within - * it, since we can't assume the root mount point has permissions that - * will let our spawned QEMU instances use it. - * - * NB the check for '/', since user may config "" to disable hugepages - * even when mounted - */ - if (qemu_driver->hugetlbfs_mount && - qemu_driver->hugetlbfs_mount[0] == '/') { - char *mempath = NULL; - if (virAsprintf(&mempath, "%s/libvirt/qemu", qemu_driver->hugetlbfs_mount) < 0) - goto out_of_memory; - - if ((rc = virFileMakePath(mempath)) != 0) { - virReportSystemError(rc, - _("unable to create hugepage path %s"), mempath); - VIR_FREE(mempath); - goto error; - } - if (qemu_driver->privileged && - chown(mempath, qemu_driver->user, qemu_driver->group) < 0) { - virReportSystemError(errno, - _("unable to set ownership on %s to %d:%d"), - mempath, qemu_driver->user, qemu_driver->group); - VIR_FREE(mempath); - goto error; - } - - qemu_driver->hugepage_path = mempath; - } - - /* Get all the running persistent or transient configs first */ - if (virDomainLoadAllConfigs(qemu_driver->caps, - &qemu_driver->domains, - qemu_driver->stateDir, - NULL, - 1, NULL, NULL) < 0) - goto error; - - conn = virConnectOpen(qemu_driver->privileged ? - "qemu:///system" : - "qemu:///session"); - - qemuReconnectDomains(conn, qemu_driver); - - /* Then inactive persistent configs */ - if (virDomainLoadAllConfigs(qemu_driver->caps, - &qemu_driver->domains, - qemu_driver->configDir, - qemu_driver->autostartDir, - 0, NULL, NULL) < 0) - goto error; - - - virHashForEach(qemu_driver->domains.objs, qemuDomainSnapshotLoad, - qemu_driver->snapshotDir); - - qemuDriverUnlock(qemu_driver); - - qemudAutostartConfigs(qemu_driver); - - qemu_driver->workerPool = virThreadPoolNew(0, 1, processWatchdogEvent, qemu_driver); - if (!qemu_driver->workerPool) - goto error; - - if (conn) - virConnectClose(conn); - - return 0; - -out_of_memory: - virReportOOMError(); -error: - if (qemu_driver) - qemuDriverUnlock(qemu_driver); - if (conn) - virConnectClose(conn); - VIR_FREE(base); - VIR_FREE(driverConf); - qemudShutdown(); - return -1; -} - -static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque) -{ - struct qemud_driver *driver = opaque; - - if (newVM) { - virDomainEventPtr event = - virDomainEventNewFromObj(vm, - VIR_DOMAIN_EVENT_DEFINED, - VIR_DOMAIN_EVENT_DEFINED_ADDED); - if (event) - qemuDomainEventQueue(driver, event); - } -} - -/** - * qemudReload: - * - * Function to restart the QEmu daemon, it will recheck the configuration - * files and update its state and the networking - */ -static int -qemudReload(void) { - if (!qemu_driver) - return 0; - - qemuDriverLock(qemu_driver); - virDomainLoadAllConfigs(qemu_driver->caps, - &qemu_driver->domains, - qemu_driver->configDir, - qemu_driver->autostartDir, - 0, qemudNotifyLoadDomain, qemu_driver); - qemuDriverUnlock(qemu_driver); - - qemudAutostartConfigs(qemu_driver); - - return 0; -} - -/** - * qemudActive: - * - * Checks if the QEmu daemon is active, i.e. has an active domain or - * an active network - * - * Returns 1 if active, 0 otherwise - */ -static int -qemudActive(void) { - int active = 0; - - if (!qemu_driver) - return 0; - - /* XXX having to iterate here is not great because it requires many locks */ - qemuDriverLock(qemu_driver); - active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1); - qemuDriverUnlock(qemu_driver); - return active; -} - -/** - * qemudShutdown: - * - * Shutdown the QEmu daemon, it will stop all active domains and networks - */ -static int -qemudShutdown(void) { - int i; - - if (!qemu_driver) - return -1; - - qemuDriverLock(qemu_driver); - pciDeviceListFree(qemu_driver->activePciHostdevs); - virCapabilitiesFree(qemu_driver->caps); - - virDomainObjListDeinit(&qemu_driver->domains); - virBitmapFree(qemu_driver->reservedVNCPorts); - - virSysinfoDefFree(qemu_driver->hostsysinfo); - - VIR_FREE(qemu_driver->configDir); - VIR_FREE(qemu_driver->autostartDir); - VIR_FREE(qemu_driver->logDir); - VIR_FREE(qemu_driver->stateDir); - VIR_FREE(qemu_driver->libDir); - VIR_FREE(qemu_driver->cacheDir); - VIR_FREE(qemu_driver->saveDir); - VIR_FREE(qemu_driver->snapshotDir); - VIR_FREE(qemu_driver->autoDumpPath); - VIR_FREE(qemu_driver->vncTLSx509certdir); - VIR_FREE(qemu_driver->vncListen); - VIR_FREE(qemu_driver->vncPassword); - VIR_FREE(qemu_driver->vncSASLdir); - VIR_FREE(qemu_driver->spiceTLSx509certdir); - VIR_FREE(qemu_driver->spiceListen); - VIR_FREE(qemu_driver->spicePassword); - VIR_FREE(qemu_driver->hugetlbfs_mount); - VIR_FREE(qemu_driver->hugepage_path); - VIR_FREE(qemu_driver->saveImageFormat); - VIR_FREE(qemu_driver->dumpImageFormat); - - virSecurityManagerFree(qemu_driver->securityManager); - - ebtablesContextFree(qemu_driver->ebtables); - - if (qemu_driver->cgroupDeviceACL) { - for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++) - VIR_FREE(qemu_driver->cgroupDeviceACL[i]); - VIR_FREE(qemu_driver->cgroupDeviceACL); - } - - /* Free domain callback list */ - virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks); - virDomainEventQueueFree(qemu_driver->domainEventQueue); - - if (qemu_driver->domainEventTimer != -1) - virEventRemoveTimeout(qemu_driver->domainEventTimer); - - if (qemu_driver->brctl) - brShutdown(qemu_driver->brctl); - - virCgroupFree(&qemu_driver->cgroup); - - qemuDriverUnlock(qemu_driver); - virMutexDestroy(&qemu_driver->lock); - virThreadPoolFree(qemu_driver->workerPool); - VIR_FREE(qemu_driver); - - return 0; -} - -typedef int qemuLogHandleOutput(virDomainObjPtr vm, - const char *output, - int fd); - -/* - * Returns -1 for error, 0 on success - */ -static int -qemudReadLogOutput(virDomainObjPtr vm, - int fd, - char *buf, - size_t buflen, - qemuLogHandleOutput func, - const char *what, - int timeout) -{ - int retries = (timeout*10); - int got = 0; - buf[0] = '\0'; - - while (retries) { - ssize_t func_ret, ret; - int isdead = 0; - - func_ret = func(vm, buf, fd); - - if (kill(vm->pid, 0) == -1 && errno == ESRCH) - isdead = 1; - - /* Any failures should be detected before we read the log, so we - * always have something useful to report on failure. */ - ret = saferead(fd, buf+got, buflen-got-1); - if (ret < 0) { - virReportSystemError(errno, - _("Failure while reading %s log output"), - what); - return -1; - } - - got += ret; - buf[got] = '\0'; - if (got == buflen-1) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Out of space while reading %s log output: %s"), - what, buf); - return -1; - } - - if (isdead) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Process exited while reading %s log output: %s"), - what, buf); - return -1; - } - - if (func_ret <= 0) - return func_ret; - - usleep(100*1000); - retries--; - } - - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Timed out while reading %s log output: %s"), - what, buf); - return -1; -} - - -/* - * Look at a chunk of data from the QEMU stdout logs and try to - * find a TTY device, as indicated by a line like - * - * char device redirected to /dev/pts/3 - * - * Returns -1 for error, 0 success, 1 continue reading - */ -static int -qemudExtractTTYPath(const char *haystack, - size_t *offset, - char **path) -{ - static const char needle[] = "char device redirected to"; - char *tmp, *dev; - - VIR_FREE(*path); - /* First look for our magic string */ - if (!(tmp = strstr(haystack + *offset, needle))) { - return 1; - } - tmp += sizeof(needle); - dev = tmp; - - /* - * And look for first whitespace character and nul terminate - * to mark end of the pty path - */ - while (*tmp) { - if (c_isspace(*tmp)) { - *path = strndup(dev, tmp-dev); - if (*path == NULL) { - virReportOOMError(); - return -1; - } - - /* ... now further update offset till we get EOL */ - *offset = tmp - haystack; - return 0; - } - tmp++; - } - - /* - * We found a path, but didn't find any whitespace, - * so it must be still incomplete - we should at - * least see a \n - indicate that we want to carry - * on trying again - */ - return 1; -} - -static int -qemudFindCharDevicePTYsMonitor(virDomainObjPtr vm, - virHashTablePtr paths) -{ - int i; - -#define LOOKUP_PTYS(array, arraylen, idprefix) \ - for (i = 0 ; i < (arraylen) ; i++) { \ - virDomainChrDefPtr chr = (array)[i]; \ - if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { \ - char id[16]; \ - \ - if (snprintf(id, sizeof(id), idprefix "%i", i) >= sizeof(id)) \ - return -1; \ - \ - const char *path = (const char *) virHashLookup(paths, id); \ - if (path == NULL) { \ - if (chr->source.data.file.path == NULL) { \ - /* neither the log output nor 'info chardev' had a */ \ - /* pty path for this chardev, report an error */ \ - qemuReportError(VIR_ERR_INTERNAL_ERROR, \ - _("no assigned pty for device %s"), id); \ - return -1; \ - } else { \ - /* 'info chardev' had no pty path for this chardev, */\ - /* but the log output had, so we're fine */ \ - continue; \ - } \ - } \ - \ - VIR_FREE(chr->source.data.file.path); \ - chr->source.data.file.path = strdup(path); \ - \ - if (chr->source.data.file.path == NULL) { \ - virReportOOMError(); \ - return -1; \ - } \ - } \ - } - - LOOKUP_PTYS(vm->def->serials, vm->def->nserials, "serial"); - LOOKUP_PTYS(vm->def->parallels, vm->def->nparallels, "parallel"); - LOOKUP_PTYS(vm->def->channels, vm->def->nchannels, "channel"); - if (vm->def->console) - LOOKUP_PTYS(&vm->def->console, 1, "console"); -#undef LOOKUP_PTYS - - return 0; -} - -static int -qemudFindCharDevicePTYs(virDomainObjPtr vm, - const char *output, - int fd ATTRIBUTE_UNUSED) -{ - size_t offset = 0; - int ret, i; - - /* The order in which QEMU prints out the PTY paths is - the order in which it procsses its serial and parallel - device args. This code must match that ordering.... */ - - /* first comes the serial devices */ - for (i = 0 ; i < vm->def->nserials ; i++) { - virDomainChrDefPtr chr = vm->def->serials[i]; - if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { - if ((ret = qemudExtractTTYPath(output, &offset, - &chr->source.data.file.path)) != 0) - return ret; - } - } - - /* then the parallel devices */ - for (i = 0 ; i < vm->def->nparallels ; i++) { - virDomainChrDefPtr chr = vm->def->parallels[i]; - if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { - if ((ret = qemudExtractTTYPath(output, &offset, - &chr->source.data.file.path)) != 0) - return ret; - } - } - - /* then the channel devices */ - for (i = 0 ; i < vm->def->nchannels ; i++) { - virDomainChrDefPtr chr = vm->def->channels[i]; - if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { - if ((ret = qemudExtractTTYPath(output, &offset, - &chr->source.data.file.path)) != 0) - return ret; - } - } - - return 0; -} - -static void qemudFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED) -{ - VIR_FREE(payload); -} - -static void -qemuReadLogFD(int logfd, char *buf, int maxlen, int off) -{ - int ret; - char *tmpbuf = buf + off; - - ret = saferead(logfd, tmpbuf, maxlen - off - 1); - if (ret < 0) { - ret = 0; - } - - tmpbuf[ret] = '\0'; -} - -static int -qemudWaitForMonitor(struct qemud_driver* driver, - virDomainObjPtr vm, off_t pos) -{ - char buf[4096] = ""; /* Plenty of space to get startup greeting */ - int logfd; - int ret = -1; - virHashTablePtr paths = NULL; - - if ((logfd = qemudLogReadFD(driver->logDir, vm->def->name, pos)) < 0) - return -1; - - if (qemudReadLogOutput(vm, logfd, buf, sizeof(buf), - qemudFindCharDevicePTYs, - "console", 30) < 0) - goto closelog; - - VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name); - if (qemuConnectMonitor(driver, vm) < 0) { - goto cleanup; - } - - /* Try to get the pty path mappings again via the monitor. This is much more - * reliable if it's available. - * Note that the monitor itself can be on a pty, so we still need to try the - * log output method. */ - paths = virHashCreate(0); - if (paths == NULL) { - virReportOOMError(); - goto cleanup; - } - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - qemuDomainObjPrivatePtr priv = vm->privateData; - ret = qemuMonitorGetPtyPaths(priv->mon, paths); - qemuDomainObjExitMonitorWithDriver(driver, vm); - - VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret); - if (ret == 0) { - ret = qemudFindCharDevicePTYsMonitor(vm, paths); - } - -cleanup: - if (paths) { - virHashFree(paths, qemudFreePtyPath); - } - - if (kill(vm->pid, 0) == -1 && errno == ESRCH) { - /* VM is dead, any other error raised in the interim is probably - * not as important as the qemu cmdline output */ - qemuReadLogFD(logfd, buf, sizeof(buf), strlen(buf)); - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("process exited while connecting to monitor: %s"), - buf); - ret = -1; - } - -closelog: - if (VIR_CLOSE(logfd) < 0) { - char ebuf[4096]; - VIR_WARN("Unable to close logfile: %s", - virStrerror(errno, ebuf, sizeof ebuf)); - } - - return ret; -} - -static int -qemuDetectVcpuPIDs(struct qemud_driver *driver, - virDomainObjPtr vm) { - pid_t *cpupids = NULL; - int ncpupids; - qemuDomainObjPrivatePtr priv = vm->privateData; - - if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) { - priv->nvcpupids = 1; - if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) { - virReportOOMError(); - return -1; - } - priv->vcpupids[0] = vm->pid; - return 0; - } - - /* What follows is now all KVM specific */ - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) { - qemuDomainObjExitMonitorWithDriver(driver, vm); - return -1; - } - qemuDomainObjExitMonitorWithDriver(driver, vm); - - /* Treat failure to get VCPU<->PID mapping as non-fatal */ - if (ncpupids == 0) - return 0; - - if (ncpupids != vm->def->vcpus) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("got wrong number of vCPU pids from QEMU monitor. " - "got %d, wanted %d"), - ncpupids, vm->def->vcpus); - VIR_FREE(cpupids); - return -1; - } - - priv->nvcpupids = ncpupids; - priv->vcpupids = cpupids; - return 0; -} - -/* - * To be run between fork/exec of QEMU only - */ -static int -qemudInitCpuAffinity(virDomainObjPtr vm) -{ - int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN; - virNodeInfo nodeinfo; - unsigned char *cpumap; - int cpumaplen; - - DEBUG0("Setting CPU affinity"); - - if (nodeGetInfo(NULL, &nodeinfo) < 0) - return -1; - - /* setaffinity fails if you set bits for CPUs which - * aren't present, so we have to limit ourselves */ - hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo); - if (maxcpu > hostcpus) - maxcpu = hostcpus; - - cpumaplen = VIR_CPU_MAPLEN(maxcpu); - if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) { - virReportOOMError(); - return -1; - } - - if (vm->def->cpumask) { - /* XXX why don't we keep 'cpumask' in the libvirt cpumap - * format to start with ?!?! */ - for (i = 0 ; i < maxcpu && i < vm->def->cpumasklen ; i++) - if (vm->def->cpumask[i]) - VIR_USE_CPU(cpumap, i); - } else { - /* You may think this is redundant, but we can't assume libvirtd - * itself is running on all pCPUs, so we need to explicitly set - * the spawned QEMU instance to all pCPUs if no map is given in - * its config file */ - for (i = 0 ; i < maxcpu ; i++) - VIR_USE_CPU(cpumap, i); - } - - /* We are pressuming we are running between fork/exec of QEMU - * so use '0' to indicate our own process ID. No threads are - * running at this point - */ - if (virProcessInfoSetAffinity(0, /* Self */ - cpumap, cpumaplen, maxcpu) < 0) { - VIR_FREE(cpumap); - return -1; - } - VIR_FREE(cpumap); - - return 0; -} - - -static int -qemuInitPasswords(virConnectPtr conn, - struct qemud_driver *driver, - virDomainObjPtr vm, - unsigned long long qemuCmdFlags) { - int ret = 0; - qemuDomainObjPrivatePtr priv = vm->privateData; - - if (vm->def->ngraphics == 1) { - if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) { - ret = qemuDomainChangeGraphicsPasswords(driver, vm, - VIR_DOMAIN_GRAPHICS_TYPE_VNC, - &vm->def->graphics[0]->data.vnc.auth, - driver->vncPassword); - } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) { - ret = qemuDomainChangeGraphicsPasswords(driver, vm, - VIR_DOMAIN_GRAPHICS_TYPE_SPICE, - &vm->def->graphics[0]->data.spice.auth, - driver->spicePassword); - } - } - - if (ret < 0) - goto cleanup; - - if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) { - int i; - - for (i = 0 ; i < vm->def->ndisks ; i++) { - char *secret; - size_t secretLen; - - if (!vm->def->disks[i]->encryption || - !vm->def->disks[i]->src) - continue; - - if (getVolumeQcowPassphrase(conn, - vm->def->disks[i], - &secret, &secretLen) < 0) - goto cleanup; - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - ret = qemuMonitorSetDrivePassphrase(priv->mon, - vm->def->disks[i]->info.alias, - secret); - VIR_FREE(secret); - qemuDomainObjExitMonitorWithDriver(driver, vm); - if (ret < 0) - goto cleanup; - } - } - -cleanup: - return ret; -} - - -#define QEMU_PCI_VENDOR_INTEL 0x8086 -#define QEMU_PCI_VENDOR_LSI_LOGIC 0x1000 -#define QEMU_PCI_VENDOR_REDHAT 0x1af4 -#define QEMU_PCI_VENDOR_CIRRUS 0x1013 -#define QEMU_PCI_VENDOR_REALTEK 0x10ec -#define QEMU_PCI_VENDOR_AMD 0x1022 -#define QEMU_PCI_VENDOR_ENSONIQ 0x1274 -#define QEMU_PCI_VENDOR_VMWARE 0x15ad -#define QEMU_PCI_VENDOR_QEMU 0x1234 - -#define QEMU_PCI_PRODUCT_DISK_VIRTIO 0x1001 - -#define QEMU_PCI_PRODUCT_BALLOON_VIRTIO 0x1002 - -#define QEMU_PCI_PRODUCT_NIC_NE2K 0x8029 -#define QEMU_PCI_PRODUCT_NIC_PCNET 0x2000 -#define QEMU_PCI_PRODUCT_NIC_RTL8139 0x8139 -#define QEMU_PCI_PRODUCT_NIC_E1000 0x100E -#define QEMU_PCI_PRODUCT_NIC_VIRTIO 0x1000 - -#define QEMU_PCI_PRODUCT_VGA_CIRRUS 0x00b8 -#define QEMU_PCI_PRODUCT_VGA_VMWARE 0x0405 -#define QEMU_PCI_PRODUCT_VGA_STDVGA 0x1111 - -#define QEMU_PCI_PRODUCT_AUDIO_AC97 0x2415 -#define QEMU_PCI_PRODUCT_AUDIO_ES1370 0x5000 - -#define QEMU_PCI_PRODUCT_CONTROLLER_PIIX 0x7010 -#define QEMU_PCI_PRODUCT_CONTROLLER_LSI 0x0012 - -#define QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB 0x25ab - -static int -qemuAssignNextPCIAddress(virDomainDeviceInfo *info, - int vendor, - int product, - qemuMonitorPCIAddress *addrs, - int naddrs) -{ - int found = 0; - int i; - - VIR_DEBUG("Look for %x:%x out of %d", vendor, product, naddrs); - - for (i = 0 ; (i < naddrs) && !found; i++) { - VIR_DEBUG("Maybe %x:%x", addrs[i].vendor, addrs[i].product); - if (addrs[i].vendor == vendor && - addrs[i].product == product) { - VIR_DEBUG("Match %d", i); - found = 1; - break; - } - } - if (!found) { - return -1; - } - - /* Blank it out so this device isn't matched again */ - addrs[i].vendor = 0; - addrs[i].product = 0; - - if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE) - info->type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI; - - if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) { - info->addr.pci.domain = addrs[i].addr.domain; - info->addr.pci.bus = addrs[i].addr.bus; - info->addr.pci.slot = addrs[i].addr.slot; - info->addr.pci.function = addrs[i].addr.function; - } - - return 0; -} - -static int -qemuGetPCIDiskVendorProduct(virDomainDiskDefPtr def, - unsigned *vendor, - unsigned *product) -{ - switch (def->bus) { - case VIR_DOMAIN_DISK_BUS_VIRTIO: - *vendor = QEMU_PCI_VENDOR_REDHAT; - *product = QEMU_PCI_PRODUCT_DISK_VIRTIO; - break; - - default: - return -1; - } - - return 0; -} - -static int -qemuGetPCINetVendorProduct(virDomainNetDefPtr def, - unsigned *vendor, - unsigned *product) -{ - if (!def->model) - return -1; - - if (STREQ(def->model, "ne2k_pci")) { - *vendor = QEMU_PCI_VENDOR_REALTEK; - *product = QEMU_PCI_PRODUCT_NIC_NE2K; - } else if (STREQ(def->model, "pcnet")) { - *vendor = QEMU_PCI_VENDOR_AMD; - *product = QEMU_PCI_PRODUCT_NIC_PCNET; - } else if (STREQ(def->model, "rtl8139")) { - *vendor = QEMU_PCI_VENDOR_REALTEK; - *product = QEMU_PCI_PRODUCT_NIC_RTL8139; - } else if (STREQ(def->model, "e1000")) { - *vendor = QEMU_PCI_VENDOR_INTEL; - *product = QEMU_PCI_PRODUCT_NIC_E1000; - } else if (STREQ(def->model, "virtio")) { - *vendor = QEMU_PCI_VENDOR_REDHAT; - *product = QEMU_PCI_PRODUCT_NIC_VIRTIO; - } else { - VIR_INFO("Unexpected NIC model %s, cannot get PCI address", - def->model); - return -1; - } - return 0; -} - -static int -qemuGetPCIControllerVendorProduct(virDomainControllerDefPtr def, - unsigned *vendor, - unsigned *product) -{ - switch (def->type) { - case VIR_DOMAIN_CONTROLLER_TYPE_SCSI: - *vendor = QEMU_PCI_VENDOR_LSI_LOGIC; - *product = QEMU_PCI_PRODUCT_CONTROLLER_LSI; - break; - - case VIR_DOMAIN_CONTROLLER_TYPE_FDC: - /* XXX we could put in the ISA bridge address, but - that's not technically the FDC's address */ - return -1; - - case VIR_DOMAIN_CONTROLLER_TYPE_IDE: - *vendor = QEMU_PCI_VENDOR_INTEL; - *product = QEMU_PCI_PRODUCT_CONTROLLER_PIIX; - break; - - default: - VIR_INFO("Unexpected controller type %s, cannot get PCI address", - virDomainControllerTypeToString(def->type)); - return -1; - } - - return 0; -} - -static int -qemuGetPCIVideoVendorProduct(virDomainVideoDefPtr def, - unsigned *vendor, - unsigned *product) -{ - switch (def->type) { - case VIR_DOMAIN_VIDEO_TYPE_CIRRUS: - *vendor = QEMU_PCI_VENDOR_CIRRUS; - *product = QEMU_PCI_PRODUCT_VGA_CIRRUS; - break; - - case VIR_DOMAIN_VIDEO_TYPE_VGA: - *vendor = QEMU_PCI_VENDOR_QEMU; - *product = QEMU_PCI_PRODUCT_VGA_STDVGA; - break; - - case VIR_DOMAIN_VIDEO_TYPE_VMVGA: - *vendor = QEMU_PCI_VENDOR_VMWARE; - *product = QEMU_PCI_PRODUCT_VGA_VMWARE; - break; - - default: - return -1; - } - return 0; -} - -static int -qemuGetPCISoundVendorProduct(virDomainSoundDefPtr def, - unsigned *vendor, - unsigned *product) -{ - switch (def->model) { - case VIR_DOMAIN_SOUND_MODEL_ES1370: - *vendor = QEMU_PCI_VENDOR_ENSONIQ; - *product = QEMU_PCI_PRODUCT_AUDIO_ES1370; - break; - - case VIR_DOMAIN_SOUND_MODEL_AC97: - *vendor = QEMU_PCI_VENDOR_INTEL; - *product = QEMU_PCI_PRODUCT_AUDIO_AC97; - break; - - default: - return -1; - } - - return 0; -} - -static int -qemuGetPCIWatchdogVendorProduct(virDomainWatchdogDefPtr def, - unsigned *vendor, - unsigned *product) -{ - switch (def->model) { - case VIR_DOMAIN_WATCHDOG_MODEL_I6300ESB: - *vendor = QEMU_PCI_VENDOR_INTEL; - *product = QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB; - break; - - default: - return -1; - } - - return 0; -} - - -static int -qemuGetPCIMemballoonVendorProduct(virDomainMemballoonDefPtr def, - unsigned *vendor, - unsigned *product) -{ - switch (def->model) { - case VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO: - *vendor = QEMU_PCI_VENDOR_REDHAT; - *product = QEMU_PCI_PRODUCT_BALLOON_VIRTIO; - break; - - default: - return -1; - } - - return 0; -} - - -/* - * This entire method assumes that PCI devices in 'info pci' - * match ordering of devices specified on the command line - * wrt to devices of matching vendor+product - * - * XXXX this might not be a valid assumption if we assign - * some static addrs on CLI. Have to check that... - */ -static int -qemuDetectPCIAddresses(virDomainObjPtr vm, - qemuMonitorPCIAddress *addrs, - int naddrs) -{ - unsigned int vendor = 0, product = 0; - int i; - - /* XXX should all these vendor/product IDs be kept in the - * actual device data structure instead ? - */ - - for (i = 0 ; i < vm->def->ndisks ; i++) { - if (qemuGetPCIDiskVendorProduct(vm->def->disks[i], &vendor, &product) < 0) - continue; - - if (qemuAssignNextPCIAddress(&(vm->def->disks[i]->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for VirtIO disk %s"), - vm->def->disks[i]->dst); - return -1; - } - } - - for (i = 0 ; i < vm->def->nnets ; i++) { - if (qemuGetPCINetVendorProduct(vm->def->nets[i], &vendor, &product) < 0) - continue; - - if (qemuAssignNextPCIAddress(&(vm->def->nets[i]->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for %s NIC"), - vm->def->nets[i]->model); - return -1; - } - } - - for (i = 0 ; i < vm->def->ncontrollers ; i++) { - if (qemuGetPCIControllerVendorProduct(vm->def->controllers[i], &vendor, &product) < 0) - continue; - - if (qemuAssignNextPCIAddress(&(vm->def->controllers[i]->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for controller %s"), - virDomainControllerTypeToString(vm->def->controllers[i]->type)); - return -1; - } - } - - for (i = 0 ; i < vm->def->nvideos ; i++) { - if (qemuGetPCIVideoVendorProduct(vm->def->videos[i], &vendor, &product) < 0) - continue; - - if (qemuAssignNextPCIAddress(&(vm->def->videos[i]->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for video adapter %s"), - virDomainVideoTypeToString(vm->def->videos[i]->type)); - return -1; - } - } - - for (i = 0 ; i < vm->def->nsounds ; i++) { - if (qemuGetPCISoundVendorProduct(vm->def->sounds[i], &vendor, &product) < 0) - continue; - - if (qemuAssignNextPCIAddress(&(vm->def->sounds[i]->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for sound adapter %s"), - virDomainSoundModelTypeToString(vm->def->sounds[i]->model)); - return -1; - } - } - - - if (vm->def->watchdog && - qemuGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) { - if (qemuAssignNextPCIAddress(&(vm->def->watchdog->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for watchdog %s"), - virDomainWatchdogModelTypeToString(vm->def->watchdog->model)); - return -1; - } - } - - if (vm->def->memballoon && - qemuGetPCIMemballoonVendorProduct(vm->def->memballoon, &vendor, &product) == 0) { - if (qemuAssignNextPCIAddress(&(vm->def->memballoon->info), - vendor, product, - addrs, naddrs) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("cannot find PCI address for balloon %s"), - virDomainMemballoonModelTypeToString(vm->def->memballoon->model)); - return -1; - } - } - - /* XXX console (virtio) */ - - - /* ... and now things we don't have in our xml */ - - /* XXX USB controller ? */ - - /* XXX what about other PCI devices (ie bridges) */ - - return 0; -} - -static int -qemuInitPCIAddresses(struct qemud_driver *driver, - virDomainObjPtr vm) -{ - qemuDomainObjPrivatePtr priv = vm->privateData; - int naddrs; - int ret; - qemuMonitorPCIAddress *addrs = NULL; - - qemuDomainObjEnterMonitorWithDriver(driver, vm); - naddrs = qemuMonitorGetAllPCIAddresses(priv->mon, - &addrs); - qemuDomainObjExitMonitorWithDriver(driver, vm); - - ret = qemuDetectPCIAddresses(vm, addrs, naddrs); - - VIR_FREE(addrs); - - return ret; -} - - -static int qemudNextFreePort(struct qemud_driver *driver, - int startPort) { - int i; - - for (i = startPort ; i < QEMU_VNC_PORT_MAX; i++) { - int fd; - int reuse = 1; - struct sockaddr_in addr; - bool used = false; - - if (virBitmapGetBit(driver->reservedVNCPorts, - i - QEMU_VNC_PORT_MIN, &used) < 0) - VIR_DEBUG("virBitmapGetBit failed on bit %d", i - QEMU_VNC_PORT_MIN); - - if (used) - continue; - - addr.sin_family = AF_INET; - addr.sin_port = htons(i); - addr.sin_addr.s_addr = htonl(INADDR_ANY); - fd = socket(PF_INET, SOCK_STREAM, 0); - if (fd < 0) - return -1; - - if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (void*)&reuse, sizeof(reuse)) < 0) { - VIR_FORCE_CLOSE(fd); - break; - } + goto error; - if (bind(fd, (struct sockaddr*)&addr, sizeof(addr)) == 0) { - /* Not in use, lets grab it */ - VIR_FORCE_CLOSE(fd); - /* Add port to bitmap of reserved ports */ - if (virBitmapSetBit(driver->reservedVNCPorts, - i - QEMU_VNC_PORT_MIN) < 0) { - VIR_DEBUG("virBitmapSetBit failed on bit %d", - i - QEMU_VNC_PORT_MIN); - } - return i; + if (privileged) { + if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) { + virReportSystemError(errno, + _("unable to set ownership of '%s' to user %d:%d"), + qemu_driver->libDir, qemu_driver->user, qemu_driver->group); + goto error; } - VIR_FORCE_CLOSE(fd); - - if (errno == EADDRINUSE) { - /* In use, try next */ - continue; + if (chown(qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group) < 0) { + virReportSystemError(errno, + _("unable to set ownership of '%s' to %d:%d"), + qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group); + goto error; } - /* Some other bad failure, get out.. */ - break; - } - return -1; -} - - -static void -qemuReturnPort(struct qemud_driver *driver, - int port) -{ - if (port < QEMU_VNC_PORT_MIN) - return; - - if (virBitmapClearBit(driver->reservedVNCPorts, - port - QEMU_VNC_PORT_MIN) < 0) - VIR_DEBUG("Could not mark port %d as unused", port); -} - - -static int -qemuAssignPCIAddresses(virDomainDefPtr def) -{ - int ret = -1; - unsigned long long qemuCmdFlags = 0; - qemuDomainPCIAddressSetPtr addrs = NULL; - - if (qemuCapsExtractVersionInfo(def->emulator, - NULL, - &qemuCmdFlags) < 0) - goto cleanup; - - if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) { - if (!(addrs = qemuDomainPCIAddressSetCreate(def))) - goto cleanup; - - if (qemuAssignDevicePCISlots(def, addrs) < 0) - goto cleanup; - } - - ret = 0; - -cleanup: - qemuDomainPCIAddressSetFree(addrs); - - return ret; -} - - -static int -qemuPrepareChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED, - virDomainChrDefPtr dev, - void *opaque ATTRIBUTE_UNUSED) -{ - int fd; - if (dev->source.type != VIR_DOMAIN_CHR_TYPE_FILE) - return 0; - - if ((fd = open(dev->source.data.file.path, - O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) { - virReportSystemError(errno, - _("Unable to pre-create chardev file '%s'"), - dev->source.data.file.path); - return -1; - } - - VIR_FORCE_CLOSE(fd); - - return 0; -} - - -struct qemudHookData { - virConnectPtr conn; - virDomainObjPtr vm; - struct qemud_driver *driver; -}; - -static int qemudSecurityHook(void *data) { - struct qemudHookData *h = data; - - /* This must take place before exec(), so that all QEMU - * memory allocation is on the correct NUMA node - */ - if (qemuAddToCgroup(h->driver, h->vm->def) < 0) - return -1; - - /* This must be done after cgroup placement to avoid resetting CPU - * affinity */ - if (qemudInitCpuAffinity(h->vm) < 0) - return -1; - - if (virSecurityManagerSetProcessLabel(h->driver->securityManager, h->vm) < 0) - return -1; - - return 0; -} - -static int -qemuPrepareMonitorChr(struct qemud_driver *driver, - virDomainChrSourceDefPtr monConfig, - const char *vm) -{ - monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX; - monConfig->data.nix.listen = true; - - if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor", - driver->libDir, vm) < 0) { - virReportOOMError(); - return -1; - } - - return 0; -} - -static int qemuDomainSnapshotSetCurrentActive(virDomainObjPtr vm, - char *snapshotDir); -static int qemuDomainSnapshotSetCurrentInactive(virDomainObjPtr vm, - char *snapshotDir); - - -#define START_POSTFIX ": starting up\n" -#define SHUTDOWN_POSTFIX ": shutting down\n" - -static int qemudStartVMDaemon(virConnectPtr conn, - struct qemud_driver *driver, - virDomainObjPtr vm, - const char *migrateFrom, - bool start_paused, - int stdin_fd, - const char *stdin_path, - enum virVMOperationType vmop) { - int ret; - unsigned long long qemuCmdFlags; - off_t pos = -1; - char ebuf[1024]; - char *pidfile = NULL; - int logfile = -1; - char *timestamp; - qemuDomainObjPrivatePtr priv = vm->privateData; - virCommandPtr cmd = NULL; - - struct qemudHookData hookData; - hookData.conn = conn; - hookData.vm = vm; - hookData.driver = driver; - - DEBUG0("Beginning VM startup process"); - - if (virDomainObjIsActive(vm)) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("VM is already active")); - return -1; - } - - /* Do this upfront, so any part of the startup process can add - * runtime state to vm->def that won't be persisted. This let's us - * report implicit runtime defaults in the XML, like vnc listen/socket - */ - DEBUG0("Setting current domain def as transient"); - if (virDomainObjSetDefTransient(driver->caps, vm, true) < 0) - goto cleanup; - - /* Must be run before security labelling */ - DEBUG0("Preparing host devices"); - if (qemuPrepareHostDevices(driver, vm->def) < 0) - goto cleanup; - - DEBUG0("Preparing chr devices"); - if (virDomainChrDefForeach(vm->def, - true, - qemuPrepareChardevDevice, - NULL) < 0) - goto cleanup; - - /* If you are using a SecurityDriver with dynamic labelling, - then generate a security label for isolation */ - DEBUG0("Generating domain security label (if required)"); - if (virSecurityManagerGenLabel(driver->securityManager, vm) < 0) { - qemuDomainSecurityLabelAudit(vm, false); - goto cleanup; - } - qemuDomainSecurityLabelAudit(vm, true); - - DEBUG0("Generating setting domain security labels (if required)"); - if (virSecurityManagerSetAllLabel(driver->securityManager, - vm, stdin_path) < 0) - goto cleanup; - - if (stdin_fd != -1) { - /* if there's an fd to migrate from, and it's a pipe, put the - * proper security label on it - */ - struct stat stdin_sb; - - DEBUG0("setting security label on pipe used for migration"); - - if (fstat(stdin_fd, &stdin_sb) < 0) { + if (chown(qemu_driver->saveDir, qemu_driver->user, qemu_driver->group) < 0) { virReportSystemError(errno, - _("cannot stat fd %d"), stdin_fd); - goto cleanup; + _("unable to set ownership of '%s' to %d:%d"), + qemu_driver->saveDir, qemu_driver->user, qemu_driver->group); + goto error; } - if (S_ISFIFO(stdin_sb.st_mode) && - virSecurityManagerSetFDLabel(driver->securityManager, vm, stdin_fd) < 0) - goto cleanup; - } - - /* Ensure no historical cgroup for this VM is lying around bogus - * settings */ - DEBUG0("Ensuring no historical cgroup is lying around"); - qemuRemoveCgroup(driver, vm, 1); - - if (vm->def->ngraphics == 1) { - if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC && - !vm->def->graphics[0]->data.vnc.socket && - vm->def->graphics[0]->data.vnc.autoport) { - int port = qemudNextFreePort(driver, QEMU_VNC_PORT_MIN); - if (port < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("Unable to find an unused VNC port")); - goto cleanup; - } - vm->def->graphics[0]->data.vnc.port = port; - } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE && - vm->def->graphics[0]->data.spice.autoport) { - int port = qemudNextFreePort(driver, QEMU_VNC_PORT_MIN); - int tlsPort = -1; - if (port < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("Unable to find an unused SPICE port")); - goto cleanup; - } - - if (driver->spiceTLS) { - tlsPort = qemudNextFreePort(driver, port + 1); - if (tlsPort < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("Unable to find an unused SPICE TLS port")); - qemuReturnPort(driver, port); - goto cleanup; - } - } - - vm->def->graphics[0]->data.spice.port = port; - vm->def->graphics[0]->data.spice.tlsPort = tlsPort; + if (chown(qemu_driver->snapshotDir, qemu_driver->user, qemu_driver->group) < 0) { + virReportSystemError(errno, + _("unable to set ownership of '%s' to %d:%d"), + qemu_driver->snapshotDir, qemu_driver->user, qemu_driver->group); + goto error; } } - if (virFileMakePath(driver->logDir) != 0) { - virReportSystemError(errno, - _("cannot create log directory %s"), - driver->logDir); - goto cleanup; - } - - DEBUG0("Creating domain log file"); - if ((logfile = qemudLogFD(driver, vm->def->name, false)) < 0) - goto cleanup; - - DEBUG0("Determining emulator version"); - if (qemuCapsExtractVersionInfo(vm->def->emulator, - NULL, - &qemuCmdFlags) < 0) - goto cleanup; - - DEBUG0("Setting up domain cgroup (if required)"); - if (qemuSetupCgroup(driver, vm) < 0) - goto cleanup; - - if (VIR_ALLOC(priv->monConfig) < 0) { - virReportOOMError(); - goto cleanup; - } - - DEBUG0("Preparing monitor state"); - if (qemuPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0) - goto cleanup; - -#if HAVE_YAJL - if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON) - priv->monJSON = 1; - else -#endif - priv->monJSON = 0; - - priv->monitor_warned = 0; - priv->gotShutdown = false; - - if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) { - virReportSystemError(ret, - _("Cannot remove stale PID file for %s"), - vm->def->name); - goto cleanup; - } - - if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) { - virReportSystemError(errno, - "%s", _("Failed to build pidfile path.")); - goto cleanup; - } - - /* - * Normally PCI addresses are assigned in the virDomainCreate - * or virDomainDefine methods. We might still need to assign - * some here to cope with the question of upgrades. Regardless - * we also need to populate the PCi address set cache for later - * use in hotplug + /* If hugetlbfs is present, then we need to create a sub-directory within + * it, since we can't assume the root mount point has permissions that + * will let our spawned QEMU instances use it. + * + * NB the check for '/', since user may config "" to disable hugepages + * even when mounted */ - if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) { - DEBUG0("Assigning domain PCI addresses"); - /* Populate cache with current addresses */ - if (priv->pciaddrs) { - qemuDomainPCIAddressSetFree(priv->pciaddrs); - priv->pciaddrs = NULL; - } - if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def))) - goto cleanup; - - - /* Assign any remaining addresses */ - if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0) - goto cleanup; - - priv->persistentAddrs = 1; - } else { - priv->persistentAddrs = 0; - } - - DEBUG0("Building emulator command line"); - vm->def->id = driver->nextvmid++; - if (!(cmd = qemuBuildCommandLine(conn, driver, vm->def, priv->monConfig, - priv->monJSON != 0, qemuCmdFlags, - migrateFrom, stdin_fd, - vm->current_snapshot, vmop))) - goto cleanup; - - if (qemuDomainSnapshotSetCurrentInactive(vm, driver->snapshotDir) < 0) - goto cleanup; - - /* now that we know it is about to start call the hook if present */ - if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) { - char *xml = virDomainDefFormat(vm->def, 0); - int hookret; - - hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name, - VIR_HOOK_QEMU_OP_START, VIR_HOOK_SUBOP_BEGIN, NULL, xml); - VIR_FREE(xml); - - /* - * If the script raised an error abort the launch - */ - if (hookret < 0) - goto cleanup; - } + if (qemu_driver->hugetlbfs_mount && + qemu_driver->hugetlbfs_mount[0] == '/') { + char *mempath = NULL; + if (virAsprintf(&mempath, "%s/libvirt/qemu", qemu_driver->hugetlbfs_mount) < 0) + goto out_of_memory; - if ((timestamp = virTimestamp()) == NULL) { - virReportOOMError(); - goto cleanup; - } else { - if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 || - safewrite(logfile, START_POSTFIX, strlen(START_POSTFIX)) < 0) { - VIR_WARN("Unable to write timestamp to logfile: %s", - virStrerror(errno, ebuf, sizeof ebuf)); + if ((rc = virFileMakePath(mempath)) != 0) { + virReportSystemError(rc, + _("unable to create hugepage path %s"), mempath); + VIR_FREE(mempath); + goto error; } - - VIR_FREE(timestamp); - } - - virCommandWriteArgLog(cmd, logfile); - - if ((pos = lseek(logfile, 0, SEEK_END)) < 0) - VIR_WARN("Unable to seek to end of logfile: %s", - virStrerror(errno, ebuf, sizeof ebuf)); - - VIR_DEBUG("Clear emulator capabilities: %d", - driver->clearEmulatorCapabilities); - if (driver->clearEmulatorCapabilities) - virCommandClearCaps(cmd); - - virCommandSetPreExecHook(cmd, qemudSecurityHook, &hookData); - - virCommandSetOutputFD(cmd, &logfile); - virCommandSetErrorFD(cmd, &logfile); - virCommandNonblockingFDs(cmd); - virCommandSetPidFile(cmd, pidfile); - virCommandDaemonize(cmd); - - ret = virCommandRun(cmd, NULL); - VIR_FREE(pidfile); - - /* wait for qemu process to to show up */ - if (ret == 0) { - if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, - _("Domain %s didn't show up"), vm->def->name); - ret = -1; + if (qemu_driver->privileged && + chown(mempath, qemu_driver->user, qemu_driver->group) < 0) { + virReportSystemError(errno, + _("unable to set ownership on %s to %d:%d"), + mempath, qemu_driver->user, qemu_driver->group); + VIR_FREE(mempath); + goto error; } -#if 0 - } else if (ret == -2) { - /* - * XXX this is bogus. It isn't safe to set vm->pid = child - * because the child no longer exists. - */ - /* The virExec process that launches the daemon failed. Pending on - * when it failed (we can't determine for sure), there may be - * extra info in the domain log (if the hook failed for example). - * - * Pretend like things succeeded, and let 'WaitForMonitor' report - * the log contents for us. - */ - vm->pid = child; - ret = 0; -#endif + qemu_driver->hugepage_path = mempath; } - if (migrateFrom) - start_paused = true; - vm->state = start_paused ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING; - - if (ret == -1) /* The VM failed to start; tear filters before taps */ - virDomainConfVMNWFilterTeardown(vm); - - if (ret == -1) /* The VM failed to start */ - goto cleanup; + /* Get all the running persistent or transient configs first */ + if (virDomainLoadAllConfigs(qemu_driver->caps, + &qemu_driver->domains, + qemu_driver->stateDir, + NULL, + 1, NULL, NULL) < 0) + goto error; - DEBUG0("Waiting for monitor to show up"); - if (qemudWaitForMonitor(driver, vm, pos) < 0) - goto cleanup; + conn = virConnectOpen(qemu_driver->privileged ? + "qemu:///system" : + "qemu:///session"); - DEBUG0("Detecting VCPU PIDs"); - if (qemuDetectVcpuPIDs(driver, vm) < 0) - goto cleanup; + qemuProcessReconnectAll(conn, qemu_driver); - DEBUG0("Setting any required VM passwords"); - if (qemuInitPasswords(conn, driver, vm, qemuCmdFlags) < 0) - goto cleanup; + /* Then inactive persistent configs */ + if (virDomainLoadAllConfigs(qemu_driver->caps, + &qemu_driver->domains, + qemu_driver->configDir, + qemu_driver->autostartDir, + 0, NULL, NULL) < 0) + goto error; - /* If we have -device, then addresses are assigned explicitly. - * If not, then we have to detect dynamic ones here */ - if (!(qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) { - DEBUG0("Determining domain device PCI addresses"); - if (qemuInitPCIAddresses(driver, vm) < 0) - goto cleanup; - } - DEBUG0("Setting initial memory amount"); - qemuDomainObjEnterMonitorWithDriver(driver, vm); - if (qemuMonitorSetBalloon(priv->mon, vm->def->mem.cur_balloon) < 0) { - qemuDomainObjExitMonitorWithDriver(driver, vm); - goto cleanup; - } - qemuDomainObjExitMonitorWithDriver(driver, vm); + virHashForEach(qemu_driver->domains.objs, qemuDomainSnapshotLoad, + qemu_driver->snapshotDir); - if (!start_paused) { - DEBUG0("Starting domain CPUs"); - /* Allow the CPUS to start executing */ - if (doStartCPUs(driver, vm, conn) < 0) { - if (virGetLastError() == NULL) - qemuReportError(VIR_ERR_INTERNAL_ERROR, - "%s", _("resume operation failed")); - goto cleanup; - } - } + qemuDriverUnlock(qemu_driver); + qemuAutostartDomains(qemu_driver); - DEBUG0("Writing domain status to disk"); - if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) - goto cleanup; + qemu_driver->workerPool = virThreadPoolNew(0, 1, processWatchdogEvent, qemu_driver); + if (!qemu_driver->workerPool) + goto error; - virCommandFree(cmd); - VIR_FORCE_CLOSE(logfile); + if (conn) + virConnectClose(conn); return 0; -cleanup: - /* We jump here if we failed to start the VM for any reason, or - * if we failed to initialize the now running VM. kill it off and - * pretend we never started it */ - virCommandFree(cmd); - VIR_FORCE_CLOSE(logfile); - qemudShutdownVMDaemon(driver, vm, 0); - +out_of_memory: + virReportOOMError(); +error: + if (qemu_driver) + qemuDriverUnlock(qemu_driver); + if (conn) + virConnectClose(conn); + VIR_FREE(base); + VIR_FREE(driverConf); + qemudShutdown(); return -1; } -static void qemudShutdownVMDaemon(struct qemud_driver *driver, - virDomainObjPtr vm, - int migrated) { - int ret; - int retries = 0; - qemuDomainObjPrivatePtr priv = vm->privateData; - virErrorPtr orig_err; - virDomainDefPtr def; - int i; - int logfile = -1; - char *timestamp; - char ebuf[1024]; - - VIR_DEBUG("Shutting down VM '%s' pid=%d migrated=%d", - vm->def->name, vm->pid, migrated); +static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque) +{ + struct qemud_driver *driver = opaque; - if ((logfile = qemudLogFD(driver, vm->def->name, true)) < 0) { - /* To not break the normal domain shutdown process, skip the - * timestamp log writing if failed on opening log file. */ - VIR_WARN("Unable to open logfile: %s", - virStrerror(errno, ebuf, sizeof ebuf)); - } else { - if ((timestamp = virTimestamp()) == NULL) { - virReportOOMError(); - } else { - if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 || - safewrite(logfile, SHUTDOWN_POSTFIX, - strlen(SHUTDOWN_POSTFIX)) < 0) { - VIR_WARN("Unable to write timestamp to logfile: %s", - virStrerror(errno, ebuf, sizeof ebuf)); - } + if (newVM) { + virDomainEventPtr event = + virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_DEFINED, + VIR_DOMAIN_EVENT_DEFINED_ADDED); + if (event) + qemuDomainEventQueue(driver, event); + } +} - VIR_FREE(timestamp); - } +/** + * qemudReload: + * + * Function to restart the QEmu daemon, it will recheck the configuration + * files and update its state and the networking + */ +static int +qemudReload(void) { + if (!qemu_driver) + return 0; - if (VIR_CLOSE(logfile) < 0) - VIR_WARN("Unable to close logfile: %s", - virStrerror(errno, ebuf, sizeof ebuf)); - } + qemuDriverLock(qemu_driver); + virDomainLoadAllConfigs(qemu_driver->caps, + &qemu_driver->domains, + qemu_driver->configDir, + qemu_driver->autostartDir, + 0, qemudNotifyLoadDomain, qemu_driver); + qemuDriverUnlock(qemu_driver); - /* This method is routinely used in clean up paths. Disable error - * reporting so we don't squash a legit error. */ - orig_err = virSaveLastError(); + qemuAutostartDomains(qemu_driver); - virDomainConfVMNWFilterTeardown(vm); + return 0; +} - if (driver->macFilter) { - def = vm->def; - for (i = 0 ; i < def->nnets ; i++) { - virDomainNetDefPtr net = def->nets[i]; - if (net->ifname == NULL) - continue; - if ((errno = networkDisallowMacOnPort(driver, net->ifname, - net->mac))) { - virReportSystemError(errno, - _("failed to remove ebtables rule to allow MAC address on '%s'"), - net->ifname); - } - } - } +/** + * qemudActive: + * + * Checks if the QEmu daemon is active, i.e. has an active domain or + * an active network + * + * Returns 1 if active, 0 otherwise + */ +static int +qemudActive(void) { + int active = 0; - /* This will safely handle a non-running guest with pid=0 or pid=-1*/ - if (virKillProcess(vm->pid, 0) == 0 && - virKillProcess(vm->pid, SIGTERM) < 0) - virReportSystemError(errno, - _("Failed to send SIGTERM to %s (%d)"), - vm->def->name, vm->pid); + if (!qemu_driver) + return 0; - if (priv->mon) - qemuMonitorClose(priv->mon); + /* XXX having to iterate here is not great because it requires many locks */ + qemuDriverLock(qemu_driver); + active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1); + qemuDriverUnlock(qemu_driver); + return active; +} - if (priv->monConfig) { - if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX) - unlink(priv->monConfig->data.nix.path); - virDomainChrSourceDefFree(priv->monConfig); - priv->monConfig = NULL; - } +/** + * qemudShutdown: + * + * Shutdown the QEmu daemon, it will stop all active domains and networks + */ +static int +qemudShutdown(void) { + int i; - /* shut it off for sure */ - virKillProcess(vm->pid, SIGKILL); + if (!qemu_driver) + return -1; - /* now that we know it's stopped call the hook if present */ - if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) { - char *xml = virDomainDefFormat(vm->def, 0); + qemuDriverLock(qemu_driver); + pciDeviceListFree(qemu_driver->activePciHostdevs); + virCapabilitiesFree(qemu_driver->caps); - /* we can't stop the operation even if the script raised an error */ - virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name, - VIR_HOOK_QEMU_OP_STOPPED, VIR_HOOK_SUBOP_END, NULL, xml); - VIR_FREE(xml); - } + virDomainObjListDeinit(&qemu_driver->domains); + virBitmapFree(qemu_driver->reservedVNCPorts); - /* Reset Security Labels */ - virSecurityManagerRestoreAllLabel(driver->securityManager, - vm, migrated); - virSecurityManagerReleaseLabel(driver->securityManager, vm); + virSysinfoDefFree(qemu_driver->hostsysinfo); - /* Clear out dynamically assigned labels */ - if (vm->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC) { - VIR_FREE(vm->def->seclabel.model); - VIR_FREE(vm->def->seclabel.label); - VIR_FREE(vm->def->seclabel.imagelabel); - } + VIR_FREE(qemu_driver->configDir); + VIR_FREE(qemu_driver->autostartDir); + VIR_FREE(qemu_driver->logDir); + VIR_FREE(qemu_driver->stateDir); + VIR_FREE(qemu_driver->libDir); + VIR_FREE(qemu_driver->cacheDir); + VIR_FREE(qemu_driver->saveDir); + VIR_FREE(qemu_driver->snapshotDir); + VIR_FREE(qemu_driver->autoDumpPath); + VIR_FREE(qemu_driver->vncTLSx509certdir); + VIR_FREE(qemu_driver->vncListen); + VIR_FREE(qemu_driver->vncPassword); + VIR_FREE(qemu_driver->vncSASLdir); + VIR_FREE(qemu_driver->spiceTLSx509certdir); + VIR_FREE(qemu_driver->spiceListen); + VIR_FREE(qemu_driver->spicePassword); + VIR_FREE(qemu_driver->hugetlbfs_mount); + VIR_FREE(qemu_driver->hugepage_path); + VIR_FREE(qemu_driver->saveImageFormat); + VIR_FREE(qemu_driver->dumpImageFormat); - virDomainDefClearDeviceAliases(vm->def); - if (!priv->persistentAddrs) { - virDomainDefClearPCIAddresses(vm->def); - qemuDomainPCIAddressSetFree(priv->pciaddrs); - priv->pciaddrs = NULL; - } + virSecurityManagerFree(qemu_driver->securityManager); - qemuDomainReAttachHostDevices(driver, vm->def); + ebtablesContextFree(qemu_driver->ebtables); -#if WITH_MACVTAP - def = vm->def; - for (i = 0; i < def->nnets; i++) { - virDomainNetDefPtr net = def->nets[i]; - if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { - delMacvtap(net->ifname, net->mac, net->data.direct.linkdev, - &net->data.direct.virtPortProfile); - VIR_FREE(net->ifname); - } + if (qemu_driver->cgroupDeviceACL) { + for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++) + VIR_FREE(qemu_driver->cgroupDeviceACL[i]); + VIR_FREE(qemu_driver->cgroupDeviceACL); } -#endif -retry: - if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) { - if (ret == -EBUSY && (retries++ < 5)) { - usleep(200*1000); - goto retry; - } - VIR_WARN("Failed to remove cgroup for %s", - vm->def->name); - } + /* Free domain callback list */ + virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks); + virDomainEventQueueFree(qemu_driver->domainEventQueue); - qemudRemoveDomainStatus(driver, vm); + if (qemu_driver->domainEventTimer != -1) + virEventRemoveTimeout(qemu_driver->domainEventTimer); - /* Remove VNC port from port reservation bitmap, but only if it was - reserved by the driver (autoport=yes) - */ - if ((vm->def->ngraphics == 1) && - vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC && - vm->def->graphics[0]->data.vnc.autoport) { - qemuReturnPort(driver, vm->def->graphics[0]->data.vnc.port); - } - if ((vm->def->ngraphics == 1) && - vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE && - vm->def->graphics[0]->data.spice.autoport) { - qemuReturnPort(driver, vm->def->graphics[0]->data.spice.port); - qemuReturnPort(driver, vm->def->graphics[0]->data.spice.tlsPort); - } + if (qemu_driver->brctl) + brShutdown(qemu_driver->brctl); - vm->pid = -1; - vm->def->id = -1; - vm->state = VIR_DOMAIN_SHUTOFF; - VIR_FREE(priv->vcpupids); - priv->nvcpupids = 0; + virCgroupFree(&qemu_driver->cgroup); - if (vm->newDef) { - virDomainDefFree(vm->def); - vm->def = vm->newDef; - vm->def->id = -1; - vm->newDef = NULL; - } + qemuDriverUnlock(qemu_driver); + virMutexDestroy(&qemu_driver->lock); + virThreadPoolFree(qemu_driver->workerPool); + VIR_FREE(qemu_driver); - if (orig_err) { - virSetError(orig_err); - virFreeError(orig_err); - } + return 0; } + +static int qemuDomainSnapshotSetCurrentActive(virDomainObjPtr vm, + char *snapshotDir); +static int qemuDomainSnapshotSetCurrentInactive(virDomainObjPtr vm, + char *snapshotDir); + + static virDrvOpenStatus qemudOpen(virConnectPtr conn, virConnectAuthPtr auth ATTRIBUTE_UNUSED, int flags ATTRIBUTE_UNUSED) { @@ -3582,7 +1223,7 @@ static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml, if (qemudCanonicalizeMachine(driver, def) < 0) goto cleanup; - if (qemuAssignPCIAddresses(def) < 0) + if (qemuDomainAssignPCIAddresses(def) < 0) goto cleanup; if (!(vm = virDomainAssignDef(driver->caps, @@ -3595,9 +1236,9 @@ static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml, if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) goto cleanup; /* XXXX free the 'vm' we created ? */ - if (qemudStartVMDaemon(conn, driver, vm, NULL, - (flags & VIR_DOMAIN_START_PAUSED) != 0, - -1, NULL, VIR_VM_OP_CREATE) < 0) { + if (qemuProcessStart(conn, driver, vm, NULL, + (flags & VIR_DOMAIN_START_PAUSED) != 0, + -1, NULL, VIR_VM_OP_CREATE) < 0) { qemuDomainStartAudit(vm, "booted", false); if (qemuDomainObjEndJob(vm) > 0) virDomainRemoveInactive(&driver->domains, @@ -3672,7 +1313,7 @@ static int qemudDomainSuspend(virDomainPtr dom) { goto endjob; } if (vm->state != VIR_DOMAIN_PAUSED) { - if (doStopCPUs(driver, vm) < 0) { + if (qemuProcessStopCPUs(driver, vm) < 0) { goto endjob; } event = virDomainEventNewFromObj(vm, @@ -3725,7 +1366,7 @@ static int qemudDomainResume(virDomainPtr dom) { goto endjob; } if (vm->state == VIR_DOMAIN_PAUSED) { - if (doStartCPUs(driver, vm, dom->conn) < 0) { + if (qemuProcessStartCPUs(driver, vm, dom->conn) < 0) { if (virGetLastError() == NULL) qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", _("resume operation failed")); @@ -3820,7 +1461,7 @@ static int qemudDomainDestroy(virDomainPtr dom) { goto endjob; } - qemudShutdownVMDaemon(driver, vm, 0); + qemuProcessStop(driver, vm, 0); event = virDomainEventNewFromObj(vm, VIR_DOMAIN_EVENT_STOPPED, VIR_DOMAIN_EVENT_STOPPED_DESTROYED); @@ -4044,7 +1685,7 @@ qemuDomainMigrateOffline(struct qemud_driver *driver, { int ret; - ret = doStopCPUs(driver, vm); + ret = qemuProcessStopCPUs(driver, vm); if (ret == 0) { virDomainEventPtr event; @@ -4315,7 +1956,7 @@ static int qemudDomainSaveFlag(struct qemud_driver *driver, virDomainPtr dom, /* Pause */ if (vm->state == VIR_DOMAIN_RUNNING) { header.was_running = 1; - if (doStopCPUs(driver, vm) < 0) + if (qemuProcessStopCPUs(driver, vm) < 0) goto endjob; if (!virDomainObjIsActive(vm)) { @@ -4525,7 +2166,7 @@ static int qemudDomainSaveFlag(struct qemud_driver *driver, virDomainPtr dom, ret = 0; /* Shut it down */ - qemudShutdownVMDaemon(driver, vm, 0); + qemuProcessStop(driver, vm, 0); qemuDomainStopAudit(vm, "saved"); event = virDomainEventNewFromObj(vm, VIR_DOMAIN_EVENT_STOPPED, @@ -4541,7 +2182,7 @@ endjob: if (vm) { if (ret != 0) { if (header.was_running && virDomainObjIsActive(vm)) { - rc = doStartCPUs(driver, vm, dom->conn); + rc = qemuProcessStartCPUs(driver, vm, dom->conn); if (rc < 0) VIR_WARN0("Unable to resume guest CPUs after save failure"); } @@ -4895,7 +2536,7 @@ static int qemudDomainCoreDump(virDomainPtr dom, /* Pause domain for non-live dump */ if (!(flags & VIR_DUMP_LIVE) && vm->state == VIR_DOMAIN_RUNNING) { - if (doStopCPUs(driver, vm) < 0) + if (qemuProcessStopCPUs(driver, vm) < 0) goto endjob; paused = 1; @@ -4914,7 +2555,7 @@ static int qemudDomainCoreDump(virDomainPtr dom, endjob: if ((ret == 0) && (flags & VIR_DUMP_CRASH)) { - qemudShutdownVMDaemon(driver, vm, 0); + qemuProcessStop(driver, vm, 0); qemuDomainStopAudit(vm, "crashed"); event = virDomainEventNewFromObj(vm, VIR_DOMAIN_EVENT_STOPPED, @@ -4925,7 +2566,7 @@ endjob: will support synchronous operations so we always get here after the migration is complete. */ else if (resume && paused && virDomainObjIsActive(vm)) { - if (doStartCPUs(driver, vm, dom->conn) < 0) { + if (qemuProcessStartCPUs(driver, vm, dom->conn) < 0) { if (virGetLastError() == NULL) qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", _("resuming after dump failed")); @@ -4952,7 +2593,7 @@ cleanup: static void processWatchdogEvent(void *data, void *opaque) { int ret; - struct watchdogEvent *wdEvent = data; + struct qemuDomainWatchdogEvent *wdEvent = data; struct qemud_driver *driver = opaque; switch (wdEvent->action) { @@ -4986,7 +2627,7 @@ static void processWatchdogEvent(void *data, void *opaque) qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", _("Dump failed")); - ret = doStartCPUs(driver, wdEvent->vm, NULL); + ret = qemuProcessStartCPUs(driver, wdEvent->vm, NULL); if (ret < 0) qemuReportError(VIR_ERR_OPERATION_FAILED, @@ -5776,8 +3417,8 @@ qemudDomainSaveImageStartVM(virConnectPtr conn, } /* Set the migration source and start it up. */ - ret = qemudStartVMDaemon(conn, driver, vm, "stdio", true, fd, path, - VIR_VM_OP_RESTORE); + ret = qemuProcessStart(conn, driver, vm, "stdio", true, fd, path, + VIR_VM_OP_RESTORE); if (intermediate_pid != -1) { if (ret < 0) { @@ -5836,7 +3477,7 @@ qemudDomainSaveImageStartVM(virConnectPtr conn, /* If it was running before, resume it now. */ if (header->was_running) { - if (doStartCPUs(driver, vm, conn) < 0) { + if (qemuProcessStartCPUs(driver, vm, conn) < 0) { if (virGetLastError() == NULL) qemuReportError(VIR_ERR_OPERATION_FAILED, "%s", _("failed to resume domain")); @@ -6136,7 +3777,7 @@ static char *qemuDomainXMLToNative(virConnectPtr conn, &qemuCmdFlags) < 0) goto cleanup; - if (qemuPrepareMonitorChr(driver, &monConfig, def->name) < 0) + if (qemuProcessPrepareMonitorChr(driver, &monConfig, def->name) < 0) goto cleanup; if (!(cmd = qemuBuildCommandLine(conn, driver, def, @@ -6202,8 +3843,8 @@ static int qemudDomainObjStart(virConnectPtr conn, goto cleanup; } - ret = qemudStartVMDaemon(conn, driver, vm, NULL, start_paused, -1, NULL, - VIR_VM_OP_CREATE); + ret = qemuProcessStart(conn, driver, vm, NULL, start_paused, -1, NULL, + VIR_VM_OP_CREATE); qemuDomainStartAudit(vm, "booted", ret >= 0); if (ret >= 0) { virDomainEventPtr event = @@ -6390,7 +4031,7 @@ static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) { if (qemudCanonicalizeMachine(driver, def) < 0) goto cleanup; - if (qemuAssignPCIAddresses(def) < 0) + if (qemuDomainAssignPCIAddresses(def) < 0) goto cleanup; if (!(vm = virDomainAssignDef(driver->caps, @@ -7894,60 +5535,6 @@ qemuDomainEventDeregisterAny(virConnectPtr conn, } -static void qemuDomainEventDispatchFunc(virConnectPtr conn, - virDomainEventPtr event, - virConnectDomainEventGenericCallback cb, - void *cbopaque, - void *opaque) -{ - struct qemud_driver *driver = opaque; - - /* Drop the lock whle dispatching, for sake of re-entrancy */ - qemuDriverUnlock(driver); - virDomainEventDispatchDefaultFunc(conn, event, cb, cbopaque, NULL); - qemuDriverLock(driver); -} - -static void qemuDomainEventFlush(int timer ATTRIBUTE_UNUSED, void *opaque) -{ - struct qemud_driver *driver = opaque; - virDomainEventQueue tempQueue; - - qemuDriverLock(driver); - - driver->domainEventDispatching = 1; - - /* Copy the queue, so we're reentrant safe */ - tempQueue.count = driver->domainEventQueue->count; - tempQueue.events = driver->domainEventQueue->events; - driver->domainEventQueue->count = 0; - driver->domainEventQueue->events = NULL; - - virEventUpdateTimeout(driver->domainEventTimer, -1); - virDomainEventQueueDispatch(&tempQueue, - driver->domainEventCallbacks, - qemuDomainEventDispatchFunc, - driver); - - /* Purge any deleted callbacks */ - virDomainEventCallbackListPurgeMarked(driver->domainEventCallbacks); - - driver->domainEventDispatching = 0; - qemuDriverUnlock(driver); -} - - -/* driver must be locked before calling */ -static void qemuDomainEventQueue(struct qemud_driver *driver, - virDomainEventPtr event) -{ - if (virDomainEventQueuePush(driver->domainEventQueue, - event) < 0) - virDomainEventFree(event); - if (qemu_driver->domainEventQueue->count == 1) - virEventUpdateTimeout(driver->domainEventTimer, 0); -} - /* Migration support. */ static bool ATTRIBUTE_NONNULL(1) @@ -8078,12 +5665,12 @@ qemudDomainMigratePrepareTunnel(virConnectPtr dconn, /* Start the QEMU daemon, with the same command-line arguments plus * -incoming unix:/path/to/file or exec:nc -U /path/to/file */ - internalret = qemudStartVMDaemon(dconn, driver, vm, migrateFrom, true, - -1, NULL, VIR_VM_OP_MIGRATE_IN_START); + internalret = qemuProcessStart(dconn, driver, vm, migrateFrom, true, + -1, NULL, VIR_VM_OP_MIGRATE_IN_START); VIR_FREE(migrateFrom); if (internalret < 0) { qemuDomainStartAudit(vm, "migrated", false); - /* Note that we don't set an error here because qemudStartVMDaemon + /* Note that we don't set an error here because qemuProcessStart * should have already done that. */ if (!vm->persistent) { @@ -8097,7 +5684,7 @@ qemudDomainMigratePrepareTunnel(virConnectPtr dconn, unixfile, false) < 0) { qemuDomainStartAudit(vm, "migrated", false); - qemudShutdownVMDaemon(driver, vm, 0); + qemuProcessStop(driver, vm, 0); if (!vm->persistent) { if (qemuDomainObjEndJob(vm) > 0) virDomainRemoveInactive(&driver->domains, vm); @@ -8324,10 +5911,10 @@ qemudDomainMigratePrepare2 (virConnectPtr dconn, * -incoming tcp:0.0.0.0:port */ snprintf (migrateFrom, sizeof (migrateFrom), "tcp:0.0.0.0:%d", this_port); - if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, true, - -1, NULL, VIR_VM_OP_MIGRATE_IN_START) < 0) { + if (qemuProcessStart(dconn, driver, vm, migrateFrom, true, + -1, NULL, VIR_VM_OP_MIGRATE_IN_START) < 0) { qemuDomainStartAudit(vm, "migrated", false); - /* Note that we don't set an error here because qemudStartVMDaemon + /* Note that we don't set an error here because qemuProcessStart * should have already done that. */ if (!vm->persistent) { @@ -8909,7 +6496,7 @@ qemudDomainMigratePerform (virDomainPtr dom, } /* Clean up the source domain. */ - qemudShutdownVMDaemon(driver, vm, 1); + qemuProcessStop(driver, vm, 1); qemuDomainStopAudit(vm, "migrated"); resume = 0; @@ -8927,7 +6514,7 @@ qemudDomainMigratePerform (virDomainPtr dom, endjob: if (resume && vm->state == VIR_DOMAIN_PAUSED) { /* we got here through some sort of failure; start the domain again */ - if (doStartCPUs(driver, vm, dom->conn) < 0) { + if (qemuProcessStartCPUs(driver, vm, dom->conn) < 0) { /* Hm, we already know we are in error here. We don't want to * overwrite the previous error, though, so we just throw something * to the logs and hope for the best @@ -9091,7 +6678,7 @@ qemudDomainMigrateFinish2 (virConnectPtr dconn, * >= 0.10.6 to work properly. This isn't strictly necessary on * older qemu's, but it also doesn't hurt anything there */ - if (doStartCPUs(driver, vm, dconn) < 0) { + if (qemuProcessStartCPUs(driver, vm, dconn) < 0) { if (virGetLastError() == NULL) qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("resume operation failed")); @@ -9113,7 +6700,7 @@ qemudDomainMigrateFinish2 (virConnectPtr dconn, goto endjob; } } else { - qemudShutdownVMDaemon(driver, vm, 1); + qemuProcessStop(driver, vm, 1); qemuDomainStopAudit(vm, "failed"); event = virDomainEventNewFromObj(vm, VIR_DOMAIN_EVENT_STOPPED, @@ -9942,8 +7529,8 @@ static int qemuDomainRevertToSnapshot(virDomainSnapshotPtr snapshot, if (qemuDomainSnapshotSetCurrentActive(vm, driver->snapshotDir) < 0) goto endjob; - rc = qemudStartVMDaemon(snapshot->domain->conn, driver, vm, NULL, - false, -1, NULL, VIR_VM_OP_CREATE); + rc = qemuProcessStart(snapshot->domain->conn, driver, vm, NULL, + false, -1, NULL, VIR_VM_OP_CREATE); qemuDomainStartAudit(vm, "from-snapshot", rc >= 0); if (qemuDomainSnapshotSetCurrentInactive(vm, driver->snapshotDir) < 0) goto endjob; @@ -9955,7 +7542,7 @@ static int qemuDomainRevertToSnapshot(virDomainSnapshotPtr snapshot, /* qemu unconditionally starts the domain running again after * loadvm, so let's pause it to keep consistency */ - rc = doStopCPUs(driver, vm); + rc = qemuProcessStopCPUs(driver, vm); if (rc < 0) goto endjob; } @@ -9976,7 +7563,7 @@ static int qemuDomainRevertToSnapshot(virDomainSnapshotPtr snapshot, */ if (virDomainObjIsActive(vm)) { - qemudShutdownVMDaemon(driver, vm, 0); + qemuProcessStop(driver, vm, 0); qemuDomainStopAudit(vm, "from-snapshot"); event = virDomainEventNewFromObj(vm, VIR_DOMAIN_EVENT_STOPPED, @@ -10473,56 +8060,32 @@ static virStateDriver qemuStateDriver = { .active = qemudActive, }; -static int -qemudVMFilterRebuild(virConnectPtr conn ATTRIBUTE_UNUSED, - virHashIterator iter, void *data) -{ - virHashForEach(qemu_driver->domains.objs, iter, data); - - return 0; -} - -static int -qemudVMFiltersInstantiate(virConnectPtr conn, - virDomainDefPtr def) -{ - int err = 0; - int i; - - if (!conn) - return 1; - - for (i = 0 ; i < def->nnets ; i++) { - virDomainNetDefPtr net = def->nets[i]; - if ((net->filter) && (net->ifname)) { - if (virDomainConfNWFilterInstantiate(conn, net)) { - err = 1; - break; - } - } - } - - return err; -} - - static void -qemudVMDriverLock(void) { +qemuVMDriverLock(void) { qemuDriverLock(qemu_driver); }; static void -qemudVMDriverUnlock(void) { +qemuVMDriverUnlock(void) { qemuDriverUnlock(qemu_driver); }; +static int +qemuVMFilterRebuild(virConnectPtr conn ATTRIBUTE_UNUSED, + virHashIterator iter, void *data) +{ + virHashForEach(qemu_driver->domains.objs, iter, data); + + return 0; +} + static virNWFilterCallbackDriver qemuCallbackDriver = { .name = "QEMU", - .vmFilterRebuild = qemudVMFilterRebuild, - .vmDriverLock = qemudVMDriverLock, - .vmDriverUnlock = qemudVMDriverUnlock, + .vmFilterRebuild = qemuVMFilterRebuild, + .vmDriverLock = qemuVMDriverLock, + .vmDriverUnlock = qemuVMDriverUnlock, }; int qemuRegister(void) { diff --git a/src/qemu/qemu_driver.h b/src/qemu/qemu_driver.h index dac0935..73da9e4 100644 --- a/src/qemu/qemu_driver.h +++ b/src/qemu/qemu_driver.h @@ -21,34 +21,9 @@ * Author: Daniel P. Berrange <berrange(a)redhat.com> */ - -#ifndef QEMUD_DRIVER_H -# define QEMUD_DRIVER_H - -# include <config.h> - -# include <libxml/xpath.h> - -# include "internal.h" - -# if HAVE_LINUX_KVM_H -# include <linux/kvm.h> -# endif - -/* device for kvm ioctls */ -# define KVM_DEVICE "/dev/kvm" - -/* add definitions missing in older linux/kvm.h */ -# ifndef KVMIO -# define KVMIO 0xAE -# endif -# ifndef KVM_CHECK_EXTENSION -# define KVM_CHECK_EXTENSION _IO(KVMIO, 0x03) -# endif -# ifndef KVM_CAP_NR_VCPUS -# define KVM_CAP_NR_VCPUS 9 /* returns max vcpus per vm */ -# endif +#ifndef __QEMU_DRIVER_H__ +# define __QEMU_DRIVER_H__ int qemuRegister(void); -#endif /* QEMUD_DRIVER_H */ +#endif /* __QEMU_DRIVER_H__ */ diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c new file mode 100644 index 0000000..1e3dea2 --- /dev/null +++ b/src/qemu/qemu_process.c @@ -0,0 +1,2412 @@ +/* + * qemu_process.h: QEMU process management + * + * Copyright (C) 2006-2011 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 + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include <config.h> + +#include <fcntl.h> +#include <unistd.h> +#include <signal.h> + +#include "qemu_process.h" +#include "qemu_domain.h" +#include "qemu_cgroup.h" +#include "qemu_capabilities.h" +#include "qemu_monitor.h" +#include "qemu_command.h" +#include "qemu_audit.h" +#include "qemu_hostdev.h" +#include "qemu_hotplug.h" +#include "qemu_bridge_filter.h" + +#include "datatypes.h" +#include "logging.h" +#include "virterror_internal.h" +#include "memory.h" +#include "hooks.h" +#include "files.h" +#include "util.h" +#include "c-ctype.h" +#include "nodeinfo.h" +#include "processinfo.h" +#include "domain_nwfilter.h" + +#define VIR_FROM_THIS VIR_FROM_QEMU + +#define START_POSTFIX ": starting up\n" +#define SHUTDOWN_POSTFIX ": shutting down\n" + +/** + * qemudRemoveDomainStatus + * + * remove all state files of a domain from statedir + * + * Returns 0 on success + */ +static int +qemuProcessRemoveDomainStatus(struct qemud_driver *driver, + virDomainObjPtr vm) +{ + char ebuf[1024]; + char *file = NULL; + + if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) { + virReportOOMError(); + return(-1); + } + + if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR) + VIR_WARN("Failed to remove domain XML for %s: %s", + vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf))); + VIR_FREE(file); + + if (virFileDeletePid(driver->stateDir, vm->def->name) != 0) + VIR_WARN("Failed to remove PID file for %s: %s", + vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf))); + + + return 0; +} + + +/* XXX figure out how to remove this */ +extern struct qemud_driver *qemu_driver; + +/* + * This is a callback registered with a qemuMonitorPtr instance, + * and to be invoked when the monitor console hits an end of file + * condition, or error, thus indicating VM shutdown should be + * performed + */ +static void +qemuProcessHandleMonitorEOF(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm, + int hasError) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr event = NULL; + qemuDomainObjPrivatePtr priv; + + VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name); + + virDomainObjLock(vm); + + if (!virDomainObjIsActive(vm)) { + VIR_DEBUG("Domain %p is not active, ignoring EOF", vm); + virDomainObjUnlock(vm); + return; + } + + priv = vm->privateData; + if (!hasError && priv->monJSON && !priv->gotShutdown) { + VIR_DEBUG("Monitor connection to '%s' closed without SHUTDOWN event; " + "assuming the domain crashed", vm->def->name); + hasError = 1; + } + + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_STOPPED, + hasError ? + VIR_DOMAIN_EVENT_STOPPED_FAILED : + VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN); + + qemuProcessStop(driver, vm, 0); + qemuDomainStopAudit(vm, hasError ? "failed" : "shutdown"); + + if (!vm->persistent) + virDomainRemoveInactive(&driver->domains, vm); + else + virDomainObjUnlock(vm); + + if (event) { + qemuDriverLock(driver); + qemuDomainEventQueue(driver, event); + qemuDriverUnlock(driver); + } +} + + +static virDomainDiskDefPtr +qemuProcessFindDomainDiskByPath(virDomainObjPtr vm, + const char *path) +{ + int i; + + for (i = 0; i < vm->def->ndisks; i++) { + virDomainDiskDefPtr disk; + + disk = vm->def->disks[i]; + if (disk->src != NULL && STREQ(disk->src, path)) + return disk; + } + + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("no disk found with path %s"), + path); + return NULL; +} + +static virDomainDiskDefPtr +qemuProcessFindDomainDiskByAlias(virDomainObjPtr vm, + const char *alias) +{ + int i; + + if (STRPREFIX(alias, QEMU_DRIVE_HOST_PREFIX)) + alias += strlen(QEMU_DRIVE_HOST_PREFIX); + + for (i = 0; i < vm->def->ndisks; i++) { + virDomainDiskDefPtr disk; + + disk = vm->def->disks[i]; + if (disk->info.alias != NULL && STREQ(disk->info.alias, alias)) + return disk; + } + + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("no disk found with alias %s"), + alias); + return NULL; +} + +static int +qemuProcessGetVolumeQcowPassphrase(virConnectPtr conn, + virDomainDiskDefPtr disk, + char **secretRet, + size_t *secretLen) +{ + virSecretPtr secret; + char *passphrase; + unsigned char *data; + size_t size; + int ret = -1; + virStorageEncryptionPtr enc; + + if (!disk->encryption) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("disk %s does not have any encryption information"), + disk->src); + return -1; + } + enc = disk->encryption; + + if (!conn) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("cannot find secrets without a connection")); + goto cleanup; + } + + if (conn->secretDriver == NULL || + conn->secretDriver->lookupByUUID == NULL || + conn->secretDriver->getValue == NULL) { + qemuReportError(VIR_ERR_NO_SUPPORT, "%s", + _("secret storage not supported")); + goto cleanup; + } + + if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW || + enc->nsecrets != 1 || + enc->secrets[0]->type != + VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) { + qemuReportError(VIR_ERR_XML_ERROR, + _("invalid <encryption> for volume %s"), disk->src); + goto cleanup; + } + + secret = conn->secretDriver->lookupByUUID(conn, + enc->secrets[0]->uuid); + if (secret == NULL) + goto cleanup; + data = conn->secretDriver->getValue(secret, &size, + VIR_SECRET_GET_VALUE_INTERNAL_CALL); + virUnrefSecret(secret); + if (data == NULL) + goto cleanup; + + if (memchr(data, '\0', size) != NULL) { + memset(data, 0, size); + VIR_FREE(data); + qemuReportError(VIR_ERR_XML_ERROR, + _("format='qcow' passphrase for %s must not contain a " + "'\\0'"), disk->src); + goto cleanup; + } + + if (VIR_ALLOC_N(passphrase, size + 1) < 0) { + memset(data, 0, size); + VIR_FREE(data); + virReportOOMError(); + goto cleanup; + } + memcpy(passphrase, data, size); + passphrase[size] = '\0'; + + memset(data, 0, size); + VIR_FREE(data); + + *secretRet = passphrase; + *secretLen = size; + + ret = 0; + +cleanup: + return ret; +} + +static int +qemuProcessFindVolumeQcowPassphrase(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virConnectPtr conn, + virDomainObjPtr vm, + const char *path, + char **secretRet, + size_t *secretLen) +{ + virDomainDiskDefPtr disk; + int ret = -1; + + virDomainObjLock(vm); + disk = qemuProcessFindDomainDiskByPath(vm, path); + + if (!disk) + goto cleanup; + + ret = qemuProcessGetVolumeQcowPassphrase(conn, disk, secretRet, secretLen); + +cleanup: + virDomainObjUnlock(vm); + return ret; +} + + +static int +qemuProcessHandleReset(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr event; + + virDomainObjLock(vm); + event = virDomainEventRebootNewFromObj(vm); + virDomainObjUnlock(vm); + + if (event) { + qemuDriverLock(driver); + qemuDomainEventQueue(driver, event); + qemuDriverUnlock(driver); + } + + return 0; +} + + +static int +qemuProcessHandleShutdown(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm) +{ + virDomainObjLock(vm); + ((qemuDomainObjPrivatePtr) vm->privateData)->gotShutdown = true; + virDomainObjUnlock(vm); + + return 0; +} + + +static int +qemuProcessHandleStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr event = NULL; + + virDomainObjLock(vm); + if (vm->state == VIR_DOMAIN_RUNNING) { + VIR_DEBUG("Transitioned guest %s to paused state due to unknown event", vm->def->name); + + vm->state = VIR_DOMAIN_PAUSED; + event = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_SUSPENDED, + VIR_DOMAIN_EVENT_SUSPENDED_PAUSED); + + if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) + VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name); + } + virDomainObjUnlock(vm); + + if (event) { + qemuDriverLock(driver); + if (event) + qemuDomainEventQueue(driver, event); + qemuDriverUnlock(driver); + } + + return 0; +} + + +static int +qemuProcessHandleRTCChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm, + long long offset) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr event; + + virDomainObjLock(vm); + event = virDomainEventRTCChangeNewFromObj(vm, offset); + + if (vm->def->clock.offset == VIR_DOMAIN_CLOCK_OFFSET_VARIABLE) + vm->def->clock.data.adjustment = offset; + + if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) + VIR_WARN0("unable to save domain status with RTC change"); + + virDomainObjUnlock(vm); + + if (event) { + qemuDriverLock(driver); + qemuDomainEventQueue(driver, event); + qemuDriverUnlock(driver); + } + + return 0; +} + + +static int +qemuProcessHandleWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm, + int action) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr watchdogEvent = NULL; + virDomainEventPtr lifecycleEvent = NULL; + + virDomainObjLock(vm); + watchdogEvent = virDomainEventWatchdogNewFromObj(vm, action); + + if (action == VIR_DOMAIN_EVENT_WATCHDOG_PAUSE && + vm->state == VIR_DOMAIN_RUNNING) { + VIR_DEBUG("Transitioned guest %s to paused state due to watchdog", vm->def->name); + + vm->state = VIR_DOMAIN_PAUSED; + lifecycleEvent = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_SUSPENDED, + VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG); + + if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) + VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name); + } + + if (vm->def->watchdog->action == VIR_DOMAIN_WATCHDOG_ACTION_DUMP) { + struct qemuDomainWatchdogEvent *wdEvent; + if (VIR_ALLOC(wdEvent) == 0) { + wdEvent->action = VIR_DOMAIN_WATCHDOG_ACTION_DUMP; + wdEvent->vm = vm; + ignore_value(virThreadPoolSendJob(driver->workerPool, wdEvent)); + } else + virReportOOMError(); + } + + virDomainObjUnlock(vm); + + if (watchdogEvent || lifecycleEvent) { + qemuDriverLock(driver); + if (watchdogEvent) + qemuDomainEventQueue(driver, watchdogEvent); + if (lifecycleEvent) + qemuDomainEventQueue(driver, lifecycleEvent); + qemuDriverUnlock(driver); + } + + return 0; +} + + +static int +qemuProcessHandleIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm, + const char *diskAlias, + int action, + const char *reason) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr ioErrorEvent = NULL; + virDomainEventPtr ioErrorEvent2 = NULL; + virDomainEventPtr lifecycleEvent = NULL; + const char *srcPath; + const char *devAlias; + virDomainDiskDefPtr disk; + + virDomainObjLock(vm); + disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias); + + if (disk) { + srcPath = disk->src; + devAlias = disk->info.alias; + } else { + srcPath = ""; + devAlias = ""; + } + + ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action); + ioErrorEvent2 = virDomainEventIOErrorReasonNewFromObj(vm, srcPath, devAlias, action, reason); + + if (action == VIR_DOMAIN_EVENT_IO_ERROR_PAUSE && + vm->state == VIR_DOMAIN_RUNNING) { + VIR_DEBUG("Transitioned guest %s to paused state due to IO error", vm->def->name); + + vm->state = VIR_DOMAIN_PAUSED; + lifecycleEvent = virDomainEventNewFromObj(vm, + VIR_DOMAIN_EVENT_SUSPENDED, + VIR_DOMAIN_EVENT_SUSPENDED_IOERROR); + + if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) + VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name); + } + virDomainObjUnlock(vm); + + if (ioErrorEvent || ioErrorEvent2 || lifecycleEvent) { + qemuDriverLock(driver); + if (ioErrorEvent) + qemuDomainEventQueue(driver, ioErrorEvent); + if (ioErrorEvent2) + qemuDomainEventQueue(driver, ioErrorEvent2); + if (lifecycleEvent) + qemuDomainEventQueue(driver, lifecycleEvent); + qemuDriverUnlock(driver); + } + + return 0; +} + + +static int +qemuProcessHandleGraphics(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm, + int phase, + int localFamily, + const char *localNode, + const char *localService, + int remoteFamily, + const char *remoteNode, + const char *remoteService, + const char *authScheme, + const char *x509dname, + const char *saslUsername) +{ + struct qemud_driver *driver = qemu_driver; + virDomainEventPtr event; + virDomainEventGraphicsAddressPtr localAddr = NULL; + virDomainEventGraphicsAddressPtr remoteAddr = NULL; + virDomainEventGraphicsSubjectPtr subject = NULL; + int i; + + virDomainObjLock(vm); + + if (VIR_ALLOC(localAddr) < 0) + goto no_memory; + localAddr->family = localFamily; + if (!(localAddr->service = strdup(localService)) || + !(localAddr->node = strdup(localNode))) + goto no_memory; + + if (VIR_ALLOC(remoteAddr) < 0) + goto no_memory; + remoteAddr->family = remoteFamily; + if (!(remoteAddr->service = strdup(remoteService)) || + !(remoteAddr->node = strdup(remoteNode))) + goto no_memory; + + if (VIR_ALLOC(subject) < 0) + goto no_memory; + if (x509dname) { + if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0) + goto no_memory; + if (!(subject->identities[subject->nidentity].type = strdup("x509dname")) || + !(subject->identities[subject->nidentity].name = strdup(x509dname))) + goto no_memory; + subject->nidentity++; + } + if (saslUsername) { + if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0) + goto no_memory; + if (!(subject->identities[subject->nidentity].type = strdup("saslUsername")) || + !(subject->identities[subject->nidentity].name = strdup(saslUsername))) + goto no_memory; + subject->nidentity++; + } + + event = virDomainEventGraphicsNewFromObj(vm, phase, localAddr, remoteAddr, authScheme, subject); + virDomainObjUnlock(vm); + + if (event) { + qemuDriverLock(driver); + qemuDomainEventQueue(driver, event); + qemuDriverUnlock(driver); + } + + return 0; + +no_memory: + virReportOOMError(); + if (localAddr) { + VIR_FREE(localAddr->service); + VIR_FREE(localAddr->node); + VIR_FREE(localAddr); + } + if (remoteAddr) { + VIR_FREE(remoteAddr->service); + VIR_FREE(remoteAddr->node); + VIR_FREE(remoteAddr); + } + if (subject) { + for (i = 0 ; i < subject->nidentity ; i++) { + VIR_FREE(subject->identities[i].type); + VIR_FREE(subject->identities[i].name); + } + VIR_FREE(subject->identities); + VIR_FREE(subject); + } + + return -1; +} + + +static void qemuProcessHandleMonitorDestroy(qemuMonitorPtr mon, + virDomainObjPtr vm) +{ + qemuDomainObjPrivatePtr priv = vm->privateData; + if (priv->mon == mon) + priv->mon = NULL; + virDomainObjUnref(vm); +} + +static qemuMonitorCallbacks monitorCallbacks = { + .destroy = qemuProcessHandleMonitorDestroy, + .eofNotify = qemuProcessHandleMonitorEOF, + .diskSecretLookup = qemuProcessFindVolumeQcowPassphrase, + .domainShutdown = qemuProcessHandleShutdown, + .domainStop = qemuProcessHandleStop, + .domainReset = qemuProcessHandleReset, + .domainRTCChange = qemuProcessHandleRTCChange, + .domainWatchdog = qemuProcessHandleWatchdog, + .domainIOError = qemuProcessHandleIOError, + .domainGraphics = qemuProcessHandleGraphics, +}; + +static int +qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm) +{ + qemuDomainObjPrivatePtr priv = vm->privateData; + int ret = -1; + + if (virSecurityManagerSetSocketLabel(driver->securityManager, vm) < 0) { + VIR_ERROR(_("Failed to set security context for monitor for %s"), + vm->def->name); + goto error; + } + + /* Hold an extra reference because we can't allow 'vm' to be + * deleted while the monitor is active */ + virDomainObjRef(vm); + + priv->mon = qemuMonitorOpen(vm, + priv->monConfig, + priv->monJSON, + &monitorCallbacks); + + if (priv->mon == NULL) + virDomainObjUnref(vm); + + if (virSecurityManagerClearSocketLabel(driver->securityManager, vm) < 0) { + VIR_ERROR(_("Failed to clear security context for monitor for %s"), + vm->def->name); + goto error; + } + + if (priv->mon == NULL) { + VIR_INFO("Failed to connect monitor for %s", vm->def->name); + goto error; + } + + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + ret = qemuMonitorSetCapabilities(priv->mon); + qemuDomainObjExitMonitorWithDriver(driver, vm); + +error: + + return ret; +} + +static int +qemuProcessLogFD(struct qemud_driver *driver, const char* name, bool append) +{ + char *logfile; + mode_t logmode; + int fd = -1; + + if (virAsprintf(&logfile, "%s/%s.log", driver->logDir, name) < 0) { + virReportOOMError(); + return -1; + } + + logmode = O_CREAT | O_WRONLY; + /* Only logrotate files in /var/log, so only append if running privileged */ + if (driver->privileged || append) + logmode |= O_APPEND; + else + logmode |= O_TRUNC; + + if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) { + virReportSystemError(errno, + _("failed to create logfile %s"), + logfile); + VIR_FREE(logfile); + return -1; + } + VIR_FREE(logfile); + if (virSetCloseExec(fd) < 0) { + virReportSystemError(errno, "%s", + _("Unable to set VM logfile close-on-exec flag")); + VIR_FORCE_CLOSE(fd); + return -1; + } + return fd; +} + + +static int +qemuProcessLogReadFD(const char* logDir, const char* name, off_t pos) +{ + char *logfile; + mode_t logmode = O_RDONLY; + int fd = -1; + + if (virAsprintf(&logfile, "%s/%s.log", logDir, name) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("failed to build logfile name %s/%s.log"), + logDir, name); + return -1; + } + + if ((fd = open(logfile, logmode)) < 0) { + virReportSystemError(errno, + _("failed to create logfile %s"), + logfile); + VIR_FREE(logfile); + return -1; + } + if (virSetCloseExec(fd) < 0) { + virReportSystemError(errno, "%s", + _("Unable to set VM logfile close-on-exec flag")); + VIR_FORCE_CLOSE(fd); + VIR_FREE(logfile); + return -1; + } + if (pos < 0 || lseek(fd, pos, SEEK_SET) < 0) { + virReportSystemError(pos < 0 ? 0 : errno, + _("Unable to seek to %lld in %s"), + (long long) pos, logfile); + VIR_FORCE_CLOSE(fd); + } + VIR_FREE(logfile); + return fd; +} + + +typedef int qemuProcessLogHandleOutput(virDomainObjPtr vm, + const char *output, + int fd); + +/* + * Returns -1 for error, 0 on success + */ +static int +qemuProcessReadLogOutput(virDomainObjPtr vm, + int fd, + char *buf, + size_t buflen, + qemuProcessLogHandleOutput func, + const char *what, + int timeout) +{ + int retries = (timeout*10); + int got = 0; + buf[0] = '\0'; + + while (retries) { + ssize_t func_ret, ret; + int isdead = 0; + + func_ret = func(vm, buf, fd); + + if (kill(vm->pid, 0) == -1 && errno == ESRCH) + isdead = 1; + + /* Any failures should be detected before we read the log, so we + * always have something useful to report on failure. */ + ret = saferead(fd, buf+got, buflen-got-1); + if (ret < 0) { + virReportSystemError(errno, + _("Failure while reading %s log output"), + what); + return -1; + } + + got += ret; + buf[got] = '\0'; + if (got == buflen-1) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Out of space while reading %s log output: %s"), + what, buf); + return -1; + } + + if (isdead) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Process exited while reading %s log output: %s"), + what, buf); + return -1; + } + + if (func_ret <= 0) + return func_ret; + + usleep(100*1000); + retries--; + } + + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Timed out while reading %s log output: %s"), + what, buf); + return -1; +} + + +/* + * Look at a chunk of data from the QEMU stdout logs and try to + * find a TTY device, as indicated by a line like + * + * char device redirected to /dev/pts/3 + * + * Returns -1 for error, 0 success, 1 continue reading + */ +static int +qemuProcessExtractTTYPath(const char *haystack, + size_t *offset, + char **path) +{ + static const char needle[] = "char device redirected to"; + char *tmp, *dev; + + VIR_FREE(*path); + /* First look for our magic string */ + if (!(tmp = strstr(haystack + *offset, needle))) { + return 1; + } + tmp += sizeof(needle); + dev = tmp; + + /* + * And look for first whitespace character and nul terminate + * to mark end of the pty path + */ + while (*tmp) { + if (c_isspace(*tmp)) { + *path = strndup(dev, tmp-dev); + if (*path == NULL) { + virReportOOMError(); + return -1; + } + + /* ... now further update offset till we get EOL */ + *offset = tmp - haystack; + return 0; + } + tmp++; + } + + /* + * We found a path, but didn't find any whitespace, + * so it must be still incomplete - we should at + * least see a \n - indicate that we want to carry + * on trying again + */ + return 1; +} + +static int +qemuProcessFindCharDevicePTYsMonitor(virDomainObjPtr vm, + virHashTablePtr paths) +{ + int i; + +#define LOOKUP_PTYS(array, arraylen, idprefix) \ + for (i = 0 ; i < (arraylen) ; i++) { \ + virDomainChrDefPtr chr = (array)[i]; \ + if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { \ + char id[16]; \ + \ + if (snprintf(id, sizeof(id), idprefix "%i", i) >= sizeof(id)) \ + return -1; \ + \ + const char *path = (const char *) virHashLookup(paths, id); \ + if (path == NULL) { \ + if (chr->source.data.file.path == NULL) { \ + /* neither the log output nor 'info chardev' had a */ \ + /* pty path for this chardev, report an error */ \ + qemuReportError(VIR_ERR_INTERNAL_ERROR, \ + _("no assigned pty for device %s"), id); \ + return -1; \ + } else { \ + /* 'info chardev' had no pty path for this chardev, */\ + /* but the log output had, so we're fine */ \ + continue; \ + } \ + } \ + \ + VIR_FREE(chr->source.data.file.path); \ + chr->source.data.file.path = strdup(path); \ + \ + if (chr->source.data.file.path == NULL) { \ + virReportOOMError(); \ + return -1; \ + } \ + } \ + } + + LOOKUP_PTYS(vm->def->serials, vm->def->nserials, "serial"); + LOOKUP_PTYS(vm->def->parallels, vm->def->nparallels, "parallel"); + LOOKUP_PTYS(vm->def->channels, vm->def->nchannels, "channel"); + if (vm->def->console) + LOOKUP_PTYS(&vm->def->console, 1, "console"); +#undef LOOKUP_PTYS + + return 0; +} + +static int +qemuProcessFindCharDevicePTYs(virDomainObjPtr vm, + const char *output, + int fd ATTRIBUTE_UNUSED) +{ + size_t offset = 0; + int ret, i; + + /* The order in which QEMU prints out the PTY paths is + the order in which it procsses its serial and parallel + device args. This code must match that ordering.... */ + + /* first comes the serial devices */ + for (i = 0 ; i < vm->def->nserials ; i++) { + virDomainChrDefPtr chr = vm->def->serials[i]; + if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { + if ((ret = qemuProcessExtractTTYPath(output, &offset, + &chr->source.data.file.path)) != 0) + return ret; + } + } + + /* then the parallel devices */ + for (i = 0 ; i < vm->def->nparallels ; i++) { + virDomainChrDefPtr chr = vm->def->parallels[i]; + if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { + if ((ret = qemuProcessExtractTTYPath(output, &offset, + &chr->source.data.file.path)) != 0) + return ret; + } + } + + /* then the channel devices */ + for (i = 0 ; i < vm->def->nchannels ; i++) { + virDomainChrDefPtr chr = vm->def->channels[i]; + if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { + if ((ret = qemuProcessExtractTTYPath(output, &offset, + &chr->source.data.file.path)) != 0) + return ret; + } + } + + return 0; +} + +static void qemuProcessFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED) +{ + VIR_FREE(payload); +} + +static void +qemuProcessReadLogFD(int logfd, char *buf, int maxlen, int off) +{ + int ret; + char *tmpbuf = buf + off; + + ret = saferead(logfd, tmpbuf, maxlen - off - 1); + if (ret < 0) { + ret = 0; + } + + tmpbuf[ret] = '\0'; +} + +static int +qemuProcessWaitForMonitor(struct qemud_driver* driver, + virDomainObjPtr vm, off_t pos) +{ + char buf[4096] = ""; /* Plenty of space to get startup greeting */ + int logfd; + int ret = -1; + virHashTablePtr paths = NULL; + + if ((logfd = qemuProcessLogReadFD(driver->logDir, vm->def->name, pos)) < 0) + return -1; + + if (qemuProcessReadLogOutput(vm, logfd, buf, sizeof(buf), + qemuProcessFindCharDevicePTYs, + "console", 30) < 0) + goto closelog; + + VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name); + if (qemuConnectMonitor(driver, vm) < 0) { + goto cleanup; + } + + /* Try to get the pty path mappings again via the monitor. This is much more + * reliable if it's available. + * Note that the monitor itself can be on a pty, so we still need to try the + * log output method. */ + paths = virHashCreate(0); + if (paths == NULL) { + virReportOOMError(); + goto cleanup; + } + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + qemuDomainObjPrivatePtr priv = vm->privateData; + ret = qemuMonitorGetPtyPaths(priv->mon, paths); + qemuDomainObjExitMonitorWithDriver(driver, vm); + + VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret); + if (ret == 0) + ret = qemuProcessFindCharDevicePTYsMonitor(vm, paths); + +cleanup: + if (paths) { + virHashFree(paths, qemuProcessFreePtyPath); + } + + if (kill(vm->pid, 0) == -1 && errno == ESRCH) { + /* VM is dead, any other error raised in the interim is probably + * not as important as the qemu cmdline output */ + qemuProcessReadLogFD(logfd, buf, sizeof(buf), strlen(buf)); + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("process exited while connecting to monitor: %s"), + buf); + ret = -1; + } + +closelog: + if (VIR_CLOSE(logfd) < 0) { + char ebuf[4096]; + VIR_WARN("Unable to close logfile: %s", + virStrerror(errno, ebuf, sizeof ebuf)); + } + + return ret; +} + +static int +qemuProcessDetectVcpuPIDs(struct qemud_driver *driver, + virDomainObjPtr vm) +{ + pid_t *cpupids = NULL; + int ncpupids; + qemuDomainObjPrivatePtr priv = vm->privateData; + + if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) { + priv->nvcpupids = 1; + if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) { + virReportOOMError(); + return -1; + } + priv->vcpupids[0] = vm->pid; + return 0; + } + + /* What follows is now all KVM specific */ + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) { + qemuDomainObjExitMonitorWithDriver(driver, vm); + return -1; + } + qemuDomainObjExitMonitorWithDriver(driver, vm); + + /* Treat failure to get VCPU<->PID mapping as non-fatal */ + if (ncpupids == 0) + return 0; + + if (ncpupids != vm->def->vcpus) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("got wrong number of vCPU pids from QEMU monitor. " + "got %d, wanted %d"), + ncpupids, vm->def->vcpus); + VIR_FREE(cpupids); + return -1; + } + + priv->nvcpupids = ncpupids; + priv->vcpupids = cpupids; + return 0; +} + +/* + * To be run between fork/exec of QEMU only + */ +static int +qemuProcessInitCpuAffinity(virDomainObjPtr vm) +{ + int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN; + virNodeInfo nodeinfo; + unsigned char *cpumap; + int cpumaplen; + + DEBUG0("Setting CPU affinity"); + + if (nodeGetInfo(NULL, &nodeinfo) < 0) + return -1; + + /* setaffinity fails if you set bits for CPUs which + * aren't present, so we have to limit ourselves */ + hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo); + if (maxcpu > hostcpus) + maxcpu = hostcpus; + + cpumaplen = VIR_CPU_MAPLEN(maxcpu); + if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) { + virReportOOMError(); + return -1; + } + + if (vm->def->cpumask) { + /* XXX why don't we keep 'cpumask' in the libvirt cpumap + * format to start with ?!?! */ + for (i = 0 ; i < maxcpu && i < vm->def->cpumasklen ; i++) + if (vm->def->cpumask[i]) + VIR_USE_CPU(cpumap, i); + } else { + /* You may think this is redundant, but we can't assume libvirtd + * itself is running on all pCPUs, so we need to explicitly set + * the spawned QEMU instance to all pCPUs if no map is given in + * its config file */ + for (i = 0 ; i < maxcpu ; i++) + VIR_USE_CPU(cpumap, i); + } + + /* We are pressuming we are running between fork/exec of QEMU + * so use '0' to indicate our own process ID. No threads are + * running at this point + */ + if (virProcessInfoSetAffinity(0, /* Self */ + cpumap, cpumaplen, maxcpu) < 0) { + VIR_FREE(cpumap); + return -1; + } + VIR_FREE(cpumap); + + return 0; +} + + +static int +qemuProcessInitPasswords(virConnectPtr conn, + struct qemud_driver *driver, + virDomainObjPtr vm, + unsigned long long qemuCmdFlags) +{ + int ret = 0; + qemuDomainObjPrivatePtr priv = vm->privateData; + + if (vm->def->ngraphics == 1) { + if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) { + ret = qemuDomainChangeGraphicsPasswords(driver, vm, + VIR_DOMAIN_GRAPHICS_TYPE_VNC, + &vm->def->graphics[0]->data.vnc.auth, + driver->vncPassword); + } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) { + ret = qemuDomainChangeGraphicsPasswords(driver, vm, + VIR_DOMAIN_GRAPHICS_TYPE_SPICE, + &vm->def->graphics[0]->data.spice.auth, + driver->spicePassword); + } + } + + if (ret < 0) + goto cleanup; + + if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) { + int i; + + for (i = 0 ; i < vm->def->ndisks ; i++) { + char *secret; + size_t secretLen; + + if (!vm->def->disks[i]->encryption || + !vm->def->disks[i]->src) + continue; + + if (qemuProcessGetVolumeQcowPassphrase(conn, + vm->def->disks[i], + &secret, &secretLen) < 0) + goto cleanup; + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + ret = qemuMonitorSetDrivePassphrase(priv->mon, + vm->def->disks[i]->info.alias, + secret); + VIR_FREE(secret); + qemuDomainObjExitMonitorWithDriver(driver, vm); + if (ret < 0) + goto cleanup; + } + } + +cleanup: + return ret; +} + + +#define QEMU_PCI_VENDOR_INTEL 0x8086 +#define QEMU_PCI_VENDOR_LSI_LOGIC 0x1000 +#define QEMU_PCI_VENDOR_REDHAT 0x1af4 +#define QEMU_PCI_VENDOR_CIRRUS 0x1013 +#define QEMU_PCI_VENDOR_REALTEK 0x10ec +#define QEMU_PCI_VENDOR_AMD 0x1022 +#define QEMU_PCI_VENDOR_ENSONIQ 0x1274 +#define QEMU_PCI_VENDOR_VMWARE 0x15ad +#define QEMU_PCI_VENDOR_QEMU 0x1234 + +#define QEMU_PCI_PRODUCT_DISK_VIRTIO 0x1001 + +#define QEMU_PCI_PRODUCT_BALLOON_VIRTIO 0x1002 + +#define QEMU_PCI_PRODUCT_NIC_NE2K 0x8029 +#define QEMU_PCI_PRODUCT_NIC_PCNET 0x2000 +#define QEMU_PCI_PRODUCT_NIC_RTL8139 0x8139 +#define QEMU_PCI_PRODUCT_NIC_E1000 0x100E +#define QEMU_PCI_PRODUCT_NIC_VIRTIO 0x1000 + +#define QEMU_PCI_PRODUCT_VGA_CIRRUS 0x00b8 +#define QEMU_PCI_PRODUCT_VGA_VMWARE 0x0405 +#define QEMU_PCI_PRODUCT_VGA_STDVGA 0x1111 + +#define QEMU_PCI_PRODUCT_AUDIO_AC97 0x2415 +#define QEMU_PCI_PRODUCT_AUDIO_ES1370 0x5000 + +#define QEMU_PCI_PRODUCT_CONTROLLER_PIIX 0x7010 +#define QEMU_PCI_PRODUCT_CONTROLLER_LSI 0x0012 + +#define QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB 0x25ab + +static int +qemuProcessAssignNextPCIAddress(virDomainDeviceInfo *info, + int vendor, + int product, + qemuMonitorPCIAddress *addrs, + int naddrs) +{ + int found = 0; + int i; + + VIR_DEBUG("Look for %x:%x out of %d", vendor, product, naddrs); + + for (i = 0 ; (i < naddrs) && !found; i++) { + VIR_DEBUG("Maybe %x:%x", addrs[i].vendor, addrs[i].product); + if (addrs[i].vendor == vendor && + addrs[i].product == product) { + VIR_DEBUG("Match %d", i); + found = 1; + break; + } + } + if (!found) { + return -1; + } + + /* Blank it out so this device isn't matched again */ + addrs[i].vendor = 0; + addrs[i].product = 0; + + if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE) + info->type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI; + + if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) { + info->addr.pci.domain = addrs[i].addr.domain; + info->addr.pci.bus = addrs[i].addr.bus; + info->addr.pci.slot = addrs[i].addr.slot; + info->addr.pci.function = addrs[i].addr.function; + } + + return 0; +} + +static int +qemuProcessGetPCIDiskVendorProduct(virDomainDiskDefPtr def, + unsigned *vendor, + unsigned *product) +{ + switch (def->bus) { + case VIR_DOMAIN_DISK_BUS_VIRTIO: + *vendor = QEMU_PCI_VENDOR_REDHAT; + *product = QEMU_PCI_PRODUCT_DISK_VIRTIO; + break; + + default: + return -1; + } + + return 0; +} + +static int +qemuProcessGetPCINetVendorProduct(virDomainNetDefPtr def, + unsigned *vendor, + unsigned *product) +{ + if (!def->model) + return -1; + + if (STREQ(def->model, "ne2k_pci")) { + *vendor = QEMU_PCI_VENDOR_REALTEK; + *product = QEMU_PCI_PRODUCT_NIC_NE2K; + } else if (STREQ(def->model, "pcnet")) { + *vendor = QEMU_PCI_VENDOR_AMD; + *product = QEMU_PCI_PRODUCT_NIC_PCNET; + } else if (STREQ(def->model, "rtl8139")) { + *vendor = QEMU_PCI_VENDOR_REALTEK; + *product = QEMU_PCI_PRODUCT_NIC_RTL8139; + } else if (STREQ(def->model, "e1000")) { + *vendor = QEMU_PCI_VENDOR_INTEL; + *product = QEMU_PCI_PRODUCT_NIC_E1000; + } else if (STREQ(def->model, "virtio")) { + *vendor = QEMU_PCI_VENDOR_REDHAT; + *product = QEMU_PCI_PRODUCT_NIC_VIRTIO; + } else { + VIR_INFO("Unexpected NIC model %s, cannot get PCI address", + def->model); + return -1; + } + return 0; +} + +static int +qemuProcessGetPCIControllerVendorProduct(virDomainControllerDefPtr def, + unsigned *vendor, + unsigned *product) +{ + switch (def->type) { + case VIR_DOMAIN_CONTROLLER_TYPE_SCSI: + *vendor = QEMU_PCI_VENDOR_LSI_LOGIC; + *product = QEMU_PCI_PRODUCT_CONTROLLER_LSI; + break; + + case VIR_DOMAIN_CONTROLLER_TYPE_FDC: + /* XXX we could put in the ISA bridge address, but + that's not technically the FDC's address */ + return -1; + + case VIR_DOMAIN_CONTROLLER_TYPE_IDE: + *vendor = QEMU_PCI_VENDOR_INTEL; + *product = QEMU_PCI_PRODUCT_CONTROLLER_PIIX; + break; + + default: + VIR_INFO("Unexpected controller type %s, cannot get PCI address", + virDomainControllerTypeToString(def->type)); + return -1; + } + + return 0; +} + +static int +qemuProcessGetPCIVideoVendorProduct(virDomainVideoDefPtr def, + unsigned *vendor, + unsigned *product) +{ + switch (def->type) { + case VIR_DOMAIN_VIDEO_TYPE_CIRRUS: + *vendor = QEMU_PCI_VENDOR_CIRRUS; + *product = QEMU_PCI_PRODUCT_VGA_CIRRUS; + break; + + case VIR_DOMAIN_VIDEO_TYPE_VGA: + *vendor = QEMU_PCI_VENDOR_QEMU; + *product = QEMU_PCI_PRODUCT_VGA_STDVGA; + break; + + case VIR_DOMAIN_VIDEO_TYPE_VMVGA: + *vendor = QEMU_PCI_VENDOR_VMWARE; + *product = QEMU_PCI_PRODUCT_VGA_VMWARE; + break; + + default: + return -1; + } + return 0; +} + +static int +qemuProcessGetPCISoundVendorProduct(virDomainSoundDefPtr def, + unsigned *vendor, + unsigned *product) +{ + switch (def->model) { + case VIR_DOMAIN_SOUND_MODEL_ES1370: + *vendor = QEMU_PCI_VENDOR_ENSONIQ; + *product = QEMU_PCI_PRODUCT_AUDIO_ES1370; + break; + + case VIR_DOMAIN_SOUND_MODEL_AC97: + *vendor = QEMU_PCI_VENDOR_INTEL; + *product = QEMU_PCI_PRODUCT_AUDIO_AC97; + break; + + default: + return -1; + } + + return 0; +} + +static int +qemuProcessGetPCIWatchdogVendorProduct(virDomainWatchdogDefPtr def, + unsigned *vendor, + unsigned *product) +{ + switch (def->model) { + case VIR_DOMAIN_WATCHDOG_MODEL_I6300ESB: + *vendor = QEMU_PCI_VENDOR_INTEL; + *product = QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB; + break; + + default: + return -1; + } + + return 0; +} + + +static int +qemuProcessGetPCIMemballoonVendorProduct(virDomainMemballoonDefPtr def, + unsigned *vendor, + unsigned *product) +{ + switch (def->model) { + case VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO: + *vendor = QEMU_PCI_VENDOR_REDHAT; + *product = QEMU_PCI_PRODUCT_BALLOON_VIRTIO; + break; + + default: + return -1; + } + + return 0; +} + + +/* + * This entire method assumes that PCI devices in 'info pci' + * match ordering of devices specified on the command line + * wrt to devices of matching vendor+product + * + * XXXX this might not be a valid assumption if we assign + * some static addrs on CLI. Have to check that... + */ +static int +qemuProcessDetectPCIAddresses(virDomainObjPtr vm, + qemuMonitorPCIAddress *addrs, + int naddrs) +{ + unsigned int vendor = 0, product = 0; + int i; + + /* XXX should all these vendor/product IDs be kept in the + * actual device data structure instead ? + */ + + for (i = 0 ; i < vm->def->ndisks ; i++) { + if (qemuProcessGetPCIDiskVendorProduct(vm->def->disks[i], &vendor, &product) < 0) + continue; + + if (qemuProcessAssignNextPCIAddress(&(vm->def->disks[i]->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for VirtIO disk %s"), + vm->def->disks[i]->dst); + return -1; + } + } + + for (i = 0 ; i < vm->def->nnets ; i++) { + if (qemuProcessGetPCINetVendorProduct(vm->def->nets[i], &vendor, &product) < 0) + continue; + + if (qemuProcessAssignNextPCIAddress(&(vm->def->nets[i]->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for %s NIC"), + vm->def->nets[i]->model); + return -1; + } + } + + for (i = 0 ; i < vm->def->ncontrollers ; i++) { + if (qemuProcessGetPCIControllerVendorProduct(vm->def->controllers[i], &vendor, &product) < 0) + continue; + + if (qemuProcessAssignNextPCIAddress(&(vm->def->controllers[i]->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for controller %s"), + virDomainControllerTypeToString(vm->def->controllers[i]->type)); + return -1; + } + } + + for (i = 0 ; i < vm->def->nvideos ; i++) { + if (qemuProcessGetPCIVideoVendorProduct(vm->def->videos[i], &vendor, &product) < 0) + continue; + + if (qemuProcessAssignNextPCIAddress(&(vm->def->videos[i]->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for video adapter %s"), + virDomainVideoTypeToString(vm->def->videos[i]->type)); + return -1; + } + } + + for (i = 0 ; i < vm->def->nsounds ; i++) { + if (qemuProcessGetPCISoundVendorProduct(vm->def->sounds[i], &vendor, &product) < 0) + continue; + + if (qemuProcessAssignNextPCIAddress(&(vm->def->sounds[i]->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for sound adapter %s"), + virDomainSoundModelTypeToString(vm->def->sounds[i]->model)); + return -1; + } + } + + + if (vm->def->watchdog && + qemuProcessGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) { + if (qemuProcessAssignNextPCIAddress(&(vm->def->watchdog->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for watchdog %s"), + virDomainWatchdogModelTypeToString(vm->def->watchdog->model)); + return -1; + } + } + + if (vm->def->memballoon && + qemuProcessGetPCIMemballoonVendorProduct(vm->def->memballoon, &vendor, &product) == 0) { + if (qemuProcessAssignNextPCIAddress(&(vm->def->memballoon->info), + vendor, product, + addrs, naddrs) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("cannot find PCI address for balloon %s"), + virDomainMemballoonModelTypeToString(vm->def->memballoon->model)); + return -1; + } + } + + /* XXX console (virtio) */ + + + /* ... and now things we don't have in our xml */ + + /* XXX USB controller ? */ + + /* XXX what about other PCI devices (ie bridges) */ + + return 0; +} + +static int +qemuProcessInitPCIAddresses(struct qemud_driver *driver, + virDomainObjPtr vm) +{ + qemuDomainObjPrivatePtr priv = vm->privateData; + int naddrs; + int ret; + qemuMonitorPCIAddress *addrs = NULL; + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + naddrs = qemuMonitorGetAllPCIAddresses(priv->mon, + &addrs); + qemuDomainObjExitMonitorWithDriver(driver, vm); + + ret = qemuProcessDetectPCIAddresses(vm, addrs, naddrs); + + VIR_FREE(addrs); + + return ret; +} + + +static int qemuProcessNextFreePort(struct qemud_driver *driver, + int startPort) +{ + int i; + + for (i = startPort ; i < QEMU_VNC_PORT_MAX; i++) { + int fd; + int reuse = 1; + struct sockaddr_in addr; + bool used = false; + + if (virBitmapGetBit(driver->reservedVNCPorts, + i - QEMU_VNC_PORT_MIN, &used) < 0) + VIR_DEBUG("virBitmapGetBit failed on bit %d", i - QEMU_VNC_PORT_MIN); + + if (used) + continue; + + addr.sin_family = AF_INET; + addr.sin_port = htons(i); + addr.sin_addr.s_addr = htonl(INADDR_ANY); + fd = socket(PF_INET, SOCK_STREAM, 0); + if (fd < 0) + return -1; + + if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (void*)&reuse, sizeof(reuse)) < 0) { + VIR_FORCE_CLOSE(fd); + break; + } + + if (bind(fd, (struct sockaddr*)&addr, sizeof(addr)) == 0) { + /* Not in use, lets grab it */ + VIR_FORCE_CLOSE(fd); + /* Add port to bitmap of reserved ports */ + if (virBitmapSetBit(driver->reservedVNCPorts, + i - QEMU_VNC_PORT_MIN) < 0) { + VIR_DEBUG("virBitmapSetBit failed on bit %d", + i - QEMU_VNC_PORT_MIN); + } + return i; + } + VIR_FORCE_CLOSE(fd); + + if (errno == EADDRINUSE) { + /* In use, try next */ + continue; + } + /* Some other bad failure, get out.. */ + break; + } + return -1; +} + + +static void +qemuProcessReturnPort(struct qemud_driver *driver, + int port) +{ + if (port < QEMU_VNC_PORT_MIN) + return; + + if (virBitmapClearBit(driver->reservedVNCPorts, + port - QEMU_VNC_PORT_MIN) < 0) + VIR_DEBUG("Could not mark port %d as unused", port); +} + + +static int +qemuProcessPrepareChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED, + virDomainChrDefPtr dev, + void *opaque ATTRIBUTE_UNUSED) +{ + int fd; + if (dev->source.type != VIR_DOMAIN_CHR_TYPE_FILE) + return 0; + + if ((fd = open(dev->source.data.file.path, + O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) { + virReportSystemError(errno, + _("Unable to pre-create chardev file '%s'"), + dev->source.data.file.path); + return -1; + } + + VIR_FORCE_CLOSE(fd); + + return 0; +} + + +struct qemuProcessHookData { + virConnectPtr conn; + virDomainObjPtr vm; + struct qemud_driver *driver; +}; + +static int qemuProcessHook(void *data) +{ + struct qemuProcessHookData *h = data; + + /* This must take place before exec(), so that all QEMU + * memory allocation is on the correct NUMA node + */ + if (qemuAddToCgroup(h->driver, h->vm->def) < 0) + return -1; + + /* This must be done after cgroup placement to avoid resetting CPU + * affinity */ + if (qemuProcessInitCpuAffinity(h->vm) < 0) + return -1; + + if (virSecurityManagerSetProcessLabel(h->driver->securityManager, h->vm) < 0) + return -1; + + return 0; +} + + +int +qemuProcessPrepareMonitorChr(struct qemud_driver *driver, + virDomainChrSourceDefPtr monConfig, + const char *vm) +{ + monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX; + monConfig->data.nix.listen = true; + + if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor", + driver->libDir, vm) < 0) { + virReportOOMError(); + return -1; + } + + return 0; +} + + +int qemuProcessStartCPUs(struct qemud_driver *driver, virDomainObjPtr vm, virConnectPtr conn) +{ + int ret; + qemuDomainObjPrivatePtr priv = vm->privateData; + + qemuDomainObjEnterMonitorWithDriver(driver, vm); + ret = qemuMonitorStartCPUs(priv->mon, conn); + if (ret == 0) { + vm->state = VIR_DOMAIN_RUNNING; + } + qemuDomainObjExitMonitorWithDriver(driver, vm); + + return ret; +} + + +int qemuProcessStopCPUs(struct qemud_driver *driver, virDomainObjPtr vm) +{ + int ret; + int oldState = vm->state; + qemuDomainObjPrivatePtr priv = vm->privateData; + + vm->state = VIR_DOMAIN_PAUSED; + qemuDomainObjEnterMonitorWithDriver(driver, vm); + ret = qemuMonitorStopCPUs(priv->mon); + qemuDomainObjExitMonitorWithDriver(driver, vm); + if (ret < 0) { + vm->state = oldState; + } + return ret; +} + + + +static int +qemuProcessFiltersInstantiate(virConnectPtr conn, + virDomainDefPtr def) +{ + int err = 0; + int i; + + if (!conn) + return 1; + + for (i = 0 ; i < def->nnets ; i++) { + virDomainNetDefPtr net = def->nets[i]; + if ((net->filter) && (net->ifname)) { + if (virDomainConfNWFilterInstantiate(conn, net)) { + err = 1; + break; + } + } + } + + return err; +} + +struct qemuProcessReconnectData { + virConnectPtr conn; + struct qemud_driver *driver; +}; +/* + * Open an existing VM's monitor, re-detect VCPU threads + * and re-reserve the security labels in use + */ +static void +qemuProcessReconnect(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque) +{ + virDomainObjPtr obj = payload; + struct qemuProcessReconnectData *data = opaque; + struct qemud_driver *driver = data->driver; + qemuDomainObjPrivatePtr priv; + unsigned long long qemuCmdFlags; + virConnectPtr conn = data->conn; + + virDomainObjLock(obj); + + VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name); + + priv = obj->privateData; + + /* Hold an extra reference because we can't allow 'vm' to be + * deleted if qemuConnectMonitor() failed */ + virDomainObjRef(obj); + + /* XXX check PID liveliness & EXE path */ + if (qemuConnectMonitor(driver, obj) < 0) + goto error; + + if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) { + goto error; + } + + /* XXX we should be persisting the original flags in the XML + * not re-detecting them, since the binary may have changed + * since launch time */ + if (qemuCapsExtractVersionInfo(obj->def->emulator, + NULL, + &qemuCmdFlags) >= 0 && + (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) { + priv->persistentAddrs = 1; + + if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)) || + qemuAssignDevicePCISlots(obj->def, priv->pciaddrs) < 0) + goto error; + } + + if (virSecurityManagerReserveLabel(driver->securityManager, obj) < 0) + goto error; + + if (qemuProcessFiltersInstantiate(conn, obj->def)) + goto error; + + if (obj->def->id >= driver->nextvmid) + driver->nextvmid = obj->def->id + 1; + + if (virDomainObjUnref(obj) > 0) + virDomainObjUnlock(obj); + return; + +error: + if (!virDomainObjIsActive(obj)) { + if (virDomainObjUnref(obj) > 0) + virDomainObjUnlock(obj); + return; + } + + if (virDomainObjUnref(obj) > 0) { + /* We can't get the monitor back, so must kill the VM + * to remove danger of it ending up running twice if + * user tries to start it again later */ + qemudShutdownVMDaemon(driver, obj, 0); + if (!obj->persistent) + virDomainRemoveInactive(&driver->domains, obj); + else + virDomainObjUnlock(obj); + } +} + +/** + * qemuProcessReconnectAll + * + * Try to re-open the resources for live VMs that we care + * about. + */ +void +qemuProcessReconnectAll(virConnectPtr conn, struct qemud_driver *driver) +{ + struct qemuProcessReconnectData data = {conn, driver}; + virHashForEach(driver->domains.objs, qemuProcessReconnect, &data); +} + +int qemuProcessStart(virConnectPtr conn, + struct qemud_driver *driver, + virDomainObjPtr vm, + const char *migrateFrom, + bool start_paused, + int stdin_fd, + const char *stdin_path, + enum virVMOperationType vmop) +{ + int ret; + unsigned long long qemuCmdFlags; + off_t pos = -1; + char ebuf[1024]; + char *pidfile = NULL; + int logfile = -1; + char *timestamp; + qemuDomainObjPrivatePtr priv = vm->privateData; + virCommandPtr cmd = NULL; + struct qemuProcessHookData hookData; + + hookData.conn = conn; + hookData.vm = vm; + hookData.driver = driver; + + DEBUG0("Beginning VM startup process"); + + if (virDomainObjIsActive(vm)) { + qemuReportError(VIR_ERR_OPERATION_INVALID, + "%s", _("VM is already active")); + return -1; + } + + /* Do this upfront, so any part of the startup process can add + * runtime state to vm->def that won't be persisted. This let's us + * report implicit runtime defaults in the XML, like vnc listen/socket + */ + DEBUG0("Setting current domain def as transient"); + if (virDomainObjSetDefTransient(driver->caps, vm, true) < 0) + goto cleanup; + + /* Must be run before security labelling */ + DEBUG0("Preparing host devices"); + if (qemuPrepareHostDevices(driver, vm->def) < 0) + goto cleanup; + + DEBUG0("Preparing chr devices"); + if (virDomainChrDefForeach(vm->def, + true, + qemuProcessPrepareChardevDevice, + NULL) < 0) + goto cleanup; + + /* If you are using a SecurityDriver with dynamic labelling, + then generate a security label for isolation */ + DEBUG0("Generating domain security label (if required)"); + if (virSecurityManagerGenLabel(driver->securityManager, vm) < 0) { + qemuDomainSecurityLabelAudit(vm, false); + goto cleanup; + } + qemuDomainSecurityLabelAudit(vm, true); + + DEBUG0("Generating setting domain security labels (if required)"); + if (virSecurityManagerSetAllLabel(driver->securityManager, + vm, stdin_path) < 0) + goto cleanup; + + if (stdin_fd != -1) { + /* if there's an fd to migrate from, and it's a pipe, put the + * proper security label on it + */ + struct stat stdin_sb; + + DEBUG0("setting security label on pipe used for migration"); + + if (fstat(stdin_fd, &stdin_sb) < 0) { + virReportSystemError(errno, + _("cannot stat fd %d"), stdin_fd); + goto cleanup; + } + if (S_ISFIFO(stdin_sb.st_mode) && + virSecurityManagerSetFDLabel(driver->securityManager, vm, stdin_fd) < 0) + goto cleanup; + } + + /* Ensure no historical cgroup for this VM is lying around bogus + * settings */ + DEBUG0("Ensuring no historical cgroup is lying around"); + qemuRemoveCgroup(driver, vm, 1); + + if (vm->def->ngraphics == 1) { + if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC && + !vm->def->graphics[0]->data.vnc.socket && + vm->def->graphics[0]->data.vnc.autoport) { + int port = qemuProcessNextFreePort(driver, QEMU_VNC_PORT_MIN); + if (port < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("Unable to find an unused VNC port")); + goto cleanup; + } + vm->def->graphics[0]->data.vnc.port = port; + } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE && + vm->def->graphics[0]->data.spice.autoport) { + int port = qemuProcessNextFreePort(driver, QEMU_VNC_PORT_MIN); + int tlsPort = -1; + if (port < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("Unable to find an unused SPICE port")); + goto cleanup; + } + + if (driver->spiceTLS) { + tlsPort = qemuProcessNextFreePort(driver, port + 1); + if (tlsPort < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("Unable to find an unused SPICE TLS port")); + qemuProcessReturnPort(driver, port); + goto cleanup; + } + } + + vm->def->graphics[0]->data.spice.port = port; + vm->def->graphics[0]->data.spice.tlsPort = tlsPort; + } + } + + if (virFileMakePath(driver->logDir) != 0) { + virReportSystemError(errno, + _("cannot create log directory %s"), + driver->logDir); + goto cleanup; + } + + DEBUG0("Creating domain log file"); + if ((logfile = qemuProcessLogFD(driver, vm->def->name, false)) < 0) + goto cleanup; + + DEBUG0("Determining emulator version"); + if (qemuCapsExtractVersionInfo(vm->def->emulator, + NULL, + &qemuCmdFlags) < 0) + goto cleanup; + + DEBUG0("Setting up domain cgroup (if required)"); + if (qemuSetupCgroup(driver, vm) < 0) + goto cleanup; + + if (VIR_ALLOC(priv->monConfig) < 0) { + virReportOOMError(); + goto cleanup; + } + + DEBUG0("Preparing monitor state"); + if (qemuProcessPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0) + goto cleanup; + +#if HAVE_YAJL + if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON) + priv->monJSON = 1; + else +#endif + priv->monJSON = 0; + + priv->monitor_warned = 0; + priv->gotShutdown = false; + + if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) { + virReportSystemError(ret, + _("Cannot remove stale PID file for %s"), + vm->def->name); + goto cleanup; + } + + if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) { + virReportSystemError(errno, + "%s", _("Failed to build pidfile path.")); + goto cleanup; + } + + /* + * Normally PCI addresses are assigned in the virDomainCreate + * or virDomainDefine methods. We might still need to assign + * some here to cope with the question of upgrades. Regardless + * we also need to populate the PCi address set cache for later + * use in hotplug + */ + if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) { + DEBUG0("Assigning domain PCI addresses"); + /* Populate cache with current addresses */ + if (priv->pciaddrs) { + qemuDomainPCIAddressSetFree(priv->pciaddrs); + priv->pciaddrs = NULL; + } + if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def))) + goto cleanup; + + + /* Assign any remaining addresses */ + if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0) + goto cleanup; + + priv->persistentAddrs = 1; + } else { + priv->persistentAddrs = 0; + } + + DEBUG0("Building emulator command line"); + vm->def->id = driver->nextvmid++; + if (!(cmd = qemuBuildCommandLine(conn, driver, vm->def, priv->monConfig, + priv->monJSON != 0, qemuCmdFlags, + migrateFrom, stdin_fd, + vm->current_snapshot, vmop))) + goto cleanup; + +#if 0 + /* XXX */ + if (qemuDomainSnapshotSetCurrentInactive(vm, driver->snapshotDir) < 0) + goto cleanup; +#endif + + /* now that we know it is about to start call the hook if present */ + if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) { + char *xml = virDomainDefFormat(vm->def, 0); + int hookret; + + hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name, + VIR_HOOK_QEMU_OP_START, VIR_HOOK_SUBOP_BEGIN, NULL, xml); + VIR_FREE(xml); + + /* + * If the script raised an error abort the launch + */ + if (hookret < 0) + goto cleanup; + } + + if ((timestamp = virTimestamp()) == NULL) { + virReportOOMError(); + goto cleanup; + } else { + if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 || + safewrite(logfile, START_POSTFIX, strlen(START_POSTFIX)) < 0) { + VIR_WARN("Unable to write timestamp to logfile: %s", + virStrerror(errno, ebuf, sizeof ebuf)); + } + + VIR_FREE(timestamp); + } + + virCommandWriteArgLog(cmd, logfile); + + if ((pos = lseek(logfile, 0, SEEK_END)) < 0) + VIR_WARN("Unable to seek to end of logfile: %s", + virStrerror(errno, ebuf, sizeof ebuf)); + + VIR_DEBUG("Clear emulator capabilities: %d", + driver->clearEmulatorCapabilities); + if (driver->clearEmulatorCapabilities) + virCommandClearCaps(cmd); + + virCommandSetPreExecHook(cmd, qemuProcessHook, &hookData); + + virCommandSetOutputFD(cmd, &logfile); + virCommandSetErrorFD(cmd, &logfile); + virCommandNonblockingFDs(cmd); + virCommandSetPidFile(cmd, pidfile); + virCommandDaemonize(cmd); + + ret = virCommandRun(cmd, NULL); + VIR_FREE(pidfile); + + /* wait for qemu process to to show up */ + if (ret == 0) { + if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + _("Domain %s didn't show up"), vm->def->name); + ret = -1; + } +#if 0 + } else if (ret == -2) { + /* + * XXX this is bogus. It isn't safe to set vm->pid = child + * because the child no longer exists. + */ + + /* The virExec process that launches the daemon failed. Pending on + * when it failed (we can't determine for sure), there may be + * extra info in the domain log (if the hook failed for example). + * + * Pretend like things succeeded, and let 'WaitForMonitor' report + * the log contents for us. + */ + vm->pid = child; + ret = 0; +#endif + } + + if (migrateFrom) + start_paused = true; + vm->state = start_paused ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING; + + if (ret == -1) /* The VM failed to start; tear filters before taps */ + virDomainConfVMNWFilterTeardown(vm); + + if (ret == -1) /* The VM failed to start */ + goto cleanup; + + DEBUG0("Waiting for monitor to show up"); + if (qemuProcessWaitForMonitor(driver, vm, pos) < 0) + goto cleanup; + + DEBUG0("Detecting VCPU PIDs"); + if (qemuProcessDetectVcpuPIDs(driver, vm) < 0) + goto cleanup; + + DEBUG0("Setting any required VM passwords"); + if (qemuProcessInitPasswords(conn, driver, vm, qemuCmdFlags) < 0) + goto cleanup; + + /* If we have -device, then addresses are assigned explicitly. + * If not, then we have to detect dynamic ones here */ + if (!(qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) { + DEBUG0("Determining domain device PCI addresses"); + if (qemuProcessInitPCIAddresses(driver, vm) < 0) + goto cleanup; + } + + DEBUG0("Setting initial memory amount"); + qemuDomainObjEnterMonitorWithDriver(driver, vm); + if (qemuMonitorSetBalloon(priv->mon, vm->def->mem.cur_balloon) < 0) { + qemuDomainObjExitMonitorWithDriver(driver, vm); + goto cleanup; + } + qemuDomainObjExitMonitorWithDriver(driver, vm); + + if (!start_paused) { + DEBUG0("Starting domain CPUs"); + /* Allow the CPUS to start executing */ + if (qemuProcessStartCPUs(driver, vm, conn) < 0) { + if (virGetLastError() == NULL) + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("resume operation failed")); + goto cleanup; + } + } + + + DEBUG0("Writing domain status to disk"); + if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) + goto cleanup; + + virCommandFree(cmd); + VIR_FORCE_CLOSE(logfile); + + return 0; + +cleanup: + /* We jump here if we failed to start the VM for any reason, or + * if we failed to initialize the now running VM. kill it off and + * pretend we never started it */ + virCommandFree(cmd); + VIR_FORCE_CLOSE(logfile); + qemuProcessStop(driver, vm, 0); + + return -1; +} + + +void qemuProcessStop(struct qemud_driver *driver, + virDomainObjPtr vm, + int migrated) +{ + int ret; + int retries = 0; + qemuDomainObjPrivatePtr priv = vm->privateData; + virErrorPtr orig_err; + virDomainDefPtr def; + int i; + int logfile = -1; + char *timestamp; + char ebuf[1024]; + + VIR_DEBUG("Shutting down VM '%s' pid=%d migrated=%d", + vm->def->name, vm->pid, migrated); + + if ((logfile = qemuProcessLogFD(driver, vm->def->name, true)) < 0) { + /* To not break the normal domain shutdown process, skip the + * timestamp log writing if failed on opening log file. */ + VIR_WARN("Unable to open logfile: %s", + virStrerror(errno, ebuf, sizeof ebuf)); + } else { + if ((timestamp = virTimestamp()) == NULL) { + virReportOOMError(); + } else { + if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 || + safewrite(logfile, SHUTDOWN_POSTFIX, + strlen(SHUTDOWN_POSTFIX)) < 0) { + VIR_WARN("Unable to write timestamp to logfile: %s", + virStrerror(errno, ebuf, sizeof ebuf)); + } + + VIR_FREE(timestamp); + } + + if (VIR_CLOSE(logfile) < 0) + VIR_WARN("Unable to close logfile: %s", + virStrerror(errno, ebuf, sizeof ebuf)); + } + + /* This method is routinely used in clean up paths. Disable error + * reporting so we don't squash a legit error. */ + orig_err = virSaveLastError(); + + virDomainConfVMNWFilterTeardown(vm); + + if (driver->macFilter) { + def = vm->def; + for (i = 0 ; i < def->nnets ; i++) { + virDomainNetDefPtr net = def->nets[i]; + if (net->ifname == NULL) + continue; + if ((errno = networkDisallowMacOnPort(driver, net->ifname, + net->mac))) { + virReportSystemError(errno, + _("failed to remove ebtables rule to allow MAC address on '%s'"), + net->ifname); + } + } + } + + /* This will safely handle a non-running guest with pid=0 or pid=-1*/ + if (virKillProcess(vm->pid, 0) == 0 && + virKillProcess(vm->pid, SIGTERM) < 0) + virReportSystemError(errno, + _("Failed to send SIGTERM to %s (%d)"), + vm->def->name, vm->pid); + + if (priv->mon) + qemuMonitorClose(priv->mon); + + if (priv->monConfig) { + if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX) + unlink(priv->monConfig->data.nix.path); + virDomainChrSourceDefFree(priv->monConfig); + priv->monConfig = NULL; + } + + /* shut it off for sure */ + virKillProcess(vm->pid, SIGKILL); + + /* now that we know it's stopped call the hook if present */ + if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) { + char *xml = virDomainDefFormat(vm->def, 0); + + /* we can't stop the operation even if the script raised an error */ + virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name, + VIR_HOOK_QEMU_OP_STOPPED, VIR_HOOK_SUBOP_END, NULL, xml); + VIR_FREE(xml); + } + + /* Reset Security Labels */ + virSecurityManagerRestoreAllLabel(driver->securityManager, + vm, migrated); + virSecurityManagerReleaseLabel(driver->securityManager, vm); + + /* Clear out dynamically assigned labels */ + if (vm->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC) { + VIR_FREE(vm->def->seclabel.model); + VIR_FREE(vm->def->seclabel.label); + VIR_FREE(vm->def->seclabel.imagelabel); + } + + virDomainDefClearDeviceAliases(vm->def); + if (!priv->persistentAddrs) { + virDomainDefClearPCIAddresses(vm->def); + qemuDomainPCIAddressSetFree(priv->pciaddrs); + priv->pciaddrs = NULL; + } + + qemuDomainReAttachHostDevices(driver, vm->def); + +#if WITH_MACVTAP + def = vm->def; + for (i = 0; i < def->nnets; i++) { + virDomainNetDefPtr net = def->nets[i]; + if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { + delMacvtap(net->ifname, net->mac, net->data.direct.linkdev, + &net->data.direct.virtPortProfile); + VIR_FREE(net->ifname); + } + } +#endif + +retry: + if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) { + if (ret == -EBUSY && (retries++ < 5)) { + usleep(200*1000); + goto retry; + } + VIR_WARN("Failed to remove cgroup for %s", + vm->def->name); + } + + qemuProcessRemoveDomainStatus(driver, vm); + + /* Remove VNC port from port reservation bitmap, but only if it was + reserved by the driver (autoport=yes) + */ + if ((vm->def->ngraphics == 1) && + vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC && + vm->def->graphics[0]->data.vnc.autoport) { + qemuProcessReturnPort(driver, vm->def->graphics[0]->data.vnc.port); + } + if ((vm->def->ngraphics == 1) && + vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE && + vm->def->graphics[0]->data.spice.autoport) { + qemuProcessReturnPort(driver, vm->def->graphics[0]->data.spice.port); + qemuProcessReturnPort(driver, vm->def->graphics[0]->data.spice.tlsPort); + } + + vm->pid = -1; + vm->def->id = -1; + vm->state = VIR_DOMAIN_SHUTOFF; + VIR_FREE(priv->vcpupids); + priv->nvcpupids = 0; + + if (vm->newDef) { + virDomainDefFree(vm->def); + vm->def = vm->newDef; + vm->def->id = -1; + vm->newDef = NULL; + } + + if (orig_err) { + virSetError(orig_err); + virFreeError(orig_err); + } +} + diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h new file mode 100644 index 0000000..4d803cc --- /dev/null +++ b/src/qemu/qemu_process.h @@ -0,0 +1,52 @@ +/* + * qemu_process.c: QEMU process management + * + * Copyright (C) 2006-2011 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 + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#ifndef __QEMU_PROCESS_H__ +# define __QEMU_PROCESS_H__ + +#include "qemu_conf.h" + +int qemuProcessPrepareMonitorChr(struct qemud_driver *driver, + virDomainChrSourceDefPtr monConfig, + const char *vm); + +int qemuProcessStartCPUs(struct qemud_driver *driver, virDomainObjPtr vm, virConnectPtr conn); +int qemuProcessStopCPUs(struct qemud_driver *driver, virDomainObjPtr vm); + +void qemuProcessAutostartAll(struct qemud_driver *driver); +void qemuProcessReconnectAll(virConnectPtr conn, struct qemud_driver *driver); + +int qemuProcessAssignPCIAddresses(virDomainDefPtr def); + +int qemuProcessStart(virConnectPtr conn, + struct qemud_driver *driver, + virDomainObjPtr vm, + const char *migrateFrom, + bool start_paused, + int stdin_fd, + const char *stdin_path, + enum virVMOperationType vmop); + +void qemuProcessStop(struct qemud_driver *driver, + virDomainObjPtr vm, + int migrated); + +#endif /* __QEMU_PROCESS_H__ */ -- 1.7.3.4

13 years, 3 months

2
3
0 / 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Devel January 2011