- Devel - libvirt List Archives

[libvirt] virDomainMemoryStats call
by Parakkal, Navin S 18 Jul '12

18 Jul '12

Hi, I'm using the API virDomainMemoryStats to fetch memory stats for a domain/guest on Fedora 16/RHEL 6.3 . I don't get these counters filled up VIR_DOMAIN_MEMORY_STAT_UNUSED, VIR_DOMAIN_MEMORY_STAT_ACTUAL_BALLOON,VIR_DOMAIN_MEMORY_STAT_AVAILABLE , VIR_DOMAIN_MEMORY_STAT_MAJOR_FAULT and VIR_DOMAIN_MEMORY_STAT_MINOR_FAULT. The version of libvirt 0.9.11 on Fedora 16 . How to get all the counters filled up ? Should I take a specific version of qemu or should I do some config settings so that these values are filled up ? P.S. Please copy me in reply Regards, Navin Parakkal HP Software

3 3

[libvirt] [PATCH] storage: netfs and iscsi need option srcSpec for resource discovery
by Guannan Ren 18 Jul '12

18 Jul '12

The option 'srcSpec' to virsh command find-storage-pool-sources is optional for logical type of storage pool, but mandatory for netfs and iscsi type. When missing the option for netfs and iscsi, libvirt reports XML parsing error due to null string option srcSpec. error: Failed to find any netfs pool sources error: (storage_source_specification):1: Document is empty (null) This patch adds a check for it, error info changed to: error: Failed to find any netfs pool sources error: internal error pool type 'netfs' need option \ --srcSpec for source discovery --- src/storage/storage_driver.c | 8 ++++++++ 1 files changed, 8 insertions(+), 0 deletions(-) diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c index fbc630d..e5f3e1b 100644 --- a/src/storage/storage_driver.c +++ b/src/storage/storage_driver.c @@ -458,6 +458,14 @@ storageFindPoolSources(virConnectPtr conn, goto cleanup; } + if ((*srcSpec == '\0') && (backend_type == VIR_STORAGE_POOL_ISCSI || + backend_type == VIR_STORAGE_POOL_NETFS)) { + virStorageReportError(VIR_ERR_INTERNAL_ERROR, + _("pool type '%s' need option --srcSpec " + "for source discovery"), type); + goto cleanup; + } + ret = backend->findPoolSources(conn, srcSpec, flags); cleanup: -- 1.7.7.5

2 2

[libvirt] [glib PATCH V2] Add bindings for virStorageDownload() and virStorageUpload()
by Jovanka Gulicoska 18 Jul '12

18 Jul '12

--- libvirt-gobject/libvirt-gobject-storage-vol.c | 85 +++++++++++++++++++++++++ libvirt-gobject/libvirt-gobject-storage-vol.h | 14 ++++ libvirt-gobject/libvirt-gobject.h | 1 + libvirt-gobject/libvirt-gobject.sym | 2 + 4 files changed, 102 insertions(+) diff --git a/libvirt-gobject/libvirt-gobject-storage-vol.c b/libvirt-gobject/libvirt-gobject-storage-vol.c index 6f60fcd..462a99e 100644 --- a/libvirt-gobject/libvirt-gobject-storage-vol.c +++ b/libvirt-gobject/libvirt-gobject-storage-vol.c @@ -349,3 +349,88 @@ gboolean gvir_storage_vol_resize(GVirStorageVol *vol, return TRUE; } + +/** + * gvir_storage_vol_download: + * @vol: the storage volume to download from + * @stream: stream to use as output + * @offset: position in @vol to start reading from + * @length: limit on amount of data to download + * @flags: extra flags, not used yet, pass 0 + * + * Returns: #TRUE of success, #FALSE otherwise + */ +gboolean gvir_storage_vol_download(GVirStorageVol *vol, + GVirStream *stream, + unsigned long long offset, + unsigned long long length, + guint flags, + GError **err) +{ + virStreamPtr st = NULL; + gboolean ret = FALSE; + + g_object_get(stream, "handle", &st, NULL); + + g_return_val_if_fail(GVIR_IS_STORAGE_VOL(vol), FALSE); + g_return_val_if_fail(GVIR_IS_STREAM(stream), FALSE); + g_return_val_if_fail(err == NULL || *err == NULL, FALSE); + + if (virStorageVolDownload(vol->priv->handle, st, offset, length, 0) < 0) { + gvir_set_error_literal(err, + GVIR_STORAGE_VOL_ERROR, + 0, + "Unable to downlaod volume storage"); + + goto cleanup; + } + + ret = TRUE; +cleanup: + if (st != NULL) + virStreamFree(st); + return ret; +} + +/** + * gvir_storage_vol_upload: + * @vol: the storage volume to upload + * @stream: stream to use as input + * @offset: position in @vol to start to write to + * @length: limit on amount of data to upload + * @flags: the flags, not set yet, pass 0 + * + * Returns: #TRUE of success, #FALSE otherwise + */ +gboolean gvir_storage_vol_upload(GVirStorageVol *vol, + GVirStream *stream, + unsigned long long offset, + unsigned long long length, + guint flags, + GError **err) +{ + virStreamPtr st = NULL; + gboolean ret = FALSE; + + g_object_get(stream, "handle", &st, NULL); + + g_return_val_if_fail(GVIR_IS_STORAGE_VOL(vol), FALSE); + g_return_val_if_fail(GVIR_IS_STREAM(stream), FALSE); + g_return_val_if_fail(err == NULL || *err == NULL, FALSE); + + if (virStorageVolUpload(vol->priv->handle, st, offset, length, 0) < 0) { + gvir_set_error_literal(err, + GVIR_STORAGE_VOL_ERROR, + 0, + "Unable to upload to stream"); + + goto cleanup; + } + + ret = TRUE; +cleanup: + if (st != NULL) + virStreamFree(st); + return ret; +} + diff --git a/libvirt-gobject/libvirt-gobject-storage-vol.h b/libvirt-gobject/libvirt-gobject-storage-vol.h index b425f0a..e156792 100644 --- a/libvirt-gobject/libvirt-gobject-storage-vol.h +++ b/libvirt-gobject/libvirt-gobject-storage-vol.h @@ -110,6 +110,20 @@ gboolean gvir_storage_vol_resize(GVirStorageVol *vol, guint flags, GError **err); +gboolean gvir_storage_vol_download(GVirStorageVol *vol, + GVirStream *stream, + unsigned long long offset, + unsigned long long length, + guint flags, + GError **err); + +gboolean gvir_storage_vol_upload(GVirStorageVol *vol, + GVirStream *stream, + unsigned long long offset, + unsigned long long length, + guint flags, + GError **err); + G_END_DECLS #endif /* __LIBVIRT_GOBJECT_STORAGE_VOL_H__ */ diff --git a/libvirt-gobject/libvirt-gobject.h b/libvirt-gobject/libvirt-gobject.h index f52cc00..b1158f7 100644 --- a/libvirt-gobject/libvirt-gobject.h +++ b/libvirt-gobject/libvirt-gobject.h @@ -44,5 +44,6 @@ #include <libvirt-gobject/libvirt-gobject-storage-pool.h> #include <libvirt-gobject/libvirt-gobject-connection.h> #include <libvirt-gobject/libvirt-gobject-manager.h> +#include <libvirt-gobject/libvirt-gobject-stream.h> #endif /* __LIBVIRT_GOBJECT_H__ */ diff --git a/libvirt-gobject/libvirt-gobject.sym b/libvirt-gobject/libvirt-gobject.sym index db32c7f..478881b 100644 --- a/libvirt-gobject/libvirt-gobject.sym +++ b/libvirt-gobject/libvirt-gobject.sym @@ -143,6 +143,8 @@ LIBVIRT_GOBJECT_0.0.8 { gvir_storage_vol_get_info; gvir_storage_vol_delete; gvir_storage_vol_resize; + gvir_storage_vol_download; + gvir_storage_vol_upload; gvir_connection_handle_get_type; -- 1.7.10.4

4 3

[libvirt] [PATCHv3] fix failure when building with --disable-debug
by Hu Tao 18 Jul '12

18 Jul '12

When building with --disable-debug, VIR_DEBUG expands to a nop. But parameters to VIR_DEBUG can be variables that are passed only to VIR_DEBUG. In the case the building system complains about unused variables. --- src/util/logging.h | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/src/util/logging.h b/src/util/logging.h index 70318d0..2e70af4 100644 --- a/src/util/logging.h +++ b/src/util/logging.h @@ -34,8 +34,21 @@ # define VIR_DEBUG_INT(category, f, l, ...) \ virLogMessage(category, VIR_LOG_DEBUG, f, l, 0, __VA_ARGS__) # else +/** + * virLogEatParam: + * + * Do nothing but eat parameters. + */ +static inline void virLogEatParam(const char *category, + const char *funcname, + long long linenr, ...) +{ + category = category; + funcname = funcname; + linenr = linenr; +} # define VIR_DEBUG_INT(category, f, l, ...) \ - do { } while (0) + virLogEatParam(category, f, l, __VA_ARGS__) # endif /* !ENABLE_DEBUG */ # define VIR_INFO_INT(category, f, l, ...) \ -- 1.7.10.2

5 4

[libvirt] [PATCH] remote: Fix locking in stream APIs
by Jiri Denemark 18 Jul '12

18 Jul '12

Remote driver needs to make sure the driver lock is released before entering client IO loop as that may block indefinitely in poll(). As a direct consequence of not following this in stream APIs, tunneled migration to a destination host which becomes non-responding may block qemu driver. Luckily, if keepalive is turned for p2p migrations, both remote and qemu drivers will get automagically unblocked after keepalive timeout. --- src/remote/remote_driver.c | 40 ++++++++++++++++++++++++++++------------ 1 file changed, 28 insertions(+), 12 deletions(-) diff --git a/src/remote/remote_driver.c b/src/remote/remote_driver.c index 3314f80..acae5d0 100644 --- a/src/remote/remote_driver.c +++ b/src/remote/remote_driver.c @@ -3976,12 +3976,14 @@ remoteStreamSend(virStreamPtr st, VIR_DEBUG("st=%p data=%p nbytes=%zu", st, data, nbytes); struct private_data *priv = st->conn->privateData; virNetClientStreamPtr privst = st->privateData; - int rv = -1; - - remoteDriverLock(priv); + int rv; if (virNetClientStreamRaiseError(privst)) - goto cleanup; + return -1; + + remoteDriverLock(priv); + priv->localUses++; + remoteDriverUnlock(priv); rv = virNetClientStreamSendPacket(privst, priv->client, @@ -3989,9 +3991,9 @@ remoteStreamSend(virStreamPtr st, data, nbytes); -cleanup: + remoteDriverLock(priv); + priv->localUses--; remoteDriverUnlock(priv); - return rv; } @@ -4004,12 +4006,14 @@ remoteStreamRecv(virStreamPtr st, VIR_DEBUG("st=%p data=%p nbytes=%zu", st, data, nbytes); struct private_data *priv = st->conn->privateData; virNetClientStreamPtr privst = st->privateData; - int rv = -1; - - remoteDriverLock(priv); + int rv; if (virNetClientStreamRaiseError(privst)) - goto cleanup; + return -1; + + remoteDriverLock(priv); + priv->localUses++; + remoteDriverUnlock(priv); rv = virNetClientStreamRecvPacket(privst, priv->client, @@ -4019,9 +4023,9 @@ remoteStreamRecv(virStreamPtr st, VIR_DEBUG("Done %d", rv); -cleanup: + remoteDriverLock(priv); + priv->localUses--; remoteDriverUnlock(priv); - return rv; } @@ -4138,12 +4142,18 @@ remoteStreamFinish(virStreamPtr st) if (virNetClientStreamRaiseError(privst)) goto cleanup; + priv->localUses++; + remoteDriverUnlock(priv); + ret = virNetClientStreamSendPacket(privst, priv->client, VIR_NET_OK, NULL, 0); + remoteDriverLock(priv); + priv->localUses--; + cleanup: virNetClientRemoveStream(priv->client, privst); virNetClientStreamFree(privst); @@ -4167,12 +4177,18 @@ remoteStreamAbort(virStreamPtr st) if (virNetClientStreamRaiseError(privst)) goto cleanup; + priv->localUses++; + remoteDriverUnlock(priv); + ret = virNetClientStreamSendPacket(privst, priv->client, VIR_NET_ERROR, NULL, 0); + remoteDriverLock(priv); + priv->localUses--; + cleanup: virNetClientRemoveStream(priv->client, privst); virNetClientStreamFree(privst); -- 1.7.11.1

2 2

Re: [libvirt] How to get the real device in guest os after attached a disk?
by Stefan Hajnoczi 18 Jul '12

18 Jul '12

On Tue, Jul 17, 2012 at 4:40 AM, Wangpan <hzwangpan(a)corp.netease.com> wrote: > I have a question as the subject above, the reason I want to know this is that, if I attach some disks on the guest, > for example, I specified /dev/vdc&/dev/vdd(target device) at the cmd line by using 'virsh attach-disk', but they may be /dev/vdb&/dev/vdc in the guest os, > so if the guest user want to detach the /dev/vdb(guest device), he\she will be confused with the two target devices /dev/vdb&/dev/vdc, > because he\she doesn't know the corresponding relation of the guest device and target device, > he\she may detach an error device /dev/vdd(target device) which corresponding to /dev/vdc(guest device). > > Could anyone give me some idea? You can use the virtio-blk serial or file system/volume labels to distinguish them. For libvirt disk <serial> information, see the Domain XML documentation: http://libvirt.org/formatdomain.html#elementsDisks Stefan

2 2

[libvirt] [PATCH] xenapi: Convert to virMacAddr
by Jiri Denemark 17 Jul '12

17 Jul '12

--- Can't we just get rid of this horrific and mostly unmaintained xenapi code completely? :-) src/xenapi/xenapi_driver.c | 3 ++- src/xenapi/xenapi_utils.c | 36 ++++++++++++++---------------------- 2 files changed, 16 insertions(+), 23 deletions(-) diff --git a/src/xenapi/xenapi_driver.c b/src/xenapi/xenapi_driver.c index 298ab21..a91f497 100644 --- a/src/xenapi/xenapi_driver.c +++ b/src/xenapi/xenapi_driver.c @@ -1512,7 +1512,8 @@ xenapiDomainGetXMLDesc(virDomainPtr dom, unsigned int flags) } xen_vif_get_record(session, &vif_rec, vif); if (vif_rec != NULL) { - if (virMacAddrParse((const char *)vif_rec->mac,defPtr->nets[i]->mac) < 0) + if (virMacAddrParse((const char *)vif_rec->mac, + &defPtr->nets[i]->mac) < 0) xenapiSessionErrorHandler(dom->conn, VIR_ERR_INTERNAL_ERROR, _("Unable to parse given mac address")); xen_vif_record_free(vif_rec); diff --git a/src/xenapi/xenapi_utils.c b/src/xenapi/xenapi_utils.c index e5a9cc5..c8b12de 100644 --- a/src/xenapi/xenapi_utils.c +++ b/src/xenapi/xenapi_utils.c @@ -453,7 +453,6 @@ createVMRecordFromXml (virConnectPtr conn, virDomainDefPtr def, char uuidStr[VIR_UUID_STRING_BUFLEN]; xen_string_string_map *strings = NULL; int device_number = 0; - char *bridge = NULL, *mac = NULL; int i; *record = xen_vm_record_alloc(); @@ -542,28 +541,21 @@ createVMRecordFromXml (virConnectPtr conn, virDomainDefPtr def, } for (i = 0; i < def->nnets; i++) { - if (def->nets[i]->type == VIR_DOMAIN_NET_TYPE_BRIDGE) { - if (def->nets[i]->data.bridge.brname) - if (!(bridge = strdup(def->nets[i]->data.bridge.brname))) - goto error_cleanup; - if (def->nets[i]->mac) { - char macStr[VIR_MAC_STRING_BUFLEN]; - virMacAddrFormat(def->nets[i]->mac, macStr); - if (!(mac = strdup(macStr))) { - VIR_FREE(bridge); - goto error_cleanup; - } - } - if (mac != NULL && bridge != NULL) { - if (createVifNetwork(conn, *vm, device_number, bridge, - mac) < 0) { - VIR_FREE(bridge); - goto error_cleanup; - } - VIR_FREE(bridge); - device_number++; + if (def->nets[i]->type == VIR_DOMAIN_NET_TYPE_BRIDGE && + def->nets[i]->data.bridge.brname) { + char *mac; + + if (VIR_ALLOC_N(mac, VIR_MAC_STRING_BUFLEN) < 0) + goto error_cleanup; + virMacAddrFormat(&def->nets[i]->mac, mac); + + if (createVifNetwork(conn, *vm, device_number, + def->nets[i]->data.bridge.brname, + mac) < 0) { + VIR_FREE(mac); + goto error_cleanup; } - VIR_FREE(bridge); + device_number++; } } return 0; -- 1.7.11.1

2 2

[libvirt] IPs via DHCP for guest domains
by Fong Vang 17 Jul '12

17 Jul '12

I'm curious to know how you are providing IPs for your Windows VMs when connected via a bridge. Since broadcasts from the VMs are going out the bridged interface, I'm thinking about implementing ebtables to block dhcp broadcasts from going in/out the interface so that dnsmasq would only respond to dhcp requests from the local guest VMs. Before I venture down this path, I thought I might ask if there's an easier way to do this with libvirt. Thanks for any help.

2 1

[libvirt] [PATCH] nwfilter: Convert to virMacAddrPtr not virMacAddrPtr*
by Michal Privoznik 17 Jul '12

17 Jul '12

My previous patch wasn't complete as it converted from (unsigned char *) to (virMacAddrPtr *). It should be (virMacAddrPtr) only. --- Again, pushed under build breaker. src/nwfilter/nwfilter_dhcpsnoop.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/nwfilter/nwfilter_dhcpsnoop.c b/src/nwfilter/nwfilter_dhcpsnoop.c index 312fada..b38e780 100644 --- a/src/nwfilter/nwfilter_dhcpsnoop.c +++ b/src/nwfilter/nwfilter_dhcpsnoop.c @@ -2200,7 +2200,7 @@ virNWFilterDHCPSnoopReq(virNWFilterTechDriverPtr techdriver ATTRIBUTE_UNUSED, const char *linkdev ATTRIBUTE_UNUSED, enum virDomainNetType nettype ATTRIBUTE_UNUSED, const unsigned char *vmuuid ATTRIBUTE_UNUSED, - const virMacAddrPtr *macaddr ATTRIBUTE_UNUSED, + const virMacAddrPtr macaddr ATTRIBUTE_UNUSED, const char *filtername ATTRIBUTE_UNUSED, virNWFilterHashTablePtr filterparams ATTRIBUTE_UNUSED, virNWFilterDriverStatePtr driver ATTRIBUTE_UNUSED) -- 1.7.8.6

1 0

[libvirt] [PATCH 0/4] qemu: configurable port boundaries for remote displays
by Martin Kletzander 17 Jul '12

17 Jul '12

This series introduces a possibility to change default minimal and maximal port numbers that are used to specify a remote display port for both VNC and SPICE. Because the code was a bit messy, PATCH 1/4 cleans up few things needed to make a clean run of PATCH 2/4, that does the main change in the code. I also noticed two things that could be changed and it made sense for me to do that, but these are nowhere near any importance, so feel free to reject them if your heart feels that way. PATCH 3/4 rewords three messages and applying it would mean that they are not translated. Even though I think it makes more sense this way, I'm not a good English speaker, so that's more like an RFC. PATCH 4/4 makes more flexible port searching available, but is used on one place and because there are so many ports, it won't be that used, I guess. P.S.: this is basically v2 for https://www.redhat.com/archives/libvir-list/2012-May/msg01081.html but it changed a lot and the name is also different, so I didn't want to create any confision. Martin Kletzander (4): qemu: Unify port-wise SPICE and VNC behavior qemu: configurable remote display port boundaries qemu: modify 3 error messages qemu: allow searching for all open ports src/conf/domain_conf.c | 2 +- src/qemu/libvirtd_qemu.aug | 4 ++ src/qemu/qemu.conf | 14 ++++++ src/qemu/qemu_command.h | 11 ++++- src/qemu/qemu_conf.c | 40 +++++++++++++++++- src/qemu/qemu_conf.h | 4 +- src/qemu/qemu_driver.c | 19 ++++----- src/qemu/qemu_process.c | 81 +++++++++++++++++++++--------------- src/qemu/test_libvirtd_qemu.aug.in | 2 + 9 files changed, 128 insertions(+), 49 deletions(-) -- 1.7.8.6

1 6