- Devel - libvirt List Archives

[libvirt] [PATCHv3 0/4] phyp: add storage management
by Eric Blake 29 Jun '10

29 Jun '10

Here's my promised refactoring of the patches. Eduardo Otubo (2): phyp: add rudimentary storage driver phyp: add storage management driver Eric Blake (2): phyp: reduce scope of driver functions phyp: optimize use of sed src/phyp/phyp_driver.c | 3542 +++++++++++++++++++++++++++++++++++------------- src/phyp/phyp_driver.h | 76 +- 2 files changed, 2621 insertions(+), 997 deletions(-)

3 13

[libvirt] [PATCH] lxc: Fix virsh console doesn't work after restarting libvirtd
by Ryota Ozaki 29 Jun '10

29 Jun '10

Because tty path is unexpectedly not saved in the live configuration file of a domain, libvirtd cannot get the console of the domain back after restarting. The reason why the tty path isn't saved is that, to save the tty path, the save function, virDomainSaveConfig, requires that the target domain is running (pid != -1), however, lxc driver calls the function before starting the domain to pass the configuration to controller. To ensure to save the tty path, the patch lets lxc driver call the save function again after starting the domain. --- src/lxc/lxc_driver.c | 9 ++++++++- 1 files changed, 8 insertions(+), 1 deletions(-) diff --git a/src/lxc/lxc_driver.c b/src/lxc/lxc_driver.c index 19d4dcb..462bc9c 100644 --- a/src/lxc/lxc_driver.c +++ b/src/lxc/lxc_driver.c @@ -1284,7 +1284,7 @@ static int lxcVmStart(virConnectPtr conn, if (lxcSetupInterfaces(conn, vm->def, &nveths, &veths) != 0) goto cleanup; - /* Persist the live configuration now we have veth & tty info */ + /* Save the configuration for the controller */ if (virDomainSaveConfig(driver->stateDir, vm->def) < 0) goto cleanup; @@ -1328,6 +1328,13 @@ static int lxcVmStart(virConnectPtr conn, goto cleanup; } + /* + * Again, need to save the live configuration, because the function + * requires vm->def->id != -1 to save tty info surely. + */ + if (virDomainSaveConfig(driver->stateDir, vm->def) < 0) + goto cleanup; + rc = 0; cleanup: -- 1.6.5.2

3 2

[libvirt] [PATCHv2 0/4] Fix domain restore problems when selinux is enforcing
by Laine Stump 29 Jun '10

29 Jun '10

Changes from V1: 1) Don't set context label of kernel *or* image file, set the context label of both. 2) Add a patch to enhance virStorageFileIsSharedFS to behave similarly to the ill-fated virFileIsOnNetworkShare. 3) Use virStorageFileIsSharedFS instead of virStorageFileIsSharedFS. Prior to this patch series, restoring a domain with selinux set to enforcing would fail, because the function that sets the label on the file to allow qemu to read it did not have the name of the file (see the comments in the individual patches). A patch from Jamie Stranboge (2b57478ef0a0a983cc6a47b98300c8359f9708d0) added the filename to the args passed down into the security driver; the first patch of this series takes advantage of that to properly set the label. Patches 2 - 4 solve a problem with restoring a domain from an NFS share - in this case the selinux functions will fail (as will functions trying to set the uid of the file, if it is a root-squashed share). The solution to this is just ignore the failure. If the security driver fails to set the label, and virStorageFileIsSharedFS() tells us that the file is on a network-shared FS, we ignore the failure, otherwise we behave as before. qemudDomainSaveFlag previously had a bit of code that detected if a particular path was on an NFS share; this code was replaced with a call to virStorageFileIsSharedFS, which is now functionally equivalent (better, even, since it detects a few other types of network filesystems).

3 10

[libvirt] [PATCH] Fix a reference leak for node devices.
by Chris Lalancette 28 Jun '10

28 Jun '10

There was one major, and a few minor bugs having to do with the reference counting of node devices in daemon/remote.c The major bug was that remoteDispatchNodeDeviceListCaps() was completely failing to unreference node devices; this would lead to many open file descriptors, which would eventually fail. The minor bugs were along the same lines, but were in rarely used error paths. Still, they should be corrected. Signed-off-by: Chris Lalancette <clalance(a)redhat.com> --- daemon/remote.c | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-) diff --git a/daemon/remote.c b/daemon/remote.c index 91faa9a..ec5f85b 100644 --- a/daemon/remote.c +++ b/daemon/remote.c @@ -5012,11 +5012,13 @@ remoteDispatchNodeDeviceGetParent (struct qemud_server *server ATTRIBUTE_UNUSED, /* remoteDispatchClientRequest will free this. */ char **parent_p; if (VIR_ALLOC(parent_p) < 0) { + virNodeDeviceFree(dev); remoteDispatchOOMError(rerr); return -1; } *parent_p = strdup(parent); if (*parent_p == NULL) { + virNodeDeviceFree(dev); remoteDispatchOOMError(rerr); return -1; } @@ -5048,6 +5050,7 @@ remoteDispatchNodeDeviceNumOfCaps (struct qemud_server *server ATTRIBUTE_UNUSED, ret->num = virNodeDeviceNumOfCaps(dev); if (ret->num < 0) { + virNodeDeviceFree(dev); remoteDispatchConnError(rerr, conn); return -1; } @@ -5076,6 +5079,7 @@ remoteDispatchNodeDeviceListCaps (struct qemud_server *server ATTRIBUTE_UNUSED, } if (args->maxnames > REMOTE_NODE_DEVICE_NAME_LIST_MAX) { + virNodeDeviceFree(dev); remoteDispatchFormatError(rerr, "%s", _("maxnames > REMOTE_NODE_DEVICE_NAME_LIST_MAX")); return -1; @@ -5083,6 +5087,7 @@ remoteDispatchNodeDeviceListCaps (struct qemud_server *server ATTRIBUTE_UNUSED, /* Allocate return buffer. */ if (VIR_ALLOC_N(ret->names.names_val, args->maxnames) < 0) { + virNodeDeviceFree(dev); remoteDispatchOOMError(rerr); return -1; } @@ -5091,11 +5096,13 @@ remoteDispatchNodeDeviceListCaps (struct qemud_server *server ATTRIBUTE_UNUSED, virNodeDeviceListCaps (dev, ret->names.names_val, args->maxnames); if (ret->names.names_len == -1) { + virNodeDeviceFree(dev); remoteDispatchConnError(rerr, conn); VIR_FREE(ret->names.names_val); return -1; } + virNodeDeviceFree(dev); return 0; } -- 1.6.6.1

5 6

[libvirt] [PATCH] lxc: Fix error handlings in lxcContainerRenameAndEnableInterfaces
by Ryota Ozaki 28 Jun '10

28 Jun '10

The function is expected to return negative value on failure, however, it returns positive value when either setInterfaceName or vethInterfaceUpOrDown fails. Because the function returns the return value of either as is, however, the two functions may return positive value on failure. The patch fixes the defects and add error messages. --- src/lxc/lxc_container.c | 13 ++++++++++--- 1 files changed, 10 insertions(+), 3 deletions(-) diff --git a/src/lxc/lxc_container.c b/src/lxc/lxc_container.c index 018f4d5..4371dba 100644 --- a/src/lxc/lxc_container.c +++ b/src/lxc/lxc_container.c @@ -255,13 +255,20 @@ static int lxcContainerRenameAndEnableInterfaces(unsigned int nveths, DEBUG("Renaming %s to %s", veths[i], newname); rc = setInterfaceName(veths[i], newname); - if (0 != rc) + if (0 != rc) { + VIR_ERROR(_("Failed to rename %s to %s (%d)"), + veths[i], newname, rc); + rc = -1; goto error_out; + } DEBUG("Enabling %s", newname); - rc = vethInterfaceUpOrDown(newname, 1); - if (0 != rc) + rc = vethInterfaceUpOrDown(newname, 1); + if (0 != rc) { + VIR_ERROR(_("Failed to enable %s (%d)"), newname, rc); + rc = -1; goto error_out; + } VIR_FREE(newname); } -- 1.6.5.2

2 1

[libvirt] [PATCH] lxc: Change VIR_ERROR to VIR_DEBUG for just a debugging message
by Ryota Ozaki 28 Jun '10

28 Jun '10

The message is actually not of error but of debugging. 02:22:56.091: error : lxcControllerMain:316 : monitor=3 client=4 appPty=19 contPty=7 --- src/lxc/lxc_controller.c | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-) diff --git a/src/lxc/lxc_controller.c b/src/lxc/lxc_controller.c index cffef52..44bcc82 100644 --- a/src/lxc/lxc_controller.c +++ b/src/lxc/lxc_controller.c @@ -313,7 +313,10 @@ static int lxcControllerMain(int monitor, fdArray[0].active = 0; fdArray[1].fd = contPty; fdArray[1].active = 0; - VIR_ERROR(_("monitor=%d client=%d appPty=%d contPty=%d"), monitor,client, appPty, contPty); + + VIR_DEBUG("monitor=%d client=%d appPty=%d contPty=%d", + monitor, client, appPty, contPty); + /* create the epoll fild descriptor */ epollFd = epoll_create(2); if (0 > epollFd) { -- 1.6.5.2

2 1

[libvirt] Live Migration with non-shared storage for kvm
by Kenneth Nagin 27 Jun '10

27 Jun '10

On Wed Apr 07 I sent an email with a patch titled Live Migration with non-shared storage for kvm. What is the status on the patch? Kenneth Nagin Ph: +972-4-8296227 Cell: 054-6976227 Fx: +972-4- 8296113 "Do not look back in anger, or forward in fear, but around in awareness." --James Thurber

6 24

[libvirt] [PATCH] build: fix regression with libvirt-api.xml generation
by Eric Blake 26 Jun '10

26 Jun '10

(EXTRA_DIST): Remove redundant listing of xml files. (html/%-%.html, html/%-virterror.html, %-api.xml, %-refs.xml): Rewrite with... (python_generated_files): ...new macro, since make didn't see through the dependency chain correctly otherwise. --- Since I'm not going to get the xsltproc stuff fixed today, but the broken python bindings are still annoying, I'm splitting this into a smaller patch. This does not fix VPATH builds (but that is no change in the status quo), but DOES fix libvirt-api.xml dependency detection, so that touching include/libvirt/libvirt.h.in will once again reflect all the way to python bindings during 'make all'. docs/Makefile.am | 20 ++++++++++++-------- 1 files changed, 12 insertions(+), 8 deletions(-) diff --git a/docs/Makefile.am b/docs/Makefile.am index 41068c6..c01e085 100644 --- a/docs/Makefile.am +++ b/docs/Makefile.am @@ -70,8 +70,8 @@ fig = \ libvirt-object-model.fig EXTRA_DIST= \ - libvirt-api.xml libvirt-refs.xml apibuild.py \ - site.xsl newapi.xsl news.xsl page.xsl ChangeLog.xsl \ + apibuild.py \ + site.xsl newapi.xsl news.xsl page.xsl ChangeLog.xsl \ $(dot_html) $(dot_html_in) $(gif) $(apihtml) $(apipng) \ $(devhelphtml) $(devhelppng) $(devhelpcss) $(devhelpxsl) \ $(xml) $(fig) $(png) \ @@ -129,19 +129,23 @@ html/index.html: libvirt-api.xml newapi.xsl page.xsl sitemap.html.in $(XMLLINT) --nonet --valid --noout html/*.html ; \ else echo "missing XHTML1 DTD" ; fi ; fi -$(devhelphtml): libvirt-api.xml $(devhelpxsl) +$(addprefix $(srcdir)/,$(devhelphtml)): $(srcdir)/libvirt-api.xml $(devhelpxsl) -@echo Rebuilding devhelp files -@if [ -x $(XSLTPROC) ] ; then \ $(XSLTPROC) --nonet -o devhelp/libvirt.devhelp \ $(top_srcdir)/docs/devhelp/devhelp.xsl libvirt-api.xml ; fi -html/%-%.html html/%-virterror.html %-api.xml %-refs.xml: $(srcdir)/apibuild.py - -srcdir=$(srcdir) $(srcdir)/apibuild.py +python_generated_files = \ + $(srcdir)/html/libvirt-libvirt.html \ + $(srcdir)/html/libvirt-virterror.html \ + $(srcdir)/libvirt-api.xml \ + $(srcdir)/libvirt-refs.xml -html/%-%.html html/%-virterror.html %-api.xml %-refs.xml: \ - $(srcdir)/../include/%/*.h \ - $(srcdir)/../src/%.c \ +$(python_generated_files): $(srcdir)/apibuild.py \ + $(srcdir)/../include/libvirt/*.h \ + $(srcdir)/../src/libvirt.c \ $(srcdir)/../src/util/virterror.c + -srcdir=$(srcdir) $(srcdir)/apibuild.py check-local: all -- 1.7.0.1

2 2

[libvirt] [PATCH] Fix more reference leaks for node devices
by Matthias Bolte 26 Jun '10

26 Jun '10

Reported by Chris Lalancette, I just found more instances of the same pattern. --- daemon/remote.c | 8 ++++++++ 1 files changed, 8 insertions(+), 0 deletions(-) diff --git a/daemon/remote.c b/daemon/remote.c index ec5f85b..cb9e83d 100644 --- a/daemon/remote.c +++ b/daemon/remote.c @@ -5126,10 +5126,12 @@ remoteDispatchNodeDeviceDettach (struct qemud_server *server ATTRIBUTE_UNUSED, } if (virNodeDeviceDettach(dev) == -1) { + virNodeDeviceFree(dev); remoteDispatchConnError(rerr, conn); return -1; } + virNodeDeviceFree(dev); return 0; } @@ -5153,10 +5155,12 @@ remoteDispatchNodeDeviceReAttach (struct qemud_server *server ATTRIBUTE_UNUSED, } if (virNodeDeviceReAttach(dev) == -1) { + virNodeDeviceFree(dev); remoteDispatchConnError(rerr, conn); return -1; } + virNodeDeviceFree(dev); return 0; } @@ -5180,10 +5184,12 @@ remoteDispatchNodeDeviceReset (struct qemud_server *server ATTRIBUTE_UNUSED, } if (virNodeDeviceReset(dev) == -1) { + virNodeDeviceFree(dev); remoteDispatchConnError(rerr, conn); return -1; } + virNodeDeviceFree(dev); return 0; } @@ -5230,10 +5236,12 @@ remoteDispatchNodeDeviceDestroy(struct qemud_server *server ATTRIBUTE_UNUSED, } if (virNodeDeviceDestroy(dev) == -1) { + virNodeDeviceFree(dev); remoteDispatchConnError(rerr, conn); return -1; } + virNodeDeviceFree(dev); return 0; } -- 1.7.0.4

3 3

[libvirt] [RESEND PATCHv3 0/3] fix build dependencies
by Eric Blake 26 Jun '10

26 Jun '10

https://bugzilla.redhat.com/show_bug.cgi?id=589465 was marked incomplete, because VIR_DOMAIN_START_PAUSED did not make it into the python bindings of 0.8.1-10.el6. Matthias Bolte was kind enough to help me figure out why (see patch 2/3 for details - but the short story was a Makefile dependency problem that failed to rebuild the API documentation after an addition to libvirt.h.in). I was a bit surprised when I could not repeat the test on the tip of my development tree, which means I already have pending patches previously posted to the list but not yet reviewed that are already sufficient to resolve the problem. Hence this resend. Eric Blake (3): build: restore operation of bit-rotted 'make cov' build: fix VPATH build of docs autobuild.sh: use VPATH build .gitignore | 4 ++ Makefile.am | 9 +++- autobuild.sh | 10 ++-- configure.ac | 1 + daemon/.gitignore | 2 - docs/Makefile.am | 48 ++++++++++++--------- docs/apibuild.py | 18 ++++++-- m4/compiler-flags.m4 | 1 - src/.gitignore | 3 - src/Makefile.am | 112 +++++++++++++++++++++++++++----------------------- tests/.gitignore | 2 - 11 files changed, 117 insertions(+), 93 deletions(-)

2 8