Jim Meyering [Tue, 17 Feb 2009 10:33:41 +0000 (10:33 +0000)]
Work around 'virGetVersion' build problem
The 'getVer' fix introducted in d88d459d [Allow remote://hostname/
style URIs for automatic driver probe...] breaks compiling libvirt
with loadable module support. Work around this to get it building again.
Signed-off-by: Maximilian Wilhelm <max@rfc2324.org>
Jim Meyering [Tue, 17 Feb 2009 10:26:43 +0000 (10:26 +0000)]
avoid compiler warning about unused parameter
* src/bridge.c (brProbeVnetHdr)
[IFF_VNET_HDR && TUNGETFEATURES && TUNGETIFF]: Use a "(void)" case
to mark the parameter as unused. Reported by Maximilian Wilhelm in
http://thread.gmane.org/gmane.comp.emulators.libvirt/11918/focus=11917
* configure.in: Enable internal WITH_BRIDGE support not just for LXC or
QEMU, but also for --with-network. Russell Haering reported that
compilation failed when configured with --without-lxc --without-qemu.
Details: http://thread.gmane.org/gmane.comp.emulators.libvirt/11928
* src/bridge.h: Use #if WITH_BRIDGE, rather than currently-equivalent
"#if defined(WITH_QEMU) || defined(WITH_LXC)"
Mark McLoughlin [Fri, 13 Feb 2009 19:10:55 +0000 (19:10 +0000)]
Fix bridge.c warnings with gcc-4.4
bridge.c: In function 'brSetInetAddr':
bridge.c:665: error: dereferencing pointer '({anonymous})' does break strict-aliasing rules
bridge.c:665: note: initialized from here
bridge.c:666: error: dereferencing pointer '({anonymous})' does break strict-aliasing rules
bridge.c:666: note: initialized from here
Fix qemud.c warnings with gcc-4.4
qemud.c: In function ‘remoteListenTCP’:
qemud.c:675: error: dereferencing pointer ‘sa.72’ does break strict-aliasing rules
qemud.c:675: note: initialized from here
qemud.c:678: error: dereferencing pointer ‘sa.73’ does break strict-aliasing rules
qemud.c:678: note: initialized from here
* src/test.c (testNetworkCreate, testNetworkDefine): Since "def" is
set to NULL immediately after any vir*AssignDef call (to indicate we
no longer own it and to ensure no clean-up path mistakenly frees it),
dereference via net->def->, not def->.
Jim Meyering [Mon, 9 Feb 2009 17:52:38 +0000 (17:52 +0000)]
libvirtd: new config-file option: unix_sock_dir
Before this change, the unix socket directory was hard-coded
to be e.g., /var/run/libvirt for euid==0 and ~/.libvirt otherwise.
With this change, you may now specify that directory in libvirtd's
config file via a line like this: unix_sock_dir = "/var/run/libvirt".
This is essential for running tests that do not impinge on any
existing libvirtd process, and in running tests in parallel.
* qemud/libvirtd.conf (unix_sock_dir): Add comment and example.
* qemud/qemud.h (struct qemud_server) [logDir]: Change type
from char[PATH_MAX] to char*.
* qemud/qemud.c (unix_sock_dir): New global
(remoteReadConfigFile): Set the global.
(qemudInitPaths): Use the global, unix_sock_dir, if non-NULL.
One minor improvement: unlink both sockets or none, never
just one of them.
(qemudCleanup): Free logDir.
(main): Use the new global rather than hard-coding "/run/libvirt".
* qemud/libvirtd.aug (sock_acl_entry): Add "unix_sock_dir".
Jim Meyering [Thu, 5 Feb 2009 16:27:51 +0000 (16:27 +0000)]
iptables.c: Use virStrerror, not strerror.
* src/iptables.c: Include "virterror_internal.h".
Use virStrerror, not strerror.
* src/iptables.c (notifyRulesUpdated): Use %s rather than
string-concatenation that made sc_unmarked_diagnostics report
a false-positive.
Jim Meyering [Thu, 5 Feb 2009 16:27:17 +0000 (16:27 +0000)]
remove duplicate *SetCloseExec and *SetNonBlock functions
* src/qemu_driver.c: Use virSetCloseExec and virSetNonBlock,
rather than qemuSet* functions. Suggested by Daniel P. Berrange.
* src/util.c (virSetCloseExec): Publicize.
* src/util.h (virSetCloseExec): Declare
* src/libvirt_private.syms: Add virSetCloseExec.
Jim Meyering [Thu, 5 Feb 2009 16:11:25 +0000 (16:11 +0000)]
qemu_driver.c: use virReportSystemError in place of some qemudLog uses
...thus eliminating many uses of strerror.
(kvmGetMaxVCPUs): Rather than merely warning, diagnose an error
and return -1 upon failure to open KVM_DEVICE. Suggested by
Daniel P. Berrange.
Jim Meyering [Fri, 30 Jan 2009 19:57:20 +0000 (19:57 +0000)]
fix qemud version reporting when qemu is not installed
* src/qemu_conf.c (uname_normalize): New function, factored out of...
(qemudBuildCommandLine): ...here. Use the new function.
(qemudExtractVersion): Use it here, rather than hard-coding "i686".
Jim Meyering [Wed, 28 Jan 2009 14:08:33 +0000 (14:08 +0000)]
libvirt_proxy: avoid potential buffer overflow
* proxy/libvirt_proxy.c (proxyReadClientSocket): Ensure that
we've read an entire virProxyPacket before dereferencing "req".
Analysis and patch by "Rasputin" <rasputin@email.ru>. Details in
<http://thread.gmane.org/gmane.comp.emulators.libvirt/11459>.
Mark McLoughlin [Tue, 27 Jan 2009 11:12:05 +0000 (11:12 +0000)]
kvm/virtio: Set IFF_VNET_HDR when setting up tap fds
IFF_VNET_HDR is a tun/tap flag that allows you to send and receive
large (i.e. GSO) packets and packets with partial checksums. Setting
the flag means that every packet is proceeded by the same header which
virtio uses to communicate GSO/csum metadata.
By enabling this flag on the tap fds we create, we greatly increase
the achievable throughput with virtio_net.
However, we need to be careful to only set the flag when a) QEMU has
support for this ABI and b) the value of the flag is queryable using
the TUNGETIFF ioctl.
It's nearly five months since kvm-74 - the first KVM release with this
feature - was released. Up until now, we've not added libvirt support
because there is no clean way to detect support for this in QEMU at
runtime. A brief attempt to add a "info capabilities" monitor command
to QEMU floundered. Perfect is the enemy of good enough. Probing the
KVM version will suffice for now.
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Jim Meyering [Tue, 27 Jan 2009 10:57:18 +0000 (10:57 +0000)]
fix errors in virReportSystemErrorFull
* src/virterror.c (virStrerror): New function.
(virReportSystemErrorFull): Don't leak "combined".
In fact, don't even attempt allocation.
Do include the result of formatted print in final diagnostic.