On Mon, Feb 14, 2011 at 20:36:24 +0800, Daniel Veillard wrote:

If you know how to configure mailman to make the list moderated, I'm all ears ... you may even win moderator status :-)

Heh, to be honest, I've never setup any mailing list so I don't have a foggiest glue how to configure it that way.

In practice you can change all current users to be moderated but if someone subscribe he gets the right to post, hence the SPAM I'm just comparing to other libvirt lists where I don't remember seeing any spam.

why I didn't tried to promote use of the list until that's fixed !

OK, sounds reasonable. Although the list is already mentioned on libvirt web site, which gives it a bit of promotion. Jirka

On Mon, Feb 14, 2011 at 09:46:01PM +1100, Justin Clift wrote:

On 14/02/2011, at 8:15 PM, Justin Clift wrote:

...

On 14/02/2011, at 3:22 PM, Daniel Veillard wrote: <snip>

...

it does fix the error messages I had on linux configure, but I'm not sure the issues on Win32 and OS-X are all sorted out yet, so please check.

Still broken on OSX. :(

Found the commit which introduces the breakage, it's this one:

************************************************************** **************************************************************

commit 2169472ab6cac3bdb1f32578de2b608a61a1d703 Author: Jiri Denemark <jdenemar@redhat.com> Date: Wed Feb 2 16:37:10 2011 +0100

qemu: Add shortcut for HMP pass through

Currently users who want to use virDomainQemuMonitorCommand() API or it's virsh equivalent has to use the same protocol as libvirt uses for communication to qemu. Since the protocol is QMP with current qemu and HMP much more usable for humans, one ends up typing something like the following:

virsh qemu-monitor-command DOM \ '{"execute":"human-monitor-command","arguments":{"command-line":"info kvm"}}'

which is not a very convenient way of debugging qemu.

This patch introduces --hmp option to qemu-monitor-command, which says that the provided command is in HMP. If libvirt uses QMP to talk with qemu, the command will automatically be converted into QMP. So the example above is simplified to just

virsh qemu-monitor-command --hmp DOM "info kvm"

Also the result is converted from

{"return":"kvm support: enabled\r\n"}

to just plain HMP:

kvm support: enabled

If libvirt talks to qemu in HMP, --hmp flag is obviously a noop.

************************************************************** **************************************************************

Now, I just need to understand how this patch causes the error:

ld: duplicate symbol _virDomainQemuMonitorCommandFlags in ./.libs/libvirt_util.a(libvirt_util_la-bitmap.o) and ./.libs/libvirt_util.a(libvirt_util_la-authhelper.o)

Suggestions and pointers welcome. :)

It is a broken enum declaration in the header libvirt-qemu.h: enum { VIR_DOMAIN_QEMU_MONITOR_COMMAND_DEFAULT = 0, VIR_DOMAIN_QEMU_MONITOR_COMMAND_HMP = (1 << 0), /* cmd is in HMP */ } virDomainQemuMonitorCommandFlags; There is a missing 'typedef' keyword there before 'enum', so instead of defining a type, it is defining a variable :-( Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|

On Mon, Feb 14, 2011 at 10:30:13PM +1100, Justin Clift wrote:

On 14/02/2011, at 9:58 PM, Daniel P. Berrange wrote:

...

It is a broken enum declaration in the header libvirt-qemu.h:

never underestimate the power of being multi platform in chasing weird bugs !

...

enum { VIR_DOMAIN_QEMU_MONITOR_COMMAND_DEFAULT = 0, VIR_DOMAIN_QEMU_MONITOR_COMMAND_HMP = (1 << 0), /* cmd is in HMP */ } virDomainQemuMonitorCommandFlags;

There is a missing 'typedef' keyword there before 'enum', so instead of defining a type, it is defining a variable :-(

Thanks Daniel (Berrange), that's perfect, it really was that simple. :) (whew!)

Just submitted the simple 1 line fix adding "typedef" before the enum, and it works on OSX now. (still working on F13 too, no breakage caused)

Daniel (Veillard) do you reckon it's worth knocking out a 0.8.8rc3 quickly or something?

okay, I will make another one tomorrow morning to accumulate a bit more patches :-) Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel@veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

2011/2/14 Zdenek Styblik <stybla@turnovfree.net>:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 02/14/11 05:22, Daniel Veillard wrote:

...

  Just a reminder that we are in feature freeze for 0.8.8, I just uploaded a new candidate version:

  ftp://libvirt.org/libvirt/libvirt-0.8.8-rc2.tar.gz

it does fix the error messages I had on linux configure, but I'm not sure the issues on Win32 and OS-X are all sorted out yet, so please check. Note that I didn't tried to update the localizations from the fedora side yet so those tarballs may be a little off from that perspective.

  thanks !

Daniel

Hello,

it seems there is a problem at Slackware64-13.1

~~~ SNIP ~~~  CCLD   libvirt_driver.la  CCLD   libvirt_driver_test.la  CCLD   libvirt_driver_remote.la  CCLD   libvirt_driver_openvz.la  CCLD   libvirt_driver_vmware.la  CCLD   libvirt_driver_vbox.la  CCLD   libvirt_driver_nwfilter.la  CCLD   libvirt_driver_security.la libtool: link: require no space between `-L' and `/usr/lib64'

Could you run make V=1 to get the full command line that produces this error? Either libvirt itself or some other library (via pkg-config) injects a '-L /usr/lib64' (instead of '-L/usr/lib64') into the LDFLAGS and libtool doesn't like it. Matthias

On Mon, Feb 14, 2011 at 06:07:59PM +0100, Zdenek Styblik wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 02/14/11 17:35, Matthias Bolte wrote: [...]

...

Could you run make V=1 to get the full command line that produces this error?

Either libvirt itself or some other library (via pkg-config) injects a '-L /usr/lib64' (instead of '-L/usr/lib64') into the LDFLAGS and libtool doesn't like it.

Matthias

~~~ SNIP ~~~ make[3]: Entering directory `/tmp/libvirt-rc/libvirt-0.8.8/src' /bin/sh ../libtool --tag=CC --mode=link gcc -I /usr/include - -I../src/conf -O2 -fPIC -L/usr/lib64 -o libvirt_driver_nwfilter.la libvirt_driver_nwfilter_la-nwfilter_driver.lo libvirt_driver_nwfilter_la-nwfilter_gentech_driver.lo libvirt_driver_nwfilter_la-nwfilter_ebiptables_driver.lo libvirt_driver_nwfilter_la-nwfilter_learnipaddr.lo -L /usr/lib64 -lpcap -ldl libtool: link: require no space between `-L' and `/usr/lib64'

What does the following print # pcap-config --cflags --libs If it includes a space, then pcap-config needs to be fixed Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|

2011/2/14 Zdenek Styblik <stybla@turnovfree.net>:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 02/14/11 18:15, Daniel P. Berrange wrote: [...]

...

What does the following print

   # pcap-config --cflags --libs

If it includes a space, then pcap-config needs to be fixed

Regards, Daniel

# pcap-config --cflags --libs - -I /usr/include

Z.

pcap-config (from libpcap 1.0.0) is pretty simple and doesn't accept multiple arguments at once. On Ubuntu 10.04 it outputs the additional space too. $ pcap-config --libs -L /usr/lib -lpcap I can reproduce the problem here. I didn't notice it until now because I didn't have libpcap-dev installed. On Ubuntu 10.10 (libpcap 1.1.1) the pcap-config script is improved, accepts multiple arguments and doesn't print the additional space (I think, judging from the source code). As pcap-config --cflags --libs seems to ignore the --libs argument I guess that Slackware64-13.1 ships an older libpcap version. So this problem is in libpcap's pcap-config script and not in libvirt. Matthias