- Devel - libvirt List Archives

[libvirt] [PATCH 00/12] A bunch of extensions to libxl driver
by Marek Marczykowski 30 May '13

30 May '13

This are some additional features to libxl driver. Some of them require change in domain config structures/syntax. Details described with each patch. The last two patches are bugfix for deadlock during daemon startup. Marek Marczykowski (12): libxl: allow script for any network interface, not only bridge libxl: PCI passthrough support libxl: nodeDevice* support for PCI devices libxl: populate xenstore memory entries at startup conf: add 'script' attribute to disk specification libxl: use disk 'script' attribute conf: support backend domain name in disk and network devices libxl: support backend domain setting for disk and net devices libxl: fill HVM SDL and VNC settings based on <graphics/> entries RFC: libxl: special 'stubdom-dm' emulator to use qemu in stub domain conf: virDomainObjListRemoveLocked function libxl: fix deadlock in libxlReconnectDomain docs/schemas/domaincommon.rng | 14 ++ src/conf/domain_conf.c | 54 ++++++++ src/conf/domain_conf.h | 5 + src/libvirt_private.syms | 1 + src/libxl/libxl_conf.c | 297 +++++++++++++++++++++++++++++++++--------- src/libxl/libxl_conf.h | 6 +- src/libxl/libxl_driver.c | 253 +++++++++++++++++++++++++++++++---- 7 files changed, 545 insertions(+), 85 deletions(-) -- 1.8.1.4

7 44

[libvirt] [PATCH] build: port qemu to cygwin
by Eric Blake 30 May '13

30 May '13

A cygwin build of the qemu driver fails with: qemu/qemu_process.c: In function 'qemuPrepareCpumap': qemu/qemu_process.c:1803:31: error: 'CPU_SETSIZE' undeclared (first use in this function) CPU_SETSIZE is a Linux extension in <sched.h>; a bit more portable is using sysconf if _SC_NPROCESSORS_CONF is defined (several platforms have it, including Cygwin). Ultimately, I would have preferred to use gnulib's 'nproc' module, but it is currently under an incompatible license. * src/qemu/qemu_conf.h (QEMUD_CPUMASK_LEN): Provide definition on cygwin. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- I'll wait for a review on this one, particularly since I'm still trying to solve another qemu failure on cygwin: qemu/qemu_monitor.c:418:9: error: passing argument 2 of 'sendmsg' from incmpatible pointer type /usr/include/sys/socket.h:42:11: note: expected 'const struct msghdr *' but argument is of type 'struct msghdr *' src/qemu/qemu_conf.h | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_conf.h b/src/qemu/qemu_conf.h index df0791e..42566b4 100644 --- a/src/qemu/qemu_conf.h +++ b/src/qemu/qemu_conf.h @@ -45,7 +45,13 @@ # include "locking/lock_manager.h" # include "qemu_capabilities.h" -# define QEMUD_CPUMASK_LEN CPU_SETSIZE +# ifdef CPU_SETSIZE /* Linux */ +# define QEMUD_CPUMASK_LEN CPU_SETSIZE +# elif defined(_SC_NPROCESSORS_CONF) /* Cygwin */ +# define QEMUD_CPUMASK_LEN (sysconf(_SC_NPROCESSORS_CONF)) +# else +# error "Port me" +# endif typedef struct _virQEMUCloseCallbacks virQEMUCloseCallbacks; typedef virQEMUCloseCallbacks *virQEMUCloseCallbacksPtr; -- 1.8.1.4

2 3

[libvirt] [PATCH] build: cast [ug]id_t when printing
by Eric Blake 30 May '13

30 May '13

This is a recurring problem for cygwin :) For example, see commit 23a4df88. qemu/qemu_driver.c: In function 'qemuStateInitialize': qemu/qemu_driver.c:691:13: error: format '%d' expects type 'int', but argument 8 has type 'uid_t' [-Wformat] * src/qemu/qemu_driver.c (qemuStateInitialize): Add casts. * daemon/remote.c (remoteDispatchAuthList): Likewise. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing under the build-breaker rule. daemon/remote.c | 4 ++-- src/qemu/qemu_driver.c | 16 ++++++++++------ 2 files changed, 12 insertions(+), 8 deletions(-) diff --git a/daemon/remote.c b/daemon/remote.c index 0e253bf..47267c2 100644 --- a/daemon/remote.c +++ b/daemon/remote.c @@ -1,7 +1,7 @@ /* * remote.c: handlers for RPC method calls * - * Copyright (C) 2007-2012 Red Hat, Inc. + * Copyright (C) 2007-2013 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 @@ -2354,7 +2354,7 @@ remoteDispatchAuthList(virNetServerPtr server ATTRIBUTE_UNUSED, } else if (callerUid == 0) { char *ident; if (virAsprintf(&ident, "pid:%lld,uid:%d", - (long long) callerPid, callerUid) < 0) { + (long long) callerPid, (int) callerUid) < 0) { virReportOOMError(); goto cleanup; } diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 4a76f14..d0dee14 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -690,25 +690,29 @@ qemuStateInitialize(bool privileged, if (chown(cfg->libDir, cfg->user, cfg->group) < 0) { virReportSystemError(errno, _("unable to set ownership of '%s' to user %d:%d"), - cfg->libDir, cfg->user, cfg->group); + cfg->libDir, (int) cfg->user, + (int) cfg->group); goto error; } if (chown(cfg->cacheDir, cfg->user, cfg->group) < 0) { virReportSystemError(errno, _("unable to set ownership of '%s' to %d:%d"), - cfg->cacheDir, cfg->user, cfg->group); + cfg->cacheDir, (int) cfg->user, + (int) cfg->group); goto error; } if (chown(cfg->saveDir, cfg->user, cfg->group) < 0) { virReportSystemError(errno, _("unable to set ownership of '%s' to %d:%d"), - cfg->saveDir, cfg->user, cfg->group); + cfg->saveDir, (int) cfg->user, + (int) cfg->group); goto error; } if (chown(cfg->snapshotDir, cfg->user, cfg->group) < 0) { virReportSystemError(errno, _("unable to set ownership of '%s' to %d:%d"), - cfg->snapshotDir, cfg->user, cfg->group); + cfg->snapshotDir, (int) cfg->user, + (int) cfg->group); goto error; } run_uid = cfg->user; @@ -752,8 +756,8 @@ qemuStateInitialize(bool privileged, if (chown(mempath, cfg->user, cfg->group) < 0) { virReportSystemError(errno, _("unable to set ownership on %s to %d:%d"), - mempath, cfg->user, - cfg->group); + mempath, (int) cfg->user, + (int) cfg->group); goto error; } } -- 1.8.1.4

1 0

[libvirt] [PATCH] build: port vbox to cygwin
by Eric Blake 30 May '13

30 May '13

I have no idea if a Cygwin app can be made to directly interact with VBoxXPCOMC.dll, but this at least lets compilation of vbox finish rather than requiring me to ./configure --without-vbox. * src/vbox/vbox_XPCOMCGlue.c (DYNLIB_NAME): Assume .dll under cygwin. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Although I'm tempted to push this under the build-breaker rule, and although I suspect no one else is trying to build libvirt on cygwin, I'll wait for a review on this one. An alternative, more conservative, patch might be to hack configure.ac to declare vbox and cygwin as incompatible. src/vbox/vbox_XPCOMCGlue.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/vbox/vbox_XPCOMCGlue.c b/src/vbox/vbox_XPCOMCGlue.c index 9719014..9cc41b0 100644 --- a/src/vbox/vbox_XPCOMCGlue.c +++ b/src/vbox/vbox_XPCOMCGlue.c @@ -3,6 +3,7 @@ */ /* + * Copyright (C) 2013 Red Hat, Inc. * Copyright (C) 2008-2009 Sun Microsystems, Inc. * * This file is part of a free software library; you can redistribute @@ -54,7 +55,7 @@ # define DYNLIB_NAME "VBoxXPCOMC.so" #elif defined(__APPLE__) # define DYNLIB_NAME "VBoxXPCOMC.dylib" -#elif defined(_MSC_VER) || defined(__OS2__) +#elif defined(_MSC_VER) || defined(__OS2__) || defined(__CYGWIN__) # define DYNLIB_NAME "VBoxXPCOMC.dll" #else # error "Port me" -- 1.8.1.4

2 2

[libvirt] [PATCH 1/3] qemu: migration: Improve p2p error if we can't open conn
by Cole Robinson 30 May '13

30 May '13

By actually showing the Open() error to the user --- src/qemu/qemu_migration.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_migration.c b/src/qemu/qemu_migration.c index 19b1236..9ac9be4 100644 --- a/src/qemu/qemu_migration.c +++ b/src/qemu/qemu_migration.c @@ -3472,7 +3472,8 @@ static int doPeer2PeerMigrate(virQEMUDriverPtr driver, qemuDomainObjExitRemote(vm); if (dconn == NULL) { virReportError(VIR_ERR_OPERATION_FAILED, - _("Failed to connect to remote libvirt URI %s"), dconnuri); + _("Failed to connect to remote libvirt URI %s: %s"), + dconnuri, virGetLastErrorMessage()); virObjectUnref(cfg); return -1; } -- 1.8.2.1

3 9

[libvirt] [PATCH] build: work around broken sasl header
by Eric Blake 29 May '13

29 May '13

Compilation on cygwin failed due to a bug in the sasl headers present on that platform (libsasl2-devel 2.1.26): In file included from rpc/virnetserverclient.c:27:0: /usr/include/sasl/sasl.h:230:38: error: expected declaration specifiers or '...' before 'size_t' * src/rpc/virnetserverclient.c (includes): Ensure size_t is defined before using sasl.h. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing under the build-breaker rule. src/rpc/virnetserverclient.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/rpc/virnetserverclient.c b/src/rpc/virnetserverclient.c index 0bd88ad..2fc4838 100644 --- a/src/rpc/virnetserverclient.c +++ b/src/rpc/virnetserverclient.c @@ -1,7 +1,7 @@ /* * virnetserverclient.c: generic network RPC server client * - * Copyright (C) 2006-2011 Red Hat, Inc. + * Copyright (C) 2006-2013 Red Hat, Inc. * Copyright (C) 2006 Daniel P. Berrange * * This library is free software; you can redistribute it and/or @@ -23,6 +23,7 @@ #include <config.h> +#include "internal.h" #if WITH_SASL # include <sasl/sasl.h> #endif -- 1.8.1.4

1 0

[libvirt] [PATCH] build: fix build without libvirtd
by Eric Blake 29 May '13

29 May '13

Building when configured --with-libvirtd=no fails with: In file included from ../src/qemu/qemu_command.h:30:0, from testutilsqemu.h:4, from networkxml2xmltest.c:14: ../src/qemu/qemu_conf.h:175:5: error: expected specifier-qualifier-list before 'virStateInhibitCallback' * src/libvirt_internal.h (virStateInhibitCallback): Move outside of conditional. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing under build-breaker rule src/libvirt_internal.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/libvirt_internal.h b/src/libvirt_internal.h index 5c83ffa..29f2043 100644 --- a/src/libvirt_internal.h +++ b/src/libvirt_internal.h @@ -1,7 +1,7 @@ /* - * libvirt.h: publically exported APIs, not for public use + * libvirt_internal.h: internally exported APIs, not for public use * - * Copyright (C) 2006-2008, 2011 Red Hat, Inc. + * Copyright (C) 2006-2013 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 @@ -27,10 +27,10 @@ # include "internal.h" -# ifdef WITH_LIBVIRTD typedef void (*virStateInhibitCallback)(bool inhibit, void *opaque); +# ifdef WITH_LIBVIRTD int virStateInitialize(bool privileged, virStateInhibitCallback inhibit, void *opaque); -- 1.8.1.4

1 0

[libvirt] [PATCH] build: fix build with newer gnutls
by Eric Blake 29 May '13

29 May '13

Building with gnutls 3.2.0 (such as shipped with current cygwin) fails with: rpc/virnettlscontext.c: In function 'virNetTLSSessionGetKeySize': rpc/virnettlscontext.c:1358:5: error: implicit declaration of function 'gnutls_cipher_get_key_size' [-Wimplicit-function-declaration] Yeah, it's stupid that gnutls broke API by moving their declaration into a new header without including that header from the old one, but it's easy enough to work around, all without breaking on gnutls 1.4.1 (hello RHEL 5) that lacked the new header. * src/rpc/virnettlscontext.c (includes): Include additional header. * configure.ac (gnutls): Check for <gnutls/crypto.h>. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing under the build-breaker rule, once my testing completes across multiple systems with various gnutls versions (1.4.1 in RHEL5, 2.12.23 in Fedora 18, 3.2.0 in Cygwin). configure.ac | 1 + src/rpc/virnettlscontext.c | 3 +++ 2 files changed, 4 insertions(+) diff --git a/configure.ac b/configure.ac index 7041710..0b3d028 100644 --- a/configure.ac +++ b/configure.ac @@ -1087,6 +1087,7 @@ if test "x$with_gnutls" != "xno"; then if test "$GNUTLS_FOUND" = "no"; then fail=0 AC_CHECK_HEADER([gnutls/gnutls.h], [], [fail=1]) + AC_CHECK_HEADERS([gnutls/crypto.h]) AC_CHECK_LIB([gnutls], [gnutls_handshake],[], [fail=1], [-lgcrypt]) test $fail = 0 && GNUTLS_FOUND=yes diff --git a/src/rpc/virnettlscontext.c b/src/rpc/virnettlscontext.c index e29c439..f2ac551 100644 --- a/src/rpc/virnettlscontext.c +++ b/src/rpc/virnettlscontext.c @@ -25,6 +25,9 @@ #include <stdlib.h> #include <gnutls/gnutls.h> +#if HAVE_GNULTLS_CRYPTO_H +# include <gnutls/crypto.h> +#endif #include <gnutls/x509.h> #include "gnutls_1_0_compat.h" -- 1.8.1.4

1 2

Re: [libvirt] [PATCH] target-i386: Disable CPUID_EXT_MONITOR when KVM is enabled
by Eduardo Habkost 29 May '13

29 May '13

(CCing libvirt people) On Tue, May 28, 2013 at 06:48:52PM +0200, Andreas Färber wrote: > Am 28.05.2013 18:46, schrieb Paolo Bonzini: > > Il 28/05/2013 18:34, Bandan Das ha scritto: > >> Eduardo Habkost <ehabkost(a)redhat.com> writes: > >> > >>> On Mon, May 27, 2013 at 02:21:36PM +0200, Paolo Bonzini wrote: > >>>> Il 27/05/2013 14:09, Eduardo Habkost ha scritto: > >>>>> On Sat, May 25, 2013 at 08:25:49AM +0200, Paolo Bonzini wrote: > >>>>>> Il 25/05/2013 03:21, Bandan Das ha scritto: > >>>>>>> There is one user-visible effect: "-cpu ...,enforce" will stop failing > >>>>>>> because of missing KVM support for CPUID_EXT_MONITOR. But that's exactly > >>>>>>> the point: there's no point in having CPU model definitions that would > >>>>>>> never work as-is with neither TCG or KVM. This patch is changing the > >>>>>>> meaning of (e.g.) "-machine ...,accel=kvm -cpu Opteron_G3" to match what > >>>>>>> was already happening in practice. > >>>>>> > >>>>>> But then -cpu Opteron_G3 does not match a "real" Opteron G3. Is it > >>>>>> worth it? > >>>>> > >>>>> No models match a "real" CPU this way, because neither TCG or KVM > >>>>> support all features supported by a real CPU. I ask the opposite > >>>>> question: is it worth maintaining an "accurate" CPU model definition > >>>>> that would never work without feature-bit tweaking in the command-line? > >>>> > >>>> It would work with TCG. Changing TCG to KVM should not change hardware > >>>> if you use "-cpu ...,enforce", so it is right that it fails when > >>>> starting with KVM. > >>>> > >>> > >>> Changing between KVM and TCG _does_ change hardware, today (with or > >>> without check/enforce). All CPU models on TCG have features not > >>> supported by TCG automatically removed. See the "if (!kvm_enabled())" > >>> block at x86_cpu_realizefn(). > >> > >> Yes, this is exactly why I was inclined to remove the monitor flag. > >> We already have uses of kvm_enabled() to set (or remove) kvm specific stuff, > >> and this change is no different. > > > > Do any of these affect something that is part of x86_def_t? > > The vendor comes to mind. I believe we can still consider the "vendor" field a special one: if other components care about the TCG/KVM difference regarding the "vendor" field, they can simply set "vendor" explicitly on the command-line. > >> I can see Paolo's point though, having > >> a common definition probably makes sense too. > > Paolo is convincing me that keeping the rest of the features exactly the same on TCG and KVM modes (and making check/enforce work for TCG as well) would simplify the logic a lot. This will add a little extra work for libvirt, that will probably need to use "-cpu Opteron_G3,-monitor" once it implements enforce-mode (to make sure the results really match existing libvirt assumptions about the Opteron_G* models), but it is probably worth it. I will give it a try and send a proposal soon. > >>> (That's why I argue that we need separate classes/names for TCG and KVM > >>> modes. Otherwise our predefined models get less useful as they will > >>> require low-level feature-bit fiddling on the libvirt side to make them > >>> work as expected.) > >> > >> Agreed. From a user's perspective, I think the more a CPU model "just works", > >> whether it's KVM or TCG, the better. > > > > Yes, that's right. But I think extending the same expectation to "-cpu > > ...,enforce" is not necessary, and perhaps even wrong for "-cpu > > ...,check" since it's only a warning rather than a fatal error. > > > > Paolo > > > > > -- > SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany > GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg -- Eduardo

1 0

[libvirt] [PATCH] qemu: match vcpu topology to maximum vcpu numbers.
by Guannan Ren 29 May '13

29 May '13

For qemu, if the -smp N or the value of maxcpus is given, it define the number of vcpu of guest whenever the vcpu topology is defined or not. But if the -smp N and maxcpus are missing, the topology can compute and define vcpus for guest automatically by math: vcpu number = sockets*cores*threads For libvirt, as <vcpu> is always mandatory, so we can ask topology to match maximum vcpu numbers. --- src/conf/domain_conf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index a9656af..ffdc6da 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -11815,10 +11815,10 @@ virDomainDefParseXML(xmlDocPtr xml, goto error; if (def->cpu->sockets && - def->maxvcpus > + def->maxvcpus != def->cpu->sockets * def->cpu->cores * def->cpu->threads) { virReportError(VIR_ERR_XML_DETAIL, "%s", - _("Maximum CPUs greater than topology limit")); + _("Topology limit does not match maximum CPUs")); goto error; } -- 1.8.1.4

2 7