- Devel - libvirt List Archives

[Libvir] don't clobber user-specified CFLAGS
by Jim Meyering 26 Mar '07

26 Mar '07

Building libvirt with CFLAGS=-g doesn't work as expected: gcc runs without -g. That was due to the following assignment that gives CFLAGS the value of an undefined variable. 2007-03-23 Jim Meyering <jim(a)meyering.net> * acinclude.m4: Remove bogus CFLAGS=$undefined_var assignment that would clobber any CFLAGS setting from e.g., ./autogen.sh CFLAGS=-g. Reported by David Lutterkort. Index: acinclude.m4 =================================================================== RCS file: /data/cvs/libvirt/acinclude.m4,v retrieving revision 1.3 diff -u -p -r1.3 acinclude.m4 --- acinclude.m4 21 Mar 2007 15:22:31 -0000 1.3 +++ acinclude.m4 23 Mar 2007 18:52:55 -0000 @@ -59,8 +59,6 @@ AC_DEFUN([LIBVIRT_COMPILE_WARNINGS],[ unset option unset try_compiler_flags - CFLAGS="$realsave_CFLAGS" - AC_ARG_ENABLE(iso-c, AC_HELP_STRING([--enable-iso-c], [Try to warn if code is not ISO C ]),,

2 1

[Libvir] [PATCH] Add the message when a little memory is set with setmaxmem.
by Masayuki Sunou 26 Mar '07

26 Mar '07

Hi The message which a cause of an error is hard to detect is displayed when "virsh setmaxmem" sets the maximum memory of an active domain less than Used Memory. -------------------------------------------------------------------------------- # virsh setmaxmem 0 4096 libvir: Xen error : failed Xen syscall ioctl 8518692 libvir: Xen Daemon error : POST operation failed: (xend.err "(22, 'Invalid argument')") libvir: error : library call virDomainSetMaxMemory failed, possibly not supported -------------------------------------------------------------------------------- This patch displays the message which is easier to detect the cause of an error to a user. -------------------------------------------------------------------------------- # virsh setmaxmem 0 4096 error: 4096 is less than current used memory. -------------------------------------------------------------------------------- Signed-off-by: Masayuki Sunou <fj1826dm(a)aa.jp.fujitsu.com> Thanks, Masayuki Sunou. -------------------------------------------------------------------------------- Index: libvirt/src/virsh.c =================================================================== RCS file: /data/cvs/libvirt/src/virsh.c,v retrieving revision 1.67 diff -u -p -r1.67 virsh.c --- libvirt/src/virsh.c 20 Mar 2007 15:31:46 -0000 1.67 +++ libvirt/src/virsh.c 22 Mar 2007 02:27:16 -0000 @@ -1493,6 +1493,7 @@ static int cmdSetmaxmem(vshControl * ctl, vshCmd * cmd) { virDomainPtr dom; + virDomainInfo info; int bytes; int ret = TRUE; @@ -1509,6 +1510,18 @@ cmdSetmaxmem(vshControl * ctl, vshCmd * return FALSE; } + if (virDomainGetID(dom) != ((unsigned int)-1)) { + if (virDomainGetInfo(dom, &info) != 0) { + virDomainFree(dom); + return FALSE; + } + if (bytes < info.memory) { + vshError(ctl, FALSE, _("%d is less than current used memory."), bytes); + virDomainFree(dom); + return FALSE; + } + } + if (virDomainSetMaxMemory(dom, bytes) != 0) { ret = FALSE; } --------------------------------------------------------------------------------

2 3

[Libvir] sprintf cleanup patch
by Daniel Veillard 23 Mar '07

23 Mar '07

Sprintf is potentially dangerous, so I replaced all instances with snprintf (even in cases where the usage was garanteed to be safe to have a coherent approach). Pending patch appended, Daniel -- Red Hat Virtualization group http://redhat.com/virtualization/ Daniel Veillard | virtualization library http://libvirt.org/ veillard(a)redhat.com | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/

2 1

[Libvir] Python bindings, errors & exceptions
by Richard W.M. Jones 23 Mar '07

23 Mar '07

I'm really confused about how the Python bindings are supposed to handle errors. Can someone explain how this is supposed to work? Case in point: currently virt-manager fails because conn.listNetworks () returns None, whereas virt-manager is expecting it to return a list of network objects. The code returns None because the underlying calls (either virConnectNumOfNetworks or virConnectListNetworks) is failing. The functions in question are: class virConnect: # libvirtclass.py # ... def listNetworks(self): """list the networks, stores the pointers to the names in @names """ ret = libvirtmod.virConnectListNetworks(self._o) return ret (the above code is automatically generated by generator.py), and: static PyObject * // libvir.c libvirt_virConnectListNetworks(PyObject *self ATTRIBUTE_UNUSED, PyObject *args) { PyObject *py_retval; char **names = NULL; int c_retval, i; virConnectPtr conn; PyObject *pyobj_conn; // ... c_retval = virConnectNumOfNetworks(conn); if (c_retval < 0) { Py_INCREF(Py_None); return (Py_None); } // ... py_retval = PyList_New(c_retval); if (names) { for (i = 0;i < c_retval;i++) { PyList_SetItem(py_retval, i, libvirt_constcharPtrWrap(names[i])); free(names[i]); } free(names); } return(py_retval); } (I've omitted some code to make the general idea clearer). The upshot is that if the underlying functions fail, at no point is an exception thrown. This is not always the case. For other functions which return C structure pointers (eg. libvirt.open which wraps virConnectOpen), the bindings automatically catch the invalid return and throw an exception. For example: def open(name): # libvirtclass.py """This function should be called first to get a connection to the Hypervisor and xen store """ ret = libvirtmod.virConnectOpen(name) if ret is None:raise libvirtError('virConnectOpen() failed') return virConnect(_obj=ret) It is my view that all errors in C code should turn into Python exceptions. One way to do that would be to have a Python virterror handler which just directly throws the exception. I don't know if this is safe because the exception would unwind through C code, and in some languages that is safe, in others it is not. Another way would be to have a Python virterror handler which remembers that an exception happened, and after each auto-generated function call we check this and raise the exception. Since I'm just starting out in Python, I don't know if there are thread or other issues with this. Comments would be welcome. Rich. -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 "[Negative numbers] darken the very whole doctrines of the equations and make dark of the things which are in their nature excessively obvious and simple" (Francis Maseres FRS, mathematician, 1759)

3 6

[Libvir] [PATCH] Xen capabilities does not detect PAE properly
by Richard W.M. Jones 23 Mar '07

23 Mar '07

-- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 "[Negative numbers] darken the very whole doctrines of the equations and make dark of the things which are in their nature excessively obvious and simple" (Francis Maseres FRS, mathematician, 1759)

2 1

Re: [Libvir] [patch 1/5] iptables: fix invalid free
by Daniel Veillard 22 Mar '07

22 Mar '07

On Wed, Mar 21, 2007 at 12:47:58PM +0000, Mark McLoughlin wrote: > In iptablesContextNew(), make sure we don't try and free an invalid > pointer if one of the iptRulesNew() fails. > > Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> > > Index: libvirt/qemud/iptables.c > =================================================================== > --- libvirt.orig/qemud/iptables.c > +++ libvirt/qemud/iptables.c > @@ -496,7 +496,7 @@ iptablesContextNew(void) > { > iptablesContext *ctx; > > - if (!(ctx = (iptablesContext *) malloc(sizeof (iptablesContext)))) > + if (!(ctx = (iptablesContext *) calloc(1, sizeof (iptablesContext)))) > return NULL; > > if (!(ctx->input_filter = iptRulesNew("filter", IPTABLES_PREFIX "INPUT"))) I usually prefer malloc + memset( , 0, ) , but this probably comes from libxml2 where I replaced malloc calls with specific wrappers (and I still have a TODO for this in libvirt though some part of libvirt are not linked to libxml2 I guess so that may make things a bit harder) What's the policy w.r.t. error reporting in qemud and libvirt related daemons in general ? I guess a failure to malloc or thisd kind of problems should be logged somewhere, right ? > @@ -518,9 +518,12 @@ iptablesContextNew(void) > void > iptablesContextFree(iptablesContext *ctx) > { > - iptRulesFree(ctx->input_filter); > - iptRulesFree(ctx->forward_filter); > - iptRulesFree(ctx->nat_postrouting); > + if (ctx->input_filter) > + iptRulesFree(ctx->input_filter); > + if (ctx->forward_filter) > + iptRulesFree(ctx->forward_filter); > + if (ctx->nat_postrouting) > + iptRulesFree(ctx->nat_postrouting); > free(ctx); > } The patch does the right thing, sounds good to me :-) Daniel -- Red Hat Virtualization group http://redhat.com/virtualization/ Daniel Veillard | virtualization library http://libvirt.org/ veillard(a)redhat.com | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/

5 18

[Libvir] [PATCH] Xen capabilities - nul-terminate the token and a couple of cleanups
by Richard W.M. Jones 22 Mar '07

22 Mar '07

-- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 "[Negative numbers] darken the very whole doctrines of the equations and make dark of the things which are in their nature excessively obvious and simple" (Francis Maseres FRS, mathematician, 1759)

1 0

[Libvir] Fix buffer overflow in dumping XML
by Daniel P. Berrange 22 Mar '07

22 Mar '07

The new bufferContentAndFree() method used for the QEMU daemon rellocs the buffer size down to release memory held by the buffer which was never used for any data. Unfortunately it reallocs it 1 byte too small, so later uses of strlen()/strcpy() either magically work, or randomly append gargage or crash the daemon depending on the phase of the moon :-) Re-allocing the buffer to relase a few bytes memory isn't really an optimization since the caller is going to free the entire block a very short while later, so this patch simply removes the realloc call. As an aside, the virBuffer functions in src/xml.c and the buffer functions in qemud/buf.c are both flawed wrt to the way they call the Grow method. The method expects the len parameter to be extra bytes needed, but several of the callers pass in the total desired length, so it allocates too much memory. There are various other non-fatal flaws which need to be cleaned up in this code, but the attached patch just focuses on the current fatal buffer overflow for now. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

3 2

[Libvir] [PATCH] virsh vcpupin guards CPU not existing.
by Masayuki Sunou 22 Mar '07

22 Mar '07

Hi When CPU not existing is set to virsh vcpupin, CPU affinity is set to all CPU. -------------------------------------------------------------------------------- # virsh vcpuinfo 0 VCPU: 0 CPU: 0 State: blocked CPU time: 428.8s CPU Affinity: y- # virsh vcpupin 0 0 100 libvir: Xen error : failed Xen syscall ioctl -1208339792 # virsh vcpuinfo 0 VCPU: 0 CPU: 0 State: blocked CPU time: 429.6s CPU Affinity: yy -------------------------------------------------------------------------------- This patch returns an error, when CPU not existing is set to virsh vcpupin. -------------------------------------------------------------------------------- # virsh vcpupin 0 0 100 error: Physical CPU 100 doesn't exist. -------------------------------------------------------------------------------- Signed-off-by: Masayuki Sunou <fj1826dm(a)aa.jp.fujitsu.com> Thanks, Masayuki Sunou. -------------------------------------------------------------------------------- Index: src/virsh.c =================================================================== RCS file: /data/cvs/libvirt/src/virsh.c,v retrieving revision 1.67 diff -u -p -r1.67 virsh.c --- src/virsh.c 20 Mar 2007 15:31:46 -0000 1.67 +++ src/virsh.c 22 Mar 2007 06:05:14 -0000 @@ -1358,6 +1358,11 @@ cmdVcpupin(vshControl * ctl, vshCmd * cm if (cpu < VIR_NODEINFO_MAXCPUS(nodeinfo)) { VIR_USE_CPU(cpumap, cpu); + } else { + vshError(ctl, FALSE, _("Physical CPU %d doesn't exist."), cpu); + free(cpumap); + virDomainFree(dom); + return FALSE; } cpulist = index(cpulist, ','); if (cpulist) --------------------------------------------------------------------------------

2 1

[Libvir] [PATCH] Add Xen Interface Version Check
by Atsushi SAKAI 22 Mar '07

22 Mar '07

Add Xen Interface version Check XEN_SYSCTL_INTERFACE_VERSION XEN_DOMCTL_INTERFACE_VERSION Currently following distribution is considered. sys dom RHEL5.0 2 3 Fedora 7 2 4 Xen-unstable 3 5 Please check virVcpuInfoPtr ipt; error message is collect or not + fprintf(stderr, "Memory allocation failed at xenHypervisorIniti()\n"); Signed-off-by: Atsushi SAKAI <sakaia(a)jp.fujitsu.com> Thanks Atsushi SAKAI --- libvirt-0.2.1/src/xen_internal.c 2007-03-16 06:36:41.000000000 +0900 +++ libvirt-0.2.1.hyper/src/xen_internal.c 2007-03-19 19:09:08.000000000 +0900 @@ -655,7 +655,6 @@ return (0); } -#ifndef PROXY /** * xenHypervisorDoV2Dom: * @handle: the handle to the Xen hypervisor @@ -697,7 +696,6 @@ return (0); } -#endif /* PROXY */ /** * virXen_getdomaininfolist: @@ -1057,6 +1055,8 @@ } return(ret); } +#endif /* !PROXY*/ + /** * virXen_getvcpusinfo: * @handle: the hypervisor handle @@ -1169,7 +1169,6 @@ } return(ret); } -#endif /* !PROXY*/ /** * xenHypervisorInit: @@ -1184,6 +1183,7 @@ hypercall_t hc; v0_hypercall_t v0_hc; xen_getdomaininfo info; + virVcpuInfoPtr ipt; if (initialized) { if (hypervisor_version == -1) @@ -1291,15 +1291,47 @@ * or the old ones */ hypervisor_version = 2; - /* TODO: one probably will need to autodetect thse subversions too */ + + ipt = malloc(sizeof(virVcpuInfo)); + if (ipt == NULL){ +#ifdef DEBUG + fprintf(stderr, "Memory allocation failed at xenHypervisorIniti()\n"); +#endif + return(-1); + } + /* Currently consider RHEL5.0 Fedora7 and xen-unstable */ sys_interface_version = 2; /* XEN_SYSCTL_INTERFACE_VERSION */ - dom_interface_version = 3; /* XEN_DOMCTL_INTERFACE_VERSION */ if (virXen_getdomaininfo(fd, 0, &info) == 1) { + /* RHEL 5.0 */ + dom_interface_version = 3; /* XEN_DOMCTL_INTERFACE_VERSION */ + if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){ #ifdef DEBUG - fprintf(stderr, "Using hypervisor call v2, sys version 2\n"); + fprintf(stderr, "Using hypervisor call v2, sys ver2 dom ver3\n"); #endif - goto done; + goto done; + } + /* Fedora 7 */ + dom_interface_version = 4; /* XEN_DOMCTL_INTERFACE_VERSION */ + if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){ +#ifdef DEBUG + fprintf(stderr, "Using hypervisor call v2, sys ver2 dom ver4\n"); +#endif + goto done; + } } + + sys_interface_version = 3; /* XEN_SYSCTL_INTERFACE_VERSION */ + if (virXen_getdomaininfo(fd, 0, &info) == 1) { + /* xen-unstable */ + dom_interface_version = 5; /* XEN_DOMCTL_INTERFACE_VERSION */ + if (virXen_getvcpusinfo(fd, 0, 0, ipt, NULL, 0) == 0){ +#ifdef DEBUG + fprintf(stderr, "Using hypervisor call v2, sys ver3 dom ver5\n"); +#endif + goto done; + } + } + hypervisor_version = 1; sys_interface_version = -1; if (virXen_getdomaininfo(fd, 0, &info) == 1) {

2 4