- Devel - libvirt List Archives

[libvirt] [PATCH] Ensure that /dev exists in the container root filesystem
by Daniel P. Berrange 19 Aug '13

19 Aug '13

From: "Daniel P. Berrange" <berrange(a)redhat.com> If booting a container with a root FS that isn't the host's root, we must ensure that the /dev mount point exists. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/lxc/lxc_container.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) diff --git a/src/lxc/lxc_container.c b/src/lxc/lxc_container.c index 0ab4026..678a9d5 100644 --- a/src/lxc/lxc_container.c +++ b/src/lxc/lxc_container.c @@ -868,7 +868,7 @@ static int lxcContainerMountProcFuse(virDomainDefPtr def ATTRIBUTE_UNUSED, static int lxcContainerMountFSDev(virDomainDefPtr def, const char *stateDir) { - int ret; + int ret = -1; char *path = NULL; VIR_DEBUG("Mount /dev/ stateDir=%s", stateDir); @@ -877,14 +877,24 @@ static int lxcContainerMountFSDev(virDomainDefPtr def, stateDir, def->name)) < 0) return ret; + if (virFileMakePath("/dev") < 0) { + virReportSystemError(errno, "%s", + _("Cannot create /dev")); + goto cleanup; + } + VIR_DEBUG("Tring to move %s to /dev", path); - if ((ret = mount(path, "/dev", NULL, MS_MOVE, NULL)) < 0) { + if (mount(path, "/dev", NULL, MS_MOVE, NULL) < 0) { virReportSystemError(errno, _("Failed to mount %s on /dev"), path); + goto cleanup; } + ret = 0; + +cleanup: VIR_FREE(path); return ret; } -- 1.8.3.1

3 2

[libvirt] [PATCH] util: Fix crash of libvirtd when running numatune with invalid nodeset
by Alex Jia 19 Aug '13

19 Aug '13

This issue is introduced by commit 0fc8909, the virBitmapIsSet() needs caller to ensure 'b < bitmap->max_bit', but it's lost in the virBitmapParse() caller, this will cause crash of libvirtd, with the patch, libvirtd no crash and can get a expected error "Failed to parse nodeset". How to reproduce? # virsh numatune foo --nodeset 1000000000 Actual result: error: Unable to change numa parameters error: End of file while reading data: Input/output error error: One or more references were leaked after disconnect from the hypervisor error: Failed to reconnect to the hypervisor GDB backtrace: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7f80ee533700 (LWP 4156)] 0x00007f80fb1b8906 in virBitmapIsSet (str=<value optimized out>, terminator=0 '\000', bitmap=0x7f80ee532978, bitmapSize=<value optimized out>) at util/virbitmap.c:158 158 return !!(bitmap->map[VIR_BITMAP_UNIT_OFFSET(b)] & VIR_BITMAP_BIT(b)); (gdb) bt #0 0x00007f80fb1b8906 in virBitmapIsSet (str=<value optimized out>, terminator=0 '\000', bitmap=0x7f80ee532978, bitmapSize=<value optimized out>) at util/virbitmap.c:158 #1 virBitmapParse (str=<value optimized out>, terminator=0 '\000', bitmap=0x7f80ee532978, bitmapSize=<value optimized out>) at util/virbitmap.c:351 #2 0x00007f80e5e7ef6b in qemuDomainSetNumaParameters (dom=<value optimized out>, params=<value optimized out>, nparams=1, flags=2) at qemu/qemu_driver.c:8357 #3 0x00007f80fb2ad7c6 in virDomainSetNumaParameters (domain=0x7f80d4000930, params=0x7f80d4000b00, nparams=1, flags=0) at libvirt.c:4163 #4 0x00007f80fbcd17ec in remoteDispatchDomainSetNumaParameters (server=<value optimized out>, client=<value optimized out>, msg=<value optimized out>, rerr=0x7f80ee532ba0, args=<value optimized out>, ret=<value optimized out>) at remote_dispatch.h:7771 #5 remoteDispatchDomainSetNumaParametersHelper (server=<value optimized out>, client=<value optimized out>, msg=<value optimized out>, rerr=0x7f80ee532ba0, args=<value optimized out>, ret=<value optimized out>) at remote_dispatch.h:7741 #6 0x00007f80fb2f6518 in virNetServerProgramDispatchCall (prog=0x7f80fd33a990, server=0x7f80fd32f3e0, client=0x7f80fd33fe30, msg=0x7f80fd33e550) at rpc/virnetserverprogram.c:435 #7 virNetServerProgramDispatch (prog=0x7f80fd33a990, server=0x7f80fd32f3e0, client=0x7f80fd33fe30, msg=0x7f80fd33e550) at rpc/virnetserverprogram.c:305 #8 0x00007f80fb2f9dd6 in virNetServerProcessMsg (srv=<value optimized out>, client=0x7f80fd33fe30, prog=<value optimized out>, msg=0x7f80fd33e550) at rpc/virnetserver.c:165 #9 0x00007f80fb2fa9f3 in virNetServerHandleJob (jobOpaque=<value optimized out>, opaque=<value optimized out>) at rpc/virnetserver.c:186 #10 0x00007f80fb20210e in virThreadPoolWorker (opaque=<value optimized out>) at util/virthreadpool.c:144 #11 0x00007f80fb2016f6 in virThreadHelper (data=<value optimized out>) at util/virthreadpthread.c:161 #12 0x00007f80f8d857f1 in start_thread (arg=0x7f80ee533700) at pthread_create.c:301 #13 0x00007f80f86bb70d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115 RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=997367 Signed-off-by: Alex Jia <ajia(a)redhat.com> --- The caller virBitmapGetBit() can make sure 'b < bitmap->max_bit', so don't need to worry about higher caller for the virBitmapGetBit(), but the virBitmapParse() is called by many XML parser function, not sure which one can crash libvirtd with read-only client then probably require a CVE, I haven't a good way to check them now and only manually check them one by one. src/util/virbitmap.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/src/util/virbitmap.c b/src/util/virbitmap.c index 7e1cd02..edbfb30 100644 --- a/src/util/virbitmap.c +++ b/src/util/virbitmap.c @@ -337,6 +337,9 @@ virBitmapParse(const char *str, if (start < 0) goto parse_error; + if ((*bitmap)->max_bit <= start) + goto parse_error; + cur = tmp; virSkipSpaces(&cur); -- 1.7.1

3 3

[libvirt] RFC: Introduce API to return configuration/state paths of the network driver
by Nehal J. Wani 18 Aug '13

18 Aug '13

Currently, there is no API which returns configuration/state paths of the network driver. Although it is a private implementation of the network driver, I don't see any harm in making the locations public because although the locations might change, there will always be a location for these files. There is a need for this API to implement method 2 of the "API to query ip addresses of a given domain", refer: http://www.mail-archive.com/libvir-list@redhat.com/msg79793.html . It is required to parse the leases file generated by dnsmasq. So, this API will be used by the qemu driver, but it can also be made public, so that, if a user wants to know get some information from a configuration file, he can get the location from libvirt and analyze it on his own. Right now, there is an alternate way to get the info: by using networkDnsmasqLeaseFileNameDefault, defined in /src/network/bridge_driver.c Since this function is static, it is part of the private implementation and not visible outside. To make it public, the following hack is possible: diff --git a/src/network/bridge_driver.c b/src/network/bridge_driver.c index a7ff602..7274861 100644 --- a/src/network/bridge_driver.c +++ b/src/network/bridge_driver.c @@ -124,7 +124,7 @@ static int networkUnplugBandwidth(virNetworkObjPtr net, static struct network_driver *driverState = NULL; -static char * +char * networkDnsmasqLeaseFileNameDefault(const char *netname) { char *leasefile; diff --git a/src/network/bridge_driver.h b/src/network/bridge_driver.h index 50258b5..40e3990 100644 --- a/src/network/bridge_driver.h +++ b/src/network/bridge_driver.h @@ -49,6 +49,8 @@ int networkDnsmasqConfContents(virNetworkObjPtr network, char **configstr, dnsmasqContext *dctx, dnsmasqCapsPtr caps); +char * networkDnsmasqLeaseFileNameDefault(const char *netname) + ATTRIBUTE_NONNULL(1); # else /* Define no-op replacements that don't drag in any link dependencies. */ # define networkAllocateActualDevice(iface) 0 @@ -57,6 +59,7 @@ int networkDnsmasqConfContents(virNetworkObjPtr network, # define networkGetNetworkAddress(netname, netaddr) (-2) # define networkDnsmasqConfContents(network, pidfile, configstr, \ dctx, caps) 0 +# define networkDnsmasqLeaseFileNameDefault(netname) 0 # endif typedef char *(*networkDnsmasqLeaseFileNameFunc)(const char *netname); Similar hack has been used so that networkAllocateActualDevice() can be called from qemu_command.c. Although the above method works, we want to have a formal API and not leave things like a hack. /* * @conn: connection object * @params: array to populate on output * @nparams: number of parameters that will be filled * @flags: not supported, user should pass 0 for now * return 0 on success -1 otherwise * Valid parameter field names: * VIR_NETWORK_CONFIG_DIR, VIR_NETWORK_AUTOSTART_DIR, VIR_STATE_DIR, * VIR_PID_DIR, VIR_DNSMASQ_STATE_DIR, VIR_RADVD_STATE_DIR * All the above will of the type VIR_TYPED_PARAM_STRING */ int virNetworkGetConfigFileName(virConnectPtr conn, virTypedParameterPtr params, int nparams, unsigned int flags) Nehal J. Wani UG3, BTech CS+MS(CL) IIIT-Hyderabad http://commandlinewani.blogspot.com

4 13

[libvirt] [PATCH] tools: Make sure to distribute conf_DATA, fix RPM build
by Cole Robinson 18 Aug '13

18 Aug '13

--- Pushed as build breaker. tools/Makefile.am | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/Makefile.am b/tools/Makefile.am index d48883c..f85c35c 100644 --- a/tools/Makefile.am +++ b/tools/Makefile.am @@ -33,6 +33,7 @@ ICON_FILES = \ EXTRA_DIST = \ $(ICON_FILES) \ + $(conf_DATA) \ virt-xml-validate.in \ virt-pki-validate.in \ virt-sanlock-cleanup.in \ -- 1.8.3.1

1 0

[libvirt] [PATCH 00/10] test: Mock snapshot APIs and misc improvements
by Cole Robinson 17 Aug '13

17 Aug '13

This series implements snapshot APIs for the test driver, and adds some misc improvements, like specifying domain state in the passed in driver XML. Cole Robinson (10): test: Split object parsing into their own functions test: Simplify args passed to testDomainStartState test: Unify object XML parsing test: Allow specifying object runstate in driver XML test: Allow specifying object transient state in driver XML test: Wire up managed save APIs snapshot_conf: Allow parsing an XML node test: Allow specifying <domainsnapshot> XML test: Implement readonly snapshot APIs test: Implement snapshot create/delete/revert APIs src/conf/domain_conf.c | 3 +- src/conf/snapshot_conf.c | 85 +- src/conf/snapshot_conf.h | 6 + src/test/test_driver.c | 4871 +++++++++++++++++++++++++++++----------------- tests/virshtest.c | 2 +- 5 files changed, 3151 insertions(+), 1816 deletions(-) -- 1.8.3.1

2 17

[libvirt] [PATCH] Add flag to BaselineCPU API to return detailed CPU features
by Don Dugger 17 Aug '13

17 Aug '13

Currently the virConnectBaselineCPU API does not expose the CPU features that are part of the CPU's model. This patch adds a new flag, VIR_CONNECT_BASELINE_SHOW_MODEL, that causes the API to explictly list all features that are part of that model. Signed-off-by: Don Dugger <donald.d.dugger(a)intel.com> --- include/libvirt/libvirt.h.in | 9 +++++++ src/cpu/cpu.c | 12 +++++---- src/cpu/cpu.h | 12 ++++++--- src/cpu/cpu_arm.c | 3 ++- src/cpu/cpu_generic.c | 3 ++- src/cpu/cpu_powerpc.c | 6 +++-- src/cpu/cpu_s390.c | 3 ++- src/cpu/cpu_x86.c | 39 ++++++++++++++++++++++++--- src/qemu/qemu_driver.c | 4 +-- tests/cputest.c | 30 +++++++++++---------- tests/cputestdata/x86-baseline-3-result.xml | 35 ++++++++++++++++++++++++ tests/cputestdata/x86-baseline-3.xml | 7 +++++ tools/virsh-domain.c | 11 +++++++- 13 files changed, 140 insertions(+), 34 deletions(-) create mode 100644 tests/cputestdata/x86-baseline-3-result.xml create mode 100644 tests/cputestdata/x86-baseline-3.xml diff --git a/include/libvirt/libvirt.h.in b/include/libvirt/libvirt.h.in index 1804c93..d0afc65 100644 --- a/include/libvirt/libvirt.h.in +++ b/include/libvirt/libvirt.h.in @@ -3890,6 +3890,15 @@ int virConnectCompareCPU(virConnectPtr conn, /** + * virConnectBaselineCPUFlags + * + * Flags when getting XML description of a computed CPU + */ +typedef enum { + VIR_CONNECT_BASELINE_SHOW_MODEL = (1 << 0), /* show model features*/ +} virConnectBaselineCPUFlags; + +/** * virConnectBaselineCPU: * * @conn: virConnect connection diff --git a/src/cpu/cpu.c b/src/cpu/cpu.c index 68125a5..c96f669 100644 --- a/src/cpu/cpu.c +++ b/src/cpu/cpu.c @@ -167,7 +167,7 @@ cpuDecode(virCPUDefPtr cpu, return -1; } - return driver->decode(cpu, data, models, nmodels, preferred); + return driver->decode(cpu, data, models, nmodels, preferred, 0); } @@ -277,7 +277,8 @@ char * cpuBaselineXML(const char **xmlCPUs, unsigned int ncpus, const char **models, - unsigned int nmodels) + unsigned int nmodels, + unsigned int flags) { xmlDocPtr doc = NULL; xmlXPathContextPtr ctxt = NULL; @@ -324,7 +325,7 @@ cpuBaselineXML(const char **xmlCPUs, doc = NULL; } - if (!(cpu = cpuBaseline(cpus, ncpus, models, nmodels))) + if (!(cpu = cpuBaseline(cpus, ncpus, models, nmodels, flags))) goto error; cpustr = virCPUDefFormat(cpu, 0); @@ -353,7 +354,8 @@ virCPUDefPtr cpuBaseline(virCPUDefPtr *cpus, unsigned int ncpus, const char **models, - unsigned int nmodels) + unsigned int nmodels, + unsigned int flags) { struct cpuArchDriver *driver; unsigned int i; @@ -395,7 +397,7 @@ cpuBaseline(virCPUDefPtr *cpus, return NULL; } - return driver->baseline(cpus, ncpus, models, nmodels); + return driver->baseline(cpus, ncpus, models, nmodels, flags); } diff --git a/src/cpu/cpu.h b/src/cpu/cpu.h index cba7149..9148871 100644 --- a/src/cpu/cpu.h +++ b/src/cpu/cpu.h @@ -49,7 +49,8 @@ typedef int const union cpuData *data, const char **models, unsigned int nmodels, - const char *preferred); + const char *preferred, + unsigned int flags); typedef int (*cpuArchEncode) (const virCPUDefPtr cpu, @@ -76,7 +77,8 @@ typedef virCPUDefPtr (*cpuArchBaseline) (virCPUDefPtr *cpus, unsigned int ncpus, const char **models, - unsigned int nmodels); + unsigned int nmodels, + unsigned int /* flags */); typedef int (*cpuArchUpdate) (virCPUDefPtr guest, @@ -145,13 +147,15 @@ extern char * cpuBaselineXML(const char **xmlCPUs, unsigned int ncpus, const char **models, - unsigned int nmodels); + unsigned int nmodels, + unsigned int /* flags */); extern virCPUDefPtr cpuBaseline (virCPUDefPtr *cpus, unsigned int ncpus, const char **models, - unsigned int nmodels); + unsigned int nmodels, + unsigned int /* flags */); extern int cpuUpdate (virCPUDefPtr guest, diff --git a/src/cpu/cpu_arm.c b/src/cpu/cpu_arm.c index cfe1a23..af1309c 100644 --- a/src/cpu/cpu_arm.c +++ b/src/cpu/cpu_arm.c @@ -48,7 +48,8 @@ ArmDecode(virCPUDefPtr cpu ATTRIBUTE_UNUSED, const union cpuData *data ATTRIBUTE_UNUSED, const char **models ATTRIBUTE_UNUSED, unsigned int nmodels ATTRIBUTE_UNUSED, - const char *preferred ATTRIBUTE_UNUSED) + const char *preferred ATTRIBUTE_UNUSED, + unsigned int flags ATTRIBUTE_UNUSED) { return 0; } diff --git a/src/cpu/cpu_generic.c b/src/cpu/cpu_generic.c index 7e3eda2..cad1782 100644 --- a/src/cpu/cpu_generic.c +++ b/src/cpu/cpu_generic.c @@ -115,7 +115,8 @@ static virCPUDefPtr genericBaseline(virCPUDefPtr *cpus, unsigned int ncpus, const char **models, - unsigned int nmodels) + unsigned int nmodels, + unsigned int flags ATTRIBUTE_UNUSED) { virCPUDefPtr cpu = NULL; virCPUFeatureDefPtr features = NULL; diff --git a/src/cpu/cpu_powerpc.c b/src/cpu/cpu_powerpc.c index d17f9ca..cd468dd 100644 --- a/src/cpu/cpu_powerpc.c +++ b/src/cpu/cpu_powerpc.c @@ -310,7 +310,8 @@ ppcDecode(virCPUDefPtr cpu, const union cpuData *data, const char **models, unsigned int nmodels, - const char *preferred ATTRIBUTE_UNUSED) + const char *preferred ATTRIBUTE_UNUSED, + unsigned int flags ATTRIBUTE_UNUSED) { int ret = -1; struct ppc_map *map; @@ -385,7 +386,8 @@ static virCPUDefPtr ppcBaseline(virCPUDefPtr *cpus, unsigned int ncpus, const char **models, - unsigned int nmodels) + unsigned int nmodels, + unsigned int flags ATTRIBUTE_UNUSED) { struct ppc_map *map = NULL; const struct ppc_model *model; diff --git a/src/cpu/cpu_s390.c b/src/cpu/cpu_s390.c index 998197c..32ab2d9 100644 --- a/src/cpu/cpu_s390.c +++ b/src/cpu/cpu_s390.c @@ -50,7 +50,8 @@ s390Decode(virCPUDefPtr cpu ATTRIBUTE_UNUSED, const union cpuData *data ATTRIBUTE_UNUSED, const char **models ATTRIBUTE_UNUSED, unsigned int nmodels ATTRIBUTE_UNUSED, - const char *preferred ATTRIBUTE_UNUSED) + const char *preferred ATTRIBUTE_UNUSED, + unsigned int flags ATTRIBUTE_UNUSED) { return 0; } diff --git a/src/cpu/cpu_x86.c b/src/cpu/cpu_x86.c index 5d479c2..6cb8263 100644 --- a/src/cpu/cpu_x86.c +++ b/src/cpu/cpu_x86.c @@ -1296,13 +1296,43 @@ x86GuestData(virCPUDefPtr host, return x86Compute(host, guest, data, message); } +static void +x86AddFeatures(virCPUDefPtr cpu, + struct x86_map *map) +{ + const struct x86_model *candidate; + const struct x86_feature *feature = map->features; + + candidate = map->models; + while (candidate != NULL) { + if (STREQ(cpu->model, candidate->name)) + break; + candidate = candidate->next; + } + if (!candidate) { + VIR_WARN("Odd, %s not a known CPU model\n", cpu->model); + return; + } + while (feature != NULL) { + if (x86DataIsSubset(candidate->data, feature->data)) { + if (virCPUDefAddFeature(cpu, feature->name, VIR_CPU_FEATURE_REQUIRE) < 0) { + VIR_WARN("CPU model %s, no room for feature %s", cpu->model, feature->name); + return; + } + } + feature = feature->next; + } + return; +} + static int x86Decode(virCPUDefPtr cpu, const union cpuData *data, const char **models, unsigned int nmodels, - const char *preferred) + const char *preferred, + unsigned int flags) { int ret = -1; struct x86_map *map; @@ -1383,6 +1413,8 @@ x86Decode(virCPUDefPtr cpu, goto out; } + if (flags & VIR_CONNECT_BASELINE_SHOW_MODEL) + x86AddFeatures(cpuModel, map); cpu->model = cpuModel->model; cpu->vendor = cpuModel->vendor; cpu->nfeatures = cpuModel->nfeatures; @@ -1610,7 +1642,8 @@ static virCPUDefPtr x86Baseline(virCPUDefPtr *cpus, unsigned int ncpus, const char **models, - unsigned int nmodels) + unsigned int nmodels, + unsigned int flags) { struct x86_map *map = NULL; struct x86_model *base_model = NULL; @@ -1691,7 +1724,7 @@ x86Baseline(virCPUDefPtr *cpus, if (vendor && x86DataAddCpuid(base_model->data, &vendor->cpuid) < 0) goto no_memory; - if (x86Decode(cpu, base_model->data, models, nmodels, NULL) < 0) + if (x86Decode(cpu, base_model->data, models, nmodels, NULL, flags) < 0) goto error; if (!outputVendor) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 4a76f14..b87c73a 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -10260,9 +10260,9 @@ qemuConnectBaselineCPU(virConnectPtr conn ATTRIBUTE_UNUSED, { char *cpu; - virCheckFlags(0, NULL); + virCheckFlags(VIR_CONNECT_BASELINE_SHOW_MODEL, NULL); - cpu = cpuBaselineXML(xmlCPUs, ncpus, NULL, 0); + cpu = cpuBaselineXML(xmlCPUs, ncpus, NULL, 0, flags); return cpu; } diff --git a/tests/cputest.c b/tests/cputest.c index 0105440..227faa0 100644 --- a/tests/cputest.c +++ b/tests/cputest.c @@ -75,6 +75,7 @@ struct data { const char *modelsName; unsigned int nmodels; const char *preferred; + int flags; int result; }; @@ -331,7 +332,7 @@ cpuTestBaseline(const void *arg) if (!(cpus = cpuTestLoadMultiXML(data->arch, data->name, &ncpus))) goto cleanup; - baseline = cpuBaseline(cpus, ncpus, NULL, 0); + baseline = cpuBaseline(cpus, ncpus, NULL, 0, data->flags); if (data->result < 0) { virResetLastError(); if (!baseline) @@ -512,12 +513,12 @@ mymain(void) } #define DO_TEST(arch, api, name, host, cpu, \ - models, nmodels, preferred, result) \ + models, nmodels, preferred, flags, result) \ do { \ static struct data data = { \ arch, api, host, cpu, models, \ models == NULL ? NULL : #models, \ - nmodels, preferred, result \ + nmodels, preferred, flags, result \ }; \ if (cpuTestRun(name, &data) < 0) \ ret = -1; \ @@ -526,31 +527,31 @@ mymain(void) #define DO_TEST_COMPARE(arch, host, cpu, result) \ DO_TEST(arch, API_COMPARE, \ host "/" cpu " (" #result ")", \ - host, cpu, NULL, 0, NULL, result) + host, cpu, NULL, 0, NULL, 0, result) #define DO_TEST_UPDATE(arch, host, cpu, result) \ do { \ DO_TEST(arch, API_UPDATE, \ cpu " on " host, \ - host, cpu, NULL, 0, NULL, 0); \ + host, cpu, NULL, 0, NULL, 0, 0); \ DO_TEST_COMPARE(arch, host, host "+" cpu, result); \ } while (0) -#define DO_TEST_BASELINE(arch, name, result) \ +#define DO_TEST_BASELINE(arch, name, flags, result) \ DO_TEST(arch, API_BASELINE, name, NULL, "baseline-" name, \ - NULL, 0, NULL, result) + NULL, 0, NULL, flags, result) #define DO_TEST_HASFEATURE(arch, host, feature, result) \ DO_TEST(arch, API_HAS_FEATURE, \ host "/" feature " (" #result ")", \ - host, feature, NULL, 0, NULL, result) + host, feature, NULL, 0, NULL, 0, result) #define DO_TEST_GUESTDATA(arch, host, cpu, models, preferred, result) \ DO_TEST(arch, API_GUEST_DATA, \ host "/" cpu " (" #models ", pref=" #preferred ")", \ host, cpu, models, \ models == NULL ? 0 : sizeof(models) / sizeof(char *), \ - preferred, result) + preferred, 0, result) /* host to host comparison */ DO_TEST_COMPARE("x86", "host", "host", VIR_CPU_COMPARE_IDENTICAL); @@ -594,11 +595,12 @@ mymain(void) DO_TEST_UPDATE("x86", "host", "host-passthrough", VIR_CPU_COMPARE_IDENTICAL); /* computing baseline CPUs */ - DO_TEST_BASELINE("x86", "incompatible-vendors", -1); - DO_TEST_BASELINE("x86", "no-vendor", 0); - DO_TEST_BASELINE("x86", "some-vendors", 0); - DO_TEST_BASELINE("x86", "1", 0); - DO_TEST_BASELINE("x86", "2", 0); + DO_TEST_BASELINE("x86", "incompatible-vendors", 0, -1); + DO_TEST_BASELINE("x86", "no-vendor", 0, 0); + DO_TEST_BASELINE("x86", "some-vendors", 0, 0); + DO_TEST_BASELINE("x86", "1", 0, 0); + DO_TEST_BASELINE("x86", "2", 0, 0); + DO_TEST_BASELINE("x86", "3", VIR_CONNECT_BASELINE_SHOW_MODEL, 0); /* CPU features */ DO_TEST_HASFEATURE("x86", "host", "vmx", YES); diff --git a/tests/cputestdata/x86-baseline-3-result.xml b/tests/cputestdata/x86-baseline-3-result.xml new file mode 100644 index 0000000..d196112 --- /dev/null +++ b/tests/cputestdata/x86-baseline-3-result.xml @@ -0,0 +1,35 @@ +<cpu mode='custom' match='exact'> + <model fallback='allow'>Westmere</model> + <feature policy='require' name='lahf_lm'/> + <feature policy='require' name='lm'/> + <feature policy='require' name='nx'/> + <feature policy='require' name='syscall'/> + <feature policy='require' name='aes'/> + <feature policy='require' name='popcnt'/> + <feature policy='require' name='sse4.2'/> + <feature policy='require' name='sse4.1'/> + <feature policy='require' name='cx16'/> + <feature policy='require' name='ssse3'/> + <feature policy='require' name='pni'/> + <feature policy='require' name='sse2'/> + <feature policy='require' name='sse'/> + <feature policy='require' name='fxsr'/> + <feature policy='require' name='mmx'/> + <feature policy='require' name='clflush'/> + <feature policy='require' name='pse36'/> + <feature policy='require' name='pat'/> + <feature policy='require' name='cmov'/> + <feature policy='require' name='mca'/> + <feature policy='require' name='pge'/> + <feature policy='require' name='mtrr'/> + <feature policy='require' name='sep'/> + <feature policy='require' name='apic'/> + <feature policy='require' name='cx8'/> + <feature policy='require' name='mce'/> + <feature policy='require' name='pae'/> + <feature policy='require' name='msr'/> + <feature policy='require' name='tsc'/> + <feature policy='require' name='pse'/> + <feature policy='require' name='de'/> + <feature policy='require' name='fpu'/> +</cpu> diff --git a/tests/cputestdata/x86-baseline-3.xml b/tests/cputestdata/x86-baseline-3.xml new file mode 100644 index 0000000..7654a1d --- /dev/null +++ b/tests/cputestdata/x86-baseline-3.xml @@ -0,0 +1,7 @@ +<cpuTest> +<cpu> + <arch>x86_64</arch> + <model>Westmere</model> + <topology sockets='1' cores='2' threads='1'/> +</cpu> +</cpuTest> diff --git a/tools/virsh-domain.c b/tools/virsh-domain.c index 0402aef..0aa8b60 100644 --- a/tools/virsh-domain.c +++ b/tools/virsh-domain.c @@ -5986,6 +5986,10 @@ static const vshCmdOptDef opts_cpu_baseline[] = { .flags = VSH_OFLAG_REQ, .help = N_("file containing XML CPU descriptions") }, + {.name = "model_features", + .type = VSH_OT_BOOL, + .help = N_("Show features that are part of the CPU model type") + }, {.name = NULL} }; @@ -5997,6 +6001,7 @@ cmdCPUBaseline(vshControl *ctl, const vshCmd *cmd) char *buffer; char *result = NULL; const char **list = NULL; + unsigned int flags = 0; int count = 0; xmlDocPtr xml = NULL; @@ -6006,6 +6011,9 @@ cmdCPUBaseline(vshControl *ctl, const vshCmd *cmd) virBuffer buf = VIR_BUFFER_INITIALIZER; int i; + if (vshCommandOptBool(cmd, "model_features")) + flags |= VIR_CONNECT_BASELINE_SHOW_MODEL; + if (vshCommandOptStringReq(ctl, cmd, "file", &from) < 0) return false; @@ -6049,7 +6057,8 @@ cmdCPUBaseline(vshControl *ctl, const vshCmd *cmd) list[i] = vshStrdup(ctl, (const char *)xmlBufferContent(xml_buf)); } - result = virConnectBaselineCPU(ctl->conn, list, count, 0); + result = virConnectBaselineCPU(ctl->conn, list, count, flags); +vshPrint(ctl, "result - %p\n", result); if (result) { vshPrint(ctl, "%s", result); -- 1.7.10.4 -- Don Dugger "Censeo Toto nos in Kansa esse decisse." - D. Gale n0ano(a)n0ano.com Ph: 303/443-3786

3 13

[libvirt] [PATCH] maint: slightly reduce configure size
by Eric Blake 17 Aug '13

17 Aug '13

Rather than inlining gl_WARN_ADD loads of time, we can shave about 17k size off of the configure script by delaying it to a cleanup shell loop. * m4/virt-compile-warnings.m4 (LIBVIRT_COMPILE_WARNINGS): Track a list of things to check, rather than inlining multiple checks. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- m4/virt-compile-warnings.m4 | 72 ++++++++++++++++++++++----------------------- 1 file changed, 36 insertions(+), 36 deletions(-) diff --git a/m4/virt-compile-warnings.m4 b/m4/virt-compile-warnings.m4 index 6bf797f..938c8bb 100644 --- a/m4/virt-compile-warnings.m4 +++ b/m4/virt-compile-warnings.m4 @@ -133,47 +133,31 @@ AC_DEFUN([LIBVIRT_COMPILE_WARNINGS],[ # Remove the ones we don't want, blacklisted earlier gl_MANYWARN_COMPLEMENT([wantwarn], [$maybewarn], [$dontwarn]) - # Check for $CC support of each warning - for w in $wantwarn; do - gl_WARN_ADD([$w]) - done - # GNULIB uses '-W' (aka -Wextra) which includes a bunch of stuff. # Unfortunately, this means you can't simply use '-Wsign-compare' # with gl_MANYWARN_COMPLEMENT # So we have -W enabled, and then have to explicitly turn off... - gl_WARN_ADD([-Wno-sign-compare]) + wantwarn="$wantwarn -Wno-sign-compare" # GNULIB expects this to be part of -Wc++-compat, but we turn # that one off, so we need to manually enable this again - gl_WARN_ADD([-Wjump-misses-init]) + wantwarn="$wantwarn -Wjump-misses-init" # GNULIB turns on -Wformat=2 which implies -Wformat-nonliteral, # so we need to manually re-exclude it. Also, older gcc 4.2 # added an implied ATTRIBUTE_NONNULL on any parameter marked # ATTRIBUTE_FMT_PRINT, which causes -Wformat failure on our # intentional use of virReportError(code, NULL). - gl_WARN_ADD([-Wno-format-nonliteral]) + wantwarn="$wantwarn -Wno-format-nonliteral" if test $lv_cv_gcc_wformat_null_works = no; then - gl_WARN_ADD([-Wno-format]) + wantwarn="$wantwarn -Wno-format" fi # This should be < 256 really. Currently we're down to 4096, # but using 1024 bytes sized buffers (mostly for virStrerror) # stops us from going down further - gl_WARN_ADD([-Wframe-larger-than=4096]) - dnl gl_WARN_ADD([-Wframe-larger-than=256]) - - # Silence certain warnings in gnulib, and use improved glibc headers - AC_DEFINE([lint], [1], - [Define to 1 if the compiler is checking for lint.]) - AH_VERBATIM([FORTIFY_SOURCE], - [/* Enable compile-time and run-time bounds-checking, and some warnings, - without upsetting newer glibc. */ - #if !defined _FORTIFY_SOURCE && defined __OPTIMIZE__ && __OPTIMIZE__ - # define _FORTIFY_SOURCE 2 - #endif - ]) + wantwarn="$wantwarn -Wframe-larger-than=4096" + dnl wantwarn="$wantwarn -Wframe-larger-than=256" # Extra special flags dnl -fstack-protector stuff passes gl_WARN_ADD with gcc @@ -182,43 +166,59 @@ AC_DEFUN([LIBVIRT_COMPILE_WARNINGS],[ *-*-linux*) dnl Fedora only uses -fstack-protector, but doesn't seem to dnl be great overhead in adding -fstack-protector-all instead - dnl gl_WARN_ADD([-fstack-protector]) - gl_WARN_ADD([-fstack-protector-all]) - gl_WARN_ADD([--param=ssp-buffer-size=4]) + dnl wantwarn="$wantwarn -fstack-protector" + wantwarn="$wantwarn -fstack-protector-all" + wantwarn="$wantwarn --param=ssp-buffer-size=4" dnl Even though it supports it, clang complains about dnl use of --param=ssp-buffer-size=4 unless used with dnl the -c arg. It doesn't like it when used with args dnl that just link together .o files. Unfortunately dnl we can't avoid that with automake, so we must turn dnl off the following clang specific warning - gl_WARN_ADD([-Wno-unused-command-line-argument]) + wantwarn="$wantwarn -Wno-unused-command-line-argument" ;; *-*-freebsd*) dnl FreeBSD ships old gcc 4.2.1 which doesn't handle dnl -fstack-protector-all well - gl_WARN_ADD([-fstack-protector]) + wantwarn="$wantwarn -fstack-protector" - gl_WARN_ADD([-Wno-unused-command-line-argument]) + wantwarn="$wantwarn -Wno-unused-command-line-argument" ;; esac - gl_WARN_ADD([-fexceptions]) - gl_WARN_ADD([-fasynchronous-unwind-tables]) - gl_WARN_ADD([-fdiagnostics-show-option]) - gl_WARN_ADD([-funit-at-a-time]) + wantwarn="$wantwarn -fexceptions" + wantwarn="$wantwarn -fasynchronous-unwind-tables" + wantwarn="$wantwarn -fdiagnostics-show-option" + wantwarn="$wantwarn -funit-at-a-time" # Need -fipa-pure-const in order to make -Wsuggest-attribute=pure # fire even without -O. - gl_WARN_ADD([-fipa-pure-const]) + wantwarn="$wantwarn -fipa-pure-const" # We should eventually enable this, but right now there are at # least 75 functions triggering warnings. - gl_WARN_ADD([-Wno-suggest-attribute=pure]) - gl_WARN_ADD([-Wno-suggest-attribute=const]) + wantwarn="$wantwarn -Wno-suggest-attribute=pure" + wantwarn="$wantwarn -Wno-suggest-attribute=const" if test "$set_werror" = "yes" then - gl_WARN_ADD([-Werror]) + wantwarn="$wantwarn -Werror" fi + # Check for $CC support of each warning + for w in $wantwarn; do + gl_WARN_ADD([$w]) + done + + # Silence certain warnings in gnulib, and use improved glibc headers + AC_DEFINE([lint], [1], + [Define to 1 if the compiler is checking for lint.]) + AH_VERBATIM([FORTIFY_SOURCE], + [/* Enable compile-time and run-time bounds-checking, and some warnings, + without upsetting newer glibc. */ + #if !defined _FORTIFY_SOURCE && defined __OPTIMIZE__ && __OPTIMIZE__ + # define _FORTIFY_SOURCE 2 + #endif + ]) + dnl Needed to keep compile quiet on python 2.4 save_WARN_CFLAGS=$WARN_CFLAGS WARN_CFLAGS= -- 1.8.3.1

2 2

[libvirt] [PATCH] libxl: refactor capabilities code
by Jim Fehlig 17 Aug '13

17 Aug '13

Cleanup the libxl capabilities code to be a bit more extensible, splitting out the creation of host and guest capabilities. This should make it easier to implement additional capabilities in the future, such as NUMA topology reporting. --- Although I can't see how changes here would affect other drivers, I nonetheless ensured libvirtd loaded when not booted to Xen :). src/libxl/libxl_conf.c | 256 +++++++++++++++++++++++-------------------------- 1 file changed, 121 insertions(+), 135 deletions(-) diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c index 827dfdd..5ea4163 100644 --- a/src/libxl/libxl_conf.c +++ b/src/libxl/libxl_conf.c @@ -64,126 +64,62 @@ static const char *xen_cap_re = "(xen|hvm)-[[:digit:]]+\\.[[:digit:]]+-(x86_32|x static regex_t xen_cap_rec; -static virCapsPtr -libxlBuildCapabilities(virArch hostarch, - int host_pae, - struct guest_arch *guest_archs, - int nr_guest_archs) +static int +libxlCapsInitHost(libxl_ctx *ctx, virCapsPtr caps) { - virCapsPtr caps; - size_t i; - - if ((caps = virCapabilitiesNew(hostarch, 1, 1)) == NULL) - goto no_memory; - - if (host_pae && - virCapabilitiesAddHostFeature(caps, "pae") < 0) - goto no_memory; - - for (i = 0; i < nr_guest_archs; ++i) { - virCapsGuestPtr guest; - char const *const xen_machines[] = {guest_archs[i].hvm ? "xenfv" : "xenpv"}; - virCapsGuestMachinePtr *machines; - - if ((machines = virCapabilitiesAllocMachines(xen_machines, 1)) == NULL) - goto no_memory; - - if ((guest = virCapabilitiesAddGuest(caps, - guest_archs[i].hvm ? "hvm" : "xen", - guest_archs[i].arch, - ((hostarch == VIR_ARCH_X86_64) ? - "/usr/lib64/xen/bin/qemu-dm" : - "/usr/lib/xen/bin/qemu-dm"), - (guest_archs[i].hvm ? - "/usr/lib/xen/boot/hvmloader" : - NULL), - 1, - machines)) == NULL) { - virCapabilitiesFreeMachines(machines, 1); - goto no_memory; - } - machines = NULL; - - if (virCapabilitiesAddGuestDomain(guest, - "xen", - NULL, - NULL, - 0, - NULL) == NULL) - goto no_memory; - - if (guest_archs[i].pae && - virCapabilitiesAddGuestFeature(guest, - "pae", - 1, - 0) == NULL) - goto no_memory; - - if (guest_archs[i].nonpae && - virCapabilitiesAddGuestFeature(guest, - "nonpae", - 1, - 0) == NULL) - goto no_memory; - - if (guest_archs[i].ia64_be && - virCapabilitiesAddGuestFeature(guest, - "ia64_be", - 1, - 0) == NULL) - goto no_memory; - - if (guest_archs[i].hvm) { - if (virCapabilitiesAddGuestFeature(guest, - "acpi", - 1, - 1) == NULL) - goto no_memory; + int err; + libxl_physinfo phy_info; + int host_pae; - if (virCapabilitiesAddGuestFeature(guest, "apic", - 1, - 0) == NULL) - goto no_memory; + err = regcomp(&xen_cap_rec, xen_cap_re, REG_EXTENDED); + if (err != 0) { + char error[100]; + regerror(err, &xen_cap_rec, error, sizeof(error)); + regfree(&xen_cap_rec); + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Failed to compile regex %s"), error); + return -1; + } - if (virCapabilitiesAddGuestFeature(guest, - "hap", - 0, - 1) == NULL) - goto no_memory; - } + if (libxl_get_physinfo(ctx, &phy_info) != 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Failed to get node physical info from libxenlight")); + return -1; } - return caps; + /* hw_caps is an array of 32-bit words whose meaning is listed in + * xen-unstable.hg/xen/include/asm-x86/cpufeature.h. Each feature + * is defined in the form X*32+Y, corresponding to the Y'th bit in + * the X'th 32-bit word of hw_cap. + */ + host_pae = phy_info.hw_cap[0] & LIBXL_X86_FEATURE_PAE_MASK; + if (host_pae && + virCapabilitiesAddHostFeature(caps, "pae") < 0) + return -1; - no_memory: - virObjectUnref(caps); - return NULL; + return 0; } -static virCapsPtr -libxlMakeCapabilitiesInternal(virArch hostarch, - libxl_physinfo *phy_info, - char *capabilities) +static int +libxlCapsInitGuests(libxl_ctx *ctx, virCapsPtr caps) { + const libxl_version_info *ver_info; char *str, *token; regmatch_t subs[4]; char *saveptr = NULL; size_t i; + virArch hostarch = caps->host.arch; - int host_pae = 0; struct guest_arch guest_archs[32]; int nr_guest_archs = 0; - virCapsPtr caps = NULL; memset(guest_archs, 0, sizeof(guest_archs)); - /* hw_caps is an array of 32-bit words whose meaning is listed in - * xen-unstable.hg/xen/include/asm-x86/cpufeature.h. Each feature - * is defined in the form X*32+Y, corresponding to the Y'th bit in - * the X'th 32-bit word of hw_cap. - */ - host_pae = phy_info->hw_cap[0] & LIBXL_X86_FEATURE_PAE_MASK; - + if ((ver_info = libxl_get_version_info(ctx)) == NULL) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Failed to get version info from libxenlight")); + return -1; + } /* Format of capabilities string is documented in the code in * xen-unstable.hg/xen/arch/.../setup.c. * @@ -209,7 +145,7 @@ libxlMakeCapabilitiesInternal(virArch hostarch, /* Split capabilities string into tokens. strtok_r is OK here because * we "own" the buffer. Parse out the features from each token. */ - for (str = capabilities, nr_guest_archs = 0; + for (str = ver_info->capabilities, nr_guest_archs = 0; nr_guest_archs < sizeof(guest_archs) / sizeof(guest_archs[0]) && (token = strtok_r(str, " ", &saveptr)) != NULL; str = NULL) { @@ -274,17 +210,80 @@ libxlMakeCapabilitiesInternal(virArch hostarch, } } - if ((caps = libxlBuildCapabilities(hostarch, - host_pae, - guest_archs, - nr_guest_archs)) == NULL) - goto error; + for (i = 0; i < nr_guest_archs; ++i) { + virCapsGuestPtr guest; + char const *const xen_machines[] = {guest_archs[i].hvm ? "xenfv" : "xenpv"}; + virCapsGuestMachinePtr *machines; - return caps; + if ((machines = virCapabilitiesAllocMachines(xen_machines, 1)) == NULL) + return -1; - error: - virObjectUnref(caps); - return NULL; + if ((guest = virCapabilitiesAddGuest(caps, + guest_archs[i].hvm ? "hvm" : "xen", + guest_archs[i].arch, + ((hostarch == VIR_ARCH_X86_64) ? + "/usr/lib64/xen/bin/qemu-dm" : + "/usr/lib/xen/bin/qemu-dm"), + (guest_archs[i].hvm ? + "/usr/lib/xen/boot/hvmloader" : + NULL), + 1, + machines)) == NULL) { + virCapabilitiesFreeMachines(machines, 1); + return -1; + } + machines = NULL; + + if (virCapabilitiesAddGuestDomain(guest, + "xen", + NULL, + NULL, + 0, + NULL) == NULL) + return -1; + + if (guest_archs[i].pae && + virCapabilitiesAddGuestFeature(guest, + "pae", + 1, + 0) == NULL) + return -1; + + if (guest_archs[i].nonpae && + virCapabilitiesAddGuestFeature(guest, + "nonpae", + 1, + 0) == NULL) + return -1; + + if (guest_archs[i].ia64_be && + virCapabilitiesAddGuestFeature(guest, + "ia64_be", + 1, + 0) == NULL) + return -1; + + if (guest_archs[i].hvm) { + if (virCapabilitiesAddGuestFeature(guest, + "acpi", + 1, + 1) == NULL) + return -1; + + if (virCapabilitiesAddGuestFeature(guest, "apic", + 1, + 0) == NULL) + return -1; + + if (virCapabilitiesAddGuestFeature(guest, + "hap", + 0, + 1) == NULL) + return -1; + } + } + + return 0; } static int @@ -874,35 +873,22 @@ error: virCapsPtr libxlMakeCapabilities(libxl_ctx *ctx) { - int err; - libxl_physinfo phy_info; - const libxl_version_info *ver_info; + virCapsPtr caps; - err = regcomp(&xen_cap_rec, xen_cap_re, REG_EXTENDED); - if (err != 0) { - char error[100]; - regerror(err, &xen_cap_rec, error, sizeof(error)); - regfree(&xen_cap_rec); - virReportError(VIR_ERR_INTERNAL_ERROR, - _("Failed to compile regex %s"), error); + if ((caps = virCapabilitiesNew(virArchFromHost(), 1, 1)) == NULL) return NULL; - } - if (libxl_get_physinfo(ctx, &phy_info) != 0) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("Failed to get node physical info from libxenlight")); - return NULL; - } + if (libxlCapsInitHost(ctx, caps) < 0) + goto error; - if ((ver_info = libxl_get_version_info(ctx)) == NULL) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("Failed to get version info from libxenlight")); - return NULL; - } + if (libxlCapsInitGuests(ctx, caps) < 0) + goto error; + + return caps; - return libxlMakeCapabilitiesInternal(virArchFromHost(), - &phy_info, - ver_info->capabilities); +error: + virObjectUnref(caps); + return NULL; } int -- 1.8.1.4

3 4

[libvirt] [PATCHv2 0/2] Fix starting of VM's on hosts without DBus compiled in or enabled
by Peter Krempa 17 Aug '13

17 Aug '13

Hi, this series fixes regression on hosts without DBus that was introduced by the refactor to use systemd.machined to create cgroups. Version 2: As errors can't be unlogged, this version enhances the error reporting code to add conditional error logging. The only remaining issue is that if a user is running DBus but doesn't use systemd, the logs are still spammed with errors. Peter Krempa (2): virerror: Add support for suppressing reporting of errors cgroup: Don't fail to start a VM when DBus isn't compiled in or running daemon/libvirtd.c | 2 +- daemon/remote.c | 2 +- src/libvirt_private.syms | 2 +- src/node_device/node_device_hal.c | 2 +- src/nwfilter/nwfilter_driver.c | 4 ++-- src/rpc/virnetserver.c | 2 +- src/util/virdbus.c | 15 ++++++++------- src/util/virdbus.h | 2 +- src/util/virerror.c | 31 ++++++++++++++++++++----------- src/util/virerror.h | 21 +++++++++++++++------ src/util/virsystemd.c | 4 ++-- 11 files changed, 53 insertions(+), 34 deletions(-) -- 1.8.3.2

2 3

Re: [libvirt] [openstack-dev] [nova] live-snapshot/cloning of virtual machines
by Andres Lagar-Cavilla 17 Aug '13

17 Aug '13

> On Fri, Aug 16, 2013 at 11:05:19AM +0100, Daniel P. Berrange wrote: >> On Wed, Aug 14, 2013 at 04:53:01PM -0700, Vishvananda Ishaya wrote: >>> Hi Everyone, >>> >>> I have been trying for some time to get the code for the live-snapshot blueprint[1] >>> in. Going through the review process for the rpc and interface code[2] was easy. I >>> suspect the api-extension code[3] will also be relatively trivial to get in. The >>> main concern is with the libvirt driver implementation[4]. I'd like to discuss the >>> concerns and see if we can make some progress. >>> >>> Short Summary (tl;dr) >>> ===================== >>> >>> I propose we merge live-cloning as an experimental feature for havanna and have the >>> api extension disabled by default. >>> >>> Overview >>> ======== >>> >>> First of all, let me express the value of live snapshoting. The >>> slowest part of the vm provisioning process is generally booting >>> of the OS. > > Like Dan I'm dubious about this whole plan. But this ^^ statement in > particular. I would like to see hard data to back this up. What we need to keep in mind is that "boot" is a small part of the picture, at least "boot" as commonly referred to in Linux. Consider a web sphere-like Java bundle of code. These things take a while to load. JiT-ed methods provide a tremendous performance boost. Nevermind if the the server constructs secondary indices to perform fast lookups of data. That is just Linux. Windows is well known for pounding storage fabrics with thousands of small reads during boot storms. Certainly a boot Windows sequence has baked in a lot of service startup sequences that prime a lot of memory content for performance objectives. Boot here means "ready to rock-n-roll", not "Cirros is up." We have live deployments that are based on bypassing the entire *application startup* sequence and have a server ready to provide high-performance responses to queries once spawned from a live saved image. > > You should be able to boot an OS pretty quickly, and furthermore it's > (a) much safer for all the reasons Dan outlines, and (b) improvements > that you make to boot times help everyone. > > [...] >>> 2. Security Concerns >>> ==================== >>> >>> There are a number of security issues with loading state from another vm. Here is a >>> short list of things that need to be done just to make a cloned vm usable: >>> >>> a) mac address needs to be recreated >>> b) entropy pool needs to be reset >>> c) host name must be reset >>> d) host keys bust be regenerated >>> >>> There are others, and trying to clone a running application as well may expose other >>> sensitive data, especially if users are snaphsoting vms and making them public. > > Are we talking about cloning VMs that you already trust, or cloning > random VMs and allowing random other users to use them? These would > lead to very different solutions. In the first case, you only care > about correctness, not security. In the second case, you care about > security as well as correctness. Case number one. The correctness issues are a hard problem, and a particularly hard one in Windows, but it is pragmatically solvable. For a common scenario in Linux, renewing dhcp leases and leveling your entropy pool are what you need. > > I highly doubt the second case is possible because scrubbing the disk > is going to take far too long for any supposed time-saving to matter. That would be very counter-productive, so yes, focusing on the first case. > > As Dan says, even the first case is dubious because it won't be correct. > >> The libguestfs project provide tools to perform offline cloning of >> VM disk images. Its virt-sysprep knows how to delete alot (but by >> no means all possible) sensitive file data for common Linux & >> Windows OS. It still has to be combined with use of the >> virt-sparsify tool though, to ensure the deleted data is actually >> purged from the VM disk image as well as the filesystem, by >> releasing all unused VM disk sectors back to the host storage (and >> not all storage supports that). > > Links to the tools that Dan mentions: > > http://libguestfs.org/virt-sysprep.1.html > http://libguestfs.org/virt-sparsify.1.html Virt-sparsify is not strictly relevant here. The disk side of live images is carried out with qcow2. Virt-sysprep is great work and highly relevant. But virt-sysprep allows us to see the argument in a different light. Have you noticed nova does not run virt-sysprep before booting an ephemeral instance from an image? (AFAIK, could be wrong, not even regenerating host ssh keys is part of the assured workflow). Furthermore, one can create arbitrary (cold, non-live) images at any time, from live instances This isn't necessarily wrong. It underpins massive deployments, it pragmatically adds value. The fundamental semantics at play with live-instances are the same: know what you are doing, ephemeral instances, bound to your tenant. So as long as we take care of not weakening cryptographic foundations and correctly reconfiguring the identity, all the principles above still apply: know what you are doing, ephemeral instances, bound to your tenant. We are working on proof-of-concept using the qemu guest agent to undertake all these tasks. It is PoC to show it's doable. It's neither clean nor asking for mainline merge (at the moment). It's not the only solution for the guest side problem. An one final very important note for the libvirt list and those who may be tuning in just now: no changes to libvirt are being asked for here. This is a nova-side tech preview feature. Best, Andres > > Note these tools can only be used on offline machines. > > Rich. > > -- > Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones > virt-top is 'top' for virtual machines. Tiny program with many > powerful monitoring features, net stats, disk stats, logging, etc. > http://people.redhat.com/~rjones/virt-top

1 0