- Devel - libvirt List Archives

[libvirt] [RFC] Refactoring bridge driver for portability
by Roman Bogorodskiy 03 Jan '13

03 Jan '13

Hi, Few weeks ago when I have submitted my all-in-one huge FreeBSD, Eric made some comments on the networking part: http://www.redhat.com/archives/libvir-list/2012-December/msg00432.html Now when we're done with most of the other things, I'd like to discuss networking in more detail. Basically, the main question that bothers me is what would be the best way to refactor network/bridge_driver.c? Currently it contains a number of things: - dnsmasq routines, static, but cross-platform, doesn't need any changes (at least major ones) - the same for radvd - minor firewalld bits, that are linux specifc - iptables routines. There are quite a lot of that code and it's also gets called from many places in the file I'm thinking about an approach that Eric mentioned for virthread.c: - keep dnsmasq and radvd in bridge_driver.c - move out all linux specific code to bridge_driver_linux.c and #include it in a way similar to virthread.c - in a similar way create bridge_driver_freebsd.c Does it sound reasonable? Roman Bogorodskiy

3 3

[libvirt] [PATCH] build: use autobuild module to make build logs nicer
by Eric Blake 03 Jan '13

03 Jan '13

A recent build failure made me realize that we could usefully add a bit more information to configure output, for aid in analysis of failed builds. Pulling in the autobuild module merely adds these four lines to configure output: configure: autobuild project... libvirt configure: autobuild revision... 1.0.1 configure: autobuild hostname... myhost configure: autobuild timestamp... 20130102T181342Z which can be useful even if not using the Autobuild project to parse build logs. * bootstrap.conf (gnulib_modules): Add autobuild. --- I'll wait for a review on this one. Hmm, I also realized that there are other things that might be useful for autobuild to log, such as uname output; I'll ask upstream if it is worth enhancing the autobuild output, and if upstream agrees, then we would pick up that enhanced information next time we sync from gnulib. bootstrap.conf | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/bootstrap.conf b/bootstrap.conf index 37a0ae1..4b45a4f 100644 --- a/bootstrap.conf +++ b/bootstrap.conf @@ -1,6 +1,6 @@ # Bootstrap configuration. -# Copyright (C) 2010-2012 Red Hat, Inc. +# Copyright (C) 2010-2013 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 @@ -20,6 +20,7 @@ gnulib_modules=' accept areadlink +autobuild base64 bind bitrotate -- 1.8.0.2

2 3

[libvirt] where is config.log when debugging a failure reported by Jenkins
by Eric Blake 02 Jan '13

02 Jan '13

Hello Jenkins folks, On the libvirt list, we came across a case where a broken commit was detected, but where the information available in the failure report was insufficient (for at least me) to reproduce the failure environment without asking more questions: http://honk.sigxcpu.org:8001/job/libvirt-build/472/ https://www.redhat.com/archives/libvir-list/2013-January/msg00037.html Is there any way that things can be tweaked to include the 'config.log' file in the web report of a failed build? Merely capturing the output of './configure' fails to provide some of the details that are intentionally redirected into the more verbose config.log file. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org

2 2

[libvirt] [PATCH] maint: update to latest gnulib
by Michal Privoznik 02 Jan '13

02 Jan '13

The update results in bootstrap being updated as well. --- .gnulib | 2 +- bootstrap | 15 ++++++++------- 2 files changed, 9 insertions(+), 8 deletions(-) diff --git a/.gnulib b/.gnulib index d245e6d..964bbc2 160000 --- a/.gnulib +++ b/.gnulib @@ -1 +1 @@ -Subproject commit d245e6ddd6ab2624d0d83acd8f111454f984f50f +Subproject commit 964bbc2d419584e93fe629ddbc40595612f62083 diff --git a/bootstrap b/bootstrap index e3e270b..c84d249 100755 --- a/bootstrap +++ b/bootstrap @@ -1,6 +1,6 @@ #! /bin/sh # Print a version string. -scriptversion=2012-07-19.14; # UTC +scriptversion=2012-12-28.10; # UTC # Bootstrap this package from checked-out sources. @@ -889,20 +889,21 @@ find "$m4_base" "$source_base" \ -depth $ -name '*.m4' -o -name '*.[ch]' $ \ -type l -xtype l -delete > /dev/null 2>&1 +# Invoke autoreconf with --force --install to ensure upgrades of tools +# such as ylwrap. +AUTORECONFFLAGS="--verbose --install --force -I $m4_base $ACLOCAL_FLAGS" + # Some systems (RHEL 5) are using ancient autotools, for which the # --no-recursive option had not been invented. Detect that lack and # omit the option when it's not supported. FIXME in 2017: remove this # hack when RHEL 5 autotools are updated, or when they become irrelevant. -no_recursive= case $($AUTORECONF --help) in - *--no-recursive*) no_recursive=--no-recursive;; + *--no-recursive*) AUTORECONFFLAGS="$AUTORECONFFLAGS --no-recursive";; esac # Tell autoreconf not to invoke autopoint or libtoolize; they were run above. -echo "running: AUTOPOINT=true LIBTOOLIZE=true " \ - "$AUTORECONF --verbose --install $no_recursive -I $m4_base $ACLOCAL_FLAGS" -AUTOPOINT=true LIBTOOLIZE=true \ - $AUTORECONF --verbose --install $no_recursive -I $m4_base $ACLOCAL_FLAGS \ +echo "running: AUTOPOINT=true LIBTOOLIZE=true $AUTORECONF $AUTORECONFFLAGS" +AUTOPOINT=true LIBTOOLIZE=true $AUTORECONF $AUTORECONFFLAGS \ || die "autoreconf failed" # Get some extra files from gnulib, overriding existing files. -- 1.8.0.2

2 2

[libvirt] [PATCH] Remove the inexistent function in comments
by Osier Yang 02 Jan '13

02 Jan '13

virNodeDeviceLookupByKey is inexistent. --- Pushed under trivial rule. --- include/libvirt/libvirt.h.in | 7 +++---- 1 files changed, 3 insertions(+), 4 deletions(-) diff --git a/include/libvirt/libvirt.h.in b/include/libvirt/libvirt.h.in index c6739d7..2de6835 100644 --- a/include/libvirt/libvirt.h.in +++ b/include/libvirt/libvirt.h.in @@ -3113,10 +3113,9 @@ typedef struct _virNodeDevice virNodeDevice; * virNodeDevicePtr: * * A virNodeDevicePtr is a pointer to a virNodeDevice structure. Get - * one via virNodeDeviceLookupByKey, virNodeDeviceLookupByName, or - * virNodeDeviceCreate. Be sure to Call virNodeDeviceFree when done - * using a virNodeDevicePtr obtained from any of the above functions to - * avoid leaking memory. + * one via, virNodeDeviceLookupByName, or virNodeDeviceCreate. Be sure + * to Call virNodeDeviceFree when done using a virNodeDevicePtr obtained + * from any of the above functions to avoid leaking memory. */ typedef virNodeDevice *virNodeDevicePtr; -- 1.7.7.6

2 2

[libvirt] [PATCH] qemu: Adapt to new log format
by Michal Privoznik 02 Jan '13

02 Jan '13

Since 586502189edf9fd0f89a83de96717a2ea826fdb0 qemu commit, the log lines reporting chardev's path has changed from: $ ./x86_64-softmmu/qemu-system-x86_64 -serial pty -serial pty -monitor pty char device redirected to /dev/pts/5 char device redirected to /dev/pts/6 char device redirected to /dev/pts/7 to: $ ./x86_64-softmmu/qemu-system-x86_64 -serial pty -serial pty -monitor pty char device compat_monitor0 redirected to /dev/pts/5 char device serial0 redirected to /dev/pts/6 char device serial1 redirected to /dev/pts/7 However, with current code we are not prepared for such change, which results in us being unable to start any domain. --- src/qemu/qemu_process.c | 33 +++++++++++++++++++++++++++++---- 1 file changed, 29 insertions(+), 4 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index eac6553..29bd082 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -1431,22 +1431,43 @@ cleanup: * * char device redirected to /dev/pts/3 * + * However, since 1.4 the line we are looking for has changed to: + * + * char device <alias> redirected to /some/path + * * Returns -1 for error, 0 success, 1 continue reading */ static int qemuProcessExtractTTYPath(const char *haystack, size_t *offset, + const char *alias, char **path) { - static const char needle[] = "char device redirected to"; - char *tmp, *dev; + static const char *needle[] = {"char device", "redirected to"}; + const char *tmp, *dev; VIR_FREE(*path); /* First look for our magic string */ - if (!(tmp = strstr(haystack + *offset, needle))) { + if (!(tmp = strstr(haystack + *offset, needle[0]))) return 1; + + tmp += strlen(needle[0]); + virSkipSpaces(&tmp); + + if (STRPREFIX(tmp, "char")) { + /* we are dealing with new style */ + tmp += strlen("char"); + if (!STRPREFIX(tmp, alias)) + return 1; + + tmp += strlen(alias); + virSkipSpaces(&tmp); } - tmp += sizeof(needle); + + if (!STRPREFIX(tmp, needle[1])) + return 1; + + tmp += strlen(needle[1]); dev = tmp; /* @@ -1569,6 +1590,7 @@ qemuProcessFindCharDevicePTYs(virDomainObjPtr vm, virDomainChrDefPtr chr = vm->def->serials[i]; if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { if ((ret = qemuProcessExtractTTYPath(output, &offset, + chr->info.alias, &chr->source.data.file.path)) != 0) return ret; } @@ -1579,6 +1601,7 @@ qemuProcessFindCharDevicePTYs(virDomainObjPtr vm, virDomainChrDefPtr chr = vm->def->parallels[i]; if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { if ((ret = qemuProcessExtractTTYPath(output, &offset, + chr->info.alias, &chr->source.data.file.path)) != 0) return ret; } @@ -1589,6 +1612,7 @@ qemuProcessFindCharDevicePTYs(virDomainObjPtr vm, virDomainChrDefPtr chr = vm->def->channels[i]; if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) { if ((ret = qemuProcessExtractTTYPath(output, &offset, + chr->info.alias, &chr->source.data.file.path)) != 0) return ret; } @@ -1608,6 +1632,7 @@ qemuProcessFindCharDevicePTYs(virDomainObjPtr vm, if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY && chr->targetType == VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO) { if ((ret = qemuProcessExtractTTYPath(output, &offset, + chr->info.alias, &chr->source.data.file.path)) != 0) return ret; } -- 1.8.0.2

4 7

[libvirt] ESX error while freeing inherited objects
by Ata Bohra 02 Jan '13

02 Jan '13

Hi All, Freeing of an object having inheritance depth >=2 throws an error such as:" internal error Call to esxVI_<grand_parent>_Free for unexpected type 'Child'. " For instance: Below code snippet is from esx_vi_generator.input (modified) object VirtualDevice Int key r Description deviceInfo o VirtualDeviceBackingInfo backing o VirtualDeviceConnectInfo connectable o Int controllerKey o Int unitNumber o end object VirtualController extends VirtualDevice Int busNumber r Int device ol end object VirtualIDEController extends VirtualController end object VirtualDeviceConfigSpec VirtualDeviceConfigSpecOperation operation o VirtualDeviceConfigSpecFileOperation fileOperation o VirtualDevice device r end object VirtualMachineConfigSpec... VirtualDeviceConfigSpec deviceChange ol..end On free of "VirtualMachineConfigSpec", I hit below error:"libvir: ESX Driver error : internal error Call to esxVI_VirtualDevice_Free for unexpected type 'VirtualIDEController'" But if I remove VirtualController and include its fields directly to VirtualIDEController, error is resolved. Please let me know if I'm missing anything, else, it seems like a bug with generated code. While debugging the error seems to come from: esx_vi_types.c: /* * Macros to implement dynamic dispatched functions */#define ESX_VI__TEMPLATE__DISPATCH(_actual_type, _actual_type_name, __type, \ _dispatch, _error_return) \ switch (_actual_type) { \ _dispatch \ \ case esxVI_Type_##__type: \ break; \ \ default: \ virReportError(VIR_ERR_INTERNAL_ERROR, \ _("Call to %s for unexpected type '%s'"), \ __FUNCTION__, _actual_type_name); \ return _error_return; \ } Thanks!Ata

1 1

[libvirt] [PATCH] ESX: append CURL headers to fix serviceContent entities
by Ata E Husain Bohra 01 Jan '13

01 Jan '13

Append curl headers with "SOAPAction" header to populate serviceContent object entities which are otherwise missing. --- src/esx/esx_vi.c | 2 ++ src/esx/esx_vi_generator.input | 3 +++ 2 files changed, 5 insertions(+) diff --git a/src/esx/esx_vi.c b/src/esx/esx_vi.c index 99c1eb1..a379183 100644 --- a/src/esx/esx_vi.c +++ b/src/esx/esx_vi.c @@ -322,6 +322,8 @@ esxVI_CURL_Connect(esxVI_CURL *curl, esxUtil_ParsedUri *parsedUri) * approx. 2 sec per POST operation. */ curl->headers = curl_slist_append(curl->headers, "Expect:"); + curl->headers = curl_slist_append(curl->headers, + _("SOAPAction: \"urn:vim25\"")); if (curl->headers == NULL) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", diff --git a/src/esx/esx_vi_generator.input b/src/esx/esx_vi_generator.input index 22c114e..236dcb3 100644 --- a/src/esx/esx_vi_generator.input +++ b/src/esx/esx_vi_generator.input @@ -1008,6 +1008,8 @@ object ServiceContent ManagedObjectReference clusterProfileManager o ManagedObjectReference complianceManager o ManagedObjectReference localizationManager o + ManagedObjectReference storageResourceManager o + ManagedObjectReference guestOperationsManager o end @@ -1073,6 +1075,7 @@ object UserSession DateTime lastActiveTime r String locale r String messageLocale r + Boolean extensionSession o end -- 1.7.9.5

3 3

[libvirt] [PATCH] sanlock: Chown lease files as well
by Michal Privoznik 31 Dec '12

31 Dec '12

Since sanlock doesn't run under root:root, we have chown()'ed the __LIBVIRT__DISKS__ lease file to the user:group defined in the sanlock config. However, when writing the patch I've forgot about lease files for each disk (this is the /var/lib/libvirt/sanlock/<md5>) file. --- src/locking/lock_driver_sanlock.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/src/locking/lock_driver_sanlock.c b/src/locking/lock_driver_sanlock.c index 75ced84..c955003 100644 --- a/src/locking/lock_driver_sanlock.c +++ b/src/locking/lock_driver_sanlock.c @@ -679,6 +679,17 @@ static int virLockManagerSanlockCreateLease(struct sanlk_resource *res) } VIR_DEBUG("Someone else just created lockspace %s", res->disks[0].path); } else { + /* chown() the path to make sure sanlock can access it */ + if ((driver->user != -1 || driver->group != -1) && + (fchown(fd, driver->user, driver->group) < 0)) { + virReportSystemError(errno, + _("cannot chown '%s' to (%u, %u)"), + res->disks[0].path, + (unsigned int) driver->user, + (unsigned int) driver->group); + goto error_unlink; + } + if ((rv = sanlock_align(&res->disks[0])) < 0) { if (rv <= -200) virReportError(VIR_ERR_INTERNAL_ERROR, -- 1.8.0.2

3 3

[libvirt] [test-API][PATCH V2] managedsave: change the checking method about bypass cache
by hongming 30 Dec '12

30 Dec '12

Check whether the O_DIRECT flag is in use on the managed save file during the duration of the managed save, since that is the real effect of the current implementation of the --bypass-cache flag. Add break condition into the while loop of get_fileflags() method in the PATCH V2 --- repos/managedsave/managedsave.py | 61 ++++++++++++++++++++------------------ 1 files changed, 32 insertions(+), 29 deletions(-) diff --git a/repos/managedsave/managedsave.py b/repos/managedsave/managedsave.py index 4a92151..70d4ebc 100644 --- a/repos/managedsave/managedsave.py +++ b/repos/managedsave/managedsave.py @@ -2,6 +2,8 @@ import os import math +import thread +import time import libvirt from libvirt import libvirtError @@ -39,39 +41,43 @@ def check_savefile_create(*args): logger.info("managed save file exists") return True -def compare_cachedfile(cachebefore, cacheafter): - """Compare cached value before managed save and its value after - managed save """ +def get_fileflags(): + """Get the file flags of managed save file""" + cmds = "cat /proc/$(lsof -w /var/lib/libvirt/qemu/save/"+guestname+".save"\ + "|awk '/libvirt_i/{print $2}')/fdinfo/1|grep flags|awk '{print $NF}'" + global fileflags + while True: + (status, output) = utils.exec_cmd(cmds, shell=True) + if status == 0: + if len(output) == 1: + logger.info("The flags of saved file %s " % output[0]) + fileflags = output[0][-5] + break + else: + logger.error("Fail to get the flags of saved file") + return 1 + + thread.exit_thread() - diff = cacheafter - cachebefore - logger.info("diff is %s " % diff) - percent = math.fabs(diff)/cachebefore - logger.info("diff percent is %s " % percent) - if percent < 0.05: +def check_fileflag(fileflags): + """Check the file flags of managed save file if include O_DIRECT""" + if int(fileflags) == 4: + logger.info("file flags include O_DIRECT") return True else: + logger.error("file flags doesn't include O_DIRECT") return False -def get_cachevalue(): - """Get the file system cached value """ - - cmds = "head -n4 /proc/meminfo|grep Cached|awk '{print $2}'" - (status, output) = utils.exec_cmd(cmds, shell=True) - if status != 0: - logger.error("Fail to run cmd line to get cache") - return 1 - else: - logger.debug(output[0]) - cachevalue= int(output[0]) - return cachevalue - def managedsave(params): """Managed save a running domain""" global logger logger = params['logger'] + global guestname guestname = params['guestname'] flags = params ['flags'] + global fileflags + fileflags = '' #Save given flags to sharedmod.data sharedmod.data['flagsave'] = flags @@ -122,17 +128,14 @@ def managedsave(params): #If given flags include bypass-cache,check if bypass file system cache if flagn % 2 == 1: - logger.info("Given flags include --bypass-cache") - os.system('echo 3 > /proc/sys/vm/drop_caches') - cache_before = get_cachevalue() - logger.info("Cached value before managedsave is %s" % cache_before) + logger.info("Given flags include --bypass-cache") + thread.start_new_thread(get_fileflags,()) + # Guarantee get_fileflags shell has run before managed save + time.sleep(5) domobj.managedSave(flagn) - cache_after = get_cachevalue() - logger.info("Cached value after managedsave is %s" % cache_after) - - if compare_cachedfile(cache_before, cache_after): + if check_fileflag(fileflags): logger.info("Bypass file system cache successfully") else: logger.error("Bypass file system cache failed") -- 1.7.7.6

2 1