October 2019 - Devel - libvirt List Archives

[libvirt] [PATCH] qemu: fix EFI nvram removal on domain undefine
by Pavel Mores 15 Oct '19

15 Oct '19

When undefining a UEFI domain its nvram file has to be properly handled as well. It's mandatory to use one of --nvram and --keep-nvram options when 'virsh undefine <domain>' is issued for a UEFI domain. To fix the bug as reported, virsh should return an error message if neither option is used and the nvram file should be removed when --nvram is given. The cause of the problem is that when qemuDomainUndefineFlags() is invoked on an inactive domain the path to its nvram file is empty. This commit aims to fix this by formatting and filling in the path in time for the nvram removal code to run properly. https://bugzilla.redhat.com/show_bug.cgi?id=1751596 Signed-off-by: Pavel Mores <pmores(a)redhat.com> --- src/qemu/qemu_domain.c | 12 +++++++++--- src/qemu/qemu_domain.h | 4 ++++ src/qemu/qemu_driver.c | 20 +++++++++++++++----- 3 files changed, 28 insertions(+), 8 deletions(-) diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index 35067c851f..1a0367cc27 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -15441,16 +15441,22 @@ qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk) } +int +qemuDomainNVRAMPathFormat(virQEMUDriverConfigPtr cfg, + virDomainDefPtr def, char **path) +{ + return virAsprintf(path, "%s/%s_VARS.fd", cfg->nvramDir, def->name); +} + int qemuDomainNVRAMPathGenerate(virQEMUDriverConfigPtr cfg, virDomainDefPtr def) { if (def->os.loader && def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH && - def->os.loader->readonly == VIR_TRISTATE_SWITCH_ON && + def->os.loader->readonly == VIR_TRISTATE_BOOL_YES && !def->os.loader->nvram) { - return virAsprintf(&def->os.loader->nvram, "%s/%s_VARS.fd", - cfg->nvramDir, def->name); + return qemuDomainNVRAMPathFormat(cfg, def, &def->os.loader->nvram); } return 0; diff --git a/src/qemu/qemu_domain.h b/src/qemu/qemu_domain.h index 01a54d4265..07725c7cc4 100644 --- a/src/qemu/qemu_domain.h +++ b/src/qemu/qemu_domain.h @@ -1207,6 +1207,10 @@ qemuDomainIsUsingNoShutdown(qemuDomainObjPrivatePtr priv); bool qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk); +int +qemuDomainNVRAMPathFormat(virQEMUDriverConfigPtr cfg, + virDomainDefPtr def, char **path); + int qemuDomainNVRAMPathGenerate(virQEMUDriverConfigPtr cfg, virDomainDefPtr def); diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index bc0ede2fb0..dcaadbdb52 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -7828,6 +7828,8 @@ qemuDomainUndefineFlags(virDomainPtr dom, int nsnapshots; int ncheckpoints; virQEMUDriverConfigPtr cfg = NULL; + char *nvram_path = NULL; + bool need_deallocate_nvram_path = false; virCheckFlags(VIR_DOMAIN_UNDEFINE_MANAGED_SAVE | VIR_DOMAIN_UNDEFINE_SNAPSHOTS_METADATA | @@ -7905,14 +7907,20 @@ qemuDomainUndefineFlags(virDomainPtr dom, } } - if (vm->def->os.loader && - vm->def->os.loader->nvram && - virFileExists(vm->def->os.loader->nvram)) { + if (vm->def->os.firmware == VIR_DOMAIN_OS_DEF_FIRMWARE_EFI) { + qemuDomainNVRAMPathFormat(cfg, vm->def, &nvram_path); + need_deallocate_nvram_path = true; + } else { + if (vm->def->os.loader) + nvram_path = vm->def->os.loader->nvram; + } + + if (nvram_path && virFileExists(nvram_path)) { if ((flags & VIR_DOMAIN_UNDEFINE_NVRAM)) { - if (unlink(vm->def->os.loader->nvram) < 0) { + if (unlink(nvram_path) < 0) { virReportSystemError(errno, _("failed to remove nvram: %s"), - vm->def->os.loader->nvram); + nvram_path); goto endjob; } } else if (!(flags & VIR_DOMAIN_UNDEFINE_KEEP_NVRAM)) { @@ -7948,6 +7956,8 @@ qemuDomainUndefineFlags(virDomainPtr dom, virDomainObjEndAPI(&vm); virObjectEventStateQueue(driver->domainEventState, event); virObjectUnref(cfg); + if (need_deallocate_nvram_path) + VIR_FREE(nvram_path); return ret; } -- 2.21.0

2 4

[libvirt] [PATCH 0/2] conf: Fix up empty 'network' cdroms
by Peter Krempa 15 Oct '19

15 Oct '19

Peter Krempa (2): conf: Reset disk type if <source> element is completely missing tests: Add test case for empty 'network' cdrom src/conf/domain_conf.c | 4 +++ ...m-empty-network-invalid.x86_64-latest.args | 34 +++++++++++++++++++ .../disk-cdrom-empty-network-invalid.xml | 30 ++++++++++++++++ tests/qemuxml2argvtest.c | 1 + ...om-empty-network-invalid.x86_64-latest.xml | 34 +++++++++++++++++++ tests/qemuxml2xmloutdata/disk-source-pool.xml | 2 +- tests/qemuxml2xmltest.c | 1 + 7 files changed, 105 insertions(+), 1 deletion(-) create mode 100644 tests/qemuxml2argvdata/disk-cdrom-empty-network-invalid.x86_64-latest.args create mode 100644 tests/qemuxml2argvdata/disk-cdrom-empty-network-invalid.xml create mode 100644 tests/qemuxml2xmloutdata/disk-cdrom-empty-network-invalid.x86_64-latest.xml -- 2.21.0

2 3

[libvirt] [PULL 00/19] Bitmaps patches
by John Snow 15 Oct '19

15 Oct '19

The following changes since commit 98b2e3c9ab3abfe476a2b02f8f51813edb90e72d: Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-request' into staging (2019-10-08 16:08:35 +0100) are available in the Git repository at: https://github.com/jnsnow/qemu.git tags/bitmaps-pull-request for you to fetch changes up to b97d9a1014b61dd0980e7f4a0c9ca1e3b0aaa761: dirty-bitmaps: remove deprecated autoload parameter (2019-10-09 17:02:45 -0400) ---------------------------------------------------------------- Pull request ---------------------------------------------------------------- John Snow (2): MAINTAINERS: Add Vladimir as a reviewer for bitmaps dirty-bitmaps: remove deprecated autoload parameter Vladimir Sementsov-Ogievskiy (17): util/hbitmap: strict hbitmap_reset block: move bdrv_can_store_new_dirty_bitmap to block/dirty-bitmap.c block/dirty-bitmap: return int from bdrv_remove_persistent_dirty_bitmap block/qcow2: proper locking on bitmap add/remove paths block/dirty-bitmap: drop meta block/dirty-bitmap: add bs link block/dirty-bitmap: drop BdrvDirtyBitmap.mutex block/dirty-bitmap: refactor bdrv_dirty_bitmap_next block: switch reopen queue from QSIMPLEQ to QTAILQ block: reverse order for reopen commits iotests: add test-case to 165 to test reopening qcow2 bitmaps to RW block/qcow2-bitmap: get rid of bdrv_has_changed_persistent_bitmaps block/qcow2-bitmap: drop qcow2_reopen_bitmaps_rw_hint() block/qcow2-bitmap: do not remove bitmaps on reopen-ro iotests: add test 260 to check bitmap life after snapshot + commit block/qcow2-bitmap: fix and improve qcow2_reopen_bitmaps_rw qcow2-bitmap: move bitmap reopen-rw code to qcow2_reopen_commit qemu-deprecated.texi | 20 ++- qapi/block-core.json | 6 +- block/qcow2.h | 19 +-- include/block/block.h | 2 +- include/block/block_int.h | 20 +-- include/block/dirty-bitmap.h | 34 ++-- include/qemu/hbitmap.h | 5 + block.c | 79 +++------ block/backup.c | 14 +- block/dirty-bitmap.c | 290 +++++++++++++++++++-------------- block/mirror.c | 4 +- block/qcow2-bitmap.c | 212 +++++++++++++++--------- block/qcow2.c | 22 ++- blockdev.c | 40 ++--- migration/block-dirty-bitmap.c | 11 +- migration/block.c | 4 +- tests/test-hbitmap.c | 2 +- util/hbitmap.c | 4 + MAINTAINERS | 3 +- tests/qemu-iotests/165 | 57 ++++++- tests/qemu-iotests/165.out | 4 +- tests/qemu-iotests/260 | 89 ++++++++++ tests/qemu-iotests/260.out | 52 ++++++ tests/qemu-iotests/group | 1 + 24 files changed, 624 insertions(+), 370 deletions(-) create mode 100755 tests/qemu-iotests/260 create mode 100644 tests/qemu-iotests/260.out -- 2.21.0

4 25

[libvirt] [virttools-planet PATCH] Add Fabiano Fidencio to Planet Virt-Tools
by Fabiano Fidêncio 15 Oct '19

15 Oct '19

Signed-off-by: Fabiano Fidêncio <fidencio(a)redhat.com> --- updater/virt-tools/config.ini | 3 +++ 1 file changed, 3 insertions(+) diff --git a/updater/virt-tools/config.ini b/updater/virt-tools/config.ini index 6b4b5ec..9375c82 100644 --- a/updater/virt-tools/config.ini +++ b/updater/virt-tools/config.ini @@ -194,3 +194,6 @@ name = Stefano Garzarella face = sgarzare.png facewidth = 89 faceheight = 96 + +[https://fidencio.org/tags/virt/index.xml] +name = Fabiano Fidêncio -- 2.23.0

4 4

[libvirt] [PATCH 00/35] Use more of GLib's macros
by Ján Tomko 15 Oct '19

15 Oct '19

3 42

[libvirt] [PATCH v3 00/19] Integrate usage of glib into libvirt
by Daniel P. Berrangé 15 Oct '19

15 Oct '19

This is a followup to a previous patch series: v0: https://www.redhat.com/archives/libvir-list/2019-August/msg01374.html v1: https://www.redhat.com/archives/libvir-list/2019-September/msg01331.html v2: https://www.redhat.com/archives/libvir-list/2019-October/msg00271.html The focus in this glib series is - Wire up pieces to facilitate interoperability around memory allocation/cleanup. - Deprecate & document existing libvirt APIs that are targetted for removal/conversion. - Illustrate some conversions and/or usage of new APIs. - Convert code that allows us to eliminate more gnulib modules There's obviously alot of conversion work that could be done, especially around memory allocation, auto cleanup and virObject stuff. I'm not intending to do that myself in the short term, as my immediate focus is on eliminating gnulib modules. Once this series is merged though, anyone is able to take on conversion jobs, so this allows the work to be spread out across the contributors. changed in v2: - First patch gets an addition to .travis.yml to add glib2 on macOS - First patch blacklists -Wbad-function-cast warning flag - Second patch updates many makefiles to refer to $(GLIB_LIBS) to fix build on platforms with stricter linkers (Ubuntu) - virIdentity conversion uses _ consistently in methods - GOptionContext conversions are dropped Daniel P. Berrangé (19): build: probe for glib-2 library in configure build: link to glib library util: use glib memory allocation functions util: use glib string allocation/formatting functions util: convert virSystemdActivation to use VIR_DEFINE_AUTOPTR_FUNC util: rewrite auto cleanup macros to use glib's equivalent src: add support for g_autoptr with virObject instances conf: convert virSecretObj APIs to use autofree util: use glib base64 encoding/decoding APIs util: convert virIdentity implementation and test suite to g_autoptr access: convert polkit driver to auto free memory admin: convert admin server code to use auto free macros rpc: convert methods using virIdentityPtr to auto free macros remote: convert methods using virIdentityPtr to auto free macros util: convert virIdentity class to use GObject libxl: convert over to use GRegex for regular expressions conf: convert over to use GRegex for regular expressions util: replace strerror/strerror_r with g_strerror build: remove use of usleep gnulib module in favour of g_usleep .travis.yml | 1 + bootstrap.conf | 8 -- build-aux/syntax-check.mk | 2 +- configure.ac | 7 +- docs/hacking.html.in | 144 ++++++++------------ libvirt.spec.in | 1 + m4/virt-compile-warnings.m4 | 23 ++++ m4/virt-glib.m4 | 36 +++++ mingw-libvirt.spec.in | 2 + src/Makefile.am | 3 + src/access/Makefile.inc.am | 4 +- src/access/viraccessdriverpolkit.c | 38 ++---- src/admin/admin_server.c | 204 ++++++++++++---------------- src/bhyve/Makefile.inc.am | 1 + src/conf/capabilities.h | 3 + src/conf/domain_capabilities.h | 3 + src/conf/domain_conf.h | 3 + src/conf/domain_event.c | 25 ++-- src/conf/snapshot_conf.h | 3 + src/conf/storage_capabilities.h | 3 + src/conf/virsecretobj.c | 68 +++------- src/datatypes.h | 15 ++ src/hyperv/hyperv_driver.c | 2 +- src/hyperv/hyperv_wmi.c | 4 +- src/interface/Makefile.inc.am | 1 + src/internal.h | 1 + src/libvirt_private.syms | 1 - src/libxl/Makefile.inc.am | 1 + src/libxl/libxl_capabilities.c | 42 +++--- src/libxl/libxl_conf.c | 3 +- src/libxl/libxl_conf.h | 2 + src/locking/Makefile.inc.am | 9 +- src/locking/lock_daemon.c | 2 +- src/locking/lock_driver_sanlock.c | 2 +- src/logging/Makefile.inc.am | 1 + src/lxc/Makefile.inc.am | 4 + src/lxc/lxc_controller.c | 2 +- src/lxc/lxc_driver.c | 2 +- src/lxc/lxc_process.c | 2 +- src/network/Makefile.inc.am | 2 + src/network/bridge_driver.c | 2 +- src/node_device/Makefile.inc.am | 5 +- src/nwfilter/Makefile.inc.am | 1 + src/nwfilter/nwfilter_dhcpsnoop.c | 4 +- src/nwfilter/nwfilter_learnipaddr.c | 2 +- src/qemu/Makefile.inc.am | 1 + src/qemu/qemu_agent.c | 6 +- src/qemu/qemu_blockjob.h | 1 + src/qemu/qemu_capabilities.h | 2 + src/qemu/qemu_command.c | 5 +- src/qemu/qemu_conf.h | 3 + src/qemu/qemu_domain.c | 8 +- src/qemu/qemu_monitor_json.c | 2 +- src/qemu/qemu_process.c | 6 +- src/qemu/qemu_tpm.c | 2 +- src/remote/Makefile.inc.am | 2 + src/remote/remote_daemon.c | 3 +- src/remote/remote_daemon_dispatch.c | 35 ++--- src/rpc/virnetserverclient.c | 57 ++++---- src/rpc/virnetserverprogram.c | 13 +- src/rpc/virnetsocket.c | 2 +- src/secret/Makefile.inc.am | 1 + src/secret/secret_driver.c | 1 - src/security/Makefile.inc.am | 1 + src/security/security_manager.c | 2 +- src/storage/Makefile.inc.am | 16 +++ src/storage/storage_backend_rbd.c | 4 +- src/storage/storage_util.c | 4 +- src/util/viralloc.c | 29 +--- src/util/viralloc.h | 14 +- src/util/virautoclean.h | 38 +++--- src/util/vircgroup.c | 2 +- src/util/virerror.c | 9 +- src/util/virerror.h | 1 + src/util/virfile.c | 2 +- src/util/virhostdev.h | 3 + src/util/viridentity.c | 92 +++++++------ src/util/viridentity.h | 7 +- src/util/virmdev.h | 3 + src/util/virnetdev.c | 2 +- src/util/virnetdevip.c | 2 +- src/util/virnetdevmacvlan.c | 2 +- src/util/virnetdevvportprofile.c | 2 +- src/util/virobject.h | 4 + src/util/virpci.c | 8 +- src/util/virpci.h | 3 + src/util/virprocess.c | 4 +- src/util/virresctrl.h | 4 + src/util/virscsi.h | 3 + src/util/virscsivhost.h | 3 + src/util/virstoragefile.h | 2 + src/util/virstring.c | 49 +++---- src/util/virstring.h | 10 +- src/util/virsystemd.c | 10 +- src/util/virsystemd.h | 5 +- src/util/virtime.c | 2 +- src/util/virusb.h | 3 + src/vbox/Makefile.inc.am | 1 + src/vbox/vbox_common.c | 2 +- src/vz/Makefile.inc.am | 1 + tests/Makefile.am | 7 +- tests/commandtest.c | 16 +-- tests/eventtest.c | 4 +- tests/fdstreamtest.c | 4 +- tests/qemumonitortestutils.c | 2 +- tests/seclabeltest.c | 4 +- tests/testutils.c | 6 +- tests/virhostcputest.c | 4 +- tests/viridentitytest.c | 45 +++--- tests/virnetserverclienttest.c | 3 +- tests/virsystemdtest.c | 2 +- tests/virtestmock.c | 4 +- tools/Makefile.am | 4 + tools/virsh-domain.c | 2 +- tools/virsh-secret.c | 17 +-- 115 files changed, 667 insertions(+), 650 deletions(-) create mode 100644 m4/virt-glib.m4 -- 2.21.0

4 26

[libvirt] [PATCH v2] util: Block SIGPIPE until execve in child process
by Wang Yechao 15 Oct '19

15 Oct '19

Libvirtd has set SIGPIPE to ignored, and virFork resets all signal handlers to the defaults. But child process may write logs to stderr/stdout, that may generate SIGPIPE if journald has stopped. So block SIGPIPE in virFork, and unblock it before execve. Signed-off-by: Wang Yechao <wang.yechao255(a)zte.com.cn> --- v1 patch: https://www.redhat.com/archives/libvir-list/2019-October/msg00720.html Changes in v2: - use pthread_sigmask to block SIGPIPE --- src/util/vircommand.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/src/util/vircommand.c b/src/util/vircommand.c index 79e1e87..bd3a582 100644 --- a/src/util/vircommand.c +++ b/src/util/vircommand.c @@ -321,6 +321,15 @@ virFork(void) virDispatchError(NULL); _exit(EXIT_CANCELED); } + + sigemptyset(&newmask); + sigaddset(&newmask, SIGPIPE); + if (pthread_sigmask(SIG_BLOCK, &newmask, NULL) != 0) { + virReportSystemError(errno, "%s", _("cannot block SIGPIPE")); + virDispatchError(NULL); + _exit(EXIT_CANCELED); + } + } return pid; } @@ -553,6 +562,7 @@ virExec(virCommandPtr cmd) struct sigaction waxon, waxoff; VIR_AUTOFREE(gid_t *) groups = NULL; int ngroups; + sigset_t set; if (cmd->args[0][0] != '/') { if (!(binary = binarystr = virFindFileInPath(cmd->args[0]))) { @@ -792,6 +802,13 @@ virExec(virCommandPtr cmd) /* Close logging again to ensure no FDs leak to child */ virLogReset(); + sigemptyset(&set); + sigaddset(&set, SIGPIPE); + if (pthread_sigmask(SIG_SETMASK, &set, NULL) != 0) { + virReportSystemError(errno, "%s", _("cannot unblock SIGPIPE")); + goto fork_error; + } + if (cmd->env) execve(binary, cmd->args, cmd->env); else -- 1.8.3.1

1 0

[libvirt] [PATCH] util: ignore SIGPIPE when writing to stderr/stdout
by Wang Yechao 15 Oct '19

15 Oct '19

libvirtd's stderr/stdout redirected to journald by default, so if the journald has stopped, libvirtd and libvirtd's child process will receive SIGPIPE signal when writing logs to stderr/stdout. journald stopped reasons: 1. manual command "systemctl stop systemd-journald.service" 2. oom killer kill it. ... Signed-off-by: Wang Yechao <wang.yechao255(a)zte.com.cn> --- src/util/virlog.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/src/util/virlog.c b/src/util/virlog.c index 4c76fbc..127e121 100644 --- a/src/util/virlog.c +++ b/src/util/virlog.c @@ -39,6 +39,7 @@ # include <sys/un.h> #endif #include <fnmatch.h> +#include <signal.h> #include "virerror.h" #include "virlog.h" @@ -732,6 +733,9 @@ virLogOutputToFd(virLogSourcePtr source ATTRIBUTE_UNUSED, if (fd < 0) return; + if (fd == STDERR_FILENO || fd == STDOUT_FILENO) + signal(SIGPIPE, SIG_IGN); + if (virAsprintfQuiet(&msg, "%s: %s", timestamp, str) < 0) return; @@ -740,6 +744,10 @@ virLogOutputToFd(virLogSourcePtr source ATTRIBUTE_UNUSED, if (flags & VIR_LOG_STACK_TRACE) virLogStackTraceToFd(fd); + + if (fd == STDERR_FILENO || fd == STDOUT_FILENO) + signal(SIGPIPE, SIG_DFL); + } -- 1.8.3.1

4 6

[libvirt] Question about migration confirm phase
by Jim Fehlig 14 Oct '19

14 Oct '19

I've been investigating a lockd lock ordering bug in a migration error handling path in the libxl driver. In the perform phase, the src calls virDomainLockProcessPause to release the lock before sending the VM to dst. In this case the send fails for other reasons and an attempt is made to reacquire the lock with virDomainLockProcessResume. But that fails since the dst has not finished cleaning up the failed VM and releasing the lock it acquired when starting to receive the VM. My immediate reaction was "why not reacquire the lock in the confirm phase", but then I saw my older comment a few lines later in the perform phase code /* * Confirm phase will not be executed if perform fails. End the * job started in begin phase. */ Is that just a bug in the implementation, or is it intended to skip the confirm phase if perform fails? Regards, Jim

2 2

Re: [libvirt] Add support for vhost-user-scsi-pci/vhost-user-blk-pci
by Cole Robinson 14 Oct '19

14 Oct '19

On 10/14/19 3:12 AM, Li Feng wrote: > Hi Cole & Michal, > > I'm sorry for my late response, I just end my journey today. > Thank your response, your suggestion is very helpful to me. > > I have added Michal in this mail, Michal helps me review my initial patchset. > (https://www.spinics.net/linux/fedora/libvir/msg191339.html) > Whoops I missed that posting, I didn't realize you had sent patches! > All concern about this feature is the XML design. > My original XML design exposes more details of Qemu. > > <vhost-user-blk-pci type='unix'> > <source type='bind' path='/tmp/vhost-blk.sock'> > <reconnect enabled='yes' timeout='5' /> > </source> > <queue num='4'/> > </vhost-user-blk-pci> > > As Cole's suggestion, the better design with all vhost-user-scsi/blk > features would like this: > > vhost-user-blk: > > <disk type='vhostuser' device='disk'> > <source type='unix' path='/path/to/vhost-user-blk.sock' mode='client'> > <reconnect enabled='yes' timeout='5' /> > </source> > <target dev='vda' bus='virtio'/> > <queue num='4'/> > <address type='pci' domain='0x0000' bus='0x00' slot='0x07' function='0x0'/> > </disk> > > vhost-user-scsi: > > <disk type='vhostuser' device='disk'> > <source type='unix' path='/path/to/vhost-user-scsi.sock' mode='client'> > <reconnect enabled='yes' timeout='5' /> > </source> > <target dev='sda' bus='scsi'/> > <queue num='4'/> > </disk> > I think my SCSI idea is wrong, sorry. vhost-user-scsi is for passing a scsi host adapter to the VM, correct? If so, then it's not really a <disk>, and so using the existing vhost-scsi support in <hostdev> is probably better. <hostdev> could possible be used for vhost-user-blk as well Can you provide some examples of full qemu command lines using vhost-user-blk and vhost-user-scsi? Just linking to examples else where is fine, but I'm wondering if there's more context Internally we already have an abstraction for vhost-scsi: <hostdev mode='subsystem' type='scsi_host'> <source protocol='vhost' wwpn='XXX'/> </hostdev> The obvious extension would be <hostdev mode='subsystem' type='scsi_host'> <source protocol='vhostuser' type='unix' path='/path/to/vhost-user-scsi.sock' mode='client'/> </hostdev> Internally implementing this will be weird. The <source> parameters are only dictated by the hostdev type= field, but in this case they would be dictated by the <source protocol=> field, and we would want to reuse the internal chardev abstraction. vhost-user-blk could be implemented similarly, but with type='storage' which is the way we pass through block devices to LXC guests, but it isn't currently supported in the qemu driver. I dunno. Maybe Michal or danpb can provide guidance > Conclusion: > 1. Add new type(vhostuser) for disk label; > 2. Add queue sub-label for disk to support multiqueue(<queue > num='4'/>) or reusing the driver label > (<driver name='vhostuser' queues='4'), which one is better? > Qemu support multiqueue like this: > -device vhost-user-scsi-pci,id=scsi0,chardev=spdk_vhost_scsi0,num_queues=4 > -device vhost-user-blk-pci,chardev=spdk_vhost_blk0,num-queues=4 > num-queues is already supported by libvirt for both <disk> and <hostdev> with <driver queues=X/>, so whether we use <disk> or <hostdev> you won't need to add any new XML here. > Another question: > When qemu is connecting to a vhost-user-scsi controller[1], there may > exist multiple LUNs under one target, > then one disklabel(<disk/>) will represent multiple SCSI LUNs, > the 'dev' property(<target dev='sda' bus='scsi'/>) will be ignored, right? > In other words, for vhost-user-scsi disk, it more likes a controller, > maybe the controller label is suitable. > Yes you are right, and this was my understanding. But then its not really a <controller> in libvirt's sense because we can't attach emulated devices to it, so it's more a fit for the existing <hostdev> vhost-user support. Unfortunately it's not really a clean fit anywhere, there will have to be some kind of compromise. And I'm not sure if <disk> or <hostdev> is right for vhost-user-blk, but hopefully others have more clear opinions. Thanks, Cole > I look forward to hearing from you as soon as possible. > > [1]: https://spdk.io/doc/vhost.html > > Feng Li > > Cole Robinson <crobinso(a)redhat.com> 于2019年10月10日周四上午6:48写道： >> >> Sorry for the late reply, and thanks Jano for pointing out elsewhere >> that this didn't receive a response. >> >> On 8/12/19 5:56 AM, Li Feng wrote: >>> Hi Guys, >>> >>> And I want to add the vhost-user-scsi-pci/vhost-user-blk-pci support >>> for libvirt. >>> >>> The usage in qemu like this: >>> >>> Vhost-SCSI >>> -chardev socket,id=char0,path=/var/tmp/vhost.0 >>> -device vhost-user-scsi-pci,id=scsi0,chardev=char0 >>> Vhost-BLK >>> -chardev socket,id=char1,path=/var/tmp/vhost.1 >>> -device vhost-user-blk-pci,id=blk0,chardev=char1 >>> >> >> Indeed that matches what I see for the qemu commits too: >> >> https://git.qemu.org/?p=qemu.git;a=commit;h=00343e4b54b >> https://git.qemu.org/?p=qemu.git;a=commit;h=f12c1ebddf7 >> >>> What type should I add for libvirt. >>> Type1: >>> <hostdev mode='subsystem' type='vhost-user'> >>> <source protocol='vhost-user-scsi' path='/tmp/vhost-scsi.sock'></source> >>> <alias name="vhost-user-scsi-disk1"/> >>> </hostdev> >>> >>> >>> Type2: >>> >>> <disk type='network' device='disk'> >>> <driver name='qemu' type='raw' cache='none' io='native'/> >>> <source protocol='vhost-user' path='/tmp/vhost-scsi.sock'> >>> </source> >>> <target dev='sdb' bus='vhost-user-scsi'/> >>> <boot order='3'/> >>> <alias name='scsi0-0-0-1'/> >>> <address type='drive' controller='0' bus='0' target='0' unit='1'/> >>> </disk> >>> >>> >>> <disk type='network' device='disk'> >>> <driver name='qemu' type='raw' cache='none' io='native'/> >>> <source protocol='vhost-user' path='/tmp/vhost-blk.sock'> >>> </source> >>> <target dev='vda' bus='vhost-user-blk'/> >>> <boot order='1'/> >>> <alias name='virtio-disk0'/> >>> <address type='pci' domain='0x0000' bus='0x00' slot='0x07' >>> function='0x0'/> >>> </disk> >>> >> >> I think wiring this into <disk> makes more sense. <hostdev> is really an >> abstraction for assigning a (typically) physical host device to the VM, >> so it handles things like hiding a PCI device from the host, and passing >> that exact device to the VM. >> >> In the vhost-user-scsi/blk case, the host device is just a special >> process running on the other side of a socket, and the device >> represented to the guest is a typical virtio device. So to me it makes >> more sense as a <disk> with a <source> that points at that socket. >> >> target bus=virtio vs bus=scsi is already used to distinguish between >> virtio-blk and virtio-scsi, so I think we can keep that bit as is, with >> the <address type=drive|pci> to match. We just need to differentiate >> between plain virtio and vhost-user >> >> network devices already have vhostuser support: >> >> <interface type='vhostuser'> >> <source type='unix' path='/tmp/vhost1.sock' mode='server|client'/> >> <model type='virtio'/> >> </interface> >> >> Internally that <source> is a virDomainChrSourceDefPtr which is our >> internal representation of a chardev. So I think something akin to this >> is the way to go. It will likely require updating a LOT of places in the >> code that check disk type= field, probably most places that care about >> whether type=NETWORK or type=!NETWORK will need to be mirrored for the >> new type. >> >> <disk type='vhostuser' device='disk'> >> <source type='unix' path='/path/to/vhost-user-blk.sock' mode='client'/> >> <target dev='vda' bus='virtio'/> >> </disk> >> >> <disk type='vhostuser' device='disk'> >> <source type='unix' path='/path/to/vhost-user-scsi.sock' mode='client'/> >> <target dev='sda' bus='scsi'/> >> </disk> >> >> - Cole > - Cole

1 0