1. 19 12月, 2019 5 次提交
  2. 18 12月, 2019 17 次提交
    • D
      conf: fix populating of fake NUMA in multi-node hosts · e67ccd3c
      Daniel P. Berrangé 提交于
      If the host OS doesn't have NUMA present, we fallback to
      populating fake NUMA info and the code thus assumes only a
      single NUMA node.
      
      Unfortunately we also fallback to fake NUMA if numactl-devel
      was not present, and in this case we can still have multiple
      NUMA nodes. In this case we create all CPUs, but only the
      CPUs in the first node have any data filled in, resulting in
      capabilities like:
      
          <topology>
            <cells num='1'>
              <cell id='0'>
                <memory unit='KiB'>15977572</memory>
                <cpus num='48'>
                  <cpu id='0' socket_id='0' core_id='0' siblings='0'/>
                  <cpu id='1' socket_id='0' core_id='0' siblings='1'/>
                  <cpu id='2' socket_id='0' core_id='1' siblings='2'/>
                  <cpu id='3' socket_id='0' core_id='1' siblings='3'/>
                  <cpu id='4' socket_id='0' core_id='2' siblings='4'/>
                  <cpu id='5' socket_id='0' core_id='2' siblings='5'/>
                  <cpu id='6' socket_id='0' core_id='3' siblings='6'/>
                  <cpu id='7' socket_id='0' core_id='3' siblings='7'/>
                  <cpu id='8' socket_id='0' core_id='4' siblings='8'/>
                  <cpu id='9' socket_id='0' core_id='4' siblings='9'/>
                  <cpu id='10' socket_id='0' core_id='5' siblings='10'/>
                  <cpu id='11' socket_id='0' core_id='5' siblings='11'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                  <cpu id='0'/>
                </cpus>
              </cell>
            </cells>
          </topology>
      
      With this new code we get something slightly less broken
      
          <topology>
            <cells num='4'>
              <cell id='0'>
                <memory unit='KiB'>15977572</memory>
                <cpus num='12'>
                  <cpu id='0' socket_id='0' core_id='0' siblings='0-1'/>
                  <cpu id='1' socket_id='0' core_id='0' siblings='0-1'/>
                  <cpu id='2' socket_id='0' core_id='1' siblings='2-3'/>
                  <cpu id='3' socket_id='0' core_id='1' siblings='2-3'/>
                  <cpu id='4' socket_id='0' core_id='2' siblings='4-5'/>
                  <cpu id='5' socket_id='0' core_id='2' siblings='4-5'/>
                  <cpu id='6' socket_id='0' core_id='3' siblings='6-7'/>
                  <cpu id='7' socket_id='0' core_id='3' siblings='6-7'/>
                  <cpu id='8' socket_id='0' core_id='4' siblings='8-9'/>
                  <cpu id='9' socket_id='0' core_id='4' siblings='8-9'/>
                  <cpu id='10' socket_id='0' core_id='5' siblings='10-11'/>
                  <cpu id='11' socket_id='0' core_id='5' siblings='10-11'/>
                </cpus>
              </cell>
              <cell id='0'>
                <memory unit='KiB'>15977572</memory>
                <cpus num='12'>
                  <cpu id='12' socket_id='0' core_id='0' siblings='12-13'/>
                  <cpu id='13' socket_id='0' core_id='0' siblings='12-13'/>
                  <cpu id='14' socket_id='0' core_id='1' siblings='14-15'/>
                  <cpu id='15' socket_id='0' core_id='1' siblings='14-15'/>
                  <cpu id='16' socket_id='0' core_id='2' siblings='16-17'/>
                  <cpu id='17' socket_id='0' core_id='2' siblings='16-17'/>
                  <cpu id='18' socket_id='0' core_id='3' siblings='18-19'/>
                  <cpu id='19' socket_id='0' core_id='3' siblings='18-19'/>
                  <cpu id='20' socket_id='0' core_id='4' siblings='20-21'/>
                  <cpu id='21' socket_id='0' core_id='4' siblings='20-21'/>
                  <cpu id='22' socket_id='0' core_id='5' siblings='22-23'/>
                  <cpu id='23' socket_id='0' core_id='5' siblings='22-23'/>
                </cpus>
              </cell>
            </cells>
          </topology>
      
      The topology at least now reflects what 'virsh nodeinfo' reports.
      The main bug is that the CPU "id" values won't match what the Linux
      host actually uses.
      Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
      Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>
      e67ccd3c
    • D
      conf: avoid mem leak re-allocating fake NUMA capabilities · fb5aaf3d
      Daniel P. Berrangé 提交于
      The 'caps' object is already allocated when the fake NUMA
      initialization takes place.
      Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>
      fb5aaf3d
    • D
      qemu_command: tidy up qemuBuildHostdevCommandLine loop · aed9bcd1
      Daniel Henrique Barboza 提交于
      The current 'for' loop with 5 consecutive 'ifs' inside
      qemuBuildHostdevCommandLine can be a bit smarter:
      
      - all 5 'ifs' fails if hostdev->mode is not equal to
      VIR_DOMAIN_HOSTDEV_MODE_SUBSYS. This check can be moved to the
      start of the loop, failing to the next element immediately
      in case it fails;
      
      - all 5 'ifs' checks for a specific subsys->type to build the proper
      command line argument (virHostdevIsSCSIDevice and virHostdevIsMdevDevice
      do that but within a helper). Problem is that the code will keep
      checking for matches even if one was already found, and there is
      no way a hostdev will fit more than one 'if' (i.e. a hostdev can't
      have 2+ different types). This means that a SUBSYS_TYPE_USB will
      create its command line argument in the first 'if', then all other
      conditionals will surely fail but will end up being checked anyway.
      
      All of this can be avoided by moving the hostdev->mode comparing
      to the start of the loop and using a switch statement with
      subsys->type to execute the proper code for a given hostdev
      type.
      Suggested-by: NJán Tomko <jtomko@redhat.com>
      Signed-off-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
      Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
      aed9bcd1
    • D
    • D
      access: report an error if no access manager is present · cccc3fc1
      Daniel P. Berrangé 提交于
      The code calling this method expects it to have reported an error on
      failure.
      Reviewed-by: NCole Robinson <crobinso@redhat.com>
      Signed-off-by: NDaniel P. Berrangé <berrange@redhat.com>
      cccc3fc1
    • M
      qemu: Don't leak hostcpu or hostnuma on driver cleanup · 39a7dff7
      Michal Privoznik 提交于
      When freeing qemu driver struct members, we forgot to free
      @hostcpu and @hostnuma members.
      Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
      Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      39a7dff7
    • M
      qemu: Reorder cleanup in qemuStateCleanup() · 7cf76d4e
      Michal Privoznik 提交于
      This function is supposed to clean up virQEMUDriver structure and
      free individual members. However, it's doing that in random order
      which makes it hard to track which members are being freed and
      which are not. Do the free in reverse order than the structure
      definition - assuming that the most important members (like
      mutex) are declared first and freed last.
      Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
      Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      7cf76d4e
    • M
      virCapabilitiesHostNUMAUnref: Accept NULL · 599f9c73
      Michal Privoznik 提交于
      Fortunately, this is not causing any problems now because glib
      does this check for us when calling this function via attribute
      cleanup. But in a future commit we will explicitly call this
      function over a struct member that might be NULL.
      Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
      Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      599f9c73
    • M
      testutilsxen: Avoid double free of driver caps · 98f1f4a4
      Michal Privoznik 提交于
      In testXLInitDriver() a dummy driver structure is filled and it
      is freed later in testXLFreeDriver(). However, it is sufficient
      to unref just driver->config because that results in
      libxlDriverConfigDispose() being called which unrefs
      driver->config->caps. There is no need to unref it again in
      testXLFreeDriver() - in fact it's undesired.
      Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
      Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      98f1f4a4
    • M
      domaincapstest: Don't leak cpu definitions · 08a7e88b
      Michal Privoznik 提交于
      When generating domain capabilities, we need to fake host CPU to
      get reproducible result. We do this by copying a pre-existent CPU
      config and setting VIR_TEST_MOCK_FAKE_HOST_CPU env variable which
      is then consumed by qemucpumock. However, we forget to free the
      CPU copy afterwards.
      
       2,196 (2,016 direct, 180 indirect) bytes in 18 blocks are definitely lost in loss record 291 of 297
          at 0x4838B86: calloc (vg_replace_malloc.c:762)
          by 0x57CB6A0: g_malloc0 (in /usr/lib64/libglib-2.0.so.0.6000.7)
          by 0x4A0F72D: virCPUDefNew (cpu_conf.c:87)
          by 0x4A0FAC7: virCPUDefCopyWithoutModel (cpu_conf.c:235)
          by 0x4A0FBBE: virCPUDefCopy (cpu_conf.c:273)
          by 0x10E3C0: testUtilsHostCpusGetDefForArch (testutilshostcpus.h:157)
          by 0x10E3C0: fakeHostCPU (domaincapstest.c:61)
          by 0x10E3C0: fillQemuCaps (domaincapstest.c:86)
          by 0x10E3C0: test_virDomainCapsFormat (domaincapstest.c:234)
          by 0x10F4BC: virTestRun (testutils.c:146)
          by 0x10DE93: doTestQemuInternal (domaincapstest.c:301)
          by 0x10E13D: doTestQemu (domaincapstest.c:332)
          by 0x1124CF: testQemuCapsIterate (testutilsqemu.c:635)
          by 0x10DCE3: mymain (domaincapstest.c:435)
          by 0x10FD8B: virTestMain (testutils.c:916)
      Signed-off-by: NMichal Privoznik <mprivozn@redhat.com>
      Reviewed-by: NDaniel Henrique Barboza <danielhb413@gmail.com>
      08a7e88b
    • D
    • P
    • P
    • P
      util: storage: Don't treat files with missing backing store format as 'raw' · 3615e8b3
      Peter Krempa 提交于
      Assuming that the backing image format is raw is wrong when doing image
      detection:
      
      1) In -drive mode qemu will still probe the image format of the backing
         image. This means it will try to open a backing file of the image
         which will fail if a more advanced security model is in use.
      
      2) In blockdev mode the image will be opened as raw actually which is
         wrong since it might be qcow. Not opening the backing images will
         also end up in the guest seeing corrupted data.
      
      Rather than attempt to solve various corner cases when us assuming the
      storage file being raw and actually being right forbid startup when the
      guest image doesn't have the format specified in the metadata.
      
      https://bugzilla.redhat.com/show_bug.cgi?id=1588373Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
      Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
      3615e8b3
    • P
      tests: storage: Remove unused test modes · a6493694
      Peter Krempa 提交于
      EXP_WARN and ALLOW_PROBE flags for the testStorageChain cases are no
      longer used so we can remove them.
      Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
      Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
      a6493694
    • P
      tests: storage: Use strict version of virStorageFileGetMetadata · 7e582fe9
      Peter Krempa 提交于
      Pass in 'true' as '@report_broken' of virStorageFileGetMetadata to make
      it fail in the tests. The most important code paths (when starting the
      VM) expect this function to fail rather than silently return partial
      data. Switch the test to exercise this more important code path.
      Signed-off-by: NPeter Krempa <pkrempa@redhat.com>
      Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
      7e582fe9
    • L
      qemu: homogenize MAC address in live & config when hotplugging a netdev · 6c17606b
      Laine Stump 提交于
      Prior to commit 55ce6564 (first in libvirt 4.6.0), the XML sent to
      virDomainAttachDeviceFlags() was parsed only once, and the results of
      that parse were inserted into both the live object of the running
      domain and into the persistent config. Thus, if MAC address was
      omitted from in XML for a network device (<interface>), both the live
      and config object would have the same MAC address.
      
      Commit 55ce6564 changed the code to parse the incoming XML twice -
      once for live and once for config. This does eliminate the problem of
      PCI (/scsi/sata) address conflicts caused by allocating an address
      based on existing devices in live object, but then inserting the
      result into the config (which may already have a device using that
      address), BUT it also means that when the MAC address of a network
      device hasn't been specified in the XML, each copy will get a
      different auto-generated MAC address.
      
      This results in the MAC address of the device changing the next time
      the domain is shutdown and restarted, which creates havoc with the
      guest OS's network config.
      
      There have been several discussions about this in the last > 1 year,
      attempting to find the ideal solution to this problem that makes MAC
      addresses consistent and accounts for all sorts of corner cases with
      PCI/scsi/sata addresses. All of these discussions fizzled out because
      every proposal was either too difficult to implement or failed to fix
      some esoteric case someone thought up.
      
      So, in the interest of solving the MAC address problem while not
      making the "other address" situation any worse than before, this patch
      simply adds a qemuDomainAttachDeviceLiveAndConfigHomogenize() function
      that (for now) copies the MAC address from the config object to the
      live object (if the original xml had <mac address='blah'/> then this
      will be an effective NOP (as the macs already match)).
      
      Any downstream libvirt containing upstream commit
      55ce6564 should have this patch as well.
      
      https://bugzilla.redhat.com/1783411Signed-off-by: NLaine Stump <laine@redhat.com>
      Reviewed-by: NMichal Privoznik <mprivozn@redhat.com>
      6c17606b
  3. 17 12月, 2019 18 次提交