[libvirt] [PATCH v8 00/18] Add support for TPM emulator (for 4.5)
Stefan Berger
stefanb at linux.vnet.ibm.com
Tue Jun 5 11:28:27 UTC 2018
On 06/05/2018 01:03 AM, Marc Hartmayer wrote:
> On Thu, May 24, 2018 at 10:25 PM +0200, Stefan Berger <stefanb at linux.vnet.ibm.com> wrote:
>> This series of patches adds support for the TPM emulator backend that
>> is available in QEMU and based on swtpm + libtpms. It allows to attach a
>> TPM 1.2 or 2 to a QEMU VM. sVirt labels are used for labeling the swtpm
>> process, its Unix socket, and log file with the same label that the
>> QEMU process gets. Besides that swtpm is added to the emulator cgroup to
>> restrict its CPU usage.
>>
>> The device XML can be changed from a TPM 1.2 to a TPM 2 and back to a
>> TPM 1.2. The device state is not removed during those changes but only
>> when the domain is undefined.
>>
>> The swtpm needs persistent storage to store its state. For that I am
>> using the uuid of the VM as part of the path since the name of the VM
>> can be changed. Logfiles, PID files, and socket names are based on the
>> name of the VM, though.
>>
>> Stefan
>>
>> v7->v8:
>> - Delaying this series for 4.5; adjusted references to 4.4
>> - Fixed a test case since version='1.2' is now formatted as well
>> - Appended patches for AppArmor and auditing
>> - Appended patches that improve / fix existing code
>> - patch for validating the TPM configuration rather than overwriting it;
>> a particular case is the CRB interface does not work with a TPM 1.2
>> - swtpm_setup can be run for a TPM 2 in unprivileged mode as well
>>
>> v6->v7:
>> - followed Jan Tomko's suggestion with resulting changing to patch
>> 10/12.
>> - re-added missing parts related to swtpm_setup and TPM that got lost
>> in v4
>>
>> v5->v6:
>> - Addressed John Ferlan's comments
>> - rebased on latest tip
>> - Added patch 12.
>>
>> v4->v5:
>> - Addressed John Ferlan's, Boris Fiuczysnki's and Marc Hartmayer's comments
>> - rebased on latest tip
>>
>> v3->v4:
>> - Addressed John Ferlan's comments
>> - Fixed bugs I found while testing
>> - rebased on latest tip
>>
>>
>> Stefan Berger (18):
>> conf: Add support for external swtpm TPM emulator to domain XML
>> qemu: Extend QEMU capabilities with 'tpm-emulator'
>> util: Implement virFileChownFiles()
>> security: Add DAC and SELinux security for tpm-emulator
>> qemu: Extend qemu_conf with tpm-emulator support
>> qemu: Extend QEMU with external TPM support
>> qemu: Add support for external swtpm TPM emulator
>> tests: Add test cases for external swtpm TPM emulator
>> security: Label the external swtpm with SELinux labels
>> conf: Add support for choosing emulation of a TPM 2
>> qemu: Add swtpm to emulator cgroup
>> news: Update news with new TPM emulator feature
>> security: Add swtpm paths to the domain's AppArmor profile
>> qemu: Run swtpm_setup in unprivileged mode for a TPM 2
>> qemu: Validate chosen TPM model rather than overwriting it
>> conf: Audit TPM emulator device at domain startup
>> conf: Use resrc=tpm in case of TPM passthrough following docs
>> conf: Use virDomainChrSourceDefClear() rather than VIR_FREE()
>>
>> docs/auditlog.html.in | 2 +-
>> docs/formatdomain.html.in | 43 +
>> docs/news.xml | 13 +
>> docs/schemas/domaincommon.rng | 17 +
>> examples/apparmor/libvirt-qemu | 3 +
>> libvirt.spec.in | 2 +
>> src/conf/domain_audit.c | 20 +-
>> src/conf/domain_conf.c | 49 +-
>> src/conf/domain_conf.h | 15 +
>> src/libvirt_private.syms | 3 +
>> src/qemu/Makefile.inc.am | 10 +
>> src/qemu/libvirtd_qemu.aug | 5 +
>> src/qemu/qemu.conf | 8 +
>> src/qemu/qemu_capabilities.c | 5 +
>> src/qemu/qemu_capabilities.h | 1 +
>> src/qemu/qemu_cgroup.c | 36 +
>> src/qemu/qemu_cgroup.h | 2 +
>> src/qemu/qemu_command.c | 34 +-
>> src/qemu/qemu_conf.c | 43 +
>> src/qemu/qemu_conf.h | 6 +
>> src/qemu/qemu_domain.c | 31 +-
>> src/qemu/qemu_extdevice.c | 180 ++++
>> src/qemu/qemu_extdevice.h | 59 ++
>> src/qemu/qemu_process.c | 16 +
>> src/qemu/qemu_security.c | 69 ++
>> src/qemu/qemu_security.h | 11 +
>> src/qemu/qemu_tpm.c | 922 +++++++++++++++++++++
>> src/qemu/qemu_tpm.h | 56 ++
>> src/qemu/test_libvirtd_qemu.aug.in | 2 +
>> src/security/security_dac.c | 7 +
>> src/security/security_driver.h | 7 +
>> src/security/security_manager.c | 36 +
>> src/security/security_manager.h | 6 +
>> src/security/security_selinux.c | 172 ++++
>> src/security/security_stack.c | 40 +
>> src/security/virt-aa-helper.c | 24 +
>> src/util/virfile.c | 55 ++
>> src/util/virfile.h | 3 +
>> tests/qemucapabilitiesdata/caps_2.11.0.s390x.xml | 1 +
>> tests/qemucapabilitiesdata/caps_2.12.0.aarch64.xml | 1 +
>> tests/qemucapabilitiesdata/caps_2.12.0.ppc64.xml | 1 +
>> tests/qemucapabilitiesdata/caps_2.12.0.s390x.xml | 1 +
>> tests/qemucapabilitiesdata/caps_2.12.0.x86_64.xml | 1 +
>> .../tpm-emulator-tpm2.x86_64-latest.args | 33 +
>> tests/qemuxml2argvdata/tpm-emulator-tpm2.xml | 30 +
>> .../tpm-emulator.x86_64-latest.args | 33 +
>> tests/qemuxml2argvdata/tpm-emulator.xml | 30 +
>> tests/qemuxml2argvtest.c | 16 +-
>> tests/qemuxml2xmloutdata/tpm-emulator-tpm2.xml | 34 +
>> tests/qemuxml2xmloutdata/tpm-emulator.xml | 34 +
>> tests/qemuxml2xmltest.c | 1 +
>> 51 files changed, 2212 insertions(+), 17 deletions(-)
>> create mode 100644 src/qemu/qemu_extdevice.c
>> create mode 100644 src/qemu/qemu_extdevice.h
>> create mode 100644 src/qemu/qemu_tpm.c
>> create mode 100644 src/qemu/qemu_tpm.h
>> create mode 100644 tests/qemuxml2argvdata/tpm-emulator-tpm2.x86_64-latest.args
>> create mode 100644 tests/qemuxml2argvdata/tpm-emulator-tpm2.xml
>> create mode 100644 tests/qemuxml2argvdata/tpm-emulator.x86_64-latest.args
>> create mode 100644 tests/qemuxml2argvdata/tpm-emulator.xml
>> create mode 100644 tests/qemuxml2xmloutdata/tpm-emulator-tpm2.xml
>> create mode 100644 tests/qemuxml2xmloutdata/tpm-emulator.xml
>>
>> --
>> 2.14.3
> Hi Stefan,
>
> if I restart libvirtd while the TPM-guest is running, the guest crashes…
> I haven’t had a closer look yet, but can you please verify the behavior
> for you? Thanks!
I do not see this with the current version (v9) I posted yesterday. Is
this happening only when a vTPM is attached or also in other cases?
Though. I had seen such behavior a while ago but then it disappeared and
iirc it wasn't limited to guests with an attached vTPM. I think it was
due to libvirt terminating the guest for some reason.
Stefan
>
> Beste Grüße / Kind regards
> Marc Hartmayer
>
> IBM Deutschland Research & Development GmbH
> Vorsitzende des Aufsichtsrats: Martina Koederitz
> Geschäftsführung: Dirk Wittkopp
> Sitz der Gesellschaft: Böblingen
> Registergericht: Amtsgericht Stuttgart, HRB 243294
More information about the libvir-list
mailing list