On 05/23/2018 08:07 AM, John Ferlan wrote:
On 05/22/2018 04:44 PM, Stefan Berger 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
>
> 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 (12):
> 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
>
> docs/formatdomain.html.in | 43 +
> docs/news.xml | 9 +
> docs/schemas/domaincommon.rng | 17 +
> libvirt.spec.in | 2 +
> src/conf/domain_audit.c | 2 +
> src/conf/domain_conf.c | 53 +-
> src/conf/domain_conf.h | 12 +
> 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 | 3 +
> 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 | 946 +++++++++++++++++++++
> 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/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 +
> 48 files changed, 2165 insertions(+), 10 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
>
This all looks good to me - thanks for the news.xml adjustment. Barring
anyone else making a late/additional review - I will look to push the
series later on today.
Thanks.
As mentioned, I will need to follow up with AppArmor support. What I am
currently experimenting with is a subprofile of the libvirt profile. The
problem with it is it's 'suboptimal' in terms of 'unspecific' paths
containing wild-cards so that this single sub profile can accommodate
the paths of all domains. This profile is static and not dynamically
generated.
diff --git a/examples/apparmor/usr.sbin.libvirtd
b/examples/apparmor/usr.sbin.libvirtd
index 3102cab382..dcab4feb6c 100644
--- a/examples/apparmor/usr.sbin.libvirtd
+++ b/examples/apparmor/usr.sbin.libvirtd
@@ -126,4 +126,15 @@
/usr/{lib,lib64,lib/qemu,libexec}/qemu-bridge-helper rmix,
}
+ /usr/bin/swtpm Cx -> usr_bin_swtpm,
+ profile usr_bin_swtpm flags=(complain) {
+ #include <abstractions/base>
+
+ /usr/bin/swtpm rm,
+
+ /run/libvirt/qemu/swtpm/*-swtpm.pid rw,
+ /run/libvirt/qemu/swtpm/*-swtpm.sock w,
+ /var/log/swtpm/libvirt/qemu/*-swtpm.log w,
+
/var/lib/libvirt/swtpm/[0-9a-f]*-[0-9a-f]*-[0-9a-f]*-[0-9a-f]*-[0-9a-f]*/{tpm1.2,tpm2}/{tpm,tpm2}-00.permall
rw,
+ }
}
A better solution would be to extend the QEMU domain profile with these
additional paths, which, as a side-effect, would give a QEMU instance
access to these paths as well. Basically QEMU and swtpm would share that
profile. To good thing is we can use specific paths (no wild cards) for
the files that the swtpm needs to access.
Yet a stricter solution would be to dynamically create a profile
specifically for the swtpm that contains only the necessary paths.
Though I think the code in src/security/security_apparmor.c is not
prepared for that and it may end up being a bigger undertaking.
Anyone who would be willing to give an opinion on the latter two cases ?
I am cc'ing Christian Ehrhard who seems to have worked on AppArmor
related code.
Stefan
John