On Wed, 1 Nov 2017 20:45:46 +0100
Wim ten Have <wim.ten.have@xxxxxxxxxx> wrote:
> On Thu, 26 Oct 2017 17:51:34 -0600
> Jim Fehlig <jfehlig@xxxxxxxx> wrote:
>
> > On 10/12/2017 01:31 PM, Wim Ten Have wrote:
> > > From: Wim ten Have <wim.ten.have@xxxxxxxxxx>
> > >
> > > Test a bidirectional xen-xl domxml to and from native for numa
> > > support administration as brought under this patch series.
> > >
> > > Signed-off-by: Wim ten Have <wim.ten.have@xxxxxxxxxx>
> > > ---
> > > .../test-fullvirt-vnuma-autocomplete.cfg | 26 +++++++
> > > .../test-fullvirt-vnuma-autocomplete.xml | 85 ++++++++++++++++++++++
> > > .../test-fullvirt-vnuma-nodistances.cfg | 26 +++++++
> > > .../test-fullvirt-vnuma-nodistances.xml | 53 ++++++++++++++
> > > .../test-fullvirt-vnuma-partialdist.cfg | 26 +++++++
> > > .../test-fullvirt-vnuma-partialdist.xml | 60 +++++++++++++++
> > > tests/xlconfigdata/test-fullvirt-vnuma.cfg | 26 +++++++
> > > tests/xlconfigdata/test-fullvirt-vnuma.xml | 81 +++++++++++++++++++++
> > > tests/xlconfigtest.c | 6 ++
> >
> > Cool! Thanks for the various configurations to test all the hairy parsing code :-).
> >
> > Reviewed-by: Jim Fehlig <jfehlig@xxxxxxxx>
> >
> > BTW I should have mentioned it while reviewing 3/4, but we should also have
> > tests for the libxl_domain_config generator, now that we have a test suite for
> > that. See tests/libxlxml2domconfigtest.c.
>
> There's a nasty issue living here. Within specific test-harness you seem to
> get through libxl_ctx_alloc() ... ?! Running as ordinary user without privileges?! (ie. not root)
>
> if (libxl_ctx_alloc(&ctx, LIBXL_VERSION, 0, log) < 0) {
>
> I'd expected this libxl_ctx_alloc() to have failed with;
> xencall: error: Could not obtain handle on privileged command interface: Permission denied
> libxl: error: libxl.c:108:libxl_ctx_alloc: cannot open libxc handle: Permission denied
>
> Funny it seems to go through and the padded memory content seem not to match a
> real libxl_ctx_alloc(). So it is bringing some kind of ctx structure which seems _NOT_ real or
> at least incorrect.
>
> From I need specific context to determine host/hypervisor phys_info() ... causing
> my libxl_domain_config added test to die with a SEGV under specific xenlight run-time.
>
> (gdb) where
> #0 0x00007ffff2e18f41 in xc.hypercall_buffer_alloc () from /lib64/libxenctrl.so.4.8
> #1 0x00007ffff2e18f98 in xc.hypercall_bounce_pre () from /lib64/libxenctrl.so.4.8
> #2 0x00007ffff2e0b962 in xc_physinfo () from /lib64/libxenctrl.so.4.8
> #3 0x00007ffff792c0bb in libxl_get_physinfo () from /lib64/libxenlight.so.4.8
> #4 0x0000000000414012 in libxlMakeVnumaList (def=0x66f720, ctx=0x668060, d_config=0x7fffffffd170)
> at libxl/libxl_conf.c:621
> #5 0x0000000000418ca6 in libxlBuildDomainConfig (graphicsports=0x666940, def=0x66f720, channelDir=0x0,
> ctx=0x668060, caps=0x669ee0, d_config=0x7fffffffd170) at libxl/libxl_conf.c:2302
> #6 0x000000000040f536 in testCompareXMLToDomConfig (
> xmlfile=0x666860 "/home/wtenhave/WORK/libvirt/vNUMA/libvirt/tests/libxlxml2domconfigdata/basic-hvm.xml",
> jsonfile=0x666790 "/home/wtenhave/WORK/libvirt/vNUMA/libvirt/tests/libxlxml2domconfigdata/basic-hvm.json")
> at libxlxml2domconfigtest.c:88
> #7 0x000000000040f851 in testCompareXMLToDomConfigHelper (data=0x6482f0 <info>) at libxlxml2domconfigtest.c:148
>
> Should I escape the libxl_get_physinfo() under test? It is not really necessary there as whole
> is academic. If you have an advise please let me know ... .
Not a nasty issue and resolved under PATCH v6 submitted today.
Regards.
- Wim.
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list
