Bug 1441646

Created attachment 1271121 [details]
GDB traceback of libvirtd during guest boot

The root cause is: The cause for the crash is: Upon a guest boot, if you copy host vendor CPUID to the guest CPU, libvirtd would crash if that host CPU had a NULL vendor field.  Indeed, from another GDB session, we could see the 'vendor_id' to be '0x0'

-----
[...]
(gdb) p host
$4 = (virCPUDef *) 0x7fa38c1d2930
(gdb) p* host
$5 = {type = 0, mode = 0, match = 0, arch = VIR_ARCH_X86_64, model = 0x7fa38c1d3510 "qemu64", vendor_id = 0x0, fallback = 0, vendor = 0x7fa38c1d34f0 "AMD", sockets = 4, cores = 1, 
  threads = 1, nfeatures = 25, nfeatures_max = 0, features = 0x7fa38c1d35d0}
(gdb) p *cpu
$6 = {type = 1, mode = 1, match = 1, arch = VIR_ARCH_NONE, model = 0x7fa3c4014ab0 "qemu64", vendor_id = 0x0, fallback = 0, vendor = 0x7fa3c40149f0 "AMD", sockets = 1, cores = 1, 
  threads = 1, nfeatures = 25, nfeatures_max = 25, features = 0x7fa3c400f5b0}
(gdb) down
#2  0x00007fa3ff4dd917 in x86Compute (host=<optimized out>, cpu=0x7fa3c400eea0, guest=0x7fa3eeb52360, message=<optimized out>) at cpu/cpu_x86.c:1604
1604                virCPUx86DataAddCPUID(&guest_model->data,
[...]
-----


After a GDB session with Jiri Denemark (thanks!), he identified the commit that fixed it upstream libvirt:

$ git show 541e9ae6d4
commit 541e9ae6d4290b9004ed73648ea663563b329b3d
Author: Jim Fehlig <jfehlig>
Date:   Fri Aug 5 15:23:47 2016 -0600

    cpu_x86: fix libvirtd crash when host cpu vendor is not available
    
    When starting a guest and copying host vendor cpuid to the guest
    cpu, libvirtd would crash if the host cpu contained a NULL vendor
    field. Avoid the crash by checking for a valid vendor in the host
    cpu before copying the cpuid to the guest cpu.
    
    For completeness, here is a backtrace from the crash
    
    (gdb) bt
    f0  0x00007ffff739bf33 in x86DataCpuid (cpuid=0x8, cpuid=0x8,
        data=data@entry=0x7fffb800ee78) at cpu/cpu_x86.c:287
    f1  virCPUx86DataAddCPUID (data=data@entry=0x7fffb800ee78, cpuid=0x8)
        at cpu/cpu_x86.c:355
    f2  0x00007ffff739ef47 in x86Compute (host=<optimized out>, cpu=0x7fffb8000cc0,
        guest=0x7fffecca7348, message=<optimized out>) at cpu/cpu_x86.c:1580
    f3  0x00007fffd2b38e53 in qemuBuildCpuModelArgStr (migrating=false,
        hasHwVirt=<synthetic pointer>, qemuCaps=0x7fffb8001040, buf=0x7fffecca7360,
        def=0x7fffc400ce20, driver=0x1c) at qemu/qemu_command.c:6283
    f4  qemuBuildCpuCommandLine (cmd=cmd@entry=0x7fffb8002f60,
        driver=driver@entry=0x7fffc80882c0, def=def@entry=0x7fffc400ce20,
        qemuCaps=qemuCaps@entry=0x7fffb8001040, migrating=<optimized out>)
        at qemu/qemu_command.c:6445
    (gdb) f2
    (gdb) p *host_model
    $23 = {name = 0x7fffb800ec50 "qemu64", vendor = 0x0, signature = 0, data = {
        len = 2, data = 0x7fffb800e720}}

diff --git a/src/cpu/cpu_x86.c b/src/cpu/cpu_x86.c
index 670b02e..ee5b57d 100644
--- a/src/cpu/cpu_x86.c
+++ b/src/cpu/cpu_x86.c
@@ -1592,7 +1592,7 @@ x86Compute(virCPUDefPtr host,
         if (!(guest_model = x86ModelCopy(host_model)))
             goto error;
 
-        if (cpu->vendor &&
+        if (cpu->vendor && host_model->vendor &&
             virCPUx86DataAddCPUID(&guest_model->data,
                                   &host_model->vendor->cpuid) < 0)
             goto error;

*** Bug 1441655 has been marked as a duplicate of this bug. ***

Some more details about this bug... libvirt stores its CPU model definitions in cpu_map.xml (installed in /usr/share/libvirt), where some models (usually older or artificial) are not defined with a specific <vendor>...</vendor> element. If libvirt decides to use one of these models as the model which best describes the host CPU, it will crash everytime it tries to start a domain.

So while this can easily be reproduced in a nested environment (it's trivial to change the host CPU nested libvirt will see), it is not completely impossible to hit this bug with a real hardware. Although the CPU would need to be either pretty old or very strange.

Verified with libvirt-3.2.0-4.el7.x86_64 and qemu-kvm-rhev-2.9.0-3.el7.x86_64 in host .
L1 xml is as below-
  <cpu mode='host-passthrough'>
    <model fallback='allow'/>
  </cpu>

L2 xml:

<cpu mode='host-model'>
    <model fallback='allow'/>
 </cpu>

L2 vm can be started successfully.

Verified with libvirt-3.2.0-4.el7.x86_64 and qemu-kvm-rhev-2.9.0-3.el7.x86_64 in host .
L1 xml is as below-
  <cpu mode='host-passthrough'>
    <model fallback='allow'/>
  </cpu>

L2 xml:

<cpu mode='host-model'>
    <model fallback='allow'/>
 </cpu>

L2 vm can be started successfully.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHEA-2017:1846

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHEA-2017:1846