** Description changed:

+ [Impact]
+ 
+  * The new version of open-vm-tools we recently SRUed had an issue
+    that if started in non-VMWare environments (protected by systemd 
+    condition-virtualization, but people fake VMware envs) would segfault.
+ 
+  * It is fine to "not work there" but segfaulting is too hard
+ 
+  * Upstream realized the issue, this is backporting the fix
+ 
+ [Test Case]
+ 
+  * Use a non-VMWare guest (e.g. a test system lxd container) and run
+     $ vmtoolsd
+ 
+  * This will segfault, with the fix it will have a clean exit.
+ 
+ [Regression Potential]
+ 
+  * All of this only affects non-VMWare environments in a tool only needed 
+    in those environments - so the risk should be low. It is closing a 
+    corner case of faked environments to keep bug reports and crashes down.
+    I can think of a regression if the detection of the VMContext would be 
+    wrong and therefore the service not properly initializing anymore in 
+    systems it should do so - so far such an issue was not seen, btu that 
+    would be the most likely potential regression.
+ 
+ [Other Info]
+  
+  * n/a
+ 
+ 
+ ---
+ 
  Hi,
  
  Crash reports like:
  vmtoolsd (11) /usr/lib/libvmtools.so.0.0.0+2839c → [stack]+1ecfc → 
/usr/lib/libvmtools.so.0.0.0+4eb05 → 
/usr/lib/open-vm-tools/plugins/common/libhgfsServer.so+ccf → 
/usr/bin/vmtoolsd+5e98 → /usr/bin/vmtoolsd+4ce3 → /usr/bin/vmtoolsd+3ea8
  
  The cmdline of the failing process had:
  /usr/bin/vmtoolsd -n vmusr --blockFd 3
  
  I have a disassembly thou which looks like:
  
  => 0x7fedb35b739c <Backdoor_InOut+28>:        in (%dx),%eax
  0x7fedb35b739d <Backdoor_InOut+29>:   xchg %rax,(%rsp)
  0x7fedb35b73a1 <Backdoor_InOut+33>:   mov %rdi,0x28(%rax)
  0x7fedb35b73a5 <Backdoor_InOut+37>:   mov %rsi,0x20(%rax)
  0x7fedb35b73a9 <Backdoor_InOut+41>:   mov %rdx,0x18(%rax)
  0x7fedb35b73ad <Backdoor_InOut+45>:   mov %rcx,0x10(%rax)
  0x7fedb35b73b1 <Backdoor_InOut+49>:   mov %rbx,0x8(%rax)
  0x7fedb35b73b5 <Backdoor_InOut+53>:   popq (%rax)
  0x7fedb35b73b7 <Backdoor_InOut+55>:   pop %rbx
- 0x7fedb35b73b8 <Backdoor_InOut+56>:   retq 
+ 0x7fedb35b73b8 <Backdoor_InOut+56>:   retq
  0x7fedb35b73b9:       nopl 0x0(%rax)
  0x7fedb35b73c0 <BackdoorHbIn>:        push %rbx
  0x7fedb35b73c1 <BackdoorHbIn+1>:      mov %rdi,%rax
  0x7fedb35b73c4 <BackdoorHbIn+4>:      push %rbp
  0x7fedb35b73c5 <BackdoorHbIn+5>:      push %rax
  0x7fedb35b73c6 <BackdoorHbIn+6>:      mov 0x30(%rax),%rbp
  
  We don't have much more yet.
  I filed an issue upstream [2] last week in case this showed up somewhere else 
as well, but haven't seen any response yet.
  
  Crash reports overview (might need ubuntu signin) are at [3] and one
  example is [4]
  
  [1]: https://bugs.launchpad.net/ubuntu/+source/open-vm-tools/+bug/1784638
  [2]: https://github.com/vmware/open-vm-tools/issues/284
  [3]: 
https://errors.ubuntu.com/?release=Ubuntu%2018.04&package=open-vm-tools&period=week&version=2%3A10.3.0-0ubuntu1~18.04.1
  [4]: 
https://errors.ubuntu.com/problem/bff98ff07d389a1a92b6cd538a823c1e2ce1a193

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1791220

Title:
  increased crash rate since 10.3 upgrade is available

To manage notifications about this bug go to:
https://bugs.launchpad.net/open-vm-tools/+bug/1791220/+subscriptions

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to